Neser, F. W. C.Ducrocq, V.van Niekerk, Michiel2025-06-042025-06-042024Thesis (Ph.D.(Animal Science))--University of the Free State, 2024http://hdl.handle.net/11660/13085Fixed regression model (FRM) analyses that consider only fixed, non-genetic effects to vary over the lactation are currently used for genetic evaluation of production traits in South African Holstein. With random regression models (RRM), the random animal and permanent environmental effects are allowed to also vary over the lactation. Hence, RRM can account for an individual component representing changes during the lactation i.e., its persistency (PERS), enabling selection for more persistent cows. Also, test-day (TD) records used for genetic evaluations come from cows in contrasted production systems. The main ones rely on full pasture (PAST) or a total mixed ration (TMR), a choice often depending on local average rainfall where herds are situated. TD records from herds were divided into two datasets based on the production system (PAST or TMR). REML was used to analyse production for each of the first three lactations under different multiple-lactation models for milk, fat and protein production, as well as fat and protein percentage. Various FRM were compared to the current FRM officially used for genetic evaluation in South Africa (saFRM). A FRM that cumulates different curves over the lactation for different fixed effects was retained based on results in the PAST dataset and was also applied to the TMR dataset. This model was then broadened to an alternative RRM (aRRM) combining for each lactation an average production and a PERS effect, after which it was compared to the current saFRM under both production systems. The aRRM for both PAST and TMR had a better goodness of fit than the current saFRM for all traits except protein percentage. The mean squared error of aRRM was lower for all traits. Generally, aRRM heritability estimates were higher than with the saFRM at the beginning and end of lactation for most traits in PAST while being mostly higher during late lactation in TMR. Overall, the h² in PAST were mostly higher than in TMR for all traits. Estimates of between-lactations genetic correlations for average production from the aRRM were generally higher. Within-lactations genetic correlations between average production and PERS for TMR from the aRRM were negative and stronger than for PAST. The extra source of information from the aRRM enables a genetic prediction of PERS and is expected to increase accuracy of genetic predictions. Different genetic parameters between the two production systems may denote a genotype x environment interaction. Dairy cattle in South Africa (SA) must produce in varying environments which can roughly be separated as total mixed ration (TMR) and pasture (PAST) production systems according to the level of rainfall in the herd area. Previous studies aimed to determine a possible genotype x environment interaction (GxE) using 305-day lactation records, finding no interaction, but different heritabilities. An alternative random regression model (aRRM) was developed using SA Holstein test-day records of herds with known production system used (TMR or PAST, analysed independently). The aRRM separates the animal genetic (additive) effect into an average-production (level) and -persistency (slope) effect and shows clear differences in the genetic variances and heritability estimates over the lactation between TMR and PAST. The aim of this study was to further investigate the existence of a possible GxE interaction, using the same dataset, methods and aRRM to differentiate each trait (milk, fat and protein production for the first three3 lactations) separately for the two production systems (TMR or PAST). The possible reranking of sires on resulting estimated breeding values from the analysis between TMR and PAST were also investigated. Genetic correlations (0.81 to 0.94) of the genetic-average-production effect between TMR and PAST were strong, while corresponding correlations of the genetic-average-persistency effect were weak (Lactation 1, 0.22 to 0.27) to moderate (Lactations 2 and 3, 0.55 to 0.68) for all traits. Minor reranking of sires between TMR and PAST are predicted for genetic-average-production, but with moderate to major reranking for genetic-average-persistency. Analyzing TMR and PAST separately using a reduced rank model fits the data significantly better, adds additional information and allows for more accurate predictions to improve genetic persistency, more so than using within-PAST estimates. The recording of TD records as TMR or PAST should be seriously contemplated as the genetic component of persistency has a significant impact on total production. Genetic improvement for persistency of milk production in South African Holstein cattle in a total mixed ration or pasture production system will be impeded because of the existence of genotype by environment interaction between the production systems. Previous studies where herds’ production system was known (unlike reality) showed that rainfall level indicates the production system preferred by producers. The study aimed to determine the effect of including average rainfall in a random regression model as a possible proxy for production system used. The animal-additive-genetic effect due to annual rainfall (class) was added to the existing additive-genetic-average-production and -persistency effect. Heritability estimates of rainfall class 1 (‘low’ rainfall) coincided with total mixed ration estimates while rainfall class 5 (‘high’ rainfall) coincided with pasture estimates in later lactations. Results suggest that genetic background responsible for average production level and persistency over the lactations are not consistent over different rainfall areas. Genomic evaluations for dairy cattle have become standard practice in 23 developed countries according to Interbull at the end of 2024. The development and use of many (tens to hundreds of thousands) single nucleotide markers (SNPs) spread over the whole genome opened new doors for improved genetic predictions. Together with the relatively newly developed single-step (ss) method that blends phenotypic, pedigree and genotypes (indicated by the SNPs of key animals) a new evaluation encompassing all this information in a single genetic analysis. Such a ss, genomic evaluation has been shown to enhance accuracy of prediction for various traits. SA Holsteins do not employ such genomic evaluations due to the relatively high cost of genotyping animals playing a significant role. Some 1221 SA Holstein animals (1143 cows) were, however, genotyped. The production system under which 535 of these cows were recorded for milk production was known. Because of the significant genotype x environment interaction between the two production systems, cumulative 295-days milk production for lactation 1 and 2 in TMR and PAST were analysed as different traits (i.e., four-trait analyses). Two-trait analyses (lactation 1 and 2, irrespective of production system used) were also carried out. Using REML, two multi-trait (MT) i.e., four-trait analyses (MT4) were done. In the first analysis (MT4) genomic information was not implemented and yielded estimated breeding values (EBVs) and their accuracies (ACC). The second analysis was the same as the first (MT4) one, except that the single-step method (ssMT4) was used, which incorporated the genomic information and yielded genomically enhanced EBVs (ssGEBVs) and their ACCs. The two-trait analyses were carried out in the same way; a MT2 and ssMT2 analysis. Heritabilities between the ssMT4 vs the MT4 analyses were similar (TMR lactation 1 estimates of 0.23 vs 0.24, respectively; TMR lactation 2 estimates of 0.16, respectively; PAST lactation 1 estimates of 0.31, respectively; PAST lactation 2 estimates of 0.27, respectively). Heritabilities for the ssMT2 vs the MT2 analyses were also similar (0.25, respectively for lactation 1, and 0.17 vs 0.18, respectively for lactation 2). Genetic correlations from the ssMT4 between TMR and PAST for lactation 1 and 2 were 0.81 and 0.70, respectively. Hence, minor to moderate reranking of animals on ssGEBVs between the two production systems might take place for lactation 1 and 2, respectively. On average, the largest and significant increase in ssGEBV vs EBV ACCs were for cows with records (0.17; no progeny in the pedigree) and especially for the corresponding cows with the lowest EBV ACCs (increases of ~0.11 to ~0.40). Dams with records (and progeny in the pedigree) had on average the second largest increase in ssGEBVs (0.06), also with the largest increases in dams with the lowest EBV ACCs (increases of ~0.12 to ~0.33). Results for dams without records followed the same tendency. In this regard, results of the ssGEBV vs EBV ACCs for the ss- and MT2 analyses followed a similar trend. These preliminary results are promising for increasing ACC of genetic prediction using ss genomic evaluations. However, more animals need to be genotyped, especially highly influential sires with genetic merits that represent the whole population. Validation of genomic predictions need to be part of future research in the SA Holstein population.enThesisUniversity of the Free State