Genetic evaluation of the number of services per conception with and without censored data in Iranian dairy cows

Document Type : Research Paper


1 Former MSc Student, Department of Animal Science, Faculty of Agriculture, University of Kurdistan, Sanandaj, Iran

2 Associate Professor, Department of Animal Science, Faculty of Agriculture, University of Kurdistan, Sanandaj, Iran

3 Associate Professor, Department of Animal Science, Faculty of Agriculture, University of Jiroft, Jiroft, Iran


Introduction: The profitability of dairy cattle herds mainly depends on the reproductive performance of cows. Because reproductive performance is effective on the amount of daily milk and calf production per cow, increasing the number of inseminations per conception, lengthening the calving interval in the herd, and causing the voluntary (due to the age of the animal) and involuntary (due to illness or fertility disorders) elimination. For evaluating fertility traits, we sometimes come across domains whose records were incomplete at the time, out of range, or delayed. These records are called censored records. The typical way to deal with such records is to delete them from the data set. However, this reduces the information available for genetic evaluation and may affect trait variance by masking actual genetic differences between animals. Another way is to use appropriate statistical methods to place censored records in genetic evaluations. The use of censored records in genetic evaluations yields more reliable genetic parameters, and it can increase the estimation accuracy of breeding values giving the support that the breeding value estimation is close to the true breeding value. The present study aimed to investigate the role of censored records in the estimation of genetic parameters of the number of inseminations per conception in Iranian dairy cows.
Materials and methods: The dataset included 29621 records collected from 2005 to 2019 by the Vahdat Cooperative of Isfahan in Iran. FoxPro software (version 9.0) was used to edit and prepare data, remove incorrect records and create contemporary groups. Five animal models, including Threshold Model (TM), Penalty Model (PM), Modified Penalty Model (MPM), Threshold-Threshold Model (TTM), and Modified Threshold-Threshold Model (MTTM), were used for the genetic evaluation of this trait. The predictive ability of the models was assessed using cross-validation. Also, a simulation study was performed to determine the appropriate model.
Results and discussion: The estimated heritability using different models was 0.3 (0.013), 0.07 (0.013), 0.07 (0.013), 0.3 (0.13), and 0.4 (0.012) for TM, PM, TTM, MPM, and MTTM, respectively. The highest correlation of breeding values for all animals was between MPM and TM (0.98). The same result was obtained for the best sire (sire with at least 20 daughters). The high correlation shows that these models work similarly in selecting the best sires. The low correlation between models indicates that the models rank sires differently. Therefore, the choice of model can affect the ranking of males. The top 10 sires showed less correlation between models than the active sires. When the number of males decreases, the selection becomes more strict. The prediction accuracy in models with censored records is higher than in models without censored records because the use of models that include censored records is due to the increased information about the trait, and the absence of bias due to deleting records improves the accuracy of estimating breeding values. Consequently, if the censoring status is not taken into account, animals with superior breeding values may be biased. The low ratio of common superior animals in different percentages between models showed that these two models rank the animals differently. Still, the high proportion of common superior animals indicates the similarity of the two models in choosing the best animals. So, model selection can be effective in ranking superior animals because when the censored record is used, the information is increased and the situation would be closer to reality. The cross-validation showed that the accuracy of the breeding value prediction by different models was higher in models with censored records than in models without censored records. The results of the simulation study showed that the models with censored records provided better genetic parameter estimates. TTM has been able to estimate expected heritability better than other models. The percentage of top animals in common between the top 10 and the top one percent of simulated animals with different models showed that the ability of the TTM model to select real top animals is more than other models, which shows that this model has identified top animals better.
Conclusions: The results of the present study showed that the censored records are important for estimating the genetic parameters of the number of inseminations per conception in Iranian dairy cows and ignoring those led to a biased estimation of the parameters. Data simulations also showed that using censored records provided more realistic genetic parameter estimates.


Main Subjects

Bahri Binabaj F., Farhangfar H. and Shamshirgaran Y. 2012. Application of poisson generalized linear model for analyzing effects of temperature and relative humidity on the number of insemination per conception of Holstein cows. Journal of Animal Science Research, 22(3): 129-136. (In Persian).
Chang Y. M., Andersen-Ranberg I. M., Heringstad B., Gianola D. and Klemetsdal G. 2006. Bivariate analysis of number of services to conception and days open in Norwegian Red using a censored threshold-linear model. Journal of Dairy Science, 89: 772-778.
Chegini A., Ghavi Hossein-Zadeh N., Hosseini Moghaddam S. H. and Shadparvar A. A. 2019. Genetic aspects of some reproductive, udder health and energy status traits in Holstein cows. Theriogenology, 130: 1-7.
Costa E. V., Ventura H. T., Veroneze R., Silva F. F., Pereira M. A. and Lopes P. S. 2019. Bayesian linear-threshold censored models for genetic evaluation of age at first calving and stayability in Nellore cattle. Livestock Science, 230: 103833.
Donoghue K. A., Rekaya R. and Bertrand J. K. 2004. Comparison of methods for handling censored records in beef fertility data: Field data. Journal of Animal Science, 82(2): 357-361.
Hou Y., Madsen P., Labouriau R., Zhang Y., Lund M. S. and Su G. 2009. Genetic analysis of days from calving to first insemination and days open in Danish Holsteins using different models and censoring scenarios. Journal of Dairy Science, 92(3): 1229-1239.
Keshavarzi H., Sadeghi-Sefidmazgi A., Ghorbani G. R., Kowsar R., Razmkabir M. and Amer P. 2020. Effect of abortion on milk production, health, and reproductive performance of Holstein dairy cattle. Animal Reproduction Science, 217: 106458.
Lázaro S. F., Varona L., e Silva F. F., Ventura H. T., Veroneze R., Brito L. C. and Lopes P. S. 2019. Censored Bayesian models for genetic evaluation of age at first calving in Brazilian Brahman cattle. Livestock Science, 221: 177-180.
Madsen P., Lidauer M. and Mäntysaari EA. 2008. Strategy for estimation of variance components for joint Nordic yield evaluation. Interbull Bulletin, 38: 36-39.
Malhado C. H. M., Malhado A. C. M., Martins Filho R., Carneiro P. L. S., Pala A. and Carrillo J. A. 2013. Age at first calving of Nellore cattle in the semi-arid region of northeastern Brazil using linear, threshold, censored and penalty models. Livestock Science, 154(1-3): 28-33.
Misztal I., Tsuruta S., Strabel T., Auvray B., Druet T. and Lee D. H. 2012. Proceedings from the 7th World Congress on Genetics Applied to Livestock Production, Montpellier, France.
Mohammadi A., Alijani S., Rafat S. A. and Abdollahi-Arpanahi R. 2020. Estimation of genetic and phenotypic parameters and trends of female fertility traits of cows and sire conception rate in Iranian Holstein. Journal of Animal Science Research, 30(3): 37-53. (In Persian).
Nishida A., Aziz M. A., Nishida S. and Suzuki K. 2006. Modelling number of services per conception of Japanese Black cattle by random regression. Journal of Animal Breeding and Genetics, 123(1): 56-63.
Pereira E., Oliveira H. N., Eler J. P., Silva J. I. D. V. and Van Melis M. H. 2007. Comparison among three approaches for evaluation of sexual precocity in Nellore cattle. Animal, 1(3): 411-418.
Rezagholivand Lahrud A., Moradi ShahrBabak M., Moradi ShahrBabak H. and Sattaei Mokhtari M. 2019. Genetic evaluation of reproductive disorders and fertility traits in Holstein cows under standard and recursive mixed models. Animal Production, 20(4): 499-511. (In Persian).
Sargolzaei M. and Schenkel F. S. 2009. QMSim: a large-scale genome simulator for livestock. Bioinformatics, 25(5): 680-681.
SAS Institute. 2014. Users Guide, Version 9.4M2: Statistics. SAS Institute, Cary, NC, USA.
Sun C., Madsen P., Nielsen U. S., Zhang Y., Lund M. S. and Su G. 2009. Comparison between a sire model and an animal model for genetic evaluation of fertility traits in Danish Holstein population. Journal of Dairy Science, 92(8): 4063-4071.
Visual FoxPro9.0. 2006. Microsoft Corporation, USA.