| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
ANIMAL GENETICS |
* Facultad de Agronomía, Universidad de la República, 12900 Montevideo, Uruguay;
Department of Animal and Dairy Science, University of Georgia, Athens 30602-2771
 |
Abstract |
|---|
 Top Abstract INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: beef cattle • calving day • calving success • censored record • genetic parameter • threshold-linear model
|
INTRODUCTION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Urioste et al., 1998) and should be included in the breeding goal. Use of reproductive information as a selection tool presents difficulties. Reproductive performance is a complex trait that has many components (Rust and Groeneveld, 2001), and no completely satisfactory measure of reproduction has yet been found. Bourdon and Brinks (1983) proposed the use of calving date as a measure of fertility in beef cattle. Conceptually, it is similar to days to calving (Meyer et al., 1990; Johnston and Bunter, 1996) or calving day (CD; Ponzoni, 1992). Calving day is more appropriate for analyses of field data from pasture-based management systems because it is not necessary to record the first joining day and can be expressed on a within herd-year basis.
Information for open cows must be included in the evaluations of CD to make the best use of the data available for reproductive performance. Johnston and Bunter (1996) assigned a fixed value to noncalving cows. Donoghue et al. (2004a, b) assumed a truncated normal distribution for the uncensored records, and made a random draw from the distribution to obtain a record for censored females. Forni and Albuquerque (2005) did not consider noncalving cows in their study, because their inclusion did not improve the identification of genetic differences between animals. No attempts have been made to analyze missing CD data with other observed reproductive traits, such as the binary trait calving success (CS).
This study assessed CD as a suitable reproductive trait to be included in national beef cattle evaluations under extensive or semiextensive conditions like those prevailing in Uruguay. Different models were fitted, genetic parameters were estimated, and genetic relationships with CS were investigated.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Data
Data for this study were obtained from the Uruguayan Aberdeen Angus Recording System. General management practices in the Aberdeen Angus herds are as follows: a) the breeding season begins with the 2-yr-old heifers (AI or service bull, depending on the body development of the heifers); b) 2 wk later, the rest of the breeding herd follows, with part of the herd being selected for AI, normally using imported semen, and the remaining part being assigned to service bulls; c) service bulls are usually the same for heifers and for adult cows; d) clean-up bulls are often the same as the service bulls; and e) synchronized AI is still very rare. A standard breeding season lasts 45 d for AI and 90 d for natural mating.
Most recorded cows are of pedigreed origin. They have an extraeconomic value because of that and will tend to stay in the herd for a long time even if they do not calve for one or even more years. They are also controlled by the Uruguayan Rural Association, which is in charge of the pedigree recording. Breeding season and corresponding service sires must be reported, and gestation lengths and calving intervals are checked to determine if they are of reasonable lengths.
Joining or mating dates and type of mating [AI or natural service (NS)] are not systematically identified, and the only readily available information related to reproduction is that of calving date. Management of the AI is identified by the specific sire registration number when the semen is imported. However, a small fraction of national bulls used in AI cannot be distinguished from the natural service sires by this procedure. Therefore, they were grouped together in the NS management group.
The trait of primary interest, CD, was defined as the number of days from the beginning of a herd’s calving season to the cow’s calving date. The trait can be expressed at every season. It differs slightly from the days-to-calving definition of the Australian workers (Johnston and Bunter, 1996) because the joining date for each cow was unavailable.
The editing process was rigorous. The initial data set had approximately 33,000 calving records and 14,000 cows in 56 herds. Calving records from cows with missing birth dates or born in the fall, from cows with an age less than 600 d at calving, from cows used as embryo transfer donors or recipients, with missing sires, with calving intervals less than 280 d, or from herds with less than 30 calving records in 3 consecutive years were initially removed. In the few cases in which the cows presented records in more than 1 herd, only the information obtained in the cows’ birth herd was used. Failure to calve was identified by means of reconstructing the individual cow’s breeding history between the first and last recorded calving. It was assumed that all cows calved in the spring breeding season, and the absence of a record in a specific year was interpreted as the cow not calving in that particular year. It was assumed that nonlactating cows remained in the herd, as is usually the case under Uruguayan conditions. The data included cows with a clearly identified first calving at the age of 2 or 3 yr, and 2 subsequent calvings. Cows with a first calving record at 4 yr of age were assumed to have failed at age 3 and were added, provided they had an interval of at least 10 mo with respect to the last recorded calving date of the other cows in the same herd in the previous year.
The current data recording provides no information on nonlactating cow mating management; therefore, nonlactating cows were randomly assigned to identified AI or NS management groups, in the same proportion between AI and NS found for cows calving within the same contemporary group (defined as a combination of herd, year, and mating management). It was assumed that mating management was a purely environmental effect (i.e., no effect of service sire). In some cases, it was not possible to find a useful criterion to assign nonlactating cows to contemporary groups, such as during the fall (only 3% of the data); we therefore assumed that the cows should have calved in the spring. Data from year 2004 (the last year of recording) were eliminated because no inferences could be made about presence or absence of cows in the herd. Records from contemporary groups with less than 5 records, or that included less than 2 records from cows that actually calved, were also deleted.
After editing, 6,763 records from 3,442 cows from 455 sires, born between 1975 and 2000, in 19 herds were available for analysis. The pedigree file had 7,748 animals, with 72% of them having both sire and dam information, and a further 10% having an identified sire.
Statistical Models
Three approaches were defined to handle CD observations on animals that failed to calve, and specific data sets were created: 1) cows were assigned a penalized CD value, corresponding to the last observed CD plus a fixed penalty of 21 d within contemporary group (PEN); 2) estimation of censored records was simultaneously obtained from a truncated normal distribution, using data for cows in the same contemporary group (CEN); and 3) records were regarded as missing, and parameters were estimated in a threshold-linear analysis, including the observed binary trait CS in different parities (TLMISS). This approach was inspired by the methodology developed by Arnason (1999) for racing performance (a linear trait) and racing status (a categorical trait) in Swedish Standardbred trotters. Whereas CEN assumes censoring by the extreme of CD, TLMISS assumes censoring of CD by an independent trait, of which only a categorical expression is observed. The TLMISS becomes similar to CEN as genetic and residual correlations in TLMISS approach 1.0.
Trivariate statistical models were fitted for the PEN and CEN data sets, treating CD at the first, second, and third calving opportunity (CD1, CD2, and CD3, respectively) as separate traits. The general model, in matrix notation, can be written as:
 | [1] |
where y is a vector of observed and predicted CD; ß is a vector of systematic effects; a is a vector of animal additive genetic effects; e is the vector of residual effects; and X and Z are the corresponding incidence matrices. The ß vector included the contemporary group effect (herd x year x mating management group) in each of the first 3 calving opportunities (173, 135, and 100 levels, respectively) and the effects of age at calving (3 levels within calving opportunity; animals within ± 1 SD, or less than –1 or more than 1 SD), and physiological status at mating (nonlactating or lactating cows). Other fixed effects, such as sex of the gestated and reared calf were initially explored using a fixed effects model, but were discarded in the final model. Residuals were assumed to follow a normal distribution, e ~ N(0, R0 
I), where
 |
Animal additive genetic effects were assumed to follow a multivariate normal distribution, a ~ N(0, G0 A), where G0 is the (co)variance matrix between animal effects, and A is the known matrix of additive relationships between animals.
For the TLMISS data set, a 6-trait model was adopted, including the threshold traits calving success at the first 3 calving opportunities (CS1, CS2, and CS3, respectively) to the above-mentioned linear CD traits (CD1, CD2, and CD3, respectively). The CS traits were defined as binary; females that calved were coded as 1, and cows without a recorded calving in a specific year, but appearing in subsequent year(s), were assigned a 0 (failure) in the corresponding year(s) between 2 identified calvings.
The model is now
 | [2] |
with definitions as in equation [1], but y now includes the observed and missing CD records and unobserved liabilities of CS, ß is a vector with corresponding levels of age at calving group and physiological status at mating, c is a vector of contemporary groups considered to be an independently and normally distributed effect, c ~ N(0, P I), and R0 is a diagonal matrix of residual variances (a value of 1 was assumed for the binary traits).
Parameters were drawn from the posterior distributions using Gibbs sampling, as implemented in the programs TM, kindly provided by Andres Legarra, INRA-Castanet Tolosan, France, and Thrgibbs1f90 by Shogo Tsuruta, University of Georgia, Athens. Based on the preGibbs diagnosis used (program Postgibbsf90 by S. Tsuruta) and on visual inspection of trace plots, a chain of 100,000 iterations was run for model [1], with a burn-in of 20,000 rounds, keeping every 50th sample for inference of posterior features. For the threshold-linear analysis, a chain of 200,000 iterations was used, with a burn-in of 40,000 rounds and a thinning interval of 100 samples.
|
RESULTS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
The average calving success for calving opportunities 1 to 3 (CS1, CS2, and CS3) was 86.3, 65.9, and 67.3% for the first, second, and third calving opportunity, respectively. National averages in Uruguay vary between 50 and 75% (MGAP-DIEA, 2003). Other descriptive statistics for the data set are given in Table 1. On average, cows began calving at 3 yr (35.4 mo) and had a new opportunity to conceive every year (at 48.2 and 60.0 mo). Observed CD records in the first calving constituted 57.4% of the total; second and third calvings were 26.1 and 16.5% of the data, respectively. As shown in Table 1, large management differences were found: 37.5% of the heifers (first calving) were inseminated, whereas equivalent figures for second and third parity were 13.9 and 18.7%, respectively. In the second calving opportunity, 27.1% of the cows were nonlactating at the beginning of the breeding season, whereas as many as 41.6% of the cows were nonlactating in the third calving opportunity.
|
There was also a trend showing differences in CD for cows exposed to AI or natural service, with the latter having greater CD records. These differences may be due to different length of the breeding season for each breeding management type. In our data, average length of the breeding season was 55 d for AI management and 82 d for natural mating, but variation was very large between herds and years.
Censored Data Sets
Posterior means and SD of genetic parameters for CD traits, using PEN and CEN data sets are presented in Table 2. In the PEN data set, nonpregnant cows were assigned a fixed penalty value (+21 d beyond the last observed CD record within contemporary group), whereas in the CEN data set values for each cow were randomly drawn from a truncated normal distribution. Both approaches showed similar estimates of (co)variances. Lower genetic and residual variation was observed in CD2 compared with CD1 and CD3; however, the uncertainty in the estimates was important.
|
Genetic correlations were positive and medium to high in magnitude (0.57 to 0.79) for the PEN data set. In the CEN data set, however, there was a high genetic correlation between CD1 and CD2 (0.91), but only a medium to low correlation (0.38) between CD2 and CD3. Despite these differences between data sets, the general picture indicates important positive correlations between CD at different parities.
Threshold-Linear Analysis with Missing CD Data
Estimates of (co)variances and genetic parameters for the threshold linear analysis of CD and CS are presented in Table 3. In this case, CD records from noncalving cows were considered as missing. An initial attempt of using the same models applied to PEN and CEN data sets did not converge; therefore, we decided to treat contemporary groups as uncorrelated random effects. Residuals were assumed uncorrelated, based on the low residual correlations estimated in the PEN and CEN data sets.
|
In agreement with results presented in Table 2
for PEN and CEN approaches, positive and high (0.82 to 0.88) genetic correlations between CD measures were found. Genetic correlations between CS measures were also positive and medium to high (0.56 to 0.80). Negative (favorable) medium to high genetic correlations between CD and CS traits (–0.54 to –0.91) were estimated in this analysis, suggesting that genetic improvement in CD (earlier CD within the corresponding calving season) would lead to improved CS for cows.
|
DISCUSSION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
All models used in this study showed a similar picture: medium heritabilities for CD traits, with the greater values (0.23 to 0.31) at the first parity (Tables 2 and 3), indicating good scope for selection. However, the uncertainty in the estimates was fairly large. Heritability estimates for CD or days to calving from field data reported in the literature range from 0.04 to 0.28 (Meyer et al., 1990; Johnston and Bunter, 1996; Mercadante et al., 2003; Donoghue et al., 2004b; Forni and Albuquerque, 2005). In a selection experiment conducted in New Zealand, Morris et al. (2000) obtained an estimate of 0.09. In the review of Koots et al. (1994), an average heritability of 0.08 was reported. Greater estimates of heritability observed in our study, compared with average literature results, may reflect differences in populations, slight differences in trait definitions, different management practices eventually confounding genetic and environmental effects, or the influence of the data structure.
Genetic associations between CD in different parities are scarcely presented in the literature. Johnston and Bunter (1996) found a genetic correlation between days to calving at parities 1 and 2 of 0.85. Mercadante et al. (2002), using random regression models on data from a selection experiment, reported very high genetic correlations (0.98 to 0.99) between days to calving for Nelore cows at ages 3, 4, and 5 yr. These results are in agreement with our findings. Similar magnitude of genetic variance across parities suggests that a simple, repeated measures model could be an alternative for implementation of a practical genetic evaluation for CD.
Heritability estimates for CS at different parities (Table 3) ranged from 0.37 to 0.42. Estimates for CS or related traits, obtained with threshold models, vary between 0.03 (Donoghue et al., 2004b) and 0.25 to 0.27 (Rust and Groeneveld, 2002; Silva et al., 2002). Urioste et al. (2006), partially with the same data as used in this study, obtained moderate estimates (0.27 to 0.44), although the credibility intervals indicated imprecise inferences. In general, use of threshold models produced greater estimated heritabilities than those obtained by linear models (Johnston and Bunter, 1996; Morris et al., 2000; Phocas and Sapa, 2004).
The negative (genetically favorable) association between CD and CS traits (Table 3) is a relevant finding. Johnston and Bunter (1996) reported a very high negative genetic correlation (–0.97), suggesting that they were genetically the same trait. Donoghue et al. (2004c), working with field data from first-calf Angus females, estimated a genetic correlation between the traits of –0.73. Our results vary between –0.54 and –0.91, pointing in the same direction. In particular, a very high (–0.91) negative genetic correlation between CD1 and CS2 suggests that cows that calve late in the first parity are greatly compromised in the subsequent calving.
Estimates of genetic parameters provide a first comparison of the models used. The PEN model, using penalized CD records for censored, noncalving cows, adding 21 d to the last calving date within contemporary group, is an ad-hoc procedure that has been successfully implemented in Australia (Graser et al., 2005), where the recording system is more detailed than the one existing in Uruguay. In the CEN approach (Donoghue et al., 2004a), each censored cow receives a different penalty, drawn from a truncated normal distribution for the uncensored records. This approach is conceptually equivalent to a general censoring model provided by Hughes (1999). However, observed CD distributions tend not to be normal, usually presenting a long right tail. Both censored approaches make a strong assumption: that cows would calve if they were given enough time. In our study, estimates are not very different between the 2 approaches, but genetic correlations are not totally consistent. This could be a consequence of these models not completely disentangling environmental and genetic effects from the data analyzed. For these reasons, the censoring models may not be the best approach for fertility traits.
On the other hand, the TLMISS approach only assumes that the CD information is missing, and censoring is by the observed and economically important trait CS. In the animal breeding context, Arnason (1999) developed a methodology for genetic evaluation of Swedish standard-bred trotters for racing performance (a linear trait) and racing status (a threshold trait with 2 categories, where nonstarting horses do not produce a record). He also conducted a simulation showing that including an all-or-none variable as a correlated trait resulted in greater correlations between true and estimated breeding values, compared with a single trait AM-BLUP model based on observed performance records alone. The methodology was further used by Thuneberg-Selonen et al. (2001) for racing performance and racing status in Finnish horses. For computational convenience, they treated the categorical trait as linear.
The TLMISS model provides a clearer and more informative picture of relationships between traits, with a more consistent estimation of genetic (co)variation between CD and CS traits. This suggests that the assumption of censoring by an independent trait is superior to the assumption that all cows would calve given enough time. Its main disadvantage may be the implementation difficulties for large data sets. However, the negative genetic relationship between CD and CS could be used in the implementation of genetic improvement for fertility traits, where CD could act as an indicator trait of CS. This latter trait has a binary phenotypic expression, whereas CD traits are linear and easily incorporated into standard procedures already developed in the Uruguayan genetic evaluation system for the Aberdeen Angus breed. It is also likely that CD would have an important degree of understanding and acceptance among breeders.
Attempts to estimate the residual correlations in TLMISS were unsuccessful. Thus, the animal environmental covariances were unaccounted for, and, subsequently, the results could have been affected. Esa Mäntysaari (Agricultural Research Centre, Jokionen, Finland, personal communication) suggested that such covariances might be estimated by introducing correlated permanent environmental effects for both traits. Because the permanent environmental effects and residual variances are confounded, he suggested fixing the residual variances at a small value, without reestimation.
In conclusion, CD is readily observed in calving cows, routinely reported when calves are registered, and shows sizeable genetic variation, independent of the model used. It is a good indicator trait for fertility, and it could be an available option for genetic evaluation of reproduction in beef cattle. Simultaneous use of CS may be the best way to account for the discrete nature of the data. Improvements in the quality of data for the Aberdeen Angus recording scheme, such as developing a full female inventory recording system, identifying joining dates, pregnancy status of heifers and cows, entry and exit dates of natural mating sires, type of service management used (AI or NS), and cow disposal codes would help to effect direct selection on the economically important female reproduction traits.
|
Footnotes |
|---|
2 Corresponding author: jurioste{at}fagro.edu.uy
Received for publication August 10, 2006. Accepted for publication April 27, 2007.
|
LITERATURE CITED |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
This article has been cited by other articles:
|
|
 |
|
  K. M. Cammack, M. G. Thomas, and R. M. Enns Review: Reproductive Traits and Their Heritabilities in Beef Cattle Professional Animal Scientist, October 1, 2009; 25(5): 517 - 528. [Abstract] [PDF]
|
|
|
|
 |
|
 C. Sun, P. Madsen, U. S. Nielsen, Y. Zhang, M. S. Lund, and G. Su Comparison between a sire model and an animal model for genetic evaluation of fertility traits in Danish Holstein population J Dairy Sci, August 1, 2009; 92(8): 4063 - 4071. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Hou, P. Madsen, R. Labouriau, Y. Zhang, M. S. Lund, and G. Su Genetic analysis of days from calving to first insemination and days open in Danish Holsteins using different models and censoring scenarios J Dairy Sci, March 1, 2009; 92(3): 1229 - 1239. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. I. Urioste, I. Misztal, and J. K. Bertrand Fertility traits in spring-calving Aberdeen Angus cattle. 2. Model comparison J Anim Sci, November 1, 2007; 85(11): 2861 - 2865. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
