S1086: Genetic aspects of beef cow adaptation to diverse U.S. production environments.

(Multistate Research Project)

Status: Active

Project Menu

SAES-422 Reports

Annual/Termination Reports:

[07/29/2025]

Date of Annual Report: 07/29/2025

Report Information

Annual Meeting Dates: 05/21/2025 - 05/22/2025
Period the Report Covers: 05/24/2024 - 05/22/2025

Participants

Brief Summary of Minutes

Accomplishments

Objective 1: Texas A&M University: Eye pigmentation: <ol> <li>Graduate student at North Dakota State University is analyzing data associated with eye pigmentation images.</li> <li>Machine learning algorithms are being implemented to quantify eye pigmentation in various eye structures.</li> <li>Over 6,000 animals with images.</li> </ol> Hair shedding: <ol> <li>Consolidation of over a dozen years of monthly records on multiple cow breed types into single spreadsheet for analyses.</li> <li>Began evaluating yearlings</li> </ol> Oklahoma State University: Study 1. Genetic Heterogeneity of Residual Variance for Growth Traits in Angus Cattle. S.T. Amorim et al. (This work is part of Dr. Amorim’s PhD thesis, and additional analysis were carried out after she joined Oklahoma State University in August 2024). Hypothesis: Birth weight (BW), weaning weight (WW) and yearling weight (YW) residual variances may be under genetic control. Objectives: Investigate the extend of genetic heterogeneity of residual variance at the pedigree level in growth traits of American Angus cattle to develop strategies to manage variation and improve uniformity of production in beef cattle breeding programs. Methods: This study utilized phenotypic and pedigree data provided by Angus Genetics Inc. The dataset included up to 75,000 records for BW and WW and nearly 50,000 for YW. Only records within three standard deviations from the mean were retained for further analysis.Weaning and yearling weights were adjusted based on Beef Improvement Federation Guidelines. Animals were born between 1986 and 2013 and raised across diverse U.S. ecoregions. Contemporary groups (CGs) were formed using combinations of sex, herd, year, and season of birth, with minimum CG sizes of 200 for BW/WW and 100 for YW. Three genetic models were fitted to estimate variance components. The first model (M1) was a traditional homoscedastic animal model assuming constant residual variance. It included direct and maternal additive genetic effects and maternal permanent environmental effects for BW and WW, while YW was modeled with only direct additive effects. The second model (M2) employed a double hierarchical generalized linear model (DHGLM) to allow for heterogeneous residual variance. This model estimated both the mean and residual variance simultaneously, using squared residuals adjusted for leverage to model individual-level variability. M2 included additive genetic effects for both trait mean and variance, and maternal permanent environmental effects for BW and WW. The third model (M3) was a Bayesian genetically structured variance model that treated residual variance as partially under genetic control. It was implemented via MCMC sampling in the GSEVM software. This model accounted for genetic and environmental influences on both the mean and residual variance, with heritability estimates varying according to environmental context. Genetic parameters estimated across models included heritability for the mean (), heritability for residual variance (), genetic coefficient of variation (), and genetic correlation between the mean and residual variance (). These estimates enabled the evaluation of the genetic potential for improving uniformity alongside trait means. Results: Our findings revealed genetic heterogeneity in residual variances across growth traits, with the genetic coefficient of variation for residual variance ranging from 0.26 in M2 to 0.35 in M3 for BW, 0.09 in M2 to 0.31 in M3 for WW, and 0.07 in M2 to 0.31 in M3 for YW. Heritability estimates for residual variance were low, ranging from 0.004 to 0.01 in M2 for all traits. Negative genetic correlations between mean (phenotype) and residual variance were observed for BW (-0.48 in M2 and -0.49 in M3) suggesting the potential to increase trait means while reducing residual variance. Positive genetic correlations indicated that selection for uniformity may be limited without a selection index. Conclusions: M2 and M3 provided similar estimates of additive genetic variance for residual variance. These results indicate the feasibility of reducing variability through selection, representing a first step in integrating growth trait uniformity into breeding goals for beef cattle. Study 2. Genetic Heterogeneity of Residual Variance for Growth Traits in Angus Cattle. S.T. Amorim et al. (This work is part of Dr. Amorim’s PhD thesis, and additional analysis were carried out after she joined Oklahoma State University in August 2024). Hypothesis: Foot Angle (FA) and  Claw Set (CS) residual variances may be under genetic control. Objectives: Investigate the extent of genetic heterogeneity of residual variances on two conformation traits in Angus cattle using genetic homogeneity (M1) and two genetic heterogeneity of residual variance models, including a double hierarchical generalised linear model (DHGLM, M2) and a genetically structured environmental variance model (M3). Methods: This study utilized phenotypic data provided by Angus Genetics Inc. A total of 190,140 phenotypes representing 118,828 animals recorded between December 2008 and October 2022 were used. FA and CS were scored subjectively using a 1-to-9 scale defined by the American Angus Association, where a score of five represents ideal conformation. FA reflects pastern angle and heel depth, while CS captures claw symmetry. Foot scores were taken at or after one year of age, typically before hoof trimming, and scored by a single evaluator within each contemporary group (CG). CGs were constructed by combining multiple systematic factors such as herd, date of measurement, management code, and diet type. CGs with fewer than 150 records or only one score category were excluded, and only one observation per animal was retained.  Three statistical models were used to estimate genetic parameters for the mean and residual variance of FA and CS. The first model (M1) was a standard homoscedastic animal model assuming constant residual variance. This model included fixed effects (CG and age) and a random additive genetic effect, and it was implemented using the AI-REML algorithm in the DMU software. The second model (M2) was a double hierarchical generalized linear model (DHGLM), which allows for heterogeneous residual variance by simultaneously modeling the trait and its squared residuals. M2 incorporated additive genetic effects for both the mean and residual variance and accounted for individual-specific environmental variability using leverage values. It was also estimated using DMU software with an iterative procedure that continued until the convergence of variance components. The third model (M3) was a Bayesian genetically structured variance, assuming that residual variance is partially under genetic control. M3 estimated both the mean and the residual variance components jointly, with direct genetic effects assumed to be normally distributed. This model was implemented in the GSEVM software using Markov Chain Monte Carlo (MCMC) algorithm. Unlike M1 and M2, the heritability estimates in M3 were conditional on environmental context due to the model’s exponential structure and were computed for each combination of fixed effects. To interpret the genetic control of residual variance, three key parameters were estimated: the heritability of residual variance (), the genetic coefficient of variation for residual variance (), and the genetic correlation between the mean and residual variance (). The  quantifies the accuracy of breeding values for residual variance. evaluates the potential for changing residual variance through selection, while  whether selection for the mean of a trait is expected to increase or decrease its variability. Results: Heritability estimates for the mean of FA and CS were consistent with values previously reported in the literature but varied across models. For FA, heritability was estimated at 0.19 in the homoscedastic model (M1), 0.11 in the DHGLM model (M2), and 0.09 in the Bayesian model (M3). For CS, corresponding heritability estimates were 0.16, 0.10, and 0.08, respectively. These findings indicate that models accounting for heterogeneous residual variance (M2 and M3) tend to produce lower heritability estimates for trait means, likely due to their more flexible variance structures. The heritability of residual variance was low (0.08 for FA and 0.001 for CS in M2) the estimates still suggest that residual variance is partially heritable and under genetic influence. The  values for residual variance were moderate for FA (0.08 in M2 and 0.06 in M3) and lower for CS (0.06 in M2 and 0.02 in M3), indicating some potential for genetic improvement of trait uniformity.  Moderate positive genetic correlations were observed between the trait mean and its residual variance in both M2 and M3. For FA, the correlations were 0.52 (M2) and 0.35 (M3), and for CS, 0.41 (M2) and 0.33 (M3). These positive correlations imply that selection to increase the mean of a trait may simultaneously increase variability, potentially limiting the effectiveness of selection for uniformity. As a result, breeding strategies such as selection indices that jointly consider trait mean and variability may be necessary to improve both performance and consistency in foot score traits.  Conclusions: FA and CS show strong potential as indicators of uniformity or resilience in beef cattle because they can be consistently measured throughout the production cycle using either traditional scoring methods or modern digital recording technologies. This study demonstrated that variability in these foot score traits can be modulated through genetic selection, opening the door to incorporating uniformity as a selection goal in beef cattle breeding programs. This is one of the first studies to evaluate genetic heterogeneity of residual variance for foot score traits in beef cattle and to estimate genetic parameters specifically related to uniformity. The estimated genetic coefficient of variation for residual variance averaged 0.07 for FA and 0.04 for CS across the more advanced models (M2 and M3), indicating that selecting for higher mean scores could also lead to a proportional increase in residual variance, approximately 7% for FA and 4% for CS.The positive genetic correlations observed between trait means and residual variances suggest that direct selection on the mean may inadvertently increase variability. This highlights the need for alternative strategies, such as the inclusion of uniformity indicators in selection indices, alongside economically important traits. University of Arkansas: Udder conformation: The objective of this study was to determine if any relationships existed between udder conformation and production traits in cows housed at the University of Arkansas beef research unit near Fayetteville.  An Angus-based fall calving cowherd (n ≈ 194) was observed, and udder scores were recorded at calving in the Fall of 2023 and at weaning in May 2024.  Cows were evaluated on a scale from 1 to 9 for udder suspension and teat size according to BIF guidelines. A score of 1 indicated a very pendulous suspension and large, balloon shaped teats and a score of 9 represented a tight suspension and refined teat size. Phenotypic data for cow age, Pre-Breeding BW, Pre-Breeding BCS, BCS of cow at weaning, BW of cow at weaning, AI pregnancy and overall pregnancy, along with calf weaning weight, and adjusted 205-day weaning weight (adjusted for dam age and calf gender) were collected. Means will be generated for suspension and teat scores by dam age. Cows receiving a suspension or teat score of 1-3 were considered undesirable, scores ranging from 4-6 were ideal and those scores 7-9 will be considered marginal. Cow performance data will be analyzed using the acceptability parameters described with CORR, GLIMMIX, MIXED, and FREQ procedures of SAS.  Significance was declared at P ≤ 0.05 and tendencies were observed at 0.05 Very little variation in frequency of cows in the ideal and marginal category but far outnumbered the frequency of cows considered unacceptable.  Overall cow frequencies at both calving and weaning for ideal, marginal, and unacceptable for udder suspension were 144, 211 and 13 respectively and teat size frequency were 195, 147 and 4 respectively.  Cows that scored the tightest udders and most refine teats were also the youngest in age and those that scored the most pendulous suspension and biggest teats were the oldest cows.  Correlation for cow age with suspension at calving was statistically different (r= -0.615, p<.0001).  The same was true about cowage with teat size at calving (r= -0.561, p<0.561). At calving, udder suspension and teat size were moderately correlated with cow age (r=-0.394, r=-0.321 respectively).  The effect of udder suspension and teat score on weaned calf performance was not significantly different (P > 0.13, 0.79 respectively). Cows with unacceptable suspension and teat size had higher pre-breeding body weights but were not statistically different from ideal and marginal cows.  Udder Suspension and Teat size did not affect weaning body weight (P = 0.812), weaning body condition scores (P = 0.588), overall AI conception rates (P = 0.753), or overall pregnancy rates (P = 0.601). Foot soundness: The objective of this study was to determine if any relationships existed between foot soundness and production traits in cows housed at the University of Arkansas beef research unit near Fayetteville. An Angus-based fall calving cowherd (n ≈ 194) was observed, and foot scores were recorded at weaning in May 2024.  Cows were evaluated on a scale from 1 to 9 for foot angle and claw set according to the American Angus Association with 1 indicating straight pasterns and short toes, and 9 indicating curled toes and crossed claws. Phenotypic data for cow age, Pre-Breeding BW, Pre-Breeding BCS, BCS of cow at weaning, BW of cow at weaning, AI pregnancy and overall pregnancy, along with calf weaning weight, and adjusted 205-day weaning weight (adjusted for dam age and calf gender) were collected. Means will be generated for suspension and teat scores by dam age. We assigned cows with a claw and angle score between 4 and 6 to have an overall “ideal” foot score. Those that scored 3 or 7 were considered “marginal” and any scores outside these were assigned “undesirable”. Cow performance data will be analyzed using the acceptability parameters described with CORR, GLIMMIX, MIXED, and FREQ procedures of SAS.  Significance was declared at P ≤ 0.05 and tendencies were observed at 0.05 Mean Cow Age of ideal Claw set was 4.5, 5.6 for marginal, and unacceptable was 6.3.  Mean cow age of ideal foot angle was 4.5, 5.4 for marginal, and unacceptable was 7.1   The frequency of ideal, marginal and unacceptable scores for both claw set, and foot angle were 212, 135 and 15 respectively.   For claw set and foot angle, statistical differences existed for pre-breeding body weight between ideal and marginal cows (1204,1278, p=0.026, 1200,1269, p=0.021). Pre-Breeding BCS, Weaning BW, Weaning BCS, and calf weaning weight was not affected by claw set or foot angle.   South Dakota State University: South Dakota State University (SDSU) continues to collect liquid water intake phenotypes on beef calves.  In 2024-2025, SDSU collected liquid water intake phenotypes on 426 beef calves.  Additionally, weight, sex, age, breed, and pedigree information is available on these calves.  Our long term goal is to create a population resource that can be used to estimate genetic parameters for water intake phenotype.  Mississippi State University: Fall calving 2004 Angus cows were bred to A.I. sires with extreme hair shedding EPDs for the third breeding season. Data were collected for pooling for the duration of the S-1086 regional project for Objective 1. Fall born 2003 Angus bulls (n=13) and heifers (n=15) were evaluated for growth and hair shedding from May to July. Hair shedding scores were highly negatively associated (-0.40) with ADG, indicating as hair shedding scores decrease ADG increases. Matings will continue to be made to produce animals with extremes in hair shedding phenotypes. This will allow for populations to be created and evaluated for production traits and their association to hair shedding. Shedding of the winter hair coat continues to show an association to increased growth as previous research has suggested. Objective 2: Texas A&M University: Estimate heterosis in Brahman-Bos taurus crosses. Table 1.  Estimates of heterosis from Texas Heterosis Retention project for pregnancy rate at first three matings and as repeated records for pregnancy rate and birth weight. B = Brahman; H = Hereford; A= Angus.  F1 and F2 = first and second (matings of F1 animals) crossbreds.  BH- (HB-)sired F2 animals had Brahman (Hereford) paternal grandsires. Table 2.  Heterosis for cow reproduction:  Genomics Heterosis preliminary results <table width="100%"> <tbody> <tr> <td width="43%"> Breed </td> <td width="8%"> N </td> <td width="33%"> Calving rate </td> <td width="13%"> SD </td> </tr> <tr> <td width="43%"> Angus </td> <td width="8%"> 51 </td> <td width="33%"> 0.8 </td> <td width="13%"> 0.4 </td> </tr> <tr> <td width="43%"> Brahman </td> <td width="8%"> 45 </td> <td width="33%"> 0.69 </td> <td width="13%"> 0.47 </td> </tr> <tr> <td width="43%"> F1 Angus sire </td> <td width="8%"> 49 </td> <td width="33%"> 0.8 </td> <td width="13%"> 0.41 </td> </tr> <tr> <td width="43%"> F1 Brahman sired </td> <td width="8%"> 43 </td> <td width="33%"> 0.93 </td> <td width="13%"> 0.26 </td> </tr> </tbody> </table> Table 3.  Calf birth weight (2025-born) means by breed group:  Genomics Heterosis preliminary results <table> <tbody> <tr> <td width="156"> Calf breed </td> <td width="156"> N </td> <td width="156"> Birth wt, kg </td> <td width="156"> SD </td> </tr> <tr> <td width="156"> Angus </td> <td width="156"> 25 </td> <td width="156"> 32.2 </td> <td width="156"> 5.28 </td> </tr> <tr> <td width="156"> Brahman </td> <td width="156"> 19 </td> <td width="156"> 32.5 </td> <td width="156"> 5.07 </td> </tr> <tr> <td width="156"> F1 Angus sire </td> <td width="156"> 12 </td> <td width="156"> 33.6 </td> <td width="156"> 4.21 </td> </tr> <tr> <td width="156"> F1 Brahman sired </td> <td width="156"> 16 </td> <td width="156"> 45.0 </td> <td width="156"> 7.44 </td> </tr> <tr> <td width="156"> ¼ Brahman </td> <td width="156"> 43 </td> <td width="156"> 32.4 </td> <td width="156"> 6.51 </td> </tr> <tr> <td width="156"> ¾ Brahman </td> <td width="156"> 36 </td> <td width="156"> 39.4 </td> <td width="156"> 6.09 </td> </tr> </tbody> </table> Objective 3: Texas A&M University: <ol> <li>Annually combine information for joint analysis for economically relevant traits</li> <li>Genotype-environment combinations (see Copley et al., 2022) relative to broad systems such as</li> <li>Artificial insemination vs. natural matings.</li> <li>Spring vs. Fall calving systems.</li> <li>Regional or longitudinal/latitudinal characterization of performances</li> <li>Herd size dynamics—implementation of genetic improvement programs and implementation of technology in small, medium, and large beef production units.</li> <li>Across-location/system estimates reported as deviations from Angus.</li> <li>Members will recruit participation from:</li> <li>Smaller university research facilities that otherwise have no hope of working in genetics.</li> <li>Private herds.</li> <li>Non-profit research groups such as the East Foundation in San Antonio and the Noble Foundation in Ardmore. </li> </ol> Oklahoma State University: Study 3. Genomic Study for Pregnancy Loss in Brahman Cattle. S.T. Amorim et al.  Hypothesis: Pregnancy loss (PL) in Brahman cattle is influenced by additive genetic variation and specific genomic regions contribute to embryo and fetal survival. Objectives: Estimate variance components and genetic parameters for PL and identify genomic regions and candidate genes associated with PL through genome-wide association analyses and functional annotation to explore biological pathways related to embryo development, implantation, fertilization, and signaling processes. Methods: Pregnancy loss data (n = 29,905) was provided by ANCP (Ribeirao Preto, Sao Paulo, Brazil). Records were classified into three reproductive stages: nulliparous, primiparous, and multiparous cows. All animals were managed under reproductive health protocols, including vaccination against brucellosis and common reproductive pathogens. The breeding season involved a combination of artificial insemination and natural mating, and pregnancy loss was identified in cows that failed to calve after confirmed pregnancy. A total of 921 animals were genotyped using a 53K SNP chip, and quality control filtered markers based on call rate, MAF, Hardy-Weinberg equilibrium, and other standard criteria, resulting in 46,342 SNPs for analysis. Variance components were estimated using a single-step genomic BLUP (ssGBLUP) procedure with a repeatability threshold animal model, incorporating fixed effects, additive genetic effects, and permanent environmental effects. Bayesian inference via Gibbs sampling was implemented in GIBBSF90+, using 1,000,000 iterations with a 200,000 burn-in and 10-step thinning. A genome-wide association study (GWAS) was conducted using the single-step methodology. Marker effects were estimated, and regions explaining more than 1% of the additive genetic variance were selected for downstream analysis. Gene annotation was performed using Ensembl Biomart (ARS-UCD1.3 build), and functional enrichment was assessed via DAVID for KEGG pathways and GO terms. Orthologous gene functions were explored in MeSH, OMIM, and OMIA databases to identify candidate biological processes underlying pregnancy loss. Results: The heritability estimate for PL was low (0.114). GWAS analyses revealed 17 candidate genomic regions containing 92 genes located on BTA4, 7, 8, 9, 11, 12, 16, 18, 19, 21, 22, and 29. These regions were enriched with genes involved in key biological processes such as embryonic development and implantation, fertilization, G protein-coupled receptor activity, embryonic brain development, olfactory signaling, and calcium signaling. Many of these genes have orthologs in humans, rats, and mice, reinforcing their potential functional relevance to pregnancy retention across species. Conclusions: This study confirms that pregnancy loss is a low-heritability trait, with environmental factors playing a dominant role in its expression. However, the detection of additive genetic variation suggests that genetic selection can still contribute to improving reproductive performance. The seventeen genomic regions were found to account for significant portions of the additive genetic variance, harboring candidate genes involved in embryonic development, RNA processing, ciliary function, immune regulation, and neuroendocrine signaling. Functional enrichment analyses supported the biological relevance of these genes by identifying pathways associated with reproduction and early embryonic development. Overall, the findings highlight the complex genetic architecture of pregnancy loss and provide valuable insights for the development of genomic selection strategies aimed at reducing reproductive failure, enhancing cow longevity, and improving the sustainability and productivity of beef cattle systems.

Publications

Impact Statements

Modern Angus-Brahman crosses had very large heterosis for calving rate and this is consistent with earlier work.
Preliminary results indicate that birth weights of Brahman-Angus crossbred calves do not follow Mendelian rules of inheritance; particularly, those crossbred calves sired by Brahman have large means and variances compared to the reciprocal cross.
Heterosis expressed by F2 Brahman Hereford cows does not conform to dominance model expectation. Those cows with Hereford paternal grandsires had calving rates as high as the F1 cows (dominance expectation is ½ of that of the F1 cows.
Implementation of the “leave one chromosome out” methodology in order to avoid double statistical fitting of an effect results in enormous genomic inflation in bovine genome wide association studies. This methodology may not be appropriate.
The exploratory analysis of Study 1 from Oklahoma State University revealed that residual variance is, at least in part, under genetic control. These findings provide important insight into the potential for improving trait uniformity of production through genetic selection. By capturing and modeling variability itself as a heritable component. The findings support the development of breeding strategies that reduce phenotypic variability while improving productivity, helping to select beef cattle that are better adapted to diverse environmental conditions and management systems.
The exploratory analysis in Study 2 from Oklahoma State University revealed that foot angle and claw set exhibit low heritability for residual variance, indicating that genetic selection could be leveraged to improve foot score uniformity. Incorporating residual variance into genetic models enhances the ability to select for ideal foot structure, while also offering insight into these traits as potential indicators of structural resilience. Improving foot score uniformity through selection may contribute to greater longevity, mobility, and overall soundness in breeding animals, ultimately supporting the adaptability and sustainability of beef cattle in diverse production environments.
Study 3 from Oklahoma State University is one of the first to characterize the genetic basis of pregnancy loss in Brahman cattle using single-step genomic approaches. Despite low heritability, genomic regions associated with pregnancy retention were identified, many of which are linked to reproductive, neuroendocrine, and immune functions. These findings provide a basis for improving reproductive efficiency through marker-assisted or genomic selection.
Udder and teat quality are among the most important functional traits of beef females. Unsound udders and teats are associated with reduced productive life and inferior calf performance, and poor udder and teat conformation is a major reason why cows are culled from the breeding herd. Understanding the implications of these scores could improve the culling process and improve production efficiency.
Sound feet are important components in cattle production systems and can influence nutritional aspects of cattle. Hoof soundness has been reported to have effects on breeding and reproductive success and both body weight and body composition. Implementing these scores can aid in selecting for more sound cows.
Hair shedding scores, although subjective, are well within the reach of both commercial and seedstock breeders. By using these scores and understanding their implications in cattle production, producers can utilize them in the match of genetic resource and production resources. This could easily increase current overall production. Monitoring coat dynamics in shedding months appears to be more critical for assessment of adaptability.