RMCP Vol. 14 Num. 1 (2023): January-March [english version]

REVISTA MEXICANA

DE CIENCIAS

PECUARIAS Volumen 14 Numero 1, Enero-Marzo 2023. Esunapublicación trimestral deacceso abierto, revisadapor pares yarbitrada, editada por el Instituto Nacional de Investigaciones Forestales, Agrícolas y Pecuarias (INIFAP). Avenida Progreso No. 5, Barrio de Santa Catarina, Delegación Coyoacán, C.P. 04010, Cuidad de México, www.inifap.gob.mx

Distribuida por el Centro de Investigación Regional Sureste, Calle 6 No. 398 X 13, Avenida Correa Racho, Col. Díaz Ordaz, Mérida Yucatán, C.P. 97130.

Editor responsable: Arturo García Fraustro Reservas de Derechos al Uso Exclusivo número 04-2022-033116571100-102. ISSN: 2448-6698, otorgados por el Instituto Nacional del Derecho de Autor (INDAUTOR).

Responsable de la última actualización de este número: Arturo García Fraustro, Campo Experimental Mocochá, Km. 25 Antigua Carretera Mérida–Motul, Mocochá, Yuc. C.P. 97454 http://cienciaspecuarias. inifap.gob.mx, la presente publicación tuvo su última actualización en diciembre de 2022.

DIRECTORIO

EDITOR EN JEFE

Arturo García Fraustro

EDITORES ADJUNTOS

Oscar L. Rodríguez Rivera Alfonso Arias Medina

EDITORES POR DISCIPLINA

Dra. Yolanda Beatriz Moguel Ordóñez, INIFAP, México

Dr. Ramón Molina Barrios, Instituto Tecnológico de Sonora, Dr. Alfonso Juventino Chay Canul, Universidad Autónoma de Tabasco, México

Dra. Maria Cristina Schneider, Universidad de Georgetown, Estados Unidos

Dr. Feliciano Milian Suazo, Universidad Autónoma de Querétaro, México

Dr. Javier F. Enríquez Quiroz, INIFAP, México

Dra. Martha Hortencia Martín Rivera, Universidad de Sonora URN, México

Dr. Fernando Arturo Ibarra Flores, Universidad de Sonora URN, México

Dr. James A. Pfister, USDA, Estados Unidos

Dr. Eduardo Daniel Bolaños Aguilar, INIFAP, México

Dr. Sergio Iván Román-Ponce, INIFAP, México

Dr. Jesús Fernández Martín, INIA, España

Dr. Maurcio A. Elzo, Universidad de Florida

Dr. Sergio D. Rodríguez Camarillo, INIFAP, México

Dra Nydia Edith Reyes Rodríguez, Universidad Autónoma del Estado de Hidalgo, México

Dra. Maria Salud Rubio Lozano, Facultad de Medicina Veterinaria y Zootecnia, UNAM, México

Dra. Elizabeth Loza-Rubio, INIFAP, México

Dr. Juan Carlos Saiz Calahorra, Instituto Nacional de Investigaciones Agrícolas, España

Dr. José Armando Partida de la Peña, INIFAP, México

Dr. José Luis Romano Muñoz, INIFAP, México

Dr. Jorge Alberto López García, INIFAP, México

Dr. Alejandro Plascencia Jorquera, Universidad Autónoma de Baja California, México

Dr. Juan Ku Vera, Universidad Autónoma de Yucatán, México Dr. Ricardo Basurto Gutiérrez, INIFAP, México

Dr. Luis Corona Gochi, Facultad de Medicina Veterinaria y Zootecnia, UNAM, México

Dr. Juan Manuel Pinos Rodríguez, Facultad de Medicina Veterinaria y Zootecnia, Universidad Veracruzana, México Dr. Carlos López Coello, Facultad de Medicina Veterinaria y Zootecnia, UNAM, México

Dr. Arturo Francisco Castellanos Ruelas, Facultad de Química. UADY

Dra. Guillermina Ávila Ramírez, UNAM, México Dr. Emmanuel Camuus, CIRAD, Francia. Dr. Héctor Jiménez Severiano, INIFAP., México Dr. Juan Hebert Hernández Medrano, UNAM, México Dr. Adrian Guzmán Sánchez, Universidad Autónoma Metropolitana-Xochimilco, México Dr. Eugenio Villagómez Amezcua Manjarrez, INIFAP, CENID Salud Animal e Inocuidad, México Dr. José Juan Hernández Ledezma, Consultor privado Dr. Fernando Cervantes Escoto, Universidad Autónoma Chapingo, México

Dr. Adolfo Guadalupe Álvarez Macías, Universidad Autónoma Metropolitana Xochimilco, México

Dr. Alfredo Cesín Vargas, UNAM, México Dra. Marisela Leal Hernández, INIFAP, México Dr. Efrén Ramírez Bribiesca, Colegio de Postgraduados, México

TIPOGRAFÍA Y FORMATO: Oscar L. Rodríguez Rivera

Indizada en el “Journal Citation Report” Science Edition del ISI . Inscrita en el Sistema de Clasificación de Revistas Científicas y Tecnológicas de CONACyT; en EBSCO Host y la Red de Revistas Científicas de América Latina y el Caribe, España y Portugal (RedALyC) (www.redalyc.org); en la Red Iberoamericana de Revistas Científicas de Veterinaria de Libre Acceso (www.veterinaria.org/revistas/ revivec); en los Índices SCOPUS y EMBASE de Elsevier (www.elsevier. com).

REVISTA MEXICANA DE CIENCIAS PECUARIAS

La Revista Mexicana de Ciencias Pecuarias es un órgano de difusión científica y técnica de acceso abierto, revisada por pares y arbitrada. Su objetivo es dar a conocer los resultados de las investigaciones realizadas por cualquier institución científica, relacionadas particularmente con las distintas disciplinas de la Medicina Veterinaria y la Zootecnia. Además de trabajos de las disciplinas indicadas en su Comité Editorial, se aceptan también para su evaluación y posible publicación, trabajos de otras disciplinas, siempre y cuando estén relacionados con la investigación pecuaria.

Se publican en la revista tres categorías de trabajos: Artículos Científicos, Notas de Investigación y Revisiones Bibliográficas (consultar las Notas al autor); la responsabilidad de cada trabajo recae exclusivamente en los autores, los cuales, por la naturaleza misma de los experimentos pueden verse obligados a referirse en algunos casos a los nombres comerciales de ciertos productos, ello sin embargo, no implica preferencia por los productos citados o ignorancia respecto a los omitidos, ni tampoco significa en modo alguno respaldo publicitario hacia los productos mencionados.

Todas las contribucionesserán cuidadosamente evaluadas por árbitros, considerando su calidad y relevancia académica. Queda entendido que el someter un manuscrito implica que la investigación descrita es única e inédita. La publicación de Rev. Mex. Cienc. Pecu. es

trimestral en formato bilingüe Español e Inglés. El costo total porpublicares de $ 7,280.00 másIVA pormanuscrito ya editado.

Se publica en formato digital en acceso abierto, por lo que se autoriza la reproducción total o parcial del contenido de los artículos si se cita la fuente.

El envío de los trabajos de debe realizar directamente en el sitio oficial de la revista. Correspondencia adicional deberá dirigirse al Editor Adjunto a la siguiente dirección: Calle 36 No. 215 x 67 y 69 Colonia Montes de Amé, C.P. 97115 Mérida, Yucatán, México. Tel/Fax +52 (999) 941-5030. Correo electrónico (C-ele): rodriguez_oscar@prodigy.net.mx.

La correspondencia relativa a suscripciones, asuntos de intercambio o distribución de números impresos anteriores, deberá dirigirse al Editor en Jefe de la Revista Mexicana de Ciencias Pecuarias, Campo Experimental Mocochá, Km. 25 Antigua Carretera Mérida–Motul, Mocochá, Yuc. C.P. 97454; garcia.arturo@inifap.gob.mx o arias.alfonso@inifap.gob.mx. Inscrita en la base de datos de EBSCO Host y la Red de Revistas Científicas de América Latina y el Caribe, España y Portugal (RedALyC) (www.redalyc.org), en la Red Iberoamericana de Revistas Científicas de Veterinaria de Libre Acceso (www.veterinaria.org/revistas/ revivec), indizada en el “Journal Citation Report” Science Edition del ISI (http://thomsonreuters. com/) y en los Índices SCOPUS y EMBASE de Elsevier (www.elsevier.com)

VISITE NUESTRA PÁGINA EN INTERNET

Artículos completos desde 1963 a la fecha y Notas al autor en: http://cienciaspecuarias.inifap.gob.mx

Revista Mexicana de Ciencias Pecuarias is an open access peer-reviewed and refereed scientific and technical journal, which publishes results of research carried out in any scientific or academic institution, especially related to different areas of veterinary medicine and animal production. Papers on disciplines different from those shown in Editorial Committee can be accepted, if related to livestock research.

The journal publishes three types of papers: Research Articles, Technical Notes and Review Articles (please consult Instructions for authors). Authors are responsible for the content of each manuscript, which, owing to the nature of the experiments described, may contain references, in some cases, to commercial names of certain products, which however, does not denote preference for those products in particular or of a lack of knowledge of any other which are not mentioned, nor does it signify in any way an advertisement or an endorsement of the referred products.

All contributions will be carefully refereed for academic relevance and quality. Submission of an article is understood to imply that the research described is unique and unpublished. Rev. Mex. Cien. Pecu. is published quarterly in original lenguage Spanish or English. Total fee charges are US $ 425.00 per article in both printed languages.

Part of, or whole articles published in this Journal may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying or otherwise, provided the source is properly acknowledged.

Manuscripts should be submitted directly in the official web site. Additional information may be mailed to Associate Editor, Revista Mexicana de Ciencias Pecuarias, Calle 36 No.215 x67 y69 Colonia Montes de Amé, C.P. 97115 Mérida, Yucatán, México. Tel/Fax +52 (999) 941-5030. E-mail: rodriguez_oscar@prodigy.net.mx. For subscriptions, exchange or distribution of previous printed issues, please contact: Editor-in-Chief of Revista Mexicana de Ciencias Pecuarias, Campo Experimental Mocochá, Km. 25 Antigua Carretera Mérida–Motul, Mocochá, Yuc. C.P. 97454; garcia.arturo@inifap.gob.mx or arias.alfonso@inifap.gob.mx.

Registered in the EBSCO Host database. The Latin American and the Caribbean Spain and Portugal Scientific Journals Network (RedALyC) (www.redalyc.org). The Iberoamerican Network of free access Veterinary Scientific Journals (www.veterinaria.org/ revistas/ revivec). Thomson Reuter´s “Journal Citation Report” Science Edition (http://thomsonreuters.com/). Elsevier´s SCOPUS and EMBASE (www.elsevier.com) and the Essential Electronic Agricultural Library (www.teeal.org) .

VISIT OUR SITE IN THE INTERNET

Full articles from year 1963 to date and Instructions for authors can be accessed via the site http://cienciaspecuarias.inifap.gob.mx

REVISTA MEXICANA DE CIENCIAS PECUARIAS

REV. MEX. CIENC. PECU. VOL. 14 No. 1 ENERO-MARZO-2023

CONTENIDO Contents

ARTÍCULOS

Articles Pág.

Identification of candidate genes and SNPs related to cattle temperament using a GWAS analysis coupled with an interacting network analysis

Identificación de genes candidatos y SNP relacionados con el temperamento del ganado utilizando un análisis GWAS junto con un análisis de redes interactuantes

Francisco Alejandro Paredes-Sánchez, Ana María Sifuentes-Rincón, Edgar Eduardo Lara-Ramírez, Eduardo Casas, Felipe Alonso Rodríguez-Almeida, Elsa Verónica Herrera-Mayorga, Ronald D. Randel ...........................................................................................1

Efecto de la consanguinidad y selección sobre los componentes de un índice productivo en ratones bajo apareamiento estrecho

Effect of consanguinity and selection on the components of a productive index, in mice under close mating

Dulce Janet Hernández López, Raúl Ulloa Arvizu, Carlos Gustavo Vázquez Peláez, Graciela Guadalupe Tapia Pérez ..................................................................................... …............23

Variabilidad genética en biomasa aérea y sus componentes en alfalfa bajo riego y sequía

Genetic variability in aerial biomass and its components in alfalfa under irrigation and drought Milton Javier Luna-Guerrero, Cándido López-Castañeda .............. ……………………………………….39

Estimación de masa de forraje en una pradera mixta por aprendizaje automatizado, datos del manejo de la pradera y meteorológicos satelitales

Estimation of forage mass in a mixed pasture by machine learning, pasture management and satellite meteorological data

Aurelio Guevara-Escobar, Mónica Cervantes-Jiménez, Vicente Lemus-Ramírez, Adolfo Kunio Yabuta-Osorio, José Guadalupe García-Muñiz ……………………………………………………………….61

Thymol and carvacrol determination in a swine feed organic matrix using Headspace SPME-GC-MS

Determinación de timol y carvacrol en una matriz orgánica de alimento para cerdo utilizando Headspace SPME-GC-MS

Fernando Jonathan Lona-Ramírez, Nancy Lizeth Hernández-López, Guillermo González-Alatorre, Teresa del Carmen Flores-Flores, Rosalba Patiño-Herrera, José Francisco Louvier-Hernández …….78

Cambios en el recuento de cuatro grupos bacterianos durante la maduración del Queso de Prensa (Costeño) de Cuajinicuilapa, México

Changes in the count of four bacterial groups during the ripening of Prensa (Costeño) Cheese from Cuajinicuilapa, Mexico

José Alberto Mendoza-Cuevas, Armando Santos-Moreno, Beatriz Teresa Rosas-Barbosa, Ma. Carmen Ybarra-Moncada, Emmanuel Flores-Girón, Diana Guerra-Ramírez ………………………..…94

Detección molecular de un fragmento del virus de lengua azul en borregos de diferentes regiones de México

Molecular detection of a fragment of bluetongue virus in sheep from different regions of Mexico

Edith Rojas Anaya, Fernando Cerón-Téllez, Luis Adrián Yáñez-Garza, José Luis Gutiérrez-Hernández, Rosa Elena Sarmiento-Salas, Elizabeth Loza-Rubio ...........................……110

Insulin-like growth factor 1 (IGF-1) concentrations in synovial fluid of sound and osteoarthritic horses, and its correlation with proinflammatory cytokines IL-6 and TNF

Concentraciones del factor de crecimiento similar a la insulina 1 (IGF-1) en el líquido sinovial de caballos sanos y osteoartríticos, y su correlación con las citoquinas proinflamatorias IL-6 y TNFα Fernando García-Lacy F., Sara Teresa Méndez-Cruz, Horacio Reyes-Vivas, Victor Manuel DávilaBorja, Jose Alejandro Barrera-Morales, Gabriel Gutiérrez-Ospina, Margarita Gómez-Chavarín, Francisco José Trigo-Tavera .......................................................…………………………………………122

Uso de células estromales mesenquimales derivadas de la gelatina de Wharton para el tratamiento de uveítis recurrente equina: estudio piloto

Use of Wharton's jelly-derived mesenchymal stromal cells for the treatment of equine recurrent uveitis: a pilot study María Masri-Daba, Montserrat Erandi Camacho-Flores, Ninnet Gómez-Romero, Francisco Javier Basurto Alcántara .................................................................................………………………………137

Escala de la producción y eficiencia técnica de la ganadería bovina para carne en Puebla, México

Scale of production and technical efficiency of beef cattle farming in Puebla, Mexico José Luis Jaramillo Villanueva, Lissette Abigail Rojas Juárez, Samuel Vargas López ……………….…154

Regresión cuantil para predicción de caracteres complejos en bovinos Suizo Europeo usando marcadores SNP y pedigrí

Quantile regression for prediction of complex traits in Braunvieh cattle using SNP markers and pedigree

Jonathan Emanuel Valerio-Hernández, Paulino Pérez-Rodríguez, Agustín Ruíz-Flores ……………….172

Análisis de crecimiento estacional de una pradera de trébol blanco (TrifoliumrepensL) Seasonal growth analysis of a white clover meadow (TrifoliumrepensL.)

Edgar Hernández Moreno, Joel Ventura Ríos, Claudia Yanet Wilson García, María de los Ángeles Maldonado Peralta, Juan de Dios Guerrero Rodríguez, Graciela Munguía Ameca, Adelaido Rafael Rojas García....................................................................................…………….190

REVISIONES DE LITERATURA Reviews

Aspects related to the importance of using predictive models in sheep production. Review

Aspectos relacionados con la importancia del uso de modelos predictivos en la producción ovina. Revisión Antonio Leandro Chaves Gurgel, Gelson dos Santos Difante, Luís Carlos Vinhas Ítavo, João Virgínio Emerenciano Neto, Camila Celeste Brandão Ferreira Ítavo, Patrick Bezerra Fernandes, Carolina Marques Costa, Francisca Fernanda da Silva Roberto, Alfonso Juventino Chay-Canul ......................................204

NOTAS DE INVESTIGACIÓN Techcnical notes

Preferencia de ocho plantas por Odocoileusvirginianusen cautiverio Preference for eight plants among captive white-tailed deer Odocoileusvirginianusin Veracruz, Mexico Hannia Yaret Cueyactle-Cano, Ricardo Serna-Lagunes, Norma Mora-Collado, Pedro Zetina-Córdoba, Gerardo Benjamín Torres-Cantú .....................................………………...............228

Rendimiento y valor nutricional de brásicas forrajeras en comparación con forrajes tradicionales

Yield and nutritional value of forage brassicas compared to traditional forages

David Guadalupe Reta Sánchez, Juan Isidro Sánchez Duarte, Esmeralda Ochoa Martínez, Ana Isabel González Cifuentes, Arturo Reyes González, Karla Rodríguez Hernández …….............237

Genetic characterization of bovine viral diarrhea virus 1b isolated from mucosal disease

Caracterización del virus de la diarrea viral bovino subtipo 1b aislado de un caso de la enfermedad de las mucosas

Roberto Navarro-López, Juan Diego Perez-de la Rosa, Marisol Karina Rocha-Martínez, Marcela Villarreal-Silva, Mario Solís-Hernández, Eric Rojas-Torres, Ninnet Gómez-Romero ...........248

Actualización: marzo, 2020

NOTAS AL AUTOR

La Revista Mexicana de Ciencias Pecuarias se edita completa en dos idiomas (español e inglés) y publica tres categorías de trabajos: Artículos científicos, Notas de investigación y Revisiones bibliográficas.

Los autores interesados en publicar en esta revista deberán ajustarse a los lineamientos que más adelante se indican, los cuales en términos generales, están de acuerdo con los elaborados por el Comité Internacional de Editores de Revistas Médicas (CIERM) Bol Oficina Sanit Panam 1989;107:422-437.

1. Sólo se aceptarán trabajos inéditos. No se admitirán si están basados en pruebas de rutina, ni datos experimentales sin estudio estadístico cuando éste sea indispensable. Tampoco se aceptarán trabajos que previamente hayan sido publicados condensados o inextensoen Memorias o Simposio de Reuniones o Congresos (a excepción de Resúmenes).

2. Todos los trabajos estarán sujetos a revisión de un Comité Científico Editorial, conformado por Pares de la Disciplina en cuestión, quienes desconocerán el nombre e Institución de los autores proponentes. El Editor notificará al autor la fecha de recepción de su trabajo.

3. El manuscrito deberá someterse a través del portal de la Revista en la dirección electrónica: http://cienciaspecuarias.inifap.gob.mx, consultando el “Instructivo para envío de artículos en la página de la Revista Mexicana de Ciencias Pecuarias”. Para su elaboración se utilizará el procesador de Microsoft Word, con letra Times New Roman a 12 puntos, a doble espacio. Asimismo se deberán llenar los formatos de postulación, carta de originalidad y no duplicidad y disponibles en el propio sitio oficial de la revista.

4. Por ser una revista con arbitraje, y para facilitar el trabajo de los revisores, todos los renglones de cada página deben estar numerados; asimismo cada página debe estar numerada, inclusive cuadros, ilustraciones y gráficas.

5. Los artículos tendrán una extensión máxima de 20 cuartillas a doble espacio, sin incluir páginas de Título, y cuadros o figuras (los cuales no deberán exceder de ocho y ser incluidos en el texto). Las Notas de investigación tendrán una extensión máxima de 15 cuartillas y 6 cuadros o figuras. Las Revisiones bibliográficas una extensión máxima de 30 cuartillas y 5 cuadros.

6. Los manuscritos de las tres categorías de trabajos que se publican en la Rev. Mex. Cienc. Pecu. deberán contener los componentes que a continuación se indican, empezando cada uno de ellos en página aparte.

Página del título Resumen en español Resumen en inglés Texto

Agradecimientos y conflicto de interés Literatura citada

7. Página del Título Solamente debe contener el título del trabajo, que debe ser conciso pero informativo; así como el título traducido al idioma inglés. En el manuscrito no es necesaria información como nombres de autores, departamentos, instituciones, direcciones de correspondencia, etc., ya que estos datos tendrán que ser registrados durante el proceso de captura de la solicitud en la plataforma del OJS (http://ciencias pecuarias.inifap.gob.mx).

8. Resumenenespañol.Enlasegundapáginasedebe incluir un resumen que no pase de 250 palabras. En él se indicarán los propósitos del estudio o investigación; los procedimientos básicos y la metodología empleada; los resultados más importantes encontrados, y de ser posible, su significación estadística y las conclusiones principales. A continuación del resumen, en punto y aparte, agregue debidamente rotuladas, de 3 a 8 palabras o frases cortas clave que ayuden a los indizadores a clasificar el trabajo, las cuales se publicarán junto con el resumen.

9. Resumen en inglés. Anotar el título del trabajo en inglés y a continuación redactar el “abstract” con las mismas instrucciones que se señalaron para el resumen en español. Al final en punto y aparte, se deberán escribir las correspondientes palabras clave (“key words”).

10. Texto.Lastrescategorías detrabajosquesepublican en la Rev. Mex. Cienc. Pecu. consisten en lo siguiente:

a)Artículoscientíficos.Deben serinformesdetrabajos originales derivados de resultados parciales o finales de investigaciones. El texto del Artículo científico se divide en secciones que llevan estos encabezamientos:

Introducción Materiales y Métodos Resultados Discusión

Conclusiones e implicaciones Literatura citada

En los artículos largos puede ser necesario agregar subtítulos dentro de estas divisiones a fin de hacer más claro el contenido, sobre todo en las secciones de Resultados y de Discusión, las cuales también pueden presentarse como una sola sección.

b) Notas de investigación. Consisten en modificaciones a técnicas, informes de casos clínicos de interés especial, preliminares de trabajos o investigaciones limitadas, descripción de nuevas variedades de pastos; así como resultados de investigación que a juicio de los editores deban así ser publicados. El texto contendrá la misma información del método experimental señalado en el inciso a), pero su redacción será corrida del principio al final del trabajo; esto no quiere decir que sólo se supriman los subtítulos, sino que se redacte en forma continua y coherente.

c) Revisiones bibliográficas. Consisten en el tratamiento y exposición de un tema o tópico de relevante actualidad e importancia; su finalidad es la de resumir, analizar y discutir, así como poner a disposición del lector información ya publicada sobre un tema específico. El texto se divide en: Introducción, y las secciones que correspondan al desarrollo del tema en cuestión.

11. Agradecimientos y conflicto de interés. Siempre que corresponda, se deben especificar las colaboraciones que necesitan ser reconocidas, tales como a) la ayuda técnica recibida; b) el agradecimiento por el apoyo financiero y material, especificando la índole del mismo; c) las relaciones financieras que pudieran suscitar un conflicto de intereses. Las personas que colaboraron pueden ser citadas por su nombre, añadiendo su función o tipo de colaboración; por ejemplo: “asesor científico”, “revisión crítica de la propuesta para el estudio”, “recolección de datos”, etc. Siempre que corresponda, los autores deberán mencionar si existe algún conflicto de interés.

12. Literatura citada. Numere las referencias consecutivamente en el orden en que se mencionan por primera vez en el texto. Las referencias en el texto, en los cuadros y en las ilustraciones se deben identificar mediante números arábigos entre paréntesis, sin señalar el año de la referencia. Evite hasta donde sea posible, el tener que mencionar en el texto el nombre de los autores de las referencias. Procure abstenerse de utilizar los resúmenes como referencias; las “observaciones inéditas” y las “comunicaciones personales” no deben usarse como

referencias, aunque pueden insertarse en el texto (entre paréntesis).

ReglasbásicasparalaLiteraturacitada Nombre de los autores, con mayúsculas sólo las iniciales, empezando por el apellido paterno, luego iniciales del materno y nombre(s). En caso de apellidos compuestos se debe poner un guión entre ambos, ejemplo: Elías-Calles E. Entre las iniciales de un autor no se debe poner ningún signo de puntuación, ni separación; después decada autor sólo se debe poner una coma, incluso después del penúltimo; después del último autor se debe poner un punto.

El título del trabajo se debe escribir completo (en su idioma original) luego el título abreviado de la revista donde se publicó, sin ningún signo de puntuación; inmediatamente después el año de la publicación, luego el número del volumen, seguido del número (entre paréntesis) dela revista yfinalmente el número de páginas (esto en caso de artículo ordinario de revista).

Puede incluir en la lista de referencias, los artículos aceptados aunque todavía no se publiquen; indique la revista y agregue “en prensa” (entre corchetes).

En el caso de libros de un solo autor (o más de uno, pero todos responsables del contenido total del libro), después del o los nombres, se debe indicar el título del libro, el número de la edición, el país, la casa editorial y el año.

Cuando se trate del capítulo de un libro de varios autores, se debe poner el nombre del autor del capítulo, luego el título del capítulo, después el nombre de los editores y el título del libro, seguido del país, la casa editorial, año y las páginas que abarca el capítulo.

En el caso de tesis, se debe indicar el nombre del autor, el título del trabajo, luego entre corchetes el grado (licenciatura, maestría, doctorado), luego el nombre de la ciudad, estado y en su caso país, seguidamente el nombre de la Universidad (no el de la escuela), y finalmente el año.

Emplee el estilo de los ejemplos que aparecen a continuación, los cuales están parcialmente basados en el formato que la Biblioteca Nacional de Medicina de los Estados Unidos usa en el IndexMedicus

Revistas

Artículoordinario,convolumenynúmero. (Incluya el nombre de todos los autores cuando sean seis o menos; si son siete o más, anote sólo el nombre de los seis primeros y agregue “etal.”).

I) Basurto GR, Garza FJD. Efecto de la inclusión de grasa o proteína de escape ruminal en el comportamiento de toretes Brahman en engorda. Téc Pecu Méx 1998;36(1):35-48.

Sólonúmerosinindicarvolumen.

II) Stephano HA, Gay GM, Ramírez TC. Encephalomielitis, reproductive failure and corneal opacity (blue eye) in pigs associated with a paramyxovirus infection. Vet Rec 1988;(122):6-10.

III) Chupin D, Schuh H. Survey of present status ofthe use of artificial insemination in developing countries. World Anim Rev 1993;(74-75):26-35.

Noseindicaelautor.

IV) Cancer in South Africa [editorial]. S Afr Med J 1994;84:15.

Suplementoderevista.

V) Hall JB, Staigmiller RB, Short RE, Bellows RA, Bartlett SE. Body composition at puberty in beef heifers as influenced by nutrition and breed [abstract]. J Anim Sci 1998;71(Suppl 1):205.

Organización,comoautor.

VI) The Cardiac Society of Australia and New Zealand. Clinical exercise stress testing. Safety and performance guidelines. Med J Aust 1996;(164):282-284.

Enprocesodepublicación.

VII)Scifres CJ, Kothmann MM. Differential grazing use of herbicide treated area by cattle. J Range Manage [in press] 2000.

Libros y otras monografías Autortotal.

VIII) Steel RGD, Torrie JH. Principles and procedures of statistics: A biometrical approach. 2nd ed. New York, USA: McGraw-Hill Book Co.; 1980.

Autordecapítulo.

IX) Roberts SJ. Equine abortion. In: Faulkner LLC editor. Abortion diseases of cattle. 1rst ed. Springfield, Illinois, USA: Thomas Books; 1968:158-179.

Memoriasdereuniones.

X) Loeza LR, Angeles MAA, Cisneros GF. Alimentación de cerdos. En: Zúñiga GJL, Cruz BJA editores. Tercera reunión anual del centro de investigaciones forestales y agropecuarias del estado de Veracruz. Veracruz. 1990:51-56.

XI) Olea PR, Cuarón IJA, Ruiz LFJ, Villagómez AE. Concentración de insulina plasmática en cerdas alimentadas con melaza en la dieta durante la inducción de estro lactacional [resumen]. Reunión nacional de investigación pecuaria. Querétaro, Qro. 1998:13.

XII) Cunningham EP. Genetic diversity in domestic animals: strategies for conservation and development. In: Miller RH etal.editors. Proc XX Beltsville Symposium: Biotechnology’s role in genetic improvement of farm animals. USDA. 1996:13.

Tesis.

XIII) Alvarez MJA. Inmunidad humoral en la anaplasmosis ybabesiosisbovinasen becerrosmantenidosen una zona endémica [tesis maestría]. México, DF: Universidad Nacional Autónoma de México; 1989.

XIV) Cairns RB. Infrared spectroscopic studies of solid oxigen [doctoral thesis]. Berkeley, California, USA: University of California; 1965.

Organizacióncomoautor.

XV) NRC. National Research Council. The nutrient requirements of beef cattle. 6th ed. Washington, DC, USA: National Academy Press; 1984.

XVI) SAGAR. Secretaría de Agricultura, Ganadería y Desarrollo Rural.Curso deactualización técnicapara la aprobación de médicos veterinarios zootecnistas responsables de establecimientos destinados al sacrificio de animales. México. 1996.

XVII) AOAC. Oficial methods of analysis. 15th ed. Arlington, VA, USA: Association of Official Analytical Chemists. 1990.

XVIII)SAS. SAS/STAT User’s Guide (Release 6.03). Cary NC, USA: SAS Inst. Inc. 1988

XIX) SAS. SAS User´s Guide: Statistics (version 5 ed.). Cary NC, USA: SAS Inst. Inc. 1985.

Publicacioneselectrónicas

XX) Jun Y, Ellis M. Effect of group size and feeder type on growth performance and feeding patterns in growing pigs. J Anim Sci 2001;79:803-813. http://jas.fass.org/cgi/reprint/79/4/803.pdf. Accessed Jul 30, 2003.

XXI) Villalobos GC, González VE, Ortega SJA. Técnicas para estimar la degradación de proteína y materia orgánica en el rumen y su importancia en rumiantes en pastoreo. Téc Pecu Méx 2000;38(2): 119-134. http://www.tecnicapecuaria.org/trabajos/20021217 5725.pdf. Consultado 30 Ago, 2003.

XXII)Sanh MV, Wiktorsson H, Ly LV. Effect of feeding level on milk production, body weight change, feed conversion and postpartum oestrus of crossbred lactating cows in tropical conditions. Livest Prod Sci 2002;27(2-3):331-338. http://www.sciencedirect. com/science/journal/03016226. Accessed Sep 12, 2003.

13. Cuadros, Gráficas e Ilustraciones. Es preferible que sean pocos, concisos, contando con los datos necesarios para que sean autosuficientes, que se entiendan por sí mismossin necesidaddeleer el texto. Para las notas al pie se deberán utilizar los símbolos convencionales.

14 Versión final. Es el documento en el cual los autores ya integraron las correcciones y modificaciones indicadas por el Comité Revisor. Los trabajos deberán ser elaborados con Microsoft Word. Las fotografías e imágenes deberán estar en formato jpg (o compatible) con al menos 300 dpi de resolución. Tanto las fotografías, imágenes, gráficas, cuadros o tablas deberán incluirse en el mismo archivo del texto. Los cuadros no deberán contener ninguna línea vertical, y las horizontales solamente las que delimitan los encabezados de columna, y la línea al final del cuadro.

15. Unavez recibidala versión final, ésta se mandará para su traducción al idioma inglés o español, según corresponda. Si los autores lo consideran conveniente podrán enviar su manuscrito final en ambos idiomas.

16. Tesis. Se publicarán como Artículo o Nota de Investigación, siempre y cuando se ajusten a las normas de esta revista.

17. Los trabajos no aceptados para su publicación se regresarán al autor, con un anexo en el que se explicarán los motivos por los que se rechaza o las modificaciones que deberán hacerse para ser reevaluados.

18. Abreviaturas de uso frecuente: cal caloría (s) cm centímetro (s) °C grado centígrado (s) DL50 dosis letal 50% g gramo (s)

ha hectárea (s) h hora (s) i.m. intramuscular (mente) i.v. intravenosa (mente) J joule (s) kg kilogramo (s) km kilómetro (s) L litro (s) log logaritmo decimal Mcal megacaloría (s) MJ megajoule (s) m metro (s) msnmmetros sobre el nivel del mar µg microgramo (s) µl microlitro (s) µm micrómetro (s)(micra(s)) mg miligramo (s) ml mililitro (s) mm milímetro (s) min minuto (s) ng nanogramo (s)Pprobabilidad (estadística) p página PC proteína cruda PCR reacción en cadena de la polimerasa pp páginas ppm partes por millón % por ciento (con número) rpm revoluciones por minuto seg segundo (s) t tonelada (s) TND total de nutrientes digestibles UA unidad animal UI unidades internacionales vs versus xg gravedades

Cualquier otra abreviatura se pondrá entre paréntesis inmediatamente después de la(s) palabra(s) completa(s).

19. Los nombres científicos y otras locuciones latinas se deben escribir en cursivas.

INSTRUCTIONS FOR AUTHORS

Revista Mexicana de Ciencias Pecuarias is a scientific journal published in a bilingual format (Spanish and English) which carries three types of papers: Research Articles, Technical Notes, and Reviews. Authors interested in publishing in this journal, should follow the belowmentioned directives which are based on those set down by the International Committee of Medical Journal Editors (ICMJE) Bol Oficina Sanit Panam 1989;107:422-437.

1. Only original unpublished works will be accepted. Manuscripts based on routine tests, will not be accepted. All experimental data must be subjected to statistical analysis. Papers previously published condensed or inextensoin a Congress or any other type of Meeting will not be accepted (except for Abstracts).

2. All contributions will be peer reviewed by a scientific editorial committee, composed of experts who ignore the name of the authors. The Editor will notify the author the date of manuscript receipt.

3. Papers will be submitted in the Web site http://cienciaspecuarias.inifap.gob.mx, according the “Guide for submit articles in the Web site of the Revista Mexicana de Ciencias Pecuarias”. Manuscripts should be prepared, typed in a 12 points font at double space (including the abstract and tables), At the time of submission a signed agreement co-author letter should enclosed as complementary file; coauthors at different institutions can mail this form independently. The corresponding author should be indicated together with his address (a post office box will not be accepted), telephone and Email.

4. Tofacilitatepeer review allpagesshouldbenumbered consecutively, including tables, illustrations and graphics, and the lines of each page should be numbered as well.

5. Research articles will not exceed 20 double spaced pages, without including Title page and Tables and Figures (8 maximum and be included in the text). Technical notes will have a maximum extension of 15 pages and 6 Tables and Figures. Reviews should not exceed 30 pages and 5 Tables and Figures.

6. Manuscripts of all three type of articles published in Revista Mexicana de Ciencias Pecuarias should contain the following sections, and each one should begin on a separate page.

Title page Abstract Text

Acknowledgments and conflict of interest Literature cited

7. Title page. It should only contain the title of the work, which shouldbe concise but informative; aswell as the title translated into English language. In the manuscript is not necessary information as names of authors, departments, institutions and correspondence addresses, etc.; as these data will have to be registered during the capture of the application process on the OJS platform (http://cienciaspecuarias.inifap.gob.mx).

8. Abstract. On the second page a summary of no more than 250 words should be included. This abstract should start with a clear statement of the objectives and must include basic procedures and methodology. The more significant results and their statistical value andthe main conclusionsshouldbe elaborated briefly. At the end of the abstract, and on a separate line, a list of up to 10 key words or short phrases that best describe the nature of the research should be stated.

9. Text. The three categories of articles which are published in Revista Mexicana de Ciencias Pecuarias are the following:

a)ResearchArticles. They should originate in primary works and may show partial or final results of research. The text of the article must include the following parts:

Introduction

Materials and Methods Results Discussion

Conclusions and implications

Literature cited

In lengthy articles, it may be necessary to add other sections to make the content clearer. Results and Discussion can be shown as a single section if considered appropriate.

b) Technical Notes. They should be brief and be evidence for technical changes, reports of clinical cases of special interest, complete description of a limited investigation, or research results which

should be published as a note in the opinion of the editors. The text will contain the same information presented in the sections of the research article but without section titles.

c) Reviews. The purpose of these papers is to summarize, analyze and discuss an outstanding topic. The text of these articles should include the following sections: Introduction, and as many sections as needed that relate to the description of the topic in question.

10. Acknowledgements. Whenever appropriate, collaborations that need recognition should be specified: a) Acknowledgement of technical support; b) Financial and material support, specifying its nature; andc) Financial relationships that could bethe source of a conflict of interest.

People which collaborated in the article may be named, adding their function or contribution; for example: “scientific advisor”, “critical review”, “data collection”, etc.

11. Literature cited. All references should be quoted in their original language. They should be numbered consecutively in the order in which they are first mentioned in the text. Text, tables and figure references should be identified by means of Arabic numbers. Avoid, whenever possible, mentioning in the text the name of the authors. Abstain from using abstracts as references. Also, “unpublished observations” and “personal communications” should not be used as references, although they can be inserted in the text (inside brackets).

Keyrulesforreferences

a. The names of the authors should be quoted beginning with the last name spelt with initial capitals, followedbytheinitialsof thefirst and middlename(s). In the presence of compound last names, add a dash between both, i.e. Elias-Calles E. Do not use any punctuation sign, nor separation between the initials of an author; separate each author with a comma, even after the last but one.

b. The title of the paper should be written in full, followedbythe abbreviated titleof the journal without any punctuation sign; then the year of the publication, after that the number of the volume, followed by the number (in brackets) of the journal and finally the number of pages (this in the event of ordinary article).

c. Accepted articles, even if still not published, can be included in the list of references, as long as the journal is specified and followed by “in press” (in brackets).

d. In the case of a single author’s book (or more than one, but all responsible for the book’s contents), the title of the book should be indicated after the

names(s), the number of the edition, the country, the printing house and the year.

e. When a reference is made of a chapter of book written by several authors; the name of the author(s) of the chapter should be quoted, followed by the title of the chapter, the editors and the title of the book, the country, the printing house, the year, and the initial and final pages.

f. In the case of a thesis, references should be made of the author’s name, the title of the research, thedegreeobtained, followed bythenameof the City, State, and Country, the University (not the school), and finally the year.

Examples

The style of the following examples, which are partly based on the format the National Library of Medicine of the United States employs in its Index Medicus, should be taken as a model.

Journals

Standard journal article (List the first six authors followed by etal.)

I) Basurto GR, Garza FJD. Efecto de la inclusión de grasa o proteína de escape ruminal en el comportamiento de toretes Brahman en engorda. Téc Pecu Méx 1998;36(1):35-48.

Issuewithnovolume

II) Stephano HA, Gay GM, Ramírez TC. Encephalomielitis, reproductive failure and corneal opacity (blue eye) in pigs associated with a paramyxovirus infection. Vet Rec 1988;(122):6-10.

III) Chupin D, Schuh H. Survey of present status of the use of artificial insemination in developing countries. World Anim Rev 1993;(74-75):26-35.

Noauthorgiven

IV) Cancer in South Africa [editorial]. S Afr Med J 1994;84:15.

Journalsupplement

V) Hall JB, Staigmiller RB, Short RE, Bellows RA, Bartlett SE. Body composition at puberty in beef heifers as influenced by nutrition and breed [abstract]. J Anim Sci 1998;71(Suppl 1):205.

Organization,asauthor

VI) The Cardiac Society of Australia and New Zealand. Clinical exercise stress testing. Safety and performance guidelines. Med J Aust 1996;(164):282284.

Inpress

VII) Scifres CJ, Kothmann MM. Differential grazing use of herbicide-treatedarea bycattle. JRangeManage[in press] 2000.

Books and other monographs

Author(s)

VIII) Steel RGD, Torrie JH. Principles and procedures of statistics: A biometrical approach. 2nd ed. New York, USA: McGraw-Hill Book Co.; 1980.

Chapterinabook

IX) Roberts SJ. Equine abortion. In: Faulkner LLC editor. Abortion diseases of cattle. 1rst ed. Springfield, Illinois, USA: Thomas Books; 1968:158-179.

Conferencepaper

X) Loeza LR, Angeles MAA, Cisneros GF. Alimentación de cerdos. En: Zúñiga GJL, Cruz BJA editores. Tercera reunión anual del centro de investigaciones forestales y agropecuarias del estado de Veracruz. Veracruz. 1990:51-56.

XI) Olea PR, Cuarón IJA, Ruiz LFJ, Villagómez AE. Concentración de insulina plasmática en cerdas alimentadas con melaza en la dieta durante la inducción de estro lactacional [resumen]. Reunión nacional de investigación pecuaria. Querétaro, Qro. 1998:13.

XII) Cunningham EP. Genetic diversity in domestic animals: strategies for conservation and development. In: Miller RH etal.editors. Proc XX Beltsville Symposium: Biotechnology’s role in genetic improvement of farm animals. USDA. 1996:13.

Thesis

XIII) Alvarez MJA. Inmunidad humoral en la anaplasmosis ybabesiosisbovinasen becerrosmantenidosen una zona endémica [tesis maestría]. México, DF: Universidad Nacional Autónoma de México; 1989.

XIV) Cairns RB. Infrared spectroscopic studies of solid oxigen [doctoral thesis]. Berkeley, California, USA: University of California; 1965.

Organizationasauthor

XV) NRC. National Research Council. The nutrient requirements of beef cattle. 6th ed. Washington, DC, USA: National Academy Press; 1984.

XVI) SAGAR. Secretaría de Agricultura, Ganadería y Desarrollo Rural.Curso deactualización técnicapara la aprobación de médicos veterinarios zootecnistas responsables de establecimientos destinados al sacrificio de animales. México. 1996.

XVII)AOAC. Official methods of analysis. 15th ed. Arlington, VA, USA: Association of Official Analytical Chemists. 1990.

XVIII) SAS. SAS/STAT User’s Guide (Release 6.03). Cary NC, USA: SAS Inst. Inc. 1988

XIX) SAS. SAS User´s Guide: Statistics (version 5 ed.). Cary NC, USA: SAS Inst. Inc. 1985.

Electronicpublications

XX) Jun Y, Ellis M. Effect of group size and feeder type on growth performance and feeding patterns in growing pigs. J Anim Sci 2001;79:803-813. http://jas.fass.org/cgi/reprint/79/4/803.pdf. Accesed Jul 30, 2003.

XXI) Villalobos GC, González VE, Ortega SJA. Técnicas para estimar la degradación de proteína y materia orgánica en el rumen y su importancia en rumiantes en pastoreo. Téc Pecu Méx 2000;38(2): 119-134. http://www.tecnicapecuaria.org/trabajos/20021217 5725.pdf. Consultado 30 Jul, 2003.

XXII)Sanh MV, Wiktorsson H, Ly LV. Effect of feeding level on milk production, body weight change, feed conversion and postpartum oestrus of crossbred lactating cows in tropical conditions. Livest Prod Sci 2002;27(2-3):331-338. http://www.sciencedirect.com/science/journal/030 16226. Accesed Sep 12, 2003.

12. Tables, Graphics and Illustrations. It is preferable that they should be few, brief and having the necessary data so they could be understood without reading the text. Explanatory material should be placed in footnotes, using conventional symbols.

13. Final version. This is the document in which the authors have already integrated the corrections and modifications indicatedbythe Review Committee. The works will have to be elaborated with Microsoft Word. Photographs and images must be in jpg (or compatible) format with at least 300 dpi resolution. Photographs, images, graphs, charts or tables must be included in the same text file. The boxes should not contain any vertical lines, and the horizontal ones only those that delimit the column headings, and the line at the end of the box

14. Once accepted, the final version will be translated into Spanish or English, although authors should feel free to send the final version in both languages. No charges will be made for style or translation services.

15. Thesis will be published as a Research Article or as a Technical Note, according to these guidelines.

16. Manuscripts not accepted for publication will be returned to the author together with a note explaining the cause for rejection, or suggesting changes which should be made for re-assessment.

17. List of abbreviations:

cal calorie (s) cm centimeter (s) °C degree Celsius DL50 lethal dose 50% g gram (s) ha hectare (s) h hour (s) i.m. intramuscular (..ly) i.v. intravenous (..ly) J joule (s) kg kilogram (s) km kilometer (s) L liter (s) log decimal logarithm Mcal mega calorie (s)

MJ mega joule (s) m meter (s) µl micro liter (s) µm micro meter (s) mg milligram (s) ml milliliter (s) mm millimeter (s) min minute (s) ng nanogram (s) P probability (statistic) p page

CP crude protein PCR polymerase chain reaction pp pages ppm parts per million % percent (with number) rpm revolutions per minute sec second (s) t metric ton (s) TDN total digestible nutrients AU animal unit IU international units vs versus xg gravidity

The full term for which an abbreviation stands should precede its first use in the text.

18. Scientific names and other Latin terms should be written in italics.

https://doi.org/10.22319/rmcp.v14i1.6077 Article

Identification of candidate genes and SNPs related to cattle temperament using a GWAS analysis coupled with an interacting network analysis

FranciscoAlejandro Paredes-Sánchez a

Ana María Sifuentes-Rincón b*

Edgar Eduardo Lara-Ramírez c

Eduardo Casas d

FelipeAlonso Rodríguez-Almeida e

Elsa Verónica Herrera-Mayorga f

Ronald D. Randel g

a UniversidadAutónoma de Tamaulipas, IA-UAMM. Mante, México.

b Instituto Politécnico Nacional. Centro de Biotecnología Genómica. Laboratorio de BiotecnologíaAnimal, Blvd. Del Maestro esq. Elías Piña. Col. Narciso Mendoza s/n. Cd. Reynosa, Tam. México.

c Instituto Mexicano del Seguro Social, Unidad de Investigación Biomédica de Zacatecas, Zacatecas, México.

d United States Department ofAgriculture. NationalAnimal Disease Center, Iowa, USA.

e UniversidadAutónoma de Chihuahua. Facultad de Zootecnia y Ecología, Chihuahua, México.

f Universidad Autónoma de Tamaulipas. IBI-UAMM, Mante, México

g TexasA&M University. AgriLife Research. Texas, USA.

*Corresponding author: asifuentes@ipn.mx

Abstract:

The objective of this study was to identify in Angus and Brangus breed animals with extreme temperament, measured as exit velocity, genomic regions and candidate genes associated with bovine temperament. The population was genotyped with the Genomic Profiler HD 150K chip and after the genome-wide association analysis, the SNPs rs133956611 (P=2.65 E-06) and rs81144933 (P=9.58 E-06) were associated with temperament. The mapping analysis of the regions close to the SNP rs81144933 identified the SNCA (alpha-synuclein) and MMRN1 (multimerin-1) genes at 222.8 and 435.9 Kb downstream respectively, while for the rs133956611 loci the gene GPRIN3 (GPRIN family-member-3) was identified at 245.7 Kb upstream, all three genes are located on the BTA6 chromosome. The analysis of SNCA protein-protein interactions allowed the identification of the genes APP (β-amyloid precursor protein), PARK7 (parkinsonismassociated-deglycase), UCHL1 (ubiquitin-C-terminal-hydrolase-L1), PARK2 (parkin-RBRE3-ubiquitin-protein-ligase), and genes of the SLC family as candidates to be associated with bovine temperament. All these candidate genes and their interacting were resequenced, which allowed the discovery of new SNPs in the SNCA and APP genes. Of these, the SNPs located in introns 5, 8 and 11 of the APP gene affect splicing site motifs. These results indicate that SNCA and its interacting genes are candidates to be related to bovine temperament.

Key words: Beef cattle, Behaviour, BTA6, Candidate genes, Temperament

Received: 18/10/2021

Accepted: 16/08/2022

Introduction

Temperament is an economically relevant trait that impacts animal welfare and traits related to productivity. Bovine temperament is considered to be the most important trait of an animal's personality and comprises a wide range of behaviors, from docility to fear and nervousness or a lack of response, attempts to escape, and aggressive behavior, in which various parameters such as general locomotor activity and reactivity to stress are observable. Temperament is affected by age, experience, sex, handling, maternal effects, environmental factors, genetics, species and breed(1,2) . To date, several genomic approaches attempted to

identify genomic regions and genes in which underlying single nucleotide polymorphisms (SNPs) are associated with temperament, a complex phenotypical trait.

Quantitative trait locus (QTL) mapping uncovered the first evidence of genomic regions associated with behavioral traits in dairy breeds(3,4) . The detection of QTLs in the genome led to the proposal of candidate genes under the genomic region encompassed by the QTL, which could potentially be responsible for the differences in trait expression. The identification of candidate genes based on their function and possible involvement in bovine temperament has been a strategy for the search for SNPs. Garza-Brenner et al(5) selected a group of 19 genes that participate in the dopamine and serotonin pathway, and through a protein-protein interaction (PPI) analysis, they identified four new interacting candidate genes (POMC, NPY, SLC18A2, and FOSFBJ), of which POMC, SLC18A2 and DRD3, HTR2A (selected based on their function) revealed SNPs associated with Exit Velocity (EV) and Pen Score (PS), which are measurements of bovine temperament in a population of Charolais cattle. The same group found that the variations in these genes (DRD3, HTR2A, and POMC) had an effect on bovine growth (birth weight) in a population of Charolais cattle, showing that the identified variations not only had an effect on bovine temperament but also on live weight traits(6) . Similarly, with the objective of evaluating the potential relationship of two of these SNPs in the DRD3 and HTR2A genes with bovine temperament and growth characteristics, and feed efficiency, a population of Angus, Brangus, and Charolais cattle with temperament assessments was analysed; the results indicated that there was no association with EV and PS, but the SNP in the HTR2A gene was associated with feed efficiency in Brangus cattle(7) .

Genome-wide association studies (GWAS), based on high-throughput single nucleotide polymorphism (SNP) genotyping technologies, are a relatively recent approach applied to genetic studies of cattle temperament and have allowed the identification of different groups of candidate genes. Lindholm-Perry et al(8) analysed a population of the Angus, Hereford, Simmental, Limousin, Charolais, Gelbvieh, and Red Angus breeds to identify genomic regions and genes associated with flight speed (FS); they determined chromosomal regions on BTA 9 and 17 associated and identified within them three genes GRIA2, GLRB, and QKI associated with nearby SNPs Valente et al(9) evaluated a Nellore population using EV to assess their temperament The NCKAP5, PARK2, DOCK1, ANTXR1, CPE, and GUCY1A2 genes were detected as potential candidates for the trait of interest. Finally, Dos Santos et al(10) used a Guzerat population in which reactivity was measured as an indicator of temperament The genes POU1F1, DRD3, VWA3A, ZBTB20, EPHA6, SNRPF, and NTN4 were proposed as candidate genes responsible for expression of the trait

In a related context, exome-specific resequencing of specific regions using next-generation sequencing (NGS) technologies has become a powerful technique that allows the identification of SNPs. This method can efficiently capture all variation in the regions of

interest. The potential effects can be assessed in an association study, which provides an effective tool to find SNPs affecting a determined trait(11) . However, due to differences in temperament phenotyping in previous studies, (i.e., each study uses different techniques to assess bovine temperament, pen score, exit velocity, reactivity, which evaluate different aspects of bovine temperament), it is not possible to link information for those genes identified as candidates, or to find a representative biological process, protein-protein interactions between these genes, or a biological path in which these genes converge to visualize how the set of genes explains bovine temperament. Thus, genomic information often remains isolated and needs to be integrated. Hence, the objective was to identify genomic regions and candidate genes associated with temperament in beef cattle through the integration of a GWAS strategy, protein-protein interaction analysis, and SNPs obtained by specific exome resequencing

Material and methods

Description of animals and biological sample sources

Data and hair samples were obtained from the biobank located at the Animal Biotechnology Laboratory CBG-IPN and were from a cattle population (n= 104) of young Angus (AN, n=63) and Brangus (BR, n=41) bulls, with an average age and bodyweight of 273 ± 38 d and 272 ± 38 kg, respectively, analysed during two centralized feed efficiency performance tests based on residual feed intake (RFI) in northern Mexico. Data recording and animal management have been previouslydescribed byGarza-Brenner et al(7) . Briefly, animals were fed in a feedlot for a period of 70 d with a pre-trial adaptation period of 20 d, weighed at the beginning and at the end of the test with intervals of 14 d in which the bovine temperament measurements were made.

From the population, a GWAS was performed using a selective genotyping approach following the strategy of the tails of the phenotypic distribution of bovine temperament measured by exit velocity (EV) because it facilitates the detection of phenotypic differences between alleles(12) Selective genotyping was achieved by selecting a group of the calmest (n=17; 10-AN and 7-BR) and most temperamental animals (n=17; 9-AN and 8-BR) based on EV values of study population. Temperament was assessed by EV measurements after a stimulus from hair sampling in a chute by measuring the rate of travel over 1.83 m (6 ft) with an infrared sensor (FarmTek Inc., North Wylie, TX, USA). The velocity was calculated as EV= distance (m)/time (s)(13,14) . It was defined the contrasting temperament groups based on

animals’ EV measurements. Animals with EV measurements ≤1.9 m/s were classified as calm, and those with EV scores ≥2.4 m/s were classified as temperamental(14,15) .

ToidentifyinformativeSNPs in candidate genes, 91animals wereused. Atotalof91animals were selected as a SNP discovery population: 18 (9 docile; 9 temperamental) of the Angus breed, 68 (44 docile; 24 temperamental) of the Brahman breed, and 5 (2 docile; 3 temperamental) of Charolais breed. From hair samples and ear notches, DNA extraction was performed using the GenElute™ extraction kit (Sigma, St. Louis, Missouri, United States).

GWAS analysis and gene discovery

Thirty-four (34) animals were genotyped using the GeneSeek Genomic Profiler HD 150K chip (Neogen, Lincoln, NE). Association analysis and identification of genomic regions associated with bovine temperament were performed with PLINK 1.9 software(16). Quality control of the genotypes was performed to identify animals with no assigned genotype or with a low genotyping rate (MIND >0.1). Allele frequency was also evaluated, and those SNPs with lower thresholds (MAF <0.01) were eliminated. Significance threshold was set at P <3×10−5.AManhattanplot was constructedusing qqman: an R package for visualization of GWAS results(17). Positions of significant SNPs were identified using the bovine Bos taurus genome (UMD 3.1.1) and Map Viewer software available at the National Center for Biotechnology Information (NCBI). Genes closest to the significant SNPs (within ~350 kb) were also identified with Map Viewer.

Pathway analysis and protein-protein interactions

For the identification of gene pathways, Gene Ontology (GO) term enrichment and proteinprotein interaction (PPI) network analysis were performed in the Ensembl genome browser(18) , Gene Ontology database(19) , and STRING database(20), respectively.

Candidate genes resequencing

With the objective of identifying SNPs in the coding regions and of the SNCA gene and its interacting genes, identified through the protein-protein interaction analysis (PPI), these genes were resequenced in the SNP discoverypopulation. As part of the sequencing strategy,

besides the exons, non-coding regions (140 bp before and after each gene-exon) were also analysed. Thus, a customized panel was designed using the Design Studio software (https://designstudio.illumina.com) (Illumina, San Diego, CA, United States) for the AmpliSeq DNA Gene Assay, in which the coding regions and the boundaries of the APP, PARK7, SLC6A2, SNCA, UCHL1, PARK2, SLC18A2, and POMC genes wereincluded, using the Bos taurus UMD 3.1.1 genome as a reference

DNA quantification was performed in all steps using the Qubit dsDNA HS Assay kit on the Qubit 3.0 fluorometer (Thermo Scientific, Massachusetts, United States). The libraries were prepared using the reference guide for custom panels AmpliSeq (Document # 1000000036408 v04) of Illumina, following the instructions for 2 pools and for 49–96 pairs of primers per pool. The quality and quantification of the libraries were carried out using the Bioanalyzer 2100 equipment (Agilent, California, United States) with theAgilent DNA 1000 kit. Sequencing (paired-end; read length 126 bp) was performed with the MiniSeq ™ Sequencing System.

Bioinformatics analysis of sequencing data

Sequence reads generated by the MiniSeq™ Sequencing System were aligned with the reference genome UMD 3.1.1 of Bos taurus using the Burrows-Wheeler aligner (BWAMEM) v0.(21) The reads were processed using Picard v1.135 (http://broadinstitute.github.io/picard) and cleaned bymarking and removing duplicate reads to generate BAM files Variations were identified using the genomic variant call format (GVCF) workflow with HaplotypeCaller(22) SNPs were generated in VCF files and filtered using the following criteria: variant confidence normalized by depth (QD) <2.0, mapping quality (MQ) <40.0, strand bias (FS) >60.0, HaplotypeScore >13.0, MQRankSum <−12.5, and ReadPosRank-Sum <−8.0(23)

Prediction of the effect of non-coding SNPs on splice sites

To study the effect of the 58 SNPs identified in the non-coding sequences from the exomespecific sequencing of the SNCA and APP genes, the online ESE finder3.0 web interface (http://krainer01.cshl.edu/cgi-bin/tools/ESE3) was used(24); the SNCA sequences NC_037333.1 and APP: NC_037328.1 were used as input, introducing them intron by intron (<5000 bp) without and with mutations, according to the location of the SNPs This process allowed to determine if the SNPs were part of a donor (5´) or acceptor (3´) splice site motif;

the programme assigns a score to the input sequence according to the loss of the consensus sequence, so that scores above a default threshold value (donor: 6.67; acceptor: 6.632) are predicted to act as a splice site, allowing the analysis of whether the SNPs affect splice sites motifs.

Results

GWAS analysis and candidate gene identification coupled to proteinprotein interaction analysis

Figure 1 depicts a Manhattan plot with the results from the GWAS analysis of SNPs evaluated for their association with temperament in Brangus and Angus cattle. Rs133956611 and rs81144933 were associated with a docile temperament (Table 1). The genes SNCA (alpha-synuclein; GenID 282857) and MMRN1 (multimerin 1; GenID 516574) are located approximately 222.8 and 435.9 Kb upstream respectively, from rs81144933; while the GPRIN3 (GPRIN family member 3; GenID 517995) gene was identified 245.7 Kb downstream of rs133956611.

Figure 1: Manhattan plot of the -log10 (p-values) for the genome-wide association with exit velocity

Table 1: SNPs associated with bovine temperament in Angus and Brangus cattle

CHR rs ID Position pb

Frecuency T D P-value

6 rs133956611 36,676,986 0.14 0.67 9.2 E-06

6 rs81144933 36,655,249 0.20 0.70 3.48 E-05

T= temperamental; D= docile.

The horizontal line corresponds to a significant threshold of P=3× 10−5 using the identified genes, we proceeded to perform a PPI analysis by querying the STRING(20) database For MMRN1, the PPI analysis showed interactions with genes such as F5 and VWF, involved in the coagulation process (Figure 2), in the Gene Ontology (GO) database, MMRN1 is annotated with the term GO:0007596, named blood coagulation. For GPRIN3, the search engineshowedinteractionsbetweenthephosphorylationprocessencodedbythe LOC790121 and OR6N1 genes with proteins that are mainly involved in cytoskeletal assembly and neurotransmission modulation (Figure 3). The GO database showed that this gene was annotated with the term GO:0031175, biological process named neuron projection development, progression of a neuron projection from its formation to the mature structure.

Figure 2: Protein-protein interactions reported for bovine MMRN1 in the STRING database

Figure 3: Protein-protein interactions reported for bovine GPRIN3 in the STRING database

Finally, SNCA protein, some GO terms identified (GO:0045920, GO:004241 and GO:0014059) were found to be involved in the regulation, synthesis, and secretion of dopamine. Interestingly, the SNCA gene was associated with the terms associated with behavior, including those related to “flight behaviour” and animal responses (through jumping, standing or walking) to internal and external stimuli (terms GO:0007610, GO:0007629 GO:0007628 GO:0007630, respectively).

The PPI analysis indicated that SNCA interacts with APP (β-amyloid precursor protein), PARK7 (parkinsonism associated deglycase), and UCHL1 (ubiquitin C-terminal hydrolase L1) proteins (P= 5.88e-06) involved in adult locomotory behaviour. In addition, the term GO:0008344 reveals strong interactions of SNCA with genes belonging to a neurotransmitter transporter family (SLC6A) in the network (Figure 4)

Figure 4: Protein-protein interactions reported for bovine SNCA in the STRING database

Red nodes annotated with the GO:0008344 term, adult locomotory behavior (p-value 5.88E-06). Green nodes annotated with the GO:0043005 term, neuron projection (p-value 0.000966). Blue nodes annotated with the 05012 KEGG pathway ID, Parkinson s disease (p-value 6.49E-11).

Based on their reported functional role, GPRIN3 and, particularly, SNCA genes could be considered as candidate genes associated with cattle temperament, the MMRN1 gene analysis indicate no obvious implications for this trait, however its identification could be important for further analysis.

Genetic variation in candidate genes

Accordingto thePPIanalysis results, it was inferred thatthe APP, PARK7, SLC6A2, UCHL1, PARK2, SLC18A2, and POMC genes were candidates associated with bovine temperament (Table 2). They were resequenced to discover genetic variation to potentially explain cattle temperament. Fifty-eight (58) SNPs were found in the non-coding regions of the SNCA and APP genes. Three SNPs were identified in introns 2 and 3 of the SNCA gene, and 55 SNPs were identified in introns 1, 5, 8, 11, 13, 14, and 17 of the APP gene (Table 3). Fifteen of the 58 SNPs were unique to the Angus breed, 1 in the SNCA gene and the remaining in the APP gene. The remaining SNPs (n= 43) were informative (polymorphic) in the Brahman and Charolais breeds, as opposed to the Angus breed in which they were uninformative (monomorphic).Theallelicfrequenciesanddistributionpattern oftheSNPsvaried according to the breed.

Table 2: Biological functions and processes associated with interacting SNCA genes

Gene Description

PARK7

No information in cattle. In humans, protects dopaminergic neurons against oxidative damage and degeneration; indirectly inhibits aggregation of αsynuclein(25); thus, mutations in this gene have been demonstrated to cause Parkinson’s disease(26) .

SLC6A2

No information in cattle. In humans, controls the action of norepinephrine that support arousal, mood, attention, and reactions to stress; thus, it has been associated with temperamental personality dimensions (novelty seeking, harm avoidance, reward dependence, and persistence)(27) .

UCHL1

PARK2

No information in cattle. In humans, it is abundantly expressed in neurons and interacts with APP, and SNPs in this gene have been implicated in the neurodegenerative disorders Parkinson’s disease and Alzheimer’s disease(28) .

In cattle, it has been associated with temperament (flight speed)(9) and in humans in the functions of dopaminergic neurons due to the mutations in this gene associated with Parkinson’s disease(29)

SLC18A2

POMC

In cattle, it has been associated with temperament (Pen Score) (GarzaBrenner et al(5). It participates in the transport of dopamine, preventing its accumulation and dopaminergic neuron death; therefore, it is a risk factor for Parkinson’s disease(30) .

In cattle, it has been associated with temperament (Pen Score)(5) . POMC is the precursor for corticotropic hormone (ACTH), which increases the expression of brain-derived neurotrophic factor (BDNF) responsible for neuron proliferation, differentiation, and survival; thus, it has been implicated in Parkinson’s disease (31) .

From the 58 SNP´s identified in the non-coding regions of SNCA and APP genes, three SNPs were part of a splice site motif according to established thresholds (donor: 6.67; acceptor: 6.632), as shown in Table 4; the identified SNPs were located in introns 5, 8, and 11. All the splice site motifs were of the acceptor type, that is, they were located at the 3´ end. The SNP g. 9770593 (C/T) did not add or abolish any splice site motif, but only increased the score value, whiletheSNPs g.9806689(G/T) and g.9845821(C/G)addedandabolishedthesplice site motifs, respectively.

Discussion

Genomic studies aimed at the exploration of cattle temperament are still scarce, mainly due to the biological complexity of the system, differences in the temperament measurement (objective/subjective), and differences between the studied cattle breeds. In this work, was usedtheGWAsasanexploratorytooltofindcandidategenesassociatedwithEV,contrasting bytemperament a pool of Angus and Brangus animals. GWAS allowed to identifya genomic region on BTA6 that harbours three candidate genes associated with EV [SNCA (Gen ID 282857), MMRN1 (Gen ID516574),and GPRIN3 (Gene ID:517995)]. Forthesegenes,Chen et al(32) reported an elevated expression of GPRIN3 in the human brain, and information from UniProtKB(33) indicates that the GPRIN3 protein may be involved in neurite outgrowth. However, the literature data (regarding function and interacting genes) strongly supports the bovine SNCA gene as a novel candidate associated with cattle temperament(9,34)

The SNCA gene is a highly conserved protein that is abundant in the brain of humans and other species like rats, mice, and monkeys(35); it is found in neurons, especiallyin presynaptic terminals(36) The molecular function of SNCA is quite ambiguous, and based on its structure, physical properties, and interacting partners, several hypotheses regarding its normal function in humans have been proposed For example, it is thought to be involved in the regulation of dopamine release and transport(34) Consequently, in humans it plays an important role in neurodegenerative disorders. According to Giasson et al(37), aggregates of SNCA protein in humans cause brain lesions that are characteristic of neurodegenerative synucleinopathies. The SNCA gene is associated, in the Kyoto Encyclopedia of Genes and Genomes (KEGG)(38), with biological pathways of neurodegenerative diseases such as Alzheimer's disease (ko05010) and Parkinson's disease (ko05012). Both diseases are important brain disorders in humans. Parkinson's disease is characterized by symptoms related to locomotion (involuntary tremor, muscle stiffness, and postural instability), as well as depression and psychosis, and it involves the progressive loss of dopaminergic neurons, with the main feature presenting as the appearance of inclusion bodies called Lewy bodies, the main component of which is SNCA(37) .

Although the pathological alterations linked to those human diseases cannot be extrapolated to this study model, this biological link provides some evidence to support the findings because the understanding of the relationship between genotype and phenotype in humans was derived from model animals with mutations in orthologous genes. Large animal species, such as dog, pig, sheep and cattle, have been some of the most important model animals, mainly because they are more similar to humans than mice (similar size, genetics, and physiology). Thus, discoveries in humans can serve as a reference to infer effects on bovine temperament(39)

Connecting gene networks to explain cattle temperament

Despite scarce attempts to identify genes and genomic regions underlying the genetic architecture of temperament, until now there have been no reports connecting the gene networks associated with this complex trait.

Protein-protein interaction analysis of the SNCA gene allowed to identify and analyse six additional genes, of which two gene members of the SLC family (SLC18A2 and SLC6A4) havealreadybeenidentifiedbyGarza-Brenner et al(5) asinteractinggenesinaprotein-protein network based on dopamine- and serotonin-related genes. These authors also found a SNP located in the SLC18A2 gene that causes a change in the amino acid sequence from alanine to threonine, with significant effects on temperament as measured by pen scores. In addition, the PPI analysis included genes in the PARK family (PARK2 and PARK7), which encode ubiquitin ligase proteins, including parkin RBR E3. The gene PARK2 was identified by Valente et al(9) as a candidate gene associated with temperament in Nellore cattle; the authors used EV as a test to evaluate bovine temperament. Multiple studies have used the GWAS strategy to identify genes that are linked to bovine temperament phenotypes(8-10), but in none of these cases has it been possible to establish interactions between the identified genes, and the information from each study seems to be isolated and independent, preventing the clarification of the genetic architecture of temperament from the information available to date. In addition, the set of candidate genes does not seem to be associated with a representative biological process that suggests participation in temperament. The identificationof SNCA in this workallowsto connect theresults of Valente et al(9) andGarzaBrenner et al(5), showing that the genes identified through different strategies (GWAS and protein-protein interaction network analysis) present an important connection. According to these results, it was explored the genetic variation in these genes in cattle with an emphasis on their coding sequences, and the results revealed a high conservation of the exonic sequences in all seven analysed genes. In humans, a low genetic variation has been reported between genes such as SNCA and UCHL1(40)

Interestingly, and according to previous reports, a high genetic variation was found in the in the non-coding regions of the bovine SNCA and APP genes.

The exact function of the amyloid beta (A4) precursor protein (APP) gene is unknown, but it has been associated with meat softness in pigs(41) , can participate in the formation of neurons, and is known for its participation in Alzheimer's disease(42). Because patients with Alzheimer's disease show the presence and accumulation of both SNCA and APP proteins, it has been proposed that they may be related in some way. Roberts et al(43) have shown that SNCA overexpression increases APP levels, and certain mutations in SNCA increase the

processing of APP, so the discovery of mutations in the coding regions of these genes could have a functional impact on them and therefore on bovine temperament.

It has been documented that approximately 21% of bovine genes are alternatively spliced(44) . In silico analyis identified three APP-SNP´s with the potential to have a functional effect in the pre-mRNA splicing process and, therefore, the expression of bovine temperament. As far as known, no different isoforms of the bovine APP gene have been reported, but splice site motifs in bovine genes have been reported to be highly conserved relative to humans(44). The human and bovine genes for APP are orthologs, having the same number of amino acids (770) and an identical amino acidic sequence. In humans, 8 different isoforms of the APP gene have been identified due to the alternative splicing in exons 7, 8, and 15, which terminates APP gene expression in neurons, resulting in the implication of a fundamental role in Alzheimer's disease(45). Here there was identified 3 SNPs that affect, add, and abolish splice site motifs in the APP gene, in introns 5, 8, and 11, so they could probably affect the final product and have an effect on the expression of bovine temperament.

In the present study, it was used the contrastingphenotype strategyto perform an exploratory GWAS analysis to identify candidate genes for temperament in cattle, and even with the small sample size limitation, the results showing a connection between SNCA and temperament are consistent with larger GWAS studies. Additionally, the coupling of these result with a PPI analysis allowed to establish connections between different genes that were previously identified within the association to the locomotor system. Fine mapping of the candidate genes predicted that the GWAS and PPI genes confirmed the existence of SNPs with the potential to affect bovine temperament. The present study provides valuable information that contributes to the -still scarce- efforts to describe the cattle temperament genetic architecture, and shows that an analytic strategy is appropriate for application in studies with a limited sample size, especially in countries where phenotyping for this complex trait is limited.

Conclusions and implications

A BTA6 genomic region (36,655,249-36,676,986 bp) neighboring the SNCA gene was associated with temperament trait in Angus and Brangus breeds. Six genes, linked to SNCA, were identified as being potentially associated with temperament. From those, the APP gene harboured three SNPs with a potential effect on the pre-mRNA splicing process and expression of bovine temperament.

Acknowledgements

This investigation was funded by research grants supported by CONACYT and IPN (project 294826, SIP 20171674) and partial financial support by CONARGEN, A.C. to support the feeding performance tests. The authors would also like to acknowledge the different herd owners and the technical staff from the Palomas complex UGRCH, who collected and provided the data and samples used in this study. The mention of trade name, proprietary products, or specified equipment does not constitute a guarantee or warranty by the USDA and does not imply approval for the exclusion of other products that may be suitable. The USDA is an Equal Opportunity Employer.

Literature cited:

1. Buchenauer D. Genetics of behaviour in cattle. In: Fries R Ruvinsky A, editors. The genetics of cattle. CABI Publishing, Wallingford;1999:365–390.

2. Mormède P. Molecular genetics of behaviour: research strategies and perspectives for animal production. Livest Prod Sci 2005;93(1):15–21.

3. Schmutz SM, Stookey JM, Winkelman-Sim DC, Waltz CS, Plante Y, Buchanan FC. A QTL study of cattle behavioural traits in embryo transfer families. J Hered 2001; 92(3):290–292.

4. Hiendleder S, Thomsen H, Reinsch N, Bennewitz J, Leyhe-Horn B, Looft C, et al Mapping of QTL for body conformation and behavior in cattle. J Hered 2003; 94(6):496–506.

5. Garza-Brenner E, Sifuentes-Rincón AM, Randel RD, Paredes-Sánchez FA, ParraBracamonte GM, Arellano-Vera W, et al. Association of SNPs in dopamine- and serotonin-pathway genes and their interacting genes with temperament traits in Charolais cows. J Appl Genetics 2016;58(3):363–371.

6. Garza-Brenner E, Sifuentes-Rincón AM, Rodríguez-Almeida FA, Randel RD, ParraBracamonte GM, Arellano-Vera W. Influence of temperament-related genes on live weight traits of Charolais cows. R Bras Zootec 2020:49:e20180121.

7. Garza-Brenner E, Sifuentes-Rincón AM, Rodríguez Almeida FA, Randel RD, ParraBracamonte GM, Arellano Vera W. Influence of genetic markers on the feeding behavior of yearling bulls. Rev Colomb Cienc Pecu 2019;32(1):14-20.

Pecu 2023;14(1):1-22

8. Lindholm-Perry AK, Kuehn LA, Freetly HC, Snelling WM. Genetic markers that influence feed efficiency phenotypes also affect cattle temperament as measured by flight speed. Anim Genet 2014;46(1):60-64.

9. Valente TS, Baldi F, Sant´Anna AC, Albuquerque LG, Da Costa MJRP. Genome-wide association study between single nucletide polymorphismos and flight speed in Nellore cattle. PLoS One 2016;11(6):1-18.

10. Dos Santos FC, Campolina PMG, De Souza PA, Ávila PMF, Ventura RV, Rosse IC, et al. Identification of candidate genes for reactivity in Guzerat (Bos indicus) cattle: a Genome-Wide Association Study. Plos One 2017;12(1):1-15.

11. Jiang L, Liu X, Yang J, Wang H, Jiang J, Liu L, et al. Targeted resequencing of GWAS loci reveals novel genetic variants for milk production traits. BMC Genomics 2014;15(1):1105.

12. Kurz JP, Yang Z, Weiss RB, Wilson DJ, Rood KA, Liu GE, Wang Z. A genome-wide association study for mastitis resistance in phenotypically well-characterized Holstein dairy cattle using a selective genotyping approach. Immunogenetics 2019;71(1):35-47.

13. Burrow HM. Measurements of temperament and their relationships with performance traits of beef cattle. Anim Breed 1997;65(7):477–495.

14. Curley KOJr, Paschal JC, Welsh THJr, Randel RD. Exit velocity as a measure of cattle temperament is repeatable and associated with serum concentration of cortisol in Brahman bulls. J Anim Sci 2006;84(11):3100-3103.

15. Burrow HM. Effect of intensive handling of zebu crossbred weaner calves on temperament. Proc Assoc Advmt Anim Breed Genet 1991;9:208-211.

16. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 2007;8(3):559-575.

17. Turner SD. qqman: an R package for visualizing GWAS results using Q-Q and manhattan plots. bioRxiv 2014;3(25):005165.

18. Zerbino DR, Achuthan P, Akanni W, Amode MR, Barrell D, Bhai J. Ensembl 2018. Nucleic Acids Res 2018;46(D1):754–761.

19. Mi H, Huang X, Muruganujan A, Tang H, Mills C, Kang D, Thomas PD. PANTHER version14: moregenomes, anewPANTHER GO-slim andimprovements inenrichment analysis tools. Nucleic Acids Res 2019;47(D1):D419-D426

Pecu 2023;14(1):1-22

20. Szklarczyk D,Morris JH,Cook H,KuhnM, Wyder S,SimonovicM, et al.TheSTRING database in 2017: quality-controlled protein-protein association networks, made broadly accessible. Nucleic Acids Res 2017;45(D1):362-368.

21. Li H, Durbin R. Fast and accurate short read alignment with Burrows–Wheeler transform. Bioinformatics 2009;25(14):1754–1760.

22. Van der Auwera GA, Carneiro MO, Hartl C, Poplin R, Del Angel G, LevyMoonshine A, et al. From FastQ data to high-confidence variant calls: the genome analysis toolkit best practices pipeline. Curr Protoc Bioinf 2013;43(1110):11101–111033.

23. Choi JW, Liao X, Stothard P, Chung WH, Jeon HJ, Miller SP, et al. Whole-genome analyses of Korean native and Holstein cattle breeds by massively parallel sequencing. PLoS One 2014;9(7):e101127.

24. Cartegni L, Wang J, Zhu Z, Zhang MQ, Krainer AR. ESEfinder: A web resource to identify exonic splicing enhancers. Nucleic Acid Res 2003;31(13):3568-3571.

25. Dolgacheva LP, Berezhnov AV, Fedotova EI, et al. Role of DJ-1 in the mechanism of pathogenesis of Parkinson's disease. J Bioenerg Biomembr 2019;51(3):175–188.

26. He L, Lin S, Pan H, Shen R, Wang M, Liu Z, et al Lack of association between DJ-1 gene promoter polymorphism and the risk of parkinson’s disease. Front Aging Neurosci 2019;11(24):1-11.

27. Narita SI, Kazuhiko N, Kenta N, Maki Y, Eiji O, Nobuyo I. Analysis of association between norepinephrine transporter gene polymorphisms and personality traits of NEOFFI in a Japanese population. Psychiatry Investig 2015;12(3):381-387.

28. Rydning SL, Backe PH, Sousa MML, Iqbal Z, Øye AM, Sheng Y, et al. Novel UCHL1 mutations reveal new insights into ubiquitin processing. Hum Mol Genet 2017;26(6):1031-1040.

29. BakhitYH, IbrahimMO, AminM,MirghaniYA,HassanMA. In silico analysisofSNPs in PARK2 and PINK1 genes that potentially cause autosomal recessive Parkinson disease. Adv Bioinformatics 2016;9313746:1-5.

30. Brighina L, Riva C, Bertola F, Saracchi E, Fermi S, Goldwurm S, et al Analysis of vesicular monoamine transporter 2 polymorphisms in Parkinson's disease. Neurobiol Aging 2013;34(6):1712.e9-13.

Pecu 2023;14(1):1-22

31. Shadrina M, Nikopensius T, Slominsky P, Illarioshkin S, Bagyeva G, Markova E, et al. Association study of sporadic Parkinson's disease genetic risk factors in patients from Russia by APEX technology. Neurosci Lett 2006;405(3):212-6.

32. Chen LT, Gilman AG, Kozasa T. A Candidate target for G protein action in brain. J Biol Chem 1999;274(38):26931–26938.

33. Chen C, Huang H, Wu CH. Protein bioinformatics databases and resources. Methods Mol Biol 2017;1558:3-39.

34. Siddiqui IJ, Pervaiz N, Abbasi AA. The Parkinson disease gene SNCA: Evolutionary and structural insights with pathological implication. Sci Rep 2016;6(1):24475.

35. Deng H, Yuan L. Genetic variants and animal models in SNCA and Parkinson disease. Ageing Res Rev 2014;15:161-76.

36. Campêlo CLDC, Silva RH. Genetic variants in SNCA and the risk of sporadic Parkinson’s disease and clinical outcomes: A review. Parkinsons Dis 2017;4318416:111.

37. Giasson BI, Duda JE, Murray IVJ, Chen Q, Souza JM, Hurtig HI, et al. Oxidative damage linked to neurodegeneration by selective alpha-synuclein nitration in synucleinopathy lesions. Science 2000;290(5493):985-989.

38. Kanehisa M, Furumichi M, Tanabe M, Sato Y, Morishima K. KEGG: New perspectives on genomes, pathways, diseases and drugs. Nucleic Acids Res 2017;45(D1):353-361.

39. Pinnapureddy AR, Stayner C, McEwan J, Baddeley O, Forman J, Eccles MR. Large animal models of rare genetic disorders: sheep as phenotypically relevant models of human genetic disease. Orphanet J Rare Dis 2015;10:107-115.

40. Dong N, Zhang X, Liu Q. Identification of therapeutic targets for Parkinson's disease via bioinformatics analysis. Mol Med Rep 2017;15(2):731-735.

41. Lobjois V, Liaubet L, San Cristobal M, Glénisson J, Fève K, Rallières J. A muscle transcriptome analysis identifies positional candidate genes for a complex trait in pig. Anim Genet 2008;39(2):147-162.

42. Masters CL, Simms G, Weinman NA, Multhaup G, McDonald BL, Beyreuther K. Amyloid plaque core protein in Alzheimer disease and Down syndrome. Proc Natl Acad Sci USA 1985;82(12):4245-4249.

Rev Mex Cienc Pecu 2023;14(1):1-22

43. Roberts HL, Schneider BL, Brown DR. α-Synuclein increases β-amyloid secretion by promoting β-/γ-secretase processing of APP. PLoS One 2017;12(2):e0171925.

44. Chacko E, Ranganathan S. Genome-wide analysis of alternative splicing in cow: Implications in bovine as a model for human diseases. BMC Genomics 2009;10(3):S11.

45. Sandbrink R, Masters CL, Beyreuther K. APP gene family. Alternative splicing generates functionally related isoforms. Ann N Y Acad Sci 1996;777:281-287.

Rev Mex Cienc Pecu 2023;14(1):1-22

Table 3: SNPs identified by specific exome sequencing in each population of APP, PARK7, SLC6A2, SNCA, UCHL1, PARK2, SLC18A2, and POMC genes

Gene Position (bp) Region

Alleles

Frecuency Angus Brahman Charolais

Ref Alt Ref Alt Ref Alt Ref Alt

36297353 Intron 3 G A 0.9924 0.0076 1.0 0.0

36297374 Intron 3 A G 0.8500 0.1500 1.0 0.0 36297422 ¥ Intron 2 T A 0.5000 0.5000 APP

SNCA

9674371 Intron 1 T C 0.9717 0.0283 1.0 0.0 9674423 Intron 1 A C 0.9403 0.0597 1.0 0.0 9674429 Intron 1 T A 0.9478 0.0522 1.0 0.0 9674430 ¥ Intron 1 T A 0.9722 0.0278 9674431* Intron 1 A T 0.9706 0.0294 0.9925 0.0075 1.0 0.0 9674437 Intron 1 T C 1.0000 0.0 0.9000 0.1000 9674448 Intron 1 T C 0.9921 0.0079 1.0 0.0 9674451 Intron 1 A G 0.9921 0.0079 0.9000 0.1000 9674455* Intron 1 G A 0.6071 0.3929 0.0093 0.9907 0.5000 0.5000 9770586* Intron 5 A G/T 0.6944 0.3056/0.0 0.8772 0.0395/0.0833 0.8000 0.2000/0.0 9770593 Intron 5 C T 0.3507 0.6493 1.0 0.0 9770633 Intron 5 G A 0.5373 0.4627 1.0 0.0 9803985* Intron 8 C T 0.9722 0.0278 0.0944 0.9056 1.0 0.0 9803991* Intron 8 A G 0.9722 0.0278 0.0909 0.9091 1.0 0.0

9806624* Intron 8 A G 0.9167 0.0833 0.0574 0.9426 0.8000 0.2000 9806672 Intron 8 T C 0.9769 0.0231 0.8000 0.2000 9806689 Intron 8 G T 0.9851 0.0149 1.0 0.0 9845631 Intron 11 C A 1.0000 0.0000 0.8000 0.2000 9845821 Intron 11 C G 0.7177 0.2823 1.0 0.0 9845862 ¥ Intron 11 G T 0.8750 0.1250 9845934 Intron 11 G A 0.9844 0.0156 1.0 0.0 9845944 Intron 11 G A 0.8750 0.1250 1.0 0.0 9845966 Intron 11 G A 0.9692 0.0308 1.0 0.0 9845980 Intron 11 A G 0.8056 0.1944 1.0 0.0 9863873* Intron 13 T C 0.9722 0.0278 0.0522 0.9478 1.0 0.0 9863960 Intron 13 T C 0.0818 0.9182 1.0 0.0

9863974¥ Intron 13 T C 0.6666 0.3333

9863983¥ Intron 13 T C 0.0 1.0

9863984¥ Intron 13 G T 0.0 1.0

9866489 Intron 13 G A 0.8433 0.1567 1.0 0.0

9866528 Intron 13 A G 0.9925 0.0075 1.0 0.0

9866542¥ Intron 13 A G 0.9722 0.02

9866545 Intron 13 C T 0.8433 0.1567 1.0 0.0

9866552¥ Intron 13 T C 0.9118 0.08

9866569 Intron 13 C T 0.8624 0.1376 1.0 0.0

9879860 Intron 13 T C 0.7881 0.2119 1.0 0.0

9880018 Intron 13 C A 0.5694 0.4306 1.0 0.0

9880025 Intron 13 G T 0.7787 0.2213 1.0 0.0

9889605¥ Intron 14 C G 0.6250 0.3750

9889627 Intron 14 G A 0.9462 0.0538 1.0 0.0 9889677* Intron 14 G T 0.0 1.0 0.9925 0.0075 0.0 1.0 9889687¥ Intron 14 T C 0.4167 0.5833

9891054¥ Intron 14 C T 0.5833 0.4167

9891056¥ Intron 14 T C 0.5000 0.5000 9891124¥ Intron 14 G T 0.4000 0.6000 9891130¥ Intron 14 A G 0.4063 0.5938 9891155¥ Intron 14 T G 0.4063 0.5938

9918483 Intron 17 C T 0.9841 0.0159 1.0 0.0 9918506* Intron 17 A G 0.1389 0.8611 0.9250 0.0750 0.2000 0.8000 9918508 Intron 17 C T 0.9655 0.0345 1.0 0.0 9918512 Intron 17 C T 0.9914 0.0086 1.0 0.0 9931517 Intron 17 C G 0.9924 0.0076 1.0 0.0 9931524 Intron 17 C G 0.9924 0.0076 1.0 0.0 9931525 Intron 17 T G 0.9924 0.0076 1.0 0.0 9931529 Intron 17 C T 1.0000 0.0000 0.9000 0.1000

* Variations present in the three populations. ¥ Specific variations in the Angus population.

Rev Mex Cienc Pecu 2023;14(1):1-22

Table 4: The ESE finder results for non-coding SNPs identified in the SNCA and APP genes

Gene Position Intron SNP

Site Sequence

Donor/ acceptor Score

APP

9770593 5

C WT CTCTCCCCTCGTCAGTGCTGTAGTTCAGGT acceptor 6.74720 T M CTTTCCCCTCGTCAGTGCTGTAGTTCAGGT acceptor 7.11480 ↑ 9806689 8 G WT T M CTTTGGATTTGCCAGGCACACTCACCTCCA acceptor 6.81380 ↑ 9845821 11 C WT CTCCTTCCACAACAGAAGGCGCTATTTTAA acceptor 6.71530 G M

The SNP nucleotide is highlighted in bold in the sequence. WT: wild type. M: sequence with non-coding SNP. ↑ indicates an increased score compared with the wild type sequence.

Effect of consanguinity and selection on the components of a productive index, in mice under close mating

Dulce Janet Hernández López a

Raúl Ulloa Arvizu a

Carlos Gustavo Vázquez Peláez a Graciela Guadalupe Tapia Pérez a*

a Universidad Nacional Autónoma de México, Departamento de Genética y Bioestadística de la Facultad de Medicina Veterinaria y Zootecnia. Ciudad de México, México.

*Corresponding author: tapiadoctora@gmail.com

Abstract:

In order to examine the influence of inbreeding depression on some productive characteristics of the laboratory mouse, 871 records were reanalyzed, which were from 20 generations in a line with narrow inbred crossing with selection for a productive index (WOFW) comparing with a line without selection, with inbred crossing (n= 135). Inbreeding coefficients (F) were calculated for each generation. In all the components of the index (reproductive life, fertile postpartum estruses and litter size), the two lines were compared, in the 15 available generations of the non-selected one, by the least squares method, grouping every five generations. The selected one was analyzed in the 20 generations for intergenerational differences with the same method. Inbreeding depression was estimated in all generations with a linear regression of consanguinity (expressed in 10 %) in all components. A significant difference (P<0.01) was observed between lines in the variables analyzed. The fertile postpartum estruses of the selected line remained constant, there was a decrease of 0.331 in the non-selected one (P<0.01). The productive index remained stable (increased 0.071) in the selected one, in the nonselectedoneit decreased (0.39)until disappearing (G15). Inbreedingdepression impacted the reproductive life of both, decreased 4.741 d in the selected one vs 7.718 d in the nonselected one (P<0.01). In the non-selected one, it affected mortality at weaning and

estrous cycle, the selection to the index counteracted that impact, probably due to the selection of genes that favor the gonadal development of mice.

Key words: Mice, Selection, Reproductive life, Number of estruses, Inbreeding depression.

Received: 05/10/2021

Accepted: 16/08/2022

Introduction

At present many genetically different mouse lines have been developed, which have particular research purposes. The inbred lines were the prototype of the genetically standardized lines, which allowed developing experiments eliminating the variability of genetic origin. Although genomics provides laboratories with the necessary tools to produce mice with the characteristics that research demands, when a characteristic has been fixed, a rigorous selection and directed mating process is needed to maintain the viability of the line, which usually leads to inbreeding depression(1,2) .

The genetic basis of this phenomenon is related to three hypotheses, namely, partial dominance (greater expression of deleterious recessive alleles), overdominance (superiority of heterozygotes over both types of homozygotes) and epistasis (greater probability of genetic combinations favorable to heterozygotes)(3) .

Breeders of purebred domestic animals use inbreeding to fix desirable genetic traits within a population or to try to eliminate deleterious traits, inbreeding depression can affect the economic income of breeders(4). Studies in mice, by offering a greater number of generations in less time, help to understand inbreeding depression, in populations where it is sought to select some characteristic.

In the mouse, a 7.2 % reduction in litter size was observed for every 10 % increase in consanguinity, under consecutive mating of complete siblings without selection(2) and, with the same increase in consanguinity in crosses between half-siblings without selection, the decrease was 6.22 % in litter size(5)

Depression was less severe in lines under directed selection than in lines without selection(6), this was observed when selecting for litter size in mice, finding that the reduction in the reproductive ability was significantly lower in inbred lines under selection, compared to that of inbred lines without selection; this is explained because

thanks to selection, there is an increase in genes related to better reproductive ability, which counteracts the inbreeding depression that causes the reduction of this ability(7). It has been seen that the behavior of an inbred line selected for litter size is similar to that of a non-inbred line, selected for the same trait. In one study, consanguinity allowed exceeding the limit of selection for large litter size, when in the selected line an inbred crossing was allowed(8) .

The number of weaned offspring per female per week (WOFW) is a productive index that is measured during and at the end of the reproductive life in each pair of mice. It is used in the founding colonies of some laboratory animal companies(9,10). Although it is recommended to select mice from families with higher WOFW to maintain laboratory lines, bynarrowinbredcrossingwith fixedcharacteristics(11),intheliteraturethereis little information on the effect of this selection in inbred mice on the variables included in it.

Therefore, the objective of this study was to evaluate the effect of inbreeding on the components of a productive index, in the animal model of laboratory mouse, during 20 generations of selection with narrow inbred crossing, as well as to evaluate whether the selection can be affected in its progress, by the effect of inbreeding depression, in the characteristics that constitute it.

Material and methods

Thepresent work is a retrospective,cross-sectional, comparativeandobservational study. Eight hundred seventy-one records of a bioterium were reanalyzed, which were taken for five years in mice with continuous selection and narrow inbred cross (brother with sister), where there were five lines selected for a productive index: (WOFW) in 20 generations (n= 871). The data, collected between 1989-1994, had been analyzed with the aim of obtaining realized heritability, for the productive index, a detailed description can be seen in Tapia-Pérez(12) .

For this study, a line of the same contemporary strain was added, which was from the same bioterium (n= 135), with narrow inbred crossing without selection until generation 15; after this generation the pairs ceased to be fertile.

The animals were housed in shoebox-type polycarbonate cages, which offer an area of 375 cm, with Cambridge-type stainless steel lid and Kraft-type rigid polyester filter; food was provided ad libitum, drinking water filtered by reverse osmosis acidified at a pH of 2.5. The air was filtered, and a temperature of 18 to 26 °C was maintained. The identification of the animals was individual, first, bymeans of notches in the ears, and the records by means of cards in each cage. These cards were then summarized in record folders called REA (Reproductive Efficiency Analysis), from which the WOFW index

was calculated every three generations, or when divergent lines were detected (this could occur in the fourth generation in each selected line), to obtain and select the subline with the highest average, which also had offspring of the third parturition, with at least two females and two males for breeding. It should be noted that those pairs who were sterile (no gestation was recorded), infertile (gestation was recorded, but not parturition) or who cannibalized their offspring had a WOFW value of zero, which was included to obtain the average since they are considered the result of inbreeding depression. As for the nonselected line, the management was similar; this line was the onlyone that remained active of the five that started at the same time of the selected ones, the other four were lost in the second generation. Reproductive management in both lines was under an intensive monogamous method, that is, a male with a female were placed in the same cage and remained together throughout their reproductive life (165 ± 3.6 d). Mating began when the animals reached sexual maturity (8-10 wk).

The pairs selected for reproduction were formed randomly, a female and a male full siblings, from the third parturition of their parents, both in the selection generation (3 or 4) and in the previous ones. An average of eight pairs per line was maintained in each generation.

Description of variables

The variables that were analyzed were:

RL: reproductive life, measured as the total days in reproduction.

FPPE: total number of fertile postpartum estruses in RL. A fertile postpartum estrus is considered when the female has a parturition in the first estrus, within 35 or less days from the previous one (since a gestation period of 21 d is assumed, with an implantation within 5 d,ifthis occursout whenthe motheris still lactatinga previous litter,it canoccur in 14 d maximum, thus: 21 + 14 =35)(9)

OBOR: total number of offspring born in RL. OWEA: total number of offspring weaned in RL.

OBPP: offspring born per parturition = �������� ������������������������

(PARTURITIONS: number of total parturitions in RL).

OWPP: offspring weaned per parturition = . �������� ������������������������

WOFW: (Productive Efficiency Index) is the number of weaned offspring per female per week = �������� ���� ×7 .

Percentage of mortality at weaning = �������� �������� �������� ��100.

Statistical analysis

For the statistical analysis, the selected line corresponds to the set of the five lines with selection for WOFW divided into: line selected for 15 generations (S15G) n= 733, the same line selected in the 20 generations (S20G) n= 871 and non-selected line, which remained until generation 15 (NS15G) n= 135, since in this one, there were five pairs, of which only one reached the third parturition and the offspring were not enough to make the crosses.

The information of five generations in each line was grouped, for all the variables, so that the data analyzed at each level correspond to those of five successive generations. They were grouped in this way to observe the result of the selection in the fifth generation, because, as explained, the selection was made at least every three generations or when divergent lines were detected, which could occur in up to four generations. Level 1 contains the sum of generations from 1 to 5; level 2, from 6 to 10 and level 3, from 11 to 15 of S15G and NS15G, while level 4 only corresponds to S20G, in generations 16 to 20.

Normality tests of the variables mentioned were performed by the Kolmogorov-Smirnov method.

The general linear model used to compare S15G and NS15G was (Model 1): �������� = ��+���� + ���� +(����)���� +��������

Where �������� is the sum of five successive generations of pairs, for each quantitative variable; ���� is the effect of the i-th selection group (i=1,2); ���� is the effect of the generation grouped every five generations (j=1,2,3); (����)���� is the effect of the interaction between the selection group and the grouped generation; �������� the random error (��~��0,���� 2).

TheanalysismodelforS20G(Model2)onlyincludedtheeffectofthegroupedgeneration gi, (i=1,2,3,4): ������ = ��+ ���� +������

Where ������ is the sum of 5 successive generations of pairs, for each quantitative variable, in the i-th generation; ������ the random error (��~��0,���� 2).

Both models were analyzed bythe least squares method. The coefficient of consanguinity was calculated for each animal, with the Pedigree Viewer© Computer Program, developed by Brian Kinghorn(13), which uses the method developed by Wright (1922)(14)

The mean inbreeding depression (�� ̂ 1) and its standard error (s.e.), of WOFW were estimated by the least squares method with the coefficients of consanguinity (Fi) of each generation, in units of 10 % for lines S15 and NS15 (i= 1,2,3, ...,15) and S20 (i= 1,2,3, ..., 20), with the following simple linear regression model. (Model 3) �� ̂ �� =��0 ̂ + ��1 ̂ ���� + ���� Where �� ̂ i is the average of each component of the index in the i-th generation; �� ̂ �� is the estimate of the intercept; ���� ̂ is the mean inbreeding depression, Fi is the coefficient of consanguinity (10 %); ���� is the random error (���� ~��µ,���� 2).

Since the increase in inbreeding was considered in units of 10 %, inbreeding depressions were related to this measure, which was chosen to be able to compare the results of this study with other mouse articles where it is calculated in that way.

The models were analyzed with the statistical package, IBM SPSS Version 22(15), the percentages ofmortalityat weaning,ofS15Gand NS15G,in eachofthethree generations grouped, were analyzed by the Chi-square test, with the online MedCalc® program(16) The P value ≤0.05 was considered as significant and P≤0.01 as highly significant.

Results

Linear models

Model 1

A highly significant effect of selection group (si) (P<0.01) was observed in all variables. Regarding the interaction (sg)ij, in the variables FPPE, OBOR, OWEA and WOFW, it was highly significant (P<0.01), in RL and OWPP, the effect of the interaction was significant (P<0.05), which did not happen in the OBPP variable (P>0.05) (Table 1).

Table 1: Least squares means (M) and standard errors (SE) of the interaction (sg)ij, in five grouped generations, for S15G and NS15G

LINE GG F(%) FPPE RL OBOR OWEA OBPP OWPP WOFW

S15GA 1 - 5 50 M 1.74a 165.2a 20.35a 17.77a 4.77a 4.16a 0.66a (n=733) SE 1.58 3.52 8.16 7.65 1.59 0.54 0.029 6 - 10 82.6 M 2.18a 129.5b 19.01a 16.09a 4.62a 3.92a 0.598a SE 1.62 3.44 10.33 10.29 2. 06 2.27 0.029

11 - 15 93.9 M 2.1a 146.2a 21.06a 19.31a 5.01a 4.60a 0.687a SE 1.5 3.32 9.78 9.61 1.92 2.00 0.028 NS15GB 1 - 5 50 M 2.65a 146.7a 17.16a 13.63a 3.76a 2.96a 0.59a (n=135) SE 1.63 6.64 7.52 7.164 1.26 1.49 0.055 6 - 10 82.6 M 1.32b 99.67b 7.76b 4.64b 3.04a 1.45b 0.12b SE 1.41 7.74 4.37 4.85 1.45 1.70 0.064 11 -15 93.9 M 0.3c 99.75b 7.85b 6.15b 3.87a 2.88a 0.20b SE 0.66 8.36 5.26 5.59 1.57 2.10 0.069 P(SG) <0.01 0.047 <0.01 0.01 0.482 0.048 <0.01

GG= grouped generations. F= consanguinity obtained in 5 generations. FPPE= number of fertile postpartum estruses. RL= reproductive life of the pair in days. OBOR= total number of offspring born in RL. OWEA= total number of offspring weaned in RL. OBPP= offspring born per parturition. OWPP= weaned offspring per parturition. WOFW= number of weaned offspring per female per week. A,B Different literals denote highly significant differences between the selection groups si (P<0.01). P(SG) is the significance calculated by the model, abc Different literals denote significant intergenerational differences (P<0.05), within line.

The means of the FPPEs of S15G increased 0.44 estruses and remained (P>0.05), while the means of NS15G decreased 2.4 estruses on average, from 1 to 5 until generations 11 to 15 (P<0.05) (Table 1 and Figure 1). Reproductive life (RL) decreased in S15G, almost 36 d (P<0.05) in generations 6 to 10, then recovered, although not at the level of the first five generations, while in NS15G it remained 47 d lower than in the first five generations (P<0.05). Both the number of offspring born and weaned, in the total reproductive life of the pairs, the lowest values were observed in generations 6 to 10 in both lines; however, only NS15G showed significant differences (P<0.05) with a decrease of 9.4 and 9 offspring, respectively. Both OBPP and OWPP were obtained as an average of all parturitions in the reproductive life of the female, grouped every five generations and were lower in NS15G, the lowest peak was observed between generations 6 to 10 in both, but it was only significant in NS15G in OWPP with a decrease of 0.7 weaned offspring (P<0.05) (Table 1).

Figure 1: Means and standard deviations of the number of fertile postpartum estruses (FPPE) with consanguinity grouped every five generations (1 to 5, 6 to 10 and 11 to 15)

S15G and NS15G are the lines with selection and without selection for WOFW in the 15 generations of the latter.

The index (WOFW) remained stable through all accumulated generations (P>0.05) in S15G, while NS15G has an abrupt drop from generations 1-5 to 6-10 (-0.47 offspring) (P<0.05), with a slight recovery in the following five grouped generations (0.08 offspring) (P<0.05); then it is lost due to high mortality (Table 1 and Figure 2).

Figure 2: Means and standard deviations of the number of weaned offspring per female per week (productive efficiency index) (WOFW) with consanguinity grouped every five generations (1 to 5, 6 to 10 and 11 to 15)

S15G and NS15G are the lines with selection and without selection for WOFW in the 15 generations of the latter.

Model 2

When S20G was analyzed, the lowest peak was observed in generations 6 to 10 (P<0.05), in almost all components, while FPPE and the WOFW index showed no significant intergenerational changes (P>0. 05) (Table 2).

Table 2: Least square means (M) and standard errors (SE) of five generations grouped, for S20G.

GG F(%) STA FPPE RL OBOR OWEA OBPP OWPP WOFW

1 to 5 50 M 1.59a 165.2a 18.64a 16.29a 4.77a 4.17a 0.616a SE 0.12 3.58 0.81 0.76 0.14 0.14 0.031

6 to 10 82.6 M 1.79a 130.0b 15.26b 12.92b 4.63a 3.92b 0.598a SE 0.12 3.49 0.78 0.74 0.14 0.15 0.030

11 to 15 93.9 M 1.78a 146.7c 17.64ab 16.17a 5.01a 4.59a 0.687a SE 0.11 3.38 0.76 0.71 0.13 0.14 0.029

16 to 20 97.4 M 1.80a 133.9bc 17.26ab 15.17ab 5.18b 4.49ab 0.675a SE 0.11 3.39 0.76 0.72 0.14 0.14 0.029

GG= grouped generations. F= consanguinity grouped into 5 generations. STA= statistic. FPPE= number of fertile postpartum estruses. RL= reproductive life of the pair in days. OBOR= total number of offspring born. OWEA= total number of offspring weaned in RL. OBPP= offspring born per parturition. OWPP= offspring weaned per parturition. WOFW= number of weaned offspring per female per week. abc Different literals denote significant intergenerational differences (P<0.05).

Model 3. Inbreeding depression

There was a highly significant effect (P<0.01) of inbreeding depression in S15G, in RL, OBPP and OWPP, while in OBOR, OWEA, FPPE and WOFW it was not significant (P>0.05); in NS15G, the characteristics FPPE, RL, OBOR, OWEA and WOFW showed a highly significant effect (P<0.01) of inbreeding depression, however, in OBPP and OWPP, it was not significant (P>0.05) (Table 3).

Table 3: Mean non-standardized regression coefficients (�� ̂ 1) and their standard error (SE), of the WOFW components and of the index itself, in NS15G, S15G and S20G, on the coefficient of consanguinity Line Statistic FPPE RL OBOR OWE A

OBPP OWP P

WOFW

S15G (n=733)

��1 ̂ SE

0.052 0.030

4.741 1.276

-0.454 0.256

-0.332 0.256

-0.466 0.084

-0.578 0.086

-0.001 0.007 NS15G (n=135)

��1 ̂ SE

��1 ̂ SE

0.331 0.066

0.047 0.026

7.718 2.507

-4.76 1.09

-1.705 0.341

-0.361 0.213

-1.325 0.326

-0.268 0.210

0.028 0.087

0.02 0.036

-0.010 0.109

0.026 0.036

-0.001 0.007

-0.062 0.015 S20G (n=871)

S15G= line with 15 generations of selection. NS15= line without selection for 15 generations. S20G= same line with selection for 20 generations. FPPE: number of fertile postpartum estruses. RL= reproductive life in days. OBOR= offspring born in the total reproductive life. OWEA= offspring weaned in the total reproductive life. OBPP= offspring born per parturition. OWPP= offspring weaned per parturition. WOFW= number of weaned offspring per female per week. Regression coefficients in bold were highly significant (P<0.01).

Reproductive life decreased in both lines for every 10 % of consanguinity, NS15G 7.718 days vs. 4.741 in S15G (P<0.01). In OBOR and OWEA, there was an effect of inbreeding depression only in NS15G (-1.705 and -1.325 offspring, respectively) (P<0.01). OBPP and OWPP were obtained as an average of the parturitions in the RLof each pair, in these variables, the inbreeding depression only affected S15G (-0.466 and -0.578 offspring, respectively) (P<0.01), while in NS15Gthere was an apparent stability(P>0.05), because the average number of parturitions in the accumulated generations was decreasing (4.2, 1.7 and 1.1), in the accumulated generations 1 to 5, 6 to 10 and 11 to 15 respectively, whilein theselectedone, theyremainedalmost unchanged(3.9,3.8and 3.6). TheWOFW index did not show inbreeding depression in S15G (P>0.05), the opposite occurred in NS15G (P<0.01). On the other hand, S20G showed highly significant inbreeding depression of -4.76 d in RL for every 10 % increase in consanguinity (P<0.01) (Table 3).

Mortality at weaning

The percentage of mortality per parturition at weaning, with respect to OBPP, had a similar behavior in both lines, with a maximum peak in generations 6 to 10, however, the non-selected line maintained a higher mortality than the selected one (52.30 %), having a difference with the selected one of 36.8 (P<0.01) (Figure 3). Survival per parturition (100 - % mortality per parturition) then declines, in the generations from 6 to 10, 31 % in the non-selected line (78.47 - 47.7 %) and 2.37 % in the selected line (87.22 - 84.85 %), remaining higher in the line with selection.

Figure 3: Mortality per parturition at weaning, in percentage with respect to OBPP, with consanguinity grouped into 5 generations

S15G and NS15G are the lines with selection and without selection for WOFW in the 15 generations of the latter.

Discussion

In the literature, there are few selection papers on long-term fertility and its effect on inbreeding depression in mice; in a study of selection of litter size at the first parturition that began in 1972, avoiding crosses between complete siblings, half-siblings or cousins, after 124 generations of selection, a consanguinity of 0.64 was found in one of its lines, which led it to greater inbreeding depression (-0.39), with a lower number of live offspring at the first parturition, for every 10 % of consanguinity(17). The results of the present work coincide with that, when the average of the offspring born alive per parturition (OBPP) was obtained in the line with selection, the effect of inbreeding depression was -0.466, a little higher than that, because in that work only the first parturition was measured. The number of offspring born per parturition, in the NS15G group, showed a non-significant effect of consanguinity (P>0.05) (Table 3), which

seemed contrary to what was expected; the explanation is that the number of parturitions in the reproductive life of the pairs was decreasing as the generations increased and when averaging the size of the litter at parturition and weaning (of the whole RL of each pair) with these, it seems that it would have remained constant (�� ̂ 1= 0.028). Another difference is that the consanguinity in that study is lower, although the number of selection generations is 124, during that time (1972 to 2007), there were several changes in the direction of selection in that work, but it was tried to avoid inbred crosses(17). No work with selection for litter size of the entire reproductive life with inbred cross was found in the literature, however, in a study with selection for an index combining litter size with birth weight, but with open crosses (not inbred) for 150 generations, litter sizes of 17.6 offspring born and 20.2 offspring on average(18) in the reproductive life of a pair were obtained, very similar to the average OBOR at the beginning of this study. The decrease in this average in the following generations in the present work is most likely due to inbreeding depression, which was -1.705 offspring for every 10 % of consanguinity (P<0.01) in the non-selected line. It should be noted that in the OBOR and OWEA characteristics, originally used to obtain the WOFW (in all the RL of each pair), there was a highly significant effect (P<0.01) of inbreeding depression in NS15G , which does not occur in S15G (Tables 1, 2 ,3). In that study, an increase in testosterone levels was observed in males, while in females there was an elevation of progesterone in one of its lines; these showed a higher number of oocytes per cycle, but a greater loss of embryos, and a decrease in reproductive life, compared to the line without selection. These results coincide with the present work since a decrease in reproductive life was also obtained in S1G and S20G (Tables 1, 2).

In the present study, it was revealed that the number of FPPE has a constant decrease over the generations studied in NS15G, with a decrease of 0.331 fertile postpartum estruses for every 10 % increase in consanguinity (P<0.01), compared to S15G which remains almost constant (P>0.05) (Table 3); this behavior is also observed in WOFW (the productive index), with a marked decrease in NS15G of generations 1 - 5 to 6 - 10, with a slight recovery in the last five; inbreeding depression was -0.062 (P<0.01) weaned offspring per female per week for every 10 % increase in consanguinity, vs. -0.001 (P>0.05)in S15G.Onestudyshowedthat adeletion ofKiss1rin theneurons oftheGnRH axis interrupts the signal of kisspeptin, the protein that induces the secretion of GnRH (gonadotropin-releasing hormone); this results in infertility due to hypogonadism, probably, consanguinityhad a negative effect on this mechanism through the interruption of this signal(19) in S15G mice; in the present work, the effects on males were not measured.Somethingverysimilarwas foundin pigs underselectionforprolificacy,when this was done with family indices, since an increase in inbreeding depression of three times more than expected without selection and a decrease in the response to selection were seen(20) .

RL was affected by consanguinity both in S15G, with a decrease of 4.741 d, and in NS15G, it decreased almost three more days (7.718 d) (P<0.01 in both) (Table 3), in a study where there was selection for longer life in mice, it was found that the increase was

related to mutations, which increased the levels of growth hormone in the GH/IGF-1 axis(21), which could be related to a decrease in the life time of the mice, as happened in the lines selected for an index involving litter size and birth weight, without consanguinity(18) in the present work, it decreased in both lines since the selection objective was different.

Postpartum mortalityis higher in NS15G, from the first five generations of narrow inbred crossing, compared to S15G, a result similar to that obtained by Sallah(5), both in the selected line and the non-selected line, with mating between half-siblings, where the highest mortality occurred in the intermediate generations. Charlesworth(22) explains this under two hypotheses: 1) dominance, where inbreeding increases the frequency of individuals that express the effects of deleterious mutations or 2) overdominance, where the homozygous would have a lower aptitude due to lack of alleles, with heterozygous advantage that they would maintain by balancing selection at intermediate frequencies in the heterozygous.

As a result of the above, the number of offspring at weaning per parturition in NS15G falls significantly in generations 6 to 10 (1.51 offspring), with a recovery in the following accumulated generations, while in the number of offspring at birth per parturition, it remains in all generations (P<0.05) (Table 1) (Figure 3). A recent study(23) revealed that, in litters with less than four offspring at birth, in non-selected mice, there is a higher mortality at weaning, and in the present study, the NS15G line showed less than four offspring at birth on average in the first five generations.

The result in these characteristics was presumable, since, on the one hand, a high inbreeding depression can be expected when performing a narrow inbred cross (brother, sister)and,on theotherhand,dueto thelowheritability(0.024to 0.063)of theproductive index(12), only limited reproduction progress can be expected; a similar result was obtained in litter size with a consanguinity of 0.61 in eight generations of selection with consanguinity in crosses between half-siblings(5)

These results lead to reflect on whether in a selection program in domestic animals, even avoiding crosses between siblings, in generations later, it could occur between relatives, and this induces inbreeding, with the counterproductive effects that were seen here.

In a Holstein cattle improvement program, it was found that with a 1 % increase in consanguinity, milk production in 305 d decreased by 36.3 kg on average, in cows aged 4 to 5 yr, and 2.42 kg of fat(24). Recently, the implementation of genomic selection was evaluated in the loss of genetic diversity in Holstein and Jersey cattle in North America, due to consanguinity; their results showed an increase in inbreeding from 1.19 to 2.06 % per generation, over a period of 10 yr in Holstein cattle, and warned about the need to implement measures to avoid inbreeding in this type of programs(25). In the Holstein population of Mexico, it was found that, with levels less than 5 % of consanguinity, no effect was detected in fat or milk protein, however, when inbreeding increased to more

than 5 %, a decrease in milk production of 260 kg per lactation was found, in addition to a loss in fat production of 11 kg and 10 kg in protein with respect to the average of groups with less than 5 %(26)

Conclusions and implications

As the results of NS15G show for a productive index (number of offspring per female per week), inbreeding depression affected its different components, especially in the reproductive characteristics, which could be regulated to a large extent by their simultaneous selection, probably due to a maintenance of genes that favored gonadal development in females and males. The work is relevant because the selection of a productive index in mice in its different components had not been analyzed, in an integral way, in addition to showing that, in a selection program with simultaneous consanguinity, the fixation of desirable alleles in the maintenance of reproductive cycles and the survival of the offspring is favored.

Literature cited:

1. Silver LM. Mouse genetics: Concepts and applications. 1st ed. Oxford, United UKA: Oxford University; 2001.

2. Falconer DS, MackayTFC. Introduction to quantitative genetics. 4th ed.Harlow, UKA: Longman; 1996.

3. Curik I, Sölkner J, Stipic N. The influence of selection and epistasis on inbreeding depression estimates. J Anim Breed Genet 2001;118:247-262. https://doi.org/10.1046/j.1439-0388.2001.00284.x

4. Leroy G. Inbreeding depression in livestock species: review and meta-analysis. Anim Genet 2014;45 (5):618-628.

5. Sallah BI, Seeland G. Einfluss von Inzucht und selektion auf die Fruchtbarkeit und das Wachstum der Maus. Arch. Tierz. Dummerstorf 2001;44(6): 671– 676.

6. Bohren BB. Designing artificial selection experiments for specific objectives. Genetics 1975; 80(1):205-220.

7. De la Fuente LF, San Primitivo F. Selection for large and small litter size of the first three litters in mice. Gênet Sêl Evo 1985;(17):251-264.

8. Eklund J, Bradford GE. Genetic analysis of a strain of mice plateaued for litter size. Genetics 1977;(85):529-542.

9. Festing FWM. Inbred strains in biomedical research. 1ra ed. London, UKA: Palgrave; 1979.

10. Hubrecht R, Kirkwood J. Handbook on care and management of laboratory animals. 8a ed. London, UKA: UFAW; 2010.

11. Benavides FJ, Guénet JL. Manual de genética de roedores de laboratorio: Principios básicos y aplicaciones. 1ra ed. Madrid, España: Universidad de Alcalá; 2003.

12.Tapia-PérezG.Respuestaalaselecciónparael númerodecríasdestetadasporsemana en líneas congénicas y singénicas de ratones de laboratorio [tesis maestría]. México, CDMX: Universidad Nacional Autónoma de México; 1995.

13. Kinghorn B, Kinghorn S. Pedigree Viewer, Ver 5.0. The University of New England 2010. https://bkinghor.une.edu.au/pedigree.htm Accessed Jan 10, 2021.

14. Wright S. Coefficients of inbreeding and relationship. The American Naturalist 1992; (56):330-338.

15. IBM SPSS Statistics for Windows, Ver 22.0. Armonk, New York: IBM Corp. 2013.

16. MedCalc Software Ltd. Comparison of proportions calculator, Ver 20.022. https://www.medcalc.org/calc/comparison_of_proportions.php. Accessed Dec 22, 2021.

17. Hinrichs D, Meuwissen THE, Odegard J, Holt M, Vangen O, Woolliams JA.Analysis of inbreeding depression in the first litter size of mice in a long-term selection experiment with respect to the age of the inbreeding. Heredity 2007;(99):81-88. https://doi.org/10.1038/sj.hdy.6800968

18. Langhammer M, Michaelis M, Hoeflich A, Sobczak A, Schoen J, Weitzel JM. Highfertility phenotypes: two outbred mouse models exhibit substantially different molecular and physiological strategies warranting improved fertility. Reproduction 2014;147(4):427-133. https://doi.org/10.1530/REP-13-0425.

19. NovairaHJ,Momodou LS,HoffmanG,KooY, KoC,WolfeA,Radovik S.Disrupted kisspeptin signaling in GnRH neurons leads to hypogonadotrophic hypogonadism. Molecular Endocrinology 2014; 28 (2): 225–238. https://doi.org/10.1210/me.20131319.

20. Toro M, Silio L, Rodrigañez J, Dobao M. Inbreeding and family index selection for prolificacy in pigs. Anim Sci 1988; 46(1): 79-85. https://doi:10.1017/S0003356100003135

21. Junnila RK, List EO, Berryman DE, Murrey JW, Kopchick JJ. The GH/IGF-1 axis in ageing and longevity. Nat Rev Endocrinol 2013;9(6):366-376. doi: 10.1038/nrendo.2013.67.

22. Charlesworth D,WillisJ.Thegenetics ofinbreedingdepression. Nat RevGenet 2009; (10):783–796.

23. Morello GM, Hultgren J, Capas-Peneda S, Wiltshire M, Thomas A, Wardle-Jones H, et al. High laboratory mouse pre-weaning mortality associated with litter overlap, advanced dam age, small and large litters. PloS one 2020;15(8):e0236290. https://doi.org/10.1371/journal.pone.0236290

24. Doekes HP, Veerkamp RF, Bijma P, De Jong G, Hiemstra SJ, Windig JJ. Inbreeding depression due to recent and ancient inbreeding in Dutch Holstein–Friesian dairy cattle. Genet Sel Evol 2019;51(54). https://doi.org/10.1186/s12711-019-0497-z.

25. Makanjuola BO, Miglior F, Abdalla EA, Maltecca C, Schenkel FS, Baes CF. Effect of genomic selection on rate of inbreeding and coancestry and effective population size of Holstein and Jersey cattle populations. J Dairy Sci 2020;103(6):5183-5199. https://doi.org/10.3168/jds.2019-18013

26. García-Ruíz A, Martínez-Marín GJ, Cortes-Hernández J, Ruíz-López FJ. Niveles de consanguinidad y sus efectos sobre la expresión fenotípica en ganado Holstein. Rev Mex Cienc Pecu 2021;12(4):996-1007. https://doi.org/10.22319/rmcp.v12i4.5681.

https://doi.org/10.22319/rmcp.v14i1.6129 Article

Genetic variability in aerial biomass and its components in alfalfa under irrigation and drought

a Colegio de Postgraduados. Postgrado en Recursos Genéticos y Productividad. Carretera México-Texcoco km. 36.5, Montecillo, Texcoco, Estado de México, México.

*Corresponding author: clc@colpos.mx

Abstract:

Drought decreases the yield of aerial biomass (BM) and its components, and the quality of forage in alfalfa. The genetic variation in BM and its components was studied in 10 varieties of alfalfa under irrigation (I) and drought (D) in a greenhouse. A randomized complete block experimental design was used, with four repetitions in I and four in D. The experimental unit was an individual plant in a PVC pipe. Sowing was carried out on March 15, 2017, and transplantingin the pipes, 20 days after sowing. The fertilization dose 60-140-00 was applied at 44, 240 and 420 dat (days after transplanting). D reduced (P≤0.01) BM, leaf dry matter yield (LDMY), number of stems (NS) and radiation use efficiency (RUE). The plants in D did not recover their productive capacity after experiencing the water deficit, even after the recovery irrigation. D also decreased (P≤0.01) the phenotypic variance for BM and its components; the additive variance was greater (P≤0.01) than the dominance variance for all traits in I and D. The BM, L:S ratio, plant height (PH), NS and RUE had higher (P≤0.01) heritability in I and D. The Genex, Atlixco, Júpiter and Milenia varieties were the most productive (P≤0.01) in D and could be used for forage production in water-scarce areas or as parental lines for forage yield improvement in selection programs.

Key words: Greenhouse, Heritability, Principal component analysis, Variance components.

Received: 22/12/2021

Accepted: 22/06/2022

Introduction

In Mexico, alfalfa (Medicago sativa L.) for forage is grown mainly under irrigation conditions and consumes large volumes of water. In regions with irrigation systems, a plant canopy of alfalfa can consume an amount of water of 10 mm day-1 at its peak of maximum development(1). In these growing conditions, the fall in the amount of precipitation over long periods of time decreases the water storage capacity in the subsoil and, therefore, the availability of irrigation. Likewise, when drought extends, the scarcity of water for irrigation is more severe and alfalfa crops may experience some degree of water stress, which can be reflected in a significant decrease in yield and forage quality(2) .

Inthenearfuture,thewaterresourcewillbelessavailablefortheproductionofalfalfaforage, due to the occurrence of frequent periods of drought, climate change and greater demands caused by the increase in the human population(3). One way to meet the demand in alfalfa forage production will be through the obtaining of new varieties with drought tolerance, high capacity of osmotic adjustment and gas exchange, high water use efficiency (e.g., more dry matter per unit of transpired or evapotranspirated water) and productive capacity(3). Alfalfa is considered a drought-resistant species, but its aerial biomass yield can fluctuate considerably under water deficit conditions; under these conditions, alfalfa has some agronomic advantages compared to other annual crops, as it has a root system that allows it to explore deeper soil layers to absorb water and tolerate drought to a greater degree; in addition to reducing the stomatal conductance and minimizing the transpiration rate(4) .

The most common reaction to a soil water deficit is the increase in the ratio of dry weight of root biomass/dry weight of aerial biomass, as a result of a greater reduction in the growth of aerial organs than in the growth of roots under drought. The increase in the root/aerial part ratio implies greater increases in root density with respect to aerial biomass, which is consequently reflected in a better capacity to maintain the water status of the plant under a given evapotranspiratory demand(5). Drought also reduces the yield of aerial biomass and its components, relative rates of growth, transpiration and elongation of the stem, chlorophyll content, relative water content, and dry weight and diameter of the root(6), and concentration of crude protein and water-soluble carbohydrates(7) .

On the other hand, drought-resistant alfalfa varieties exhibit high concentration of watersolublecarbohydratesinstorageorgansunderconditionsofseverewaterstress.Thissituation is combined with a water conservation strategy that implies less evapotranspiration in the initial phases of drought stress, due to a limited development of the root system that results in more available moisture, for its use under severe conditions of water stress(8). Biomass accumulation rates in plant roots and aerial organs were higher in 2-yr-old grasslands and aerial biomass accumulation was higher and maintained the best soil moisture conditions in 4-yr-old grasslands, once the crop reached the maximum development of the root system and cover of the soil surface(9). Drought-tolerant germplasm shows a lower degree of wilting underinitial conditions ofwaterdeficit, moreplants with thegreen plant canopyundersevere water stress conditions and more stems per plant under stress conditions or favorable moisture conditions(3). Despite the existence of a wide genetic variability in morphological and physiological traits associated with drought resistance, it is difficult to achieve the combination of adaptive traits to specific environments in the same variety with wide adaptation to environments vulnerable to drought(8) .

The genetic improvement of drought resistance and the yield of aerial biomass and its components requires special attention to traits with high heritability, general combining ability, additive genetic effects, maternal genetic effects, low genotype*environment interaction and ease of selection. In the analysis of the genetic variation of a population of the same species, additive genetic variance is the most important because it is the main determinant of the genetic properties observable in the population and of the response to selection(10). The additive variance is the only one that can be estimated directly from the observations made in the population and can be used in the estimation of heritability, which represents the reliability of the phenotypic value as an indication of the reproductive value, which determines its influence on the next generation(10). The similarity observed in the heritability values, for the traits measured in the plant under irrigation and drought, can be used as an indication of the effectiveness in the selection of new progenies, regardless of the selection environment(10). Broad-sense heritability (H2) measures the contribution of the genotype to the total phenotypic variance (���� 2); theoretically, it can varyin a range from zero, when there is no genetic variation present, to 1, when all the observed variation is genotypic in origin(11)

Selection for drought resistance can be achieved by increasing water use efficiency, drought severity index, mean productivity, harmonic mean, geometric mean, stress tolerance index, modified stress tolerance index, superiorityindex and abiotic tolerance index in water deficit conditions(12). Selection for morphological components of aerial biomass yield can be achieved by including the number of secondary stems and crown diameter per plant in the selection criteria(13). Other components of aerial biomass yield with moderate to high heritability that could be successfully used in selection to increase yield are absolute growth

rate, radiation use efficiency, number of stems, L:S ratio and plant height, in addition to the presence of maternal genetic effects favorable to aerial biomass yield(14). The selection of new varieties with drought resistance and high yield of aerial biomass and its components can be achieved by identifying the genetic traits with greater heritability and contribution to the productivity of the genotype. The objective of the present research was to study the genetic variability in the production of aerial biomass and its components, in commercial varieties of alfalfa under irrigation and drought in greenhouse conditions.

Material and methods

An experiment was carried out under irrigation and drought conditions in a greenhouse with a metal structure and transparent glass without whitewashing, and with a mechanical ventilation system in the College of Postgraduates, Montecillo, Texcoco, State of Mexico (19° 29’ N, 98° 53’ W and altitude of 2,250 masl) in the 2017-2019 period. The locality is characterized by having a subhumid temperate climate with long cool summer (Cb (wo) (w) (i´)g), average annual rainfall of 637 mm and winter rainfall of less than 5 %; average annual temperaturewithfluctuationsfrom12to18°C andthermaloscillationbetween5and7°C(15) Thegeneticmaterialusedincludedthe followingcommercialvarietiesofalfalfa:SanMiguel, Oaxaca, Atlixco, Aragón, Victoria, Genex, Júpiter, Milenia, San Isidro and Cuf 101, with germination percentage greater than 95 %. A randomized complete block experimental design was used, with four repetitions and two soil moisture treatments (irrigation and drought). The experimental unit was an individual plant transplanted in a cylindrical polyethylene bag inside a PVC pipe 1 m high and 4” in diameter, to favor the expression of the genetic potential of the morphological characteristics of the variety. The sowing was carried out on March 15, 2017, by placing five seeds of each variety in individual cells of seedbed boxes. At 20 days after sowing (das), the most vigorous seedling of each cell was selected and transplanted individually into the PVC pipes. The PVC pipes were filled with dry soil of sandy-loamy texture, bulk density of 1.12 T m-3 and pH of 7.3; 18.8 and 0.22 % of organic matter and total nitrogen; 176.3 mg kg-1 and 2,420 mg kg-1 of phosphorus and potassium; 54.6 Cmol(+) kg-1 and 0.53 dS m-1 of cation exchange capacity and electrical conductivity; and 52 and 38.2 % of field capacity (FC) and permanent wilting percentage (PWP) (Central University Laboratory, Chapingo Autonomous University, Chapingo, Mexico, 2016). The fertilization dose 60-140-00 was applied at 44 days after transplantation (dat), using urea and calcium triple superphosphate as sources of nitrogen and phosphorus, diluted in the irrigation water; a second and third fertilization was done at 240 and 420 dat with the same dose of fertilizer. Two treatments of soil moisture were used: irrigation, where the soil water content remained close to FC from the date of transplantation (20 das) to 406 dat (I1) and from 406 dat until the end of the experiment (798 dat) (I2), and drought, where

the application of water to plants was suspended in a first period for 61 d [345 to 406 dat; March to May 2018; (D1)] and a second period for 68 days [620-688 dat; November 2018 to February 2019; (D2)]. Recovery irrigation (RI) was applied to the plants at the end of the treatments of D1 (406 dat, RI1) and D2 (688 dat, RI2).

Cuts were made in the aerial part of the plant every 5 wk in the autumn-winter period and every four weeks in the spring-summer period, at a height of 5 cm above ground level. In eachcut, theplant height (PH,cm)was measuredfromthesoil surfaceto thelast leafexposed on the highest stem with a ruler graduated to 5 mm; in addition, the total number of stems (NS) was counted and the leaf:stem ratio (L:S) was determined in a subsample of four secondary stems, by dividing the leaf dry weight (LDW) by the stem dry weight (SDW), obtained after a drying period of 48 h at a temperature of 65 °C (L:S = LDW/SDW). The total dry matter yield (TDMY, g) or aerial biomass (BM) was calculated by adding the dry weight of leaves and secondary stems of the subsample used to determine the L:S ratio, and the dry weight of the leaves and secondary stems of the remaining sample of the plant. The leaf dry matter yield (LDMY, g) was represented by the dry weight of leaves. The radiation use efficiency (RUE, g d DM MJ-1) was calculated by dividing the TDMY by the solar radiation accumulated daily (data obtained from the meteorological station of the Chapingo Autonomous University) during the period between subsequent cuts(16). The maximum and minimum air temperature in the greenhouse was recorded daily with a maximum and minimummercurycolumnthermometer,Taylorbrandmodel5458P,placednext totheplants at a height of 2 m above floor level. The maximum temperature during the study ranged from 19 to 40 °C and the minimum from -4 to 15 °C, with an average of 32 and 8.5 °C. The water content in the soil was determined by the gravimetric method every third day with a Tor-Rey electronic balance, PCR Series model. In irrigation, the water content of the soil was kept close to FC, by adding water in each weighing during the experiment, while in drought, the plants were treated in the same way as in irrigation, except in the periods in which the application of water was suspended [345 to 406 (D1) and 620-688 (D2) dat] and only the decrease in soil weight in each PVC pipe (data not shown) was recorded.

The phenotypic variance (σ�� 2) and its components were estimated for the variables measured in all the cuts in irrigation (I1 and I2) and drought (D1 and D2), under the following statistical model(17,18):

Yijk = µ + DCi + R(DC)ij + Gk + G*DCik + Eijk

Where, Yijk is the value of the response variable; μ is the overall mean;

DCi is the effect of the date of cut;

R(DC)ij is the effect of repetitions within the date of cut; Gk is the effect of genotypes; G*DCik is the effect of the interaction between genotypes and dates of cut; Eijk is the experimental error.

Estimates of phenotypic variance and its components were made under the assumption of Hardy-Weinberg equilibrium, linkage equilibrium and absence of epistasis(17,19). The values of phenotypic variance (σ�� 2) and its components, and heritability (h2) were obtained from the values of the expectations of the mean squares of the analysis of phenotypic variance and its components as follows: σ�� 2 = σ�� 2 + σe 2 + σ��∗���� 2

Where, σ�� 2 is the additive variance (σA 2= (M1 – M2)/r*d), σe 2 is the environmental variance (σe 2 = M3) and σ��∗���� 2 is the variance of the interaction of genotypes*dates of cut (σ��∗���� 2 = (M2 – M3)/r); M1, M2 and M3 represent the expectations of the mean squares, d represents the date of cut and r represents the number of repetitions(17)

Narrow-sense heritability (h2) was calculated according to the following equation: h2 = (σ�� 2) / (σ�� 2). Where, σ�� 2 is the additive variance and σ�� 2 is the phenotypic variance.

The dominance variance (���� 2) was estimated(17) by using the additive variance (����) between half-sib families(20): ���� 2 = 3 4 ���� 2 +���� 2 and ���� 2 = 1 4 ���� 2

Where, ���� 2 is the genetic variance and the value of ���� 2 is obtained as follows(20): ���� 2 = 1 4 ���� 2

Narrow-sense heritability(h2) was calculated under the assumption that the varieties used are a random and representative sample of the genetic variability of alfalfa and considering that this is an allogamous species(17). Thus, the component of variance obtained from the mathematical expectation of the mean square of the factor of varieties is an estimator of the additive variance(21) .

The data obtained were analyzed with the GLM(22) procedure, version for Windows 10, with a completely randomized design in factorial arrangement. The means of soil moisture

treatments, genotypes and genotypes within soil moisture treatments were compared with the honest minimum significant difference (HMSD, P<0.05) according to the following model:

Yij = µ + Ti + Gj + T*Gjj + Eij Where, Yij is the value of the response variable; μ is the overall mean; Ti represents soil moisture treatments; Gj represents genotypes; T*Gjj represents the interaction between soil moisture treatments and genotypes; Eij is the experimental error(23)

Results and discussion

The soil moisture treatments were different (P≤0.01) in total dry matter yield and leaf dry matter yield in the cuts made between 406 and 798 dat; differences (P≤0.01) in the L:S ratio at 406, 434, 462, 490 and 686 dat; differences (P≤0.01) in plant height at 406, 434, 462, 686, 742, 770 and 798 dat; and differences (P≤0.01) in number of stems and radiation use efficiency between 406 and 798 dat (Table 1). The varieties showed differences (P≤0.01) in total dry matter yield, L:S ratio, plant height and radiation use efficiency in all cuts made between 112 and 798 dat; differences (P≤0.01) in leaf dry matter yield and number of stems in all cuts, except for cuts made at 245, 406, 434, 553 and 588, and 140 dat. The interaction of soil moisture treatments*varieties showed differences (P≤0.01) in total dry matter yield at 112, 140, 210, 406 and 746 dat and differences (P≤0.05) at 175, 315, 434 and 770 dat; differences (P≤0.01) in leaf dry matter yield at 112, 140 and 210 dat, and differences (P≤0.05) at 175, 742 and 770 dat; differences (P≤0.01) in the L:S ratio at 112, 140, 175, 210, 245, 280, 315, 406, 434, 490, 686, 770 and 798 dat, differences (P≤0.05) at 588 dat; differences (P≤0.01) in plant height at 112, 245, 280, 490, 742 and 798 dat, and differences (P≤0.05) at 112, 210, 315 and 406 dat; differences (P≤0.01) in number of stems at 175, 315 and 434 dat, and differences (P≤0.05) at 140, 245, 462, 518 and 686 dat; and differences (P≤0.01) in radiation use efficiency at 140, 210, and 742 dat, and differences (P≤0.05) at 112, 175, 315, 434, and 770 dat.

The comparison of the total dry matter yield and its components in irrigation vs. drought showed that the water deficit of the soil in D1 and D2 reduced (P≤0.01) the total dry matter yield and leaf dry matter yield, number of stems and radiation use efficiency from 406 to 798

dat; plants under drought did not recover their productive capacity after experiencing the water deficit in D1 and D2 with respect to plants under irrigation (I1 and I2), even after recoveryirrigations(RI1 andRI2)(Figure1).The L:Sratioinplantsunderdrought washigher (P≤0.01) than in irrigation (I1 and I2), and these differences between irrigation and drought were more noticeable during the application of drought (D1 and D2). The plant height in D1 andD2 waslower(P≤0.01)thaninirrigation(I1 andI2)andsubsequentlyrecoveredits growth capacity with respect to its behavior in irrigation. The survival of alfalfa through periods of water deficit in field conditions depends on the length and intensity of the drought, the genotype, the type of soil (water capacity of the soil and depth of the root system) and the environment (salinity and temperature); its survival to short periods (2-3 weeks) without irrigation is reflected in its high recovery capacity when receiving irrigation again and producing normal yields in subsequent years(24). The greater recovery capacity of alfalfa when receiving water after experiencing periods of water deficit(24) may be due to the fact that plants that grow in field conditions have greater access to moisture and nutrients in the soil profile, unlike plants that grow in greenhouse conditions in pots or PVC pipes, where plant roots grow in an environment limited in soil volume, moisture and nutrients; this is reflected in a reduction in the accumulation of aerial biomass due to a decrease in stomatal conductance, transpiration and assimilation(3). The high values in the L:S ratio in drought could be due to a lower partition of assimilates to the stem with respect to the leaf; plants subjected to water stress show some morphological changes in response to water deficit, by reducing the loss or increasing the absorption of water to maintain the water status of the tissue(25). Plant height was the only morphological characteristic that showed recovery capacity after water application (RI1 and RI2), reaching values similar to those observed in plants under irrigation; soil water deficit affects different morphological characteristics of plants, such as plant height, stem diameter, number, size and area of leaves, dry matter production, assimilate partitioning, flower and fruit production, and physiological maturity(25) .

Figure 1: Yield of total dry matter (a) and leaf dry matter (b), leaf:stem ratio (c), plant height (d), number of stems (e) and radiation use efficiency (f) in 18 cuts in irrigation (R1=I1 and R2=I2) and drought (S1=D1 and S2=D2), average of 10 varieties of alfalfa

Montecillo, Texcoco, State of Mexico [RR1=Recovery irrigation in I1 (RI1); RR2=Recovery irrigation in I2 (RI1); *(P≤0.05); **(P≤0.01); ns (not significant)].

On the other hand, in irrigation (I1 and I2), a wide variability (P≤0.01) was observed between genotypes for total dry matter yield (Figures 2a and 3a), L:S ratio (Figures 2c and 3c), plant height (Figures 2d and 3d) and radiation use efficiency (Figures 2f and 3f) in all cuts in I1 (112 to 434 dat) and I2 (462 to 798 dat). The Genex, Atlixco, Júpiter, Oaxaca, San Miguel and Milenia varieties produced more (P≤0.01) total dry matter yield than the other varieties in all cuts in I1 (Figure 2a),and onlytheGenex,Atlixco,JúpiterandMilenia varieties showed

high (P≤0.01) total dry matter yield in I2 (Figure 3a). The high total dry matter yield in the Genex, Atlixco, Júpiter, Oaxaca, San Miguel and Milenia varieties (Figure 2a) was accompanied by high (P≤0.01) leaf dry matter yield (Figure 2b), plant height (Figure 2d), number of stems (Figure 2e) and radiation use efficiency (Figure 2f) in I1. The high (P≤0.01) total dry matter yield of the Genex, Atlixco, Júpiter and Milenia varieties (Figure 3a) was also accompanied by high (P≤0.01) leaf dry matter yield (Figure 3b), plant height (Figure 3d), number of stems (Figure 3e) and radiation use efficiency (Figure 3f) in I2. The Victoria, Aragón and San Isidro (Figure 2c), and Aragón and San Isidro (Figure 3c) varieties showed a higher (P≤0.01) L:S ratio than the other varieties in I1 and I2. In a study with 11 alfalfa cultivars under greenhouse irrigation conditions, it was determined that BCB, ALF and AFR varieties showed higher yields of total dry matter, root dry matter, stem elongation rate, relative water content and root diameter than the other alfalfa varieties(6). The varieties F 1412-02, F 1535-03, Roxana and F 2007-08, and F 1414-02, F 1711-05, F 1715-05 and F 2010-08 stood out from a group of 74 genotypes under greenhouse irrigation conditions, producing higher total dry matter yield, plant height and number of stems than the rest of the varieties(4) .

Figure 2: Yield of total dry matter (a) and leaf dry matter (b), leaf:stem ratio (c), plant height (d), number of stems (e) and radiation use efficiency (f) in nine cuts in irrigation (I1), for 10 varieties of alfalfa

R1= Irrigation in the cutting period from 112 to 406 dat (I1).

Figure 3: Yield of total dry matter (a) and leaf dry matter (b), leaf:stem ratio (c), plant height (d), number of stems (e) and radiation use efficiency (f) in nine cuts in irrigation (I2), for 10 varieties of alfalfa

R2= Irrigation in the cutting period from 462 to 798 dat (I2).

In drought, a wide variability (P≤0.01) was also observed between genotypes for total dry matter yield (Figures 4a and 5a), L:S ratio (Figures 4c and 5c), plant height (Figures 4d and 5d) and radiation use efficiency (Figures 4f and 5f) in all cuts in D1 (112 to 406 dat) and D2 (462 to 798 dat). The Genex, Atlixco, Júpiter, Oaxaca, San Miguel and Milenia varieties produced higher (P≤0.01) total dry matter yield than the other varieties in all cuts in D1 (Figure 4a), and onlythe Genex, Atlixco,Júpiter and Milenia varieties showed high (P≤0.01) total drymatter yield in D2 (Figure 5a). The high total drymatter yield of the Atlixco, Júpiter,

Oaxaca, San Miguel and Milenia varieties (Figure 4a) was accompanied by higher (P≤0.01) leaf dry matter yield (Figure 4b), plant height (Figure 4d), number of stems (Figure 4e) and radiation use efficiency (Figure 4f) in I1. In I2, the highest (P≤0.01) total dry matter yield of the Genex, Atlixco, Júpiter and Milenia varieties (Figure 5a) was also accompanied by high (P≤0.01) leaf dry matter yield (Figure 5b), plant height (Figure 5d), number of stems (Figure 5e) and radiation use efficiency (Figure 5f). The Milenia, Victoria, Cuf-101, Aragón and San Isidro (Figure 4c), and Victoria, Aragón and San Isidro (Figure 5c) varieties showed a higher (P≤0.01) L:S ratio than the other varieties in I1 and I2. Other studies in different varieties of alfalfa under greenhouse drought detected genotypes that reduce less stem elongation, relative growth rate and aerial biomass with respect to irrigation, in addition to maintaining greater root growth capacity, relative water content, chlorophyll content and water use efficiency(6). The Gold Queen variety produced higher yield of dry matter and water-soluble carbohydrates and was more drought-resistant than the Suntory variety under field conditions; drought decreased crude protein content and increased fiber fraction in response to water deficiency in the two alfalfa varieties(7). The Amerist (USA), Sardi10 and Siriver (Australia), and Melissa (France) genotypes showed greater drought tolerance than other alfalfa varieties, because they produced thinner leaves, accumulated more proline and potassium, and maintained greater efficiency in the use of water in conditions of water deficiencies(26).TheAragonand San Isidrovarieties consistentlyshowedhighaveragevalues for the L:S ratio in irrigation and drought; this morphological characteristic of the plant is highly appreciated as an estimator of forage quality and can be used to improve yield, and dry matter quality in lines, half-sib families or clones in large populations, considering its high values of narrow-sense heritability (h2=0.75)(27) .

Figure 4: Yield of total dry matter (a) and leaf dry matter (b), leaf:stem ratio (c), plant height (d), number of stems (e) and radiation use efficiency (f) in nine cuts in drought (D1), for 10 varieties of alfalfa

S1= Drought in the cutting period from 112 to 406 dat (D1).

Figure 5: Yield of total dry matter (a) and leaf dry matter (b), leaf:stem ratio (c), plant height (d), number of stems (e) and radiation use efficiency (f) in nine cuts in drought (D2), for 10 varieties of alfalfa

S2= Drought in the cutting period from 462 to 798 dat (D2).

The phenotypic variance for total dry matter yield and leaf dry matter yield, L:S ratio, plant height, number of stems and radiation use efficiency in irrigation (I1 and I2) was higher (P≤0.05) than in drought (D1 and D2). The phenotypic variance for the total dry matter yield and its components was greater (P≤0.05) than the other components of variance in irrigation and drought. However, environmental variance contributed more (P≤0.05) to phenotypic variance than genetic variance in both irrigation and drought. The additive genetic variance was greater (P≤0.05) than the dominance genetic variance for all traits measured in plants in

irrigation and drought. The variance of the interaction was lower than the phenotypic, environmental and additive genetic variances, for all the traits measured in the plants in irrigation and drought (Table 2). In autotetraploid alfalfa, similar results were obtained when estimating the components of variance; the dominance variance was much lower than the additive variance for the yield of dry matter and its components(28). The additive variance was significantly greater than zero and the genetic variance for dry matter yield was mainly additive in an F1 population of alfalfa under controlled growth conditions(29). Heritability(h2) was low forleafdrymatter yield to moderatefortotaldrymatter yield, L:S ratio,plantheight, number of stems, and radiation use efficiency in irrigation and drought (Table 2). These heritability values are similar to those obtained for aerial biomass and plant height in annual alfalfa (Medicago sativa subsp. falcata) under field conditions(28) and could be useful in improving the yield of alfalfa dry matter with the support of genomic selection(27)

Table 2: Estimated genetic parameters for total dry matter yield (TDMY) and leaf dry matter yield (LDMY), leaf:stem ratio (L:S), plant height (PH), number of stems (NS) and radiation use efficiency (RUE) in irrigation (I1 and I2), and drought (D1 and D2), average of 10 varieties of alfalfa

Genetic parameters

TDMY LDMY L:S PH NS RUE Irrigation I1 and I2

Phenotypic variance (���� 2) 3.6 (0.7) 0.5 (0.1) 0.01 (0.001) 86.4 (7.6) 16.0 (1.6) 0.021 (0.001)

Genotypic variance (���� 2) additive (���� 2) 1.2 (0.4) 0.1 (0.05 0.005 (0.0003) 31.6 (1.2) 4.5 (0.8) 0.01 (0.001) dominance (���� 2) 0.3 0.02 0.001 7.9 1.1 0.002 interaction (����∗���� 2 ) 0.7 0.06 0.002 12.3 2.5 0.002 Environmental variance (���� 2) 1.7 (0.4) 0.4 (0.08) 0.004 (0.0008) 42.6 (6.9) 9.0 (1.6) 0.01 (0.001) Heritability (h2) 0.3 (0.04) 0.2 (0.04) 0.4 (0.04) 0.4 (0.03) 0.3 (0.04) 0.4 (0.04)

Drought D1 and D2

Phenotypic variance (���� 2) 1.5 (0.2) 0.2 (0.03) 0.01 (0.001) 61.6 (5.8) 11.5 (0.8) 0.015 (0.007)

Genotypic variance (���� 2) additive (���� 2) 0.5 (0.02) 0.04 (0.004) 0.004 (0.0003) 20.4 (2.0) 4.1 (0.3) 0.046 (0.005) dominance (���� 2) 0.1 0.01 0.001 5.1 1.0 0.001 interaction (����∗���� 2 ) 0.2 0.04 0.004 13.7 1.8 0.002

Environmental variance (���� 2) 0.8 (0.2) 0.1 (0.03) 0.001 (0.0003) 27.5 (4.9) 5.5 (0.8) 0.008 (0.2)

Heritability (h2) 0.3 (0.04) 0.2 (0.04) 0.4 (0.03) 0.3 (0.04) 0.4 (0.04) 0.3 (0.04)

The analysis of principal components (PC1 and PC2) identified two components that explain the largest proportion of the total variation (75.8 %) shown in the experiment. PC1 explained 56.2% of the variation and had a positive correlation with total dry matter yield (r= 0.52), leaf dry matter yield (0.50), number of stems (r= 0.42), radiation use efficiency (r= 0.40) and plant height (r= 0.34), and negative correlation with L:S ratio (r= -0.19). PC2 explained only 19.6 % of the observed variability and had a positive correlation with the L:S ratio (r= 0.78) and leaf dry matter yield (r=0.31), and negative correlation with plant height (r= -0.49) (Figure 6). Additionally, total dry matter yield was positively related to the number of stems and leaf dry matter yield, and negatively related to plant height; plant height was negatively related to L:S ratio. The variability observed for yield of dry matter and its components in the present study was similar to that observed in a group of 27 populations and cultivars of alfalfa under field conditions, where PC1 contributed 58.2 % of the total variability and showed positive association with dry and green matter yield, vigor, growth habit, regeneration of the plant and width of the central leaflet(30). Other results in irrigated and rainfed alfalfa in the field showed a PC1 with 54.3 % of the total variability and positive association with the diameter of lateral roots and number of lateral or branched roots(31). It is interesting to note the similarity in the values observed for PC1 and the variability between genotypes in these studies, and the traits of the plant that had the greatest positive association with this component, especially with dry matter yield.

Figure 6: Biplot plane of dry matter yield vs. total dry matter yield (RMST), leaf dry matter yield (RMSH), L:S ratio (H:T), number of stems (NT), plant height (AP) and radiation use efficiency (EUR) in irrigation (I1 and I2) and drought (D1 and D2), on average of 10 varieties of alfalfa in greenhouse conditions

Conclusions and implications

The drought decreased the total dry matter yield and its components, and plants under soil water deficit conditions did not recover theirproductivecapacityafterexperiencingthe water deficiencies of the soil, even after recovery irrigation. In contrast, the L:S ratio was higher in plants in drought than in irrigation and plant height was the only component of yield that regained its growth capacity after recovery irrigation. Soil water deficit also reduced phenotypic variance for total dry matter yield and its components; environmental variance was greater than genetic variance in irrigation and drought. Additive variance was greater than dominance variance for all traits measured in irrigation and drought. Total dry matter yield, L:S ratio, plant height, number of stems, and radiation use efficiency had higher heritability in irrigation and drought. Leaf dry matter yield, number of stems, radiation use efficiency and plant height were positively related to total dry matter yield. The most productive varieties could be used for forage production in water-scarce areas and/or as parental lines for forage yield improvement in selection programs. Future research work on this topic requires confirmation under field conditions.

Literature cited:

1. Guitjens JC. Alfalfa. In: Stewart BA, Nielsen DR, editors. Irrigation of agricultural crops. American Society of Agronomy, Inc. Madison, Wisconsin USA; Monograph Number 30 in the series of Agronomy; 1990:537-596.

2. Lauriault L, Marsalis M, Contreras-Govea F, Angadi S. Managing alfalfa during drought. Cooperative Extension Service, College of Agricultural and Environmental Sciences, New Mexico State University Las Cruces, New Mexico; 2009 (Circular 646):4.

3. Luna-Guerrero MJ, López-Castañeda C, Quero-Carrillo AR, Herrera-Haro JG, OrtegaCerrilla ME, Martínez-Hernández PA. Water relations and gas exchange in lucerne under drought conditions. Rev Mex Cienc Agríc 2020; Special Publication Number 24:81-92.

4. Petcu E, Schitea M, Drăgan L, Bǎbeanu N. Physiological response of several alfalfa genotypes to drought stress. Rom Agric Res 2019;36:107-118.

5. Blum A. Crop responses to drought and the interpretation of adaptation. Plant Grow Reg 1996;20:135–148.

6. Anower MR, Boe A, Auger D, Mott IW, Peel MD, Xu L, Kanchupati P, Wu Y. Comparative drought response in eleven diverse alfalfa accessions. J Agron Crop Sci 2017;203:1-13.

Pecu 2023;14(1):39-60

7. Liu Y, Wu Q, Ge G, Han G, Jia Y. Influence of drought stress on alfalfa yields and nutritionalcomposition.BMCPlantBiology2018;18(13):1-9.doi:10.1186/s12870-0171226-9.

8. AnnicchiaricoP,Pecetti L, Tava A.Physiological andmorphological traits associated with adaptation of lucerne (Medicago sativa) to severely drought-stressed and to irrigated environments. Ann Appl Biol 2013; 162:27–40. doi:10.1111/j.17447348.2012.00576.x.

9. Huang Z, Liu Y, Cui Z, Fang Y, He H, Liu BR, Wu GL. Soil water storage deficit of alfalfa (Medicago sativa) grasslands along ages in arid area (China). Field Crop Res 2018;221:1-6.

10. Falconer DS. Introducción a la genética cuantitativa. México: Cía. Editorial Continental, SA de CV: 1984.

11.HillJ,BeckerHC,TigerstedtPMA.Quantitativeandecological aspects ofplantbreeding. London: Chapman & Hall; 1998.

12. Bellague D, Hammedi-Bouzina MM, Abdelguerfi A. Measuring the performance of perennial alfalfa with drought tolerance indices. Chil J Agric Res 2016;76(3):273-284. doi:10.4067/ S0718-58392016000300003.

13. Márquez-Ortiz JJ, Lamb JFS,Johnson LD, Barnes DK, Stucker RE.Heritabilityof crown traits in alfalfa. Crop Sci 1999;39:38-43.

14. Luna-Guerrero MJ, López-Castañeda C, Hernández-Garay A. Genetic improvement of aerial alfalfa biomass and its components: half-sib family selection. Rev Mex Cienc Pecu 2020;11(4):1126-1141. doi.org/10.22319/rmcp.v11i4.5344.

15. García E. Modificaciones al sistema de clasificación climática de Köppen. Serie Libros Núm. 6, Instituto de Geografía, UNAM. México, DF; 2004.

16. Luna-Guerrero MJ, López-Castañeda C, Hernández-Garay A, Martínez-Hernández PA, Ortega-Cerrilla ME. Evaluación del rendimiento de materia seca y sus componentes en germoplasma de alfalfa (Medicago sativa L.). Rev Mex Cienc Pecu 2018;9(3):486-505. doi.org/10.22319/rmcp.v9i3.4440.

17. Molina-Galán JD. Introducción a la genética de poblaciones y cuantitativa (algunas implicaciones en genotecnia). México, DF: AGT Editor, SA; 1992.

18. Márquez-Sánchez F, Sahagún-Castellanos J. Estimation of genetic variances with maternal half-sib families. Maydica 1994;39(3):197-201.

Pecu 2023;14(1):39-60

19. Melendres-Martínez JI, Valdivia-Bernal R, Lemus-Flores C, Medina-Torres R, GarcíaLópez M, Ortiz-Caton M, et al. Estimación de parámetros genéticos de maíz bajo mejoramientoporselecciónrecíprocarecurrente.RevMexCiencAgríc2018;9(7):13271337.

20. Galicia-Juárez M. Varianza genética y mapeo molecular de rendimiento y calidad nutricional en familias de medios hermanos en Medicago sativa [tesis maestría]. Texcoco, México: Colegio de Postgraduados; 2012.

21.HillJ,BeckerHC,TigerstedtPMA.Quantitativeandecologicalaspects ofplantbreeding. London: Chapman & Hall; 1998.

22. SAS (Statistical Analysis System), Version 9.4 para Windows. SAS Institute Inc., Cary, NC, USA; 2012.

23. Hinkelmann K, Kempthorne O. Design and analysis of experiments. Volume 1: Introduction to experimental design. USA: A John Wiley and Sons, Inc; 2008.

24. Orloff S, Putnam D, Bali K. Drought strategies for alfalfa. Agriculture and Natural Resources, UC, USA. Publication 8522; 2015:1-9 (http://anrcatalog.ucanr.edu/).

25. Anjum SA, Ashraf U, Zohaib A, Tanveer M, Naeem M, Ali I, Tabassum T, Nazir U. Growth and developmental responses of crop plants under drought stress: a review. Zemdirbyste-Agriculture 2017;104(3):267-276. doi:10.13080/z-a.2017.104.034.

26. Benabderrahim MA, Hamza H, Haddad M, Ferchichi A. Assessing the drought tolerance variability in Mediterranean alfalfa (Medicago sativa L.) genotypes under arid conditions. Plant Biosystems 2015; 149 (2):395-403. doi.org/10.1080/11263504.2013.850121.

27. Annicchiarico P. Alfalfa forage yield and leaf/stem ratio: narrow-sense heritability, genetic correlation, and parent selection procedures. Euphytica 2015;205(2):409–420. doi:10.1007/s10681-015 1399-y.

28. Riday H, Brummer EC. Narrow sense heritability and additive genetic correlations in alfalfa subsp. falcata. J Iowa Academy Sci 2007;114(1-4):28-34.

29. Bowley SR, Christie BR. Inheritance of dry matter yield in a heterozygous population of alfalfa. Can J Plant Sci 1981;61:313-318.

30. Tucak M, Popović S, Ćupić T, Šimić G, Gantner R, Meglić V. Evaluation of alfalfa germplasm collection by multivariate analysis based on phenotypic traits. Rom Agric Res 2009;26:47-52.

31. Odorizzi A, Basigalup D, Arolfo V, Balzarini M. Análisis de la variabilidad de caracteres de raíz en poblaciones de alfalfa (Medicago sativa L.) con alto número de raíces laterales. AgriSci 2008;25(2):65-73.

Rev Mex Cienc Pecu 2023;14(1):39-60

Table 1: Factors of variation, degrees of freedom (DF) and significance of total dry matter yield (TDMY) and leaf dry matter yield (LDMY), leaf:stem ratio (L:S), plant height (PH), number of stems (NS) and radiation use efficiency (RUE) in irrigation (I1) and drought (D1) (112-434 dat), and in I2 and D2 (462-798 dat)

Characteristic DF 112 140 175 210 245 280 315 406 434 462 490 518 553 588 686 742 770 798

TDMY (g DM plant-1)

A 1 ns ns ns ns ns ns ns ** ** ** ** ** ** ** ** ** ** **

B 9 ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** **

A*B 9 ** ** * ** ns ns * ** * ns ns ns ns ns ns ** * ns

LDMY (g DM plant-1)

A 1 ns ns ns ns ns ns ns ** ** ** ** ** ** ** ** ** ** **

B 9 ** ** ** ** ns ** ** ns ns ** * ** ns ns * ** ** **

A*B 9 ** ** * ** ns ns ns ns ns ns ns ns ns ns ns * * ns L:S ratio

A 1 ns ns ns ns ns ns ns ** ** ** ** ns ns ns ** ns ns ns

B 9 ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** **

A*B 9 ** ** ** ** ** ** ** ** ** ns ** ns ns * ** ns ** ** PH (cm)

A 1 ns ns ns ns ns ns ns ** ** ** ns ns ns ns ** ** ** **

B 9 ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** **

A*B 9 ** * ns * ** ** * * ns ns ** ns ns ns ns ** ns ** NS

A 1 ns ns ns ns ns ns ns ** ** ** ** ** ** ** ** ** ** ** B 9 ** ns ** ** ** ** ** * ** ** ** ** ** ** ** ** ** **

A*B 9 ns * ** ns * ns ** ns ** * ns * ns ns * ns ns ns

RUE (g DM MJ-1)

A 1 ns ns ns ns ns ns ns ** ** ** ** ** ** ** ** ** ** **

B 9 ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** **

A*B 9 * ** * ** ns ns * ns * ns ns ns ns ns ns ** * ns

A=Soil moisture treatments (Irrigation=I1 and I2, and Drought=D1 and D2); B=Genotypes; A*B Interaction of soil moisture treatments*genotypes; *(P≤0.05); **(P≤0.01); ns (not significant). D1 (345-406 dat) and D2 (620-688 dat).

https://doi.org/10.22319/rmcp.v14i1.6162 Article

Estimation of forage mass in a mixed pasture by machine learning, pasture management and satellite meteorological data

Aurelio Guevara-Escobar a

Mónica Cervantes-Jiménez a*

Vicente Lemus-Ramírez b

Adolfo Kunio Yabuta-Osorio b

José Guadalupe García-Muñiz c

a Universidad Autónoma de Querétaro Facultad de Ciencias Naturales. 76230 Juriquilla, Santiago de Querétaro, Querétaro, México.

b Universidad Nacional Autónoma de México. Facultad de Medicina Veterinaria y Zootecnia, Centro de Enseñanza, Investigación y Extensión en Producción Animal en Altiplano CEIEPAA. Querétaro, México.

c Universidad Autónoma Chapingo Departamento de Zootecnia, Posgrado enProducción Animal. Estado de México, México.

* Corresponding author: monica.cervantes@uaq.mx

Abstract:

Measuring forage mass (FM) in the pasture, prior to grazing, is critical to determining the daily allocation of forage in pastoral animal production systems. FM is estimated by cutting forage in known areas, using allometric equations, or with the use of remote sensors (RS); however, the accuracy and practicality of the different methods for estimating FM is variable. The objective was to obtain predictive models using environmental and pasture management variables to predict FM. Regression models were fitted to estimate FM based on variables of pasture management (PM) or measurements obtained byRS, such as reflectance, air temperature, and rainfall. A mixed pasture grazed by beef cattle was studied for three years. With 80 % of data, models were built by ordinary least squares (OLS) or by machine learning (ML) algorithms. The remaining 20 % of the data was used to validate the models using the coefficient of determination

and average bias between estimated and observed values. The base model of study was the relationship between pasture height before grazing and FM, this model was fitted using OLS; the r2 was 0.43. When models that included PM variables were fitted, the r2 was 0.45 for OLS and 0.63 for ML. When fitting models with PM and RS variables, the r2 was 0.71 for OLS and 0.96 for ML. ML-fitted model ensembles reduced the bias of FM estimates oftheexamined pasture.Overall,MLmodels betterrepresentedtherelationship betweenpastureheight before grazingand FM than OLS models, when fitted with pasture management variables and RS information. ML models can be used as a tool for daily decision-making in pastoral production systems.

Key words: Alfalfa, Forage, Rain, Lucerne, Temperature, Remote sensors.

Received: 08/03/2022

Accepted: 18/07/2022

Introduction

Animal production using grazed pastures depends on the rate of accumulation of forage mass (FM), as well as on the timely allocation of an adequate stocking rate to take advantage of the FM; other important aspects are nutritional quality and seasonality in the rate of accumulation of FM. Cost-effective management of a pasture through direct grazing involves, among other things, implementing grazing management without compromising vegetation cover regrowth, as well as accurately knowing the FM in the pasture before and after grazing(1). Traditionally, FM is measured directly with forage cuts in quadrants of known area, distributed in a spatially representative manner and in a sufficient number that represents the variability of the vegetation cover in the pasture(2,3) . The cutting of quadrants is laborious and therefore methods and devices have been developed for the indirect estimation of FM(4-6) Pasture canopy height, measured with a sward stick, is useful to represent the FM, although the relationship may be different depending on the botanical composition, density of the pasture canopy and season of the year(7-9).Theheightofthecompressedforagemeasuredwitharisingplatemeterestimates the FM considering the density of the canopy and is a very common practice at the farm level in countries such as New Zealand(2). The relationship between canopy height and FM in ryegrass and white clover pastures is well known and routinely applied in New Zealand(10); for pastures with other forage species such as alfalfa, more research is needed to determine the relationship between canopy height and FM(8)

Remote sensing (RS) by orbital satellites measures spectral reflectance, the proportion of incident energy reflected by the Earth’s surface at different wavelengths; these measurements have been associated with vegetation activity processes(11). With RS information, it is also possible to estimate environmental variables such as temperature,

rainfall, and others(12). The wide availability and free access of RS products is an opportunity to explore crop dynamics and establish relationships with productive parameters, such as FM. The time series available for different RS products allow retrospective studies to be made, which is valuable for evaluating pasture management practices and regional grassland studies. However, the spatial scale of measurement is coarse in some RS sensors and is an important disadvantage in studies such as the one described in this research.

Recently, a variety of machine learning (ML) algorithms have been incorporated into regression analysis and they are an alternative to ordinary least squares (OLS) regression. Photosynthesis in ecosystems, named gross primary productivity and net primary productivity (when discounting losses by respiration), has been modeled with empirical or mechanistic approaches, from OLS models to those that simulate ecophysiological processes at the global level based on RS(13). Net primary productivity includes photosynthetic partitioning into aerial and root biomass and therefore does not reflect the FM available for grazing. Lang et al(14) estimated arid grassland production using measurements from rainfall RS sensors, spectral reflectance obtained from the Landsat 7 satellite and random forest; a ML algorithm. Using Neural Networks, another ML-type algorithm, Chen et al(15) related the spectral reflectance measured by the Sentinel-2 satellite and FM on dairy farms of Tasmania in Australia. In these studies, the coefficient of determination (r2) in different models was between 0.6 and 0.7. Conceptually, it is important to incorporate humidity conditions, in the short or medium term, to explain the carrying capacity of the grassland(16), since water is the main limiting resource of plants in arid and semi-arid environments. The conditions of water availability for plants can be represented by the precipitation (P) that occurred, water available in the soil or vapor deficit in the atmosphere. However, to explain the FM, not only the P occurred in the period of accumulation of the FM (month in which the FM was measured) is important, but also the humidity conditions that occurred in previous months.

In the present work, the relationship between FM and pasture height was examined as a baseline to compare other models that used meteorological variables obtained by RS or in conjunction with variables representative of pasture management (PM) conditions; such as the grazing and rest periods of the grazed area or the pasture height itself. In particular, the usefulness of models to predict FM based on previous rainfall and temperature conditions in different time windows was explored; for example, the P accumulated in the previous month, in two months or three months before the measurement of the FM. The objective was to obtain a predictive model of FM that could be incorporated into grazing planning. For this purpose, three years of measurements on a mixed of alfalfa-grass pasture grazed by beef cattle, were used.

Material and methods Site

The study was carried out at the Centro de Enseñanza, Investigación y Extensión en Producción Animal en el Altiplano, run by Facultad de Medicina Veterinaria y Zootecnia fromtheUniversidadNacionalAutónomadeMéxico Thesiteislocatedat20°36’13.88” N, 99° 55’ 02.91” W and altitude of 1,913 m asl. The climate is extreme dry Ganges type without dry spells, BS1 0w(e)g, according to the historical climatological records (1951 to 2006) of climatological station 22025; the closest to the site, where the annual averages of precipitation and temperature are 458 mm and 23.5 °C(17) .

The pasture was established in 2004 with a mixture of 50 % alfalfa (Medicago sativa) and grasses such as orchard grass (Dactylis glomerata), tall fescue (Festuca arundinaceae) and perennial ryegrass (Lolium perenne). The grazing area was 19 ha divided into 16 paddocks of equal size and delimited through mobile electric fence. The pasture was irrigated with a center-pivot sprinkler system; however, there were no records of the irrigation sheet or calendar. The grazing mob was made up of 88 dams of the Limousin breed and their calves. The grazing time in each division was established based on: the estimation of FM, proximate chemical analysis of FM samples, and the dry matter (DM) allowance for the mob in each turn. Reproductive management was mainly with artificial insemination and year round calving

Data

From 2008 to 2010, 399 FM observations were obtained prior to grazing of the allocated grazing area. Each FM observation corresponded to the beginning of a grazing cycle of the mob. The observations were considered experimental units, and each consisted of eight random measurements obtained with the modified quadrat technique; to protect the alfalfa regrowth the pasture samples were cut to 10 cm height in an area of 0.25 m2(18) . Forage samples were dehydrated in a forced-air oven for 48 h to determine the DM content and the data was expressed in kg DM ha-1. In each grazing cycle, the following were recorded: the height of the pasture (H_pasture), the date of grazing (Day_grazing and Month_grazing), grazing time (G_time), resting time of the grazed area from the previous grazing (R_time), month of the beginning of growth in the previous grazing cycle (Month_beg_grow) and the average monthly pasture accumulation rate of DM (PAR, kg DM ha-1 d-1). These variables were collectively referred to as pasture management (PM) variables.

Using the Application for Extracting and Exploring Analysis Ready Samples of the Land Processes Distributed Active Archive Center of the National Aeronautics and Space Administration (NASA), the MCD43A4 version 6(19) product was requested. The MCD43A4 product is generated from measurements made by Moderate-Resolution

Imaging Spectroradiometer (MODIS) sensors at a spatial resolution of 500 m2. This product consists of seven reflectance bands adjusted by the Bidirectional Reflectance Distribution Function and produced daily, which are a moving average of the contiguous 16 days measurements. Data from eight contiguous pixels corresponding to the polygon of coordinates: 99.93 W, 20.60 N to 99.92 W, 20.61 N, were downloaded. The radiation spectrum (nm) covered by bands one to seven is (b1-b7): 620-670, 841-876, 459-479, 545-565, 1230-1250, 1628-1652 and 2105-2155. Rainfall data were from the 3IMERG version 6 product of the Global Precipitation Measurement Mission of NASA obtained through the Giovanni portal (https://giovanni.gsfc.nasa.gov/giovanni). The P data (mm) was the monthly accumulated for the coordinate 99.92 W, 20.60 N; the spatial resolution of 3IMERG is 10 km2. Through the Giovanni portal, the MODIS MOD11A2 version 6 product of daily surface temperature during the day (LST_d) and night (LST_n) was also obtained.

For MODIS, good quality was determined according to the quality data accompanying the respective products. In the R(20) language, a code was generated to find the measurement dates of the MCD43A4 closest to the measurement date of the FM. Using Qgis v3.16.4(21) andasatelliteimage from Google Maps(22) as aguidingtemplate,avector layer corresponding to the area of irrigation by central pivot was determined; the circle comprised different area of the sampled pixels of the MCD43A4. For each reflectance band, the average corresponding to the vector was obtained using the extract function of the raster package.

Variable generation

The reflectance in the bands b2 and b1 is associated with the ability of vegetation to absorb photosynthetically active light and there are different indices to represent this activity of the vegetation. The normalized vegetation index (NDVI) and the enhanced vegetation index (EVI) were calculated using the spectral bands of the MCD43A4 product:

With the time series of P, the following variables were calculated: the P accumulated in the previous month (P_lag_1), the P accumulated in the previous two months (P_lag_2) and so on until the P accumulated in six previous months: (P_lag_3, P_lag_4, P_lag_5 and P_lag_6). For LST_d and LST_n, the average of the previous month (LST_x_avg_1), of the previous two months (LST_x_avg_2) or of the previous three months (LST_x_avg_3), where x represents the indicative d or n, for day or night, was calculated. These variables represented the prevailing environment before measuring FM.

Modeling

The baseline model for comparison was the linear regression between FM and H_pasture. Four modeling scenarios according to the type of algorithm were explored: ML or OLS and the type of variables available for modeling: using only explanatory variables of RS origin (ML_RS and OLS_RS) or RS variables and those of PM (ML_RS_PM and OLS_RS_PM). The models were trained with 80 % of observations chosen randomly and 20 % were reserved for evaluation. Model evaluation is a black box concept about the relevance of the result of the model(23). The statistical procedures were carried out in the R language, the name of the packages is indicated where relevant. An orthogonal regression (major axis regression) model was fitted between observed values and predicted values using the smatr 3 package, since observed FM values are measured with error(24). The following were calculated: coefficient of determination (r2), root mean square error (RMSE), the Akaike (AIC) and Bayesian (BIC) information criteria, deviance, and bias. These quantitative indicators, as well as graphical evaluation, are techniques commonly used to evaluate mathematical models for predictive purposes(25) .

In the case of OLS, the variance inflation value (VIF) was used to identify multicollinearity using the stepAIC and vif(26) functions; 10.0 was the maximum allowed value of VIF to retain variables in the OLS multiple regression model. The significance level was set at 0.05 for parametric analyses and residual analysis of the OLS regression.

The ML model was generated with the h2o.automl function of the H2O(27) package, it produces a set of models with different algorithm realizations: deep learning (DL), feedforward artificial neural network (NN), general linear models (GLMs), gradientboosting machine (GBM), extreme gradient-boosting (XGBoost), default distributed random forest (DRF) and extremely randomized trees (XRT). Each individual model can be used to predict the response, but also to generate two types of model ensemble: one is from all the algorithms used in the generated models, and the second type of ensemble only considers the best models of each class or family of algorithms; both types of ensembles generally produce better predictions than individual models(23)

The h2o.automl function was run twenty times with the following parameters: a) max_runtime_secs = 500, the maximum runtime before training a final ensemble of models, b) nfolds = 15, number of folds for cross-evaluation (k-folds), c) seed = a random integer value with value between 1 and 50; each of the runs used a randomly chosen seed value, d) nthreads = 50, the number of available processing threads, e) max_mem_size = 100GB, the available RAM in Gigabytes. The approximate runtime was 50 min on an equipment with dual Xeon 2680 v4 processor with 14 cores and double thread each and 128 GB of RAM.

With the h2o.explain function, the importance of the variables in the individual ML models and dependence figures was obtained(27). Deviance was used as a goodness-of-fit statistic to rank the generated models. Machine learning has two elements for supervised

learning: training loss and regularization. The training task attempts to find the best parameters for the model while minimizing the training loss function; this function could be the mean square error or others. The regularization term controls the complexityof the model, helping to reduce overfitting. Overfitting becomes apparent when the model performs accurately during training, but accuracy decreases during the evaluation of the model. A good model needs extensive fitting of parameters by running the algorithm several times to explore the effect on regularization and accuracy of cross-validation(28) .

In this research, the function of training loss was the deviance, which is a likelihood generalization ofthesum ofsquares oftheerror;lowerornegativevalues indicateabetter performance of the model(29)

Results and discussion

The average of FM of the pasture was 2,134 kg DM ha-1 with a seasonal pattern of lower production in winter and higher production in summer (Figure 1a). FM was different among the three years 2,121, 1,770 and 2,392 kg ha-1 for 2008 to 2010 (P<0.05). The rainfall was 636, 382 and 552 mm, respectively. The greatest amount of rainfall was from Julyto September;for2010,Februarywas atypical with 151 mm (Figure1b) andpossibly positivelyimpactingtheFMfromMarchinthatyear.TherainfallrecordedbytheIMERG product in 2008 and 2010 was higher than that recorded by the climatological station closest to the study site; this rainfall estimate was considered accurate because this product has shown good agreement with terrestrial precipitation records(30). The seasonal behavior of the FM suggested an important effect of rainfall, even in the case of this irrigated pasture. April and Maywere the months with the highest average LST_d (Figure 1c). The difference between LST_d and LST_n was greater from April to May (28.5 and 27.3 °C) and lower in July to September (17.3, 16.4 and 15.6 °C); which indicates the site’s extreme characteristic of the climate during the spring. These environmental conditions were also reflected in seasonal changes in pasture management on rest days, forage height, and PAR (Figure 2).

Figure 1: Environmental variables and production of a mixed alfalfa-grass mixed pasture grazed by beef cattle: a) forage mass (FM), b) rainfall (P) and c) diurnal (●) and nocturnal (○) surface temperature (LST)

Figure 2: Management of a mixed alfalfa-grass mixed pasture grazed by beef cattle during 2008 (●), 2009 (○) and 2010 (■): a) Rest days before grazing, b) rate of accumulation of forage (PAR) of the period, c) pasture height

In MA regression, the intercept was numerically close to 0 in the ML_RS_PM scenario and its slope was equal to 1, a model with slope equal to 1 and intercept equal to 0 indicates good fit. The lower value of the RMSE, AIC, BIC and deviance suggested a better representation of the FM with the ML_RS_PM scenario (Table 1). Regarding deviance analysis, the comparison between two or more models will be valid if they fit the same data set, this requirement was not met because the predicted values of FM were inherently different for each model generated. The difference of deviances is distributed approximately as X2 with degrees of freedom equal to the difference in the number of parameters between the models(14), with this difference being 0 for the case of simple linear regression models used to represent the relationship between estimated and predicted values in each modeling scenario. For these two reasons, deviance analysis was not possible; therefore, the selection of the best model was based solely on the numerical value of the goodness-of-fit measures. The worst model was the simple regression between FM and H_pasture, not only according to the goodness-of-fit means but also in the graphical representation of the estimated vs. observed values (Figure 3).

Table 1: Goodness-of-fit measures between observed and estimated FM values resulting from modeling scenarios using algorithms of ordinary least squares (OLS) or machine learning (ML) in combination with explanatory variables related to pasture management (PM) alone or in conjunction with remote sensing variables (PM_RS) OLS_height OLS_RS OLS_RS_PM ML_RS ML_RS_PM

r2 0.40 0.49 0.67 0.70 0.97

RMSE 361.0 341.0 269.0 259.0 78.0

AIC 734.0 724.0 686.0 691.0 542.0

BIC 738.0 728.0 690.0 695.0 546.0 Deviance 8079684.0 6874194.0 4377078.0 4003784.0 363954.0

Bias -3.4 47.1 16.5 -35.1 -1.3

CI 2.5 % -95.9 -39.2 -52.7 -43.5 -21.2

CI 97.5 % 89.0 133.5 85.7 96.4 18.6

MA intercept -1799.0 -2044.0 -594.0 -735.0 27.0

CI 2.5 % -3386.0 -3395.0 -1137.0 -1257.0 62.0

CI 97.5 % -831.0 -1162.0 -162.0 -316.0 112.0

MA slope 1.9 2.0 1.3 1.4 1.0

CI 2.5 % 1.4 1.6 1.1 1.2 0.9

CI 97.5 % 2.6 2.7 1.6 1.6 1.0

r2= coefficient of determination; RMSE= root mean square error; AIC= Akaike information criterion; BIC= Bayesian information criterion; MA= major axis regression; CI= confidence interval.

Figure 3: Evaluation between observed and estimated values of FM using algorithms of ordinary least squares (OLS) or machine learning (ML)

a) OLS, predictor variable forage height; b) OLS_RS scenario; c) OLS_RS_PM scenario; d) ML_RS scenario; e) ML_RS_PM scenario. Coefficient of determination (r2), root mean square error (RMSE), bias and its 95 % confidence interval (CI=IC).

The PAR and H_pasture variables of PM were the most important (Table 2), both in the ML and OLS models; the variable R_time was much less important (Table 2). The most important RS variables were: LST_n, P, P_lag_3 or P_lag_5, LST_d_avg_3 or LST_n_avg_3; indicating the relevance of the environmental conditions of precipitation and temperature not only of the current month, but of the conditions preceding the measurement of the FM. Reflectance (b1 – b7) and vegetation indices were incorporated into ML models, but the stepwise procedure did not choose them for OLS. Compared with PAR and H_pasture, reflectance variables were of low importance in the RS_PM scenarios of ML. Spectral reflectance bands were more important than EVI and NDVI; this finding coincides with the FM study for mixed pastures of temperate climate(15) Although the prediction of fresh biomass in Brachiaria pastures based on the NDVI with r2= 0.73(31) was considered adequate.

Table 2: Important variables included in the scenarios using two possible algorithms: ordinary least squares (OLS) or machine learning (ML) and two types of explanatory variable: only remote sensors (RS) or pasture management variables and RS (RS_PM) Machine learning (ML) Ordinary least squares (OLS)

Remote sensors (RS)

Variable

Remote sensors (RS)_Pasture management (PM) RS RS_PM

LST_d_avg_3 0.081 0.023 0.036 0.027 LST_n_avg_3 0.064 0.017 LST_d 0.036 0.036 LST_n 0.161 0.007 0.060 b1 0.027 0.008 b2 0.034 0.014 b3 0.028 0.003 b4 0.033 0.004 b5 0.044 0.008 b6 0.048 0.010 b7 0.096 0.008 P 0.058 0.008 0.048 P_lag_3 0.099 0.023 P_lag_5 0.270 NDVI 0.001 EVI 0.018 0.001 H_pasture 0.303 0.231 Month_beg_grow 0.006 0.020 R_time 0.101 0.072 PAR 0.417 0.368

For ML models the sum of importance is 1, for OLS models the sum of importance is equal to the r2

The partial dependence that existed between the prediction of FM and the value of some of the most important variables in some ML models is shown in Figure 4, in the ML_RS scenario and in Figure 5 for the ML_RS_PM scenario. The ensembles of ML models had lower deviance compared to some ML algorithms in the two scenarios and were therefore considered better representations of the FM. The partial dependence figures indicate how the explanatory variable influences the predictions of one of the models or ensembles, after standardizing the effect of other variables. For linear regression models (such as the GLM model obtained by ML), the figure is a straight line with slope equal to the parameter of the model(32). FM depended directly and proportionally on the variables PAR, H_pasture and R_time in different models even for a GLM model (pink line), but for variables P_lag_3 and LST_d, the dependence differed between the GLM model and ML models, particularly the DL-type model (dark green line) which was the best individual ML model (Figure 5). The interpretation of the figures is improved with the frequency histogram of the observations, depending on the value of the variable. Where there was less frequency of data, it was interpreted that dependence was not supported by sufficient evidence. An example of this situation was the dependence of LST_n in Figure 4, where the DL-type model has an abrupt ascent, but the last two class intervals of the histogram have few observations.

Figure 4: Partial dependence of FM and: A) monthly average of nocturnal surface temperature (LST_n), B) precipitation accumulated in the previous three months (P_lag_3), C) reflectance band b7 of the MODIS MCD43A4 product, D) monthly average of the diurnal surface temperature in the previous three months (LST_d_avg_3)

The gray bars are the data frequency according to class intervals of the variable. Only models of lower deviance (value in parentheses) obtained by machine learning in the scenario using only variables measured with remote sensors (ML_RS) are shown.

Figure 5: Partial dependence of FM and: A) rate of accumulation of forage (PAR), B) forage height (H_pasture), C) pasture rest days (R_time), D) monthly average of the diurnal surface temperature (LST_d)

The gray bars are the data frequency according to class intervals of the variable. Only models of lower deviance (value in parentheses) obtained by machine learning in the scenario using variables measured with remote sensors and of pasture management (ML_RS_PM) are shown.

The ML_RS_PM scenario included the PAR variable, and this could be a limitation for the practical application of the model. To clarify this aspect, an ML model was built without this variable and using the same training data, resulting in an r2 of 0.76, RMSE of 232.2 and bias of –35.6 (CI –94.4 to 23.1), being better than that obtained in the ML_RS scenario (data not shown). This result has two aspects of importance: other variables available for modeling can replace a variable identified as the most important and second, it is possible to incur into a local optimal solution, even when the ML algorithm explored a solution space with different optimization parameters. A possible alternative would be to increase the number of times the h2o.automl function is run and increase the value of the max_runtime_secs constant.

Despite the coarse spatial resolution of the MODIS and GPM remote sensors (250 m2 and 10 km2), the FM was adequately estimated in the ML_RS scenario (Figure 3d), the r2= 0.70 of this model was within the range recently reported in the literature for ML models that estimate biomass with RS data(14,15) or gross primary productivity(33). A model based on RS data is only attractive for the management of large grasslands. When RS variables

were used in combination with pasture management variables that are easy to measure (H_pasture) or record (R_time and Month_beg_grow), the estimate was very good (Figure 3e); the r2= 0.97 was similar to the r2= 0.96 of biomass estimated from 30 m spatial resolution RS data(5). The prediction of the forage mass obtained with forage height measurements improved when pasture management variables and local meteorological data were incorporated into an ML algorithm of random forest (r2= 0.82), this approach was judged practical for producers, albeit the cost of meteorological instruments(2) .

Despitebeinganirrigated pasture,thepreviousshort-termrainwasimportantinformation for OLS and ML models. In a recent study, it was identified that the spatial-temporal variation of gross primary productivity was not only explained by reflectance bands of the MODIS MCD43A4 but was also related to the vapor pressure deficit(33). Similar to the result of these authors, here it was useful to include other reflectance bands besides b1, b2 and vegetation indices such as NDVI and EVI. From a practical point of view, the model of the ML_RS_PM scenario was considered very feasible to implement as it used routine measurements of the management of the pasture and NASA’s remote sensor data which are publicly accessible.

Animal production under grazing is sustainable when feed consumption that meets nutrient needs is ensured. In grazing management, this depends on adjusting the stoking rate according to the phenological stage of the plant, to the FM before and after grazing andtotheforagethatisdecidedtoleaveas residualpasturemass.Forbeefcattle,adequate FM before grazing can be set at 2,500 kg ha-1 and FM after grazing around 1,200 kg ha-1(10); although these thresholds will depend on the reproductive and physiological stage of the animal, the season of year and different pasture management strategies for feed rationing, phenological control or balance in botanical composition(34). For these reasons, it is important that the predictive model of FM fits well at the extremes of its range and with the exception of the ML_RS_PM scenario, there was an overestimation of the FM when it was less than approximately 1,500 kg ha-1 (Figure 3).

Pasture mass is spatially variable given by differences in soil moisture and fertility, dung deposition, alterations in the plant community by selective grazing and other factors. Forage quadrant cuttings are limited to represent and capture this spatial variability in pastures and therefore the statistical method of sampling is important. Sensors on board unmanned aerial vehicles or drones are an alternative to capture variability in vegetation reflectance on the spatial scale of centimeters, but the cost of multispectral equipment, dataprocessingand operational limitation to cover theterritory (35),in addition to theneed for a calibration function for forage mass, must be considered.

Conclusions and implications

The prediction of FM had lower bias with ML models than with OLS models, especially when remote sensors and pasture management variables were incorporated in the models. ML ensembles had lower deviance compared to some of the individual ML models. The use of RS variables predicted FM similarly to the relationship between H_pasture and FM, althoughthe MLmodel had lowerbias. The models exploredwould haveto be tested in other pasture conditions in order to have a spatial application, be able to represent ecosystems and to value the environmental service of carbon capture. At the local farm scale, these models could be applicable for everyday use in farm feed budgeting or retrospective evaluationoffarmpasture management. In these cases, theresults presented here are promising.

Acknowledgements

The study was the product of the support for sabbatical leave of the first author by the Autonomous University of Querétaro.

Literature cited:

1. Sheath GW, Hay RJM, Giles KH. Managing pastures for grazing animals. In: Nicol, AM, editor. Livestock feeding on pasture. NZ Soc Anim Prod occasional publication. 1987;65–74.

2. Murphy D, O’Brien B, Hennessy D, Hurley M, Murphy M. Evaluation of the precision of the rising plate meter for measuring compressed sward height on heterogeneous grassland swards. Precis Agric 2021;22(3):922–946.

3. Radcliffe J. Cutting techniques for pasture yields on hill country. Proc NZ Grassland Association. 1971;33:91–104.

4. Jáuregui JM, Delbino FG, Bonvini MIB, Berhongaray G. Determining yield of forage crops using the Canopeo mobile phone app. J NZ Grassl 2019;41–46.

5. Marsett RC, Qi J, Heilman P, Biedenbender SH, Watson MC, Amer S, et al. Remote sensing for grassland management in the arid southwest. Rangel Ecol Manag 2006;59(5):530–540.

6. O’Donovan M, Dillon P, Rath M, Stakelum G. A comparison of four methods of herbage mass estimation. Ir J Agric Food Res 2002;17–27.

7. Mills A, Smith M, Moot DJ. Relationships between dry matter yield and height of rotationally grazed dryland lucerne. J NZ Grassl 2016;(78):185–196.

8. MootDJ,YangX,TaHT,BrownHE,TeixeiraEI, SimRE, et al.Simplifiedmethods for on-farm prediction of yield potential of grazed lucerne crops in New Zealand. NZ J Agric Res 2021;65(4-5)1–19.

9. Robertson S. Mass to height relationships in annual pastures and prediction of sheep growth rates. Anim Prod Sci 2014;54(9):1305–1310.

10. Nicol AM, Nicoll GB. Pastures for beef cattle. In: Nicol, AM. editor. Feeding livestock on pasture. Society of Animal Production. Lincoln, New Zealand. 1987;119–131.

11. Zhang Y, Ye A. Would the obtainable gross primary productivity (GPP) products stand up? A critical assessment of 45 global GPP products. Sci Total Environ 2021;783:146965

12. Jiao W, Wang L, McCabe MF. Multi-sensor remote sensing for drought characterization: current status, opportunities and a roadmap for the future. Remote Sens Environ 2021;256:112313.

13. Anav A, Friedlingstein P, Beer C, Ciais P, Harper A, Jones C, et al. Spatiotemporal patterns of terrestrial gross primary production: A review. Rev Geophys 2015;53(3):785–818.

14. Lang M, Mahyou H, Tychon B. Estimation of rangeland production in the arid oriental region (Morocco) combining remote sensing vegetation and rainfall indices: challenges and lessons learned. Remote Sens 2021;13(11):2093.

15. Chen Y, Guerschman J, Shendryk Y, Henry D, Harrison MT. Estimating pasture biomass using sentinel-2 imagery and machine learning. Remote Sens 2021;13(4):603.

16. Hacker R, Smith W. An evaluation of the DDH/100 mm stocking rate index and an alternative approach to stocking rate estimation. Rangel J 2007;29(2):139–148.

17. CICESE C de IC y de ES de E. Base de datos climatológica nacional (CLICOM). [Internet]. Tequisquiapan, Querétaro; 2021. Estación 22025: Consultada 6 Mar, 2021. http://clicom-mex.cicese.mx/.

18. Hodgson J. Grazing management. Science into practice. Longman Group UK Ltd. 1990.

19. Schaaf C, Wang Z. MCD43A4 MODIS/Terra+ Aqua BRDF/Albedo Nadir BRDF AdjustedRefDailyL3Global 500 m V006. NASA EOSDIS LandProcesses DAAC. 2015.

Pecu 2023;14(1):61-77

20. R Development Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing. 2009. https://www.r-project.org

21. QGIS.org. QGIS Geographic Information System. QGIS Association. 2021. http://www.qgis.org

22. Google maps. Mapa satelital. México; 2021.

23. Pasquel D, Roux S, Richetti J, Cammarano D, Tisseyre B, Taylor JA. A review of methods to evaluate crop model performance at multiple and changing spatial scales. Precis Agric 2022;23:1489–1513.

24. Warton DI, Duursma RA, Falster DS, Taskinen S. smatr 3-an R package for estimation and inference about allometric lines. Methods Ecol Evol 2012;3(2):257–259.

25. Tedeschi LO. Assessment of the adequacy of mathematical models. Agric Syst 2006;89(2):225–247.

26. Hall P, Gill N, Kurka M, Phan W, Bartz A. Machine learning interpretability with H2O driverless AI. Bartz A. Editor. California, U.S.: H2O.ai Inc.; 2019.

27. LeDell E, Poirier S. H2o automl: Scalable automatic machine learning. In 2020. https://www.automl.org/wp-ontent/uploads/2020/07/AutoML_2020_paper_61.pdf.

28. Mitchell R, Frank E. Accelerating the XGBoost algorithm using GPU computing. PeerJ Comput Sci 2017;3:e127.

29. McElreath R. Statistical rethinking: A bayesian course with examples in R and STAN. Boca Raton, FL. U.S.: CRC Press; 2020.

30. WangJ,PetersenWA,WolffDB.Validationofsatellite-basedprecipitation products from TRMM to GPM. Remote Sens 2021;13(9):1745.

31. Bretas IL, Valente DS, Silva FF, Chizzotti ML, Paulino MF, D’Áurea AP, et al. Prediction of aboveground biomass and dry‐matter content in Brachiaria pastures by combining meteorological data and satellite imagery. Grass Forage Sci 2021;76(3):340–352.

32. FriedmanJH.Greedyfunctionapproximation:agradientboostingmachine.AnnStat 2001;1189–1232.

33. Joiner J, Yoshida Y. Satellite-based reflectances capture large fraction of variability in global gross primary production (GPP) at weekly time scales. Agric For Meteorol 2020;291:108092.

34. Griffiths W, Dodd M, Kuhn-Sherlock B, Chapman D. Management options to recover perennial ryegrass populations and productivity in run-out pastures. NZGA: Research and Practice Series. 2021;17.

35. Ahmad A, Ordonez J, Cartujo P, Martos V. Remotely piloted aircPARt (RPA) in agriculture: A pursuit of sustainability. Agronomy 2021;11(1):7.

https://doi.org/10.22319/rmcp.v14i1.5394

Article

Thymol and carvacrol determination in a swine feed organic matrix using Headspace

SPME-GC-MS

Fernando Jonathan Lona-Ramírez a

Nancy Lizeth Hernández-López a

Guillermo González-Alatorre a

Teresa del Carmen Flores-Flores a

Rosalba Patiño-Herrera a

José Francisco Louvier-Hernández a*

a Tecnológico Nacional de México en Celaya. Departamento de Ingeniería Química. Guanajuato, México.

*Corresponding author: francisco.louvier@itcelaya.edu.mx

Abstract:

In recent years, oregano essential oil has been used as an animal food additive due to its antifungal and antibacterial properties as well as s synthetic antibiotic substitute. It is desirable to develop fast and effective thymol and carvacrol quantification method in a swine feed organic matrix. In this work, a performance comparison between the Soxhlet solvent extraction technique using petroleum ether and ethyl acetate and the head space-solid phase microextraction (HS-SPME) technique is made A 24 design of experiments is performed for defining HS-SPME parameters: equilibrium temperature of 40 ºC, extraction temperature of 40 ºC, ionic strength of 0.57 M, and extraction time of 40 min. The HS-SPME method is more efficient for extracting thymol and carvacrol extraction from an organic matrix Limits of detection and quantification values using Soxhlet extraction with ethyl acetate were 3.7 and 12.5 μL-1 for thymol and 1.4 and 4.7 μg L-1 for carvacrol, respectively; while LOD and LOQ for HS-SPME were 0.9 and 3.1 μg L-1 for thymol and 0.6 and 1.9 μg L-1 for carvacrol,

respectively. Thehead space-solid phasemicroextraction methodhas thepotential forquality control in the industry for active compounds present in oregano’s essential oil as an additive into an organic matrix.

Key words: Essential oil, Thymol, Carvacrol, Oregano, Origanum, HS-SPME-GC-MS.

Received: 17/08/2020

Accepted: 02/09/2021

Introduction

People use herbal spices as food flavor enhancers and medicinal aids since antiquity(1) , mainly for their biological activity. Oregano is one of the most important herbs, which is the common name from a wide variety of plant genera and species worldwide, but usually referred to as Origanum in the Lamiaceae (Labiatae) family(2) or Lippia graveolens in the Verbenaceae family Oregano’s essential oil (OEO) has been used as a food additive due to its antimicrobial activity attributed to its high monoterpenes content such as thymol and carvacrol, the latter generally recognized as a safe food additive(3-6) Due to the banning of antibiotics by the European Commission, OEO has received increased attention from the poultry and swine industry for improving natural defenses and strengthening animal organisms with favorable results(7-10) OEOcan beincorporated into theswinefeed bymixing the oil and the organic matrix. However, a confident quantification method is required for quality control

Thymol and carvacrol(11) show antibacterial(4,12) , antioxidant(13) , and fungicide activity(3,4) and are two of the main components of the OEO. Thus, they can serve as markers for quantification. The quality control method begins with a solvent extraction of the volatile compounds from the feed matrix; however, organic solvents are neither environmentally friendly nor acceptable for food processing. Some other extraction technologies, such as supercritical carbon dioxide extraction, require high-cost equipment and high-pressure operational conditions(14,15) . Thus, it is desirable to develop quick and effective thymol and carvacrol quantification method inside a swine feed organic matrix. In this paper propose the Head Space Solid Phase Micro Extraction (HS-SPME) technique along with the gas chromatography-mass spectroscopy (GC-MS) method since HS-SPME is an effective, nonexpensive, and environmentally friendly technique for the detection and quantification of volatile compounds(16,17) .

As is known, this technique has not been used for thymol and carvacrol detection in an organic matrix added with OEO, but only to quantify these active compounds in pure oil(18-20) . This work, compare a solvent extraction technique using petroleum ether or ethyl acetate in a Soxhlet extractor(21) with an HS-SPME technique from the swine feed flour matrix to quantifythe thymol and carvacrol of the extract using a GC-MS system. It was used nitrosopiperidine (NPIP) as an internal standard for absorbing signal variations due to the extraction method and the equipment itself(22) and calculate the limit of detection (LOD) and the limit of quantification (LOQ) for assessing the extraction technique’s effectiveness. Finally, it was performed a design of experiments using R(23) and RStudio(24) software.

Material and methods Reagents

Thymol (100.0%), carvacrol (99.9%), nitrosopiperidine (99.9%), and sodium chloride were obtained from Sigma Aldrich (St. Louis, USA). Analytical grade (ACS) petroleum ether and ethyl acetate were obtained from Fermont (Monterrey, México). Tridistilled water from MERCK was used in HS-SPME experiments The carrier gas used for GC-MS was ultrahigh purity (grade 5.0) helium from Praxair. Polyacrylate (PA) fibers for SPME were obtained from Sigma Aldrich. A local industry provided the swine feed flour added with OEO

Chromatographic method

AnAgilentgaschromatographmodel7890Acoupledwithamassspectrometermodel5975C with a positive pole ion, single quadrupole with electron impact ionization (EI) source were used for detection and identification. An HP-INNOWax capillary column (30 m, 0.25 mm ID, and 0.5 μm thickness polyethylene glycol film; Alltech) was used for compound separation. Transfer line temperature was set at 250 ºC and GC injector port at 260 °C on splitless mode. The oven temperature was initially set at 60 °C for 3 min, then raised to 250 ºC at a rate of 20 °C per minute and kept there for 3 min. MS was programmed both on scan and SIM mode, with a solvent delay time of 8 min. Scan mode was set from 20 to 300 m/z while SIM mode was set to 114 m/z (characteristic ions of nitrosopiperidine) for the time interval of 8 to 11.5 min and immediately shifted to follow the 135 and 150 m/z signals (characteristic ion of thymol and carvacrol) until the end of the analysis.

Sample preparation

A local industry provided the swine feed flour samples added with oregano’s essential oil during the manufacturing process. The samples were stored in hermetic plastic bags until

used. Two different techniques for extracting thymol and carvacrol were used in this work: (i) solvent extraction using a Soxhlet distillation apparatus, and (ii) HS-SPME using a PA fiber. Three different solvents were used: ethyl acetate and petroleum ether for Soxhlet extraction and deionized water for the HS-SPME technique. 5.0 mg L-1 of NPIP was added to every sample as an internal standard.

Soxhlet solvent extraction

A sample of 10 g of swine feed was put into a Soxhlet extraction apparatus with 130 mL of solvent (either ethyl acetate or petroleum ether). The mixture was heated until five cycles were completed, then an aliquot of this extract was stored for analysis. A new fresh solvent was immediately added to perform another five cycles using the same sample, and another extract aliquot was taken. A third distillation step with more fresh solvent was made, and a third aliquot was taken. Thus, each sample (10 g of swine feed) was subjected to extraction three times using two different solvents.

Moreover, it was used two quantification methods using (a) an external standard calibration curve and (b) a standard addition method. The calibration curve for the external standard is made using known concentrations for thymol and carvacrol (2, 4, 6, 8, and 10 mg L-1). For the standard addition method, 0.5 mL of extract aliquot was mixed with 0.5 mL of solvent with different thymol and carvacrol concentrations (2, 4, 6, 8, and 10 mg L-1). In all cases, the sample amount injected into the GC was 1.0 μL.

Head space-solid phase microextraction

The HS-SPME technique involves some parameters such as equilibrium time (teq), equilibrium temperature (Teq), extraction time (text), extraction temperature, (Text), and ionic strength (I) The equilibrium time and temperature are the time and temperature at which the sample is left to reach equilibrium between the solid phase (swine feed matrix) and the vial’s headspace. The extraction time and extraction temperature correspond to the time and temperature at which the microfiber is in contact with the headspace adsorbing volatile compounds. Ionicstrengthisameasureoftheconcentrationofionsinasolutionandmodifies the equilibrium of the system. It is necessary to determine the effect of these parameters on the signal obtained in GC-MS. To evaluate this effect, it was added thymol and carvacrol standards in water (along with 5 mg L-1 of NPIP as the internal standard) to form a 10 mg L-1 solution.

For thymol and carvacrol quantification in swine feed, a sample of 0.5 g powder swine feed was added to 15 mL glass vials with PTFE/Silicone septum with the required NaCl content and 10 mL of water with different thymol and carvacrol concentrations to perform the

standard addition technique analysis. External standard calibration curves were not possible to perform in the HS-SPME due to the interactions between volatile compounds of the powder swine feed matrix in the gas phase and the fiber during the extraction. The relative area between thymol or carvacrol and the added standard (NPIP) was calculated and used as the response variable to evaluate the performance of the extraction method.

Design of experiments

A 24 factorial analysis was performed to evaluate the effect of the equilibrium temperature (40–50 ºC), extraction temperature (40–50 ºC), extraction time (20–40 min), and ionic strength (0.57–2.28 mole L-1 of NaCl). Equilibrium time is fixed at a sufficiently long time to assure equilibrium (Table 1).

Table 1: Level values of the factors for the design of the experiment

Factor Low level -1 High level +1

Equilibrium temperature, Teq (ºC) 40 50 Extraction temperature, Text (ºC) 40 50 Extraction time, text (min) 20 40 Ionic strength, I (mole L-1) 0.57 2.28

A 24 factorial design of experiments with a single replicate consists of 16 experimental runs. The analysis of variances of the complete model (main factors and all possible interaction combinations) gives no residuals, Fo, and P-values since the degree of freedom of the error is equal to zero and there is no estimate of the internal error. So, the negligible three- and four-order interactions are used to estimate error. Moreover, after evaluating ANOVA of main effects and two-factor interactions, the significant factors are defined, and another ANOVA analysis is performed taking in account only the factors that are significant. A regressionmodelisthenevaluated,andresidualsandcontourplotsareplottedusingR-studio.

Results

Thymol and carvacrol mass spectrum identification

A sample of 0.5 g of powder swine feed was put in a vial with 10 mL of water and NPIP as theinternalstandard.AnHS-SPMEprocesswasperformedtoidentifythepresenceofthymol and carvacrol, as shown in Figure 1. Spectra from scan mode were analyzed with the NUST/EPA/NIH mass spectral library for confirmation with a 90 % concordance between the experimental and theoretical spectrum. Retention times of internal standard, thymol, and

carvacrol were 10.3, 12.3, and 12.5 min, respectively. The retention time is obtained by following their respective characteristic ion: 144 for NPIP, 135 for thymol, and 150 for carvacrol. It is important to note that the HP-Innowax column was appropriate for good separation between thymol and carvacrol due to its isomeric nature.

Figure 1: Chromatograms of powder swine feed using ethyl acetate and HS-SPME

Calibration curves

For comparison purposes, three different methodologies for thymol and carvacrol quantification were performed: (a) Soxhlet extraction using organic solvents and calibration with an external standard, (b) Soxhlet extraction using organic solvents and calibration by standard addition, and (c) HS-SPME with water as solvent and calibration using standard addition. The use of an external standard and standard addition is intended for sensibility comparison.

Figure 1 shows the comparison of SIM chromatograms using HS-SPME and Soxhlet extraction with ethyl acetate solvent. For the HS-SPME technique, the sensibility increases almost nine times when compared with the solvent extraction technique, even using less sample quantity during the micro-extraction process, which proves the effectiveness d advantage of the HS-SPME methodology.

The obtained signal [relative area = (thymol or carvacrol area) / (internal standard area)] and its relative standard deviation (RSD) for all the experiments of the factorial design is shown in Figure 2. There are seven experiments with an RSD <15 % for both analytes, but only two with an RSD <5.5 %, experiments #1 and #12. A high signal is desirable, so it was identified fourexperiments with ahighrelativearea. Experiments #4, #12,and#16 showacombination

of high signal (relative area) with low dispersion (RSD) values. All these experiments were performed at a salt content of 0.57 M and extraction time of 40 min, but extraction temperature and equilibrium temperature of 40 and 50 ºC. It is interesting to note that ionic strength (salt content) and extraction time are the factors in common, and they are significant factors as will be seen later.

Figure 2: Relative area and dispersions (relative standard deviation) for thymol and carvacrol for each experiment of the factorial design for improving process parameters

In Figure 3 it can be observed the calibration curves for (a) the use of thymol and carvacrol as external standards for the Soxhlet extraction technique; (b) the use of NPIP as added standard for the Soxhlet extraction technique; and (c) the use of NPIP as added standard for HS-SPME technique. It is not possible to use external standards for the SPME technique. It is noteworthy that the signal of the relative area is in the order of tens for carvacrol either with external standard or addition standard, while the signal of relative area for thymol is in the order of units for external and addition standards. But for HS-SPME the signal of the relative area is in the order of hundreds for both thymol and carvacrol, which again confirms the increased sensibility of one order of magnitude (two orders for thymol) of this technique.

Figure 3: Soxhlet extraction external standard calibration curves (a), Soxhlet extraction with standard addition curves (b), and HS-SPME with standard addition curves (c), for carvacrol and thymol quantification

It should be noted that relative area values (y-axis) are greater for the HS-SPME technique compared to the Soxhlet extraction technique. Soxhlet extraction using ethyl acetate and HS-SPME using water as solvents.

Discussion

Design of experiments

Table 2 shows the analysis of variance for main effects and two-factor interactions for carvacrol and thymol quantification, and it is possible to observe that only extraction time and salt content as well as the interaction between them are significant for both analytes. Table 3 shows the analysis of variance considering only ionic strength, extraction time, and ionic strength–extraction time interaction factors for carvacrol and thymol. Considering only the significant factors and interactions, the regression model for the carvacrol HS-SPME extraction is: �������������������������������� =13.9782+9.0768��+1.3297�������� 0.6242��×��������

With an R2 of 0.8558, meaning that 85.6 % of the data variability is explained by the model with a randomly distributed residuals plot (not shown). A contour plot in Figure 4, shows that when extraction time is at a high level, there is a strong negative effect of salt content, meaning that the relative area of carvacrol is higher when salt content is lower; moreover, when extraction time is at a low level, there is a still negative effect of salt content but weaker than at the high level of extraction time. Considering only the significant factors and interactions, the regression model for the thymol HS-SPME extraction is: ��ℎ���������������������� = 95790+55841��+ 08329��������−04008��×��������

With an R2 of 0.8401, meaning that 84 % of the data variability is explained by the model with a randomly distributed residuals plot (not shown). A contour plot in Figure 5, shows that when extraction time is at a high level, there is a strong negative effect of salt content, meaning that the relative area of thymol is higher when salt content is lower; moreover, when extraction time is at a low level, there is a still negative effect of salt content but weaker than at the high level of extraction time. This is the same for thymol and carvacrol, the only difference is that carvacrol shows a 1.5 times higher relative area signal than thymol.

Table 2: ANOVA of main effects and two-factor interactions for the design of the experiment

Carvacrol

DF

Thymol

Sum of squares Mean squared F0 P-value DF

Sum of squares Mean squared F0 P-value

Teq (ºC) 1 0.6 0.6 0.016 0.90550 1 0.2 0.2 0.011 0.92026

Text (ºC) 1 1.4 1.4 0.038 0.85218 1 0.7 0.7 0.042 0.84494

I (M) 1 1089.2 1089.2 29.208 0.00293 1 484.9 484.9 30.822 0.00261

text (min) 1 310.1 310.1 8.315 0.03444 1 109.7 109.7 6.974 0.04593

Teq x Text 1 39.0 39.0 1.046 0.35339 1 28.3 28.3 1.798 0.23760

Teq x I 1 13.7 13.7 0.368 0.57070 1 6.6 6.6 0.416 0.54715

Teq x text 1 3.6 3.6 0.096 0.76966 1 0.1 0.1 0.006 0.94100

Text x I 1 2.8 2.8 0.075 0.79541 1 1.4 1.4 0.089 0.77805

Text x text 1 65.0 65.0 1.742 0.24407 1 33.1 33.1 2.105 0.20652

text x I 1 455.8 455.8 12.222 0.01736 1 187.9 187.9 11.941 0.01813

Residuals 5 186.5 37.3 5 78.7 15.7

DF= degrees of freedom.

Table3 showstheanalysis ofvariance evaluatedonlywith thesignificant factorsoftheDOE, i.e., ionic strength, extraction time, and ionic strength-extraction time interaction factors.

Table 3: ANOVA of the significant factors for the for the design of the experiment

Carvacrol Thymol

DF Sum of squares Mean squared F0 P-value DF

Sum of squares Mean squared F0 P-value

I (M) 1 1089.2 1089.2 41.83 <0.001 1 484.9 484.9 39.065 <0.001 text (min) 1 310.1 310.1 11.91 0.00480 1 109.7 109.7 8.839 0.01163 text x I 1 455.8 455.8 17.50 0.00127 1 187.9 187.9 15.134 0.00215

Residuals 12 312.5 26.0 12 148.9 12.4

DF= degrees of freedom.

For choosing the best operational parameters, should be always select a high extraction time and low salt content. Since the other two factors are not significant, it can work at any equilibrium temperature and extraction temperature; so, was decided to work at low equilibrium and extraction temperatures for economics.

Table 4: Table of effects for the design of the experiment

Carvacrol

Thymol

Effect T-value P-value Effect T-value P-value

Ionic strength, M -16.502 -6.467 <0.001 -11.0102 -6.250 <0.001 Extraction time, min 8.804 3.451 0.00480 5.2372 2.973 0.01163 text x I -10.674 -4.183 0.00127 -6.853 -3.890 0.00215

For both analytes, an increase in salt content (or ionic strength) results in a decrement in the relative area, probablybecause the solution is close to saturation. Also, a high extraction time (text) benefits the relative area signal which is in good agreement with the concept that extraction increases as extraction time increments.

Table 5 shows thymol and carvacrol content in the powder swine feed when comparing the two extraction techniques and the solvents used. The total thymol and carvacrol content for each solvent is calculated by adding the measured quantity of each of the three consecutive extractions. Regarding Soxhlet solvent extraction, petroleum ether was not efficient in extracting thymol and carvacrol from the powder swine feed, as indicated by the low quantification obtained of both components, but especially for thymol. Petroleum ether extracted 52 to 55 % less thymol and 19 to 22 % less carvacrol than ethyl acetate Interestingly, there is a selective extraction capability of both solvents for thymol over carvacrol Again, the HS-SPME technique shows an improved extraction capacity for both thymol and carvacrol, and there are extracted with no selectivity.

Table 5: Comparison of the total thymol and carvacrol content in powder swine feed measured by different extraction techniques

Technique Soxhlet extraction HS-SPME

Standard method Standard addition External standard Standard addition Solvent Ethyl acetate Petroleum ether Ethyl acetate Petroleum ether water Analyte Thymol Carvacrol Thymol Carvacrol Thymol Carvacrol Thymol Carvacrol Thymol Carvacrol

First extract, mg L-1 4.25 0.67 1.99 0.48 4.03 0.58 1.92 0.40 - -

Second extract, mg L-1 0.63 0.16 0.33 0.18 0.79 0.15 0.24 0.16 - -

Total content, mg L-1 4.88 0.82 2.32 0.67 4.82 0.73 2.16 0.56 3.25 4.17

Content in swine feed, mg kg-1 63.45 10.71 30.22 8.66 62.60 9.43 28.12 7.30 65.00 83.40

The calibration method also shows some differences. Calibration with an external standard shows a concentration value that is 9 % less on average than that using standard addition

calibration; the standard addition method has an improved uncertainty, but it is an expensive method since the standard must be added to each sample.

Figure 4: Contour plot for ionic strength and extraction time for carvacrol extraction using HS-SPME

TheHS-SPME technique shows thehighest concentrationforthymol andcarvacrol.Thetotal thymol content agrees with the total thymol content obtained bySoxhlet extraction with ethyl acetate of about 63-65 mg/kg; however, carvacrol’s total content is very different. Carvacrol quantification by HS-SPME has a value of 83.40 mg/kg, while quantification using Soxhlet extraction with ethyl acetate is 10.71 mg/kg, which is eight times lower than the HS-SPME result. It might be related to the steric behavior of thymol and carvacrol (stereoisomers) and interactions with the fiber material (polyacrylate).

Figure 5: Contour plot for ionic strength and extraction time for thymol extraction using HS-SPME

Method validation

Limits of detection (LOD) and quantification (LOQ) were estimated to evaluate the performance of the extraction methods and were calculated using the baseline noise and the signal, defined as three times the relation signal/noise for LOD and ten times for the LOQ (16). Figure 3 shows the calibration curves for each component (thymol and carvacrol) for the three different solvents used LOD and LOQ values using Soxhlet extraction with ethyl acetate were 3.7 and 12.5 μL-1 for thymol and 1.4 and 4.7 μg L-1 for carvacrol, respectively TheHS-SPME techniquegivesbetterresults forboth substances since LOD, and LOQvalues were 0.9 and 3.1 μg L-1 for thymol and 0.6 and 1.9 μg L-1 for carvacrol, respectively The linearity of data was estimated via the linear correlation coefficient, where the lowest value found was 0.9892.

Conclusions and implications

Oregano’s essential oil is positively identified in the swine feed powder using characteristic volatile compounds thymol and carvacrol using two different extraction methods: Soxhlet and HS-SPME. Among the organic solvents for Soxhlet extraction, petroleum ether was not suitable since it only extracted about 50 and 10 % of the total thymol and carvacrol content, respectively (relative to HS-SPME quantification) Furthermore, regarding the use of ethyl acetate in Soxhlet extraction, this solvent was able to extract all the thymol, but not the carvacrol, showing some sort of selectivity. For the HS-SPME technique, a 24-factorial design of experiments was performed to evaluate process parameters and obtain the highest possible S/N ratio. The proper conditions are equilibrium temperature (Teq) of 40 ºC, extraction temperature (Text) of 40 ºC, ionic strength (I) of 0.57 M, and extraction time (text) of 40 min. HS-SPME showed a nine-times better extraction performance compared to Soxhlet extraction, even with smaller sample amounts, with a limit of detection and quantification of 0.9 and 3.1 μg L-1 for thymol, and 0.6 and 1.9 μg L-1 for carvacrol, respectively. The results show that the HS-SPME method is more efficient for thymol and carvacrol extraction from an organic matrix and has the potential for a quality-control technique in the food industry to quantify the active compounds of oregano’s essential oil when used as an additive to an organic matrix such as swine feed.

Acknowledgments

To CONACYT (The National Council of Science and Technology) for the financial support awarded to doctoral student Fernando Jonathan Lona Ramírez (grant Number: 344837) and Tecnológico Nacional de México (TecNM) for research grant number 5267.14-P to carry out this study. To Alimentos Aicansa SA for providing swine feed samples with OEO as an additive

Conflict of interest

The authors declare no conflict of interest.

Literature cited:

1. Sánchez-Ruíz JF, Tejeda-Rosales ME, Sánchez-Tejeda JF, Sánchez-Tejeda MG. Pharmacy, medicine and herbolary in the Florentine Codex. Rev Mex Cienc Farm 2012;43(3):55-66.

2. Kintzios SE. Oregano. In: Peter KV editor. Handbook of herbs and spices Vol. 2. Cambridge, UK: Woodhead; 2004:215-229.

3. Ahmad A, Khan A, Akhtar F, Yousuf S, Xess I, Khan LA, Manzoor N. Fungicidal activity of thymol and carvacrol by disrupting ergosterol biosynthesis and membrane integrity against Candida. Eur J Clin Microbiol Infect Dis 2011;30(1):41-50.

4. LioliosCC,GortziO,LalasS,TsaknisJ,Chinou I. Liposomalincorporationofcarvacrol and thymol isolated from the essential oil of Origanum dictamnus L. and in vitro antimicrobial activity. Food Chem 2009;112(1):77-83.

5. MarcheseA, Orhan IE., DagliaM, BarbieriR, Di LorenzoA,Nabavi SF,Gortzi O, Izadi M, Nabavi SM. Antibacterial and antifungal activities of thymol: A brief review of the literature. Food Chem 2016;210:402-414.

6. Pesavento G, Calonico C, Bilia AR, Barnabei M, Calesini F, Addona R, et al. Antibacterial activity of Oregano, Rosma-rinus and Thymus essential oils against Staphylococcus aureus and Listeria monocytogenes in beef meatballs. Food Control 2015;54:188-199.

7. Bozkurt M, Bintaş E, Kırkan Ş, Akşit H, Küçükyılmaz K, Erbaş G, et al. Comparative evaluation of dietary supplementation with mannan oligosaccharide and oregano essential oil in forced molted and fully fed laying hens between 82 and 106 weeks of age. Poult Sci 2016;95(11):2576-2591.

8. Franciosini MP, Casagrande-Proietti P, Forte C, Beghelli D, Acuti G, Zanichelli D, et al. Effects of oregano (Origanum vulgare L.) and rosemary (Rosmarinus officinalis L.) aqueous extracts on broiler performance, immune function and intestinal microbial population. J Appl Anim Res 2016;44(1):474-479.

9. Scocco P, Forte C, Franciosini MP, Mercati F, Casagrande-Proietti P, Dall’Aglio C, et al. Gut complex carbohydrates and intestinal microflora in broiler chickens fed with oregano (Origanum vulgare L.) aqueous extract and vitamin E. J Anim Physiol Anim Nutr (Berl) 2017;101(4):676-684.

2023;14(1):78-93

10. Zeng Z, Zhang S, Wang H, Piao X. Essential oil and aromatic plants as feed additives in non-ruminant nutrition: a review. J Anim Sci Biotechnol 2015;6:7.

11. Russo M, Galletti GC, Bocchini P, Carnacini A. Essential oil chemical composition of wild populations of italian oregano spice (Origanum vulgare ssp. hirtum (Link) Ietswaart):  A preliminaryevaluation of their use in chemotaxonomyby cluster analysis. 1. Inflorescences. J Agric Food Chem 1998;46(9):3741-3746.

12. de Oliveira Nóbrega R, de Castro Teixeira, AP, Araújo de Oliveira W, de Oliveira Lima E, Oliveira Lima I. Investigation of the antifungal activity of carvacrol against strains of Cryptococcus neoformans. Pharm Biol 2016;54(11):2591-2596.

13. Safaei-Ghomi J, Ebrahimabadi AH, Djafari-Bidgoli Z, Batooli H. GC/MS analysis and in vitro antioxidant activity of essential oil and methanol extracts of Thymus caramanicus Jalas and its main constituent carvacrol. Food Chem 2009;115(4):15241528.

14. Díaz-Maroto MC, Pérez-Coello MS, Cabezudo MD. Supercritical carbon dioxide extraction of volatiles from spices: Comparison with simultaneous distillation–extraction. J Chromatogr A 2001;947(1):23-29.

15. Hossain MB, Barry-Ryan C, Martin-Diana AB, Brunton NP. Optimisation of accelerated solvent extraction of antioxidant compounds from rosemary (Rosmarinus officinalis L.), marjoram (Origanum majorana L.) and oregano (Origanum vulgare L.) using response surface methodology. Food Chem 2011;126(1):339-346.

16. Lona-Ramirez FJ, Gonzalez-Alatorre G, Rico-Ramírez V, Perez-Perez, MCI, CastrejónGonzález EO. Gas chromatography/mass spectrometry for the determination of nitrosamines in red wine. Food Chem 2016;196:1131-1136.

17. Méndez-Pérez D, González Alatorre G, Botello Álvarez E, Escamilla Silva E, Alvarado JFJ. Solid-phase microextraction of N-nitrosodimethylamine in beer. Food Chem 2008; 107(3):1348-1352.

18. Adams A, Kruma Z, Verhé R, De Kimpe N, Kreicbergs V. Volatile profiles of rapeseed oil flavored with basil, oregano, and thyme as a function of flavoring conditions. J Am Oil Chem Soc 2011;88(2):201-212.

19. Karami-Osboo R, Miri R, Asadollahi M, Jassbi AR. Comparison between head-space spmeandhydrodistillation-gc-msofthevolatilesof Thymus daenensis.JEssentOilBear Pl 2015;18(4):925-930.

Rev Mex Cienc Pecu 2023;14(1):78-93

20. Stashenko EE, Martínez JR. Sampling volatile compounds from natural products with headspace/solid-phase micro-extraction. J Biochem Biophys Methods 2007;70(2):235242.

21. Pothier J, Galand N, El Ouali M, Viel C. Comparison of planar chromatographic methods (TLC, OPLC, AMD) applied to essential oils of wild thyme and seven chemotypes of thyme. II Farmaco 2001;56(5-7):505-511.

22. Pawliszyn J. Solid phasemicroextraction: Theoryandpractice.NewYork, USA: WileyVCH; 1997.

23. R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing, 2021. http://www.R-project.org/

24. RStudio Team. RStudio: Integrated Development Environment for R. Boston, Massachusetts, USA: RStudio. PBC; 2020. http://www.rstudio.com/

https://doi.org/10.22319/rmcp.v14i1.6121 Article

Changes in the count of four bacterial groups during the ripening of Prensa (Costeño) Cheese from Cuajinicuilapa,

Mexico

José Alberto Mendoza-Cuevas a

Armando Santos-Moreno a

Beatriz Teresa Rosas-Barbosa b

Ma. Carmen Ybarra-Moncada a Emmanuel Flores-Girón a Diana Guerra-Ramírez c*

a Universidad Autónoma Chapingo. Departamento de Ingeniería Agroindustrial. Carretera México-Texcoco km 38.5, Texcoco, Estado de México. México.

b Universidad de Guadalajara. Centro Universitario de Ciencias Biológicas y Agropecuarias, Zapopan, Jal. México.

c Universidad Autónoma Chapingo. Departamento de Preparatoria Agrícola. Texcoco, Estado de México. México.

* Corresponding author: guerrard@correo.chapingo.mx

Abstract:

Prensa Cheese, also called Costeño, is made in an artisanal way from raw cow’s milk in the Costa Chica region of the state of Guerrero. In order to know the characteristics of Mexican artisanal cheeses, the objective of this research was to analyze the changes in the count of aerobicmesophilicbacteria(AMB),totalcoliform(TCs)microorganisms,lacticacidbacteria (LAB) and coagulase-positive staphylococci (CPS), during the ripening (5, 30, 60 and 90 d) of Prensa cheeses, made by four different cheese factories (A, B, C and D) of Cuajinicuilapa, Guerrero, Mexico. A portion (25 g) of each cheese sample was homogenized with peptone

diluent (225 mL) and dilutions from 10-1 to 10-6 were prepared with which 3MTM PetrifilmTM plates were sown. After incubating under different conditions, depending on the type of microorganism, AMB, TCs, LAB and CPS counts were made. The results showed that as the ripening time of the Prensa Cheese progressed, the microbial load decreased: AMB from 4 to 2, TCs from 6 to 3, LAB from 6 to 2 and CPS from 5 to 2 log10 CFU g-1. The changes in the counts of the bacterial groups studied can be attributed to the physicochemical and microbiological transitions typical of cheese maturation and to the characteristics of the microbiota present in each of the cheese factories. The results of this research provide elements for the microbial characterization of Mexican artisanal cheeses.

Key words: Lactic acid bacteria, Aerobic mesophilic bacteria, Coagulase-positive staphylococci, Raw milk, Microbiota, Total coliform microorganisms, Artisanal cheeses.

Received: 14/12/2021

Accepted: 02/09/2022

Introduction

Around the world, cheese, in addition to being a rich source of nutrients, is an essential food used in the local gastronomy of different societies(1,2). Currently, around 1,833 varieties of cheese located in 74 countries are known(3); this diversity is determined by the technological processes used for its preparation, such as the origin of milk, fat-protein ratio, types of cultures and coagulating agents; the shape, size of the cheese and the maturation conditions(4,5,6) .

Cheese ripening consists of its storage, under certain conditions of temperature and moisture, for a period of time that can range from 3 to 7 d, up to 2 yr(5,7). The maturation process, in addition to providing sensory characteristics, is a method of conservation(5,6,8) . In this stage, biotic and abiotic changes that have a direct impact on the microbiota present in the cheese occur(5,7,9) .

Most artisanal cheeses are made from raw milk, with spontaneous fermentation, nontechnified preparation processes and varied maturation times(5,7,10) .

In Mexico there are about 40 artisanal cheeses(7), among them are matured cheeses such as Cotija from the Sierra JalMich, Añejo cheese from Zacazonapan, Maduro cheese from Veracruz, Chihuahua cheese and artisanal cheese from the Ojos Negros region of Baja

California, of which some aspects of their microbiology have been published(11-14) . Prensa cheese (PC) is made with unpasteurized cow’s milk, commercial liquid rennet and salt; it goes through a pressing stage whose duration varies at the discretion of the manufacturer from 1 to 3 days, then it is left to mature for periods of up to three months. The cheese thus obtained has color variations between white and yellow(15) . It is generally rectangular or circular in shape, its consistency is firm, and its weight is 1 to 14 kg per piece (Figure 1)(15)

Figure 1: Prensa cheese in rectangular and cylindrical shapes

PC has been produced for more than 100 years in southwestern Mexico, mainly in the Costa Chica region of the state of Guerrero in the municipalities of Cuajinicuilapa and Ometepec, as well as in the municipality of Pinotepa Nacional, on the coast of the state of Oaxaca(15)

According to INEGI(16), the climate of the Costa Chica region is warm subhumid, and its temperatures range from 22 to 28 °C.

Studies are currentlybeingcarried out to identifythe characteristics of artisanal cheeses from Mexico(7,15), the objective of this research was to analyze the changes that occur in the count of aerobic mesophilic bacteria, total coliform microorganisms, lactic acid bacteria and coagulase-positive staphylococci, during the ripening (5, 30, 60 and 90 d) of prensa cheeses made by four cheese factories (A, B, C and D) of Cuajinicuilapa, Guerrero, Mexico.

Material and methods

Cheese samples

Samples of PC made in an artisanal way in the municipality of Cuajinicuilapa, Guerrero, Mexico (16°28′18′ N, 99°24′55′ W), were analyzed in July 2018. Based on a targeted

sampling, four cheese factories were selected as sampling units, which will henceforth be named A, B, C and D. Four freshly made cheeses, weighing 1 kg, were purchased from each cheese factory. The samples were moved to the municipality of San Marcos, Guerrero (16°47′46′ N, 99°23′05′ W) to a space with characteristics similar to those of the cheese factories of Cuajinicuilapa. In this place, the samples of the cheeses from the cheese factories A, B, C and D (four of each cheese factory) were left to mature for 5, 30, 60 and 90 d. After the ripening time, each batch was transferred in polyethylene bags, inside coolers with refrigerant, to the laboratory. The samples were kept in refrigeration at 4 °C until analysis. The maximum ripening time was 90 d because after that period the flavors intensify, and local consumers avoid it because they prefer softer flavors.

Sample preparation

Each of the cheese samples (25 g) was mixed with 225 mL of peptone diluent, the mixture was homogenized for 2 min (VWR® symphony D S41 Vortex, VWR International) and dilutions from 10-1 to 10-6 were made by transferring 1 mL of the sample to vials containing 9 mL of peptone diluent(17) .

Microorganism count

The following culture media (3M PetrifilmTM plates) were used: aerobe count (AC No. of catalog 6400), coliform count (TC No. of catalog 6410), lactic acid bacteria (No. of catalog 6461) and staph express for coagulase-positive staphylococci (No. of catalog 6493); 1 mL of the corresponding dilution was placed in each of the plates(18-21) .

All counts were done in duplicate. For AMB, dilutions 10-3 and 10-4 were sown and the medium was incubated at 35 ± 2 °C for 48 ± 3 h(18). TC microorganisms were studied based on dilutions 10-2 and 10-3, being incubated at 35 ± 1 °C for 24 ± 2 h(19). The determination of LAB was made by inoculating the media with dilutions from 10-3 to 10-6 and incubating at 35 ± 2 °C for 48 ± 3 h(20). The CPS study was conducted from dilutions 10-2 to 10-4 and an incubation at 37 ± 1 °C for 24 ± 3 h(21) .

Once the incubation time was completed, the growth was reviewed and the plates containing between 15 and 300 colonies were counted, the mean of the two repetitions was obtained and this average was multiplied by the inverse of the dilution with which the plate was inoculated(22). The result of the count was reported as log10 of the number of colony-forming units per gram (log10 CFU g-1)(23)

Statistical analysis

The statistical analysis was based on a design with repeated means and completely random distribution of treatments, tested over time, whose probabilistic model corresponds to: �������� =��+���� +���� +(����)���� +�������� (24)

Where: ��+���� +���� +(����)���� is the mean of treatment i at time k, which contains the effects of treatment, time, and the time × treatment interaction; �������� is the random error associated with the measurement at time k on j assigned to treatment i

The effect of the treatments (cheese factories A, B, C and D) was evaluated through the ripening time (5, 30, 60 and 90 d) with four repetitions, generating 64 experimental units, each consisting of 25 g of cheese.

The response variables evaluated were: total count of aerobic mesophilic bacteria (AMB), totalcoliforms(TCs),lacticacidbacteria(LAB)andcoagulase-positivestaphylococci(CPS). The data were analyzed using a mixed model(24,25) whose random effect corresponds to the maturation time and the fixed effect to the cheese factories. The Tukey-Kramer method (P<0.05) was applied to identifythe effect of the treatments. Analyses were performed in the SAS package version 9.1 (SAS Institute, Inc., Cary, NC, USA).

Results and discussion

During the time of ripening, a decrease in the count of the different microorganisms was observed. The highest counts of all microbial groups were reached at 5 d of ripenig while the minimum values occurred at 90 d (Table 1).

Rev Mex Cienc Pecu 2023;14(1):94-109

Table 1: Count of bacterial groups, log10 CFU g-1, during the ripening of Prensa cheese made in four cheese factories (A, B, C and D) of Cuajinicuilapa, Guerrero, Mexico

Bacterial

group Time (days)

Aerobic mesophilic bacteria (log10 CFU g-1)

Total coliforms (log10 CFU g-1)

Lactic acid bacteria (log10 CFU g-1)

Cheese factories A B C D

5 6.3143 ±0.2973 Cc 4.6802 ±0. 2973 Ba 5.8948 ±0.2973 Cb 5.9077 ±0.2973 Cb 30 5.1156 ±0.2973 Bb 4.3884 ±0. 2973 Ba 4.4981 ±0.2973 Ba 5.1043 ±0.2973 Bb 60 4.0021 ±0.2973 Ab 3.1505 ±0. 2973 Aa 4.4047±0.2973 Bbc 4.6703±0.2973 ABc 90 4.3178 ±0.2973 Ab 2.5638 ±0. 2973 Aa 2.4005 ±0.2973 Aa 4.3597 ±0.2973 Ab

5 4.1093 ±0.1002 Cb 2.1505 ±0.1002 Aa 2.4203 ±0.1002 Aa 4.7211 ±0.1002 Db 30 3.7726 ±0.0541 Cc 2.8838 ±0.0541 Ba 3.7916 ±0.0541 Cc 3.0878 ±0.0541 Cb 60 2.3138 ±0.2854 Bb 2.3763 ±0.2854 ABb 2.2698 ±0.2854 Ab 1.5753 ±0.2854 Ba 90 <1 ±0.0440 Aa 2.3451 ±0.0440 Ab 2.0753 ±0.0440 Ab <1 ±0.0440 Aa

5 6.5457 ±0.0651 Dc 5.9454 ±0.0651 Cb 5.5084 ±0.0651 Ba 6.7366 ±0.0651 Cc 30 5.8336 ±0.0651 Cb 5.8231 ±0.0651 Cb 5.1193 ±0.0651 Ba 6.1241 ±0.0651 Cb 60 3.5524 ±0.0651 Ba 3.7918 ±0.0651 Ba 3.1945 ±0.0651 Aa 5.0914 ±0.0651 Bb 90 <1 ±0.0651 Aa <1 ±0.0651 Aa 3.2258 ±0.0651 Ab 4.2394 ±0.0651 Ac

Coagulase-positive staphylococci (log10 CFU g-1)

5 5.6562 ±0.0540 Cb 3.7456 ±0.0540 Ba 5.6918 ±0.0540 Cb 5.8746 ±0.0540 Cb 30 3.6276 ±0.3811 Bb 2.2500 ±0.3811 Aa 3.7271 ±0.3811 Bb 3.8389 ±0.3811 Bb 60 2.6945 ±0.0438 Aa 2.7143 ±0.0438 Aa 2.7311 ±0.0438 Aa 2.8063 ±0.0438 Aa 90 2.5951 ±0.0524 Aa 2.4203 ±0.0524 Aa 2.6945 ±0.0524 Aa 2.5951 ±0.0524 Aa

Means with lowercase letter in rows and means with uppercase letter in columns, followed by different letter, indicate statistical significance (Tukey, P<0.05).

Aerobic mesophilic bacteria

In the AMB counts, there were significant differences (P<0.05) between the cheese factories (Table 1). The gradual decrease in AMB from day 5 to 90 was close to 2 log10 CFU g-1, for cheese factories A, B and D; while for cheese factory C, it was 3.49 log10 CFU g-1 .

Most of the AMB values found in the PC are within the range of 4 to 9 logarithms CFU g-1 , reported for cheeses made from raw milk and matured for 60 or more days(26). Since the maturation process of cheeses involves the multiplication of the microorganisms present, AMB concentrations of 4 to 9 logarithms (104 to 109 CFU g-1) are expected in this type of products, without this implying a deterioration of the food or suggesting that non-sanitary conditions occurred during its preparation or storage(26,27,28). In the region of Ojos Negros in the state of Baja California, based on a study that included matured cheeses from 22 cheese factories, made with raw milk, it is reported that AMB were found in a range of 4.6 to 7.2 log10 CFU g-1(14) .

In studies on the ripening of artisanal Cotija cheese, salted and matured at temperatures of 14 °C to 32 °C, Chombo(11) reports the following variations in AMB: 8.3, 7.0, 3.5 and 4.7 log10 CFU g-1, on d 8, 30, 60 and 90, while Magallón(29) found 5.3 and 1.8 log10 on d 30 and 90, respectively. The counts found in the PC at 30 and 90 d are very close to those reported for Cotija cheese that is ripened in temperature ranges similar to those of PC.

The decrease in AMB was common in the cheeses from the four cheese factories (Table 1), reflecting a certain homogeneity in the preparation processes and the ripening conditions of the cheeses. The statistical difference (P<0.05) between cheese factories suggests quantitative or qualitative variations in the microbiota of cheese generated by milk and the microenvironments of each cheese factory. It should be noted that, between d 5 and 60, it is observed that the AMB of cheeses from cheese factory A descend 2.31 log10 CFU g-1 , however, between d 60 and 90 they remain unchanged; while the AMB of cheeses from cheese factory C, from d 60 to 90, show a reduction of 2 log10 CFU g-1 .

The development of AMB in cheeses from cheese factory A shows the ability of bacteria to adapt and survive, while the decrease in AMB in the cheeses from cheese factory C exhibits a loss of viability with the release of enzymes that contribute to the generation of flavors and textures(5). This suggests that it is appropriate to study the relationship of the organoleptic characteristics of prensa cheese, between cheese factories and between different concentrations of AMB over maturation time, as well as the relationship of AMB with the shelf life of cheese at room temperature. On the other hand, AMB studies are useful as an initial stage in the search for starter cultures from artisanal cheeses. Variations in salt concentration and moisture may have influenced the survival of AMB. The results suggest

that, in cheeses from cheese factories A and D, there are bacteria adapted for long-term survival. While in cheeses from cheese factories B and C, bacteria of short survival may be present, and therefore can contribute more quickly to the production of tastes, smells and textures (pleasant or unpleasant)(5). Another possibility is that the reduction of AMB in cheeses from cheese factories B and C is due to the presence of substances with antimicrobial action, caused by the metabolism of microorganisms, or by the biochemical changes that occur in the cheese, derived from the proteolysis of casein to give rise to peptides with antimicrobial activity(5,30)

Total coliform bacteria

Total coliform counts in the cheeses showed significant differences (P<0.05) between cheese factories during the ripening period (Table 1). The highest counts were found on d 5 (4.72 log10 CFU g-1) and the lowest on d 90 (<1 log10). Cheese, being a solid sample, hinders a direct count, so it is necessaryto make an initial dilution that leads to the minimum detection level being 10 CFU g-1, so the absence of growth was reported as < 1 log10.

Two different dynamics were observed, cheeses from cheese factories A and D had the highest initial TC loads, which decreased at 30 d of maturation to reach 3.72 to 3.10 log10 CFU g-1, respectively (Table 1). In cheeses from cheese factories B and C, the TCs showed initial levels of 2 logarithms, which rose during the first month and remained with slight variations to coincide on d 90 with values very close to each other.

The dynamics of TCs in cheeses from cheese factories A and D have been reported in semihard cheeses and are characterized by a progressive decrease in coliforms as ripened progresses(31), which is attributed to the decrease in pH due to the fermentation of lactose(32) In ripened cheeses such as Cheddar, coliforms die at a rate of 0.3 log10 CFU g-1 per week and in Gouda cheese at 0.7 log10 per week(33). Therefore, the dynamics observed in cheeses from cheese factories B and C is atypical, because between d 5 to 30, there is an increase of 0.73 and 1.37 log10, respectively, followed on d 60 to 90 by very small decreases, 0.03 and 0.19 log10, respectively (Table 1). This suggests that there are common aspects between the two cheese factories that favor the selection of bacteria that persist during ripening, such as water and milk quality, personnel or variations in the preparation or cleaning processes.

The accepted levels of total coliforms in matured cheeses are less than 100 CFU/g (< 2 log10 CFU g-1)(34), values that were reached between d 60 and 90 (Table 1) in cheese factories A and D. According to Metz(32), cheeses made with good quality raw milk, under hygienic sanitary conditions, applying good manufacturing practices and properly ripened, will have

low levels of total coliform bacteria, fecal coliforms, enterococci, Enterobacteriaceae and Escherichia coli.

The activity of coliform organisms in cheeses can adversely affect their sensory characteristics(28,35). However, it has been observed that certain genera of coliforms contribute positively to the texture and sensory characteristics of the product; in addition to the fact that some strains of Hafnia contribute to the accumulation of aromas, and generation of flavors(32,35,36). The persistence of coliforms suggests the possibility of the participation of these microorganisms in the organoleptic characteristics of cheeses from cheese factories B and C, an aspect that has not been addressed in studies of Mexican cheeses.

Lactic acid bacteria

During the ripening period, the LAB counts of the cheeses from the cheese factories studied showed significant differences (P<0.05) (Table 1). This microbial group had the highest counts of the entire study. On d 5 the counts ranged between 5.50 and 6.73 log10 CFU g-1 , these values decreased from day 30. From day 5 to 60, the reductions were 2.99, 2.15, 2.31 and 1.77 log10, for cheeses from cheese factories A, B, C and D, respectively.

During the ripening of the Spanish artisanal cheeses Casar de Cáceres, Afuega’l Pitu and Cabrales, decreases in lactococcal counts of 2 to 3 log10 CFU g-1 were reported between d 0 and 60, while in “La Serena” cheese there was only a reduction of less than one logarithm, this is attributed to the fact that this cheese had a low salt content during the first weeks of ripening(5). This suggests that the reductions observed in the PC from d 5 to 60 are consistent with what happens with homofermentativelacticacid bacteriain cheeses madein an artisanal way with native microbiota(5). The concentrations of LAB found in the PC from d 60 to 90 are lower than those reported in ripened cheeses from Europe, which have values of 7 to 9 log10 CFU g-1(5,37,38) .

With differences in the type of cattle and geographical areas, Cotija cheese and PC share temperatures, preparation and ripening processes. In Cotija cheese, small increases and decreases have been found in LAB counts; one study reports 2.6 log10 CFU g-1 on d 30 with an increase to 2.9 log10 on d 90(29), another study indicates 5.9 log10 on d 60, which decreases to 5.0 log10 on d 90(11). The above data suggest that, in both PC and Cotija cheese, LAB counts tend to be lower than in other ripened cheeses; this could be explained by the temperatures at which theyare ripened, which favors a greater loss of moisture that generates values of water activity and moisture/salt ratio that are inhibitory for LAB(5) .

Coagulase-positive staphylococci

The CPS counts of the cheeses from the cheese factories studied showed significant differences (P<0.05) during the ripening period (Table 1). From d 5 to 60, a continuous decrease in CPS was observed in cheeses from cheese factories A, C and D, followed by a stabilization from d 60 to 90. In cheese factory B, there was a decrease from d 5 to 30 followed by an increase from d 30 to 60 and a stabilization from d 60 to 90 (Figure 2). The death rates of CPS (average decrease log10 CFU g-1 divided between week of ripening)(39) in cheeses from cheese factories A and C were 0.30 log10 and 0.23 and 0.31 log10 in cheese factories B and D, respectively.

Figure 2: Antagonism and change in lactic acid bacteria (LAB) concentration with respect to coagulase-positive staphylococci (CPS)

For cheeses made from raw milk, the accepted limits of coagulase-positive staphylococci are 104 to 105 CFU g-1, which is equivalent to 4-5 log10 CFU g-1(40), limits that were exceeded on day 5 in cheese factories A, C and D but that were reached again on day 30 and remained until day 90 (Table 1). CPS counts greater than 4 log10 show the need to apply corrective measures in the hygiene of the processes of milk collection, cheese making and the selection of raw materials(40). Values of 105 CFU g-1 or higher lead to study the presence of staphylococcal toxin in cheeses(40), since being thermostable, it can persist even when staphylococci have died. Concentrations of 106 CFUs g-1 are usually needed to produce enough toxin (one nanogram per gram of cheese) to cause a disease outbreak(5) .

Reductions of Staphyloccocus aureus of 1 to 3 logarithms have been reported in different cheeses(41,42,43), figures that coincide with reductions in CPS during PC ripening.

The death rate of S. aureus in Manchego cheese made with raw milk from d 1 to 60 is 0.404 log10 CFU g-1(39), the death rates of CPS in cheese factories A, C and D (0.49, 0.50 and 0.52) were close to this value, suggesting that the decrease was as expected for this type of cheese (matured Prensa paste cheeses). No factors that explain the lower death rate observed in cheese factory B (0.36) were identified.

The development and survival of S. aureus are affected by factors such as: physicochemical changes that occurred during the ripening process, secondarymetabolites generated by LAB, as well as the composition of the product, storage period and temperature(44,45) Staphyloccocus aureus is inhibitedbyLABthroughnutrientcompetition,productionoflactic acid, hydrogen peroxide and production of antimicrobial substances(46), which may explain the decrease in CPS in the first 30 d of ripening, a period in which LAB levels were higher (Figure 2). Between d 30 and 60, storage conditions at room temperature could increase moisture loss, which changes the moisture/salt ratio, being inhibitory for LAB(5), this favors S. aureus, which could explain its slight increase in cheese factory B and the suspension of its decrease in cheese factories A, C and D.

Conclusions and implications

Prensa cheese is a cheese made in an artisanal way, ripened in warm subhumid climate with the participation of native lactic acid bacteria, whose concentrations are lower than those reported for European cheeses, but similar to those reported in Cotija cheese. Statistical differences in microbial counts at different times show the changes that occur as cheese matures. Meanwhile, the statistical differences between the cheese factories suggest the existence of microbiomes specific to each cheese factory, which could be able to generate variants of PC among different artisanal producers, even when they have similar production processes. The changes in the counts of the bacterial groups studied can be attributed to physicochemical changes and successions in the bacterial populations typical of the maturation of the cheese and to the characteristics of the microbiota present in each of the cheese factories. It is convenient to explore whether the shelf life of this cheese extends beyond 90 days. The finding of coliforms that persist during ripening shows the need to investigate whether this is an exceptional case, and bacteria from this group contribute to the pleasant characteristics of the cheese or are related to its deterioration. The data on reduction and survival of coagulase-positive staphylococci generated in this research can serve as a reference to initiate and evaluate improvement programs in this type of cheese factories. Althoughthis researchincludedfourcheesefactoriesinthemainPC-producingmunicipality, the information generated can serve as a reference for the characterization of this artisanal cheese.

Acknowledgements and conflicts of interest

This research work was possible thanks to the scholarship granted by the National Council of Science and Technologyfor the master’s studies of the first author. We also appreciate the comments made on the manuscript by Doctors Angélica Luis Juan Morales and Ricardo Alaniz de la O. None of the authors has any conflict of interest with respect to this publication.

Literature cited:

1. Cidón CD, Canut E. Quesos españoles, ases en la mesa, comodines en la cocina. 1a ed. Madrid, ESP: Everest; 2003.

2. Yescas C, Santacruz J. Quesos mexicanos. 1a ed. México, Distrito Federal: Larousse; 2013.

3. Cheese.com specialty cheeses. Alphabetical list, find over 1833 specialty cheeses from 74 countriesintheworld´s greatestcheeseresource.https://www.cheese.com/alphabetical/. Accessed Jan 26, 2021.

4. KongoJM, MalcataFX.Cheese: Types ofcheeses -soft. In: CaballeroB, ToldráF, Pinglas PM, editors. Encyclopedia of food and health. 1rst ed. Oxford, GB: Academic Press; 2015:768–773.

5. Fox PF, Guinee TP, Cogan TM, McSweeney PLH. Fundamentals of cheese science. 2nd ed. New York, USA: Springer; 2017.

6. Santiago-López L,Aguilar-ToaláJE,Hernández-MendozaA,Vallejo-Cordoba B, Liceaga AM, González-Córdova AF. Invited review: Bioactive compunds produced during cheese ripening and health effects associated with aged cheese consumption. J DairySci 2018;101(5):3742–3757.

7. Villegas GA, Santos MA, Cervantes EF. Los quesos mexicanos tradicionales. 1a ed. Texcoco, MEX: Universidad Autónoma Chapingo; 2016.

8.Alejo-MartínezK,Ortiz-HernándezM,Recino-MetelinBR,González-CortésN, JiménezVera R. Tiempo de maduración y perfil microbiológico del queso de poro artesanal. ReIbCi 2015;2(5):15–24.

9. Sánchez-Valdés JJ, Colín-Navarro V, López-González F, Avilés-Nova F, Castelán-Ortega OA, Estrada-Flores JG. Diagnóstico de la calidad sanitaria en las queserías artesanales del municipio de Zacazonapan, Estado de México. Salud Publ Mex 2016;58(4):461–467.

10. Vasek O, Cardozo M, Fusco AJ. Producción artesanal de quesos. Sistema de transformación agroalimentario en la región correntina (Argentina). IV Congreso internacional de la red SIAL. Mar del Plata, provincia de Buenos Aires.2006:1–32.

11. Chombo-Morales P, Kirchmayr M, Gschaedler A, Lugo-Cervantes E, VillanuevaRodríguez S. Effects of controlling ripening conditions on the dynamics of the native microbial population of Mexican artisanal Cotija cheese assessed by PCR-DGGE. LWT-Food Sci Technol 2016;65:1153–1161.

12. Ramírez-Rivera EJ, Ramón-Canul LG, Torres-Hernández G, Herrera-Corredor JA, Juárez-Barrientos JM, Rodríguez-Miranda J, et al. Tipificación de quesos madurados de cabra producidos en la zona montañosa central del estado de Veracruz, México. Agrociencia 2018;52(1):15–34.

13. Sánchez-Gamboa C, Hicks-Pérez L, Gutiérrez-Méndez N, Heredia N, García S, NevárezMoorillón GV. Microbiological changes during ripening of Chihuahua cheese manufactured with raw milk and its seasonal variations. Foods 2018; 7(9):153.https:// www.mdpi.com/2304-8158/7/9/153. Accessed Jan 30, 2021.

14. Silva-Paz LE, Medina-Basulto G. E., López-Valencia G, Montaño-Gómez MF, VillaAngulo R, et al Caracterización de la leche y queso artesanal de la región de Ojos Negros, Baja California, México Rev Mex Cienc Pecu 2020;11(2):553-564. https://cienciaspecuarias.inifap.gob.mx/index.php/Pecuarias/article/view/5084 Consultado 30 Ene, 2021.

15. Sandoval-Alarcón F. Caracterización y análisis de la productividad del queso de prensa de la Costa Chica de Guerrero y Oaxaca [tesis maestría]. Texcoco, Estado de México: Universidad Autónoma Chapingo; 2016.

16. INEGI. Instituto Nacional de Estadística y Geografía. Guerrero, Clima. Recuperado de:http://www.cuentame.inegi.org.mx/monografias/informacion/gro/territorio/clima.as px?tema=me&e=12. Consultado 02 Sep, 2020.

17. Romero-Castillo PA, Leyva-Ruelas G, Cruz-Castillo JG, Santos-Moreno A. Evaluación de la calidad sanitaria de quesos crema tropical mexicano de la región de Tonalá, Chiapas. Rev Mex Ing Quím 2009;8(1):111–119.

18. 3M México. Placas Petrifilm™ para el recuento de aerobios AC, guía de interpretación. Ciudad de México, México: 3M. 2017.

19. 3M México. Placas Petrifilm™ para el recuento coliformes, guía de interpretación. Ciudad de México, México: 3M. 2017.

20. 3M México. Placas para el recuento de bacterias ácido lácticas 3M® Petrifilm®, guía de interpretación. Ciudad de México, México: 3M. 2017.

21. 3M México. Placas Petrifilm™ Staph Express para recuento de Staphylococcus aureus, guía de interpretación. México DF, México: 3M. 2009.

22. Camacho A, Giles M, Ortegón A, Palao M, Serrano B, Velázquez O. Técnicas para el análisis microbiológico de alimentos, cuenta en placa de bacterias. 2a ed. México, DF, México: Facultad de Química.; 2009. http://depa.fquim.unam.mx/amyd/archivero/TecnicBasicas-Cuenta-en-placa_6527.pdf. Consultado 2 Sep, 2020.

23. FAO, WHO. Statistical aspects of microbiological criteria related to foods, a risk managers guide. 1st ed. Rome, ITA: Food and Agriculture Organization of the United Nations; 2016.

24. Little RC, Milliken GA, Stroup WW, Wolfinger RD, Schabenberger O. SAS® for mixed models. 2nd ed. Cary, North Carolina, USA: SAS Institute Inc; 2006

25. Stroup WW, Milliken GA, Claassen EA, Wolfinger RD. SAS® for mixed models: Introduction and basic applications. 3rd ed. Cary, North Carolina, USA: SAS Institute Inc; 2018.

26. Brooks JC, Martinez B, Stratton J, Bianchini A, Kroksrom R, Hutkins R. Survey of raw milk cheeses for microbiological quality and prevalence of foodborne pathogens. Food Microbiol 2012;31(2):154–158.

27. ICMSF. International Commission on Microbiological Specifications for Foods. Microorganismos de los alimentos 1. Técnicas de análisis microbiológico. 2da ed. Zaragoza, ESP: Acribia; 2000.

28. Fernández-Escartín E. Microbiología e inocuidad de los alimentos. 2da ed. Querétaro, MEX: Universidad Autónoma de Querétaro; 2008.

29. Flores-Magallón R, Oliva-Hernández AA, Narváez-Zapata AA. Characterization of microbial traits involved with the elaboration of the Cotija cheese. Food Sci Biotechnol 2011;20(4):997–1003.

30. López-Expósito I, Miralles B, Amigo L, Hernández-Ledesma B. Health effects of cheese components with a focus on bioactive peptides. In: Frias J, Martinez-Villaluenga C, Peñas E, editors. Fermented foods in health and disease prevention. 1st ed. London, UK: Academy Press; 2017:239–273.

31. Asperger H, Brandl E. The significance of coliforms as indicator organisms in various types of cheese. Antonie van Leeuwenhoek 1983;48:635–639.

Pecu 2023;14(1):94-109

32. Metz M, Sheehan J, Feng PCH. Use of indicator bacteria for monitoring sanitary quality of raw milk cheeses – A literature review. Food Microbiol 2020;85(2020): 103283. https://www.sciencedirect.com/science/article/abs/pii/S0740002018311213?via%3Dih ub Accesed Jan 30, 2021.

33. Fox PF, Guinee TP, Cogan TM, McSweeney PLH. Fundamentals of cheeses science. 1st ed. Gaithersburg, Maryland, USA: Aspen Publisher 2000.

34.NACMCF. National Advisory Committe on Microbiological Criteria for Foods. NACMCF-Report-Process-Control-061015 (1) response to questions posed by the Department of Defense regarding microbiological criteria as indicators of process control or insanitary conditions, Washington DC, USA: United States Department of Agriculture; 2015. https://www.fsis.usda.gov/sites/default/files/media_file/202007/NACMCF-Report-Process-Control-061015.pdf . Accesed Aug 07, 2021.

35.Martin NH,TrmčićA, HsiehTH,BoorKJ,Wiedmann,M. Theevolvingroleofcoliforms as indicators of unhygienic processing conditions in dairy foods. Front Microbiol 2016;7:1549 https://www.frontiersin.org/articles/10.3389/fmicb.2016.01549/full Accessed Aug 07, 2021.

36. Trmčić A, Chauhan K, Kent DJ, Ralyea RD, Martin NH, Boor KJ, et al. Coliform detection in cheese is associated with specific cheese characteristics but no association was found with pathogen detection. J Dairy Sci 2016;99(8):6105–6120.

37. Khalid NM, Marth EH. Lactobacili – their enzymes and role in ripening and spoil age of cheese- A review. J Dairy Sci 1990;73(10):2669–2684.

38. Tavaria FK, Reis PJM, Malcata FX. Effect of dairy farm and milk refrigeration on microbiological and microstructural characteristics of matured Serra da Estrella cheese. Int Dairy J 2006;16(8):895–902.

39. Nuñez M, Bautista L, Medina M, Gaya P. Staphylococcus aureus, thermostable nuclease and staphylococcal enterotoxins in raw ewes' milk Manchego cheese. J Appl Bacteriol 1988;65(1):29–34.

40. Commission Regulation (EC) No. 2073/2005 of 15 November 2005 on microbiological criteria of foodstuffs L338. Official Journal of the European Union 2005 L338: 1–26. https://eur-lex.europa.eu/legal-content/EN/ALL/?uri=CELEX%3A32005R2073 Accessed Sep 2, 2020.

41. Kirdar SS, Yurdakul O, Kalit S, Kalit M. Microbiological changes throughout ripening of Keş cheese. J Cent Eur Agric 2018;19(1):61–71.

42. Cardoso VM, Dias RS, Soares BM, Clementino LA, Araújo CP, Rosa CA. The influence of ripening period length and season on the microbiological parameters of a traditional Brazilian cheese. Braz J Microbiol 2013;44(3):743–749.

43. Çolaklar M, Taban BM, Aytaç SA, Barbaros H, Gürsoy A, Akçelik N. Application of bacteriocin-like inhibitory substances (BLIS)- Producing probiotic strain of Lactobacillus plantarum in control of Staphylococcus aureus in White-Brined cheese production. J AgrSci 2019;25(2019):401–408.

44. Bellio A, Astegiano S, Traversa A, Bianchi DM, Gallina S, Vitale N, et al. Behaviour of Listeria monocytogenes and Staphylococcus aureus in sliced, vacuum-packaged raw milk cheese stored at two different temperatures and time periods. Int Dairy J 2016; 57:15–19.

45. Stecchini MA, Sarais I, de Bertoldi M. The influence of Lactobacillus plantarum culture inoculation on the fate Staphylococcus aureus and Salmonella typhimurium in Montasio cheese. Int J Food Microbiol 1991;14(2):99–109.

46. Haines WC, Harmon LG. Effect of selected lactic acid bacteria on growth of Staphylococcus aureus and production of enterotoxin. Appl Microbiol 1973;25(3):436–441.

https://doi.org/10.22319/rmcp.v14i1.6014 Article

Molecular detection of a fragment of bluetongue virus in sheep from different regions of Mexico

Edith Rojas Anaya a

Fernando Cerón-Téllez b

Luis Adrián Yáñez-Garza c

José Luis Gutiérrez-Hernández b

Rosa Elena Sarmiento-Salas c

Elizabeth Loza-Rubio b*

a Instituto Nacional de Investigaciones Forestales Agrícolas y Pecuarias (INIFAP). Centro Nacional de Recursos Genéticos. México.

b INIFAP. Centro Nacional de Investigación Disciplinaria en Salud Animal e Inocuidad (CENID-SAI), Campus Ciudad de México. Carretera México-Toluca Km 15.5, Colonia Palo Alto. 05110. Alcaldía Cuajimalpa de Morelos. Ciudad de México. México.

c Universidad Nacional Autónoma de México, FMVZ. México.

*Corresponding author: eli_rubio33@hotmail.com; loza.elizabeth@inifap.gob.mx

Abstract:

Bluetongue disease (BTD) affects various species of wild and domestic ruminants. In Mexico, the disease, caused bythe bluetongue virus (BTV) is still regarded as exotic, despite the fact that antibodies have been detected on several occasions. The objective was to establish molecular techniques using a synthetic gene, including the genes NS1 and NS3 as positive controls for the diagnosis of BTV in samples of sheep from different regions of the country. A total of 320 total whole blood samples were obtained from sheep. The samples obtained were evaluated byend-point RT-PCR and real-time RT-PCR, the conditions having

been established by the work group. Twelve sheep samples were found to be positive for the detection of NS1; these samples were sequenced, and a fragment of 101 base pairs was obtained.Uponalignment,wereobtainedidentitieswithsequencesreportedinGenBankwith NS1 fragments ranging from 89% (p= 1e-12)to 98 % (p=4e-13), correspondingto serotypes 10, 11 and 12. From these samples, two positive sheep samples were obtained using realtime PCR (RT-PCR): one from Chiapas (Chiapas breed), and the other, from Tamaulipas (Suffolk breed). The results of the RT-PCR were corroborated by CPA-SENASICA. This work provides evidence, for the first time in Mexico, of the importance of using a synthetic gene as a positive control to perform BSL-2 detection in official laboratories, which in a health emergency is of utmost importance.

Key words: Bluetongue disease, Bluetongue virus, Diagnosis, NS1 and NS3 genes, Sheep, Synthetic gene.

Received: 01/07/2021

Accepted: 31/08/2022

Introduction

Bluetongue virus (BTV) belongs to the genus Orbivirus and the family Reoviridae, and causes bluetongue disease (BTD) affecting both domestic and wild ruminants(1) . The virus has a negative-sense double-stranded RNA (dsRNA) genome consisting of 10 segments(2). It is a non-enveloped virus with an icosahedral capsid, with a diameter of approximately 90 nm. The genome codes for the structural proteins that make up the external and internal capsid or core (VP1 - VP7), and the four non-structural proteins, called non-structural (NS) that are involved in the replication, maturation, or exit of the virion from infected cells). Nonstructural genes are highly conserved across the genus(3,4). The NS1 gene encodes for a protein of the same name, which is expressed in the largest quantity during BTV replication and is the most abundant cytoplasmic protein. On the other hand, the NS3 gene encodes for theNS3proteinthatactsasaviroporin, whichisrelatedtocelllysis(3,4).Becauseoftheabove, the two genes have been used as targets in screening assays for the identification of BTV(5) . The virus is transmitted by the bite of mosquitoes of the genus Culicoides spp; therefore, the occurrence of the disease is associated with the spread of this vector, although other vectors such as ticks have been reported(6); it is known that the virus can remain viable throughout the life of the vector.

Currently, 28 different BTV serotypes have been described worldwide(7), and the virus is distributed in practically all countries where cattle and sheep are raised. Bluetongue disease (BTD) can occur both subclinically and clinically, especially in sheep, since in cattle it is mostly asymptomatic. In countries where the disease is endemic, it causes severe economic losses to producers(8). The name "blue tongue" was given to this disease by Africans who observed cyanosis on the tongue of some animals; however, this sign is not observed in all infected animals, as the signs vary between species and depend on the strain. Lesions such ashyperemiaand edemaofthelipsandface, oral erosionsandulcers,andthetypicalcyanosis of the tongue are due to infection of the endothelial cells that allow increased cell permeability(9)

The World Animal Health Organization (OIE)(10) classifies bluetongue disease as a notifiable disease; therefore, timely diagnosis is important. The degree of severity of the disease depends on the serotype, the virus strain, and the species, age and immune status of the animal, with sheep and white-tailed deer being the most affected(11); in sheep, the incubation period of BTV is six to eight days; on the other hand, cattle rarely show clinical signs, but maintain a prolonged viraemia(12) . Deer can also be infected by a closely related orbivirus responsible for epizootic hemorrhagic disease(13). BTD is not contagious and is only transmitted by Culicoides insects; its distribution is therefore associated with the prevalence of the vector. Up to five serotypes have been identified in North America; however, seven serotypes have been reported only in the United States(14). The occurrence of the virus in the Americas is mainly associated with the presence of two vector species, C. sonorensis and C. insignis(15). In Mexico, although the disease is considered exotic, in the 1980s, positive serology to the virus was reported in both sheep and cattle in different regions of the country(16,17). On the other hand, in 2015, the detection of a viral genome fragment in three Culicoides species (C. variipennis, C. sonorensis, and C. occidentalis) was published(18). The VP2 gene is used to define the 28 BTV serotypes described so far(7) .

Finally, in February 2021, the Secretary of Agricultural Development, Fisheries and Aquaculture (SEDPA) of Oaxaca in the Municipality of San Pedro Mixtepec notified the CPA of oral lesions in sheep. The CPA detected two bluetongue virus-positive samples using the RT-PCR technique and the notification was made in the CPA's AVISE Newsletter(19)

BTD is notifiable to the OIE, mainly because new outbreaks lead to movement and trade restrictions, resulting in severe economic losses. However, active surveillance is implemented worldwide to detect BTV infection through different tests such as virus isolation or other screening or serological tests(11). The detection method par excellence is the isolation of the virus in permissive cell cultures, for subsequent genetic analysis of the virus to determine the serotype present in the sample of the affected animal. In virus endemic areas, vector control is recommended to prevent the spread of the virus, in addition to vaccination programs. Live attenuated vaccines have been used in the United States and

Europe. The objective of this work was to establish molecular techniques using a synthetic gene as a positive control that included the NS1 and NS3 genes, in order to subsequently evaluate sheep samples from different regions of the country.

Material and methods Samples

Aconveniencesamplingofapparentlyhealthysheepwas carriedout,obtaining3mlofwhole blood with anticoagulant (heparin) from 320 individuals from five states of the country (Chiapas, Coahuila, Estado de México, Morelos, and Tamaulipas). Samples were obtained duringthe summer of 2016 to 2018. Sampling was performed on males and breeding females between one and five years of age. Table 1 describes the total number of samples analyzed.

Table 1: Blood samples obtained from sheep in five Mexican states analyzed for molecular detection of bluetongue virus by RT-PCR State Species Sex Breed Total

Total 320 F= female; M= male; N/D= no data.

Synthetic gene

Since BTD is considered exotic in Mexico, a synthetic gene was designed to be used as a positive control in order to avoid using the inactivated virus or its genetic material in a BSL2 laboratory. For this purpose, two fragments of the viral genome were inserted into the pUC57 vector (GeneScript, USA) one corresponding to the NS1 gene (354 bp), and the other, corresponding to the NS3 gene (300 bp) , based on the BTV-11 sequences reported in GenBank (KF986511 and KM580467, respectively). For use as a positive control in the molecular assays, the plasmid concentration was set to 100 ng/µl.

Genetic material extraction

Viral RNA was extracted from 250 l of the blood sample using the Trizol LS® reagent (Ambion, USA), following the manufacturer's instructions with some modifications to the protocol The RNA obtained was stored at -70 °C until use.

Molecular assays

Constitutive gene. In order to verify the quality of the RNA thus obtained, a fragment of the constitutive GAPDH gene was amplified by RT-PCR using the primers and conditions reported by González-Arto M, et al(20). Complementary DNA synthesized from viral RNA with the M-MLV Reverse Transcriptase kit was used as a template (Invitrogen, USA).

Detection of a fragment of gene NS3. The RNA extracted from the samples served as a templateforthedetection ofafragmentoftheNS3 geneoforbiviruses using apairofprimers and the probe recommended in the OIE Manual10. RT-PCR was carried out according to a one-step amplification protocol established in our laboratory, utilizing the iTaq Universal Probe One-Step Kit (Bio-Rad, USA).

Detection of a fragment of gene NS1. Inorderto corroboratethepresenceof theBTV genome in real-time PCR positive samples, a protocol was established for the detection of a fragment of the NS1 gene using primers described in the OIE Manual(10). The iProof HF Master Mix Kit was utilized for this purpose (Bio-Rad, USA). The products for the amplification of

positives to this protocol were purifiedin agarose gels andsequenced accordingto the Sanger method at the IBT-UNAM Synthesis and Sequencing Unit.

Sequencing. Sequencing results were analyzed with NCBI's BLAST tool. The obtained sequences were compared with 29 sequences reported in the Gene Bank of bluetongue virus and with sequence AM745001.1 for the epizootic hemorrhagic fever virus as an outgroup

The alignment was performed using the “Multiple alignment program for amino acid or nucleotide sequences” (MAFFT version 7, AIST). A phylogenetic analysis was performed using Bayesian methods (Markov Chain Monte Carlo) and the alignment was carried out with Mesquite in MrBayes software (Open source)

Results

As an assay to evaluate the quality of the genetic material, the amplification of a fragment of approximately 400 bp of the ovine GAPDH gene was carried out as previously above. All samples used for detection of a viral genomic fragment were positive for GAPDH amplification by RT-PCR, which indicates that the genetic material was intact and in good condition for use in RT-PCR assays (Figure 1).

Figure 1: RT-PCR for amplification of the sheep’s constitutive GAPDH gene

Lane 1, 50bp fragment size marker (Low Mass ladder); Lanes 2-7, sheep samples. Amplification products were run on a 1.5% agarose gel

As for the assay for the detection of a fragment of the NS1 gene by RT-PCR endpoint, the identification of this gene was obtained in 12 samples of sheep blood from the states of Chiapas, Coahuila, and Tamaulipas. In order to harmonize the methods, it was decided to use primers suggested by the OIE, as described above. The analysis of the sequences obtained showed in the alignment an identity with sequences reported in GenBank with the NS1 fragment from 89 % (p= 1e-12) to 98 % (p= 4e-13), corresponding to serotypes 10, 11 and 12.

Figure 2 shows the results of the phylogenetic inference, depicting the clustering of the samples primarily with serotypes 10 and 11. The positive results were corroborated by realtime PCR in two of the samples, one from Chiapas and the other from Tamaulipas; as mentioned above, RT-PCR uses the gene NS3.

Figure 2: Phylogenetic inference of bluetongue virus NS1-positive sheep samples

The dendogram was obtained using the alignment of 100 bp of the NS1 gene from the sequences of the positive samples in this study and 30 sequences obtained from GenBank belonging to serogroups 10, 11, and 12 Sheep samples positive by end-point RT-PCR are identified as follows:

OT= Tamaulipas sheep: OT1, OT2, OT3, OT4, OT5. OC= Chiapas sheep: OC1, OC2, OC3, OCF10, OCF39, OC5, OC4. The breeds are indicated by color: Creole , Pelibuey , Suffolk , and Dorper . The samples were taken between 2016 and 2017. The sequence of the epizootic hemorrhagic fever virus (AM745001) was used as outgroup, indicated by .

The positive result of RT-PCR detection was corroborated by the SENASICA – CPA laboratory and notified to SIVE. In this laboratory, the detection of a fragment of gene NS3

also serves as a positive control, using viral RNA obtained from the supernatant of cell cultures infected with the virus that are subsequently inactivated (CPA personal communication). With respect to the tests performed in collaboration with the official laboratory to corroborate the results of the animals that tested positive, the synthetic control proposed in this work showed that it could be used without problem in any BSL2 laboratory in order to perform virus detection in official BSL2 laboratories.

Discussion

Asmentionedabove,BTDin different hosts canbesubclinical,anddetectionofthecausative agent in sheep populations can be complex(21). Therefore, for this study, clinically healthy animals were considered for sampling, where the status of viraemia in animals and signs may ormaynot havebeen observed, dependingontheviral loadorsubtypes involved. In addition, sampling was carried out in those Federal Entities of the country located in areas where the vector transmitting the virus is present in a climate suitable for its development(22) , from sheep that were close to cattle farms, as bovines can be a healthy carrier of the virus.

Regarding the diagnosis of the causative agent of BTD, the recommended method is byvirus isolation in a cell culture or in embryonated eggs(23) However, different versions of RT-PCR have been developed that can be used to detect BTV, specifically the Orbivirus serogroup, and to determine the BTV serotype. These molecular approaches are much faster than traditional virological and immunological methods, which can take up to four weeks to provide information on serogroups and serotypes. Currently, there are targeted assays mainly for VP1, NS1, NS2, VP6, and NS3 proteins. None of these proteins is related to virus serotyping, and theyarestronglyconserved amongBTVserotypes, whilesome,suchas NS3, have a higher degree of conservation among orbiviruses. Therefore, these assays lack the potential to classify isolates(24) .

Inaddition,eachtechniqueoffersarangeofvirusor genomedetection;forinstance,Bonneau et al(25) report that the RT-PCR assay is capable of detecting the genome within a period ranging from 3 to 122 days. Therefore, it is important to make the recommendation to carry out sampling and surveillance campaigns not only on ruminants but also on the potential vectors that are reported as transmitters of the virus.

BTD is considered exotic in Mexico; however, this status should be reconsidered taking into account the various notifications made since the 1980s to the present date in different regions of the country; this would allow to assess the presence of the virus in different hosts and vectors using a variety of methods. In 1981, Moorhead et al(26) determined the presence of

antibodies by immunoprecipitation in sheep slaughtered at the slaughterhouse, finding 8.5 % positivity in serum. Subsequently, Vilchis et al (1986)(27) , using immunodiffusion, demonstrated 27.4 % seropositivity in the animals sampled. Stott et al(28) reported seropositivity of 6 %, 35 % and 60 % in three independent studies on cattle from different states of the country. The most recent scientific publication by Lozano-Rendón JA and his work group(18) proved a 14.4 % molecular detection of the NS1 gene of BTV in Culicoides vectors in the state of Nuevo León

As for the results presented in this research, a 3.75 % positivity rate was detected in the samples of clinically healthy sheep using the same NS1 gene as Lozano-Rendón et al(15) However, this studywas conducted on the vector, where the probabilityof demonstrating the presence of the virus is greater than in sheep, where the viraemia time is shorter. This NS1 gene, as already mentioned, is one of the most conserved among the different BTV serotypes(29). The results of the detection of a fragment of the viral genome in sheep samples in this study are consistent with those reported this year by the CPA(19) .

On the other hand, the detection rate is similar to that described in older reports using ruminant samples. The results reported in the present work, as well as those presented by other authors, show the need to change the status of the disease, as well as to implement virus surveillance systems in both the vectors and the main hosts of the virus, whether these be domestic or wild animals.

Conclusions and implications

A synthetic positive control is presented herein as an alternative to viral RNA, which can only be utilized in the BSL3 laboratory of the country's official agencies. The use of such synthetic positive control would enlarge the network of laboratories capable of implementing the viral detection technique to determine the real status of the disease in the country.

Acknowledgments

The authors are grateful to Roberto Navarro López, MSc; to Marcela Villarreal Silva, PhD; to Mariana García Plata, MSc, and to Martín García Osorio, DVM, for their collaboration in corroborating the results in the CPA-SENASICA Laboratory. This research was financed by INIFAP project No. 12583634008 and the validated form No. 914545716.

Literature cited:

1. Qing-Long G, Wang Q, Yang XY, Li DL, Zhao B, Ge GY, et al. Seroprevalence and risk factors of the bluetongue virus in cattle in China from 1988 to 2019: A comprehensive literature review and meta-analysis. Front Vet Sci 2021;7:550381 doi: 10.3389/fvets.2020.550381.

2. Maclachlan NJ. Bluetongue: History, global epidemiology, and pathogenesis. Prev Vet Med 2011;102(2):107– 111.

3. Chacko N, Mohanty NN, Biswas SK, Chand K, Yogisharadhya R, et al. A coiled-coil motif in non-structural protein 3 (NS3) of bluetongue virus forms an oligomer. Virus Genes 2015;51(2):244–251 doi 10.1007/s11262-015-1230-9.

4. Roy P. Bluetongue virus proteins and particles and their role in virus entry, assembly and release. Adv Virus Res 2015;64:69–123. doi.org/10.1016/S0065-3527(05)64004-3.

5. Schwartz-Cornil PP, Mertens V, Contreras B, Hemati F, Pascale E, Breard et al. Bluetongue virus: virology, pathogenesis and immunity. Vet Res 2008;39(5):46. doi: 10.1051/vetres:2008023.

6. Sperlova A, Zendulkova D. Bluetongue: a review. Vet Med 2011;56:430–452.

7. Bumbarov V, Golender N,Jenckel M, WernikeK,Beer M, Khinich E, ZaleskyO,Erster O. Characterization of bluetongue virus serotype 28. Transb Emer Dis 2020;67(1):171–182. doi.org/10.1111/tbed.133338.

8. Maclachlan NJ, Drew CP, Drew CP, Darpel KE, Worwa G. The pathology and pathogenesis of bluetongue. J Comp Pathol 2009;141:1-16. doi.org/10.1016/j.jcpa.2009.04.003.

9. Drew CP, Heller MC, Mayo C, Watson JL, Maclachlan NJ. Bluetongue virus infection activatesbovinemonocyte-derivedmacrophages andpulmonaryarteryendothelialcells. Vet Immunol Immunopathol 2010;136(3–4):292–296. doi:10.1016/j.vetimm.2010.03.006.

10. OIE. Enfermedades, infecciones e infestaciones de la Lista de la OIE en vigor en 2019. http://www.oie.int/es/sanidad-animal-en-el-mundo/enfermedades-de-la-lista-de-la-oie2018/. Consultado 16 Nov, 2019.

11. Rojas JM, Rodríguez-Martín D, Martín V, Sevilla N. Diagnosing bluetongue virus in domestic ruminants: current perspectives. Vet Med Res Report 2019;10:17–27.

12. Barratt-Boyes SM, Maclachlan NJ. Dynamics of viral spread in bluetongue virus infected calves. Vet Microbiol 1994;40(3–4):361–371. doi.org/10.1016/03781135(94)90123.

13. Falconi C, López-Olvera JR, Gortázar C. BTV infection in wild ruminants, with emphasis on red deer: a review. Vet Microbiol 2011;151(3-4)209-219. doi.org/10.1016/j.vetmic.2011.02.011.

14. Drolet S, Rijn P, Howerth E, Beer M, Mertens P. A review of knowledge gaps and tools for Orbivirus research. Vector-borne Zoon Dis 2015;15(6): 339-347. doi: 10.1089/vbz.2014.1701.

15. Gay GC. Orbiviruses: A gap analysis. Vector Borne Zoonotic Dis- 2015;15(6):333-334. doi:10.1089/vbz.2015.28999.cgg.

16. Suzan VM, Misao O, Romero EA, Yosuke M. Prevalence of bovine herpesvlrus-1, paraenfluenza-3, bovine rotavirus, bovine viral diarrhoea, bovine adenovirus 7, bovine leukemia virus and bluetongue virus antibodies in cattle in Mexico. Jpn J Vet Res 1983;31(3-4): 125-132.

17. Vilchis C, GayJ, Batalla D. Determinación de anticuerpos contra el virus de lengua azul en ovinos por la técnica de inmunodifusión. Tec Pecu Méx 1986;51:116-121.

18. Lozano-Rendón JA, Contreras-Balderas AJ, Fernández-Salas I, Zarate-Ramos J, Avalos-Ramírez R. Molecular detection of bluetongue virus (BTV) and epizootic hemorrhagic disease virus (EHDV) in captured Culicoides spp. in the northeastern regions of Mexico. Afr J Microbiol Res 2015;9(45):2218-2224.

19. Boletín Informativo de la CPA. AVISE. No 9 Febrero, 2021. https://issuu.com/boletinavise/docs/boletin_avise_ed09_febrero

20. Gonzalez-Arto M, Hamilton dos STR, Gallego M, Gaspar-Torrubia E, Aguilar D, Serrano-Blesa E, et al. Evidence of melatonin synthesis in the ram reproductive tract. Andrology 2016;4(1):167-171. doi: 10.1111/andr.12117.

21. Celma CC, Bhattacharya B, Eschbaumer M, Wernike K, Beer M, Roy P. Pathogenicity study in sheep using reverse-genetics-based reassortant bluetongue viruses Vet Microbiol 2014;174(1-2):139-47. doi: 10.1016/j.vetmic.2014.09.012.

22. SIAP. Producción por Estado. 2016. http://infosiap.siap.gob.mx/anpecuario_siapx_gobmx/apecnal.jsp?id=5.

23. McHolland LE, Mecham JO. Characterization of cell lines developed from field populations of Culicoides sonorensis (Diptera: Ceratopogonidae). J Med Entomol 2003;40(3):348-51. doi: 10.1603/0022-2585-40.3.348.

Pecu 2023;14(1):110-121

24. Maan S, Maan NS, Belaganahalli MN, Potgieter AC, Kumar V, Batra K, et al. Development and evaluation of Real Time RT-PCR assays for detection and typing of Bluetongue Virus. PLoS ONE 2016;11(9): e0163014. doi: 10.1371/journal.pone.0163014

25. Bonneau KR, DeMaula CD, Mullens BA, Maclachlan NJ. Duration of viraemia infectious to Culicoides sonorensis in bluetongue virus-infected cattle and sheep. Amsterdam: Elsevier Scientific Publishing Co; 2002.

26. Moorhead JR. Estudio de la presencia de anticuerpos precipitantes contra el virus de la Lengua Azul en ovinos y bovinos sacrificados en el Rastro de Ferrería de la Ciudad de México, DF [tesis]. Facultad de Medicina Veterinaria y Zootecnia, Universidad Nacional Autónoma de México;1981.

27. Vilchis CM, Gay GJ, Batalla D. Determinación de anticuerpos contra el virus de lengua azul en ovinos por la técnica de inmunodifusión. Tec Pecu Méx 1986;51:116-121.

28. Stott JL, Blanchard-Channell M, Osburn BI, Riemann HP, Obeso RC. Serologic and virologic evidence of bluetongue virus infection in cattle and sheep in Mexico. Am J Vet Res 1989;50(3):335–340.

29. Mertens PP, Diprose J, Maan S, Singh KP, Attoui H, et al. Bluetongue virus replication, molecular and structural biology. Vet Italiana 2004;40(4):426-437.

https://doi.org/10.22319/rmcp.v14i1.6241 Article

Insulin-like growth factor 1 (IGF-1) concentrations in synovial fluid of sound and osteoarthritic horses, and its correlation with proinflammatory cytokines IL-6 and TNF

Fernando García-Lacy F. a

Sara Teresa Méndez-Cruz b

Horacio Reyes-Vivas b

Victor Manuel Dávila-Borja c

Jose Alejandro Barrera-Morales d

Gabriel Gutiérrez-Ospina e

Margarita Gómez-Chavarín f*

Francisco José Trigo-Tavera g

a UniversidadNacionalAutónomadeMéxico FacultaddeMedicina Veterinaria yZootecnia Departamento de Medicina, Cirugía y Zootecnia para Équidos. Ciudad de México. México.

b Instituto Nacional de Pediatría Laboratorio de Bioquímica Genética. Ciudad de México. México.

c Instituto Nacional de Pediatría. Laboratorio de Oncología Experimental. México.

d SEDENA. Centro Ecuestre de Alto Rendimiento. Ciudad México. México.

e Universidad Nacional Autónoma de México. Departamento de Fisiología. Instituto de Investigaciones Biomédicas. Ciudad de México. México.

f Universidad Nacional Autónoma de México. Facultad de Medicina. Departamento de Fisiología. Ciudad de México. Mexico.

g UniversidadNacionalAutónomadeMéxico.FacultaddeMedicinaVeterinaria yZootecnia. Departamento de Patología. Ciudad de México. México.

*Corresponding author: margaritachavarin@gmail.com

Abstract:

Insulin-like growth factor I (IGF-1) is the most important known growth factor for cartilage repair in horses. It promotes mitosis of chondrocytes, collagen II expression, and extra cellular matrix production. Osteoarthritis (OA) is the most common musculoskeletal condition that causes lameness and poor performance in sport horses. A total of 11 lame horses wereclinicallyand radiographicallyevaluated,andall wereconfirmedto sufferafront metacarpophalangeal lameness by a positive flexion test, a low-4-point nerve block and an intraarticular block. Total protein, IGF-1, IL-6 and TNF were determined by ELISA, demonstrating changes and different correlations between clinical condition, radiographic changes and degree of inflammation. All horses with joint associated pain and therefore associated lameness, demonstrated a significant increase of total protein (P<0.0001) and IGF-1 concentration (P<0.05). Concentrations of IL-6 and TNF between controls and lame horses demonstrated significant differences (P<0.01 and P<0.001 respectively). Horses with less radiographic changes, demonstrated the highest IGF-1 expression in synovial fluid, and horses with more chronic OA conditions had very similar IGF-1 expression levels than control joints. In all lame joints, it was identified by Western blot a lighter isoform of IGF-1 (~7.5 kDa) which was inflammation related and it is the molecular weight of the mature peptide, and all control joints expressed a heavier isoform (~12 kDa). This finding could lead to new research for sequencing and targeting the isoform which is not expressed during an inflammatory process within a joint, and to have a better understanding of its role in the horse’s joint.

Key words: Insulin growth factor 1 (IGF-1), Horse, Osteoarthritis (OA), Lame

Received: 21/05/2022

Accepted: 07/09/2022

Introduction

Insulin-like growth factor I (IGF-1) is the most important known growth factor for cartilage repair in horses, because it stimulates proteoglycan synthesis, and therefore extra cellular matrix (ECM), and promotes mitosis of chondrocytes. It has an important growth-promoting

activity not only in articular cartilage, but in several tissues, mainly in muscle, bones and brain. It activates the mitogen-activated protein kinase (MAPK) pathway, having several effects in promoting: cell survival, growth, proliferation, protectionto hypoxia,inflammation regulation in muscle injuries and in bone growth plates(1-4) It has also a key role in brain development, along with estradiol, regulating a variety of developmental and neuroplastic events(5) In structure, it is very similar to insulin When insulin is instilled in a joint, IGF-1 expression in synovial fluid is enhanced(6)

Injuries of the articular cartilage are normally repaired by substitution with fibrocartilage, which leads to loss of function of the joint resulting in osteoarthritis (OA)(7). Lameness is the most frequent reason for which equine practitioners are required from horse owners, and OA represents more than 60 % of all lameness cases in sport horses(8). The main problem in OA is inflammation, conditioning an imbalance between catabolism and anabolism in the articular cartilage. In this particular tissue, the only cellular component is constituted by chondrocytes, which are responsible of ECM synthesis, in order to maintain adequate cartilage function.

There is evidence regarding exogenous efficacy of IGF-1 in vitro, which enhances proteoglycan synthesis by stimulated chondrocytes. Other therapies, such as chondrocyte transplantation from mature and neonatal chondrocytes, gene therapystrategies to upregulate IGF-1 expression by transfected chondrocytes, require general anesthesia, a surgical procedure and therefore specialized equipment and personnel(9) .

On a pilot study conducted by the authors in which 13 synovial fluid samples obtained from different joints (distal interphalangeal joints, metacarpophalangeal joints, shoulder joints, tarsometatarsal joints, and stifles) from horses with associated lameness AAEP (American Association of Equine Practitioners) grade: 2/5, no radiographic changes but a positive response on 1-minute flexion test. By ELISA, a significant increase of IGF-1 and a positive correlation between total protein and IGF-1 levels in synovial fluid (data not shown) were found. In this study it was obtained synovial fluid samples from 21 horses with different degrees of OA (confirmed by intraarticular block and radiographic changes) in the metacarpo-phalangeal joint (MCPJ), where IGF-1 and total protein correlated positively in horses with acute OA, and negatively in horses with chronic OA and marked bone remodeling. In horses with mild of non-radiogrphic changes (acute OA), IGF-1 correlated negativelywith interleukin-6 (IL-6) and tumoral necrosis factor alpha (TNF). Interestingly, were able to find by western blot, at least two functional isoforms of IGF-1 expressed in synovial fluid, one present only in control horses, and the other in lame horses.

To our knowledge, there is no information regarding IGF-1 fluctuations on naturally occurring OA. There are no in vivo studies regarding IGF-1 levels during OA. Perhaps this paper can help practitioners to understand the role of IGF-1 for this particular condition and

could be used as a baseline for further studies on IGF-1 concentration and its possible use as an alternative treatment.

Material and methods

Synovial fluid samples were obtained from Warmblood and Thoroughbred horses (n=11) from two different disciplines: showjumpers (Warmblood) (n= 8), and race horses (Thoroughbred) (n= 3) with a mean age of 10.5 yr old and a mean weight of 520 kg. Control (Ctrl) samples were obtained from two geldings, Warmblood horses of 5 and 7 yr old. No more control horses were available for the study, since they all were sound, it was not easy to obtain consent from the owners to sample their joints. A complete lameness evaluation and radiographic assessment were performed in all control horses in order to be included in this study. None of them showed signs of front limb lameness and were negative to passive and active flexion tests (30 sec). Additionally, none of them presented any radiographic changes associated with joint pathology in the metacarpophalangeal joint.

Lameness evaluation

A clinical evaluation was performed on all horses included in this study, in order to find evidence of lameness associated with the metacarpo-phalangeal joint of the front limbs. Evaluation consisted on static observation, palpation and passive flexion response; dynamic evaluation ofwalk and trot onastraight line andlunged onhardandsoftsurface. All included horses demonstrated a 2 and 3/5 lameness (AAEP), with a positive flexion test (1 min). Additionally, all horses were positive to low-4-point block (lateral and medial palmar nerves and lateral and medial metacarpal nerves), using 2 and 1.5 ml respectively of 2 % mepivacaine (Carbocaine, Zoetis Inc.) and further intra-articular block of the metacarpalphalangeal joint, using a volume of 6 ml of 2 % Mepivacaine (Carbocaine, Zoetis Inc.) as previously described(10) Any horse negative to these blocks, was excluded from the study.

Synovial fluid collection

All synovial fluid samples were obtained from the metacarpal-phalangeal joint MCPJ joints of lame horses, using an aseptic technique on the palmaro-lateral approach as previously

described and 5 d after the intraarticular (IA) block(10). Samples were obtained from healthy horses and were used as controls (n= 4).

Radiographic evaluation

All selected horses were radiographically evaluated from the MCPJ, using four standard views (dorso-palmar, latero-medial, dorso-lateral palmaro-medial, and dorso-medial palmaro-lateral) in order to assess the radiologic condition of all horses. Three different grades of radiologic changes were determined associated with the clinical condition of the horse (Table 1).

Table 1: Grade of severity and its relation on clinical and radiographic findings on horses included in this study

Grade Radiographic and clinical findings

I Non to minor changes associated with joint pain and lameness: Irregularity and loss of normal homogeneity of the sagittal ridge of MTCIII.

II Moderate changes associated with joint pain and lameness: Osselets (osteophytes) on P1 and MTCIII.

III Severe changes associated with severe lameness and decrease of motion range: suprachondilar or subchondral lysis, osteophytes and new bone formation with periostic reaction and loss of articular space. (Modified from: Verwilghen D, et al. 2009)(11)

Protein concentration determination

The concentration of total protein from all synovial fluid samples was obtained by using the BCA Protein Assay Kit, (Pierce BCA Protein Assay Kit cat. 23225), according with the manufacturer’s instructions. For each sample the final concentration was 100 g/50 L for the ELISA procedure.

IGF-1 concentration analysis

Determination of IGF-1 concentration in synovial fluid samples of control and osteoarthritic horses was made with 50 L using a commercial ELISA kit (Horse IGF1 ELISA kit, #MBS017382, MyBio-Source®) following the manufacturer’s instructions.

Interleukin 6 (IL-6) concentration

A quantitative determination of IL-6 in synovial fluid of all samples was performed using a commercial ELISA kit (Horse interleukin-6 ELISA kit, cat. #: CSB-E16634Hs), which is a sandwich immunoassay technique, where the plates are coated with a specific horse IL-6 antibody, then, a specific biotin-conjugated antibody for IL-6 and then avidin conjugated horseradish peroxidase (HRP) are added. Protocol is performed following manufacturer’s instructions.

Tumoral necrosis factor alpha (TNF) concentration

Determination of TNF concentration in synovial fluid samples of control and osteoarthritic horses was made with 100 l using a commercial ELISA kit (Equine TNF ELISA kit, cat #: ESS0017 Invitrogen) following the manufacturer’s instructions

Western Blot analysis

Equal amounts of protein (100 µg per lane), were subjected to a 16 % SDS-PAGE (90V for 30 min and 120 V for 3.5 h). Precision Plus Protein Dual Color Standards marker was used, containing ten prestained recombinant proteins (10 to 250 kD), including eight blue-stained bands and two pink reference bands (25 and 75 kD). After electrophoresis, gels were transferred using a semi-dry transfer system (271mA for 15 min) to PVDF (0.45uM) (BioRad) membranes, which were blocked using 4% skim milk diluted in PBS (pH 7.4) and incubated on a shaker at 37 oC, 120 rpm for 2 h. After blocking, membranes were washed 3x (for 5 min each) using PBS containing 0.05% Tween-20. As a primary antibody, a goat polyclonal anti-IGF-1 (1:1000) (Sta. Cruz #Sc-1422) was used, incubated on a shaker first at 37 oC, 120 rpm for 2 h, and left overnight at 4 oC; membranes were washed again as

previously described and as a secondary antibody, a polyclonal anti-goat IgG (1:5000) (Millipore #AP180B) was used, and incubated on a shaker at 37 oC, 120 rpm for 2 h and a final wash of the membranes was performed. Proteins were detected by using an enhanced chemiluminescencemethodandvisualizedusinga high-resolution Imaging System(Bio-Rad ChemiDoc). Membranes were incubated to a 1:1 dilution of luminol and peroxidase (Merck Millipore, Luminata # WBLUF0500), and exposed at various times, where the optimum time of exposure was 35 seconds.

Results

A total of 45 horses were examined, from which only 11 horses (22 samples) were included in this study, and 2 horses (4 samples) as controls. All horses varied from each other in degrees of lameness and radiographic changes, and all responded positively to the digital flexion test, low-4-point block and intraarticular (IA) block of the fetlock joint. Six joints were scored as grade I, five joints were scored as grade II and 8 joints were grade III (Figure 1)

Figure 1: Representative radiographs from horses scored with various grades

IGF-1 concentration

All horses with joint associated pain, lameness, and less radiographic changes, demonstrated a significant increase of IGF-1 concentration (P<0.05) (Figure 2). All samples were repeated by pairs and read three times in a 5, 10 and 15-min period with no difference between measurements (data not shown) and the values of linear regression and standard curve were: P<0.001; r2=0.9931.

IL-6 and TNF analysis

Concentrations of IL-6 between controls and lame horses, showed a significant difference as well (P<0.01). TNF concentrations between controls and lame horses showed even more significant differences in terms of concentration (P<0.001), being higher on lame horses with more severe changes in the affected joints (Grade III). A Pearson’s correlation analysis was performed demonstrating a positive correlation between total protein and IGF-1 concentrations (r= 1), which was seen in grade I and II horses, whereas in grade III this correlation is negatively or inversely proportional. In other words, the worse changes a joint had (as seen in grade III horses), the less IGF-1 concentration in synovial fluid was observed.

Figure 2: A) IGF-1 determination between control (sound) and lame horses, demonstrating a significant difference (P<0.05)*. B) Concentrations of IL-6 between controls and lame horses,showedasignificantdifferenceas well(P<0.01)**.C)TNF concentrationsbetween controls and lame horses showing significant differences in concentration (P<0.001)***.

Western blot analysis

Horses with less radiographic changes demonstrated a higher IGF-1 concentration concordantly with the ELISA results for IGF-1 (Grade I and II). Horses with more severe radiographic changes and a chronic state of the pathologic condition (Grade III), were the ones with the lowest IGF-1 concentrations in both ELISA and WB analysis. Interestingly, with this analysis it was able to identify in all samples, two different bands, one of ~12 kDa which was seen only in control (normal) horses with no joint pathology, and another of ~7.5 kDa seen in all lame horses (Figure 3).

Figure 3: Representative photograph of Western blot analysis for IGF-1, demonstrating a difference in molecular weight between synovial fluid samples indicating the existence of two different isoforms present in normal joints and during an inflammatory process

1: Protein marker (marking 10 kDa); 2: Samples from a control horse and a horse with OA; 3: Control horse; 4 & 5: Two different samples from horses with OA.

Discussion

Sport horses are exposed to excessive loads to their joints and soft tissue structures. The joint that can udergo traumatic OA depends on the discipline in which the horse performs. There’s evidence regarding interventions such as joint injections on acute phases of the disease that can help modify its course and prevent further damage while the horse is still performing(12) . Impact loads due to exercise are responsible of damaging articular cartilage by first cracking the surface, and depending on the force applied and the time it is being applied, the depth and therfore the severity of the development of the disease (OA) are produced. Characterization

of mechanical consequences of impact injuries to articular cartilage has been proven to develop damage, by continuously and directly stressing the joint structures(13) .

When inflammation occurs, chondrocytes migrate to the lesion site in an attempt to regenerate the defect by forming groups of cells or clusters with the ability to synthesize ECM de novo. Since the cellular component (chondrocytes) of the articular cartilage is only 1-2 % of the whole tissue, they are unable to repair the damaged area, because their ability to synthesize ECM is surpassed by the matrix metalloprotease (MMP) activity which degrades the already damaged ECM aggravating the condition by increasing necrosis and activating local inflammation by releasing intracellular components which act as damage associated molecular patterns (DAMPs) and proinflammatory cytokines such as prostaglandins (PGs), nitrous oxide (NO), interleukin-1 (IL-1), interleukin-6 (IL-6), tumoral necrosis factor alpha (TNF) and substance P. Particularly, TNF inhibits IGF-1 expression by increasing ECM catabolism, and blocking AKT pathway via activating JNK pathway. If the cartilage defect reaches subchondral bone, the cartilage repairs forming a low-quality articular cartilage called fibrocartilage(2,4,7) .

Factors that contribute to the inflammation cascade other than citokines, include extracellular vesicles, which playan important role on promotingjoint inflammation and are also involved on apoptosis and ECM degradation. These vesicles are exosomes, microvesicles and apoptotic vesicles, which are all released to the articular cavity (into the synovial fluid), and have intimate relation with cell-cell communication during the inflammatory process(7,8,14) . The aim of this study was to compare IGF-1 concentration in synovial fluid from sound (control) horses and horses with different degrees of lameness and joint pathology (OA) in the MCPJ It was hypothesized that the more severe and chronic conditions of the joint, the highest IGF-1 levels in synovial fluid would be found, because of the joint’s high demand for repairing the defect was higher than in horses with mild changes The rationale behind the hypothesis was: to our knowledge, there is still no data available regarding IGF-1 concentrations and its correlation with a particular clinical condition in horses, so it was conducted a pilot study, where a total of 13 synovial fluid samples were collected from different joints of different horses. All of these horses were high performance show jumpers with a positive flexion test from the sampled joints (no radiographic evaluation was conducted on any of these horses). What was found, it was a significant increase of total protein, with a positive correlation (Pearson’s correlation, P=0.0229) on IGF-1 levels in synovial fluid when compared to control samples (synovial fluid obtained from sound horses). This gave sufficient information to hypothesize that horses with more severe clinical signs and more chronic joint pathology would have higher IGF-1 levels when compared to control horses.

Interestingly, with the results obtained, this hypothesis was refuted. It was encountered that the horses with more severe radiographic changes and thus, the more chronic inflammatory

conditions within the joint, were the ones that demonstrated a decrease on IGF-1 concentrations, very similar to what control horses had.

Similar results were seen on studies in which experimentally induced lesions on articular cartilage in horses, make an acute peak of mRNA expression of igf-1, and at 4 weeks tend to decrease When IGF-1 decreases, TGF- predominates and it is responsible of new bone formation and activation of quiescent lymphocytes to Th17(7)

Equine Insulin-like growth factor 1 (IGF-1) has been widelystudied, there are several studies whereitsimportancein cellproliferation,growth andsurvival,repairandextracellularmatrix production is well documented, although there are not enough studies regarding the different isoformsandtheirfunctionality(15).ItisknownthatthemRNAundergoespost-transcriptional modifications (alternate splicing) which generates different isoforms along with posttranslation modifications. IGF-1 propeptides are encoded by multiple alternately spliced transcripts including C-terminal extension peptides called E-peptides, and N-terminal signal peptides. When an immature protein has signal peptide, mature peptide and E peptide is called pre-proIGF-1, and when the signal peptide is eliminated leaving only the mature peptide and the E peptide, is called pro-IGF-1. These E-peptides control the bioavailabilty of mature IGF-1, by binding to the ECM due to their highly positive charge, preventing its systemic circulation and therefore, its local use. They also modulate mature IGF-1 re-entry to the cell in a murine muscle cell-line(1) .

In humans, three different IGF-1 isoforms have been identified (IGF-1Ea, IGF-1Eb and IGF1Ec, also known as mecano-growth factor or MGF), and have been proposed to have various functions in muscle repair(16)

Nixon, et al(17) described the igf1 gene consisting of 5 exons with 4 intron sequences, which undergo both post-transcriptional and post-traslational modifications, where the translated proteins resulting from alternate splicing of exon 4 form a smaller propeptide (105 aminoacids) transcript named Pre-proIGF-1A consisting of signal peptide (encoded byexons 1 and 2), mature peptide (encoded by exons 2 and 3), and a C-terminal E-peptide encoded by exons 3 and 5); and when exon 4 is not alternately spliced, a larger transcript is translated forming Pre-proIGF1B (111 aminoacids)(17). To our knowledge, this was the last research paper published regarding post-transcriptional and post-traslational modifications and alternate splicing of IGF-1 mRNA in horses. It was conducted a bioinformatic analysis of igf1 gene undergoing different types of alternate splicing, which according to Le, et al(18) are: exon skipping, intron retention, mutually exclusive exons and alternative 5’ donor or 3’ acceptor sites. This analysis revealed that IGF-1 mRNA consisted of not 5, but 4 exons and 3 introns, which transcipts form 4 isoforms: variant 1 (exons 1-3), variant 2 (exons 2 and 3), variant 3 (exons 1-3, a 93 pb intron retention, and exon 4) and variant 4 (exons 2-4)(18) .

The Western blot analysis demonstrated the presence of at least two different functional isoforms of IGF-1, where the one seen in all normal horses is heavier (~12 kDa) than the one seen in all horses with different degrees of OA (7.5 kDa). Probably the lighter one is the mature form of IGF-1, although aminoacid sequencing techniques must be carried out in order to confirm this statement. With this result, can be presumed that the expression of these two different functional isoforms depends on inflammation.

This could lead to a new line of research which can focus on determine by advanced sequencing techniques the exact isoforms of IGF-1 and to target overexpression of the isoform which is not present when there is an inflammatory process of the joint, and its role on repairing cartilage defects

Articular cartilage does not regenerate by itself, since is the only connective tissue in mammals that does not have either blood and lymphatic vessels, or nerves(19). Therefore, it is virtually impossible to regenerate after an injury, so it is repaired via substitution with fibrous tissue, which generates a low quality fibrous cartilage called fibrocartilage. There have been several treatments to improve cartilage regeneration, in humans, osteochondral allograft transplantation has proven to be effective in function improvement and overall repair with graft survivorship of up to 80 % of the patients who had undergone previous surgical treatment: Microfracture, cartilage debridement, forage, abrasion chondroplasty, osteochondral and periosteal grafts, cartilage flap reattachment, among others(7,17,20) .

Local anestethics and steroids have been used widely by practitioners in the field, for diagnostic and therapeutic reasons respectively. However, excessive use of these components, have been proven to damage articular cartilage. Intraarticular injection using local anesthetics and steroids have make a growing concern about inducing potential toxicity to chondrocytes and synoviocytes. Sherman et al(21) , conducted an interesting comparisson of lidocaine, bupivacaine, betamethasone acetate, methylprednisolone acetate, and triamcinolone acetonide in a canine model. They found that in vitro, 1 and 0.5 % lidocaine, 0.2 and 0.25 % bupivacaine, betamethasone acetate and methylprednisolone acetate were severely chondrotoxic and synoviotoxic when compared with 0.625 % bupivacaine and triamcinolone(21)

Conclusions and implications

For this reason, treatmentwise, the main goal is to have alternatives that could be used in the field by clinicians, that can provide an alternative other than steroids that can also enhance cartilage repair without the need of getting the horse under general anesthesia and still have

an effect that lead to horses having a long lasting sport career. This paper provides important information that can serve as a base for further research regarding IGF-1 isoforms and their role in cartilage repair.

Acknowledgements

The authors would like to thank all CEAR, SEDENA personnel, QFB Alberto Enrique Fernández Molina and MVZ Jorge Rodríguez Lezama. Fernando García Lacy is a doctoral student at Doctorado en Ciencias de la Producción y Salud Animal de la Facultad de Medicina Veterinaria y Zootecnia de la Universidad Nacional Autónoma de México and received a scholarship from CONACYT. The work reported in this manuscript is part of his doctoral dissertation.

Conflicts of interest

The authors declare that there are no conflicts of interest.

Literature cited:

1. Pfeffer LA, Brisson BK, Hanquin L, Barton ER. The Insulin-like Growth Factor (IGF-1) E-peptides modulate cell entry to mature IGF-1 protein. Mol Biol Cel 2009;20:38103817.

2. Choukair D, Hügel U, Sander A, Uhlmann L, Tönshoff B. Inhibition of IGF-1-related intracellular signaling pathways by proinflammatory citokines in growth plate chondrocytes. Ped Res 2014;76(3):245-251.

3. Liu Q, Guan JZ, Sun Y, Le Z, Zhang P, Yu D, et al Insulin-like growth factor 1 receptormediated cell survival in hypoxia depends on the promotion of autophagyvia supression of the PI3K/Akt/mTOR signaling pathway. Mol Med Rep 2017;15:2136-2142.

4. Tonkin J, Temmerman L, Sampson RD, Gallego-Colon E, Barberi L, Bilbao D, et al. Monocyte/Macrophage-derivedIGF-1orchestratesmurineskeletal muscleregeneration and modulates autocrine polarization. Am Soc Gene Cell Ther 2015;23(7):1189-1200.

5. García-Segura LM, Arévalo MA, Azcoitia I. Interactions of estradiol and insulin-like growth factor-I signaling in the nervous system: New advances. Prog Brain Res 2010;181:251-272.

Pecu 2023;14(1):122-136

6- García-Lacy F, Gutiérrez-Olvera L, Bernad M, Fortier L, Trigo-Tavera FJ, GómezChavarín M, et al. Pharmacokinetic analysis of intraarticular injection of insulin and its effect on IGF-1 expression in synovial fluid of healthy horses. Rev Mex Cienc Pecu 2022;13(2):391-407.

7. Fortier LA, Balkman CE, Sandell LJ, Ratcliffe A, Nixon A. Insulin-like growth factor-1 gene expression patterns during spontaneous repair of acute articular cartilage injury. J Orth Res 2001;19:720-728.

8. Frisbie D. Future directions in treatment of joint disease in horses. Vet Clin Equine 2005;21:713-724.

9. Aguilar IN, Trippel SB, Shuiliang S, Bonassar LJ. Comparison of efficacy of endogenous and exogenous IGF-I in stimulating matrix production and mature chondrocytes Cartilage 2015;6(4):264-272.

10. Moyer W, Schumacher J, Schumacher J. A guide to equine joint injection and regional anesthesia. Yardley, PA: Veterinary Learning Systems, USA. 2007.

11. Verwilghen D, Busoni V, Gangl M, Franck T, Lejeune JP, Vanderheyden L, et al. Relationship between biochemical markers and radiographic scores in the evaluation of the osteoarticular status of Warmblood stallions. Res Vet Sci 2009;87(2):319-328.

12. Chu CR, Beynnon BD, Buckwalter JA, Garrett WE Jr, Katz JN, Rodeo SA. Closing the gap between benck and nedside research for early arthritis therapies (EARTH): report from the AOOSSM/NIH U-13Post-joint injuryosteoarthritis conference II. AmJ Sports Med 2011;39(7):1569-1578.

13 Bonnevie ED, Delco ML, Fortier LA, Alexander PG, Tuan RS, Bonassar LJ. Characterization of tissue response to impact loads delivered using a hand-held instrument for studying articular cartilage injury. Cartilage 2015;6(4):226-232.

14. Buzas EI, Gyrgöry B, Nagy G, Falus A, Gay S. Emerging role of extracellular vesicles in inflammatory diseases. Nat Rev Reumatol 2014;10(6):3563-3564. 15. www.uniprot.com. . https://www.afternic.com/forsale/uniprot.com?utm_source=TDFS&utm_medium=sn_ affiliate_click&utm_campaign=TDFS_Affiliate_namefind_direct8&traffic_type=CL3 &traffic_id=Namefind. 16. Phillipou A, Papageorgiou E, Bogdanis G, Halapas A, Sourla A, Maridaki M, et al Expression fo IGF-1 isoforms after exercise-induced muscle damage in humans: Characterization of the MGF E peptide actions in vitro. In vivo 2009;23:567-576.

17. Nixon AJ, Brower-Toland BD, Sandell LJ. Primary nucleotide structure of predominant and alternate splice forms of equine insulin-like growth factor I and their gene expression patterns in tissues. AJVR 1999;60(10):1234-1241.

18. Le, KQ, Prabhakar B, Hong, WJ. et al. Alternative splicing as a biomarker and potential target for drug discovery. Acta Pharmacol Sin 2015;36:1212–1218.

19. Geneser F. Tejido esquelético. Cap 12, Geneser F. Histología. 3rd ed. Madrid, España: Editorial Médica Panamericana; 2001.

20. Briggs DT, Sadr KN, Pulido PA, Bugbee WD. The use of osteochondral allograft transplantation for primary treatment of cartilage lesions in the knee. Cartilage 2015;6(4):203-207.

21. Sherman SL, Khazai RS, James CH, Stoker AM, Flood DL, Cook JL. In vitro toxicity of local anesthetics and corticosteroids on chondrocyte and synoviocyte viability and metabolism. Catilage 2015;6(4):233-240.

https://doi.org/10.22319/rmcp.v14i1.6273 Article

Use of Wharton's jelly-derived mesenchymal stromal cells for the treatment of equine recurrent uveitis: a pilot study

María Masri-Daba a*

Montserrat Erandi Camacho-Flores b

Ninnet Gómez-Romero c,d

Francisco Javier Basurto Alcántara c

a UniversidadNacionalAutónomadeMéxico.FacultaddeMedicina Veterinaria yZootecnia. Departamento de Medicina, Cirugía y Zootecnia para Équidos. Ciudad de México, México.

b UniversidadNacional AutónomadeMéxico. FacultaddeMedicinaVeterinaria yZootecnia. Posgrado en Ciencias de la Producción y de la Salud Animal. Ciudad de México, México.

c UniversidadNacionalAutónomadeMéxico.FacultaddeMedicina Veterinaria yZootecnia. Departamento de Microbiología e Inmunología. Ciudad de México, México.

d Comisión México-Estados Unidos para la prevención de fiebre Aftosa yotras enfermedades exóticas de los animales. Ciudad de México, México.

*Corresponding author: masri@unam.mx

Abstract:

Equine recurrent uveitis (ERU) is a disease that affects 2 to 25 % of equines worldwide, 56% of which go blind; therefore, it is considered the most common cause of blindness in horses. ERU is a spontaneous immune-mediated condition characterized by recurrent intraocular inflammatory events. Currently, there is no treatment for horses with this disease. Mesenchymal stromal cells (MSCs) derived from various tissues, such as Wharton's jelly (WJ), have demonstrated their ability to modulate the immune response by negatively regulating the inflammatory process. The objective of this pilot study was to evaluate the

effect of using MSCs derived from WJ as a treatment for ERU. The WJ was obtained and processed according to previously described methodologies for obtaining EMF. The horses involved in this study received a dose of 5x106 MSCs in the subpalpebral area. The research evaluated the concentration of interleukins (IL: IL-1, IL-2, IL-10, IFN-, and TNF) in tear samples obtained before treatment inoculation, 30 min after the inoculation, and 7 days post inoculation. No significant changes in IL concentration were observed suggesting a decrease in pro-inflammatory ILs. However, horses with ERU treated with MSCs exhibited a positive response to therapy, evidenced by a decrease in signs of ERU. The results obtained suggest that treatment of ERU with WJ-derived MSCs is a safe alternative with promising results.

Keywords: Wharton's Jelly, Mesenchymal Stromal Cells, Equine recurrent uveitis, Therapeutics.

Received: 27/06/2022

Accepted: 01/08/2022

Introduction

Mesenchymal stromal cells (MSCs) are characterized by their ability to differentiate into various cell lineages; therefore, theymaybe involved in the regeneration of damaged tissues. Another important characteristic of MSCs is that they have anti-inflammatory properties and regulate the immune response by producing a set of immunomodulatory factors such as interleukin 6 (IL-6), prostaglandin E2 (PEG2), and nitric oxide(1,2). The secretion of these factors inhibits the proliferation of activated T lymphocytes, reduces the secretion of proinflammatory cytokines, and increases the population of regulatory T lymphocytes (Tregs)(2,3,4) .

In equines, MSCs can be obtained from bone marrow, adipose tissue, amniotic membrane, umbilical cord blood, and foal umbilical cord tissue known as Wharton's jelly (WJ)(5). WJ is the primitive mucous connective tissue of the umbilical cord that lies between the amniotic epithelium and the umbilical vessels; it consists of a hyaluronic acid and chondroitin sulfatebased substance with a high concentration of MSCs(6). Due to their molecular characteristics, such as the absence of the expression of histocompatibility molecules I and II, these cells offer a unique advantage for autologous and allogeneic application(7) . In particular, WJ is considered an important source of MSCs, both in humans and in other species, with great potential in the therapeutics of various inflammatory and immune-mediated conditions, such as equine recurrent uveitis (ERU).

ERU,alsoknownasmoonblindness,isadiseaserecognizedastheleadingcause ofblindness in horses. It has been reported to be prevalent in 2 to 25 % of equines in the USA(8) . It is characterized by recurrent episodes of intraocular inflammation or low levels of persistent inflammation,predominantlyintheiris,ciliarybody, andchoroid(9) Thisdiseasehasanacute presentation that includes signs such as miosis, decreased intraocular pressure, and iris adhesions, while its chronic presentation results in the development of cataracts, glaucoma, and blindness.

The triggering or etiological factors of ERU remain unknown; however, it has been reported that genetic components as well as Leptospira interrogans infections may be involved in the development of this condition(8,10). Subsequently, the signs that occur in ERU are the result of T lymphocyte activation, specificallyTh1 and Th17, causing destruction of the uveal tract of the eye(11,12,13). Currently, there is no cure for ERU; therefore, treatment focuses on decreasing inflammation with the goal of preserving vision, limiting the recurrence of episodes, and reducing pain with anti-inflammatory and mydriatic drugs(14) .

It has been shown that the use of MSCs in immune diseases of dogs, cats, and horses can induce the switch of proinflammatory T lymphocyte subsets to regulatory T lymphocytes(15,16,17). Therefore, the use of WJ-derived MSCs in the treatment of ERU is a promising alternative. This article describes the procurement, culture, characterization and differentiation of MSCs derived from umbilical cord WJ of foals at foaling and their preliminary use in horses with ERU.

Material and methods

MSC procurement and cultivation

WJ MSCs were collected from 26 foals of full English blood mares (EBM) aged 5 to 20 yr. The umbilical cords were collected was performed following the delivery of the placenta. They were handled and processed under sterile conditions at the Tissue Engineering, Cell Therapy, and Regenerative Medicine Unit of the National Rehabilitation Institute of Mexico.

In short, a 15 to 20 cm fragment of each cord, still wrapped in amnion, was taken, followed by two washes with iodine solution interspersed with washes of sterile physiological saline solution (PSS). The sections were then cut to approximately 5 cm and stored at 4 ºC in phosphate buffered solution (PBS) with penicillin (10,000 U/ml), amphotericin B (25 µg/ml) and streptomycin (10,000 µg/ml)forprocessing in thelaboratory.Subsequently, the WGwas separated from the umbilical cord tissue and deposited in a Petri dish with PBS where cuts were made to facilitate enzymatic digestion. The latter was carried out in 10 ml of Dulbecco's modified Eagle's medium (DMEM) solution with collagenase (0.8 mg/ml) in incubation at

37 ºC for 1 hr. After the incubation time had elapsed, the cells were centrifuged at 700 xg for 7 min at 37 ºC, the supernatant was decanted and the cells were resuspended in DMEM supplemented with 10 % fetal bovine serum (FBS) and 1% penicillin, amphotericin B and streptomycin (10,000 U/ml; 25 µg/ml; 10,000 µg/ml). The primary culture was maintained in 25 cm2 cell culture bottles incubated with 5 % CO2 at 37 ºC, and three passages were performed once 80 % confluence was reached.

EMF characterization

Cells obtained before the third passage were subjected to surface phenotype evaluation in order to corroborate their mesenchymal profile byflow cytometry. It was used 2.5 x 105 cells containedin round-bottom polystyrenetubes resuspended in 1ml PBS.For cell labeling, cells were incubated for 1h with specific primary antibodies for detection of CD90 (FITC Mouse Anti-Human CD90 Clone 5E10 555595), CD73 (APC Mouse Anti-Human CD73 Clone AD2560847), CD105 (PE Mouse Anti-Human CD105 Clone 266 560839), CD45 (FITC Mouse Anti-Human CD45 Clone G44-26 555478), CD34 (PE Mouse Anti-Human CD166 Clone 34 559263), CD14 (PerCP Mouse Anti-Human CD14 Clone MφP9 340585), and MHC-II (APC Mouse Anti-Human HLA-DR Clone G46-6 559866). Cells were then washed twice and analyzed using the FACS-Calibur Becton and Dickinson flow cytometer.

EMF differentiation

WJ MSCs were grown in 12-well plates at a density of 5x104 using DMEM supplemented with 5% SFB and 1% antibiotic, under the same culture conditions as previously mentioned. After 48 h, the culture medium was replaced by adipogenic, osteoblastic, and chondrogenic medium as described below.

For the adipose lineage induction, after 48 h of incubation the cell culture medium was replaced by differentiation medium formulated with DMEM supplemented with 0.5% SFB, dexamethasone (1 mM), 3-isobutylmethylxanthine (0.5 mM), insulin (10 %), indomethacin (50 mM), as well as penicillin (10,000 U/ml), amphotericin B (25 µg/ml), and streptomycin (10,000 µg/ml). Medium changes were performed every third day for 21 d. Finally, cell differentiation was evaluated using Nile red staining.

MSC differentiation into osteoblastic lineage was carried out using DMEM culture medium supplemented with SFB 1%, and with dexamethasone (100 nM), ascorbic acid (0.05 mM), 10 mM/L -glycerophosphate and BMP-7 (10 ng/ml). Likewise, the medium was changed everythird day for 21 d, and osteogenic differentiation was evaluated byVon Kossa staining.

The chondrogenic lineage was obtained by using DMEM culture medium added with insulin (10 %), ascorbic acid (1 mg/ml), transforming β growth factor (TGF-β) (10 ng/ml), sodium pyruvate (1 %), and bone morphogenic protein 2 (BMP-2) (100 ng/ml). The evaluation of the differentiation was carried out with Alcian blue staining.

Horses

A total of 15 EBM horses aged 2 to 7 yr were included for this study. Twelve clinically healthyhorses were used as a control group. As part of the experimental group, 3 horses with at least one episode of ERU with characteristic signs such as miosis, iris hyperpigmentation, blepharospasm, corneal edema, aqueous flame, hypopyon, hyphema, epiphora, photophobia, fibrin in the anterior chamber, conjunctival hyperemia, and scleral injection were considered. The horses used in this study underwent a strict general physical and complete ophthalmological examination consisting of threat reflex assessment, pupillary response, consensual reflex, Schirmer's test, corneal sensibility, flourescein stain, Jones test, rose Bengal stain and fundus observation.

Tear sample collection and EMF inoculation

Once the control and experimental groups were formed, the horses were sedated using intravenous xylazine at a dose of 0.3 a 0.5 mg/kg. Subsequently, 100 µl of tears were collected using a sterile capillary tube without additives; the sample was placed in sterile vials and stored at -80 ºC until use.

Using an insulin syringe, the inoculum, PBS and 5x106 MSCs were collected for six horses of the control group, as well as for the three horses with ERU of the experimental group; both in a volume of up to 200 l. This part of the procedure was carried out under sterile conditions. Prior to inoculum application, the area was aseptically cleaned with alcohol, avoiding direct contact with the eye. A 25 G needle was inserted into the subpalpebral area and the syringe was connected to inoculate the contents. The second and third tear samples were taken 30 min and one week post inoculation, respectively.

Evaluation of interleukins in tear samples

The evaluation of interleukins IL-1, IL-2, IL-10, IFN- and TNF in tear samples obtained before and after the treatment was performed by multiplex enzyme-linked immunoassay (ELISA), in order to determine whether there are changes in the pattern of interleukins detected.

For the multiplex ELISA test, the Equine Cytokine/Chemokine Magnetic Bead Panel Milliplex MAP Kit was used, following the manufacturer's instructions on the Luminex equipment (Bio-Plex 200, Bio-Rad Laboratories, EU). Briefly, 200 l of wash buffer was added to each of the 96 wells of the plate, the plate was covered and incubated in agitation for 10 min at 20 ºC; the contents of the wells were then discarded, removing the excess by turning the plate over and tapping it on a bed of absorbent towels. Subsequently, 25 l of the standard and cores were added to the corresponding wells. 25 l of assay buffer were added to the sample wells. Subsequently, 25 l of matrix solution were added to all wells, and 25 l of tear sample were added to the corresponding wells. Finally, 25 l of the bead mixture were added to each of the wells of the plate, which was incubated for 18 h under agitation at 4 ºC covered with aluminum foil.

After the incubation time had elapsed, the entire contents of the plate were discarded, and the plate was washed three times. Subsequently, 25 l of interleukin detection antibodies were added to all wells, the plate was sealed and incubated in agitation at room temperature for 1 h. Then 25 l of streptavidin-phycoerythrin were added to all wells, the plate was sealed and incubated in agitation at room temperature for 30 min. Once this step was completed, the contents were again decanted, and three washes were performed. To each of the wells, 150 l of "Seath Fluid" were added and agitated for 5 min at room temperature. Finally, the plaque was read to estimate the concentration of interleukins detected in the tear samples corresponding to the three sampling times.

Statistical analysis

For the analysis of the results of the interleukin concentration, nonparametric statistics were used according to the results obtained from the homogeneity of variance (analysis of residuals) and normal distribution (Shapiro-Wilk test) tests included in the Prism 8.0 statistical program (GraphPad, Software Inc., EEUU).

In particular, the Mann Whitney U test was used to compare the concentration of cytokines between the media used (PBS and MSC) in the control group. Subsequently, this test was repeated to compare the concentration of cytokines between the control and experimental groups. Additionally, the Kruskal Wallis test followed by Dunn's multiple comparison test was performed to determine possible changes in cytokine concentration at the different times evaluated (baseline, after 30 min, and after 7 d) in both groups. In all cases, a value of P<0.05 was considered significant.

Results

MSC procurement and cultivation

Primary cultures of MSCs obtained from WJ initially showed a rounded morphology and clustered in clusters of up to 100 m. Once attached, after 120 h they acquired fibroblastoid morphology (Figure 1).

Figure 1: Primary culture of mesenchymal stromal cells obtained from Wharton's jelly

a) Rounded morphology (white arrow) of MSCs and formation of cell clusters (black arrow); b) adherence and fibroblastoid morphology of MSCs (black arrow); c) 80 % confluence of the EMF monolayer with characteristic morphology (black arrow); c) 80 % confluence of the EMF monolayer with characteristic morphology.

Characterization of EMF

The identification of positive MSC surface markers was carried out by flow cytometry. CD90, CD73, and CD105 markers should be expressed, while negative markers are CD14, CD34,CD45,and MHC-II. Figure2 shows the phenotype characterization ofMSCs obtained from WJ. The lack of expression of CD45, CD34, and CD14 markers is shown; so is the expression of the positive markers CD73 and CD90. On the other hand, the MScs samples evaluated showed a reduction in the expression of the CD105 marker.

Figure 2: Phenotype of cultured MSCs

Flow cytometry analysis of cultured WJ-derived MSC protein expression labeled with anti CD45 (green), CD34 (turquoise), CD14 (pink), CD73 (red), and CD90 (blue) antibodies. The histogram in purple indicates the intensity of the fluorescence of MSCs labeled with the control antibody. Open histograms indicate positive reactivity with the indicated antibody.

MSC differentiation

MSCs were treated with three formulations of culture medium to evaluate their differentiation into adipose, osteoblastic, and chondrogenic cell lineages. Differentiation was assessed by cell staining (Figure 3); shown are representative images of a) adipocytes stained with Nile red to detect fattyacid vacuoles within cells; b) osteoblasts stained with Von Kossa stain to identify calcium deposits; c) chondrocytes detected with Alcian blue stain, mucopolysaccharide staining is visualized in the extracellular matrix of this tissue.

Figure 3: MSC differentiation to a) adipose, b) osteoblastic, and c) chondrogenic lineage

Evaluation of horses with ERU

Horse 1. Warmblood, 12-yr-old castrated male Retinto. He presented signs of uveitis and started treatment with betamethasone, cyclosporine A, and artificial tears. Subsequently, the MSC treatment protocol was started, and onlythe left eye was injected; throughout the week, no clinical changes were observed.

Horse 2. Friesian, 15-yr-old whole male. He exhibited acute uveitis in the left eye, and had been previouslytreated withprednisoloneandcyclosporineA.He showedsigns such as pain, edema, vascularization, epiphora, and blepharospasm. (Figure 4). One week after the MSC treatment, he exhibited improvement from the clinical point of view; all the previous signs decreased slightly, and he showed a better mood.

Figure 4: Horse 2

a) shows blepharospasm and epiphora, b) shows edema and vascularization; palpebral margins are green due to previous fluorescein staining. c) shows a decrease in blepharospasm of the left eye d) shows decreased edema.

Horse 3. Apaloosa, entire male, 21 yr old. He exhibited an acute condition, but received no treatment. Both eyes had miosis, epiphora, corneal edema, and neovascularization, scleral and conjunctival injection, both eyes were treated with MSCs and maintained for one week with a topical mydriatic. At 7 d both eyes were mydriatic, without pain, minor epiphora, edema, and conjunctival and scleral injection. Three days after the second visit, he started medical treatment, but he exhibitedevenlessedemaand was much more comfortable (Figure 5)

Figure 5: Horse 3

a) and b) The pictures show acute signs of uveitis of both eyes, with edema, neovascularization, and conjunctival hyperemia. c) and d) show decreased edema and hyperemia.

Evaluation of interleukins in tear simples

Comparison between the media used (PBS and MSC) for the control group showed no differences in anyof the interleukins measured. However, when comparingthe concentration of these at different times (basal, after 30 min and after 7 d), it was found that the concentration of IL-1⍺ with the use of PBS as vehicle, showed statistically significant differences between the basal measurement and the one performed 7 d later (Table 1).

Since no statistically significant differences were found between the two media applied to the control group, the measurements were considered within a single group and contrasted with the data obtained in the experimental group. In this regard, no statistically significant

differences were found in the concentration of ILs at any of the three times when they were used when comparing the control group with the treated experimental group.

Discussion

In the field of regenerative medicine, interest in MSC research has increased over the last decade.Thesecells, also knownas mesenchymalprogenitorcells, have the abilityto promote tissue regeneration, modulate the immune response, and regulate the inflammatory process(18). They are also considered to be cell populations with the ability to self-renew and differentiate into various types of connective tissue cells(19). Consequently, they have the potential to act at sites of inflammation by synthesizing interleukins that participate in the modulation of this process(20)

Specifically, previous studies have described the immunomodulatory effect of MSCs in horses; additionally,it has beenshownthat equine-derivedMSCs, compared to thoseofother species, have a greater capacity to inhibit the proliferation of activated T lymphocytes and to decrease the production of IFN- and TNF(1,3). Likewise, it has been reported that the use of equine MSCs as a treatment induces lymphocyte apoptosis and reduces IL-2 receptor (CD25) expression in lymphocytes T CD4+ .

In equine medicine, they are currently used, above all, to treat diseases of the locomotor system, skin wounds, equine metabolic syndrome, asthma, laminitis, neurological, and ophthalmological problems(21). Within the latter, ERU is considered the leading cause of blindness in equines and is described as an autoimmune inflammatory disease with characteristics similar to human uveitis(22). Horses suffering from ERU are characterized by an inflammatory phenotype of Th1 lymphocytes (CD4+ IFN-); therefore, the described data suggest that the use of MScs is a suitable alternative in the treatment of ERU, as well as other immune-mediated diseases(23). Together, ocular therapeutic benefits have been documented in equine and other species(24,25,26) .

In particular, it has been proven that the application of MSCs in rabbits with corneal surface damage accelerates the corneal healing process, reduces oxidative stress, and suppresses the production of proinflammatory interleukins, resulting in a decrease in corneal opacity and neovascularizationofthe affected area(27). In rats, theuse ofMSCs in the treatment of corneal burns and reconstruction of the corneal surface has yielded positive results(25)

On the other hand, its use in horses as a treatment for immune-mediated keratitis has yielded promisingresults; it was observedthat3out of4horses submittedto this therapyhadpositive results evidenced by the decrease in opacity, irregularity, and vascularization on the corneal surface; in addition to maintaining the corneal disease stable for up to one year after MSC

treatment. Therefore, it is considered a new alternative immunomodulatory therapy for this condition(24). Likewise, in another study on equine immune-mediated keratitis, a satisfactory responseto MSC inoculation in theophthalmicarteryandsubconjunctival topical application three times a day for three weeks were reported(28)

In the present study, was evaluated parameters that would indicate the benefits of subpalpebral administration of EMF in the treatment of ERU. Once the treatment with WJderived MSC was applied in horses with ERU, tear samples were analyzed to evaluate the pattern of interleukins present in the sample before and after treatment (30 min and 7 d after). However, no significant changes in interleukin concentrations were observed at the different times evaluated where a decrease in proinflammatory interleukins and an increase in antiinflammatoryinterleukins were expected, as that horses with ERU are said to maintain a high concentration of IL-10, IL-1, IFN-, IL-6, and IL-17 in tears(23) .

These results maybe associated with the route of administration, where MSC inoculation via the subpalpebral route shows less efficacy in resolving the condition, probably because it reaches the site of action (eye) in an insufficient proportion, in an inadequate dosage and frequencyoftreatmentapplication,andat aninadequatestageofthedisease,thereforehaving a low capacity to affect the inflammatory process at that level, since most reports indicate a significant improvement when the treatment is applied at the acute phase of the disease(23) .

On the other hand, the absence of proinflammatory IL detection in tear samples may be due to the peak concentration occurring at different periods of the disease than those evaluated in this study.

These findings are useful when choosing the route of administration for MSC treatment. Although success stories of the use of the subconjunctival route have been described, there is a need to evaluate and compare additional routes of administration that may provide better results for short- and long-term efficacy. In contrast, intravenous administration of MSCs has been reported to be completely safe; however, it is not known whether or not it yields better results(29). Therefore, further studies are needed to establish the necessary conditions for the treatment of ERU.

Subsequently, in the three patients with ERU who were part of the present pilot study, the effect of treatment with subpalpebral MSCs was evaluated. No clinical changes indicating improvement were observed in horse 1. In contrast, horse 2 showed improvement of the clinical signs presented (pain, edema, vascularization, epiphora, and blepharospasm), including improvement in mood 7 d after post-treatment. In the case of horse 3, there was improvement in some clinical signs such as less epiphora, reduction of edema and conjunctival and scleral injection, as well as improvement in mood.

Two of the three treated horses showed improvement and decrease in clinical signs of ERU 7 days post treatment. The positive results obtained when applyingequine-derived WJ MSCs highlight the importance of further studies to establish a uniform treatment and the development of an efficient MSC application protocol to achieve better results. For example, treatment options promoting a long-lasting response and treatment efficacy are enhanced by a specific number of, or multiple applications with a standardized MSC dose, as well as by co-application with local immunosuppressive therapy(30)

Conclusions and implications

This pilot study describes the experimental use of WJ-derived MSCs for the treatment of ERU. Although there were certain limitations, e.g. in the number of animals analyzed that might allow us to reach firm conclusions, the obtainment of positive results in the respective clinical presentations without generating adverse effects reaffirms the use of MSCs as a viable alternative to the treatment of ERU. Despite its promising results, controlled studies of MSC treatment must be carried out in order to demonstrate and confirm the benefits of the MSC treatment for ERU.

Financing

Funding for the study was granted by Project No. IN228919 of the Support Program for Research and Technological Innovation Projects (Programa de Apoyo a Proyectos de Investigación e Innovación tecnológica, PAPIIT), “Use of allogenous stem cells for the treatment of equine recurrent uveítis” ("Uso de células troncales alógenas para el tratamiento de uveítis recurrente equina") of the Facultyof VeterinaryMedicine and Animal Husbandry-Universidad Nacional Autónoma de México (Facultad de Medicina Veterinaria y Zootecnia-Universidad Nacional Autónoma de México).

Conflict of interest

The authors declare that they have no conflict of interest in regard to the publication of this article.

Literature cited:

1. Carrade DD, Lame MW, Kent MS, Clark KC, Walker NJ, Borjesson DL. Comparative analysis of the immunomodulatory properties of equine adult-derived mesenchymal stem cells. Cell Med 2012;4(1):1-11. doi: 10.3727/215517912X647217.

Pecu 2023;14(1):137-153

2. Martínez‐Montiel MDP, Gómez‐Gómez GJ, Flores AI. Therapy with stem cells in inflammatory bowel disease. World J Gastroenterol 2014;20:1211‐1227.

3. Carrade Holt DD, Wood JA, Granick JL, Walker NJ, Clark KC, Borjesson DL. Equine mesenchymal stem cells inhibit T cell proliferation through different mechanisms depending on tissue source. Stem Cells Dev 2014;23:1258‐1265.

4. Le Blanc K, Davies LC. Mesenchymal stromal cells and the innate immune response. Immunol Lett 2015;168:140‐146.

5. Iacono E, Rossi B, Merlo B. Stem cells from foetal adnexa and fluid in domestic animals: an update on their features and clinical application. Reprod Dom Anim 2015; 50:353–64. doi: 10.1111/rda.12499

6. Weiss ML, Troyer DL. Stem cells in the umbilical cord. Stem Cell Rev 2016;2:155–162.

7. Iacono E, Pascucci L, Rossi B, Bazzucchi C, Lanci A, Ceccoli M, et al. Ultrastructural characteristics and immune profile of equine MSCs from fetal adnexa. Reproduction 2017;154:509–519. doi: 10.1530/REP-17–0032.

8. Gilger BC, Hollingsworth SR. Diseases of the uvea, uveitis, and recurrent uveitis. In: Gilger BC, editor. Equine ophthalmology. Hoboken, NJ: John Wiley & Sons, Inc. 2016: 369–415. doi: 10.1002/9781119047919.ch8.

9. Gilger BC, Deeg C. Chapter 8‐Equine recurrent uveitis. Gilger BC editor. Equine Ophthalmology, 2nd ed. Saint Louis, MO: W.B. Saunders; 2011:317‐349

10. SauvageAC,MonclinSJ,ElansaryM, HansenP,GrauwelsMF.Detectionofintraocular Leptospira spp. by real-time polymerase chain reaction in horses with recurrent uveitis in Belgium. Equine Vet J 2019;51:299–303.

11. Deeg CA. Ocular immunology in equine recurrent uveitis. Vet Ophthalmol 2008;11(Suppl 1):61‐65.

12. Deeg CA, Ehrenhofer M, Thurau SR, Reese S, Wildner G, Kaspers B. Immunopathology of recurrent uveitis in spontaneously diseased horses. Exp Eye Res 2002;75:127‐133.

13. Gilger BC, Malok E, Cutter KV, Stewart T, Horohov DW, Allen JB. Characterization of T‐lymphocytes in the anterior uvea of eyes with chronic equine recurrent uveitis. Vet Immunol Immunopathol 1999;71:17‐28

14. Gilger BC, Michau TM. Equine recurrent uveitis: new methods of management. Vet Clin North Am Equine Pract 2004;20:417–27. doi: 10.1016/j.cveq.2004.04.010.

15. Kol A, Walker NJ, Nordstrom M, Borjesson DL. Th17 pathway as a target for multipotent stromal cell therapy in dogs: Implications for translational research. PLoS One 2016;11:e0148568.

16. Arzi B, Mills-Ko E, Verstraete FJM, Kol A, Walker NJ, BadgleyMR, et al. Therapeutic efficacy of fresh, autologous mesenchymal stem cells for severe refractory gingivostomatitis in cats. Stem Cells Transl Med 2016;5:75–86.

17. Holt DDC, Wood JA, Granick JL, Walker NJ, Clark KC, Borjesson DL. Equine mesenchymal stem cells inhibit T cell proliferation through different mechanisms depending on tissue source. Stem Cells Dev 2014;23:1258–1265.

18. Stewart MC, Stewart AA. Mesenchymal stem cells: characteristics, sources, and mechanisms of action. Vet Clin North Am Equine Pract 2011;27(2):243-61. doi: 10.1016/j.cveq.2011.06.004.

19. Madrigal M, Rao KS, Riordan NH. A review of therapeutic effects of mesenchymal stem cell secretions and induction of secretory modification by different culture methods. J Transl Med 2014;12:260.

20. Meirelles LS, Fontes AM, Covas DT, Caplan AI. Mechanisms involved in the therapeutic properties of mesenchymal stem cells. Cytokine Growth Factor Rev 2009;20(5-6):419-27. doi: 10.1016/j.cytogfr.2009.10.002.

21. CequierA,Sanz C,RodellarC,Barrachina L.The usefulness ofmesenchymal stem cells beyond the musculoskeletal system in horses. Animals (Basel). 2021;11(4):931. doi:10.3390/ani11040931.

22. Malalana F, Stylianides A, McGowan C. Equine recurrent uveitis: Human and equine perspectives. Vet J 2015;206(1):222-9. doi: 10.1016/j.tvjl.2015.06.017.

23. Saldinger LK, Nelson SG, Bellone RR, Lassaline M, Mack M, Walker NJ, Borjesson DL. Horses with equine recurrent uveitis have an activated CD4+ T-cell phenotype that can be modulated bymesenchymal stem cells in vitro. Vet Ophthalmol 2020;23(1):160170. doi: 10.1111/vop.12704.

24. Davis AB, Schnabel LV, Gilger BC. Subconjunctival bone marrow-derived mesenchymal stem cell therapy as a novel treatment alternative for equine immunemediated keratitis: A case series. Vet Ophthalmol 2019;22(5):674-682. doi: 10.1111/vop.12641.

25. Jiang TS, Cai L, Ji WY, Hui YN, Wang YS, Hu D, Zhu J. Reconstruction of the corneal epithelium with induced marrow mesenchymal stem cells in rats. Mol Vis 2010;14;16:1304-16.

26. Dodi PL. Immune-mediated keratoconjunctivitis sicca in dogs: current perspectives on management. Vet Med (Auckl) 2015;30;6:341-347. doi: 10.2147/VMRR.S66705.

27. Cejkova J, Trosan P, Cejka C, Lencova A, Zajicova A, Javorkova E, Kubinova S, Sykova E, Holan V. Suppression of alkali-induced oxidative injury in the cornea by mesenchymal stem cells growing on nanofiber scaffolds and transferred onto the damaged corneal surface. Exp Eye Res 2013;116:312-23. doi: 10.1016/j.exer.2013.10.002.

28. Marfe G, Massaro-Giordano M, Ranalli M, Cozzoli E, Di Stefano C, Malafoglia V, Polettini M, Gambacurta A. Blood derived stem cells: an ameliorative therapy in veterinary ophthalmology. J Cell Physiol 2012;227(3):1250-1256. doi: 10.1002/jcp.22953.

29. Kol A, Wood JA, Carrade HDD, Gillette JA, Bohannon-Worsley LK, Puchalski SM, et al. Multiple intravenous injections of allogeneic equine mesenchymal stem cells do not induce a systemic inflammatory response but do alter lymphocyte subsets in healthy horses. Stem Cell Res Ther 2015;6(1):73. doi: 10.1186/s13287-015-0050-0.

30. Zhang J, Huang X, Wang H, Liu X, Zhang T, Wang Y, Hu D. The challenges and promises of allogeneic mesenchymal stem cells for use as a cell-based therapy. Stem Cell Res Ther 2015;6:234. doi: 10.1186/s13287-015-0240-9.

Rev Mex Cienc Pecu 2023;14(1):137-153

Table 1: Measurement of interleukins by median and range technique (<Quantification level)

Control group(n=12) Experimental group(n=3)

Basal 30 minutes 7 days Basal 30 minutes 7 days Median Range Median Range Median Range Median Range Median Range Median Range

IL-1α 43.27 38.7153.20 41.62 36.9551.00 38.71* 36.1746.36 40.39 38.3145.32 41.26 35.1943.78 38.31 36.1739.06

IFN-γ 33.94 7.450175.3 18.52 3.900186.6 28.3 3.900341.5 93.12 25.2297.09 45.18 11.1795.31 108.3 14.78329.7 IL-2 1.323 0.453410.75 0.781 0.1316 -14.56 0.8325 0.3679 - 30.42 1.383 0.5992 - 1.729 0.8926 0.42291.611 1.516 0.93072.082 IL-10 19.52 6.70970.25 19.43 7.69446.30 17.84 6.70952.23 18.59 10.3419.46 12.43 7.85714.38 14.89 11.4928.56

TNF-α 14.63 2.20056.80 2.998 2.20052.03 25.61 2.20070.88 11.53 2.20019.22 2.797 2.20015.64 13.86 8.01517.99

IL-1⍺ (*) detected 7 d post-treatment showed a statistically significant decrease compared to the baseline measurement (P=0.0356). Values are reported in pg/ml.

https://doi.org/10.22319/rmcp.v14i1.5537 Article

Scale of production and technical efficiency of beef cattle farming in Puebla, Mexico

José Luis Jaramillo Villanueva a*

Lissette Abigail Rojas Juárez a Samuel Vargas López a

a Colegio de Postgraduados Campus Puebla. Boulevard Forjadores de Puebla No. 205, Santiago Momoxpan, Municipio de San Pedro Cholula, 72760, Puebla, México

*Corresponding authotr jaramillo@colpos.mx

Abstract:

The objective of this study was to estimate the degree of technical efficiency and identifythe factors of inefficiency of beef cattle production in the Sierra Norte of Puebla, Mexico. The data were generated bysurveying a statistical sample of 180 bovine production units (BPUs). Technical efficiency was estimated using the Stochastic Production Frontier and the explanationofinefficiencywasestimatedwithamultiplelinearregressionmodel.Theresults indicatethatthe sizeoftheBPU is positivelycorrelatedwith efficiency; the small BPU group showed an average efficiency of 0.72, the medium ones 0.75 and the large ones 0.85. Feed and labor costs can be reduced, while maintaining the same level of production. The significant (P≤0.05) explanatoryvariables of inefficiencyare schooling, technical assistance, experience, and administrative management.

Key words: Cattle, Technical efficiency, Production scale, Production frontier.

Received: 04/10/2019

Accepted: 17/09/2020

Introduction

According to official data(1), in 2017 Mexico produced 3.5 million tons of live cattle and 1.9 million tonnes of beef. National consumption for 2019 was 1.83 million tonnes. National production, in the last 15 yr, shows a mean growth rate (MGR) of 1.6 %, while demand grew at a MGR of 0.21, reflecting a fall in consumption, explained by the increase in prices(2). In this regard, per capita consumption went from 18 kg in 2007 to 15.1 in 2017. However, in 2017 imports totaled 136 thousand tonnes(3) .

In Mexico, non-specialized beef production presents difficulties in being profitable, especiallythat carriedout bysmallandmedium-sizedbovineproductionunits(BPUs),which obtain negative or very low rates of return(4). This type of BPU was one million in 2018. Accordingto the 2014 National Agricultural Survey (5), 62 % of the BPUs have 1 to 10 heads, 26 % from 11 to 35, 9.9 % from 36 to 120, and 1.6 % more than 120 heads. Therefore, approximately 88 % of BPUs are small. Given the importance of this sector and of cattle to generate family income, it is necessary to support their development through the analysis of the technical-economic factors that have a greater impact on their productivity(6)

A factor that negatively affects the economic profitability of small farmers is the low productivity and technical efficiency at the level of BPU(7). Another important factor is the growth rate of inputs, which is higher than that of the price of the output(8). Therefore, the challenges posed by the problems described can be addressed through the improvement of the productive efficiency of BPUs. Productive efficiency can improve the profitability of BPUs through lower costs and greater supply to the market.

Productive efficiency(9) is defined as the situation in which a cattle production unit (CPU) that produces a single product can improve its production only if it increases the use of at least one of its inputs. The literature on efficiency focuses on two aspects; measurement of technical and economic efficiency and sources of inefficiency. Efficiency studies have been carried out in a wide variety of agricultural production activities; grains(10); vegetables(11) , dairy(12), and coffee(13). In the world, few studies have addressed efficiency in beef cattle(14,15,16). In these it was found that there are significant deviations from the efficient production frontier.

InMexico,Morales-Hernández et al(17) conductedtheonlyavailablestudyof beefproduction efficiency in Mexico. They found that for small producers, as factors of production increase by a certain proportion, production grows less than proportionally. On the other hand, for the large ones, as the factors increased by a certain proportion, production grew in greater proportion. It is not necessaryto increase the amount of feed or the area of pasture to increase

the total amount of beef, but the number of animals.

The study of the efficiency of BPUs and the sources of inefficiency are therefore important from a practical and political point of view. On the one hand, farmers could use this information to improve the productivityof their farm. On the other hand, policymakers could focus interventions to improve producer income(18)

The objective of this study was to address this gap in knowledge by estimating the degree of efficiency, and to identify the factors of inefficiency of beef cattle production in the Sierra Norte of Puebla, Mexico, from an econometric perspective.

Material and methods

For the present study, seven municipalities of the Sierra Norte of Puebla were selected (Table 1). The study area was located at coordinates 19° 59' 10'' and 20° 34' 20'' N; 97° 19' 97'' and 97° 47' 98'' W. The altitude ranged from 10 to 1,700 m asl. The climate is warm humid with abundant rainfall all year round, except the municipality of Xicotepec, which has a humid semi-warm climate. The vegetation is composed of pasture (35 %), jungle (13 %) and forest (6 %)(19). These municipalities contribute 32.1 % of cattle production at the state level(3)

The methodology consisted of four stages: the first was the knowledge of the region, where the survey of the area was carried out, and interviews were conducted with leading producers and technicians to know general aspects of cattle farming; the second was the design of the sampling, of a simple random type, with proportional distribution, according to the number of producers in each municipality. The population used corresponds to 60,020 BPUs, reliabilitywas 95 % and accuracy was 7.5 % of the herd size mean, resulting in a sample size of 180 BPUs. The third stage consisted of the design, testing and application of questionnaires, distributed proportionally in the municipalities of the study (Table 1). The fourth stage was the statistical analysis of the data derived from the questionnaire, which were organized into sociodemographic, technological, and economic variables.

Table 1: Distribution of the sample size

Municipality Population (N) Participation (%) Sample (n)

Francisco Z. Mena 6791 11.31 54

Venustiano Carranza 11898 19.82 36

Tenampulco 3909 6.51 27 Pantepec 17919 29.86 20

Xicotepec 4734 7.89 18

Jalpan 8860 14.76 14

Ayotoxco de Guerrero 5909 9.85 12 Total 60020 100 180

The economic characterization of the cattle production units with the aforementioned variables is very useful for producers, since it allows them to know the behavior of their company and they can make decisions in their activities to minimize costs, improve productivity and profitability of the company. Therefore, it is important to distinguish between accounting costs and economic costs.

The cost accounting perspective emphasizes expenditures incurred, historical costs and depreciation. Economic costs represent the opportunitycost of the factors of production. One way to differentiate between these two approaches is to analyze how the costs of various factors (labor, capital, or business services) and the accounting or monetary costs, which are the costs incurred by the production unit for the purchase of inputs and assets at market prices(20), are defined.

For the purposes of this research, the total costs (TC) are the result of the sum of fixed costs (FC) and variables costs (VC) (TC = FC + VC). Fixed costs are those charges assumed by the production unit regardless of its level of production, including the option of zero productions. Variable costs are those that change depending on the level of production of the LPU. Total costs include: the cost of total labor, based on the sum of eventual labor (brush clearing and fertilizer application), and permanent labor (commonly known as payment for the cowboy and the flotante), which they require annually for cattle handling; cost of inputs (feed, medicines and others); and the cost of machinery and equipment (including depreciation rate of each asset, considering a value of 10 % per year).

The basis for defining the strata of herd size was the segmentation of livestock units of SAGARPA(21), which considers a stratum A made up of 20 heads or less, stratum B from 21

to 50 heads, and stratum C made up of a herd greater than 50 heads. The above to serve the CPUs in a differentiated way. Once the groups were formed, the following were carried out: econometric analysis; estimation of the stochastic production frontier and estimation of an explanatory model of inefficiency.

Stochastic frontier model

The assumption of a production of a stochastic nature means that the level of production of a unit of production is limited superiorly by a stochastic frontier, which can be modeled as in Equation 1: �� =��(��)+��,�� =�� �� (1) Where the error term is composed of two parts; a random perturbation v, symmetric that is assumed to be identically and independentlydistributed with mean 0, and u is a non-negative error term, which is distributed independently of v, following a one-tailed distribution(22) The random component represents events that are not controllable by the CPU (climatic, social, economic, and political phenomena), while u collects the distance of each company to its stochastic frontier, representing a measure of technical inefficiency(23). Therefore, the Stochastic Production Frontier (SPF) is described by Equation 2: ��∗=��(��)+�� (2) For SPFs, the technical efficiency index for enterprise i can be calculated with Equation 3: ������ = ���� ��(��)+���� (3)

The SPF is first proposed in the 1970s of the last century(24,25) where they considered(24) the case in which u is semi-normally distributed, that is, �� |��(0,����)| and v normally distributed.Theimplications at theconceptuallevel of PFbeingstochasticareveryimportant for the interpretation of inefficiency. As Schmidt(24) says, “the farmer whose harvest is devastated by drought or a storm is unfortunate with our measure, but inefficient with the usual measure”. An important limitation of the first estimates of SPF is that only the average efficiency of the sample was calculated, and it was not possible to obtain a measure of the efficiency of each company. Later developments(26) managed to find a measure of individual efficiency using the conditional distribution of u in ε. The technical efficiency index for each firm i is:

������ =������[ ��(����|��1)] (4)

The most commonly used measure of TE is the ratio of observed production and the corresponding stochastic production frontier, as in Equation 5:

������ = ���� ������(���� ´ ��+���� = ������(���� ´ ��+���� ����) ������(���� ´ ��+����) =������( ����) (5)

This measure of technical efficiency takes a value between zero and one. It measures the output of the i-th CPU relative to the output that a fully efficient CPU could produce using

the same input vector. The first step in calculating the TE is to estimate the parameters of the stochastic production frontier model:

Estimation of parameters

Because model 9.2 includes random terms; the symmetric error (vi) and a non-negative random variable (ui), the selected estimation method includes assumptions about both terms. Each vi is distributed independentlyof each ui and both are uncorrelated with the explanatory variables. Additionally, the noise component vi is assumed to have properties identical to those of the classical linear regression model. The inefficiency component has similar properties except that it has a non-zero mean (ui ≥0), so Ordinary Least Squares cannot be used. One solution is to make some distribution assumptions regarding the two error terms and estimate the model using the maximum likelihood (ML) method.

Half-normal model

ML estimators were obtained(24) under the following assumptions: vi=iidN(0,σv 2) and ui=iidN+(0,σu 2) . This indicates that the vi are normal random variables distributed independently and identically with means and variances zero and the ui are semi-normal random variables distributed independently and identically with scale parameter. That is, the probability density function (pdf) of each ui is a truncated version of a normal random variable that has zero mean and variance ���� 2

The log-likelihood function was parameterized(24) for this half-normal model in terms of ��2 =���� 2 +���� 2 and ��2 =���� 2/���� 2 ≥0. If �� =0, there are no technical inefficiency effects and all deviations from the frontier are due to noise. Using this parameterization, the maximum likelihood function is represented in Equation 6: In��(��|��,��,��)= 1 2 In (����2 2 )+∑ InΦ( ������ �� ) 1 2��2 ∑ ���� 2 1 ��=1 1 ��=1 (6) Where, y is an output vector; ���� ≡���� ���� ≡In���� ������ is the compound error term; and ��(��) is a cumulative distribution function (cfd) of the standard normal random variable evaluated at x.

The empirical analysis is based on the estimation of a Cobb-Douglas production function in which both output and inputs are expressed in logarithmic form (Equation 7), so that the estimated coefficients are interpreted as elasticities(27)

Ln(Yi)=β0+β1Ln(ARE)+β2Ln(LA)+β3Ln(ASS)+β4Ln(HEA)+β5Ln(FEED)+ε (7)

In this model, the dependent variable (Yi) is the value of cattle production of the CPUs. The explanatory variables are;

ARECAT is the area for cattle, in hectares owned by the BPU.

LA is the cost of the labor used in production.

ASS is the value of assets; value of machinery, equipment, and production facilities used in the cattle activity.

HEA is the expenditures in health; veterinary supplies and services.

FEED is the cost of feeding; cost of meadow maintenance and supplementary feeding.

Model of individual efficiencies

The estimated model of individual efficiencies (Equation 7) considers the measures of inefficiency estimated in the first stage as a dependent variable. Explanatory variables are a set ofvariablesthathypotheticallyaffect theperformanceof theCPU(6). The literaturereports as the most common explanatory variables the age of the head of the CPU, they level of schooling, experience in the activity under study, characteristics of the CPU, administration, and environmental factors, among the most cited(28-31). The multiple regression model was that described in Equation 8:

Ui=δ0+δ1Ln(Age)+δ2Ln(Schoo)+δ3Ln(Exper)+δ4Ln(Admon)+δ5Ln(TA)+ϑi (8)

Where: Age is the age of the head of the CPU; Schoo is the level of schooling (in years) of the head of the CPU; Exper are the years of experience in the cattle activity; Admon is a dummy variable that takes the value of zero if the CPU does not have an administration system and one if theyhave an administration system; TA is technical assistance, zero if they did not receive technical assistance and one if they received the service. The variables of the stochastic frontier model and of the individual inefficiency model are showed in Table 2.

Table 2: Variables used in the stochastic frontier production model Concepts Frequency Percentage

Gender of the head Woman 22 12.0 Man 162 88.0

Schooling of the head of the CPU Primary education 69 37.3 Junior High school 63 34.1 High school 33 17.8 Professional 20 10.8 Administration They do not have a system 114 61.6 They have a system 71 38.4

Technical assistance They did not receive 124 67.0 They did receive 55 33.0

Technological level Low 94 50.8 Medium 45 24.3 High 46 24.9

Strata [number of animal units (A.U.)] 20 or less 89 48.1 21 to 50 60 32.4 50 or higher 36 19.5

Variable Mean Standard deviation

Age of the head 56.0 13.4 Experience 22.2 13.3 Animal units 62.5 88.6 Meadow area, ha 64.9 129.4 Labor cost, $ 37,837 19,354 Health, $ 10,680 3,292 Feeding costs, $ 125,477 72,226 Assets; annual depreciation, $ 35,260 10,500 Net income, $ 83,488 20,824 Benefit cost (B/C) 1.31 0.26

Results and discussion

The owners of the BPUs in the Sierra Norte region of Puebla have an average age of 56 yr and range from 25 to 86 yr. The average schooling is 8 years; just under half of producers have completed primary education, 28.6 % finished junior high school and 29.2 % completed high school. The above characteristics are similar to those previouslyreported(32) for the rural population of the state of Puebla. The experience of producers in the production of cattle was 27 yr, and they have received technical assistance in topics of feeding, animal health and carrying capacity.

Half of the CPUs (50.8 %) are dedicated exclusively to the production of live cattle, 22.2 % are supported by other commercial activities (leases, businesses, and transport), 16.8 % are supported by agricultural and fruit activities (coffee, banana, corn, orange, beans, and vanilla), and 10.3 % report other non-agricultural activities. The percentage of household income generated by non-agricultural productive activities was 55 %, a result similar to that reported in previous studies(33)

The average herd size was 73 heads, with a minimum of 4 and a maximum of 657, which shows a great heterogeneity between the production units, hindering the conditions to compete and achieve a better production process(34). The average area held by the CPUs for grazing was 64 ha and the value of their assets was $135,261 (vehicles, mill, warehouse, milking machine, silo, corral, drinkers, feeder, and scale). The average annual income reported was $83,666, equivalent to 10 % of the herd, for the sale of weaning calves and discarded animals. In the cost structure, feed represented 60 % of the total cost of production, contracted and family labor 18 %, fixed costs and depreciation of assets 17 %, and 5 % was the cost of health.

Results of the econometric model

The results of the stochastic frontier model, using the full sample, are shown in Table 3. The variables had the expected sign, according to economic theory. The positive sign means that increasing the use of the production factor increases production, while the magnitude of the coefficient accounts for the relative importance of each independent variable in explaining the dependent variable.

Table 3: Results of the fit of the stochastic frontier model

Explanatory variable Coefficient SE t-statistic [95% confidence interval]

Area of pastures (ARE) 0.025 0.015 1.71* -0.023 0.073 Labor (LA) 0.263 0.068 3.89** 0.430 0.696 Value of Assets (ASS) 0.365 0.046 7.87** 0.456 0.274 Health (HEA) 0.411 0.081 5.07** 0.152 0.670 Feeding (FEED) 0.195 0.016 11.82** 0.053 0.327 Intercept -1.777 0.498 -3.57** -2.753 -0.802 sig2v -3.481 0.287 -12.13 -4.044 -2.919 sig2u -2.607 0.369 -7.06 -3.331 -1.884 sigma_v 0.175 0.025 0.132 0.232 sigma_u 0.272 0.050 0.189 0.390 sigma2 0.105 0.021 0.063 0.146 lambda and lambda2 1.370/1.88 0.072 1.408 1.688

gamma: �� =���� 2/���� 2 0.74 SE= standard error; * and** significant at 10 % and 5 % respectively.

The variables LA, ASS, HEA, and FEED are significant at 5 %. Area for cattle (ARECAT) was also found significant(30) when studying factors influencing technical efficiency in southeastern Kenya in 2013; a 10 % increase in area for cattle resulted in a 29 % increase in cattle production. The LA variable was found to be significant by several authors(31,35,36). In a study in Botswana(36) conducted with four strata of producers, they found that increasing the amount of labor by 10 % increases producers’ profits by 15 % and 18 %, respectively. The ASS variable has not been identified as significant in the studies reviewed. In the present study, ASS has a positive effect on the production of cattle PUs, as expected by economic theory(20). The variables HEA and FEED were also reported as significant(14,30,31) .

Regarding the fit of the model (7), the estimated stochastic production frontier showed a normal distribution of residuals (Shapiro-Wilks test), no serial correlation of errors (DurbinWatson), no heteroscedasticity of variance and no autocorrelation or multicollinearity problems. In the values obtained from the general fitted model (Table 3), it was determined that cattle production presents increasing returns to scale (the sum of the coefficients is greater than the unit). To confirm this result, the test was performed for returns to scale, where a value of P= 0.03 < 0.05 was obtained, this causes the existence of constant returns to scale to be rejected(6)

Regarding the inefficiencies of model 8, it was observed that the variance parameters of the maximum likelihood (ML) function are estimated from the total variance model defined as: ���� 2 =���� 2 +���� 2 and the estimated value in the model for the total variance (���� 2) was 0.105. While the lambda value (%) resulted in 1.370, which shows that the variance of the

efficiencies is greater than the variance of the random perturbations at 88 % (��2 1) and the gamma value obtained from the relationship between the variances �� =���� 2/���� 2 states that 73.9 % of the total variance is explained by the variance of the inefficiencies.

The results of the stochastic frontier model for each stratum of cattle producers are shown in Table 4. Similar to the general model, the models for each estimated stratum showed normal distribution of residuals, no serial correlation of errors, no heteroscedasticity of variance, and no autocorrelation. The variables ARECAT, LA, ASS, HEA, and FEED are significant at 5 % in strata two and three.

Table 4: Results of the Stochastic Frontier model for the strata of CPU Stratum 1 Stratum 2 Stratum 3

Variable Coefficient Z-value Coefficient Z-value Coefficient Z-value

ARE 0.094 1.85 0.027 2.27 0.073 3.31 LA 0.121 1.76 0.086 2.17 0.163 4.55 ASS 0.116 3.33 0.204 3.83 0.210 4.33 HEA 0.118 2.18 0.158 3.55 0.194 8.95 FEED 0.654 13.64 0.607 14.07 0.670 2.35 Constant 0.641 1.08 0.752 1.59 -1.267 -2.96 /lnsig2v -3.704 -24.7 -4.206 -23.02 -37.972 -0.06 /lnsig2u -13.129 -0.07 -13.419 -0.07 -2.339 -9.92 sigma_v 0.157 0.122 0.000 sigma_u 0.001 0.001 0.310 sigma2 0.025 0.015 0.096 lambda 0.009 0.010 5.460

In stratum 1, only HEA and FEED were significant. One possible explanation is that small producers have lower quality pastures, without agronomic management, use family labor, little specialized, and the value of their assets is very low, reflecting low-technified CPUs. The feed variable is the one that has the greatest weight in explaining the production of the CPUs for the three strata. The value of assets has twice as much relative weight in strata two and three than in strata one, which means that these CPUs not only have greater investment in assets, but that it is modern and generates greater productivity. The models for strata 2 and 3 show increasing returns to scale, but not the model of stratum 1 which has decreasing returns to scale. In this regard(37), in a study in the United States of America, it was found that as the size of the CPU increases, TE increases, which showed evidence of economies of scale. A possible explanation for the result of stratum 1 is that small producers have a low level of capitalization, low-skilled labor, and since they have little pasture area, they make intensive use, overexploiting the resource(38,39)

Frequency distribution of technical efficiency(TE) by UPG stratum

The TE range for cattle producers was between 0.50 and 0.95. Of the total of the 185 CPUs, 29 % have values between 0.50 and 0.70, 63 % between 0.71 and 0.90, and only 8 % TE values greater than 0.90. Table 5 shows that stratum 3 presents most of the values of 0.91 or more. In this regard(40) , it was found that the CPUs with the largest number of animal units and the largest area for cattle presented the highest values of technical efficiency.

Table 5: Frequency distribution (percentages) of technical efficiency (TE) by CPU strata

Strata (no. of heads) TE (0.50 - 0.70) TE (0.71-0.90) TE (> 0.91) Average Stratum 1 (20 or less) 47.2 22.2 13.3 0.712 Stratum 2 (21 to 50) 41.5 33.3 0.0 0.751 Stratum 3 (greater than 50) 11.3 44.4 86.7 0.844 General 100.0 100.0 100.0 0.789

Results of individual inefficiencies

Table 6 shows the results of the individual inefficiencies model according to Equation (8). The significant variables, at different levels of significance, and with a negative coefficient, wereSchoo, Exper,Admon andTA. Thenegativesign ofthecoefficients indicatesan inverse relationship between the value of the explanatory variable and the value of the inefficiency. In this regard, previous studies(28,30,36) have reported results that support the results of this study. It was found that more years of schooling reduces inefficiency in values very similar to those reported in this research. Similarly, in the case of the Admon(6,14,41) variable, they found an inverse relationship between having an administration system and inefficiency. For TA(28,30,41), they reported that receiving this service contributes to reducing the inefficiency of the CPUs. In the present study, Age is not significant, a result supported by what was found in the literature(30)

Table 6: General explanatory model of inefficiency

Explanatory variable Coefficient Standard error t-value Interval

Age (age) 0.02 0.0212 1.1 -0.042 – 0.042 Schooling (Schoo) -0.23 0.0635 3.6 0.010 – 0.635 Experience (Exper) -0.12 0.0739 1.7 -0.012 – 0.024 Administration (Admon) -0.23 0.0824 2.5 -0.001 - 0.048 Technical Assistance (TA) -0.22 0.0136 14.9 0.176 - 0.230 Constant -0.47 0.799 -0.6 -0.626 – (-0.310) Fit (R2)/R2 adjusted 0.7929 / 0.7859 Heteroscedasticity(CookWeisberg) Prob> Ji2=0.000 Normality: (ShapiroWilk) 0.00002 Inflation factor variance 1.59

The above results suggest that reducing inefficiencyshould be addressed by providing public technical assistance services, an activity that, in Mexico, has been at very low levels since the nineties. In this regard, in a study on the use of livestock innovations in Sinaloa(7), it was reported that only 3 % of the PUs receive technical assistance services, and of these, the CPUs represent only 19.3 %. Training in the management of the CPU, including administrative services, should also be a central aspect, in addition to the technological issues of cattle farming.

Results of the technical inefficiency model by CPU strata

Table 7 reports the results of the model of technical inefficiency by strata of CPUs. For stratum 1, Age and Exper are significant, but not Schoo, Admon and TA. The producers of this stratum have low schooling, 6 years on average, have experience, and most do not have administration systems and do not receive any type of technical assistance services. For stratum 2, Schoo, Exper and TA are significant. It was observed that the years of schooling increase significantly for the producers of this stratum. Finally, for stratum 3, four variables are significant. It should be noted that the values of the coefficients are in the range of 0.13 to 0.28, which shows an important effect of these variables to reduce inefficiency. Therefore, improving administration systems and the quality of technical assistance are aspects that can lead these CPUs to be highly efficient(14,30,41) .

Table 7: Results of the technical inefficiency model by CPU strata

Stratum 1 Stratum 2 Stratum 3

Variable Coef. t SE Coef. t SE Coef. t SE Age -0.066 -2.15* 0.031 0.017 0.42 0.041 0.065 1.38 0.047 Schoo -0.001 -0.03 0.007 -0.184 -2.08* 0.088 -0.142 -2.59* 0.055 Exper -0.027 -2.30* 0.012 -0.126 -2.09* 0.060 -0.197 -4.28* 0.046 Admon 0.033 1.63 0.020 0.017 0.81 0.020 -0.281 -5.73* 0.049 TA 0.108 1.42 0.076 -0.150 -7.34* 0.020 -0.134 -5.56* 0.024 Constant -0.238 -2.10 0.113 -0.481 -2.97 0.162 -0.700 -3.89* 0.179 R2/R2 Adj. 0.7935 / 0.7884 0.8027 / 0.7904 0.8214 / 0.7945 D-W 0.0719 0.0005 0.0247 Normality 0.01219 0.69848 0.17108 VIF 1.4 1.23 1.7 SE= standard error; D-W= Durbin-Watson; VIF= variance inflation factor

Conclusions and implications

Theproductionoflivecattleinthestudyregioniscarriedoutwith ahighdegreeofefficiency, however, there is significant room for improvement, especially in small producers. The most efficient producers have more schooling, receive technical assistance services, use administration systems, have more pasture area, more heads and use better animal health systems. Labor, health, food, and asset costs can be reduced while maintainingthe same level of production. Small producers, which are the largest subsector in number, can improve their production by attending to food and health aspects, with the other variables constant. The use of technical assistance services reduces inefficiency, through a more intensive and appropriate use of available livestock technology. Due to the above, it is advisable to make these services extensive and permanent to all farmers, especially small farmers. The positive relationship between herd size and productive efficiency may be related to the benefits of economies of scale, in the case of medium and large producers, so financing to increase the herd can generate production and efficiency gains.

Literature cited:

1. SIAP. Sistema de Información Agroalimentaria y Pesca. 2017. Avance mensual de la producción pecuaria. Consultado 13 marzo, 2019. http://infosiap.siap.gob.mx/repoAvance_siap_gb/pecConcentrado.jsp.

2. FIRA. Fideicomisos Instituidos en Relación con la Agricultura. 2017. Panorama Agroalimentario. Panorama Agroalimentario Carne de bovino 2017.pdf Consultado 22 enero, 2019.

Pecu 2023;14(1):154-171

3. SIAP. Sistema de Información Agroalimentaria y Pesca. 2018. Cosechando números del Campo; carne de bovino. http://www.numerosdelcampo.sagarpa.gob.mx/publicnew/productosPecuarios/cargarP agina/1 Consultado 24 Ene, 2019.

4. Jaramillo-Villanueva JL, Escobedo-Garrido JS, Carranza-Cerda I. Oportunidades estratégicas para el desarrollodel sectoragropecuario enPuebla, sistemas de producción y procesos de agregación de valor. 1ra ed. México: Plaza y Valdés SA de CV. 2017.

5. ENA. Encuesta Nacional Agropecuaria. 2015. Existencias de ganado bovino según rangos de edad por entidad federativa. https://www.inegi.org.mx/programas/ena/2014/default.html#Tabulados. Consultado 22 Ene, 2019.

6. Veloso-Contreras F, Cabas-Monje J, Velasco-Fuenmayor J, Vallejos-Cartes R, Gil-Roig JM. Eficiencia técnica de los pequeños productores bovinos de la región centro sur de Chile. Revista Científica 2015;25(2):99-106.

7. Cuevas RV, Baca del Moral J, Cervantes EF, Espinosa GJA, Aguilar ÁJ, Loaiza MA. Factores que determinan el uso de innovaciones en la ganadería de doble propósito en Sinaloa, México. Rev Mex Cienc Pecu 2013;4(1):31-46.

8. Ayala-Garay AV, Rivas-Valencia P, Cortes-Espinoza L, De la O-Olán M, EscobedoLópez D, Espitia-Rangel E. La rentabilidad del cultivo de amaranto (Amaranthus spp.) en la región centro de México. CIENCIA ergo-sum 2014;21(1):47-54.

9.KoopmansTC. Ananalysisofproductionasanefficient combinationof activities.Activity Analysis of Production and Allocation 1951;(13):33-97.

10. Martey E, Wiredu AN, Etwire PM, Kuwornu JK. The impact of credit on the technical efficiency of maize-producing households in Northern Ghana. Agric Finance Rev 2019;79(3):304-322.

11. Wiboonpongse A, Sriboonchitta S, Rahman S, Calkins P, Sriwichailumphun T. Joint determination of the choice of planting season and technical efficiency of potato in northern Thailand: A comparison of Greenes versus Heckmans sample selection approach. Afr J Bus Manag 2012;6(12):4504-4513.

12. Cabrera VE, Solís D, del Corral J. The effect of traditional practices in the efficiency of dairy farms in Wisconsin. South J Agric Econ Association Annual Meeting Orlando FL, February 6-9; 2010.

Pecu 2023;14(1):154-171

13. Cárdenas G, Vedenov DV, Houston JE. Analysis of production efficiency of Mexican coffee-producing districts. AAEA Annual Meetings, Selected Paper #134280 Providence, RI July 2005.

14. Otieno DJ, Hubbard LJ, Ruto E. Determinants of technical efficiency in beef cattle production in Kenya. Selected Paper prepared for presentation at the International Association of Agricultural Economists (IAAE) Triennial Conference Foz do Iguacu Brazil August 2012;18-24.

15. Iraizoz B, Bardaji I, Rapun M. The Spanish beef sector in the 1990s: impact of the BSE crisis on efficiency and profitability. Appl Econ 2005;37(4):473-484.

16. Trestini S. Technical efficiency of Italian beef cattle production under a heteroscedastic non-neutral production frontier approach. Paper presented at the 10th Joint Conference J Food Agric Environ, Duluth Minnesota 2006 August:27-30.

17. Morales-Hernández JL, González-Razo FDJ, Hernández-Martínez J. Función de producción de la ganadería de carne en la zona sur del Estado de México. Rev Mex Cienc Pecu 2018;9(1):1-13.

18. Solís D, Bravo-Ureta B, Quiroga R. Technical efficiency among peasant farmers participating in natural resource management programs in Central America. J Agric Econ 2009;60:202-219.

19. INEGI. Instituto Nacional de Estadística, Geografía. 2009. Censo Agropecuario 2007. http://www.inegi.org.mx/est/contenidos/proyectos/Agro/ca2007/Resultados_Ejidal/def ault.aspx. Consultado 18 Jun, 2019.

20. Nicholson W. Teoría microeconómica. Principios básicos y ampliaciones Novena edición Cengage Learning 2012;212-224.

21. SINIIGA. Sistema Nacional de Identificación Individual de Ganado. 2012. Estratificación por UPs y vientres bovinos, 32. Veracruz, México. http://ugrnv.com.mx/web/wp-content/uploads/2012/06/Siniiga%20Presentacion.pdf. Consultado 17 Jun, 2019.

22.Alvarez PA. Lamedición delaeficiencia ylaproductividad. Primera ed. Madrid,España: Ediciones Pirámide; 2001.

23. Greene WH. Simulated likelihood estimation of the normal-gamma stochastic frontier function. J Product Anal 2003;19(2):179–190.

24. Aigner DJ, Lovell K, Schmidt P. Formulation and estimation of stochastic frontier production function models. J Econom 1977;6(1):21–37.

25. Meeusen W, Van Den Broeck J. Efficiency estimation from Cobb-Douglas production functions. Int Econ Rev 1977;18(2):435–444.

26. JondrowJ,Lovell CK,Materov IS,Schmidt P.Ontheestimation oftechnical inefficiency in the stochastic frontier production function model. J Econom 1982;19(2-3):233-238.

27. Kumbhakar SC, Lovell CAK. Stochastic Frontier Analysis. Cambridge University Press. Cambridge, New York, Melbourne 2003. http://dx. doi. org/10.1017/cbo9781139174411.

28. Velasco FJ, Ortega SL, Sánchez CE, Urdaneta F. Análisis de sensibilidad del nivel tecnológico adoptado en fincas ganaderas de doble propósito del Estado Zulia, Venezuela. Rev Cient 2010;20(1):67-73.

29. Melo-Becerra LA, Orozco-Gall AJ. Technical efficiency for Colombian small crop and livestock farmers: A stochastic metafrontier approach for different production systems. J Product Anal 2017;47(1):1-16.

30. Kibara MJ, Kotosz B. Tecnnical efficiency estimation in the livestock industry: Case study of the southern rangelands of Kenya. Challenges in National and International Economic Policies 2018;97-114.

31. LatruffeL, BalcombeK,DavidovaS,ZawalinskaK.Determinantsoftechnicalefficiency of crop and livestock farms in Poland. Appl Econ 2004;36(12):1255-1263.

32. INEGI. Instituto Nacional de Estadística, Geografía. 2018. Censos y Conteos de Población y Vivienda. https://www.inegi.org.mx/programas/ccpv/2010/. Consultado 17 Ene, 2019.

33. Mora-Rivera J, Martínez-Domínguez M, Jaramillo-Villanueva JL, Chávez-Alvarado MA. Participación en el sector no agropecuario en el México rural: una perspectiva de género. R Bras Est Pop Belo Horizonte 2017;34(2):367-389.

34. Escalante S, Roberto I, Catalán H. Situación actual del sector agropecuario en México: perspectivas y retos. Economía Informa 2008; 350. http://www.economia.unam.mx/publicaciones/econinforma/pdfs/350/01escala nte.pdf.Consultado22May,2019.

35. Rakipova AN, Gillespie JM, Franke DE. Determinants of technical efficiency in Louisiana beef cattle production. J ASFMRA 2003;99-107.

36. Bahta S, Baker D. Determinants of profit efficiency among smallholder beef producers in Botswana. Int Food Agribusiness Manag Rev 2015;18(3):107-130.

37. Sabasi D, Shumway CR, Astill GM. Off‐farm work and technical efficiency on US dairies. Agric Econ 2019;1-15

38. Perfetti JJ, Balcázar A, Hernández A, Leibovich J. Políticas para el desarrollo de la agricultura en Colombia. Fedesarrollo, Sociedad de Agricultores de Colombia (SAC), Incoder, Finagro, Banco Agrario. 2013.

39. Cano CG, Vallejo C, Caicedo E, Amador JS, Tique EY. El mercado mundial del café y su impacto en Colombia. Borradores de Economía 2012;710

40. Qushim B, Gillespie JM, Bhandari BD, Scaglia G. Technical and scale efficiencies of US grass-fed beef production: whole-farm and enterprise analyses. J Agric Appl Econ 2018;50(3):408-428.

41.CeyhanV,HazneciK.Economicefficiencyofcattle-fatteningfarmsin Amasyaprovince, Turkey. J Anim Vet Adv 2010;9(1):60-69.

https://doi.org/10.22319/rmcp.v14i1.6182 Article

Quantile regression for prediction of complex traits in Braunvieh cattle using SNP markers and pedigree

a Universidad Autónoma Chapingo. Posgrado en Producción Animal. Carretera Federal México-Texcoco Km 38.5, 56227, Texcoco, Estado de México, México.

b Colegio de Postgraduados. Socio Economía Estadística e Informática. Carretera Federal México-Texcoco Km 36.5, 56230, Texcoco, Estado de México.

*Corresponding author: perpdgo@colpos.mx

Abstract:

Genomic prediction models generally assume that errors are distributed as normal, independent, andidenticallydistributedrandom variableswith zeromeanandequal variance. This is not always true, in addition there may be phenotypes distant from the population mean, which are usually removed when making the prediction. Quantile regression (QR) deals with statistical aspects such as high dimensionality, multicollinearity and phenotypic distribution different from the normal one. The objective of this work was to compare QR using marker and pedigree information with alternative methods such as genomic best linear unbiased prediction (GBLUP) and single-step genomic best linear unbiased prediction (ssGBLUP) to analyze the birth (BW), weaning (WW) and yearling (YW) weights of Braunvieh cattle and simulated data with different degrees of asymmetry and proportion of outliers. The predictive capacity of the models was assessed by cross-validation. The predictive performance of QR both with marker information alone and with information of markers plus pedigree, with the actual dataset, was better than the GBLUP and ssGBLUP methodologiesforWW andYW.ForBW,GBLUP andssGBLUP werebetter,however,only

quantiles 0.25, 0.50 and 0.75 were used, and the BW distribution was not asymmetric. In the simulated data experiment, correlations between “true” marker effects and estimated effects, as well as “true” and estimated signal correlations were higher when QR was used compared to GBLUP. The advantages of QR were more noticeable with asymmetric distribution of phenotypes and with a higher proportion of outliers, as was the case with the simulated dataset.

Key words: Quantile regression, GBLUP, ssGBLUP.

Received: 30/03/2022

Accepted: 04/08/2022

Introduction

The main motivation of the quantile regression (QR) method is that most models for genetic evaluation assume normality, which is not always true. Another problem is that sometimes phenotypic records very far from the population average are considered as recording errors or outliers and therefore removed from the analyses, seen from the genomic point of view, valuable information of markers associated with certain regions of DNA with strong influence on characteristics of interest is being lost.

With the QR method, robust results and a broad vision of the explanatory variables on the dependent ones are obtained(1). The data generated from omics experiments are often complex andlarge, so thereis a statistical challengeto extract relevant biological information from the large volume of data(2,3). Using a robust approach such as QR makes inference less biased and less subject to false positives(2). Recent studies using QR describe various applicationssuchas eticassociationstudies(4),populationgenetics(5),geneexpression(6,7),and genomic selection(8–10) .

One of the first studies where QR was used to predict individual genetic merit was presented by Nascimento et al(11), who used simulated data, finding advantages when using QR compared to conventional methodologies. In the same year(12), results using QR to adjust growth curves with data from pigs and molecular markers were published; not only did they successfully adjust the growth curves, but they identified important markers associated with the studied characteristic. Another similar work by the same team of researchers was presented by Nascimento et al(13), but with bean data. Recently, Pérez-Rodríguez et al(10) extended the quantile regression model to include pedigree information through the use of theadditivegeneticrelationshipmatrix,furtherimprovingthepredictiveabilityofthemodels

and at the same time identifying the proportions of the variances attributed to markers, relationships between individuals and the residual, which allows a precise partitioning of the phenotypic variance to be obtained.

Theobjectiveofthepresentstudywastostudythepredictivepowerofthequantileregression model using simulated data and actual data (birth, weaning and yearling weights) from Braunvieh cattle and the following models were considered: 1) QR with information of SNP molecular markers (QRM), 2) QR simultaneously including molecular marker information and genealogical information derived from pedigree (QRH); 3) GBLUP which, like QRM, only included molecular marker information, and 4) single-step genomic evaluation (ssGBLUP) which included marker and pedigree information.

Material and methods

Genotypes

The information used was from 300 animals (236 females, 64 males) born from 2001 to 2016 in eight herds located in Eastern, Central and Western Mexico. Hair samples were collected for genotyping by the company GeneSeek (Lincoln, https://www.neogen.com/, NE, USA), using the GeneSeek® Genomic Profiler Bovine LDv.4 panel, with 30,000 and 50,000 SNP markers, 150 animals with each Chip. Genotyping was performed on two separate occasions, initially 150 individuals with the 30K Chip and later another group of 150 individuals with the50KChipsincethe30KChipwasnotavailableatthetime.TheSNPsincommonbetween the 30K and 50K chips (12,835 SNPs) were used. The proportions of missing values were calculated for each marker and for each individual. The average of missing values per individual was 2.09 % with a standard deviation of 7.50 %. The average call rate (not missing proportion for each marker) was 97.90 % with a standard deviation of 4.66 %. Markers with a call rate of less than 95 % were removed. The genotypes were recoded as AA= 0, AB= 1 and BB= 2, from which a matrix with 300 rows (individuals) and 12,835 columns (markers) was obtained, whose cells take values in the set {0,1,2, }, where “ ” denotes a missing value. For the 12,835 common markers of the two chips, the missing values were randomly imputed, generating samples of the ����������������(2,��) distribution, where �� is the frequency of the major allele, calculated from the observed marker genotypes. Monomorphic markers or those with minor allele frequency (MAF) less than 0.04 were removed. After quality control, 9,628 of the 12,835 SNPs in common between the two chips were available for further analyses.

Phenotypes

The phenotypic and pedigree information of the Braunvieh cattle population was obtained from the database of the Mexican Association of Breeders of Registered Swiss Cattle. Records of birth (BW), weaning (WW) and yearling (YW) weights were used for analysis. Phenotype editing was similar for BW, WW and YW, records of animals not genetically related to those genotyped or with missing information for herd, dam’s age and management were discarded. Contemporary groups (CG) were defined by removing animals in CG of 2 individuals or with variance equal to zero. For BW, the CGs were defined by combining the effects of the herd (8 herds), year (1998 to 2016) and season of birth; the seasons of birth were defined considering the Julian calendar, from 80 to 171d, spring; from 172 to 264 d, summer; from 265 to 354 d, autumn; from 355 to 366 d and from 1 to 79 d, winter. After editing data, for BW, 330 records were obtained. For WW and YW, the CGs were defined by combining the effects of the herd (6 herds), year (from 1998 to 2016), season of birth (same as BW) and management. In the case of WW, the management groups were defined in three ways: calves fed their mother’s milk; their mother’s milk plus balanced feed; and milk from theirmother andnurseplus abalanceddiet. For YW,themanagement groupswere defined in three ways: grazing animals; grazing animals with feed concentrate; and housed animals with a balanced diet. The edition of WW and YW data ended with 267 and 232 records for further analyses. Table 1 shows a summary of the number of animals genotyped, and phenotyped for BW, WW and YW. Figure 1 shows the violin plots for BW, WW and YW, the sample mean is represented by the red dot and the sample median by the horizontal line within the box, from the plot, it is clear that the response variables have an asymmetric distribution and the circles with solid filling in it suggest the presence of outliers. Table 1:

Figure 1: Violin plots of birth (PN=BW), weaning (PD=WW) and yearling (PA=YW) weights in a Braunvieh cattle population

The sample mean is represented by the red dot and the sample median by the horizontal line inside the box

Models

Quantile regression model with markers (QRM)

The model for quantile regression is: ���� =��+���� ����+����, where ���� is the value of the phenotype of the i-th animal; �� is an intercept; ���� �� =(����1,…,������) represents the i-th row of the marker matrix, ��=(��1,…,����)�� is the vector of regression coefficients associated with markers and ���� are independent random variables such that their quantile �� ∈(0,1) is zero. The estimation of the regression coefficients for a fixed interest quantile �� is obtained by solving the following minimization problem: ������{∑ ���� �� ��=1 (���� �� ���� ����)+��∑ |����| �� ��=1 }, where ∑ |����| �� ��=1 is the sum of the absolute values of the regression coefficients; �� is the penalty parameter that controls the intensity of regularization; and ����() is the function defined as(1): ����(����)={��×���� If���� ≥0 (1 ��)×���� If���� <0, where ���� =���� �� ���� ����. After estimating the parameters of the model, the breeding values estimated by markers (GEBV) are obtained by the following expression:

��������(��)=�� ̂ ��(��)=∑ �������� ̂ ��(��) �� ��=1 , where �� ̂ ��(��) is the effect of the j-th marker, defined by the functional relationship obtained for the quantile of interest. The QR model can be extended to include other terms, in particular for growth characteristics, the following model is used: ���� =��+���� ����+���� ����+���� ����+����, where ���� is the value of the phenotype of the analyzed characteristic (BW, WW or YW) of the i-th animal, �� is an intercept; ���� �� =(����1, ,������) the i-th row of the incidence matrix for fixed effects (sex, dam’s age, management), ��=(��1, ,����)�� the regression coefficients for fixed effects, ���� �� =(����1,...,������) the i-th row of the incidence matrix for random effects of contemporary group (54, 43 and 37 for BW, WW and YW), ��=(��1,...,����)�� random effects of contemporary group, the rest of the terms as described above.

GBLUP

The model is given by: ���� =��+���� ����+���� ����+���� ����+����, where ���� �� =(����1, ,������) is the i-th row of the matrix that connects phenotypes with genotypes, ��=(��1,...,����)�� is the vector of random effects for animals. Additive, contemporary group and residual genetic variances are assumed ������(��)=������ 2 , ������(��)= �������� 2 , and ������(��)=������ 2 , respectively. The matrix of genomic relationships, ��, is calculated asdescribedbyLopez-Cruz et al(14) andPérez-Rodríguez et al(15);briefly, G=WW’/p,where W is the standardized and centered marker matrix (each marker centered by subtracting the mean allele frequency and standardized by dividing by the standard deviation of the sample of the allele frequency), p is the total number of markers, ���� normal and independent random variables with normal distribution with mean 0 and variance ���� 2 .

Single-step quantile regression (QRH) model

This method is considered an extension of the quantile model for a relationship matrix constructed using matrices of relationships for genotyped and non-genotyped animals and of whichapedigreeisavailable.Theresultingmatrixisknownintheliteratureasmatrix H(16,17) , this matrix is given by: �� 1 =�� 1 +[�� �� �� ���� 1 ������ 1], where, Agg is a submatrix of A for genotyped animals, Ga = βG + α; �� and �� are obtained by solving the system of equations: {������(��������(��))��+�� =������(��������(������)) ������(��)��+�� =������(������) .

The QRH model is given by: ���� =��+���� ����+���� ����+���� ����+����, where ������(��)=���� 2��, the rest of the terms as described above.

Single-step GBLUP regression (ssGBLUP) model

The ssGBLUP model is equivalent to the GBLUP model described above with the difference that the genomic relationship matrix G is replaced with the extended genetic relationship matrix H, it is assumed that ������(��)=������ 2 .

Cross-validation

The predictive capacity of the models was evaluated by cross-validation, which was performed as follows. The dataset was divided into five groups of the same size {��1,��2,…,��5}, 80 % of the data was used for training of the model, the remaining 20 % for validation. For example, {��2} is used as a validation group and the set {��1,��3,…,��5} for training of the model. The models were fitted using the training set, and the fitted model was used to obtain predictions for the validation set. This procedure was repeated five times and predictions were obtained for each group. Correlations between observed and predicted phenotypes were calculated and averaged for the test sets(18). Note that because these are actual values, the true breeding values are not known, but only the observed phenotypes are available, the fitted model provides predictions for breeding values and predictions of other fixed and environmental effects, with which a prediction of the phenotype is obtained, which is contrasted with the true value of the phenotype.

Simulation

In order to evaluate the predictive power of the QR model against GBLUP, an asymmetric data simulation with the presence of outliers was also carried out; the simulation of the present work is analogous to that used by Pérez-Rodríguez et al(10). The main idea is to highlight that the quantile regression model works adequately in the presence of atypical observations, inhomogeneous variances and response variables with responses with asymmetric distribution. In the context of selection, for example, it is not unusual to have asymmetric distributions for phenotypes due to the process itself, since, if one selects for some characteristic Y, and if there is in addition to this another characteristic of interest O, then the conditional distribution of Y |O>o(19) is asymmetric. In the context of genomic selection, it is also common to find subsets of observations that differ significantly from the restandtheseobservationscouldbeconsideredatypical.Montesinos-López et al(20) proposed a model with Laplace errors and showed that it predicts adequately even in the presence of outliers, the proposed model is a special case of the quantile regression model that is studied

in the present work. The 9,628 SNPs resulting from the quality control described above for 300 animals wereconsidered,thesimulation ofthedatawas carriedout consideringthelinear model: ���� =��+∑ ���������� 9,628 ��=1 +����, where �� =1, ,300, with �� =39 for BW, it was assumed that the errors come from a biased normal distribution (������)with mean0, variance ��2 (scale parameter ��)andasymmetryindex ��1, that is ����~������(0,��,��1), with �� =√1 ℎ2 , ℎ2 with a value of 0.35, ��1 = √2 ����3 (4 �� 1)(1 2��2 �� ) 3/2 , �� ∈{0950,0975,0999} were considered, leading to different degrees of positive bias. Only positive values of ��1 were considered since the negativebias is obtained simplybychangingthe signofthe ���� ′ �� andthereforetheconclusions obtained for the case of positive bias will also be valid for the negative case(21,22). Fifty markers with non-zero effect were fixed, simulating them from a normal distribution with mean 0 and variance √1 ℎ2⁄50, the rest of the markers were set at 0; the positions of the sampled markers were taken at random. To introduce outliers in the phenotypes, a certain proportion of the residues of ����~������(0,3,��1) were randomly generated, two proportions were considered, 5 and 10 %,so samples from amixtureoftwocomponents ofbiasednormal distributions were taken. Six datasets were generated, three different asymmetry coefficients 0.950, 0.975, 0.999 with their two alternatives of outlier proportion 5 % and 10 %. The asymmetric normal distribution has been used in genomic prediction(22) and its use in channeled selection has also been suggested(23). Once the data were generated, the QRM model was fitted with �� ={0.25,0.50,0.75} to compare it with GBLUP. The selection of quantiles was made according to Nascimento et al(11), who consider these three possibilities when the distribution of phenotypes is asymmetric �� ∈{0.25,0.75} or when the distribution is symmetric ��0.50, since our fundamental interest in this work focused on the modeling of possibly asymmetric data and with the presence of outliers. The selection of the parameters was also made for computational convenience since the fitting of the model is done by using intensive computational techniques based on Markov chain Monte Carlo, as mentioned in the section on software and fitting of the models. For each analysis, the correlation between true and estimated ��′��, the correlation between true ���� and estimated ���� ̂ signals and the component of variance associated with the residuals for each model, which is a way to evaluate the goodness of fit of the models, were calculated. The Deviation Information Criterion (DIC) was also considered, which can be used to select candidate models; models with lower DIC are preferred to models with higher DIC(24) .

Software and model fitting

The quantile regression models were fitted using a computational strategy similar to that described by Pérez-Rodríguez et al(10). Adaptations of algorithms to include fixed and random effects do not present great computational difficulty. The codes for the fitting of the

models were developed in the programming languages R(25) and C. The codes for the fitting of the models were organized in such a way that they can be easily run from the statistical software R and are available by requesting them to the first author of the present study. In all cases, three quantiles were selected, �� ={0.25,0.50,0.75}. The GBLUP and ssGBLUP models were fitted with the BGLR library of R(26) .

Results

Real data

Tables 2, 3, and 4 show the results of the experiment conducted with BW, WW, and YW data from a Braunvieh cattle population, evaluated under two scenarios 1) with marker informationonly,and2)markerandpedigreeinformation. In general,thehighestcorrelations between observed and predicted values were obtained with QR, except for BW, where the correlations of GBLUP and ssGBLUP were higher than those obtained with QRM and QRH, however, the correlations of QRM �� =0.75 and QRH �� =0.75 were close to those obtained with GBLUP and ssGBLUP (0.7902 vs 0.7924), (0.6981 vs 0.7055), respectively. The lowest MSE values were obtained with QRM �� =0.75 and QRH �� =0.75 in the WW characteristic, while in the BW and YW characteristics, the lowest values were obtained with GBLUP and ssGBLUP. The variance components associated with the error obtained with QRM and QRH were lower than those obtained with GBLUP and ssGBLUP. In general, the lowest DIC values were obtained with QRM �� =0.75 and QRH �� =0.75, except for BW with the markers-only scenario, where the lowest DIC was obtained with QRM �� =025.

Table 2: Averages of Pearson correlation and standard deviation (in parentheses) between observed phenotypic values (��) and predicted phenotypic values (�� ̂), mean squared error, variance components associated with the error (���� 2 , ���� 2 ) and deviation information criterion (DIC) for birth weight

Model Cor(��,�� ̂) MSE ���� �� or ���� �� DIC

QRM ��=�� ���� 0.7521 3.9973 2.7260 513.5014 (0.0753) (1.6108) (1.9762) (531.5701)

QRM ��=��.���� 0.5619 7.3249 8.6297 970.7680 (0.1501) (0.4561) (0.2660) (6.9791)

QRM ��=��.���� 0.7902 3.6535 2.4268 716.4237 (0.0766) (0.0943) (0.4829) (35.7161)

GBLUP 0.7924 2.3269 3.0035 803.0675 (0.0874) (0.2063) (0.5578) (31.9814)

QRH ��=��.���� 0.6713 3.5026 2.3645 872.3949 (0.1329) (1.2848) (1.9670) (432.0737)

QRH ��=��.���� 0.6816 2.9988 2.7372 659.1450 (0.1253) (0.7769) (1.8239) (1079.8674)

QRH ��=��.���� 0.6981 4.1405 2.8610 1077.2027 (0.1140) (0.6187) (0.8666) (60.6781)

ssGBLUP 0.7055 2.4463 3.2641 1189.4282 (0.1191) (0.2204) (0.4244) (26.5023)

Cor(��,�� ̂ )=correlation between observed and predicted phenotypes, MSE=mean squared error, ���� 2 or ���� 2=components of variance associated with the error, DIC=deviation information criterion.

Table 3: Averages of Pearson correlation and standard deviation (in parentheses) between observed phenotypic values (��) and predicted phenotypic values (�� ̂), mean squared error, variance components associated with the error (���� 2 , ���� 2 ) and deviation information criterion (DIC) for weaning weight

Model Cor(��,�� ̂) MSE ���� �� or ���� �� DIC

QRM ��=��.���� 0.5661 476.5293 419.4138 1550.5339 (0.2212) (17.4612) (23.3216) (13.9644)

QRM ��=��.���� 0.5695 357.7328 396.8138 1576.8871 (0.2307) (8.9681) (47.7433) (21.5826)

QRM ��=��.���� 0.5493 175.1298 67.9660 737.2216 (0.2196) (47.6181) (82.0807) (1150.7340)

GBLUP 0.5677 294.5807 376.7794 1583.2355 (0.2377) (36.6279) (24.1379) (16.2187)

QRH ��=��.���� 0.4816 644.1278 551.5150 1962.1296 (0.0672) (50.8464) (64.8091) (20.9916)

QRH ��=��.���� 0.4797 366.5940 356.9005 1537.7760 (0.0274) (56.8604) (238.5303) (903.3492)

QRH ��=��.���� 0.3918 216.1753 5.9471 -706.1573

(0.0544) (53.2417) (11.7834) (2034.7757)

ssGBLUP 0.4712 303.0404 421.8316 1982.3314 (0.0502) (37.6933) (55.2774) (21.9229)

Cor(��,�� ̂ )=correlation between observed and predicted phenotypes, MSE=mean squared error, ���� 2 or ���� 2=components of variance associated with the error, DIC=deviation information criterion.

Table 4: Averages of Pearson correlation and standard deviation (in parentheses) between observed phenotypic values (��) and predicted phenotypic values (�� ̂), mean squared error, variance components associated with the error (���� 2 , ���� 2 ) and deviation information criterion (DIC) for yearling weight

Model Cor(��,�� ̂) MSE ���� �� or ���� �� DIC

QRM ��=�� ���� 0.5421 1037.6529 953.6807 1487.1104 (0.1350) (175.2648) (261.8652) (35.8873)

QRM ��=��.���� 0.5341 868.3651 964.4477 1524.0511 (0.1355) (34.0429) (113.1832) (12.4648)

QRM ��=��.���� 0.5115 938.8244 700.7849 1284.0829 (0.1290) (241.2205) (465.2109) (402.9787)

GBLUP 0.5330 725.7579 924.8388 1526.7596 (0.1389) (71.3999) (90.0089) (11.6346)

QRH ��=��.���� 0.5306 1277.9493 1172.2877 1850.7122 (0.1411) (44.0948) (108.7991) (17.2025)

QRH ��=��.���� 0.5098 894.4148 1061.3157 1883.6773 (0.1700) (35.3996) (129.4702) (15.4422)

QRH ��=��.���� 0.5027 915.1871 666.8830 1706.4933 (0.1748) (162.7629) (413.5046) (209.8455)

ssGBLUP 0.4712 778.6416 1071.3096 1891.9029 (0.0502) (84.9871) (128.2878) (17.5592)

Cor(��,�� ̂ )=correlation between observed and predicted phenotypes, MSE=mean squared error, ���� 2 or ���� 2=variance components associated with the error, DIC=deviation information criterion.

Simulated data

The results of the simulation exercise where QR is compared with GBLUP under different degrees of asymmetry and proportions of outliers are shown in Table 5. Column 2 records the correlations between the “true” marker effects and the estimated marker effects, the correlations obtained with QR were higher than those obtained with GBLUP. Column 3 shows the correlations between the “true signals” and the estimated ones, the highest correlations were obtained with QR. Column 4 records the estimation of the variance components associated with the error and column 5 the DIC values, the lowest values in both columns were obtained with QR �� =0.75

Table 5: Averages of Pearson correlation and standard deviation (in parentheses) between “true” and estimated marker effects, “true” and estimated signals, variance components associated with the error and DIC values for simulated data with different degrees of asymmetry and proportion of outliers

Model Cor(��,�� ̂ ) Cor(����,���� ̂ ) ���� �� or ���� �� DIC

��=�� ����, 5% outliers

QR ��=�� ���� 0.0784 0.4963 0.6821 620.5455 (0.0034) (0.0336) (0.1806) (49.3305)

QR ��=�� ���� 0.0766 0.4643 0.6644 665.8219 (0.0042) (0.0493) (0.0703) (16.3032)

QR ��=�� ���� 0.0606 0.4269 0.1438 290.6870 (0.0132) (0.0386) (0.1421) (148.9695)

GBLUP 0.0722 0.4910 0.7375 691.6503 (0.0064) (0.0398) (0.0723) (19.9391)

��=�� ����, 10% outliers

QR ��=��.���� 0.0614 0.4369 0.4683 407.6496 (0.0183) (0.0329) (0.4030) (330.6304)

QR ��=��.���� 0.0728 0.4579 0.7947 706.7931 (0.0045) (0.0420) (0.1063) (20.5797)

QR ��=�� ���� 0.0574 0.4061 0.4482 381.4644 (0.0092) (0.0399) (0.3225) (474.7138)

GBLUP 0.0654 0.4556 0.8717 731.9104 (0.0057) (0.0314) (0.0890) (21.8563)

��=��.������, 5% outliers

QR ��=��.���� 0.0773 0.4835 0.5578 582.4254 (0.0087) (0.0562) (0.2523) (83.0548)

QR ��=��.���� 0.0771 0.4689 0.6369 662.0337 (0.0074) (0.0515) (0.0868) (23.8018)

QR ��=��.���� 0.0598 0.4169 0.2398 219.1691 (0.0128) (0.0450) (0.2033) (444.5060)

GBLUP 0.0703 0.4804 0.7316 692.6392 (0.0056) (0.0333) (0.0831) (24.0645)

��=��.������, 10% outliers

QR ��=�� ���� 0.0731 0.4386 0.8739 677.0858 (0.0081) (0.0789) (0.1077) (23.5472)

QR ��=��.���� 0.0734 0.4529 0.8154 711.2935 (0.0078) (0.0615) (0.0845) (14.9809)

QR ��=��.���� 0.0541 0.3945 0.3628 385.6030 (0.0056) (0.0583) (0.2572) (393.1935)

GBLUP 0.0640 0.4491 0.8913 736.7880 (0.0077) (0.0517) (0.0654) (14.8343)

��=��.������, 5% outliers

QR ��=�� ���� 0.0615 0.5286 0.1535 205.6973

(0.0144) (0.0271) (0.1657) 277.5807

QR ��=�� ���� 0.0741 0.5514 0.4860 614.2282 (0.0037) (0.0167) (0.0663) 15.7647

QR ��=�� ���� 0.0467 0.4855 0.0166 -271.4761 (0.0112) (0.0150) (0.0192) 288.4509

GBLUP 0.0737 0.5428 0.5305 625.9703 (0.0030) (0.0121) (0.0353) 11.3632

��=�� ������, 10% outliers

QR ��=��.���� 0.0768 0.4807 0.7817 650.8593 (0.0080) (0.0687) (0.0888) 22.8417

QR ��=�� ���� 0.0696 0.4630 0.6154 511.5287 (0.0148) (0.0600) (0.3369) 412.6645

QR ��=�� ���� 0.0507 0.3967 0.0204 -160.1660 (0.0031) (0.0505) (0.0127) 213.0462

GBLUP 0.0659 0.4649 0.7876 709.7240 (0.0065) (0.0418) (0.0528) 14.8566

Cor(��,�� ̂ )=correlation between “true” and estimated marker effects, Cor(����,���� ̂ )=correlation between “true” and estimated signals, ���� �� or ���� ��=variance components associated with the error, DIC=deviation information criterion.

Discussion

In this study, QR analysis methodologies were compared with GBLUP and ssGBLUP. This comparison was made with simulated phenotypes with different degrees of asymmetry and proportions of outliers and actual data for birth, weaning and yearling weights.

Real data

The observed and predicted phenotype correlations obtained from cross-validation with actual data were higher when using QRM and QRH in the WW and YW characteristics. For BW, the highest correlations were obtained with GBLUP and ssGBLUP; however, in this study, only three quantiles 0.25, 0.50 and 0.75 were tested, there is evidence in other studies where QR is better than GBLUP, as in the case of the work of Nascimento et al(4), who compared QR with models such as BLASSO, BayesB and RR-BLUP. These authors found a 15.15 % gain in the predictive capacity of QR compared to RR-BLUP, it should be noted that, mathematically, RR-is equivalent to GBLUP, in addition to the fact that the datasets used in this experiment presented asymmetry.

The values of the mean squared error (MSE) measure the average of the squared error, that is, the difference between the estimator and what is estimated, so low values are preferred; the MSE averages of QRM and QRH were lower than those obtained with GBLUP and ssGBLUP only for WW. The residual variance estimator is an indication that how well or

poorly the model fits the observed data, low values are preferred; the smallest variance components of the error were obtained with QRM and QRH for the three characteristics analyzed. Finally, the DIC value is used to select candidate models and, like MSE and error variance components, low values are preferred. The lowest DIC values were obtained with QRM �� =0.75 and QRH �� =0.75, except in the marker-only scenario and BW, where the lowest DIC was obtained with QRM �� =0.25.Themeansquared error,the residual variance and the DIC are values that help to choose the best fit model. When examining these values together, it is observed that QRM and QRH are better in some of them, while in others they are not, that is, QR has a performance equal to or greater than GBLUP and ssGBLUP; although it should be noted that only three quantiles were tested and that QR has advantages when used in asymmetric data and outliers, for this case there were only outliers and the distributions did not present asymmetry. Mendes et al(27) compared QR with the Bayesian method of LASSO (BLASSO), these authors reported a 6.7 % and 20.0 % increase in accuracy and considered quantiles 0.15 and 0.45 in the evaluation of carcass yield and bacon thickness, respectively, however, the characteristics evaluated in their study were asymmetric.

In the analysis of real data, a limitation of the present study is the sample size, which can impact the variability of the parameters estimated with the models and consequently the variability of the predictions, however, all the models were fitted using the same information and therefore the comparison of the predictive capacity of the models is considered reasonable, the ideal would be to have large sample sizes, but, due to economic limitations, this is not always possible. On the other hand, it is currently very common to use prediction models in which the number of phenotypic records is smaller than the number of predictors (SNPS), that is �� ≪��, even in this context, numerous studies have shown that Bayesian methods provide sophisticated tools that allow deriving reasonable predictions as long as the regularization parameters are selected properly, for example using cross-validation methods(28–30)

Simulated data

In the simulated data experiment, the correlations between “true” marker effects and estimated effects as well as correlations of “true” and estimated signals were higher when QR was used compared to GBLUP. These results are similar to those obtained by other researchers(10), who simulated data with three different coefficients of asymmetry 0.75, 0.95, 0.999 with 5 % and 10 % of outliers and found that the correlations obtained with QR were higher than those obtained with Bayesian ridge regression (BRR), in addition, this pattern was more evident with a greater asymmetry and proportion of outliers. In this study, simulations with asymmetry coefficients of 0.950, 0.975, 0.999 were carried out and the quantiles with which higher correlations were obtained varied between 0.25 and 0.50; the

advantage of QR is that different quantiles can be tested, obtaining better results depending on the quantile used, this advantage in the abilityto predict the effects of markers and signals has been taken advantage of by other researchers(4) , who found no trait association using the traditional GWAS model of single SNP, but, when using QR with extreme quantiles such as 0.1, the model was able to find up to 7 SNPs associated with the characteristics studied.

Thecoefficients ofvarianceofthe errorindicatehowwell theproposedmodel fits thestudied data, the smaller this value, the better the fit, the DIC is another value that is used to compare candidate models. Models with a smaller DIC are preferred to models with a larger DIC(24)

The lowest residual variance estimators and DIC values were obtained with QR �� =075, perhaps this is because high asymmetry coefficients 0.950, 0.975, 0.999 were used in the simulation, so therefore a quantile that fits best is expected to be the highest, in this case 0.75. QR performed equally or better than GBLUP and ssGBLUP to predict growth characteristics BW, WW and YW, the advantages of this method are more noticeable when the data are more biased and present a higher proportion of outliers, as in the case of the simulation experiment.

Conclusions and implications

The predictive performance of QR both with marker information alone and with information of markers plus pedigree, with the actual dataset, was better than the GBLUP and ssGBLUP methodologies for WW and YW. For BW, GBLUP and ssGBLUP were better; however, only quantiles 0.25, 0.50 and 0.75 were used, and the BW distribution was not asymmetric. In the simulated data experiment, correlations between “true” marker effects and estimated effects, as well as correlations of“true”and estimated signals werehigher whenQR was used compared to GBLUP. The advantages of QR were more noticeable with asymmetric distribution of phenotypes and with a higher proportion of outliers, as was the case with the simulated dataset.

Acknowledgements

To the National Council of Science and Technology, Mexico, for the financial support for the first author during his doctoral studies. The authors also thank the Mexican Association of Breeders of Registered Swiss Cattle for allowing the use of their databases, and the cooperating breeders for their kind cooperation in this study.

Conflicts of interest

The authors declare that there are no conflicts of interest.

Literature cited:

1. Koenker R, Bassett G. Regression quantiles. Econometrica 1978;46(1):33. https://doi.org/10.2307/1913643.

2. Briollais L, Durrieu G. Application of quantile regression to recent genetic and -omic studies. Hum Genet 2014;133(8):951–966. https://doi.org/10.1007/s00439-014-1440-6.

3. Wang L, Wu Y, Li R. Quantile regression for analyzing heterogeneity in ultra-high dimension. J Am Stat Assoc 2012;107(497):214-222. https://doi.org/10.1080/01621459.2012.656014

4. Nascimento M, Nascimento ACC, Silva FF, Barili LD, Do Vale NM, Carneiro JE, Cruz CD, Carneiro PCS, Serão NVL. Quantile regression for genome-wide association study of flowering time-related traits in common bean. PLoS One 2018;13(1):1-14. https://doi.org/10.1371/journal.pone.0190303.

5. Fisher E, Schweiger R, Rosset S. Efficient construction of test inversion confidence intervals using quantile regression. J Comput Graph Stat 2016;29:140-148, http://arxiv.org/abs/1612.02300

6. Logan JAR, Petrill SA, Hart SA, Schatschneider C, Thompson LA, Deater-Deckard K, de Thorne LS, Bartlett C. Heritability across the distribution: An application of quantile regression. Behav Genet 2012;42(2):256–267. https://doi.org/10.1007/s10519-0119497-7.

7. Vinciotti V, Yu K. M-quantile regression analysis of temporal gene expression data. Stat Appl Genet Mol Biol 2009;8(1). https://doi.org/10.2202/1544-6115.1452.

8. Gianola D, Cecchinato A, Naya H, Schön CC. Prediction of complex traits: Robust alternatives to best linear unbiased prediction. Front Genet 2018;9:195. https://doi.org/10.3389/fgene.2018.00195

9. Oliveira GF, Nascimento ACC, Nascimento M, Sant’Anna IdeC, Romero JV, Azevedo CF, Bhering LL, Moura ETC. Quantile regression in genomic selection for oligogenic traits in autogamous plants: A simulation study. PLoS One 2021;16(1):1-12. https://doi.org/10.1371/journal.pone.0243666

10. Pérez-Rodríguez P, Montesinos-López OA, Montesinos-López A, Crossa J. Bayesian regularized quantile regression: A robust alternative for genome-based prediction of skewed data. Crop J 2020;8(5):713-722. https://doi.org/10.1016/j.cj.2020.04.009.

11. Nascimento M, e Silva FF, de Resende MDV, Cruz CD, Nascimento ACC, Viana JMS, Azebedo CF,Barroso LMA. Regularizedquantileregression appliedto genome-enabled prediction of quantitative traits. Genet Mol Res 2017;16(1). https://doi.org/10.4238/GMR16019538.

Pecu 2023;14(1):172-189

12. Barroso LMA, Nascimento M, Nascimento ACC, Silva FF, Serão NVL, Cruz, CD, et al. Regularized quantile regression for SNP marker estimation of pig growth curves, J. Anim Sci Biotechnol 2017;8:59. https://doi.org/10.1186/s40104-017-0187-z

13. Nascimento AC, Nascimento M, Azevedo C, Silva F, Barili L, Vale N, et al. Quantile regression applied to genome-enabled prediction of traits related to flowering time in the common bean. Agronomy 2019;9(12):1-11. https://doi.org/10.3390/agronomy9120796.

14. López-Cruz M, Crossa J, Bonnett D, Dreisigacker S, Poland J, Jannink JL, Singh RP, Autrique E, de los Campos G. Increased prediction accuracy in wheat breeding trials using a Marker × Environment interaction genomic selection model. G3 Genes Genom Genet 2015;5(4):569–582. https://doi.org/10.1534/g3.114.016097.

15. Pérez-Rodríguez P, Crossa J, Rutkoski J, Poland J, Singh R, Legarra A, et al Single-step genomic and Pedigree Genotype × Environment interaction models for predictingwheat lines in international environments. Plant Genome 2017;10(2). https://doi.org/10.3835/plantgenome2016.09.0089.

16. Christensen O, Lund M. Genomic relationship matrix when some animals are not genotyped. Genet Sel Evol 2010;42(2):1-8. http://www.gsejournal.org/content/42/1/2

17. Aguilar I, Misztal I, Johnson DL, Legarra A, Tsuruta S, Lawlor TJ. Hot topic: A unified approach to utilize phenotypic, full pedigree, and genomic information for genetic evaluation of Holstein final score. J Dairy Sci 2010;93(2):743-752. https://doi.org/10.3168/jds.2009-2730.

18. Crossa J, Pérez P, de los Campos G, Mahuku G, Dreisigacker S, Magorokosho C. Genomic selection and prediction in plantbreeding. J Crop Improv 2011;25(3):239-261. https://doi.org/10.1080/15427528.2011.558767.

19. Arnold BC, Beaver RJ. Hidden truncation models. Shankhya. Indian J Stat 2000;62(1):23–35. http://www.jstor.org/stable/25051286. Accessed Jul 6,2022.

20. Montesinos-López A, Montesinos-López OA, Villa-Diharce ER, Gianola D, Crossa J. A robust Bayesian genome-based median regression model. Theor Appl Genet 2019;132(5):1587-1606. https://doi.org/10.1007/s00122-019-03303-6

21. Pérez-Rodríguez P, Villaseñor-Alva JA. On testing the skew normal hypothesis. J Stat Plan Inference 2010;140(11):3148-3159. https://doi.org/10.1016/j.jspi.2010.04.013.

22. Pérez-Rodríguez P, Acosta-Pech R, Pérez-Elizalde S, Cruz CV, Espinosa JS, Crossa J. A Bayesian genomic regression model with skew normal random errors. G3 Genes|Genom|Genet 2018;8(5):1771–1785. https://doi.org/10.1534/g3.117.300406.

Pecu 2023;14(1):172-189

23. Domínguez-Viveros J. Parámetros genéticos en la varianza residual de variables de comportamiento en toros de lidia. Arch Zoot 2020;69(267):354–358. https://doi.org/10.21071/az.v69i267.5354

24. Spiegelhalter DJ, Best NG, Carlin BP, van der Linde A. Bayesian measures of model complexity and fit. J R Stat Soc: Series B Stat Methodol 2002;64(4):583–639. https://doi.org/10.1111/1467-9868.00353.

25. R Core Team. R: A language and environment for statistical computing. R Foundation for statistical computing 2021. Vienna, Austria. https://www.R-project.org/.

26. Pérez P, de los Campos G. Genome-wide regression and prediction with the BGLR statistical package. Genetics 2014;198(2):483-495. https://doi.org/10.1534/genetics.114.164442

27. Mendes dos Santos P, Nascimento ACC, Nascimento M, Fonseca e Silva F, Azevedo CF, Mota RR, et al. Use of regularized quantile regression to predict the genetic merit of pigs for asymmetric carcass traits. Pesqui Agropecu Bras 2018;53(9):1011–1017. https://doi.org/10.1590/S0100-204X2018000900004

28. Gianola D. Priors in whole-genome regression: The Bayesian alphabet returns. Genetics 2013;194(3):573-596. https://doi.org/10.1534/genetics.113.151753.

29.delosCamposG,HickeyJM,Pong-WongR,DaetwylerHD,CalusMPL.Whole-genome regression and prediction methods applied to plant and animal breeding. Genetics 2013;193(2):327-345. https://doi.org/10.1534/genetics.112.143313.

30. Ferragina A, de los Campos G, Vazquez AI, Cecchinato A, Bittante G. Bayesian regression models outperform partial least squares methods for predicting milk components and technological properties using infrared spectral data. J Dairy Sci 2015;98(11):8133-8151. https://doi.org/10.3168/jds.2014-9143.

https://doi.org/10.22319/rmcp.v14i1.5187 Article

Seasonal growth analysis of a white clover meadow (Trifolium repens L.)

Edgar Hernández Moreno a

Joel Ventura Ríos b

Claudia Yanet Wilson García c

María de los Ángeles Maldonado Peralta d*

Juan de Dios Guerrero Rodríguez e

Graciela Munguía Ameca a

Adelaido Rafael Rojas García d

a Colegio de Postgraduados - Campus Montecillo. km 36.5 Carretera México-Texcoco, 56250, Texcoco, Estado de México, México.

b Universidad Autónoma Agraria Antonio Narro. Departamento de Producción Animal. Saltillo Coahuila, México.

c Universidad Autónoma Chapingo. Sede San Luis Acatlán. San Luis Acatlán, Guerrero, México.

d Universidad Autónoma de Guerrero. Facultad de Medicina Veterinaria y Zootecnia N° 2. Cuajinicuilapa, Guerrero, México.

e Colegio de Postgraduados – Campus Puebla, México.

*Corresponding author: mmaldonado@uagro.mx

Abstract:

The objective of the present study was to assess a growth analysis of white clover (Trifolium repens L.) and determine the optimal harvest time per season. The experiment was carried

out at the Colegio de Postgraduados, Campus Montecillo, Texcoco, Mexico. Twenty-four 3.7 X 1.7 m plots were used, distributed in a completely randomized design, with eight treatments and three replicates per station. The treatments consisted of successive weekly cuts, during a regrowth cycle of 8 wk, in each season of the year. At the beginning of the study, a uniform cut was made and the residual forage was determined. The evaluated variables were: accumulation of dry matter, botanical and morphological composition, and leaf area index of white clover. The highest forage accumulation (P<0.05) occurred in the eighth week in spring (2,688 kg DM ha-1). Leaf production was higher (P<0.05) in spring, autumn and winter. The highest leaf area index was reached in the eighth week in spring(3.0; P<0.05). It is recommend exploitingthewhiteclovermeadowin thesixth weekofthespringsummer period and in the seventh week of autumn-winter.

Key words: Growth analysis, Trifolium repens L., Dry matter accumulation, Botanical composition.

Received: 11/12/2021

Accepted: 06/07/2022

Introduction

In the central zone of Mexico, white clover (Trifolium repens L.), perennial ryegrass (Lolium perenne L.), orchard grass (Dactylis glomerata L.), and alfalfa (Medicago sativa L.) planted on 171,520 hectares are the forage species that have exhibited the best performance under grazing conditions in pure or mixed pastures(1,2,3).However, due to its chemical composition, its persistence resulting from its creeping growth habit, and its adaptability to temperate zones, white clover is the species of greatest agronomic importance among the almost 300 species of the genus Trifolium(4). In addition, it can also improve soil fertility by supplying nitrogen in a proportion of up to 450 kg N ha-1 through symbiotic fixation(5,6,7) .

Forage production patterns in Mexico are influenced by climate variations, with temperature and precipitation being the main factors(8); therefore, it is important to know the seasonal growth patterns of the most widely used forage species in each one of the ecological regions of the country(9) . Previous works mention that the management strategies of a meadow, intensity and frequency through cutting or grazing can modify the botanical composition, yield, and nutritional quality(10,11) The severity in the use of the meadow can modify the

carbohydrate reserves in the plant, which affects the growth pattern, reducing the number of stems, number of sprouts, and number of leaves in the plant(12) .

Evaluating the seasonal growth of a pure or mixed meadow helps to understand the behavior of plants and the association of different species, since the balance between growth rate and tissue loss varies with the season throughout the year(9). Dry matter yield, crop growth rate, leaf area index, botanical and morphological composition, intercepted radiation, plant height, leaf: stem ratio, and leaf: non-leaf component are structural variables that help to understand thebehaviorofameadowandmustbeconsideredtounderstandtheresponseunderamowing or grazing system(13,14)

In a study conducted by Moreno et al(15), they reported that, when associated with irrigated grasses, white clover produced an average of 1,581 kg DM ha-1 (P<0.05) in its first year of evaluation, while Maldonado et al(3) reported an increase of 376 % (equivalent to 7,532 kg DM ha-1) in their fourth year of evaluation under mixed pastures with irrigation due to their stoloniferous growth and persistence in the meadow. In another research(9), the highest leaf area index occurred at week five in summer (P0.05), and the leaf was the largest component in spring. There is little research analyzing the growth of white clover in Mexico. The objective of the present study was to evaluate growth analysis of white clover (Trifolium repens L.) in order to determine the optimal physiological moment of grazing in each season of the year.

Materials and methods

The study was performed in a white clover meadow in the experimental field of the College of Postgraduates (Colegio de Postgraduados), in Montecillo, Texcoco, State of Mexico, at 19º 29' N and 98º 53' O, at 2,240 m asl. Broadcast sowing was carried out in February 2009 with a viable pure seed density of 6 kg ha-1. The local climate is temperate sub-humid, with an average annual precipitation of 636.5 mm, and a rainfall regime in the summer (from June to October) and an average annual temperature of 15.2 ºC(16). The local soil is sandy loam, with a slightly alkaline (pH 7.8) and 2.4 % organic matter(17) .

In themiddle of each seasonof2012, a uniform grazingwas made andlater a growth analysis was carried out; the treatments consisted of an eight-week growth analysis in spring-summer and a nine-week one in autumn-winter, since low temperatures promote a slower forage growth. Sheep were used as defoliators until the remaining forage was left at a height of approximately 5 cm above ground level, and, for better management, an electric fence was established in the experimental plots.

The plots were distributed in a completely randomized design with three replications. Twenty-four 3.7 x 1.7 m plots were drawn, among which the treatments were randomly distributed. During the dry season, the meadows were irrigated by gravity at field capacity; 16 irrigations were carried out every 2 wk with approximately 32 mm for each one, which gave a total of 512 mm of water, and the meadows were not fertilized.

Climate data

The monthly averages of outdoor temperature (maximum and minimum) and monthly rainfall during the study period were obtained from the agrometeorological station of the College of Postgraduates (Colegio de Postgraduados), located at a distance of 100 m from the experimental site (Figure 1). The maximum monthly temperature ranged from 22.1 to 30.2 °C, while the minimum temperature was from -2.6 to 11 °C. The highest temperature occurred in spring, registering a maximum of 30.2 °C in April, and the lowest temperature, of -2.6 °C was recorded in December. The accumulated precipitation from March 2012 to April 2013 was 312.3 mm, 70 % of which occurred in June, July, August, and September 2012, accumulating a precipitation of 220 mm.

Figure 1: Average maximum and minimum monthly temperature and accumulated precipitation per month during the study period (March 2012 - April 2013)

Dry matter accumulation

After uniform grazing, three 0.25 m2 squares were cut at a height of 5 cm from the ground in each experimental plot for eight weeks. The forage harvested in each quadrant was washed and dried in labeled paper bags in a forced air oven at 55 °C during 72 h in order to estimate the amount of dry matter per hectare at the various regrowth ages.

Botanical and morphological composition

For the purpose of determining the botanical composition of the forage, a subsample of this was collected for a dry matter yield of approximately 20 %(11) and was separated into the following components: dead material, weeds, grasses and white clover. The morphological components of white clover (leaf, petiole, runner and flower) were separated. Each separate component was dried in a labeled paper bag and left in a forced air oven at 55 °C during 72 h in order to estimate its dry weight.

Leaf area index

The leaf area index was calculated byseparating the leaves of five stolons were separated for each week’s replicate and placing them in a leaf area integrator (LI 3100 LI-COR Inc.) from which the leaf area readings in cm2 were obtained. These readings together with the number of stolons per square meter allowed estimating the leaf area index by means of the following formula:

LAI= LA * SD Where: LAI= leaf area index; LA=leaf area per stem; and SD= stolon density (m-2).

Statistical analysis

The data were analyzed by the GLM procedures of SAS(18), for a completely randomized experimentaldesign,wherethetreatmentsweretheweeksofevaluationwiththreerepetitions per season and a regression analysis for each variable. The means were compared with the Tukey test (α= 0.05).

Results and discussion

Dry matter accumulation

Figure 2 shows the results of the accumulated dry matter yield, which increased with regrowth age, reaching its maximum in the eighth week of spring (2,688 kg DM ha-1; P<0.05), the seventh week of summer (2,241 kg DM ha-1; P<0.05), the eighth week of fall (1,781.3 kg; P<0.05), and the sixth week of winter (1,643 kg; P<0.05). The biomass accumulated in springwas higher by20 % (447 kg DM ha-1), 51 % (907 kg), and 64 % (1,045 kg), compared to summer, autumn and winter, respectively (P<0.05). The leaf for spring, autumn and winter increased as the regrowth weeks increased and it was greater than the petiole, however, in summer there was a higher petiole yield and less leaf. During the evaluation period in winter (March 3, 2013), an intense frost occurred; this coincided with the sixth week of regrowth, which limited the biomass yield for said sampling, drastically increasing the dead material.

For their part, in their assessment of white clover in the highlands of Mexico, Gutiérrez et al(9) mention an accumulation of forage as the age of the plant increased in all seasons, reaching the maximum yield for spring, autumn, and winter in the eighth week, with 2,953, 1,592, and 1,791 kg DM ha-1, respectively, and in the seventh week for summer with 1,971 kg DM ha-1 (P<0.05).

When establishing associations of white clover with perennial ryegrass (Lolium perenne) and orchard grass (Dactylis glomerata L.), Moreno et al(15) found a maximum white clover yield of 513 kg DM ha-1 in their first sampling year. However, Maldonado et al(3) registered a significant increase in the yield of white clover in these same associations in their third and fourth year of production, adding up to an average of 7,220 kg DM ha-1. These same authors mention that white clover dominates over time in the meadow because it is a kind of stoloniferous growth habit that allows rapid growth compared to grasses, which are tufted. In another study(19), they reported that 65% of the annual yield occurred in spring and summer, 23 % in winter, and autumn was the season that exhibited the lowest value, of 12 % (P<0.05).

According to various authors(18,20), white clover requires temperatures of 18 to 30 °C, the optimal being 24 °C, and precipitations of 750 mm for best performance. These temperatures were reached in spring (Figure 1), favoring the growth of the meadow as a result of the increase in the leaf area per plant and probably due to the increase in leaf appearance and elongation rates(19).Conversely,thedrymatter yieldwas low in winter; in this regard,various

authors (3,12,21) arguethatlowtemperatures limit growthandforage accumulation, dueto their direct influence on a lower leaf appearance and foliar rate.

Botanical and morphological composition

Based on the morphological components and the estimated accumulated biomass yield, the plant morphologywas variable(P<0.05).Thehighest leafpercentage, of 68 %,was produced in the third week in spring; likewise, in summer the highest percentage of leaf, of 46 to 40 %, was producedfromthefirst to thethirdweek,decreasingdrasticallyin thefourth week, to 30 % (P<0.05). In autumn, the highest leaf percentage occurred in the first week (70 %) and remained until the seventh week (59 %), after which it decreased. Finally, in winter, the highest leaf percentage (60 %) occurred in the fifth week (Figure 3).

On the other hand, the greatest contribution of the petiole, of 38 %, occurred in summer (P<0.05), while the highest percentage of stolon was reported in spring, being greater only in the first two weeks of growth, when it amounted to 20 % in average (P<0.05). As for pastures, the largest percentage, an average of 15 %, occurred in the summer. The contribution of weeds and flowers was minimal in all seasons and weeks of regrowth, being 3 % in average.

Winter was the season that reported the largest amount of dead material and from the seventh week on there was a drastic increase, of 100 %, since there was a decrease in temperature (Figure1)frost causingthedeath of whiteclover. Inthis regard, Brockand Tilrock(8) mention that all plants have an optimal growth temperature and when these surpass it or decrease drastically, there may be cell death, which causes a drastic increase in dead material. On the other hand, as the age of the plant increased, the dead material also built up (P<0.05), due to the maturation of the senescent leaves of the lower strata(9) .

The large proportion of leaves with respect to the petiole and stolon indicates that it is a highquality forage since this allows it to be more digestible. In addition to this, as observed in all the assessment stages, the content of flower was minimal (P>0.05), which indicates that this forage is not precocious and allows it to increase its nutritional value in the first weeks. As has been shown, the association of this legume with grasses confers it forage value, since it augments the total yield per surface unit to up to 14 t DM ha-1; however, these qualities can also be affected by the season of the year(19,22)

In addition to this, white clover under grazing conditions is less susceptible to loss of apical meristems due to the horizontal arrangement of leaves and leaf primordia, which allow efficient capture of solar radiation compared to Gramineae(8,14) .

Leaf area index

The leaf area index (LAI) allows to estimate the photosynthetic capacity of a plant per unit area and helps to understand its ability to assimilate solar energy and transform it into dry matter yield(23). The highest LAI was reported during spring and autumn, with 3.0 and 2.6, respectively, in week eight, while in summer and winter the highest index was obtained in week five with 2.6 (P<0.05). On the other hand, in all seasons there was a close relationship between the LAI and the regrowth weeks, the highest r2, of 0.97, occurring in spring, and the lowest, of 0.83, in summer (Figure 4).

The temperature range of 22 to 30 °C during the spring and summer and the higher precipitation in June, July, August, and September resulted in 70 % of the total accumulated precipitationintheexperimentalperiod,contributingtothe greater growthoftheplant,which profited from the biochemical and photosynthesis processes for its optimal development. However, the conditions were not favorable in the fall and winter, when the low temperatures ranged from -2 to 11 °C, causing a reduction of the tissue turnover in winter and thereby affecting the growth and development of the plant, which exhibited the lowest LAI and the lowest yield of accumulated biomass during this season.

In a trial to assess white clover(9), the highest leaf area index, of 3.0, was observed in the eighth week of spring; later, in the fifth week of summer, it was 1.7, and in the eighth week of autumn and of winter, it had a value of 1.4 and 1.6, respectively results that agree with those of this research. In a study directed by Zaragoza et al(1), they reported the highest LAI (P<0.05) for the alfalfa crop in week five of spring (3.5), of summer (2.8), and of autumn (2.0), and in the sixth week of winter(1.9). However, the values were different when evaluatingtheorchardgrass,sincethehighest LAI(P<0.05)occurredinweeksix ofregrowth in spring(2.3), summer(1.4) and autumn (1.1), and in the seventh week of winter (1.0). Other researches on alfalfa(23,24) reported behaviors similar to those observed in this experiment, since the highest values of LAI (P<0.05), of 3.3 and 4.9, respectively, were recorded in spring-summer.

The LAI varies for each crop and depends on the environmental conditions present. Matthew et al(14) points out that the LAI is optimal when the net forage production is at a maximum

point and the highest LAI is reached simultaneously; however, the LAI can be affected indirectly by low temperatures, according to the type of crop and the time of sampling.

Conclusions and implications

The accumulated biomass yield was higher in the spring and lower in autumn and winter. As the regrowth age increased, so did the dry matter. It is recommend profiting from the defoliation of the meadow in the sixth week of spring and summer and the seventh week of autumn and winter, based on the fact that an adequate dry matter yield, greater leaf component, and less dead material were obtained at those times, consequently optimizing the forage nutrients.

Literature cited:

1. Zaragoza EJ, Hernández GA, Pérez PJ, Herrera HJG, Osnaya GF, Martínez HPA, González MSS, et al. Análisis de crecimiento estacional de una pradera asociada alfalfapasto ovillo. Téc Pecu Méx 2009;47(2):173-188.

2. Parfitt RL, Couper J, Parquinson R, Schon NL, Stevenson BA. Effect of nitrogen fertilizer on nitrogen pools and soil communities under grazed pastures. NZ J Agr Res 2012;55(3):217-233.

3. MaldonadoPMÁ,Rojas GAR,Torres SN,HerreraPJ,Joaquín CS,VenturaRJ,Hernández GA, Hernández GFJ. Productivity of orchard grass (Dactylis glomerata L.) alone and associated with perennial ryegrass (Lolium perenne L.) and white clover (Trifolium repens L.). Rev Brasi Zootec 2017;46(12):890-895.

4. Randazzo CP, Rosso BS, Pagano EM. Identificación de cultivares de trébol blanco (Trifolium repens L.) mediante SSR. J Basic Appl Gen 2013;24(1):19-26.

5. Black AD, Laidlaw AS, Moot DJ, O’Kiely P. Comparative growth and management of white and red clovers. Irish J Agric Food Res 2009;48(2):149-166.

6. Phelan P, Keogh B, Casey A, Necpalova M, Humphreys. The effects of treading by dairy cows on soil properties and herbage production for three white clover‐based grazing systems on a clay loam soil. Grass Forage Sci 2012;68(4):548-563.

7. Unkovich M. Nitrogen fixation in Australian dairy systems: review and prospect. Crop Past Sci 2012;(63):787-804

Pecu 2023;14(1):190-203

8. Brock JL, Tilbrook JC. Effect of cultivar of white clover on plant morphology during the establishment of mixed pastures under sheep grazing. NZ J Agric Res 2000;43(3):335343.

9. Gutiérrez-Arenas AF, Hernández-Garay A, Vaquera-Huerta H, Zaragoza-Ramírez JL, Luna-Guerrero MJ, Reyes-Castro S, Gutiérrez-Arenas DA. Análisis de crecimiento estacional de trébol blanco (Trifolium repens L.). Agroproductividad 2018;11(5):62-68.

10. Olivo CJ, Ziech MF, Meinerz GR, Agnolin CA, Tyska D, Both JF. Valor nutritivo de pastagens consorciadas com diferentes espécies de leguminosas. Rev Brasi Zoot 2009;38(8):1543-1552.

11. Rojas GAR, Hernández GA, Ayala W, Mendoza PSI, Joaquín CS, Vaquera HH, Santiago OMA. Comportamiento productivo de praderas con distintas combinaciones de ovillo (Dactylis glomerata L.), ballico perene (Lolium perenne L.) y trébol blanco (Trifolium repens L.). Rev Fac Cienc Agra 2016;48(2):57-68.

12. Rojas GAR, Hernández GA, Rivas JMA, Mendoza PSI, Maldonado PMA. Joaquín CS. Dinámica poblacional de tallos de pasto ovillo (Dactylis glomerata L.) yballico perenne (Lolium perenne L.) asociados con trébol blanco (Trifolium repens L.). Rev Fac Cienc Agra 2017;49(2):35-49.

13. Rojas GAR, Torres SN, Joaquín CS,Hernández-GarayA, Maldonado PMA, Sánchez SP. Componentes del Rendimiento en diferentes variedades de alfalfa (Medicago sativa L.). Agrociencia 2017;51(7):697-708.

14. Matthew C, Lemaire G, Sackville-Hamilton NR, Hernández-Garay A. A modified selfthinning equation to describe size/density relationships for defoliated swards. Ann Botany 1995;76(6):579–587.

15. Moreno CMA, Hernández-Garay A, Vaquera HH, Trejo LC, Escalante EJA, Zaragoza RJL. Joaquín TBM. Productividad de siete asociaciones y dos praderas puras de gramíneas yleguminosas en condiciones de pastoreo. Rev Fito Mex 2015;(38):101-108.

16. García E. Modificaciones al sistema de clasificación climática de Koppen. 4ed. Universidad Nacional Autónoma de México. México, DF. 2004;217.

17. Ortiz SC. Colección de Monolitos. Depto. Génesis de suelos. Edafología, IRENAT. Colegio de Postgraduados. Montecillo, Texcoco, Estado de México. 1997.

18.SAS,Institute.2009.SAS/STAT® 9.2.UseGuideRelease.Cary,NC:SAS Institute.USA.

19. Castro RR, Hernández GA, Pérez PJ, Hernández GJ, Quero CAR, Enríquez QJF. Comportamiento productivo de cinco asociaciones gramíneas-leguminosas bajo condiciones de pastoreo. Rev Fito Mex 2012;35(1):87-95.

20. Lane LA, Ayres JF, Lovett JV. The pastoral significance, adaptive characteristics, and grazing value of white clover (Trifolium repens L.) in dryland environments in Australia: a review. Austr J Exper Agric 2000;40(7):1033-1046.

21. Hernández GA, Hodgson J, Matthew C. Effect of spring grazing management on perennial ryegrass/white clover pastures. 1. Tissue turnover and herbage accumulation. NZ J Agr Res 1997;40:25-35.

22. Karsten HD, Carlassare M. Describing the botanical compositions of a mixed species northeastern U. S. Pasture rotationally grazed by cattle. Crop Sci 2002;(42):882-889.

23. Rojas GAR, Hernández GA, Joaquín CS, Maldonado PMA, Mendoza PSI, Álvarez VP. Joaquín TBM. Comportamiento productivo de cinco variedades de alfalfa. Rev Mex Cienc Agríc 2016;7(8):1855-1866.

24. Álvarez-Vázquez P, Hernández-Garay A, Mendoza-Pedroza SI, Rojas-García AR, Wilson-García CY, Alejos-de la Fuente JI. Producción de diez variedades de alfalfa (Medicago sativa L.) a cuatro años de establecida. Agrociencia 2018;52:841-851.

Rev Mex Cienc Pecu 2023;14(1):190-203

Figure 2: White clover growth curves by season and by morphological component during a growth cycle of 8 and 9 weeks

Figure

3: Botanical and morphological composition of white clover during an 8 and 9 week growth cycle. mm= dead material

Rev Mex Cienc Pecu 2023;14(1):190-203

Figure 4: Leaf area index of white clover during a growing season

Spring= 4.67/(1+0.11*exp(-0.0619t)) r= 0.97

Summer= 0.03+0.11t-0.00152 r= 0.83

Autumn= 3.13/(1+26.85*exp(-0.0666t)) r= 0.98

Winter= 1.70/(1+18.91*exp(-0.1142t)) r= 0.96

https://doi.org/10.22319/rmcp.v14i1.6126 Review

Aspects related to the importance of using predictive models in sheep production. Review

Antonio Leandro Chaves Gurgel a*

Gelson dos Santos Difante a

Luís Carlos Vinhas Ítavo a

João Virgínio Emerenciano Neto b

Camila Celeste Brandão Ferreira Ítavo a

Patrick Bezerra Fernandes a

Carolina Marques Costa a

Francisca Fernanda da Silva Roberto c

Alfonso Juventino Chay-Canul d

a Universidade Federal de Mato Grosso do Sul, Faculdade de Medicina Veterinária e Zootecnia. Avenida Senador Filinto Müler, 2443 - Pioneiros, 79074-460, Campo Grande, Mato Grosso do Sul, Brasil.

b Universidade Federal do Rio Grande do Norte, Unidade Acadêmica Especializada em Ciências Agrárias. Macaíba, Rio Grande do Norte, Brasil.

c Universidade Federal da Paraíba, Centro de Ciências Agrárias. Areia, Paraíba, Brasil.

d Universidad Juárez Autónoma de Tabasco, División Académica de Ciencias Agropecuarias. Villahermosa, Tabasco, México.

* Corresponding author: antonioleandro09@gmail.com

Abstract:

Sheep production systems face numerous challenges, which make decision-making a process fraught with risks and uncertainties. Modelling is a helpful tool in this respect, as it allows decision-makers to evaluate the behaviour of variables and their

interrelationships, in addition to using previous or related information to predict results and simulate different scenarios. The advent of prediction models has made it possible to monitor the weight of an animal and determine the best time for its sale Additionally, it allows producers to estimate the weights of the carcass and major marketable cuts before slaughter. All this information is directly associated with the profitability and success of the production activity. Therefore, in view of the different applications of mathematical models in production systems, this literature review examines concepts in modelling studies and the importance of using prediction models in meat sheep production. Furthermore, it addresses the practical application of modelling studies in predicting dry matter intake and carcass traits of meat sheep through correlated variables.

Key words: Biometric measurements, Carcass, Intake prediction, Mathematical equations, Meat sheep farming, Tropical pasture

Received: 22/12/2021

Accepted: 22/06/2022

Introduction

Sheep farming for meat production in Brazil has expanded in the last decade because of the increased demand for this type of meat in the market Thus, producers have sought ways to establish production systems capable of efficiently generating quality meat at a low cost(1,2,3). However, the productivity of these animals in Brazil is still incipient due to deficiencies in genetic and nutritional management, poor financing, inadequate management systems of the various rearing stages, and low ability to properly organize the production chain(4). Another relevant fact is that over 50 % of sheep production is carried out on natural pastures without management(4). These peculiarities make the decision-making process in sheep production systems fraught with risks and uncertainties.

Despitebeingan inherent characteristicof animal productionsystems, therisk(likelihood of occurrence of an event) can be minimized through the adoption of tools that help decision-making(5). In this sense, a lack of knowledge will result in the impossibility of estimating risk (uncertainty). Some information within and outside the production system is essential to reduce uncertainties associated with decision making(6). Outside the production system, the producer has little control over the actions that impact the profitability of production. In contrast, actions within the farm will directly impact the success of the activity.

In this context, an adequate methodology for decision-making analysis requires accurate information about theproblem as well as efficiencyin handlingthesystem, sotheplanned goals can be achieved(7). Modelling is a tool that can aid the decision-making process, as it allows decision-makers to evaluate the behaviour of variables and their interrelationships, in addition to using previous or related information to predict results and simulate different scenarios(8).

The use of mathematical models allows producers to estimate some important information that would be difficult to obtain in practical terms, e.g., herbage intake using correlated variables(9) Modelling also makes it possible to monitor the weight of an animal and determine the best time for its sale(10,11). In addition, it allows producers to estimate the weight of the carcass and major cuts before slaughter(12,13). All this information is directly associated with the profitability and success of the production activity.

Thus, because of the different applications of mathematical models in production systems, this literaturereviewexamines concepts inmodellingstudies andtheimportance of using predictive models in the production of meat sheep.

Mathematical models

The use of mathematical models has become an indispensable tool for public policymakers and scientists(14). Pool(15) suggested that the act of modelling would become a third domain of science, joining the traditional domains of theory and experimentation. In this sense, important political decisions, such as the effect of global warming on terrestrial biology(16,17), public health, and pandemic management(18) , have come to depend heavily on modelling studies. In addition, researchers have started to use modelling in the most diverse fields of science, e.g. medicine(19), economics(20) , physics(21), chemistry(22) , engineering(23), law(24) , animal science(25,26) , and many others. There are several concepts for mathematical models Hamilton(14) defined them as the expression of theory, which provides a possibility of comparing the theory with data obtained in the physical environment. For Tedeschi(25) , models are mathematical representations of mechanisms that govern natural phenomena that are not fully known, controlled, or understood. More recently, Tedeschi and Mendez(8) postulated that mathematical models are arithmetic representations of the behaviour of real devices and life processes. All these authors also considered that models are an abstraction and a representation of reality(8,14,25)

The use or non-use of mathematics defines whether the model is predictive or descriptive, respectively. Descriptive models theoretically address the performance of variables and their interrelationships. In contrast, predictive models are aimed at using prior

information to predict results or simulate different scenarios(8). In this respect, Tedeschi and Mendez(8) categorized mathematical (predictive) models into three main classes: in a temporal context, models can be classified as static or dynamic; in a natural context, as empirical or mechanistic; and, in a behavioural context, as deterministic and stochastic.

Dynamic models are those that describe changes that have occurred and the obtained responses as a function of time. The non-linear models used to describe animal growth have a dynamic character(27,28) . Static models, on the other hand, are those that generate a response for a fixed instant, that is, they do not include time as a variable(8). However, Tedeschi and Mendez(8) warn that the concept of static versus dynamic depends on the time scale used, as a biological phenomenon can be better represented by a dynamic model when dailychanges occur, but when years are used as a time variant, a static model may work better than a dynamic model, since daily changes are irrelevant to the variable of interest.

Empirical models are obtained from observational data. These models are applied in experimental studies that evaluate dose-response relationships, e.g., the effect of nitrogen rates on the crude protein content of forage plants. Thus, it is possible to estimate the concentration of a crude protein (variable Y) at any nitrogen rate (variable X) through polynomial regression fitting(29). Mechanistic models, on the other hand, consider the underlying conceptual mechanisms and the combination of elements from different hierarchical levels. The main purpose of these models is to explain how an element at a higher level behaves or responds to a range of elements at a lower level. This type of model can be better exemplified in the modelling of the herbage accumulation dynamics of a given forage plant(30). In this case, the mechanistic model seeks to explain the sequence of actions of abiotic factors at the level of molecules, cells, tissues, organs, tillers, plants, clumps, and the forage canopy.

Stochastic models are those that associate a risk or probability with the decision. Stochasticity is associated with a lack of understanding of the biological phenomenon Accordingly, a greater understanding of the phenomenon would translate into a less stochastic model. An example of a stochastic model was developed by Nadal-Roig et al(5) to address tactical decisions, plan production, increase flexibility, and improve the coordination and overall production of swine under the uncertainty associated with the price of animal sales. The authors concluded that the stochastic model was efficient in predicting the best scenario for the production system. Furthermore, due to the market uncertainty of the sales price for swine, the stochastic version led to more precise and realistic results than the deterministic version.

In contrast, deterministic models do not associate anyprobability with a given estimate(8) . Therefore, whenever the model is run without changes in the input variables, the same output information will be obtained. An example of the use of this type of model was proposedbytheNRC(31) to estimatedrymatterintakebysheep.Accordingto theNRC(31) , the dry matter intake (kg/day) of sheep is determined by the following input variables:

adult body weight (kg), body weight, and standard reference weight, which would correspond to an adult animal. If female, it should be considered non-pregnant and nonlactating, with a body score of 2.5 on a scale of 1 to 5, and having already undergone complete skeletal growth. In this way, every time the model is run using the same input information, the predicted intake will be the same. However, the model does not give any probability that the predicted intake will indeed be observed.

Regardless of the type of model, its basic structure is composed of variables, parameters, and constants, but not all models exhibit these three components simultaneously. Variables dependent or independent are changed according to the individual; parameters vary depending on the model, and the constants do not vary in any situation. Inthissense,twoprocessesarecommonlyused forthecreationofmodels:1)Establishing ideasandconceptsthroughanin-depthstudyoftheliteratureandthencreatingparameters for the model variables, or 2) Analyzing experimental data that explain a biological phenomenon and then combining them into an equation(25). In both situations, proper statistical analysis to assess the fit of the models is an indispensable step.

Evaluation of mathematical models

There are conceptual differences between the terms ‘validation’ , ‘verification’, and ‘evaluation’ of mathematical models(14,25). However, the term ‘model validation’ was frequently questioned by researchers(25,32) Because models are considered an abstraction of reality and an approximation of the real system(8,14,25), it is impossible to prove that all model components will truly predict the behaviour of a biological system. Tedeschi(25) proposed the terms ‘evaluation’ or ‘test’ to indicate the degree of robustness of the model based on pre-established criteria. The author also highlighted that mathematical models cannot be proved valid, except if they are suitable for the purpose for which they are intended, under certain conditions.

In modelling studies, a protocol must be followed to define the best prediction model for the established goal. Thus, the process first requires an extensive review of the literature on the topic addressed. After a theoretical understanding of the phenomenon to be modelled is achieved, the next step is to adjust, evaluate, and compare the defined models and, finally, interpret the results and make inferences about the application of the selected models. Therefore, it is understood that evaluation is afundamental stepin theadjustment of prediction models(25), as this step defines whether the model is suitable for its intended purpose. According to Hamilton(14), model evaluation is a comparison of predicted over observed data, which uses statistical tools to support conclusions.

Accuracy and precision are two important concepts when evaluating mathematical models. Accuracy indicates the proximity of predicted to observed mean values. Precision, on the other hand, is the model's ability to consistently predict values(25) .

Therefore, an accuratebut not precisemodel (situation 1in Figure1) estimates anaverage value close to the true average value, but with a high standard deviation. In contrast, a model with low accuracy and high precision (situation 4 in Figure 1) predicts a mean different from the observed data but denotes a low standard deviation in the predictions. In situations 2 and 4 (Figure 1) the models are equally accurate, yet only model 2 shows the characteristics of accuracy and precision, as the points are distributed compactly in the center of the target.

Figure 1: Schematic representation of precision and accuracy concepts (Adapted(25))

The first and simplest assessment of the goodness-of-fit of models (precision and accuracy) is moment analysis of predicted and observed data. In this type of assessment, a good model is expected to estimate mean, maximum, and minimum values as well as data variance and standard deviation close to the observed values(33). Spearman's correlation coefficient value has also been used initially to assess the classification of predicted and observed data values. This coefficient assesses whether the highest predicted value is also the highest observed value, thus creating a classification among all data(34)

Linear regression between observed and predicted values is commonly used to evaluate models. The hypothesis that the predicted data are equal to the observed data is tested by the regression equation Y = β0 + β1 × X, where Y is the observed value; β0 and β1 represent the intercept and slope of the regression equation, respectively; and X is the value predicted by the equations. Model-predicted values are plotted on the X-axis, whereas observed values are plotted on the Y-axis(25). In this graph format, the data points located above and below the equality line indicate overestimation or underestimation by the model, respectively(26) .

To test the hypothesis (β0 = 0 and β1 = 1), Dent and Blackie(35) suggested simultaneously

evaluating whether the intercept is different from zero and the slope is different from one. For this purpose, the simultaneous F test for the identity of the regression parameters predicted by observed data was used(36). However, Tedeschi(25) warned that the F test is valid only for deterministic models and should not be used for stochastic models. In addition, due to the assumption that the data are independent, which is not always observed in a modelling study, the simultaneous F test can result in errors of acceptance or rejection of the tested hypothesis(36) .

After obtaining the linear regression, it is possible to calculate the coefficient of determination of the equation (R2). The coefficient of determination indicates the percentage of variation in Y that is explained byX. Therefore, R2 evaluates the proximity of the data to the fitted regression line. It is noteworthy that the interpretation of the R2 value is often wrong(33) . When used in isolation, this information is not a good indicator of the quality of the model, as R2 measures the precision and not the accuracy of the equation. Coupled with this is the fact that a high coefficient of determination does not necessarily imply that there is a linear relationship between predicted and observed data since the relationship can be curvilinear(25)

Another way to evaluate the regression equation is by the mean square error (MSE), which evaluates the precision of the adjusted linear regression using the difference between the observed values and the values estimated by the regression. Analla(37) recommended MSE as the best criterion to select the model with the best fit when comparing several models. It should be noted that although several methods are used to assess the adequacy of the regression equation, its use may generate ambiguous results when data do not show the normal distribution and in cases in which residual errors are low(25). In this context, some additional evaluations are carried out.

The MSE is similar to the mean squared error of prediction (MSEP). The fundamental difference between the two parameters is that MSEP is the difference between the observed values and the values predicted by the model, while MSE, as seen above, is the difference between the observed values and the values estimated by the regression. Tedeschi(25) considered MSEP the most common and reliable measure to determine the predictive accuracy of a model; however, the author warned that its reliability will decrease as the number of observations decreases. In addition, the author highlighted that MSEP does not provide any information about the precision of the model and that a disadvantage of MSEP is that deviations are weighted by their squared values, which removes the negative data, thus giving greater emphasis to larger values.

Bunke and Droge(38) proposed a decomposition of MSEP that takes into account the source of variation of the parameters. By this fractionation, MSEP is divided into mean error, systematic error, and random error. When most errors are attributed to the mean error, it means that there is a deficiency in the placement of the equality line, which can be corrected with an additive correction factor. Systematic error, on the other hand, indicates a fault in line displacement, which can be corrected with a multiplicative

correction.

The model's coefficient of determination (CD), which shows the proportion of the total variance of observed values that is explained by the predicted data, has long been used to evaluate mathematical models. However, the CD has been replaced by the concordance correlation coefficient (CCC) in studies of continuous variables(39,40). The CCC simultaneously assesses the accuracy and precision of equations, which makes it a powerful measure. The CCC value is obtained by an equation of two components: 1) Correlation coefficient, which measures precision; and 2) Bias correction factor, which indicates accuracy(41)

Numerous statistical techniques are used to assess the precision and accuracy of models. However, no technique used in isolation is capable of adequately evaluating model performance(25). Therefore, the best way to assess the predictive performance of a model is to associate it with a set of statistical methods. It is important to emphasize that this review addresses the main methods used in modelling studies predicting the dry matter intake and carcass traits of sheep(39,40,42). A further discussion on the evaluation of models from a statistical point of view was presented by Neter et al(33) and Tedeschi(25)

Application of predictive models in meat sheep production

Due to the diverse applications of mathematical models in sheep production systems, this literature review will address the application of modellingstudies in predicting drymatter intake by grazing sheep as well as the body weight and carcass traits of sheep through biometric measurements. This information is difficult to obtain under practical conditions; however, it is directly associated with the profitability and success of the production activity. It is highlighted that the possibilities of using modelling in sheep production are as diverse as possible and it would be difficult to summarize all this information in a single review.

Prediction of dry matter intake by grazing sheep

In the case of feedlot animals, the chemical and physical characteristics of ingredients that make up the diets and their interactions have a great effect on dry matter intake (DMI)(43,44). In short, the animal's energy demand defines the consumption of diets with high caloric density(45). On the other hand, when the animal is fed diets of low nutritional valueandlowenergydensity,thephysicalcapacityofthegastrointestinaltractdetermines the potential for DMI(46). In this respect, Mcdowell(47) mentioned that herbage intake is primarily influenced by body size since the size of the animal, is positively correlated

with the nutritional requirements of maintenance(42,43,45) , followed by energy density and the rate of digestion of the diet. Furthermore, the author observed that DMI is positively correlated with organic matter digestibility.

The neutral detergent fibre (NDF) content of a diet or herbage is an efficient parameter to express the action of these two mechanisms to control drymatter intake, as it is positively related to the rumen-fill effect and inversely related to the energy concentration of the diet(48) .

Animal-related factors such as breed, sex, age, body weight, physiological stage (growth, pregnancy, or lactation), and body composition influence the nutrient requirements and intakeof sheep(45).Mertens(48) suggestedthat nutrient intakedepends on important factors related to feeding management (feed availability, linear trough area, feed accessibility, frequency of supply, physical form, and processing), in addition to environmental conditions and animal welfare related to the energy concentration of the diet(48) .

Regarding grazing animals, in addition to all the aforementioned factors acting on dry matter intake, the complex interactions between animal and pasture characteristics affect the nutrient intake rate(49). Feeding behaviour is known to be the most efficient way to demonstrate the interactions between pasture structure and herbage intake(50) .

According to a mechanistic view, the dailyDMI for grazing sheep is the result of the time spent by the animal in searching and prehending the herbage and the intake rate during this period(50), which, in turn, is the product of biting rate and bite weight. The rate and weight of a bite change when the amount of herbage per bite (bite volume) is changed. The bite volume is sensitive to oscillations in bite depth and herbage bulk density, which in turn is determined bycanopyheight and herbage mass (Figure 2). The pasture structure (herbage mass, height, etc.) also changes the time spent by the animal on the grazing activity(51,52) .

Figure 2: Schematic representation of the feeding behaviour of grazing animals (Adapted53)

To exemplify the relationship between pasture structure, feeding behaviour, and intake, one must simply imagine a situation with limitations on the herbage supply. In this circumstance, there is a reduction in bite-size, whereas grazing time and biting rate increase(52). Therefore, to some extent, it is possible to obtain constant herbage intake in pastures with different canopies. Nonetheless, if the herbage allowance is too low, the increase in grazing time will not be able to maintain intake due to the reduction in intake rate(54)

Thus, because of the existence of several factors influencing the DMI of grazing sheep, the modelling of this parameter becomes very complex. For this reason, most sheep DMI prediction models are obtained from experiments conducted in feedlot conditions(31,42,43,45) . This may lead to inconsistencies if they are used to predict the DMI of grazing sheep, as they do not consider the characteristics of the pasture and the interactions between the animal and the forage plant.

Most models used to predict intake by grazing animals are mechanistic, focusing on the digestive process and the selectivity of ingestion under grazing conditions, and they mainly consider pasture height or the amount of herbage removed(50,55). Pittroff and Kothmann(56) undertook acomparative analysis of quantitativemodels predictingthefeed intake of sheep and observed that about 55 % of the equations took into account some

pasture trait, with a predominance of herbage availability. The researchers concluded that there is a weak conceptual framework in the development of the models.

Of the models presented in the review by Pittroff and Kothmann(56) , that developed by Freer et al(57) focused more on pasture characteristics. The equation predicts intake by sheep as the product of potential feed intake (Imax) and the proportion of that potential (relative intake) that the animal can obtain from the available amount of feed. Imax was defined as the amount ingested (kg/d of DM) when the animals are allowed unrestricted access to feed with a DM digestibility of at least 80 %, which depends on the standard weight of an adult animal (standard reference weight) and the ratio between body weight and standard reference weight (equation 1). It is noteworthy that in the case of tropical grasses, the minimum digestibility of 80 % is hardly reached(58,59,60)

(1) Imax= 0.04 × SRW × Z × (1.7 - Z)

WhereSRW=standardreference weight; Z= relativeanimal size, theratioofbodyweight to standard reference weight.

Relative intake was described as the product of two feed attributes: relative availability and relative ingestibility. For grazing animals, relative availability is mainly predicted from the herbage mass, whereas relative ingestibilityis predicted from the digestibilityof the pasture collected by grazing simulation (hand plucking)(57) .

To simulate the ruminant intake dynamics during grazing, Baumont et al(50) developed a theoretical model of an intake rate that combines the pasture structure and the animal's decision to graze or perform other activities. The authors defined dry matter intake as the sum of instantaneous intake rates, which, in turn, are determined as a function of potential intake rates in grazing horizons, preferences that determine the proportions selected in both pasture horizons, and animal satiety levels (equation 2).

(2) IR= (PREFi × PIRi) + (PREFi+1× PIR +

1) / SL

Where IR= intake rate (g DM/min); PREFi and PREFi + 1= relative preference determined from a grazing decision sub-model that defines how the animal distributes intake between the highest available horizon (i) and the next available horizon (i + 1), according to the relative preferences PREFi and PREFi+1; SL= satiety level; PIRi= potential intake rate (g DM/min) obtained from the time taken by the animal to perform the bite and the weight of that bite in the highest available horizon (i), and the next available horizon.

McCall(61) proposed a model to estimate herbage intake in pastures where perennial ryegrass is the predominant forage species. The author modelled the actual DMI of sheep on pasture as a function of the maximum intake multiplied by the correction factor (equation 3). The correction factor is obtained from herbage allowance and the animal's

potential intake (equations 4 and 5). Herbage allowance was estimated by harvesting the forage contained within 1-m2 metal frames.

(3) HI= Imax × M

(4) M= A × EXP (-1 016 × EXP (1 0308 × HAM)

(5) A= 1-1.42 × (-0.00198 × HM),

Where HI= herbage intake (kg/day); Imax= maximum herbage intake (kg/day); M= correction factor; EXP= exponential (2.7182); HAM= herbage allowance divided by maximum intake; and HM= herbage mass minus dead material (kg/ha).

Medeiros(62) used the model proposed by McCall(61) to estimate the intake of sheep on Cynodon spp. under different grazing intensities in a continuous grazing system and concluded that the McCall(61) model overestimated the animals’ intake. This overestimation indicated by Medeiros could be due to the type of grass since McCall worked with a C3 grass that is more digestible than the C4 with which Medeiros worked. Thus, Medeiros(62) suggested replacing the green herbage allowance (leaf + stem) in the equation with a green leaf allowance. Only then was the estimated intake statistically equal to that observed.

Similarly, Gurgel et al(9) evaluated different models predicting DMI in tropical pastures using the adjustment factor proposed by McCall(61) and concluded that the equations do not accurately predict the DMI of meat sheep and generate overestimated values in tropical climate pastures. The authors proposed that the DMI estimate for lambs on tropical pasture should consider the following model (equation 6):

(6) DMI (% LW)= 7.16545 - 0.21799 × LW + 0.00273 × LW2 - 0.00688 × GT + 0.000007 × GT2 + 0.00271 × GHA

Where DMI= dry matter intake (% LW); LW= live weight (kg); GT= grazing time (min/day); and GHA= green herbage allowance (kg DM/100 kg LW), which corresponds to herbage allowance minus dead material.

Therefore, the models proposed for a temperate climate do not correctly estimate herbage intake by sheep on tropical pasture. In this way, studies to estimate intake by sheep in tropical regions are necessary, especially in systems that adopt pasture as the primary source of nutrients, as this information is of fundamental importance for nutritional planning.

Prediction of body weight and carcass traits of sheep through biometric measurements

Body weight is one of the main pieces of information that guides decision-making in production systems due to its direct relationship with the nutritional requirements of animals(31). In addition, monitoring the growth curve of ruminants makes it possible to identify the phases in which the animal is more capable of converting the consumed feed into body tissue and the best time for its sale(10,11,63) .

Animal growth is evaluated using direct measuring equipment, such as livestock scales. However, due to the conditions in which traditional sheep production systems operate(4) , the direct determination of the animals’ body weight often represents a challenge for producers because of the high cost of acquiring and maintaining scales(64,65,66,67) In most cases, this causes producers to market animals based on visual scores, which leads to errors in the estimation of body weight and affects the profitability of production systems(68) .

The estimation of body weight by indirect methods can be an easily adopted, low-cost alternative. In this sense, biometric measurements are a viable option to predict body weight due to the correlation between these traits and the body weight of animals(65,69) . This methodconsists ofdevelopingmathematical models thatallow producersto estimate bodyweight using some biometric measurements (Figure 3) from linear and multiple regression analyses. These body measurements can be obtained with a horse measuring stick and a measuring tape(12,41), easy-to-handle and inexpensive instruments that do not require sophisticated periodic maintenance.

Themainbiometricmeasurements(Figure3)evaluatedinsheepareas follows(70):withers height (WH) – from the highest point of the withers to the ground (1); rump height (RH) – the vertical distance from the highest point of the rump to the ground (2); body length (BL) – from the scapulohumeral joint to the caudal part of the ischium (3); chest width (CW) – the measurement between the tips of the scapulae (4); rump width (RW) – the distance between the ischial tuberosities (5); heart girth (HG) – taken around the chest cavity (6); abdominal circumference (AC) – taken around the abdominal cavity (7); leg length (LL) – taken from the ischial tuberosity to the ground (8); and leg circumference (LC) – taken around the middle portion of the thigh (9).

Figure 3: Main biometric measurements performed on sheep

Withers height (1); rump height (2); body length (3); chest width (4); rump width (5); heart girth (6); abdominal circumference (7); leg length (8); leg circumference (9)

Some studies were conducted to develop linear and multiple equations to estimate the bodyweight of sheep from biometric measurements(65,66,71,72,73) The authors concluded that HG is the most important biometric measurement for predicting animal body weight (Table 1). In contrast, Canul-Solís et al(66) used RW to estimate the body weight of Pelibuey sheep. However, when more than one measurement is used, the predictive capacity of the equations increases(68,69,73,74) .

Table

1:

Equations to predict the body weight (BW) of sheep using biometric measurements (cm)

Author Breed Equation

Chay-Canul et al(65)

Canul-Solís et al(66)

R2

Pelibuey BW (kg) = -47.97 + 1.07 × HG 0.86

Pelibuey BW (kg) = - 19.17 + 3.46 × RW 0.96

Málková et al(68) Charolais; Kent; crossbred BW = (kg) = -3.997 + 0.225 × HG 0.78

Málková et al(68) Charolais; Kent; crossbred BW (kg) = -4.672 + 0.243 × CBC + 0.198 × HG 0.80

Gurgel et al(69) Santa Inês BW (kg) = 0.45 × HG - 0.58× AC + 0.005 × AC2 + 0.002 ×RH2 0.88

Kumar et al(71) Harnali BW (kg) = -63.72 + 1.23 × HG 0.87

Worku(73) Kebeles; ArsiBale BW (kg) = -39.51 + 0.91× HG 0.71

Worku(73) Kebeles; ArsiBale BW (kg) = 45.77 + 0.59 × HG + 1.99 × CBC + 0.30 × CD + 0.5 × RH 0.81

Grandis et al(74) Texel BW = (kg) -107.16 + 1.40 × HG + 0.60 × WH 0.88

HG= heart girth, RW= rump width; CBC= cannon bone circumference; AC= abdominal circumference; RH= rump height; CD= chest depth; WH= withers height.

Another way to use biometric measurements to predict the body weight of sheep is from body volume, which is obtained by the formula used for calculating the volume of a cylinder, including the HG and BL measurements(75):

Radius (cm)= HG / 2π, Body volume (dm3)= (π × r2 × BL) / 1000, where, r= radius of the circumference (cm); π= 3.1416; HG= heart girth (cm); and BL= body length (cm).

Salazar-Cuytun et al(67) compared three equations (linear, quadratic, and exponential) to assess therelationship betweenbodyvolumeand weight in Pelibueylambs andewes. The authors observed a correlation coefficient of 0.89 between body volume and weight. Additionally, the quadratic model was found to have the best performance, according to the adequacy assessment. Le Cozler et al(76) reported that body volume is strongly correlated with weight in lactating Holstein cows.

In addition to being an efficient method to estimate bodyweight, biometricmeasurements are used to predict sheep carcass traits(12,40,77) Determining the yield of carcass or major cuts before slaughter is valuable information for production systems, as it allows the producer to estimate the gross income of the farm. In this regard, the use of biometric

measurements taken before slaughter is of greater interest in commercial production conditions due to the low additional cost for producers(40,78,79) .

Because it is directly related to producer remuneration, the carcass weight has been the variable most predicted by biometric measurements, with slaughter weight explaining 47.0 to 99.0%ofthevariationin ruminant carcass weight(79,80).However, when biometric measurements are used in association with body weight in linear and multiple equations to predict carcass weight, there is an increase in the coefficient of determination(40). In this respect, Gurgel et al(81) showed that the measurements of CW, LC, and RW, together with body weight, explained 91.0 % of the variation in the carcass weight of Santa Inês lambs finished on tropical pasture. For Pelibuey lambs, Bautista-Díaz et al(78) recommended an equation to estimate the carcass weight that is associated with the measurements of BL, HG, and AC and abdominal width (R2= 0.89). In predicting the hot carcass weight of Morada Nova lambs, Costa et al(12) recommended an equation without using body weight as an independent variable. According to the authors, the measurements of BL, WH, CW, AC, and bodycondition scores are the most important in predicting the carcass weight of the studied sheep (R2= 0.80).

Sheep meat is sold mostly in the form of half carcasses or whole carcasses Nevertheless, one way to add value to the meat is by selling it through cuts obtained by sectioning the carcass(82).Thus, the carcass is initiallydivided into themajor cuts ofshoulder,neck,loin, leg, and rib, which are smaller and facilitate marketing, conservation at home, and preparation for consumption(3,82,83) .

Biometric measurements are highly correlated with the major cuts of the carcass(2) Therefore, studies were developed to test the hypothesis that biometric measurements would be efficient in predicting the yield of these cuts. Shehata(13) developed regression models to predict the weight of the major cuts of the carcass of Barki lambs from biometric measurements and found that HG explained 67.0 % of the variation in leg weight, and when HG was associated with BL, this value rose to 72.0 %. In addition, Shehata(13) observed that the HG and BL precisely estimate the weights of the loin roast, shoulder, and loin chop cuts. Abdel-Moneim(84) indicated BL as an efficient variable to predict the shoulder weight of Barki sheep.

The application of biometric measurements is not restricted to predicting carcass weight and major cuts. When used in equations, these measurements estimate the amount of internal fat and carcass trimmings, ribeye area, and the yield of non-carcass components, muscles, bones, and adipose tissue(12,40,77). Thus, the monitoring of biometric measurements is a management tool that can help production systems increase revenues and shorten the time needed for animals to reach slaughter weight.

It is noteworthy that, for the most part, these measurements are carried out on feedlotfinished animals and/or in wool sheep, which does not represent the reality of production systems in tropical regions, since tropical forage grasses are the food base of small and

large ruminants and are responsible for most of the meat produced in the tropics. Therefore, modelling studies must be developed to estimate the weight and carcass traits of hair sheep finished in tropical pastures through biometric measurements, taking into account that genotype, sex, age, rearing system, and health can change carcass traits and composition(85,86)

Conclusions and implications

Despite its low adoption rate, modelling has great potential to help in decision-making in meat sheep production Modelling is a tool capable of predicting the dry matter intake, bodyweight, carcass weight, and major marketable cuts of sheep with high precision and accuracy, through correlated measurements. These equations can be used by researchers, producers, technicians, and the meat industry, thus facilitating activity planning. However, further research is warranted to increase the databases so that the equations can be applied in the most diverse scenarios. In addition, more studies are needed to predict herbage intake using information more easilyobtained in practical production conditions.

Literature cited:

1. Trindade TFM, Difante GS, Emerenciano Neto JV, Fernandes LS, Araújo IMM, Véras ELL, et al. Biometry and carcass characteristics of lambs supplemented in tropical grass pastures during the dry Season. Biosci J 2018;34(1):172-179.

2. Gurgel ALC, Difante GS, Emerenciano Neto JV, Costa MG, Dantas JLS, et al Supplementation of lamb ewes with different protein sources in deferred marandu palisadegrass (Brachiaria brizantha cv.Marandu)pasture. ArqBrasMedVet Zootec 2020;72(5):1901-1910.

3. Araújo CGF, Costa MG, Difante GS, Emerenciano Neto JV, Gurgel ALC, et al. Carcass characteristics, meat qualityand composition of lambs finished in cultivated pastures. Food Sci Technol 2021; Ahead of Print: 1-6.

4. Hermuche PM, Maranhão RLA, Guimarães RF, Carvalho Júnior OA, Gomes RAT, PaivaSR,McmanusC.DynamicsofsheepproductioninBrazil. IntJGeoinformatics 2013;2(3):665-679.

5. Nadal-Roiga E, Plà-Aragonèsa EM, Pagès-Bernausa A, Albornoz VM. A two-stage stochastic model for pig production planning in vertically integrated production systems. Comput Electron Agric 2021;176:e105615.

6. Calvano MPCA, Brumatti RC, Barros JC, Garcia MV, Martins KR, Andreotti R. Bioeconomic simulation of Rhipicephalus microplus infestation in different beef cattle production systems in the Brazilian Cerrado. Agric Syst 2021;194:e103247, 2021.

7. Calvano MPCA, Brumatti RC, Garcia MV, Barros JC, Andreotti A. Economic efficiency of Rhipicephalus microplus control and effect on beef cattle performance in the Brazilian Cerrado. Exp Appl Acarol 2019;79:459-471.

8. Tedeschi LO, Menendez HM. Mathematical modeling in animal production. In: Bazer FW, Lamb GC, Wu G editors. Animal agriculture sustainability, challenges and innovations. 1rst ed. Cambridge: Academic Press; 2020:431-453.

9. Gurgel ALC, Difante GS, Emerenciano NJV, Santana JCS, Fernandes PB, Santos GT, et al Prediction of dry matter intake by meat sheep on tropical pastures. Trop Anim Health Prod 2021;53:e479.

10. Silva FL, Alencar MM, Freitas AR, Packer IU, Mourão GB. Curvas de crescimento em vacas de corte de diferentes tipos biológicos. Pesqui Agropecu Bras 2011;46(3):262-271.

11. Sousa JER, Façanha DAE, Bermejo LA, Ferreira JB, Paiva RDM, Nunes SF, Souza MSM. Evaluation of non-linear models for growth curve in Brazilian tropical goats. Trop Anim Health Prod 2021;53:e198.

12. Costa RG, Lima AGVDO, Ribeiro NL, Medeiros AND, Medeiros GRD, Gonzaga Neto S, Oliveira RL. Predicting the carcass characteristics of Morada Nova lambs using biometric measurements. Rev Bras Zootec 2020;49:e20190179.

13. Shehata MF. Prediction of live body weight and carcass traits by some live body measurements in Barki lambs. Egyptian J Anim Prod 2013;50(2):69-75.

14. Hamilton MA. Model validation: an annotated bibliography. Commun Stat - Theory Methods1991;20(7):2207-2266.

15. Pool R. Is it real, or is it Cray?. Science 1989;244(4811):1438-1440.

16. Devi S, Mishra RP. A mathematical model to see the effects of increasing environmental temperature on plant–pollinator interactions. Modelo Earth Syst Environ 2020;6:1315-1329.

17. Mandal S, Islam MS, Biswas MHA, Akter S. A mathematical model applied to investigatethepotentialimpact ofglobal warmingonmarineecosystems. Appl Math Model 2022;101:19-37.

18. Dover DC, Kirwin EM, Hernandez-Ceron N, Nelson KA. Pandemic Risk Assessment Model (PRAM): a mathematical modeling approach to pandemic influenza planning. Epidemiol Infect 2016;144:3400-3411.

19. Waters SL, Schumacher LJ, Haj AJE. Regenerative medicine meets mathematical modelling: developing symbiotic relationships. Regen Med 2021;6:e24.

20. Brandt AR.Reviewofmathematicalmodelsoffutureoilsupply:Historical overview and synthesizing critique. Energy 2010;35:3958-3974.

21. Madadelahi M, Acosta-Soto LF, Hosseini S, Martinez-Chapa SO, Madou MJ. Mathematical modeling and computational analysis of centrifugal microfluidic platforms: a review. Lab Chip 2020;20:1318-1357.

22. Yu PY, Craciun G. Mathematical analysis of chemical reaction systems. Isr J Chem 2018;58:733-741.

23. Shamsi M, Mohammadi A, Manshadia MKD, Sanati-Nezhad, A. Mathematical and computational modeling of nano-engineered drug delivery systems. J Control Release 2019;307:150-165.

24. Eriksson K. The accuracy of mathematical models of justice evaluations. J Math Sociol 2012;36:125-135.

25. Tedeschi LO.Assessment oftheadequacyofmathematical models. AgricSyst 2006; 89:225-247.

26. Zanetti D, Prados LF, Menezes ACB, Silva BC, Pacheco MVC, Silva FAZ, et al. Prediction of water intake to Bos indicus beef cattle raised under tropical conditions. J Anim Sci 2019;97:1364-1374.

27. Richards FJ. A flexible growth function for empirical use. J Exp Bot. 1959;10:290301

28. Fernandes HJ, Tedeschi LO, PaulinoMF, Detmann E,Paiva LS,Valadares SC,Silva AG, Azevêdo JAG. Evaluation of mathematical models to describe growth of grazing young bulls. Rev Bras Zootec 2012;41:367-373.

29. Leite RG, Cardoso AS, Fonseca NVB, Silva MLC, Tedeschi LO, Delevatti LM, et al Effects of nitrogen fertilization on protein and carbohydrate fractions of Marandu palisadegrass. Sci Rep 2021;11:e14786.

30. Brunetti HB, Boote KJ, Santos PM, Pezzopane JRM, Pedreira CGS, Lara MAS, et al. Improving the CROPGRO Perennial Forage Model for simulating growth and biomass partitioning of guineagrass. Agron J 2021;113:1-16.

31. NRC. National Research Council. Nutrient requirements of small ruminants: sheep, goats, cervids and new world camelids. Washington, DC, USA: National Academy Press; 2007.

32. Oreskes N, Shrader-Frechette K, Belitz K. Verification, validation, and confirmation of numerical models in the earth sciences. Science 1996;263:641-646.

33. Neter J, Kutner MH, Nachtsheim CJ, Wasserman W. Applied linear statistical models. New York, USA: McGraw-Hill Publishing Co.; 1996.

34. Agresti, A. Categorical data analysis. New Jersey, USA: John Wiley & Sons. 2002.

35. Dent JB, Blackie MJ. Systems simulation in agriculture. London: Applied Science; 1979.

36. Mayer DG, Stuart MA, Swain AJ. Regression of real-world data on model output: an appropriate overall test of validity. Agric Syst 1994;45:93-104.

37. Analla M. Model validation through the linear regression fit to actual versus predicted values. Agric Syst 1998;57:115-119.

38. Bunke O, Droge B. Estimators of the mean squared error of prediction in linear regression. Technometrics 1984;26:145-155.

39. Morais MG, Menezes BB, Ribeiro CB, Walker CC, Fernandes HJ, Souza ARDL, et al. Models predict the proportion of bone, muscle, and fat in ewe lamb carcasses from in vivo measurements of the 9th to 11th rib section and of the 12th rib. Semin Cienc Agrar 2016;37:1081-1090.

40. Bautista-Díaz E, Mezo-Solis JA, Herrera-Camacho J, Cruz-Hernández A, GomezVazquez A, Tedeschi LO, et al. Prediction of carcass traits of hair sheep lambs using body measurements. Animal 2020;10:e1276.

41. King TS, Chinchilli VM. A generalized concordance correlation coefficient for continuous and categorical data. Stat Med 2001;20:2131-2147.

42. Cabral LS, Neves EMO, Zervoudakis JT, Abreu JG, Rodrigues RC, Souza AL, Oliveira IS. Nutrients requirements estimative for sheep in Brazilian conditions. Rev Bras Saúde Prod Anim 2008;9:529-542.

43. Vieira PAS, Pereira LGR, Azevêdo JAG, Neves ALA, Chizzotti ML, Santos RD, et al. Development ofmathematical models to predictdrymatterintakein feedlot Santa Ines rams. Small Ruminant Res 2013;122:78-84.

44. Allen MS. Effects of diet on short-term regulation of feed intake by lactating dairy cattle. J Dairy Sci 2000;83:1598-1624.

45. Oliveira AP, Pereira ES, Pinto AP, Silva AMA, Carneiro MSS, Mizubuti IY, et al Estimativas dos requisitos nutricionais e utilização do modelo Small Ruminant Nutrition System para ovinos deslanados em condições semiáridas. Semin Cienc Agrar 2014;35:1985-1998.

46. Romera AJ, Gregorini P, Beukes PC. Technical note: a simple model to estimate changes in dietary composition of strip-grazed cattle during progressive pasture defoliation. J Dairy Sci 2010;93:3074-3078.

47. McDowell LR. Nutrient requirements of ruminants. In: McDowell LR. Nutrition of grazing ruminants in warm climates. Cambridge: Academic Press 1985:21-36.

48. Mertens DR. Regulation of forage intake. In: Fahey Junior GC. Forage quality, evaluation and utilization. Am Soc Agron 1994:450-492.

49. Carvalho PCF. Harry Stobbs Memorial Lecture: Can grazing behavior support innovations in grassland management?. Trop Grassl-Forrages 2013;1:137-155.

50. Baumont R, Cohen-Salmão D, Prache S, Sauvant D. A mechanistic model of intake and grazing behaviour in sheep integrating sward architecture and animal decisions. Anim Feed Sci Technol 2004;112:5-28.

51. Bremm C, Carvalho PC, Fonseca L, Amaral GA, Mezzalira JC, Perez NB, et al Diet switching by mammalian herbivores in response to exotic grass invasion. Plos One 2016;11:e0150167.

52. Gonçalves RP, Bremm C, Moojen FG, Marchi D, Zubricki G, Caetano LAM, et al. Grazing down process: The implications of sheep's ingestive behavior for sward management. Livest Sci 2018;214:202-208.

53. Hodgson J. Grazing management. Science into practice. Longman Group UK Ltd., 1990.

54. Guzatti GC, Duchini PG, Sbrissia AF, Mezzalira JC, Almeida JGR, Carvalho PCF, Ribeiro-Filho HMN. Changes in the short-term intake rate of herbage by heifers grazing annual grasses throughout the growing season. Grassl Sci 2017;63:255-264.

55. Gregorini P, Beukes PC, Romera AJG, Hanigan MD. A model of diurnal grazing patterns and herbage intake of a dairy cow, MINDY: Model description. Ecol Model 2013;270:11-29.

56. Pittroff W, Kothmann MM. Quantitative prediction of feed intake in ruminants: I. Conceptual and mathematical analysis of models for sheep. Livest Prod Sci 2001;71:131-150.

57. Freer M, Moore AD, Donnelly JR. GRAZPLAN: Decision support systems for Australian grazing enterprises II. The animal biology model for feed intake, production and reproduction and the Graz Feed DSS. Agric Syst 1997;54:77-126.

58. Leal ES, Ítavo LCV, Valle CB, Ítavo CCBF, Dias AM, Difante GS, et al. Influence of protodioscin content on digestibility and in vitro degradation kinetics in Urochloa brizantha cultivars. Crop Pasture Sci 2020;72:278-284.

59. Ítavo LCV, Ítavo CCBF, Valle CB, Dias AM, Difante GS, Morais MG, et al Brachiaria grasses in vitro digestibility with bovine and ovine ruminal liquid as inoculum. Rev Mex Cienc Pecu 2021;12:1045-11060.

60. Euclides VPB, Montagner DB, Araújo AR, Pereira MA, Difante GS, Araújo IMM, et al Biological and economic responses to increasing nitrogen rates in Mombaça guinea grass pastures. Sci Rep 2022;12:1937.

61. Mccall DG. A systems approach to research planning to North Island hill country. [Doctoral thesis]. New Zealand, DF: Massey University; 1984.

62. Medeiros HR. Avaliação de modelos matemáticos desenvolvidos para auxiliar a tomada de decisão em sistemas de produção de ruminantes em pastagens. [Doctoral thesis]. Brazil, SP: Universidade de São Paulo; 2003.

63. Gurgel ALC, Difante GS, Emerenciano NJV, Fernandes HJ, Itavo LCV, Itavo CCBF, et al. Evaluation of mathematical models to describe lamb growth during the pre-weaning phase. Semin Cienc Agrar 2021;42:2119-2126.

64. Conrado VDC, Arandas JKG, Ribeiro MN. Modelos de regressão para predição do peso da raça Canindé através de medidas morfométricas. Arch Zootec 2015;64:277280.

65. Chay-Canul AJ, García-Herrera RA, Salazar-Cuytún R, Ojeda-Robertos NF, CruzHernández A, Fonseca MA, Canul-Solís JR. Development and evaluation of equations to predict body weight of Pelibuey ewes using heart girth. Rev Mex Cienc Pecu 2019;10:767-777.

66. Canul-Solis J, Angeles-Hernández JC, García-Herrera RA, Razo-Rodríguez D, LeeRangleHA,Piñeiro-Vázquez AT, et al.Estimationofbodyweight in hairewes using an indirect measurement method. Trop Anim Health Prod 2020;52:2341-2347.

67. Salazar-Cuytun R, Garcia-Herrera RA, Munoz-Benitez AL, Ptacek M, PortilloSalgado R, Bello-Perez EV, Chay-Canul AJ. Relationship between bodyvolume and body weight in Pelibuey ewes. Trop Subtrop Agroecosyst 2021;24:e125.

68. Málková A, Ptáček M, Chay-Canul A, Stádník L. Statistical models for estimating lamb birth weight using body measurements. Ital J Anim Sci 2021;20:1063-1068.

69. Gurgel ALC, Difante GS, Emerenciano NJV, Santana JCS, Dantas JLS, Roberto FFS, et al Use of biometrics in the prediction of body weight in crossbred lambs. Arq Bras Med Vet Zootec 2021;73:261-264.

70. Oliveira DP, Oliveira CAL, Martins ENM, Vargas Junior FM, Barbosa-Ferreira M, Seno LO, et al. Morphostructural characterization of female and young male of naturalized Sul-mato-grossenses “Pantaneiros” sheep. Semin Cienc Agrar 2014;35: 73-986.

71. Kumar S, Dahiya SP, Malik ZS, Patil CS. Prediction of body weight from linear body measurements in sheep. Indian J Anim Res 2018;52:1263-1266.

72. Huma ZE, Iqbal F. Predicting the body weight of Balochi sheep using a machine learning approach. Turk J Vet Anim Sci 2019;43:500-506.

73. Worku A. Body weight had highest correlation coefficient with heart girth around the chest under the same farmers feeding conditions for Arsi Bale sheep. Int J Food Sci Technol 2019;5:6-12.

74. Grandis FA, Fernandes Junior F, Cunha LFC, Dias CBA, Ribeiro ELA, Constantino C, et al. Relação entre medidas biométricas e peso corporal em ovinos da raça Texel. Vet Zootec 2018;25:1-8.

75. Paputungan U, Hendrik MJ, Utiah W. Predicting live weight of Indonesian LocalBali cattle using body volume formula. Livest Res Rural Dev 2018;30:8

76. Le Cozler Y, Allain C, Xavier C, Depuille L, Caillot A, Delouard JM, et al. Volume and surface area of Holstein dairy cows calculated from complete 3D shapes acquired using a high-precision scanning system: Interest for body weight estimation. Comput Electron Agric 2019;165:e104977.

77. Gomes MB, Neves MLMW, Barreto LMG, Ferreira MA, Monnerat JPIS, Carone GM, Morais JSASC. Prediction of carcass composition through measurements in vivo and measurements of the carcass of growing Santa Inês sheep. Plos One 2021;16:1-17.

78. Bautista-Díaz E, Salazar-Cuytun R, Chay-Canul AJ, Herrera RAG, Piñeiro-Vázquez ÁT, Monforte JGM, et al. Determination of carcass traits in Pelibuey ewes using biometric measurements. Small Ruminant Res 2017;147:115-119.

79. Alves AAC, Pinzon AC, Costa RM, Silva MSS, Vieira EHM, Mendonça IB, et al. Multiple regression and machine learning based methods for carcass traits and saleable meat cuts prediction using non-invasive in vivo measurements in commercial lambs. Small Ruminant Res 2019;171:49-56.

80. Castilhos AM, Francisco CL, Branco RH, Bonilha SFM, Mercadante MEZ, Meirelles PRL, et al. In vivo ultrasound and biometric measurements predict the empty body chemical composition in Nellore. J Anim Sci 2018;96:1678-1687.

81. Gurgel ALC, Difante GS, Emerenciano Neto JV, Araujo CGF, Costa MG, Itavo LCV, et al. Prediction of carcass traits of Santa Inês lambs finished in tropical pastures through biometric measurements. Animal 2021;11:e2329.

82. Costa RG, Ribeiro NL, Cavalcante ITR, Roberto FFS, Lima PR. Carne de caprinos e ovinos do Nordeste: Diferenciação e agregação de valor. Rev Cient Prod Anim 2019;21:25-33.

83. Oliveira JPF, Ferreira MA, Alves AMSV, Melo ACC, Andrade IB, Urbano SA, et al Carcass characteristics of lambs fed spineless cactus as a replacement for sugarcane. Asian Australas J Anim Sci 2018;31:529-536.

84. Abdel-Moneim AY. Body and carcass characteristics of Ossimi, Barki and Rahmani ram lambs raised under intensive production system. Egypt J Sheep Goats Sci 2009;4:1-16.

85. Ekiz B, Yilmaz A, Ozcan M, Kocak O. Effect of production system on carcass measurements and meat quality of Kivircik lambs. Meat Sci 2012;90:465-471.

86. Hopkins DL, Mortimer SI. Effect of genotype, gender and age on sheep meat quality and a case study illustrating integration of knowledge. Meat Sci 2014;98:544-555.

https://doi.org/10.22319/rmcp.v14i1.5123

Technical note

Preference for eight plants among captive white-tailed deer Odocoileus virginianus in Veracruz, Mexico

Hannia Yaret Cueyactle-Cano a

Ricardo Serna-Lagunes a*

Norma Mora-Collado a

Pedro Zetina-Córdoba b

Gerardo Benjamín Torres-Cantú a

aUniversidadVeracruzana FacultaddeCienciasBiológicas yAgropecuariasregiónOrizabaCórdoba, Calle Josefa Ortiz de Domínguez s/n Peñuela. Amatlán de Los Reyes, Veracruz, México.

b Universidad Politécnica de Huatusco. México.

*Corresponding author: rserna@uv.mx

Abstract:

Wild white-tailed deer Odocoileus virginianus consume a diversity of high energy plants Captivedeer,however, donot haveaccess to this diversity,which mayaffect their productive capacity. A cafeteria test was used to evaluate intake of and preference for eight plant species among captive deer in Veracruz, Mexico. Three replicates were done of five consecutive days of feeding with the selected plants followed by a 15-d evaluation period. One kilogram of material from each plant species was offered each day and intake recorded. Physicochemical analyses were done of all eight species. Intake results were evaluated with an analysis of variance and a Tukey test, and a partial least squares regression analysis was applied to relate intake to plant characteristics. Intake was highest for four plants: Zapoteca acuelata, Bidens pilosa, Pennisetum purpureum and Parthenium hysterophorus. Preference for these species was determined bytheir fiber and protein contents, and °Brix and pH levels.

Diversifying the diet of captive deer could provide additional feed options for producers and increase animal productivity parameters

Key words: Proximate analysis, Diet, Cervidae, Intake.

Received: 23/10/2019

Accepted: 20/08/2021

White-tailed deer (Odocoileus virginianus; Artiodactyla: Cervidae) is distributed throughout the Americas, from Canadian forests, to coniferous and xerophytic forests in the US, in most forests in Mexico and even in portions of South America(1). It is widely hunted in Mexico(2) , and is raised in Wildlife Conservation Management Units (Unidades de Manejo para la Conservación de la Vida Silvestre - UMA) to produce trophies, meat, skin, brood stock, and ornaments, among other products(3) .

In the wild, O. virginianus is an opportunistic selective herbivore, foraging a selection of plant parts (e.g. shoots, fruits, leaves, bark, and seeds), especially those with high nutritional value(4). When the dry season occurs in deciduous tropical forests plant abundance decreases and their nutritional quality diminishes(5). Under these circumstances, O. virginianus can experience deficiencies in development, such as a lower than standard weight, become prone to disease and limit its reproduction(4) These same responses are often observed in captive O. virginianus Captive deer, fed diets based on sheep and commercial deer feeds as well as alfalfa(6) , produce single rather than twin births, have low birth weight offspring, and longer intervals between births(7)

Adult deer require 5.5 to 9 % crude dietary protein for adequate physiological development(8,9) Protein requirements may be related to ontogeny(9), since captive fawns require between 13 and 20 % protein for adequate development, while, for optimal antler development, 15 to 18 % protein is required(9) Females require from 11 to 18 % protein in pre-breeding, mating, pregnancy, lactation, and to increase offspring count(10) Diet diversification in O. virginianus UMAs is imperative to complement basic feed nutritional value and improve productive characteristics(11). If animal feed preferences, nutrients contained in preferred plants and the nutritional requirements of animals at given weights can be interrelated, then animal productive behavior can be estimated(11) .

Estimates of the nutritional content of plants consumed by wild deer have been done using various methodologies(12,13), but none have been done for captive deer. Cafeteria tests allow quantification and analysis of how animals modify dietary behavior to balance their nutritional needs. Essentially a multiple choice test, animals are offered one or several plants

and their nutritional preferences documented(14) . The present study objective was to use a cafeteria test to quantifythe dietarypreferences of captive O. virginianus offered eight plants as feed.

The studywas carried out at the El Pochote UMA (Secretaría de Medio Ambiente yRecursos Naturales registry: UMA-IN-CR-0122-VER/og), located in Ixtaczoquitlán municipality, in the state of Veracruz, Mexico (coordinates: 18°52’13.70” N; 97°02’59.97” W; 1,137 m asl) Regional climate is predominantly semi-warm humid (Cwa) with abundant summer rains, an average annual temperature of 18 to 24 °C, and average annual rainfall of 1,900 to 2,600 mm Vegetation near the UMA consists of remnant semi-evergreen tropical forest and secondary vegetation.

Experimental animals were two-year-old deer (3 males and 3 females, n = 6), all healthy and with similar body conditions. The cafeteria feeding trial was done over a 60-d period, that is, three replicates of 5 d feeding with the eight selected plants, followed by a 15-d evaluation. Feeding with the selected plants was done for five consecutive days at 0900 h. Independent feeders were randomlydistributed within the pen, and 1 kg fresh material (leaves, shoots and green branches) from each of the tested plants placed in separate feeders (Table 1). To reduce animal subjectivity (deer tend to repeat feeding behaviors), feeder positions were changed daily. After 2 h, the feeders and the remaining plant material were removed from the pen. Intake was quantified with the equation consumption = grams material offered – grams material rejected.

Table 1: Intake (grams) of eight tested plants species by captive white-tailed deer O. virginianus during a cafeteria feeding trial

Plant species Mean Standard deviation Standard error Coefficient of variation Min Max Bidens pilosa 999.6 0.69 0.4 0.07 998.8 1000 Bursera simaruba 516 112.93 65.2 21.89 393 615

Fetusca sp 594.4 44.39 25.63 7.47 559 644.2

Pennisetum purpureum 975.67 23.86 13.78 2.45 949 995 Phartenium hysterophorus 966.27 33.00 19.05 3.45 928.8 991

Saccharum officinarum 797.47 10.71 6.18 1.34 787 808 Vachelia farnesiana 616.4 43.99 25.4 7.14 587.2 667 Zapoteca acuelata 1000 0 0 0 1000 1000

Proximate analyses were done of the eight tested plant species. Three samples of 100 g of mixed material were collected from each plant and incinerated for 2 h at 600 °C. Organic matter, ash, °Brix, pH and acidity were estimated; crude protein was quantified with the Kjeldahl method (N x 6.25) and ether extract in a Soxhtel extractor(15). The intake and physicochemical analysis data were analyzed with descriptive statistics using a central tendency Intake levels by animal were analyzed with an analysis of variance (ANOVA) and a Tukey means test (α=0.05). A partial least squares (PLS) regression analysis was applied in which the dependent variable was intake per plant species, the categorical variables were the eight plants, and the predictor variables were each plant’s physicochemical characteristics. All analyses were run with the Infostat ver 2017 software

Theaverageintakeresults (Table1)showed Bursera simaruba to havethehighest coefficient of variation and the lowest average intake. The ANOVA identified Zapoteca aculeata, Bidens pilosa, Parthenium hysterophorus and Pennisetum purpureum as having the highest intake (correlation coefficient: R²= 0.96, coefficient of variation= 5.94; P<0.05; Table 2). These levels exceeded those of the other evaluated plants (Tukey: minimum significant difference=135.68 g, error=2204.01, gl=16; Figure1) Thisissupportedbythecoefficients of variation, since only these four plants were clearly preferred by the animals. The tested plant species varied in terms of protein, fiber and °Brix (Table 3). The PLS regression analysis explained 61.7 % of the correlation for intake preference of V. farnesiana, B. pilosa, Z. acuelata and S. officinarum, which was related to fiber and protein contents and °Brix level (Figure 2).

Table 2: ANOVA results for plant intake by captive O. virginianus

Source of variation Sum of squares Degrees of freedom Mean square F P-value Plant species 883335.23 7 126190.75 54.77 <0.0001 Error 36864.08 16 2304.01 Total 920199.31 23

Figure 1: Tukey means test results identifying plants with highest intake by O. virginianus

Consumption

Plant species

Table 3: Average physicochemical values of eight plants fed captive O. virginianus Plant species Moistur e (%)

Protein (%) Fat (%) Fiber (%) Ash (%) pH Bri x (°)

Acidity

Bidens pilosa 48.937 18.15 4.728 23.94 1.505 5.5 7.8 0.224 Bursera simaruba 58.437 8.88 3.484 6.03 1.902 5.3 2.7 0.352 Phartenium hysterophor us

63.174 16.02 6.475 39.04 2.202 6.0 2.4 0.032

Saccharum officinarum 63.510 11.19 4.555 17.03 1.164 4.6 6.8 0.256

Vachellia farnesiana 48.016 18.1 0.474 29.04 2.245 5.0 4.5 0.192

Pennisetum purpureum 48.795 14.1 3.011 46.4 2.438 6.0 3.1 0.032

Zapoteca aculeata 41.771 20.5 5.224 22.06 0.352 4.5 9.3 0.64

Festuca sp. 32.375 15.02 6.873 48.02 1.432 4.3 7.8 0.16

Figure 2: Relationship of plant physicochemical characteristics to intake by O. virginianus

%

Consumption 1 Consumption 2

Consumption 3

Of the tested plants, Z. aculeata, B. pilosa, P. purpureum, P. hysterophorus and S. officinarum were preferred by the captive O. virginianus. That these include two herbaceous plants and a grass is of note since wild deer consume mostly shrubs and trees, choose herbaceous species only seasonally, and consume few grasses throughout the year(16,17) . Voluntaryconsumption of bushy, herbaceous and grassyplants reflects nutritional need(18,19) , and is focused on species with the best physicochemical characteristics(20), such as carbohydrates (°Brix) and fiber, both vital to digestibility(21) .

All eight tested plant species meet deer protein requirements according to ontogenic stage. To reach above-average weight male deer require 15 % dietary protein(21), and females require 13 %(22) . In young males, optimal growth requires from 13 to 16 % protein, while 20 % will augment their reproductive activity(22) .

These results suggest that at least five of the tested plant species could be used to diversify the diet of captive O. virginianus, which represents more dietary options for UMAs in this region. In addition, the tested plants have physicochemical characteristics that make them apt for use as deer feed while meeting the productive and reproductive requirements of O. virginianus.

Acknowledgements

The research reported here was financed by the project “Caracterización de recursos zoogenéticos de las altas montañas, Veracruz: aplicación de la filogeografía y modelación ecológica2 (PRODEP: 511-6/18-9245/PTC-896). The authors thank María del Rosario Dávila for carrying out the physicochemical analyses.

Literature cited:

1. Ortega S, Mandujano S, Villarreal J, Di Mare MI, López-Arévalo H, Molina M, CorreaViana M. Managing White-tailed deer: Latín América. In: Hewitt DG editor. Biology and management of White-tailed deer. Boca Ratón, Fl, USA: CRC Press. 2011:565-585.

2. Mandujano S, Delfín-Alfonso CA, Gallina S. Comparison of geographic distribution models of white-tailed deer Odocoileus virginianus (Zimmermann, 1780) subspecies in Mexico: biological and management implications. Therya 2010;1:41-68.

3. Gallina S, Mandujano S, Bello J, López-Fernández H, Weber M. White-tailed deer Odocoileus virginianus (Zimmermann, 1780). In: Barbanti DJM, González S editors. Neotropical Cervidology: Biology and Medicine of Latin American Deer. FUNEP/IUCN. 2009:101-118.

4. Ramírez-Lozano RG. Nutrición del venado cola blanca. Universidad Autónoma de Nuevo León. Monterrey, Nuevo León, México. 2004.

5. Arceo G, Mandujano S, Gallina S, Pérez-Jiménez LA. Diet diversity of white-tailed deer (Odocoileus virginianus) in a tropical dry forest in México. Mamm 2005;69:159-168.

6. Fulbright TE, Ortega-Santos JA. Ecología y manejo de venado cola blanca. Texas A&M University Press. 2007.

7. Henke SE, Demaris S, Pfister JA. Digestive capacity and diets of White-tailed deer and exotic ruminants. J Wild Management 1998;52:595-598.

8. Holter JB, Hayes HH, Smith SH. Protein requirement of yearling white-tailed deer. J Wild Management 1979;1979:872-879.

9.SmithSH,HolderJB, HayesHH,SilverH.Proteinrequirementsofwhitetaileddeerfawns. J Wild Management 1975;39:582-589.

Pecu 2023;14(1):228-236

10. Jones PD, Strickland BK, Demarais S, Wang G, Dacus DC. Nutrition and ontogeny influence weapon development in a long-lived mammal. Canad J Zool 2018;99:1-8.

11. Plata FX, Ebergeny S, Resendiz JL, Villarreal O, Bárcena R, Viccon JA, Mendoza GD. Palatabilidad y composición química de alimentos consumidos en cautiverio por el venado cola blanca de Yucatán (Odocoileus virginianus yucatanensis). Archiv Med Vet 2009;41:123-129.

12. Miller R, Kaneene JB, Fitzgerald SD, Schmitt SM. Evaluation of the influence of supplemental feeding of white-tailed deer (Odocoileus virginianus) on the prevalence of bovine tuberculosis in the Michigan wild deer population. J Wild Dis 2003;39:84-95.

13. Champagne E, Moore BD, Côté SD, Tremblay JP. Spatial correlations between browsing on balsam fir by white‐tailed deer and the nutritional value of neighboringwinter forage. Ecol Evol 2018;8(5):2812-2823.

14. Hartley A, Jones GE. Process oriented supplier development: building the capability for change. Inter J Purch Mat Management 1997;33:24-29.

15. AOAC. Official Methods of Analysis. Association of Official Analytical Chemists. 15th ed. Washington, DC, USA. 1990.

16. Gallina S. White-tailed deer and cattle diets in La Michilia, Durango, Mexico. J Range Management 1993;46:487-492.

17. Granados D, Tarango L, Olmos G, Palacio J, Clemente F, Mendoza G. Dieta y disponibilidad de forraje del venado cola blanca Odocoileus virginianus thomasi (Artiodactyla: Cervidae) en un campo experimental de Campeche, México. Rev Biol Trop 2014;62:699-710.

18. Ramírez GR, Quintanilla JB, Aranda J. White-tailed deer food habits in northeastern Mexico. Small Ruminant Res 1997;25:141-146.

19. López-Pérez E, Serrano-Aspeitia N, Aguilar-Valdés BC, Herrera-Corredor A. Composición nutricional de la dieta del venado cola blanca (Odocoileous virginianus ssp. mexicanus) en Pitzotlán, Morelos. Rev Chap serie Cienc Forest Amb 2012;18:219229.

20. Aguilera-Reyes U, Sánchez-Cordero V, Ramírez-Pulido J, Monroy-Vilchis O, GarcíaLópez GI, Janczur M. Hábitos alimentarios del venado cola blanca Odocoileus virginianus (Artiodactyla: Cervidae) en el Parque Natural Sierra Nanchititla, Estado de México. Rev Biol Trop 2013;61:243-253.

21. Clemente F, Riquelme E, Mendoza GD, Bárcena R, González S, Ricalde R. Digestibility of forage diets of white-tailed deer (Odocoileous virginianus, Hays) using different ruminal fluid inocula. J Appl Anim Res 2005;27:71-76.

22. Ullrey DE, Youatt WG, Johnson HE, Fay LD, Bradley BL. Protein requirement of whitetailed deer fawns. J Wild Management 1967;31:679-685.

https://doi.org/10.22319/rmcp.v14i1.6225

Technical note

Yield and nutritional value of forage brassicas compared to traditional forages

David Guadalupe Reta Sánchez a*

Juan Isidro Sánchez Duarte b

Esmeralda Ochoa Martínez b

Ana Isabel González Cifuentes c

Arturo Reyes González b

Karla Rodríguez Hernández b

a Instituto Nacional de Investigaciones Forestales, Agrícolas y Pecuarias (INIFAP). Campo Experimental Delicias. Km. 2 Carretera Delicias-Rosales. 33000, Centro, Cd. Delicias, Chihuahua, México.

b INIFAP. Campo Experimental La Laguna. Matamoros, Coahuila, México.

c Universidad Juárez del Estado de Durango. Facultad de Agricultura y Zootecnia. Gómez Palacio, Durango, México.

*Corresponding author: reta.david@inifap.gob.mx

Abstract:

The high nutritional value of brassicas can increase productivity in traditional forage productionsystems. The objectiveof thestudywas to comparethe nutritionalvalue and yield of dry matter (DM) and nutrients between forage brassicas and traditional autumn-winter species. The forage brassicas were Winfred, Hunter and Graza radish and the traditional forages were oats, triticale, barley, wheat and berseem clover. The study was conducted in Matamoros, Coahuila, Mexico in the 2018-2019 cycle, under a randomized complete block experimentaldesignwith fourrepetitions. Theregrowthcapacity,thenutritional composition

of the forage and the yields of DM and nutrients were determined. All species showed regrowth capacity, with three cuts in berseem clover in 154 d, and with two cuts in brassicas (150-154 d) and cereals (133-144 d). The brassicas had nutritional composition similar to that of berseem clover and better than that of cereals, mainly due to their higher content of net energyof lactation (NEL; 6.57 to 7.32 MJ kg-1 DM). The DM yields of the brassicas were similar to those observed in traditional forages; however, due to their high nutritional composition, the brassicas were equal to or superior in production of crude protein (CP) (1,608 to 2,986 kgha-1) and NEL (62,819 to 84,044 MJ ha-1) to traditional forages. In general, forage brassicas can increase nutrient yield with respect to cereals and berseem clover, especially in the production of NEL (27.5 to 47.3 %).

Key words: Alternative crops, Dry matter, Regrowth, Crude protein, Energy.

Received: 30/04/2022

Accepted: 11/07/2022

Intensive cow’s milk production is one of the main economic activities in the Comarca Lagunera, Mexico. The forage required by livestock is produced in a production system where the main crops are corn, sorghum, alfalfa, oats and triticale. The production of these crops faces problems of water scarcity, salinity in the soil and high environmental temperatures(1), conditions that will worsen in the next decades due to climate change(2). This situation makes it necessaryto look for new crop options that allow increasing the nutritional value and yields of dry matter and nutrients. An alternative is to increase forage production in autumn-winter using species with regrowth capacity, and good nutritional and production characteristics.

In the Comarca Lagunera, cereals in autumn-winter are produced with one or two cuts in the stages of booting or beginning of heading, which are usually ensiled. Forage brassicas that include species of canola, rapeseed, turnips, suede, kale and radish are a viable alternative for the region due to their production potential, nutritional quality, in addition to their capacity for regrowth(3,4) and silage of their forage(5,6). Brassicas produce 8,000 to 15,000 kg ha-1 of dry matter (DM) in a period of 80 to 150 days after sowing (das). This means that their DM yields may be equal to or higher than autumn-winter forage cereals(3,7). The main benefit of the brassicas is their ability to produce forage with high nutritional value for a relatively long period, since the crude protein (CP) content and the digestibility of DM(8) do not decrease markedly with age. The CP content in brassica forage varies from 134 to 255 g kg-1; the digestibility of DM fluctuates from 85 to 93 %(8,9); the neutral detergent fiber (NDF)

content reaches values from 199 to 516 g kg-1(9,10); and it has high concentrations of energy (NEL) (1.79 to 1.87 Mcal kg-1 DM)(11) .

In studies with stabled dairy cows, it is indicated that brassica forage can be used in the diet of dairy cows without effects on milk production and composition(12,13). Other studies show positive effects of brassica forage with increases in milk production, without negative effects on cow health(14,15). In addition, in studies where the inclusion of brassica forage did not affect milk production and composition, an increase in profitability was observed when pasture silage and commercial concentrates were replaced with forage brassicas(15,16). It is also reported that the use of brassica forage has a favorable environmental effect due to the lower methane production compared to ruminants fed on pasture-based diets(11,17). The objective of the study was to compare the nutritional value and yield of dry matter (DM) and nutrients between forage brassicas and traditional species during the autumn-winter cycle.

The studywas carried out at the La Laguna Experimental Station (CELALA, for its acronym in Spanish) of the National Institute of Forestry, Agricultural and Livestock Research (INIFAP, for its acronym in Spanish), located in Matamoros, Coahuila, Mexico (103° 13’ 42” W and 25° 31’ 41” N, at an altitude of 1,100 m asl). The soil of the experimental site has a clayey-loamy texture, with a depth greater than 1.8 m, water availability values of 150 mm m-1(18), organic carbon content of 0.75 % and a pH of 8.14(1). The preparation of the land consisted of a fallow, double harrowing and leveling of the terrain with laser. Before sowing, each experimental plot was manually fertilized with granular ammonium sulfate and monoammonium phosphate at doses of 50 kg N and 80 kg P2O5, respectively.

Sowing was done manually on October 12, 2018, on this date, sowing irrigation with a 15 cm irrigation sheet was also applied. Eight days after sowing, an overirrigation with a 6 cm sheetwas appliedtofacilitatetheemergenceofseedlings.Thespeciesandcultivarsevaluated were the following: oats (Avena sativa L.), Cuauhtémoc variety; triticale (x Triticosecale Wittmack), Río Nazas variety; barley (Hordeum vulgare L.), Narro 95 variety; wheat (Triticum aestivum L.), AN265 variety; berseem clover (Trifolium alexandrinum L.), Multicut variety; brassica, Winfred cultivar (Brassica oleracea L. x Brassica rapa L.); Hunter cultivar (Brassica rapa L. x Brassica napus L.) and forage radish, Graza cultivar (Raphanus sativus L.x Brassica oleracea L., Raphanus maritimus L.).Duringtheproduction cycle, six supplemental irrigations with a total sheet of 75 cm were applied in oats, triticale, wheat, clover, and Hunter brassica; while in barley, Winfred brassica and Graza radish, five supplemental irrigations with a sheet of 63 cm were applied. The nitrogen fertilization dose (250 kg ha-1) was also completed with 55 kg ha-1 at 33 das, 90 kg ha-1 after the first cut in each species between 77 and 112 das, and 55 kg ha-1 before the second cut between 112 and 135 das.

A randomized complete block experimental design with four repetitions was used. The experimental plot consisted of 20 furrows 0.18 m apart and 6 m long. The useful plot for determining forage yield was 14.4 m2, harvesting 16 central furrows of 5 m in length. At harvest, fresh forage and DM yields were determined. The DM content was obtained in a random sample of 0.72 m2, sampling two of the central furrows of each plot of 2 m in length. The sampled plants were dried at 60 °C in a forced-air oven until reaching constant weight.

DM yield was determined by multiplying the fresh forage yield by the DM content of each plot. In cereals, two harvests were made in the booting stage; clover was harvested three times in the vegetative stage, while the cultivars of brassica and radish were harvested twice in the vegetative stage. The leaf area index (LAI) was determined weekly in all plots of the experiment. For this, an AccuPAR ceptometer model Lp-80 PAR/LAI (Decagon Devices, Inc., Pullman, WA, USA) was used. Three readings per plot were taken between 1200 and 1400 h solar time. Three measurements were made above and below the canopy, parallel to the soil surface. The sensor was placed at an angle of 45° with respect to the furrows.

Plants sampled for the determination of DM content were also used to analyze the nutritional value of forage. The dry samples were ground in a Wiley® mill (Thomas Scientific, Swedesboro, NJ, USA) with a 1 mm mesh. The nitrogen content in each sample was determined using the Dumas combustion method number 990.03 of AOAC, in which the Thermo Scientific Flash 2000 equipment was used, and the result was multiplied by 6.5 to obtain the percentage of crude protein (CP)(19). The neutral detergent fiber (NDF) and the acid detergent fiber (ADF) were obtained according to Goering and Van Soest(20). The content of net energy of lactation (NEL) was estimated following the methodology proposed by Weiss et al(21). CP and NEL yields per hectare were determined by multiplying the CP and NEL contents by the DM yield per hectare estimated for each experimental plot.

For the evaluation of regrowth capacity, the data on DM yield and LAI were analyzed by harvest, using the MIXED procedure for repeated measurements of SAS (P≤0.05)(22). For DM, CP and NEL yields, data from thetwo orthree harvests in each crop were added together to perform the statistical analysis. For the data on the nutritional value of the forage, a weighted average of each parameter evaluated in the harvests carried out was obtained, considering the DM yields. Analyses of variance (P≤0.05) were performed for the variables of nutritional composition and yields of DM and nutrients. The means of these parameters were compared with the protected Fisher’s least significant difference test (P≤0.05). The analysis of the information was performed with the SAS(22) statistical program.

All the species evaluated had regrowth capacity, but berseem clover was superior with three cuts in 156 days. The rest of the species produced two cuts; where the alternative species Winfred brassica, Hunter brassica and Graza radish required the total available period (150 to 154 d); while cereals produced the cuts between 133 and 144 d. This behavior of cereals

allows starting earlier the preparation of the land for the next crop in the spring cycle. However, if this is not so important in the production system, the later harvest of alternative crops does not represent a disadvantage in the use of irrigation water, since these crops required less or equal irrigation sheet than that used in cereals (63 to 75 cm of water sheet).

The regrowth capacity of the hybrids of brassica and the forage radish for the production of two or three harvests in this studyhas also been observed in other works, where it is indicated that several grazings can be carried out in these crops(3,4). Their good regrowth capacity is observed in the little or no reduction of the LAI in regrowth and the higher yields of DM in regrowths comparedtothefirstharvestinWinfred brassica,HunterbrassicaandGrazaradish (Table 1).

Table 1: Growth cycle, dry matter yield recovery (DMY) and leaf area index (LAI) at regrowth after the first cut in traditional and alternative crops evaluated in the autumnwinter cycle of 2018-2019

Treatments Cycle (days)

DMY (kg ha-1) LAI

Cut 1 Cut 2 Cut 3 Cut 1 Cut 2 Cut 3

Cuauhtémoc oat 144 4694 a 6550 a - 6.08 a 4.48 bRío Nazas triticale 141 3718 b 5684 a - 4.20 a 3.55 a -

Narro 95 barley 133 4089 a 5697 a - 5.98 a 5.76 aAN265 wheat 144 4779 a 6534 a - 5.64 a 2.92 b -

Berseem clover 156 3924 a 4183 a 2094 b 3.65 b 6.19 a 3.10 b Winfred brassica 150 4586 b 7430 - 7.20 a 6.26 bHunter brassica 154 3391 a 5178 - 5.82 a 6.30 aGraza radish 154 4483 a 5999 - 6.44 b 8.03 a -

ab Means followed by different letters in each row are significantly different (Tukey-Kramer P≤0.05).

The regrowth capacity observed in traditional crops is in accordance with what is commonly observed in other studies carried out in the Comarca Lagunera. In berseem clover, it has been reported that the Multicut variety produces up to 13.1 t ha-1 of DM in six cuts(23). In cereals such as triticale, oat and barley, it has been observed that they have good capacity to regrowth(24,25), with two to three cuts(26). Generally, greater capacity is observed in winter genotypes, followed by facultative ones and lower in spring ones(27,28). In the present study,

the spring cultivars of Cuauhtémoc oat, Río Nazas triticale and Narro 95 barley had regrowth capacitysimilar to that observed in the facultative wheat AN265, which had a lower recovery of LAI due to its later growth cycle. This represents a disadvantage in an intensive forage production system, since AN265 wheat did not reach its maximum growth in regrowth as spring cereals did.

Of the traditional crops, berseem clover had the best forage nutritional composition, with lower concentrations of NDF (417 g kg-1) and ADF (289 g kg-1), as well as higher contents of CP (286 g kg-1) and NEL (6.44 MJ kg-1 DM) with respect to the values observed in all cereals. Among cereals, Rio Nazas triticale was outstanding for its lower ADF content (372 g kg-1), and higher concentrations of NEL (5.52 MJ kg-1 DM) and CP (189 g kg-1) (Table 2).

Table 2: Nutritional composition of traditional and alternative crops evaluated in the autumn-winter cycle of 2018-2019

Treatments

CP (g kg-1) NDF (g kg-1) ADF (g kg-1) NEL (MJ kg-1 DM)

Cuauhtémoc oat 148.7 d 612.3 a 395.8 c 5.27 e Río Nazas triticale 189.1 c 606.6 a 372.2 d 5.52 d Narro 95 barley 204.7 c 567.3 b 488.7 a 4.27 g

AN265 wheat 165.1 d 628.6 a 418.7 b 5.02 f

Berseem clover 286.4 a 417.1 d 288.6 e 6.44 c Winfred brassica 248.8 b 431.3 d 239.5 f 6.99 b

Hunter brassica 187.8 c 277.0 e 210.4 g 7.32 a Graza radish 198.4 c 456.6 c 280.7 e 6.57 c

CP= crude protein; NDF= neutral detergent fiber; ADF= acid detergent fiber; NEL= net energy of lactation; DM= dry matter.

†Means followed by different letters in each column are significantly different (MSD P≤0.05).

The alternative crops, brassicas and radish, had a better nutritional composition than that observed in cereals, due to their high CP content, lower fiber concentration and higher NEL content. In CP concentration, Winfred brassica (249 g kg-1) exceeded cereals (149 to 205 g); while Hunter brassica (188 g) and Graza radish (198 g) obtained values similar to or higher than those observed in cereals. In berseem clover, the CP content (286 g kg-1) was higher than that observed in the alternative crops, while in concentration of NEL, Winfred brassica

and Graza radish (6.57 to 6.99 MJ kg-1 DM) were higher than that obtained in berseem clover (Table 2).

The results of the nutritional composition of the present study in the forage of brassicas and radish were within the typical range observed in forage brassicas of other works, which were characterized mainly by their high contents of CP (134 to 255 g kg-1)(8,9) and NEL (7.49 to 7.82 MJ kg-1 of DM)(11). However, in this studyin Winfred brassica and Graza radish, higher ADF and NDF contents than those obtained in previous studies were observed, with ADF values of 118 to 217 g kg-1 and 166 to 334 g in NDF(10,11,29). It has been indicated that these NDF concentrations do not meet the minimum values for the proper functioning of the rumen in cows (350 g)(30). In the present study, NDF values in Winfred brassica (431 g) and Graza radish (457 g) were greater than 350 g, and similar to those observed in berseem clover (417 g); while in Hunter brassica (277 g), NDF values were lower than this amount. The high content of NEL in the forage of Hunter and Winfred brassicas, Graza radish and berseem clover was associated with the lower contents of ADF and NDF, in relation to the values observed in cereals harvested in the booting stage.

The alternative crops, Winfred brassica and Graza radish, were outstanding in DM yield (12,016 to 10,482 kg ha-1). These yields were similar to those obtained by berseem clover (10,201 kg) and to the best cereals, Cuauhtémoc oat, Narrro 95 barley and AN265 wheat (9,786 to 11,313 kg). In nutrient production, only berseem clover obtained CP yields (2,871 kg) similar to those of Winfred brassica (2,986 kg), the rest of the crops obtained lower CP yields (1,608 to 2,082 kg). In yield of NEL, Winfred brassica (84,044 MJ) exceeded all other crops evaluated (from 41,689 to 68,722 MJ ha-1) (Table 3).

Table 3: Yields of dry matter (DM), crude protein (CP) and net energy of lactation (NEL) in traditional and alternative crops evaluated in the autumn-winter cycle of 2018-2019

Treatments DM (kg ha-1) CP (kg ha-1) NEL (MJ ha-1)

Cuauhtémoc oat 11244 ab 1672 b 59442 bc

Río Nazas triticale 9402 bc 1781 b 52074 cd

Narro 95 barley 9786 abc 1996 b 41689 d

AN265 wheat 11313 ab 1854 b 57045 bc

Berseem clover 10201 abc 2871 a 65923 bc

Winfred brassica 12016 a 2986 a 84044 a

Hunter brassica 8569 c 1608 b 62819 bc

Graza radish 10482 abc 2082 b 68722 b

abc Means followed by different letters in each column are significantly different (MSD P≤0.05).

The DM yields obtained in the brassicas with two cuts are similar to the best yields reported in other studies in brassicas (10,134 to 14,000 kg ha-1)(31,32). This level of yield in brassicas, and their higher contents of CP and NEL with respect to cereals resulted in higher yields of these nutrients per hectare. In relation to berseem clover with a high CP content, the brassicas obtained similar CP yields for their high DM yield; however, in NEL yields, Winfred brassica was superior to all species as a result of a combined effect of a high NEL content (Table 2) and a high DM production (Table 3).

An aspect to highlight in the study was the ability of forage brassicas to produce yields of DM and nutrients similar to or greater than those obtained with traditional species, with irrigation sheets (63 to 75 cm) less than or equal to those used in traditional crops. These results are important in a forage production system such as that of the Comarca Lagunera, which has a shortage of water for irrigation.

In conclusion, forage brassicas have the potential to increase productivity in forage production in autumn-winter due to their high nutritional value, good regrowth capacity and high production of DM and nutrients. Of the species evaluated, Winfred brassica was outstanding with respect to traditional crops mainly due to its higher content and production of NEL (27.5 to 47.3 %).

Pecu 2023;14(1):237-247

Literature cited:

1. Santamaría CJ, Reta SDG, Chávez GJFJ, Cueto WJA, Romero PRJI. Caracterización del medio físico en relación a cultivos forrajeros alternativos para la Comarca Lagunera. Libro Técnico Núm. 2. INIFAP-CIRNOC-CELALA. México; 2006.

2. Andrade VM, Montero MJ. Nuevas proyecciones de cambio de precipitación y temperatura para el siglo XXI en el Norte de México. Herrera E, López M, Carrillo J editores. Memorias del segundo congreso cambio climático del Estado de Chihuahua. Primera ed. 2014:26-35.

3. Bell LW, Watt LJ, Stutz RS. Forage brassicas have potential for wider use in drier, mixed crop-livestock farming systems across Australia. Crop Pasture Sci 2020;71(10):924943. https://doi.org/10.1071/cp20271

4. Umami N, Prasojo YS, Haq MS. Morphological characteristics and biomass production Brassica rapa var. Marco during the dry season. Anim Prod 2022;24(1):31-36. https://doi.org/10.20884/1.jap.2022.24.1.107

5. Sánchez DJI, Serrato CJS, Reta SDG, Ochoa ME, Reyes GA. Assessment of ensilability and chemical composition of canola and alfalfa forages with or without microbial inoculation. Indian J Agric Res 2014;48(6):421-428. https://doi.org/10.5958/0976058x.2014.01325.0.

6. Kilic U, Erisek A, Garipoğlu AV, Ayan I, Onder H. The effects of different forage types on feed values and digestibilities in some brassica fodder crops. Turkish J Agric Natural Sci 2021;8(1):94-102. https://doi.org/10.30910/turkjans.747031.

7. Watt LJ, Bell LW, Cocks BD, Swan AD, Stutz RS, Toovey A, De Faveri J. Productivity of diverse forage brassica genotypes exceeds that of oats across multiple environments within Australia´s mixed farming zone. Crop & Pasture Sci 2021;72(5):393-406. https://doi.org/10.1071/CP21034

8. Villalobos LA, Brummer JE. Forage brassicas stockpiled for fall grazing: yield and nutritive value. Crop Forage Turfgrass Management 2015;1(1):1-6. https://doi.org/10.2134/cftm2015.0165.

9. Dillard SL, Billman ED, Soder KJ. Assessment of forage brassica species for dairy and beef-cattle fall grazing systems. Appl Anim Sci 2020;36(2):157-166. https://doi.org/10.15232/aas.2019-01921.

10. Omokanye A,Hernández G, LardnerHA,Al-Maqtari B, Singh Gill K, LeeA.Alternative forage feeds for beef cattle in Northwestern Alberta, Canada: forage yield and nutritive value of forage brassicas and forbs. J Appl Anim Res 2021;49(1):203-210. https://doi.org/10.1080/09712119.2021.1933990.

Pecu 2023;14(1):237-247

11. Dillard SL, Roca-Fernández AI, Rubano MD, Elkin KR, Soder KJ. Enteric methane production and ruminal fermentation of forage brassica diets fed in continuous culture. J Anim Sci 2018;96(4):1362-1374. doi: 10.1093/jas/sky030

12. Keim JP, Castillo M, Balocchi O, Pulido R,Pacheco D, Muetzel S. Brief communication: milk production responses and rumen fermentation of dairy cows supplemented with summer brassica crops. NZ J Anim Sci Prod 2018;78:122-124.

13. Vargas-Bello-Pérez E, Geldsetzer-Mendoza C, Ibáñez RA, Rodríguez JR, AlvaradoGillis C, Keim JP. Chemical composition, fatty acid profile and sensory characteristics ofchanco-stylecheesefromearlylactationdairycowsfedwinterBrassicacrops.Animal 2021;11(1):107. https://doi.org/10.3390/ani11010107.

14. Williams SRO, Moate PJ, Deighton MH, Hannah MC, Wales WJ, Jacobs JL. Milk production and composition, and methane emissions from dairy cows fed lucerne hay with forage brassica or chicory. Anim Prod Sci 2016;56(3):304-311. https://doi.org/10.1071/AN15528.

15. Keim JP, Daza J, Beltrán I, Balocchi OA, Pulido RG, Sepúlveda-Varas P, Pacheco D, Berthiaume R. Milk production responses, rumen fermentation, and blood metabolites ofdairycows fedincreasing concentrations of forage rape (Brassica napus ssp. biennis). J Dairy Sc 2020;103(10):9054-9066. https://doi.org/10.3168/jds.2020-18785

16. Castillo-Umaña M, Balocchi O, Pulido R, Sepúlveda-Varas P, Pacheco D, Muetzel S, Berthiqume R, Keim JP. Milk production responses and rumen fermentation of dairy cows supplemented with summer brassicas. Animal 2020;14(8):1684-1692. https://doi.org/10.1017/S175173112000021X.

17. Sun XZ. Invited review: glucosinolates might result in low methane emissions from ruminants fed brassica forages. Frontiers in Vet Sci 2020;7:588051. https://doi.org/10.3389/fvets.2020.588051.

18. Santamaría CJ, Reta SDG, Orona CI. Reducción del rendimiento potencial de maíz forrajeroencalendarioscontres ycuatroriegos. TerraLatinoamericana2008;26(3):235241.

19. AOAC (Association of Official Agricultural Chemists). Official methods of analysis. Dumas method (99003). 15th edition Washington DC, USA. 2005.

20. Goering HK, Van Soest PJ. Forage fiber analysis Apparatus, reagents, procedure and some applications Agric Handbook 379 ARS. Washington, DC, USDA; 1970.

Pecu 2023;14(1):237-247

21. Weiss WP, Conrad HR, St-Pierre NR. A theoretically-based model for predicting total digestible nutrient values of forages and concentrates. Anim Feed Sci Technol 1992;39(1-2):95-110. https://doi.org/10.1016/0377-8401(92)90034-4

22. SAS Institute. The SAS system for windows, release 93. Cary, NC: Statistical Analysis Systems Inst; 2011.

23. Núñez HG, Quiroga GHM, Márquez OJ de J, de Alba AA. Producción ycalidad de trébol de Egipto (Trifolium alexandrinum L.) para ganado lechero en el Norte y Centro de México. Agrociencia 1997;31(2):157-164.

24. Keles G, Ates S, Coskun B, Koc S. Re-growth yield and nutritive value of winter cereals. In: Proc 22nd Int Grassland Cong. 2013.

25. Wilson GCY, López ZNE, Ortega CME, Ventura RJ, Villaseñor MHE, Hernández GA. Acumulación de forraje, composición morfológica e intercepción luminosa en dos variedades de avena. Interciencia 2018;43(9):630-636.

26. Zamora VVM, Lozano del RAJ, López BA, Reyes VMH, Díaz SH, Martínez RJM, Fuentes RJM. Clasificación de triticales forrajeros por rendimiento de materia seca y calidad nutritiva en dos localidades de Coahuila. Téc Pecu Mex 2002;40(3):229-242.

27. Lozano del RAJ, Rodríguez SA, Díaz SH, Fuentes RJM, Fernández BJM, Fernando NMJM, Zamora VVM. Producción de forraje y calidad nutritiva en mezclas de triticale (X Triticosecale Wittmack) y ballico anual (Lolium multiflorum L.) en Navidad, N.L. Téc Pecu Méx 2002;40(1):17-35.

28. Ye CWE, Díaz SH, Lozano del RAJ, Zamora VVM, Ayala OMJ. Agrupamiento de germoplasma de triticale forrajero por rendimiento, ahijamiento y gustosidad. Téc Pecu Méx 2001;39(1):15-29.

29. Villalobos L, Brummer J. Evaluation of Brassicas for fall forage. In: Proc Western States Alfalfa and Forage Symp, Reno, NV, December, 2013. UC Cooperative Extension, Plant Science Department University of California, Davis, CA 95616.

30. Kolver ES. Nutrition guidelines for the high producing dairy cow. Proc Ruakura Farmers Conference; 2002.

31. Stewart AV, Moorhead AJ. The development of a fodder radish suitable for multiple grazing. Agronomy NZ 2004;34:1-7.

32. McGrath S, Sandral G, Holman B, Friend M. Lamb growth rates and carcass characteristics of White Dorper and crossbred lambs grazing traditional and novel pastures during spring in souther Australia. Anim Prod Sci 2020;61(11):1151-1159. https://doi.org/10.1071/AN18769

https://doi.org/10.22319/rmcp.v14i1.6200

Technical note

Genetic characterization of bovine viral diarrhea virus 1b isolated from mucosal disease

Roberto Navarro-López a

Juan Diego Perez-de la Rosa b

Marisol Karina Rocha-Martínez b

Marcela Villarreal-Silva a

Mario Solís-Hernández a

Eric Rojas-Torres a Ninnet Gómez-Romero a* aComisiónMéxico-EstadosUnidosparalaprevencióndefiebreAftosa yotrasenfermedades exóticas de los animales, Carretera México-Toluca Km 15.5 Piso 4 Col. Palo Alto. Cuajimalpa de Morelos. 05110. Ciudad de México. México. b Centro Nacional de Servicios de Constatación en Salud Animal (CENAPA), Morelos, México. *Corresponding author: ninnet.gomez.i@senasica.gob.mx; ninna_gr@hotmail.com

Abstract:

This report describes a fatal case of mucosal disease in a two-year-old bull. For causal agent detection, scab, whole blood, and feces samples were tested by RT-PCR, PCR, ELISA, and viral isolation. RT-PCR positive amplification was obtained in blood samples for bovine viral diarrhea virus (BVDV). Viral isolation from the scab samples confirmed BVDV as the causative agent of the clinical manifestations. Subsequently, genetic characterization based on phylogenetic analysis of three partial sequences revealed the presence of BVDV subgenotype 1b in analyzed samples. Due to the development of clinical manifestation

named mucosal disease, these findings suggest the detection of BVDV persistently infected (PI) bull; therefore, these results demonstrate the importance of establishing BVDV control programs that rely on testing the presence of PI in cattle from Mexico.

Key words: Bovine viral diarrhea virus, Cattle, Mucosal disease, Persistent infection, Mexico. Received: 18/04/2022

Accepted: 18/07/2022

Bovine viral diarrhea (BVD) remains one of the most common endemic diseases of cattle and other ruminant populations worldwide. Furthermore, BVD has a significant economic impact on the cattle industry due to its negative effects on cattle reproduction and health conditions(1,2).BVDis causedbyapositive-sensesingle-strandedRNAgenomevirustermed bovine viral diarrhea virus (BVDV), belonging to the Flaviviridae family within the Pestivirus genus. BVDV is currently divided into three species: Pestivirus A (Bovine viral diarrhea virus 1, BVDV-1), Pestivirus B (Bovine viral diarrhea virus 2, BVDV-2), and Pestivirus H (HoBi-like pestivirus), which are segregated into subgenotypes(3) . Pestivirus A is subdivided into up to 21 subgenotypes (1a to 1u), Pestivirus B, and Pestivirus H into four subgenotypes each (a to d)(4) . Further, BVDV strains are classified in cytopathic (CP) and non-cytopathic (NCP) biotypes according to their effect on replication and morphological changes induced in cell culture. This classification is relevant because cytopathogenicity in vitro is not related to cytopathogenicity in vivo. Thus, NCP strains are predominant in the field, involved in most natural infection cases and persistent infections. In contrast, CP strains are rare and isolated almost exclusively from a fatal form of BVD named mucosal disease (MD)(5) .

BVDV infection is characterized by clinical manifestations, including respiratory, gastrointestinal, and reproductive disorders. However, reproductive failures such as abortions, mummification, stillbirth, congenital defects, and the birth of persistentlyinfected animals (PI) are considered of major economic importance(6)

PIanimalsaregeneratedasaresultoftransplacentalinfectionwithNCPBVDVstrainduring the first 125 d of gestation. Such animals acquire immunologic tolerance towards the infecting BVDV strain and develop persistent infection; hence, a PI calf will not induce an immune response by antibodies or T-cells against the virus(7). Additionally, PI cattle shed the virus in body secretions like nasal and oral discharges, milk, urine, feces, and semen

throughout their entire lives. Therefore, they are considered a permanent source of viral infection and play an essential role in BVD pathogenesis and epidemiology(8) .

Calves born as PI appear normal and sometimes as weak animals but are characterized by reduced growth rates, immunosuppression, and high death rate(2). Moreover, PI has increased morbidity and mortality rates owing to susceptibility to other diseases and may eventually die from pneumonia or MD. Most PI calves succumb to MD, usually between 6 to 24-mo old(9,10) Nevertheless, older PI cattle of 3, 5, and 7-yr old have been previously reported, implying a broader viral dissemination period(2,11,12)

MD is a sporadic fatal condition restricted to PIcattle that occurs when the PIcausative NCP BVDV mutates into CP as a result of a recombination event or when the PI animal is coinfected with an antigenically homologous related strain of CP BVDV(13,14) Therefore, both biotypes can be consistently found in animals with MD(15,16) . The outcome of MD is death occurring within two weeks after the onset of the clinical signs. Erosions and extensive ulceration of the gastrointestinal tract are the main lesions found(17) . Conversely, late MD onset after several months has also been described(18). Other clinical signs include anorexia, fever, dehydration, diarrhea, dermatitis, necrosis of lymphoid tissue, poor condition, and death(19) .

This case report describes the onset of MD in a two-year bull with severe clinical signs suggesting the description of PI cattle from Mexico for the first time.

On June 2021, a 2-year-old bull was reported with 15 d course of clinical signs including anorexia, depression, ptyalism, severe hemorrhagic watery diarrhea, dehydration, nasal discharge, and deep and extensive ulceration in muzzle, nares, lips, gums, and hard palate (Figures 1, 2, and 3). The affected animal belonged to a traditional backyard farm located in Texcoco, State of Mexico, Mexico. The farm kept four bovines, four horses, six dogs, and threepigs,apparentlyhealthyatthereport.Nosimilarclinicalmanifestations wereregistered in the neighboring farms prior to the event. According to the owner, no animal mobilization among nearby farms, and new animals were introduced.

Figure 1: Two years old bull with mucosal disease presentation showing erosive lesions in nasal discharge, extensive ulceration in muzzle and nares

Figure 2: Erosive lesions in lips and gums

Figure 3: Superficial erosions in hard palate

Scab samples from skin lesions, whole blood, and feces samples were obtained and submitted for diagnosis to the Immunology, Cellular and Molecular Biology Laboratory from the Comisión México-Estados Unidos para la prevención de fiebre Aftosa y Otras enfermedades exóticas de los Animales (CPA). The case report was identified with the number CPA-0861-21. Main vesicular cattle diseases were considered for differential diagnosis, including foot and mouth disease (FMD), vesicular stomatitis (VS), malignant catarrhal fever disease (MCF), and BVD using RT-PCR, PCR, ELISA, and virus isolation. Negative results were obtained on viral isolation in cell culture, RT-PCR, and ELISA for FMD and VS. Similarly, the MCF virus was not detected by PCR in surveyed samples.

Conversely, BVDVwas isolatedfrom scabsamples, and positiveamplificationwas obtained from whole blood samples using RT-PCR. Consequently, the BVDV isolate was submitted to the Molecular Biology Laboratory of the Centro Nacional de Servicios de Diagnóstico en Salud Animal (CENASA) for partial sequencing. The 5'UTR, Npro, and E2 BVDV sequences obtained were deposited in GenBank under accession numbers OM812936, OM812937, and OM812938, respectively. Moreover, phylogenetic analysis was performed based on5'UTR,Npro, andE2regions. Partial 5'UTR (360bp), Npro (504bp),and E2(1482 bp) sequences obtained in this study were compared to BVDV reference strains to characterize BVDV isolate. The evolutionary history was inferred using the Maximum likelihood method with a Kimura 2-parameter substitution model(20) for 5'UTR and Npro sequences, and a Tamura 3-parameter substitution model(21) for the E2 sequences was conducted in commercial software MEGA7 using 1000 bootstrap replicates each (Figure

4). A discrete gamma distribution with two categories was used to model evolutionary rate differences among sites, with some sites being evolutionary invariable for Npro and E2 sequences.

Figure 4: Phylogenetic tree based on 5'UTR region (a), Npro (b), and E2 (c) sequences

Phylogenetic inference was conducted in MEGA 7 according to maximum likelihood method. Analysis was supported by 1000 bootstraps replicates. Reference sequences are identified by GenBank accession number. Mexican nucleotide sequences are highlighted with symbol ""

BVD continues to be a significant concern to the cattle industry, with substantial economic impact mainly associated with reproductive disorders(2). Depending on the stage of pregnancy at the time of infection, transplacental infections with BVDV NCP strains may result in the birth of immunotolerant PI calves. These animals are consistently viremic, BVDV spreads through most organs in the animal, but no apparent lesions are developed(22) Consequently, PI cattle sustain lifelong viral replication and excretion in all body secretions(23) Thus, PI animals represents the main transmission and maintenance source of BVDV within and between herds. Moreover, NCP BVDV can also be transmitted from acutely infected cattle and by fomites such as contaminated surgical and handling material, rectal examination, bovine sera used in embryo transfer, and vaccine production, infected semen, and contaminated vaccines(24-27) .

Further, BVDV infections directly impact PI animals' fertility, i.e., PI bulls can produce semen of acceptable quality. However, they are associated with poor fertility related to spermatozomal abnormalities and low motility(28) . Likewise, BVDV infections alter ovarian function by causing hypoplasia and reduced ovulations in PI cows(29). Nevertheless, bulls and PI cows can still sire normal PI offspring, which may recirculate BVDV in susceptible dams(30) .

Continual exposure of healthy animals to BVDV from a PI animal may lead to the perpetuation of BVDV infections(31); thus, herd infertility, immunosuppression, and generation of new PI calves may arise(32). Furthermore, acute NCP infections compromises herd fertility by producing retarded and reducing follicle growth(33) and diffuse interstitial ovaritis(34), and conception failure by preventing embryo implantation(22). In addition, embryonic death before d 79 of gestation in pregnant cows or congenital malformations between days 79 and 150 can also occur(35)

In areas where adequate BVDV control measures are implemented,the estimatedprevalence of PI animals is around 1%-2 %(1); however, no report of MD outbreaks nor presentation in the Mexican bovine population has been previously described. In addition, the current proportion of PI's calves in the country remains unknown. Recently, limited information regarding the BVDV genetic characterization and prevalence in Mexico has begun to be surveyed(36) .

In the present study, it was described a case of MD by BVDV-1b affecting a beef bull in which ulcerative lesions in the gastrointestinal tract were predominant. BVDV-1b is currently defined as the most common strain found in the field; thus, it is considered the predominant subgenotype worldwide, followed by 1a and 1c(4). BVDV-1b is also described as the most prevalent strain in PI calves(37). Similar to these studies, the genetic characterization of the virus isolated from the evaluated bull in this study, reveals the identification of BVDV subgenotype 1b. The latter correlates to a previous study where BVDV-1b was described as an endemic virus circulating in Mexican cattle, together with 1a, 1c, and 2a(38). Despite these initial efforts to report BVDV cases, BVD remains a nonregulated disease hence no control strategies nor prevention measures are officially implemented.

Consequently, vaccination protocols are based on voluntary procedures, and monitoring and biosafety measures are applied depending on cattle producers' BVD knowledge. The evaluated bull from this clinical case belongs to a farm where scarce sanitary measures and no vaccination practices against BVDV are applied. BVDV positive tests and clinical presentation suggest an MD case developed in a PI bull of 2 yr old.

The latter has important implications for BVD control in the nation. These results confirm the presence of BVDV-1b circulating in Mexican cattle, similar to the findings reported by Gómez-Romero et al(38). Clinical presentation from the case highlights the severe outcome of MD and the relevance of underdiagnoses of PI animals and, therefore, BVDV epidemiological status. Furthermore, national BVD case reports will impulse the developmentofcontrolstrategiesthatallowproducerstodetectBVDVandremovePIcalves fromtheherd.Moreover,whenvaccinationisapplied,thechoiceofaspecificvaccineshould be evaluated for protection provided against circulating BVDV. In Mexico, the recent

addition of BVDV-1b as vaccine antigen has been included in one commercial vaccine; however, vaccination alone is not adequate for the BVD control programs. The finding of BVDV-1b in a non-vaccinated bull demonstrates the crucial role of biosecurity and disease surveillance to mitigate the effects of BVDV infections in cattle populations.

Conflict of interest

The authors declared no conflict of interest regarding the authorship or publication of this manuscript

Literature cited:

1. Houe H. Epidemiological features and economical importance of bovine virus diarrhoea virus (BVDV) infections. Vet Microbiol 1999;64(2-3):89-107. doi: 10.1016/s03781135(98)00262-4. Erratum in: Vet Microbiol 2003;93(3):275-6.

2. Houe H. Economic impact of BVDV infection in dairies. Biologicals 2003;31(2):13743. doi: 10.1016/s1045-1056(03)00030-7.

3. SmithDB,MeyersG,BukhGJ,GouldEA,MonathT,Scott-MuerhoffA, et al.Proposed revision to the taxonomy of the genus Pestivirus, family Flaviviridae. J General Virol 2017;98(8), 2106–2112. https://doi.org/10.1099/jgv.0.000873.

4. Yesilbag K, Alpay G, Becher P. Variability and global distribution of subgenotypes of bovine viral diarrhea virus. Viruses 2017;9(6):128. doi: 10.3390/v9060128.

5. Neill J. Molecular biology of bovine viral diarrhea virus. Biologicals 2013;41:2–7. doi: 10.1016/j.biologicals.2012.07.002.

6. Houe H, Lindberg A, Moenning V. Test strategies in bovine viral diarrhea virus control and eradication campaigns in Europe. J Vet Diagn Invest 2006;18(5):427-36. doi: 10.1177/104063870601800501.

7. Meyers G, Thiel HJ. Molecular characterization of pestiviruses. Adv Virus Res 1996; 47:53-118. doi: 10.1016/s0065-3527(08)60734-4.

8. Brodersen BW. Bovine viral diarrhea virus infections: manifestations of infection and recent advances in understanding pathogenesis and control. Vet Pathol 2014;51(2):45364. doi: 10.1177/0300985813520250.

9. Odeón AC, Leunda MR, Faverín C, Boynak N, Vena MM, Zabal O. In vitro amplification of BVDV field strains isolated in Argentina: effect of cell line and culture conditions. Rev Argent Microbiol 2009;41:79–85.

Pecu 2023;14(1):248-259

10. Uzal FA, Platnner BL, Hostetter JM. Alimentary system in pathology of domestic animals. In: Maxie, MG, editor. Jubb, Keneddy and Palmers pathology of domestic animals.;6th ed. St. Louis, Missouri: Academic Press Inc; 2016:122–130.

11. Brock KV, Grooms DL, Ridpath JF, Bolin SR. Changes in levels of viremia in cattle persistently infected with bovine viral diarrhea virus. J Vet Diagn Invest1998;10:22-26.

12. Bedekovic T, Lemo N, Lojkic I, Cvetnicz Z, Cac Z, Madic J. Bovine viral diarrhoea: A seven year old persistently infected cow - a case report. Veterinarski Arch 2012;82 (6):637-643.

13. Brownlie J. The pathways for bovine virus diarrhoea virus biotypes in the pathogenesis of disease. Arch Virol Suppl 1991;3:79-96. doi: 10.1007/978-3-7091-9153-8_10.

14. Tautz N, Thiel HJ. Cytopathogenicity of pestiviruses: cleavage of bovine viral diarrhea virus NS2-3 has to occur at a defined position to allow viral replication. Arch Virol 2003;148(7):1405-12. doi: 10.1007/s00705-003-0106-9.

15. Bolin SR, McClurkin AW, Cutlip RC, Coria MF. Severe clinical disease induced in cattle persistently infected with noncytopathic bovine viral diarrhea virus by superinfection with cytopathic bovine viral diarrhea virus. Am J Vet Res 1985;46(3):573-6.

16. Kummerer B, Tautz MN, Becher P, Thiel H, Meyers G. The genetic basis for cytopathogenicity of pestiviruses. Vet Microbiol 2000;77(1-2):117-28. doi: 10.1016/s0378-1135(00)00268-6.

17. Baker JC. The clinical manifestations of bovine viral diarrhea infection. Vet Clin North Am Food Anim Pract 1995;11(3):425-45. doi: 10.1016/s0749-0720(15)30460-6.

18. FritzemeierJ, Haas L, LieblerE, MoennigV, Greiser-Wilke I. Thedevelopmentofearly vs. late onset mucosal disease is a consequence of two different pathogenic mechanisms. Arch Virol 1997;142(7):1335-50. doi: 10.1007/s007050050164.

19. Wilhelmsen CL, Bolin SR, Ridpath JF, Cheville NF, Kluge JP. Lesions and localization of viral antigen in tissues of cattle with experimentally induced or naturally acquired mucosal disease, or with naturally acquired chronic bovine viral diarrhea. Am J Vet Res 1991;52(2):269-75.

20. Kimura M. A simple method for estimating evolutionary rate of base substitutions through comparative studies of nucleotide sequences. J Molec Evol 1980;16:111-120.

21. Tamura K. Estimation of the number of nucleotide substitutions when there are strong transition-transversion and G + C-content biases. Molec Biol Evol 1992;9:678-687.

Pecu 2023;14(1):248-259

22. Fray MD, Paton DJ, Alenius S. The effects of bovine viral diarrhoea virus on cattle reproduction in relation to disease control. Anim Reprod Sci 2000;60-61:615-27. doi: 10.1016/s0378-4320(00)00082-8.

23. Kameyama K, Konishi M, Tsutsui T, Yamamoto T. Survey for detecting persistently infected cattle with bovine viral diarrhea in Japan J Vet Med Sci 2016;78(8):1329–31.

24. Gunn HM. Role of fomites and flies in the transmission of bovine viral diarrhoea virus. Vet Rec 1993;132(23):584-5. doi: 10.1136/vr.132.23.584.

25. Brock KV, Redman DR, Vickers ML, Irvine NE. Quantitation of bovine viral diarrhea virus in embryo transfer flush fluids collected from a persistently infected heifer. J Vet Diagn Invest 1991;3(1):99-100. doi: 10.1177/104063879100300127

26. Lang-Ree JR, Vatn T, Kommisrud E, Loken T. Transmission of bovine viral diarrhoea virus by rectal examination. Vet Rec 1994;135(17):412-3. doi: 10.1136/vr.135.17.412.

27. Gómez-Romero N, Velázquez-Salinas L, Ridpath JF, Verdugo-Rodríguez A, BasurtoAlcántara FJ. Detection and genotyping of bovine viral diarrhea virus found contaminating commercial veterinary vaccines, cell lines, and fetal bovine serum lots originating in Mexico. Arch Virol 2021;166(7):1999-2003. doi: 10.1007/s00705-02105089-9.

28. Revell SC, Chasey D, Drew TD, Edwards S. Some observations on the semen of bulls persistentlyinfected with bovine virus diarrhoea virus. Vet Rec 1988;123(5):122-5. doi: 10.1136/vr.123.5.122.

29. Grooms DL, Ward LA, Brock KV. Morphologic changes and immunohistochemical detection of viral antigen in ovaries from cattle persistently infected with bovine viral diarrhea virus. Am J Vet Res 1996;57(6):830-3.

30. Meyling A, Jensen AM Transmission of bovine virus diarrhoea virus (BVDV) by artificial insemination (AI) with semen from a persistently-infected bull. Vet Microbiol 1988;17(2):97-105. doi: 10.1016/0378-1135(88)90001-6.

31. Roeder PL, Harkness JW. BVD virus infection: prospects for control. Vet Rec 1986;118(6):143-7. doi: 10.1136/vr.119.6.143.

32. Hamers C, Lecomte C, Kulcsar G , Lambot M, Pastoret PP. Persistently infected cattle stabilise bovine viral diarrhea virus leading to herd specific strains. Vet Microbiol 1998;61(3):177-82. doi: 10.1016/s0378-1135(98)00185-0.

33. Grooms DL, Brock KV, Pate JL, Day ML. Changes in ovarian follicles following acute infection with bovine viral diarrhea virus. Theriogenology. 1998;49(3):595-605. doi: 10.1016/s0093-691x(98)00010-7.

34. Ssentongo YK, Johnson RH, Smith JR. Association of bovine viral diarrhoea-mucosal disease virus with ovaritis in cattle. Aust Vet J 1980;56(6):272-3. doi: 10.1111/j.17510813.1980.tb05722.x.

35. Windsor P. Abnormalities of development and pregnancy. Noakes ED, et al, editors. England, Vet Rep Obst (Tenth ed), W.B. Saunders, 2019; ISBN 9780702072338, https://doi.org/10.1016/B978-0-7020-7233-8.00009-4.

36. Gómez-Romero N, Ridpath JF, Basurto-Alcántara FJ, Verdugo-Rodríguez A. Bovine viral diarrhea virus in cattle from Mexico: Current Status. Front Vet Sci 2021;8:673577. doi: 10.3389/fvets.2021.673577.

37. Fulton RW, Whitley EM, Johnson BJ, Ridpath JF, Kapil S, Burge LJ, Cook BJ, Confer AJ. Prevalence of bovine viral diarrhea virus (BVDV) in persistently infected cattle and BVDV subtypes in affected cattle in beef herds in south central United States. Can J Vet Res 2009;73(4):283-91.

38. Gómez-Romero N, Basurto-Alcántara FJ, Verdugo-Rodríguez A, Bauermann FV, Ridpath JF, Genetic diversity of bovine viral diarrhea virus in cattle from Mexico. J Vet Diagn Invest 2017;29(3):362-365. doi: 10.1177/1040638717690187

Rev. Mex. Cienc. Pecu. Vol. 14 Núm. 1, pp. 1-259, ENERO-MARZO-2023

CONTENIDO CONTENTS

ARTÍCULOS / ARTICLES

Pags.

Identification of candidate genes and SNPs related to cattle temperament using a GWAS analysis coupled with an interacting network analysis Identificación de genes candidatos y SNP relacionados con el temperamento del ganado utilizando un análisis GWAS junto con un análisis de redes interactuantes Francisco Alejandro Paredes-Sánchez, Ana María Sifuentes-Rincón, Edgar Eduardo Lara-Ramírez, Eduardo Casas, Felipe Alonso Rodríguez-Almeida, Elsa Verónica Herrera-Mayorga, Ronald D. Randel......……………………………………………………………………........………………... 1

Efecto de la consanguinidad y selección sobre los componentes de un índice productivo en ratones bajo apareamiento estrecho Effect of consanguinity and selection on the components of a productive index, in mice under close mating Dulce Janet Hernández López, Raúl Ulloa Arvizu, Carlos Gustavo Vázquez Peláez, Graciela Guadalupe Tapia Pérez………………………........ 23

Variabilidad genética en biomasa aérea y sus componentes en alfalfa bajo riego y sequía Genetic variability in aerial biomass and its components in alfalfa under irrigation and drought Milton Javier Luna-Guerrero, Cándido López-Castañeda 39

Estimación de masa de forraje en una pradera mixta por aprendizaje automatizado, datos del manejo de la pradera y meteorológicos satelitales Estimation of forage mass in a mixed pasture by machine learning, pasture management and satellite meteorological data Aurelio Guevara-Escobar, Mónica Cervantes-Jiménez, Vicente Lemus-Ramírez, Adolfo Kunio Yabuta-Osorio, José Guadalupe García-Muñiz 61

Thymol and carvacrol determination in a swine feed organic matrix using Headspace SPME-GC-MS Determinación de timol y carvacrol en una matriz orgánica de alimento para cerdo utilizando Headspace SPME-GC-MS Fernando Jonathan Lona-Ramírez, Nancy Lizeth Hernández-López, Guillermo González-Alatorre, Teresa del Carmen Flores-Flores, Rosalba Patiño-Herrera, José Francisco Louvier-Hernández 78

Cambios en el recuento de cuatro grupos bacterianos durante la maduración del Queso de Prensa (Costeño) de Cuajinicuilapa, México Changes in the count of four bacterial groups during the ripening of Prensa (Costeño) Cheese from Cuajinicuilapa, Mexico José Alberto Mendoza-Cuevas, Armando Santos-Moreno, Beatriz Teresa Rosas-Barbosa, Ma. Carmen Ybarra-Moncada, Emmanuel Flores-Girón, Diana Guerra-Ramírez….... 94

Detección molecular de un fragmento del virus de lengua azul en borregos de diferentes regiones de México Molecular detection of a fragment of bluetongue virus in sheep from different regions of Mexico Edith Rojas Anaya, Fernando Cerón-Téllez, Luis Adrián Yáñez-Garza, José Luis Gu�érrez-Hernández, Rosa Elena Sarmiento-Salas, Elizabeth Loza-Rubio 110

Insulin-like growth factor 1 (IGF-1) concentrations in synovial fluid of sound and osteoarthritic horses, and its correlation with proinflammatory cytokines IL-6 and TNFα Concentraciones del factor de crecimiento similar a la insulina 1 (IGF-1) en el líquido sinovial de caballos sanos y osteoartríticos, y su correlación con las citoquinas proinflamatorias IL-6 y TNFα Fernando García-Lacy F., Sara Teresa Méndez-Cruz, Horacio Reyes-Vivas, Victor Manuel Dávila- Borja, Jose Alejandro Barrera-Morales, Gabriel Gu�érrez-Ospina, Margarita Gómez-Chavarín, Francisco José Trigo-Tavera……..….... 122

Uso de células estromales mesenquimales derivadas de la gelatina de Wharton para el tratamiento de uveítis recurrente equina: estudio piloto Use of Wharton&#39;s jelly-derived mesenchymal stromal cells for the treatment of equine recurrent uveitis: a pilot study María Masri-Daba, Montserrat Erandi Camacho-Flores, Ninnet Gómez-Romero, Francisco Javier Basurto Alcántara………………… 137

Escala de la producción y eficiencia técnica de la ganadería bovina para carne en Puebla, México Scale of production and technical efficiency of beef cattle farming in Puebla, Mexico José Luis Jaramillo Villanueva, Lisse�e Abigail Rojas Juárez, Samuel Vargas López…………………………………….... 154

Regresión cuantil para predicción de caracteres complejos en bovinos Suizo Europeo usando marcadores SNP y pedigrí Quantile regression for prediction of complex traits in Braunvieh cattle using SNP markers and pedigree Jonathan Emanuel Valerio-Hernández, Paulino Pérez-Rodríguez, Agus�n Ruíz-Flores…………………… 172

Análisis de crecimiento estacional de una pradera de trébol blanco (Trifolium repens L) Seasonal growth analysis of a white clover meadow (Trifolium repens L.) Edgar Hernández Moreno, Joel Ventura Ríos, Claudia Yanet Wilson García, María de los Ángeles Maldonado Peralta, Juan de Dios Guerrero Rodríguez, Graciela Munguía Ameca, Adelaido Rafael Rojas García....................................................................................…………………………………………………………….……………...... 190

REVISIONES DE LITERATURA / REVIEWS

Aspects related to the importance of using predictive models in sheep production. Review Aspectos relacionados con la importancia del uso de modelos predictivos en la producción ovina. Revisión Antonio Leandro Chaves Gurgel, Gelson dos Santos Difante, Luís Carlos Vinhas Ítavo, João Virgínio Emerenciano Neto, Camila Celeste Brandão Ferreira Ítavo, Patrick Bezerra Fernandes, Carolina Marques Costa, Francisca Fernanda da Silva Roberto, Alfonso Juven�no Chay-Canul……....……....……....……....……....……....……....……....……....……....……....……....……....... 204

NOTAS DE INVESTIGACIÓN / TECHNICAL NOTES

Preferencia de ocho plantas por Odocoileus virginianus en cautiverio

Preference for eight plants among captive white-tailed deer Odocoileus virginianus in Veracruz, Mexico Hannia Yaret Cueyactle-Cano, Ricardo Serna-Lagunes, Norma Mora-Collado, Pedro Ze�na-Córdoba, Gerardo Benjamín Torres-Cantú..……..………………..………..………..………..……….........………...……....…. 228

Rendimiento y valor nutricional de brásicas forrajeras en comparación con forrajes tradicionales Yield and nutritional value of forage brassicas compared to traditional forages David Guadalupe Reta Sánchez, Juan Isidro Sánchez Duarte, Esmeralda Ochoa Mar�nez, Ana Isabel González Cifuentes, Arturo Reyes González, Karla Rodríguez Hernández...........................…………. 237

Genetic characterization of bovine viral diarrhea virus 1b isolated from mucosal disease Caracterización del virus de la diarrea viral bovino subtipo 1b aislado de un caso de la enfermedad de las mucosas Roberto Navarro-López, Juan Diego Perez-de la Rosa, Marisol Karina Rocha-Mar�nez, Marcela Villarreal-Silva, Mario Solís-Hernández, Eric Rojas-Torres, Ninnet Gómez-Romero........................…………. 248