Genetic evaluation of crossbred Bos indicus cow temperament at parturition

Abstract

Cow temperament at parturition may be mostly a measure of aggressiveness. The heritability of cow temperament at parturition in Bos taurus cows has been reported to be low. The objectives of this study were to estimate the heritability of cow temperament at parturition, conduct a genome-wide association analysis of cow temperament at the time of parturition, and estimate the correspondence of cow temperament at the time of parturition with cow productive performance and early-life temperament traits in Bos indicus crossbreds. Cow temperament was assessed from 1 to 5 indicating increasing levels of aggressiveness of cows (937 cows and 4,337 parturitions) from 2005 to 2022. Estimates of heritability and repeatability were 0.12 ± 0.024 and 0.24 ± 0.018. The estimates of proportion of phenotypic variance were 0.13 ± 0.019 and 0.02 ± 0.011 for permanent and maternal permanent environmental components, respectively. Estimates of heritability for maximum lifetime temperament score and proportions of temperament scores >1 were 0.18 ± 0.07 and 0.13 ± 0.072. Within cycles (generations), 2-yr-old cows had lower temperament score means than cows in most other age categories. There were low to moderate positive estimates of unadjusted correlation coefficients (r = 0.22 to 0.29; P < 0.05) of unadjusted temperament score with temperament measured on the same females when they were 8 mo old. There were low to moderate positive estimates of correlation coefficients (r = 0.09 to 0.37; P < 0.05) of unadjusted temperament score with calving rate, weaning rate, weaning weight per cow exposed, and weaning weight per 454 kg cow weight at weaning. Cows with the lowest temperament score had lower (P < 0.05) calving and weaning rate than cows in other temperament categories. Within 3 of 5 cycles, cows with the lowest temperament score (totally docile) had lower (P < 0.05) weaning weight per cow exposed than cows in other temperament categories. There were 2 SNP on BTA 4 associated with maximum lifetime temperament score (FDR < 0.05). The non-genetic influence of a cow’s mother was documented in her own temperament measured at the time of calving; this may be a consequence of learned behavior. Less aggressiveness displayed by cows at the time of calving may be accompanied by lower reproductive and maternal performance.

Cow temperament at the time of parturition is influenced by non-genetic life experience, even that of her own dam. Docility at the time of parturition was associated with lower reproduction and productivity.

Introduction

Maternal instinct in cows may manifest itself in aggressive or defensive behavior, especially in relation to human interactions. Cattle temperament has been measured by both objective and subjective methods. There is much documentation of favorable relationships between cattle temperament at various times of production and the corresponding associations with performance in different traits such as reproductive efficiency, growth, weight, meat quality, milk yield, and other traits (Hoppe et al., 2010; Cziszter et al., 2016; Alvarenga et al., 2022); that is, favorable temperament corresponds to favorable production metrics. There is potential for selection for improved temperament during handling to impact the maternal protectiveness of beef cows (Turner and Lawrence, 2007; Turner et al., 2013); however, cow temperament after calving was lowly heritable (Buddenberg et al., 1986; Vallée et al., 2015). Cow temperament traits at the time of calving corresponded to some maternal behaviors in Gyr cattle (Ribeiro Vicentini et al., 2022); more protectiveness was associated with increased time eating in a feeder during the pre-calving period and cows that spent more time lying down in the pre-calving period were less attentive to human intervention with their newborn calves. However, Zebu crossbred cow temperament varied little across the peripartum period (Orihuela and Ungerfeld, 2020). Differences among breeds and the relationship of cow temperament with some production measures were investigated in Bos taurus cows in Arkansas (Buddenberg et al., 1986; Sandelin et al., 2005; Sims et al., 2023). Those impacts may be different in Bos indicus cattle. The objectives of this work were to (1) estimate the heritability of cow temperament at parturition, (2) identify genomic regions associated with cow temperament at parturition, and (3) estimate the correspondence of cow temperament at the time of parturition with their temperament evaluated when they were weaned calves and with measures of cow productive performance.

Materials and Methods

Cattle

All animal procedures were conducted with multiple successive approvals from the Institutional Animal Care and Use Committees of both Texas A&M University and Texas A&M AgriLife Research. Five groups of 1/2 Nellore 1/2 Angus crossbred females were evaluated from 2005 through 2022 at the Texas A&M AgriLife Research Center at McGregor. Some of these populations were previously described (Riley et al., 2013, 2014; Hulsman Hanna et al., 2014). Purebred Nellore bulls and purebred Angus cows were bred to produce the F₁ founders. Cycle 1 animals were F₂ from 14 full sibling embryo transfer families (5 F₁ bulls and 14 F₁ cows) born in the fall or spring from 2003 through 2007. One of the F₁ sires had only two project calves; records from those were excluded from data. All other cycles were produced by natural service. Cycle 2 cows were also F₂ Nellore-Angus; however, Angus-sired F₁ bulls and cows were also used to produce these. Cycle 2 cows were born from 2010 to 2015. Subsequent cycles were generations that followed the first cycle; that is, bulls and cows from the preceding cycle produced the very next cycle in this order: 1, 3, 4, and 5. Cycles 3, 4, and 5 were therefore F₃, F₄, and F₅ Nellore-Angus. Cycle 3 cows were born from 2009 to 2013. Cycle 4 cows were born in 2014 and 2015. Cycle 5 cows were born in 2017 and 2018.

The breeding season in all years was from the beginning of May until the beginning of July; each breeding herd contained multiple sires, one bull per 20 to 25 cows. Cows from all cycles were exposed to Angus bulls as yearlings. Fall-born Cycle 1 cows were first exposed to bulls at approximately 18 mo of age to align them with a spring calving system. Cows in each cycle were then exposed to bulls in the same cycle to produce animals for the next generation. Target numbers of calves to produce in the subsequent generation decreased across time. Other than years where they were mated to produce calves in the next generation, cows in Cycles 1, 3, 4, and 5 were exposed to Angus or Hereford bulls. Cycle 2 cows were always exposed to Angus or Hereford bulls. Calves were born each year from mid-February to mid-April. Calves were weaned at approximately 7 mo of age in October of each year. In 2011 and 2022, calves were weaned early in August or September because of drought conditions. Cows were removed from the project for health reasons or for two failures to conceive and wean a calf. After 14 yr of age, cows were removed for a single failure to conceive and wean a calf. The majority of the F₃ cows were removed in 2015 from the project after it was determined that sufficient F₄ calves had been produced. A few cows died and a few were so aggressive and dangerous that they were removed from the project. Some cows in all cycles had records through 2022; distributions of records within cycles by age are presented in Table 1.

Table 1.

Numbers of records within cycles by cow age

Cow age, yr			Cycle1
	1	2	3	4	5	Total
2	176	126	238	72	43	655
3	216	99	158	72	44	589
4	218	154	145	86	27	630
5	213	148	69	74		504
6	214	126	43	61		444
7	221	114	13	55		403
8	200	83	11	26		320
9	171	51	11			233
10	140	33	8			181
11	106	18				124
12	79					79
13	57					57
14	47					47
15	34					34
16	22					22
17	12					12
18	3					3
Total	2,129	952	696	446	114	4,337

¹Cycles 1 and 2 were distinct populations of F₂ Nellore-Angus. Cycles 3 through 5 were females produced by inter se matings of F₂ (Cycle 1 only), F₃, and F₄, respectively, bulls and cows.

Inbreeding was avoided as much as possible. Cycle 1 cows were pastured as two groups by family; sons from two of the four F₁ bulls were bred to daughters of the other two F₁ bulls. Inbreeding was similarly minimized in production of the F₄ generation; that is, daughters from two of the sire lineages were bred to sons of the other two sire lineages. Cycles 4 and 5 cows were maintained as single groups.

Traits

Temperament scores were assigned to cows at the time their calves were weighed and tagged, almost always within 1 d after calving, by consensus, of one or more of three evaluators across all years. Scores were assigned based upon the cow’s behavior at that time: 1 = totally docile, 2 = protective, but not aggressive, 3 = moderately aggressive when calf is disturbed, 4 = very aggressive when calf is disturbed, and 5 = very aggressive even when calf is not disturbed. Half scores were also assigned (e.g., 3.5). Most cows had multiple scores corresponding to parturitions. The maximum temperament score recorded for each cow across her productive lifetime was evaluated as a single trait. Because some cows had the same maximum score recorded multiple times, the earliest of those was designated the maximum. For each cow, a distinct trait was calculated as the proportion of the temperament scores for all calvings across her productive lifetime that were greater than 1. Data are summarized by trait and cycle (generation) in Table 2.

Table 2.

Numbers of cows and records in each cycle and unadjusted means (SD) for temperament at parturition

Cycle1			Cow temperament score
	Cows	Records	All2	Maximum3	Proportion > 14
1	286	2,129	2.64 (1.38)	3.66 (1.34)	0.78 (0.32)
2	228	952	2.34 (1.14)	3.11 (1.22)	0.71 (0.30)
3	273	696	2.38 (1.27)	2.77 (1.44)	0.64 (0.41)
4	103	446	2.17 (1.10)	2.93 (1.18)	0.63 (0.33)
5	56	114	1.37 (0.66)	1.64 (0.79)	0.28 (0.33)
Total	946	4,337	2.45 (1.29)	3.07 (1.40)	0.68 (0.36)

¹Cycles 1 and 2 were distinct populations of F₂ Nellore-Angus. Cycles 3 through 5 were females produced by inter se matings of F₂ (Cycle 1 only), F₃, and F₄, respectively, bulls and cows.

²Temperament score: 1 = totally docile, 2 = protective, but not aggressive, 3 = moderately aggressive when calf is disturbed, 4 = very aggressive when calf is disturbed, and 5 = very aggressive even when calf is not disturbed.

³Maximum temperament score of each cow (across lifetime records) at the age of earliest occurrence (some cows had more than one score that was the maximum value).

⁴Proportion of cow temperament scores (across lifetime records) greater than 1.

Genotypes

Genotypes for 777,962 SNPs from the BovHD array (Illumina Inc., San Diego, CA) or genotypes for 52,785 SNPs from Version 1 of the BovineSNP50 array (Illumina Inc., San Diego, CA) were available for 14 Nellore bulls and 9 Angus cows that were the grandparents and other ancestors of Cycle 1, 4 Nellore-Angus F₁ bulls and 14 Nellore-Angus F₁ cows (i.e., parents of Cycle 1). Genotypes from the BovineSNP50 array were obtained for F₂ females from Cycle 1. Genotypes for 49,629 SNPs from the IDB V.3 array (Weatherbys Scientific, Newhall, Naas, Co. Kildare, Ireland) were available for reciprocal F₂ females from Cycle 2, F₃ females from cycle 3, and F₄ females from cycle 4. Genotypes for 47,843 SNPs from the GGP 50K array (Neogen Genomics, Lincoln, NE) were obtained for 59 F₅ females. All genotypes were in Illumina top orientation. The SNP manifest for Version 1 of the BovineSNP50 array was based on coordinates from the Btau4.0 assembly of the bovine genome. The SNP manifests for the BovHD array, IDB V.3 array, and GGP 50K array were based on the UMD3.1 assembly. All SNP coordinates were converted to the ARS-UCD1.2 assembly based on coordinates from Schnabel (2018). Prior to merging genotypes from different arrays, genotypes were removed for markers and animals with call rates less than 90%, and for markers with a minor allele frequency < 5%. After merging, duplicates of variants that shared the same coordinate and allele code were removed. Parentage was determined from identity-by-descent computations in PLINK v1.9 (Chang et al., 2015). Across the arrays, ~99,000 SNPs with unique positions and genotypes were extracted from the BovHD data for the purebred and F₁ founders (n = 38), as a reference panel for imputation. Genotypes for each autosome were converted to variant call format (vcf) and phased with eagle v2.4.1 (Loh et al., 2016) using the reference panel by invoking four threads and assuming 1 Mb was equivalent to 1 cM. Phased genotypes were then imputed to 99K density using Minimac3 (Das et al., 2016). Prior to GWAS, SNP that deviated (P < 0.0001) from Hardy-Weinberg proportions were removed (Wigginton et al., 2005).

Statistical analyses

Traits were analyzed using animal models with ASReml 4.1 (Gilmour et al., 2015). Fixed effects investigated include various parameterizations of cow age, year of record, cycle, and calf sex. Fixed effects were determined with a criterion of P < 0.15 of the F statistic while modeling only an additive genetic effect. After the fixed effects were confirmed, depending upon the analyzed trait, various random effect structures were investigated including the additive genetic, maternal additive genetic, permanent environmental and maternal (based on the dam of the cow) permanent environmental variances, as well as the additive genetic-maternal additive genetic covariance. A combination of pedigree and marker information was used to construct the hybrid relationship matrix using the ASRgenomics package (Gezan et al., 2022) in the R Statistical computing environment. Final random structures were determined by likelihood ratio tests of nested models and the fixed effects reconfirmed. All genetic variances and genetic relationships were omitted from final models to generate residuals for each trait for genome-wide association analyses.

Residuals were dependent variables in genome-wide association analyses with univariate procedures of GEMMA (Zhou and Stephens, 2012). Residuals for repeated measures per cow were averaged. Covariances among animals were modeled using a standardized genomic relationship matrix constructed with the SNP markers. Candidate genes were identified for each marker with detected associations using the R statistical program package Map2NCBI (Hulsman Hanna and Riley, 2014).

Unadjusted correlation coefficients of cow temperament traits with production traits were estimated using R Statistical Software. Estimated correlation coefficients included the relationship of each cow temperament trait with (1) the same female’s temperament scores for aggressiveness, flightiness, gregariousness, nervousness, and overall temperament (Cycles 1, 2, and 3 cows only) at weaning (Riley et al., 2014), (2) cow body condition score (BCS) 1 to 9 scale indicating increasing amounts of fat cover evaluated visually (Herd and Sprott, 1986), (3) birth weight, weaning weight, and weaning BCS (1 to 9 scale) of the calf of the cow, (4) the cow’s lifetime calving and weaning rates (each cow was given values of 1 or 0 for success or failure to give birth or wean a calf in a given year), (5) calf weaning weight per cow exposed to bulls for breeding (i.e., those that did not wean a calf were given a weaning weight value of 0), and (6) calf weaning weight per 454 kg cow weight (i.e., 1,000 lb) as a measure of efficiency. Residuals were generated for each record from mixed models that included additive genetic and permanent environmental (or maternal permanent environmental effects for calf traits) and fixed effects including parameterizations of year, cow age, and cycle. Calf sex was included in analyses of calf weight traits. Calving and weaning rate analyses were analyzed as binomially distributed and employing a logit link function. Residuals or actual measurements were evaluated as averaged values for each cow and directly (cow temperament score and other traits at each observed time).

Various covariates were evaluated after final models were determined. These included linear and quadratic covariates in separate models: pedigree and genomic inbreeding coefficients, calf birth weight, and calving date within year.

In separate analyses, each cow’s maximum lifetime temperament score was used to create a 5-level fixed effect. Half scores (e.g., 1.5, 3.5) were rounded down to assign to levels. This effect was evaluated as a single effect and as nested within cycles. Animal models with fixed and random effects as described above were used to accompany this fixed effect in cow productivity traits: calving and weaning rates, and calf weaning weight per cow exposed to bulls for breeding, calf weaning weight per 454 kg cow weight.

Results

Repeated records analyses

The fixed effects of cycle and year were included in all models (P < 0.001). Inspection of data distribution by age of cows resulted in creation of five categories of cow age that correspond to BIF (2023) guidelines: 2, 3, 4, 5 to 10, and over 10 years of age to reduce the number of categories. Not all cows in all cycles had records in each age category. Therefore, cow age category was nested in levels of cycle (P < 0.001). Temperament score means are presented in Table 3. Within each cycle, 2-yr-old cows had significantly lower temperament scores than those in most other age categories. Within 2-yr-olds, the Cycle 1 (F₂) had a significantly higher mean than cows in Cycle 4. Likelihood ratio tests supported the additive genetic component (P < 0.001), the permanent environmental (P < 0.001), and the maternal permanent environmental component (P = 0.04). Estimates of proportion of phenotypic variance explained by these components were 0.12 ± 0.024, 0.13 ± 0.019, and 0.02 ± 0.011, respectively. The estimate of repeatability was 0.24 ± 0.018.

Table 3.

Adjusted temperament score¹ means by age categories and cycle

	Cycle3
Age2, yr	1	2	3	4	5
2	1.99 ± 0.15a, y	1.48 ± 0.14a, xy	1.50 ± 0.12a, xy	1.09 ± 0.18a, x	1.41 ± 0.23a, xy
3	2.35 ± 0.13b, y	2.19 ± 0.15b, xy	1.79 ± 0.14a, x	1.77 ± 0.18b, x	1.95 ± 0.23ab, xy
4	2.45 ± 0.15b	2.71 ± 0.14c	2.30 ± 0.13b	2.10 ± 0.18b	2.70 ± 0.27b
5 to 10	2.58 ± 0.10b	3.09 ± 0.11d	2.57 ± 0.14b	2.83 ± 0.16c
>10	2.82 ± 0.15b	3.08 ± 0.27cd

¹Temperament score: 1 = totally docile, 2 = protective, but not aggressive, 3 = moderately aggressive when calf is disturbed, 4 = very aggressive when calf is disturbed, and 5 = very aggressive even when calf is not disturbed.

²Cow age categories that correspond to BIF (2023) guidelines.

³Cycles 1 and 2 were distinct populations of F₂ Nellore-Angus. Cycles 3 through 5 were females produced by inter se matings of F₂ (Cycle 1 only), F₃, and F₄, respectively, bulls and cows.

^{a, b, c}Means in the same column that do not share a common superscript differ (P < 0.05) after correction for multiple comparisons.

^{x, y}Where present, means in a row that do not share a common superscript differ (P < 0.05) after correction for multiple comparisons.

Single record analyses

The maximum lifetime temperament score and the proportion of all temperament scores greater than 1 were single records per cow. The effect of cow age nested in cycles (P < 0.001) presented a similar pattern to that from repeated records analyses of temperament scores (Table 4). Evidence of the effect appeared to be attributed to lower means for the maximum lifetime temperament score of 2-yr-old cows for all cycles and for the maximum lifetime temperament score of 2- and 3-yr-old Cycle 1 cows relative to the means for the same ages of cows in the other cycles.

Table 4.

Means for maximum temperament score¹ by cow age category and cycle

	Cycle3
Age2, yr	1	2	3	4	5
2	2.98 ± 0.19a, y	1.96 ± 0.19a, x	1.97 ± 0.13a, x	1.38 ± 0.32a, x	1.05 ± 0.25a, x
3	3.33 ± 0.16a, y	3.26 ± 0.23bc, y	2.53 ± 0.16b, x	2.08 ± 0.26b, x	1.46 ± 0.32ab, x
4	3.60 ± 0.19ab	3.25 ± 0.19b	3.22 ± 0.17c	2.80 ± 0.26bc	2.50 ± 0.35b
5 to 10	3.95 ± 0.13b	3.88 ± 0.15c	3.76 ± 0.22c	3.42 ± 0.20c
> 10	3.84 ± 0.26ab

¹Maximum lifetime temperament score of each cow (across lifetime records) at the age of earliest occurrence (some cows had more than one score that was the maximum value).

²Cow age categories that correspond to BIF (2023) guidelines.

³Cycles 1 and 2 were distinct populations of F₂ Nellore-Angus. Cycles 3 through 5 were females produced by inter se matings of F₂ (Cycle 1 only), F₃, and F₄, respectively, bulls and cows.

^a–dMeans in the same column that do not share a common superscript differ (P < 0.05) after correction for multiple comparisons.

^{x, y}Where present, means in a row that do not share a common superscript differ (P < 0.05) after correction for multiple comparisons.

Means by cycle for proportion of temperament scores greater than 1 are presented in Table 5. Parameterizations of cow age were not included in analyses of the proportion of temperament scores greater than 1 because those values were calculated across all the records a cow had across years and ages. Cows in Cycles 1 and 2 had a higher (P < 0.05) proportion of temperament scores greater than 1 than cows in the other cycles. Cycle 5 cows had a lower (P < 0.05) proportion of temperament scores greater than 1 than those in Cycles 3 and 4. Additive genetic effects were the only random effects included in analyses of these traits.

Table 5.

Means for proportion of temperament scores greater than 1 by cycle

Cycle1	N	Proportion > 12
1	286	0.78 ± 0.028a
2	228	0.71 ± 0.036a
3	273	0.62 ± 0.035b
4	103	0.60 ± 0.055b
5	56	0.26 ± 0.054c

¹Cycles 1 and 2 were distinct populations of F₂ Nellore-Angus. Cycles 3 through 5 were females produced by inter se matings of F₂ (Cycle 1 only), F₃, and F₄, respectively, bulls and cows.

²Proportion of cow temperament scores (across lifetime records) greater than 1 (1 represents totally docile).

^a–cMeans that do not share a common superscript differ (P < 0.05).

Estimates of heritability for maximum lifetime temperament score and for the proportion of temperament scores greater than 1 were 0.18 ± 0.07 and 0.13 ± 0.072, respectively.

Genome-wide association

Two markers on BTA 4 were associated with maximum lifetime temperament score while controlling FDR < 0.05 (Table 6). Two additional markers on BTA 4 and a single marker on BTA 11 were identified under relaxed significance requirements (0.05 < FDR < 0.2). One SNP was identified (0.05 < FDR < 0.2) for proportion of scores > 1 on BTA 14. The candidate gene nearest to the SNP with strongest association on BTA 4 was cordon-bleu WH2 repeat protein (COBL).

Table 6.

Markers and candidate genes associated with cow temperament traits

BTA1	SNP	Mb	MAF2	P-value	Candidate gene	Distance3, kb
Maximum lifetime temperament score4
4	BovineHD0400000882	3.6	0.43	1  × 10–7	LOC112446320	179.2
	BovineHD0400000863	3.6	0.22	8.9 × 10–6*	LOC112446320	179.2
	BovineHD0400001166	4.5	0.39	9  × 10–7*	cordon-bleu WH2 repeat (COBL)	109.2
	BovineHD0400001472	5.2	0.42	1  × 10–5	growth factor receptor bound 10 (GRB10)	0
11	Hapmap51367-BTA-99264	79.7	0.04	4.7 × 10–6	LOC112448881	273.9
Proportions of temperament scores greater than 1
14	ARS-BFGL-NGS-30322	47.1	0.16	4.6 × 10–6	solute carrier family 30 member 8 (SLC30A8)	0

¹ Bos taurus chromosome number.

²minor allele frequency.

³Distance to closest boundary of the candidate gene. A value of 0 indicates that the SNP was within the gene boundaries.

⁴Maximum lifetime temperament score of each cow (across lifetime records) at the age of earliest occurrence (some cows had more than one score that was the maximum value).

^*FDR < 0.05; all others: 0.05 < FDR < 0.2.

Correspondence of cow temperament with other traits

Cows from Cycles 1, 2, and 3 had temperament records as weaned calves (Riley et al., 2014). Unadjusted correlation coefficients of cow temperament score at parturition (average of all records per cow) paired with aggressiveness, nervousness, flightiness, gregariousness, and overall total scores at approximately at 8 mo of age (1 mo after weaning) using unadjusted records (Table 7) and residuals (not presented) were positive and of similar moderate magnitude (r = 0.2 to 0.3; P < 0.05). Estimated correlation coefficients of single values of cow temperament at parturition (maximum lifetime score, proportion > 1, or average temperament score) with cow productivity measures were positive and moderate (Table 8). The largest of those were for maximum lifetime temperament score with cow productivity traits (r = 0.23 to 0.37). Estimates of correlation coefficients for residuals of these traits were much lower in magnitude (not presented). Estimates of correlation coefficients of cow temperament score (not averaged; each record) with calf birth and weaning weight, cow weight and BCS were 0.14, 0.1, 0.18, and 0.03, respectively (P < 0.05). The estimate of the correlation coefficient for cow temperament with calf BCS at weaning did not differ from 0 (P = 0.8). Estimated correlation coefficients of cow temperament with the same traits using residuals (not presented) were of the same sign but 1/4 to 1/2 the magnitude of the estimates using unadjusted values.

Table 7.

Estimates of correlation coefficients (P < 0.05) of unadjusted temperament scores at the time of calving and unadjusted temperament traits when cows were approximately 8 mo of age¹

	Agg	Ner	Fli	Gre	Ovt
Maximum lifetime temperament score2	0.262	0.235	0.247	0.221	0.259
Proportions of temperament scores > 13	0.281	0.289	0.291	0.265	0.291
Temperament score average4	0.288	0.265	0.278	0.252	0.287

¹Agg = aggressiveness; Ner = nervousness; Fli = flightiness; Gre = gregariousness; Ovt = overall temperament; 1 to 9 score; average of 4 evaluators; higher scores indicate worse temperament (Riley et al., 2014).

²Maximum temperament score of each cow (across lifetime records) at the age of earliest occurrence (some cows had more than one score that was the maximum value).

³Proportion of cow temperament scores (across lifetime records) greater than 1.

⁴Cow temperament scores at parturition were averaged for estimation of correlation coefficients.

Table 8.

Estimates of correlation coefficients (P < 0.05) of unadjusted temperament scores and unadjusted cow productivity traits

	Calving rate	Weaning rate	WWCE1	WW1K1
Maximum lifetime temperament score2	0.232	0.309	0.365	0.235
Proportion of temperament scores > 13	0.094	0.223	0.245	0.155
Temperament score average	0.173	0.229	0.249	0.166

¹WWCE = weaning weight per cow exposed; WW1K = weaning weight per 454 kg (1,000 lb) cow weight.

²Maximum temperament score of each cow (across lifetime records) at the age of earliest occurrence (some cows had more than one score that was the maximum value).

³Proportion of cow temperament scores (across lifetime records) greater than 1.

Analyses of cow productivity traits were conducted to assess the effect of maximum lifetime cow temperament score at calving as a five-level fixed effect. Those cows in the maximum lifetime temperament score category 1 had lower (P < 0.05) calving and weaning rates than cows in the other maximum temperament score categories (Table 9). Although no maximum temperament category differences were observed in this significant nested effect (there were some differences between cycles) from analyses of weaning weight per 454 kg cow weight (Table 10), cows in three of the five cycles in the maximum temperament category 1 had lower mean weaning weight per cow exposed in the most recent breeding season than some of the other categories (Table 11).

Table 9.

Means for lifetime cow reproduction by maximum lifetime temperament score

Temperament score1	Calving rate	Weaning rate
1	0.78 ± 0.05a	0.64 ± 0.06a
2	0.83 ± 0.04b	0.80 ± 0.04b
3	0.88 ± 0.03b	0.83 ± 0.04b
4	0.89 ± 0.04b	0.87 ± 0.04b
5	0.88 ± 0.03b	0.83 ± 0.04b

¹Temperament score: 1 = totally docile, 2 = protective, but not aggressive, 3 = moderately aggressive when calf is disturbed, 4 = very aggressive when calf is disturbed, and 5 = very aggressive even when calf is not disturbed. Values represent the maximum of each cow across her lifetime scored at 1 d post calving.

^{a, b}Means in a column that do not share a superscript differ (P < 0.01).

Table 10.

Means for weaning weight per 454 kg cow weight (kg) by maximum lifetime temperament score

	Cycle2
Temperament score1	1	2	3	4	5
1	192 ± 10.8xy	163 ± 8.1x	210 ± 5.7x	224 ± 9.8x	232 ±  7.6x
2	193 ±  4.8xy	182 ± 5.2x	204 ± 4.8xy	213 ± 7.1y	216 ±  8.7y
3	196 ±  5.0xy	187 ± 5.2x	201 ± 5.1x	220 ± 6.2y	211 ± 11.2xy
4	192 ±  5.7xy	184 ± 6.0x	218 ± 7.1x	214 ± 7.8xy	176 ± 30.0xy
5	199 ±  3.8x	173 ± 5.7x	213 ± 5.9x	205 ± 9.6x

¹Temperament score: 1 = totally docile, 2 = protective, but not aggressive, 3 = moderately aggressive when calf is disturbed, 4 = very aggressive when calf is disturbed, and 5 = very aggressive even when calf is not disturbed. Maximum temperament score of each cow (across lifetime records) at the age of earliest occurrence (some cows had more than one score that was the maximum value).

²Cycles 1 and 2 were distinct populations of F₂ Nellore-Angus. Cycles 3 through 5 were females produced by inter se matings of F₂ (cycle 1 only), F₃, and F₄, respectively, bulls and cows.

^{x, y}Means in a row that do not share a common superscript differ (P < 0.05) after correction for multiple comparisons. No differences of maximum lifetime temperament category within cycles were observed after correction for multiple comparisons.

Table 11.

Means for weaning weight per cow exposed to bulls in the most recent breeding season (kg) by maximum lifetime temperament score

	Cycle2
Temperament score1	1	2	3	4	5
1	169 ± 9.8y	99 ± 10.9a, x	137 ± 7.5a, xy	111 ± 13.8a, x	139 ±  9.8xy
2	176 ± 4.2	154 ±  5.5b	160 ± 5.9ab	165 ±  8.2b	144 ± 12.0
3	177 ± 4.4	171 ±  5.1b	163 ± 6.0ab	168 ±  6.4b	152 ± 16.8
4	177 ± 5.2	176 ±  5.6b	166 ± 9.0ab	165 ±  7.7b	204 ± 45.1
5	173 ± 3.2	164 ±  5.8b	174 ± 5.9b	177 ± 10.3b

¹Temperament score: 1 = totally docile, 2 = protective, but not aggressive, 3 = moderately aggressive when calf is disturbed, 4 = very aggressive when calf is disturbed, and 5 = very aggressive even when calf is not disturbed. Maximum temperament score of each cow (across lifetime records) at the age of earliest occurrence (some cows had more than one score that was the maximum value).

²Cycles 1 and 2 were distinct populations of F₂ Nellore-Angus. Cycles 3 through 5 were females produced by inter se matings of F₂ (cycle 1 only), F₃, and F₄, respectively, bulls and cows.

^{a, b, c}Where present, means in a column that do not share a superscript differ (P < 0.05).

^{x, y}Where present, means in a row that do not share a common superscript differ (P < 0.05) after correction for multiple comparisons.

Discussion

The estimates of heritability for these cow temperament traits at the time of parturition were low. Buddenberg et al. (1986) reported an estimate of 0.06 for an 11-category scale as assessment of temperament in purebred cows of four Bos taurus breeds in Arkansas. Vallée et al. (2015) reported heritability of 0.19 for aggression at the time of parturition in Charolais cows. Estimates of heritability for temperament measured after weaning in the females of the present study and half-sibling males were much higher (0.4 to 0.5). Estimates of heritability for many temperament traits span almost the entire parameter space (e.g., 0.08 for flight speed; Burrow and Corbet, 2000; Silva et al., 2002; to 0.7 for flight distance, the “minimum tolerated approach distance of a handler to segregated bulls in a large yard”, Fordyce et al., 1996). Most estimates, however, are in the range from 0.15 to 0.45 (Burrow, 2001; Phocas et al., 2006; Alvarenga et al., 2022); these include many ways of assessing temperament at different time points. Alternatively, unfavorable heterosis for cattle temperament (Riley et al., 2007, 2010; Chase et al., 2017) supports the importance of nonadditive genetic variation. Although there was no active selection for temperament, a few bulls from different cycles were not used to produce calves in the subsequent cycle because of their aggressive temperament. A few cows were also removed because of their aggressiveness.

Accumulation of life experience appears to influence cow temperament more than the additive genetic contribution. The estimate of permanent environmental variance as a proportion of the phenotypic variance from repeated records analyses was larger than the estimate of heritability. This result appears consistent with temperament evaluations of (mostly) Bos taurus crossbred calves in Mississippi (Littlejohn et al., 2018) and in Brahman calves in East Texas (Schmidt et al., 2014). The significant maternal permanent environmental variance, although small as a proportion of the phenotypic variance, suggests that heifers acquire some of their dam’s temperament that is independent of inheritance. Torres-Vásquez and Spangler (2016) reported maternal permanent environmental variance as 0.008 of the phenotypic variance of chute score of weaned Hereford calves. However, Sims et al. (2023) reported no relationship of cow temperament after calving with the temperament measured later of their calves.

Omission of the maternal permanent environmental component resulted in an almost equal corresponding increase in heritability. Failure to model the maternal permanent environmental variance resulted in overestimation of heritability and could result in overestimation of selection response. The repeatability estimate (0.26) was relatively low, and of course suggests that additional records have substantial value for selection programs. Repeatability of cow temperament at parturition in Bos taurus cows ranged from 0.09 (Buddenberg et al., 1986) to 0.3 (Hoppe et al., 2008).

Genome-wide association

The protein produced by the candidate gene, cordon-bleu (COBL), for the most strongly associated SNP on BTA 4 (Table 5) has a documented construction role in neural tubes (Gasca et al., 1995; Carroll et al., 2003). Expression dynamics of COBL in neuronal cells was associated with branching of neurites (Ahuja et al., 2007; Kessels et al., 2011). Growth factor receptor binding protein 10 (GRB10) is a component of some cellular signaling mechanisms and has a regulatory (suppressive) role in β-cell activities in the pathogenesis of diabetes (Riedel, 2004; Zhang et al., 2012). Also relative to pathogenesis of diabetes, the gene solute carrier family 30 member 8 (SLC30A8) was characterized as an insulin support molecule (Chimienti et al., 2004; Tamaki et al., 2013).

Others reported associations of markers on BTA 4, 11, and 14 with bovine temperament. Kolbehdari et al. (2008) reported an association of temperament of Holstein cows with a more distal (~40 Mb, relative to associated SNP in the present study) region on BTA 4 and suggested neuronal cell adhesion molecule (NRCAM) as a candidate gene. Chen et al. (2021) identified markers on BTA 11 as associated with Holstein temperament. Brahman exit velocity was associated with SNP in the VRK serine/threonine kinase 2 (VRK2), FA complementation group L (FANCL) and transfer RNA cysteine (anticodon ACA) (TRNAC-ACA) genes (Paredes-Sánchez et al., 2020); all were more centric than the associated SNP in the present study (~13 to 35 Mb). A region (~25 Mb more distal) on BTA 14 was associated with Guzerat temperament with a proposed candidate gene KIAA1429 ortholog (KIAA1429) (Dos Santos et al., 2017). Alvarenga et al. (2023) reported more centric regions (~35 Mb) on BTA 14 as associated with Angus temperament; candidate genes included thymocyte selection associated high mobility group box (TOX), zinc finger and AT-hook domain containing (ZFAT), family with sequence similarity 135 member B (FAM135B), and potassium two pore domain channel subfamily K member 9 (KCNK9) (Alvarenga et al., 2023). Alvarenga et al. (2021) reviewed literature and detailed previously reported markers and candidate genes on BTA 14.

Age and other potential influences

The cow age effect was primarily supported by lower temperament scores of 2-yr-old cows, although there were slight differences between the age effect between cows of the different cycles. This may be because 2-yr-old cows have not fully developed a strong maternal protective instinct at their first parturition; a similar thought was discussed in the review of Nevard et al. (2022). Hoppe et al. (2008) observed a similar age effect in Bos taurus cows. Sandelin et al. (2005) condensed the 11-category cow temperament score of Buddenberg et al. (1986) to 4 categories. Although slightly different as descriptive categories from the present study, Sandelin et al. (2005) also reported less aggressive (“indifferent” or “apathetic”) temperament for the youngest (3-yr-old) cows in that work. Most of the Cycle 3 females were removed from the project after the 2015 production year. This is reflected in the larger SE for cows in this cycle in the 5- to 10-yr age category (Table 3). Cycle 5 cows have not had as many opportunities to be evaluated because of their age.

There may be other influences on cow temperament. Although inbreeding was avoided, it accumulated in this closed population; estimates of inbreeding from the genomic relationship matrix ranged from –0.02 for Cycle 1 (including males) to 0.03 in the F₅ cattle. Inbreeding levels were much higher in Cycles 4 and 5. To assess the effect of inbreeding, inbreeding coefficients based on pedigree and those from the diagonal elements of the genomic relationship matrix were fit as covariates in separate analyses in the repeated records model as described above. No regression coefficient, neither linear nor quadratic, across or within cycles, explained much trait variation (P > 0.42). It might be reasonable to consider that dystocia could negatively affect cow temperament. Only 16 of the 4,337 parturitions involved dystocia of any degree. To further assess this possibility, additional repeated records analyses were conducted that fit birth weight as a covariate. Four of the linear regression coefficient estimates within cycle (P = 0.03) were negative. The estimates for cycles 1, 2, and 3 were –0.01 (average SE 0.006), suggesting that lower cow temperament scores were associated with higher birth weights; cycles 4 (0.01 ± 0.02) and 5 (–0.05 ± 0.03) did not differ from 0. Birth weights of Bos taurus calves from cows that were more aggressive/attentive were generally higher than cows that were less aggressive, but a single difference between cow temperament categories (“very aggressive” and “indifferent”) was reported (Sims et al., 2023). It may be that heavier birth weights result in vigorous calves that prompt maternal protectiveness, as in the results of Sims et al. (2023), or the larger calf promotes some aspect of complacency and less aggressiveness, as in the present results. Some aspect of weather or temperature could be associated with cow temperament. Generally, temperatures become warmer across the February to April Central Texas calving season. Calving date was evaluated as linear and quadratic covariates in attempt to model the weather changes across the period. These regression coefficients did not explain significant variation in temperament score (P > 0.5).

Correspondence with other traits

Cow temperament after parturition was presumed to be a distinct trait from temperament measured earlier in life. Pre-calving measures of temperament did not correspond to various assessments of maternal temperament or behavior (Aitken, 2011). However, the correlation of cow temperament at the time of parturition was positive and moderate with the five assessments of calf temperament after weaning (Table 6). This relationship was not strong enough to consider that such a temperament evaluation would be a good representation of future cow temperament.

The most docile cows at the time of parturition had lower levels of productivity. Sandelin et al. (2005) reported higher calf survival from “very aggressive” cows. However, no differences in calf weaning weight by maternal temperament category were noted in Bos taurus cattle (Hoppe et al., 2008; Sims et al., 2023). Most reported research asserts that docile cattle temperament is associated with higher levels of productivity (Hoppe et al., 2010; Cziszter et al., 2016; Alvarenga et al., 2022); this is mostly for temperament of young or growing cattle measured in a variety of ways. Some of those may be related to excessive movement or activity, for example, in mature cows in the chute at the time of artificial insemination (Cooke et al., 2017). Objective temperament assessments may be influenced by the size or the condition of the animal; for example, in growing cattle, bigger, heavier animals may just be slower (Elzo et al., 2009). In our previous work with this research herd, there were minimal relationships between temperament at weaning with any production or beef product traits (Riley et al., 2019a, b, 2020; Baker et al., 2022). Even though there were positive correlations with temperament scores at weaning, cow temperament at the time of calving must be considered a distinct trait, and very different from a conventional response-to-humans type evaluation. Fear may always be influential on maternal behavior (Grandinson, 2005; Turner and Lawrence, 2007). It may be that threat to offspring provokes a stronger response than threat to self.

Glossary

Abbreviations

BCS

body condition score

BTA

Bos taurus chromosome

FDR

false discovery rate

SNP

single-nucleotide polymorphism

vcf

variant call format

Conflict of Interest Statement

None declared.