ORIGINAL

Estimation of genetic and phenotypic parameters for production traits in Holstein and Jersey from Colombia

 

Estimación de parámetros genéticos y fenotípicos para características de producción en ganado Holstein y Jersey Colombiano

 

Juan Rincón F,1,2* Ph.D, Juan Zambrano A,1 Ph.D, Julián Echeverri,1 Ph.D.

1Universidad Nacional de Colombia. Facultad de Ciencias Agropecuarias, Departamento de producción Animal, Grupo BIOGEM: Biodiversidad y genética molecular. Calle 59 A N° 63-20 B50, Medellín, Colombia.
2Universidad Tecnológica de Pereira, Facultad de Ciencias de la Salud, profesor del programa de Medicina Veterinaria y zootecnia, Carrera 27 N° 10-02.

*Correspondence: jcrincon@unal.edu.co

Received: September 2014; Acepted: March 2015.


ABSTRACT

Objective. Determine the genetic and phenotypic parameters for milk yield, fat percentage, protein percentage and somatic cell score. Materials and methods. 18134 lactation records were used to Holstein and 1377 lactations for Jersey in different herds. The (co) variance components and genetic parameters were estimated using the software Multiple Trait Derivative-Free Restricted Maximum Likelihood MTDFREML. Results. The Holstein and Jersey heritability’s (and standard error) for milk yield were: 0.16 (0.082) and 0.15 (0.306), 0.30 (0.079) and 0.37 (0.319) for protein percentage, 0.32 (0.076) and 0.46 (0.313) for fat percentage and for somatic cell score were: 0.01 (0.054) and 0.01 (0.233), respectively. The largest genetic correlations were found between the percentage of fat and percentage of protein, with values of 0.82 (0.126) and 0.98 (0.852) for Holstein and Jersey respectively. The lowest correlations were between fat percentage and somatic cell score with -0.01 (1.147) and -0.01 (1. 734). Phenotypic correlations were generally found low and repeatability showed a significant effect of permanent environment on milk production per lactation. Conclusions. It is important to emphasize the development of research to help guide breeding programs in the tropics, using selection indices of multi-traits.

Key words: Dairy cattle, genetic correlations, heritability, phenotype (Source: AIMS).


RESUMEN

Objetivos. Determinar los parámetros genéticos y fenotípicos para producción de leche, porcentaje de grasa, porcentaje de proteína y puntaje de células somáticas. Materiales y métodos. Se utilizó información de 18134 lactancias para Holstein y 1377 para Jersey de diferentes hatos del departamento de Antioquia (Colombia). La determinación de los componentes de varianza, covarianza y los parámetros genéticos se realizó mediante el método de máxima verosimilitud restricta libre de derivadas usando el programa MTDFREML. Resultados. La heredabilidad y el error estándar en Holstein y Jersey para producción de leche fueron 0.16 (0.082) y 0.15 (0.306), para porcentaje de proteína 0.30 (0.079) y 0.37 (0.319), para el porcentaje de grasa de 0.32 (0.076) y 0.46 (0.313) y para el puntaje de células somáticas fue de 0.01 (0.054) y 0.01 (0.233), respectivamente. Las mayores correlaciones genéticas encontradas fueron entre porcentaje de grasa y porcentaje de proteína, con valores de 0.82 (0.126) y 0.98 (0.852) para Holstein y Jersey, respectivamente. La menores correlaciones fueron obtenidas entre porcentaje de grasa y puntaje de células somáticas con valores de -0.01 (1.147) y -0.01 (1.734), respectivamente. Las correlaciones fenotípicas encontradas por lo general fueron bajas y la repetibilidad evidenció un efecto importante del ambiente permanente sobre la producción de leche por lactancia. Conclusiones. El presente trabajo encuentra algunas diferencias con los reportes de parámetros genéticos en otros países, lo que resalta la importancia del desarrollo de trabajos de investigación que permitan orientar los programas de mejoramiento genético en el trópico.

Palabras clave: Correlaciones genéticas, Fenotipo, Ganado lechero, Heredabilidad (Fuente: AIMS).


INTRODUCTION

The increase in the human population has increased the need for more food derived from animal products, which creates a need to seek more efficient and productive animals from different perspectives. Some important variables in dairy milk yield can be improved by using adequate selection programs according to the specific conditions where they are performed, and precise knowledge of genetic and phenotypic parameters is critical in planning and developing adequate selection strategies (1).

Estimating genetic and phenotypic parameters is necessary to determine the degree of variation due to genetics and environment, and also to predict genetic associations between two or more variables. These parameters are commonly required to construct selection indexes and predict correlated responses and to perform more efficient genetic evaluations (2).

Animal production in Holstein and Jersey cattle requires one of the highest technical levels and they are the specialized breeds most used for milk production in the high Colombian tropics. However, estimating genetic and phenotypic parameters, and consequently genetic evaluations, have presented serious difficulties, principally due to the limited amount of periodic and trustworthy production records for characters of economic importance (3). As a consequence, selection decisions according to the needs and conditions specific to the tropics have faced great uncertainty, especially in young cows and bulls (1).

While selection of dairy cattle in Colombia has been oriented towards milk production, the variables associated with its compositional quality play a very important role in genetic selection programs due to traits that have important economic impacts for the cattle industry (4). Additionally, the somatic cell score (SCS) is considered to be a variable of milk quality, and is also associated with susceptibility that can be present in cows with subclinical mastitis (5), a disease that causes huge economic loss around the world (6).

Having genetic and phenotypic parameters available that are based on precise information that is more representative of conditions specific to the high Colombian tropics, which involves not just milk production but is also related to the compositional quality of milk and udder health, is greatly important for genetic improvement programs for Holstein and Jersey cattle. The objective of this study was to determine the genetic and phenotypic parameters for milk production, fat percentage, protein percentage, and somatic cell score (SCS) in Holstein and Jersey cattle in Antioquia, Colombia.

MATERIALS AND METHODS

Location. This investigation was based on information obtained from 156 Holstein farms and 36 Jersey farms from 18 municipalities in the department of Antioquia, at 2000 and 2600 m.a.s.l. and average temperatures between 13 and 17°C. The conditions of handling, feeding and health were variable in different production systems, as well as topography and geographic location; however, all the animals were pastured and fed with balanced commercial feed.

Using information supplied by these farms in the milk control program at the Medellin headquarters of the Universidad Nacional de Colombia, productive data for fat percentage (PG), protein percentage (PP), milk yield in lactation (PL) and somatic cell count (RCS) were used to estimate genetic and phenotypic parameters, after confirming their veracity. All the information from the herds was organized and saved in Control 1 software version 1.0 (7).

The information collected was reviewed in order to determine reliability, and records that were considered suspect were removed from the definitive analysis. Additionally, all totally atypical and physiologically abnormal data was removed, which was determined while processing and storing the information.

The RCS values were transformed into somatic cell score (SCS) by means of the mathematical equation: SCS=[LOG2 (RCS/100000) + 3] in order to improve normalcy in the data (5). The information was edited in order to obtain a record base that would be the most reliable and appropriate to the local situations.

Sample size. Once the information was edited, 18134 animal records remained; 7723 animals with PL records, 5709 with PG records, 5866 with PP and 5769 with RCS. In the case of Jersey cattle, 1.377 animal records were used: 820 with PL records, 455 with PG, 475 with PP and 471 with RCS. The number of animals in the kinship matrix for the Holstein breed was 9099 and for the Jersey breed was 801.

Statistical analysis. A descriptive analysis was performed in which the measurements, standard deviation, and variation coefficients were estimated for each variable in the two breeds. Also, normality suppositions were validated, as well as independence and homoscedasticity, using different procedures of the SAS/STAT program (8).

Components of variance, heritability and repeatability. Preliminary analysis was used to test models that included different fixed effects and covariables in order to determine the statistical models that best fit the estimates of genetic parameters of the different productive traits evaluated.

The components of variance and genetic parameters were estimated based on an independent univariate animal model for each trait. This was done by means of the restricted maximum likelihood method (REML) using the SAS statistical package (8) and MTDFREML (9) that determine solutions based on the equations of mixed models (MME) described by Henderson (10).

The univariate animal model used included the following fixed effects: the herd, number of births and contemporary group (with municipality, year of birth and season of birth). As a covariable, it included the duration of lactation (only for PL) and milk yield in the case of PP and PG variables. Finally, as random effects, the permanent environment and direct additive genetic effect of the animal was included.

The model used was the following:
Y=Xβ + Zα + Wπ + e

Where:

Y = vector of observations for PL, PP, PG and SCS; β = unknown vector of fixed effects (the herd, number of births and contemporary group and as covariables the duration of lactation for PL and milk yield for PG and PP); α = unknown vector of random animal effects; πi = unknown vector of permanent random environmental effects; e = unknown vector of random residual effects; X = Incidence matrix of fixed effects, Z= Incidence matrix of random animal effects and W = incidence matrix of random permanent environmental effects.

The above model assumes that the random effects of the animal, permanent environment and residual effects are independently distributed with zero mean and variances , y , respectively.

Using the above model, a narrow sense of heritability was determined, which was estimated as: and repeatability was estimated using the equation: (11). The heritability and repeatability parameters with their respective standard errors were obtained using MTDFREML software (9).

It is important to keep in mind that the univariate model makes it possible to break down the variability of the character without taking into account the effect of other secondary numeric variables. For this reason, a unicaracteristic model was made to compare and define just one estimate, since the bicharacteristic model generates different estimates according to the paired set of variables that are evaluated in the model. It is also important to mention that the estimates for the bicharacteristic model depend on evaluating individuals that present information for both traits, so that the size can change with each set of variables evaluated in the model.

Components of (co)variance and genetic and phenotypic correlations. The (co)variance components were estimated using a derivative free restricted maximum likelihood method (10) using MTDFREML software (9) and a bicharacteristic model that includes fixed effects for both variables: herd, birth number and contemporary group (with municipality, year of birth and season of birth). Random effects that were included were the permanent environmental effect and the additive effect of the animal. The matrix model used was:

Where: Yi = vector of n observations for each i trait (PL, PP, PG y SCS); βi = solution vector for fixed effects (herd, birth number and contemporary group, and as covariables lactation duration for PL and milk yield for PG and PP); αi = solution vector for random animal effects; πi = solution vector for random permanent environmental effects; e = vector for residual wastes; Xi = Matrix of incidence of fixed effects, Zi = matrix of incidence for random animal effects and Wi = Matrix of incidence of random permanent environmental effects.

The above model assumes that the random effects (random animal effects, random permanent environmental effects and residual random effect) are independently distributed with zero mean and variances,,and respectively.

The genetic (rg) and phenotypic (rp) correlations among the different traits evaluated were determined using the following equations:

Where:= genetic covariance for traits i and j; =genetic variance for trait i; =genetic variance for trait j; = phenotypic covariance for traits i and j; = phenotypic variance for traits i and = phenotypic variance for trait j.

RESULTS

Descriptive analysis. With the descriptive evaluation it was possible to determine that the mean for PL was 5524±2156 and 4234±1869 liters/lactation for Holstein and Jersey breeds, respectively, with a coefficient of variation of 39.0% for Holstein and 44.2% for Jersey, which was the trait that most varied between the two breeds (Table 1). On the other hand, the trait that varied the least was PP, with variation coefficients of 9.2% and 8.8% for Holstein and Jersey, respectively. All the traits presented variation coefficients according to what was expected for the production systems in which the investigation was done. In general, the estimates obtained make it possible to observe a greater PL in the Holstein breed. However, milk produced in the Jersey breed presents greater percentages of fat and protein.

Heritability and repeatability. The trait that presented the greatest heritability was PG, with values of 0.32 and 0.46 for Holsteins and Jerseys, respectively; the lesser heritability was SCS with values of 0.01 I both breeds. The results show that heritability for SCS were lower (h2<5%), however the repeatability is much higher with values of 0.26 and 0.41 for Holstein and Jersey, respectively, which shows that the permanent environment had a very important effect on this trait.

The heritability for PG and PP were average (0.32 and 0.30) for Holstein, however the Jersey breed presented higher heritability (0.46 and 0.47, respectively) for the traits mentioned (Table 2). Regarding milk yield per lactation, the heritability was 0.16 and 0.15 for Holsteins and Jerseys. However, repeatability was much higher, (0.30 and 0.32, respectively), which showed an important effect in the permanent environment on this trait in the two breeds.

Genetic and phenotypic correlations. The highest genetic correlation was found between PG and PP, with values of 0.82 and 0.98 for Holstein (Table 3) and Jersey (Table 4), respectively, which indicates a very strong positive genetic association between the two milk components. On the other hand, the phenotypic correlation between the two traits was lower than the genetic due to environmental factors that affect them.

The lowest genetic correlation obtained was for PG and SCS (-0.01 for the two breeds). However, the phenotypic correlation was a little higher (0.12 and 0.15 for Holstein and Jersey).

The genetic correlation between PL and PP and between PL and PG in Holstein cattle was negative, with values at -0.40 and -0.28, respectively (Table 3). In the two breeds, the phenotypic correlations between PL and PG were negative, however the phenotypic correlation between PL and PP were positive but very low in Jersey cattle, although it should be kept in mind that they presented a high standard error (Table 4). In general, all the phenotypic correlations were lower than the genetic ones.

DISCUSSION

The standard error of heritability in the different variables evaluated in Jersey cattle was very high, which suggests implications concerning the reliability of the estimates. In Holstein cattle, a lower standard error was obtained in the estimates mainly due to a greater sampling size and reliability of the records.

The milk quality traits (PP and PG) presented mean heritability values (greater than 20%) and a low permanent environmental effect in the two breeds. However, the Jersey breed presented greater heritability values than Holstein, both for PG and for PP, although with a greater standard error. Additionally, heritability for PG was greater than for PP, which is in agreement with some authors (12), although not with others (13). However, both traits were highly correlated.

Milk yield per lactation (PL) presented heritability lower than 20% in the two evaluated breeds, showing an important influence from the permanent environment on this trait, which made possible a greater repeatability. Additionally, it is important to make clear that PL was the trait that had the greatest number of repeated records per animal, with an average of 1.9, even with cows that had 7 lactations, and more than 68% of the animals had production records for more than one lactation.

For SCS, the low heritability for this trait was confirmed (0.01) both in the Holstein and Jersey breeds, suggesting the high influence of environmental factors on this variable 814). However, repeatability was greater in the two breeds, showing the important effect that the permanent environment has. These results agree with those found in other studies (14,15).

The mean values obtained for milk quality traits (PP and PG) are in agreement with the values found in literature that catalogues them as medium to high heritability (16,13). However, they are less than some reported in Colombia (2) and greater than other reports in other parts of the world (12,17). The estimated repeatability for these characteristics is also found in concordance with different reports presented for dairy cattle, showing that the permanent environment has a lesser importance on solids present in the milk (18).

Milk yield per lactation showed heritability of 0.16 and 0.15 for Holstein and Jersey respectively, but with a greater standard error than for Jersey cattle, values which were lower than cases found in literature (12,19,20) that show values above 0.2 and that indicate that milk yield is a trait with medium heritability. However, some authors have found similar values to those obtained in this study, confirming a lower heritability in PL when compared with PG and PP and cataloguing the trait as medium or low heritability (21). On the other hand, the repeatability for this trait is double in magnitude when compared with heritability, which shows the huge effect that the permanent environment has. Similar results have been reported in other studies (22).

The somatic cell score (SCS) presented low heritability (0.01) and the highest standard error in both breeds, which is according to what is found in literature (15,21). These results suggest that the traits association with udder health are not highly inheritable, which constitutes a problem for genetic improvement programs. However, the majority of reports suggest a slightly higher heritability than that found in this study (14,16). Most research has encountered difficulties in improving this trait and in many cases suggests the need to reinforce genetic improvement programs with other methods in order to obtain a more accelerated genetic progress, such as assisted selection using molecular markers and genomic selection.

On the other hand, higher values for repeatability were found, suggesting the important effect the permanent environment has, which agrees with some reports (14,15). However, genetic progress in these traits is greatly affected by the environment, so good herd management can greatly improve this situation.

The genetic correlation, understood as the probability that two different characteristics are influenced by the same genes, is seen as the association of genetic values of one trait on those of another (11). Keeping this in mind, this study determined the existing correlations between productive traits (PL, PP, PG and SCS) in Holstein and Jersey breeds, finding a high correlation between traits associated with milk quality (PP and PG) but a medium to low correlation in the rest.

Genetic correlations between PG and PP were 0.82 and 0.98 for Holstein and Jersey, respectively, which agrees with the results obtained by other authors that have reported high genetic correlations between these two traits (greater than 70%) in different milk breeds (17,22), although in Colombia lower correlations have been estimated (23). The magnitude and direction found in this genetic correlation means that when one of this traits is selected, the other is also selected, resulting in a joint genetic progress. However, it is important to remember that the size of the sample used and the standard error obtained in the Jersey breed is very high, and should therefore be used with caution due to reliability problems.

The genetic correlation between PP and PL was negative in both breeds; however, the importance of the association was much greater in Holstein than Jersey cattle. The above results agree with what has been found in literature, where negative associations for these two traits are well known (17,23). However, the Jersey breed has lower values than those reported in the majority of work done on dairy cattle, although it should be kept in mind that the standard error in all estimates is very high, which implies that the results obtained in this breed should be taken with precaution due to reliability problems attributed mainly to the small sample size and the reliability of the data.

Genetic association between PL and PG was also negative for both breeds, agreeing with that found in literature (24) and showing difficulty to improve the quality of milk and quantity at the same time, although estimations for the Jersey breed show the same problem associated with standard error that was mentioned previously.

Genetic correlation between PP and SCS was 0.28 and 0.19 for Holstein and Jersey, respectively, which, although still low, agree with what was reported by VanRaden et al (19). However, it is important to keep in mind that the standard error for the correlation between the two breeds is high, and that in general the estimates for SCS were more reliable when compared with other traits.

Other genetic correlations were lower (less than 0.1) and with a higher standard error, which coincides with the majority of the results presented in literature (24,25). However, it is important to mention that genetic correlations are estimated parameters for a specific population under particular environmental and genetic conditions, and can therefore present variations between populations (18).

Milk yield (PL) was weakly associated with SCS, contradicting the hypotheses that indicate improving milk yield increases the somatic cell score and susceptibility to mastitis. Additionally, some studies have found correlations similar to those reported in this work, reinforcing the idea that highly productive animals that are resistant to mastitis can be achieved through genetic improvement (19,25).

Finally, the estimated phenotypic correlations were generally weak, which coincides with some reports in literature (22,25). However, the values of these correlations are very variable, according to the environment where they were estimated, so that it is common to find higher values in highly controlled production conditions (17). Greater values were obtained for association among the fat percentage (PG) and protein percentage (PP) (0.48 and 0.25) for Holstein and Jersey, respectively, which reflects the environmental important on the phenotypic expression of a trait, due to the genetic correlation being high, is diluted by the environment. The results obtained in this investigation agree with those found by other authors for dairy cattle (19,25). It is important to highlight that the phenotypic correlations are generally low, due to the influence of multiple environmental factors and in the majority of the cases the genetic correlations are superior to phenotypic correlations. To conclude, we highlight the importance of developing research studies that can orient genetic improvement programs under tropical conditions in order to obtain selection indexes that include more than one trait.

REFERENCIAS

1. Echeverri J. Evaluación genética de toros para características de importancia económica en la industria lechera. [Tesis Doctoral]. Medellín: Facultad de Ciencias Agrarias, Universidad de Antioquia; 2011.

2. Ruales FR, Manrique C, Cerón ME. Fundamentos en mejoramiento animal. Medellín: L Vieco e Hijas Ltda; 2007.

3. Echeverri JJ, Quijano JH. Estimación de la heredebilidad y repetibilidad para el porcentaje proteína y grasa de la leche en una población de ganado Holstein del departamento de Antioquia, Colombia. Rev Fac Nac Agron 2009; 62(Supl 1):77 -78.

4. Miglior F, Muir BL, Van Doormaal BJ. Selection indices in Holstein cattle of various countries. J Dairy Sci 2005; 88(3):1255–1263.

5. Bradley A, Green M. Use and interpretation of somatic cell count data in dairy cows. In practice 2005; 27(6):310-315.

6. Seegers H, Fourichon C, Beaudeau F. Production effects related to mastitis and mastitis economics in dairy cattle herds. Vet Res 2003; 34(4):475–491

7. Echeverri J, Lopez A, Parra J. Control 1: Manejo y control de producción para hatos lecheros. [Programa de ordenador]. versión 1. Medellín: Editorial Universidad Nacional de Colombia; 2010.

8. SAS Institute Inc. SAS/STAT User’s Guide: Version 9. 5th edition. Cary, North Carolina, USA: SAS Institute Inc; 2003.

9. Boldman KG, Kriese LA, Van Vleck LD, Van Tassell CP, Kachman SD. A manual for use of MTDFREML: A set of programs to obtain estimates of variances and covariances [Draft]. Washington (DC): ARS, USDA. 1995.

10. Martínez C, Manrique C, Elzo M. La evaluación genética de vacunos: una percepción histórica. Rev Colomb Cienc Pecu 2012; 25:293-311.

11. Arboleda E, Elzo M, Hurtado-Lugo N, Vergara O, Cerón-Muñoz M. Modelación aplicada a las ciencias animales: II Evaluaciones genéticas. Medellín: Biogénesis Fondo Editorial; 2012.

12. Abe H, Masuda Y, Suzuki M. Relationships between reproductive traits of heifers and cows and yield traits for Holsteins in Japan. J Dairy Sci 2009; 92:4055–4062.

13. Makgahlela ML, Banga CB, Norris D, Dzama K, Ng’ambi JW. Genetic correlations between female fertility and production traits in South African Holstein cattle. S Afr Anim Sci 2007; 37(3):180-188.

14. Koivula M, Negussie E, Mäntysaari EA. Genetic parameters for test-day somatic cell count at different lactation stages of Finnish dairy cattle. Livest Sci 2004; 90: 145–157.

15. Mostert BE, Banga C, Groeneveld E, Kanfer FHJ. Breeding value estimation for somatic cell score in South African dairy cattle. S Afr Anim Sci 2004; 34(2):32-34.

16. König S, Chongkasikit N, Langholz HJ. Estimation of variance components for production and fertility traits in Northern Thai dairy cattle to define optimal breeding strategies. Arch Tierz 2005; 48(3):233-246.

17. Toghiani S. Genetic relationships between production traits and reproductive performance in Holstein dairy cows. Arch Tierz 2012; 55(5):458-468.

18. Pereira JC. Melhoramento genético aplicado à produção Animal. 5a ed. Belo Horizonte: FEPMVZ Editora; 2008.

19. VanRaden PM, Sanders AH, Tooker ME, Miller RH, Norman HD, Kuhn MT, et al. Development of a national genetic evaluation for cow fertility. J Dairy Sci 2004; 87:2285–2292.

20. Galeano AP, Manrique C. Estimación de parámetros genéticos para características productivas y reproductivas en los sistemas doble propósito del trópico bajo colombiano. Rev Med Vet Zoot 2010; 57:119-131.

21. Dal Zotto R, De Marchi M, Dalvit C, Cassandro M, Gallo L, Carnier P, et al. Heritabilities and genetic correlations of body condition score and calving interval with yield, somatic cell score, and linear type traits in brown Swiss cattle. J Dairy Sci 2007; 90:5737–5743.

22. Kadarmideen HN, Thompson R, Coffey MP, Kossaibati MA. Genetic parameters and evaluations from single- and multiple-trait analysis of dairy cow fertility and milk production. Livest Sci 2003; 81:183–195.

23. Echeverri JJ, Quijano JH. Estimación de las correlaciones genéticas y ambientales entre el porcentaje de proteína y grasa de la leche con algunas características de importancia lechera. Rev Fac Nac Agron 2009; 62(Supl 1):78–79.

24. Carlén E, Strandberg E, Roth A. Genetic Parameters for Clinical Mastitis, Somatic Cell Score, and Production in the First Three Lactations of Swedish Holstein Cows. J Dairy Sci 2009; 87:3062–3070.

25. Pantelic V, Petrovic MM, Stevica Aleksic, Ostojic D, Sretenovic L, Nuvakovic Z. Genetic correlations of productive and reproductive traits of Simmental cows in Republic of Serbia. Arch Zootech 2008; 11(4):73-78.