Using the difference in actual and expected calf liveweight relative to its dam liveweight as a statistic for interherd and intraherd benchmarking and genetic evaluations

Abstract

The importance of improving the efficiency of beef production systems using both genetic and management strategies has long been discussed. Despite the contribution of the mature beef herd to the overall cost of production in the sector as a whole, most strategies for improving (feed) efficiency have focused on the growing animal. The objective of the present study was to quantify the phenotypic and genetic variability in several novel measures that relate the weight of a calf to that of its dam and vice versa. Two novel residual traits, representing the deviation in calf weight relative to its expectation from the population based on its dam’s weight (DIFFcalf) or the deviation in the weight of the dam relative to its expectation from the population based on its calf’s weight (DIFFdam), were calculated while simultaneously accounting for some nuisance factors in a multiple regression model. Four supplementary traits were also calculated, namely, 1) the deviation in calf weight from its expectation expressed relative to the weight of the dam (DIFFcalf_ratio), 2) the deviation in dam weight from its expectation relative to the weight of the dam (DIFFdam_ratio), 3) DIFFcalf-DIFFdam, and 4) the simple ratio of calf weight to its dam’s weight (RATIOcalfdam). Genetic and residual variance components for each of the 6 traits were estimated using animal–dam linear mixed models. The phenotypic SD for DIFFcalf was 42 kg and, when expressed relative to the weight of the dam (i.e., DIFFcalf_ratio), was 0.07. The genetic SD for DIFFcalf and DIFFcalf_ratio was 16.66 kg and 0.02, respectively. The direct and maternal heritability estimated for DIFFcalf was 0.28 (SE = 0.04) and 0.11 (SE = 0.02), respectively, and for DIFFcalf_ratio was 0.24 (SE = 0.04) and 0.17 (SE = 0.03), respectively. The genetic SD for DIFFdam was 47.09 kg; the direct heritability was 0.50 (SE = 0.03), and the dam repeatability was 0.75 (SE = 0.01). The genetic SD for RATIOcalfdam was 0.03; the direct and maternal heritability was 0.24 (SE = 0.04) and 0.24 (SE = 0.03), respectively. The suggested traits outlined in the present study provide useful metrics for benchmarking dam–calf efficiency; in addition, the genetic variability detected in these traits suggest genetic progress for more efficient dam–calf pairs is indeed possible.

Introduction

The importance of improving the efficiency of beef production systems using both genetic and management approaches has been discussed for many decades (Koch et al., 1963; Archer et al., 1999; Berry and Crowley, 2013). Although noticeable improvements in the efficiency of the beef sector have been documented in the United States at least (Capper, 2011), consumers are demanding further and more accelerated improvement (Capper, 2012). The contribution of the mature beef herd to the overall cost of production in the beef sector is now well recognized (Montaño-Bermudez et al., 1990); despite this, most strategies for improving efficiency have focused on the growing animal (Berry and Crowley, 2013). The focus on the growing animal may be simply attributable to such a strategy being viewed as an easier approach because growing cattle are mainly confined to feedlots where performance can be relatively easily measured. Beef dams, on the other hand, generally reside outdoors grazing rangelands for the majority of the calendar year; obtaining a measurement of any type on such animals is usually resource intensive.

Calf-to-weaning beef production systems are generally pasture based (Stonehouse et al., 2003; McHugh et al., 2010; Capper, 2012) and thus synonymous with pasture-based dairy production systems. The main output in calf-to-weaning production systems, other than the cull cow, is the weaned calf. One of the main consumers of input in such systems is the feed intake requirements of the mature herd, which can be approximated from dam size; incidentally, dam intake will also affect the achievable stocking rate and thus output per hectare. In the present study, we propose several novel residual traits, similar to the concepts investigated by Byerly (1941) and Koch et al. (1963), which relate the weight of a calf to the mean of the population for a given dam weight (and vice versa); the objective of the study was to quantify the phenotypic and genetic variabilities in these statistics, thus providing insights into the potential for improvement. For comparison purposes, the weight of the calf relative to the weight of the dam, as advocated by Dinkel and Brown (1978), was also generated.

Materials and Methods

Data

A total of 1,473,238 liveweight records from 1,179,966 beef dams collected from 41,053 herds between the years 2007 and 2017, inclusive were extracted from the Irish Cattle Breeding Federation database. In addition, 1,857,715 calf liveweight observations from 55,222 herds between the years 2007 to 2017, inclusive were also available. Dams were defined as females that had calved at least once but were less than 12 yr of age when weighed (McHugh et al., 2011). Only dams weighing between 300 and 1,000 kg were retained. In the present study, calves were defined as singleton-born calves with a weight record taken between 3 and 9 mo of age; only records of calves weighing between 80 and 500 kg were retained. Only liveweight records relating to both the dam and calf were retained if both the dam and her calf were weighed within 14 d of each other on the same farm. This resulted in the retention of 25,955 dam–calf weight pairs.

Data were also available on dam parity, breed composition of all animals, and the ancestry information of each dam and calf. Records were discarded if the herd of birth for the calf was unknown, or the sire, dam, or maternal–grandsire of the calf were unknown. The breed proportions of both dam and calf for the 8 most common breeds (Aberdeen Angus, Belgian Blue, Charolais, Friesian, Hereford, Holstein, Limousin, and Simmental) were available. Dam parity was subsequently categorized as 1, 2, 3, 4, or ≥5. Heterosis and recombination loss coefficients were calculated for both the dam and calf as $1-\sum _{i-1}^x\mathrm{s}\mathrm{i}\mathrm{r}{\mathrm{e}}_i.\mathrm{d}\mathrm{a}{\mathrm{m}}_i$ (Van Raden, 1992) and $1-\sum _{i-1}^x\frac{{\mathrm{s}\mathrm{i}\mathrm{r}\mathrm{e}}_i^2+{\mathrm{d}\mathrm{a}\mathrm{m}}_i^2}{2}$ (Van Raden and Sanders, 2003), respectively, where sire_i and dam_i represent the proportion of breed i in the sire and dam, respectively. Heterosis and recombination loss coefficients were subsequently grouped into distinct classes based on the frequency distribution of the respective coefficients in the edited data set. Heterosis was categorized into 3 classes: <50%, 50% to 99%, and 100%. For recombination loss, 2 classes were formed: <25% and 25% to 50%.

Each dam–calf weight combination was allocated to a contemporary group based on herd-year-week of weighing; only records from contemporary groups with at least 5 records were retained. Following all edits, 11,254 dam–calf weights from 856 contemporary groups and 551 beef herds remained (Fig. 1).

Figure 1.

Number of calf–dam weight records by week of the calendar year.

Phenotype Creation

In addition to calf and dam weight, a number of novel residual phenotypes were created to represent the actual weight of the calf relative to expectation based on the weight of the dam at the same point in time, or vice versa. In the present study, DIFFcalf was defined as the difference in actual weight of the calf relative to its expectation based on the weight of the dam, estimated from a multiple linear regression model described later, fitted to the entire population. A second phenotype, here termed DIFFdam, was also defined from a multiple regression model as the difference between the actual weight of the dam relative to its expectation based on the weight of her calf. For the derivation of DIFFcalf, the following model was initially fitted:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{C}\mathrm{a}\mathrm{l}\mathrm{f}\mathrm{W}{\mathrm{T}}_i=\mu +\mathrm{C}{\mathrm{G}}_j+{\beta }_{DamWT}\beta \mathrm{D}\mathrm{a}\mathrm{m}\mathrm{W}{\mathrm{T}}_k+{\beta }_{Age}\beta \mathrm{A}\mathrm{g}{\mathrm{e}}_i+e_{ijk}}\end{array}}$$ (1)

where CalfWT_i = weight of calf i, µ = the population mean, CG_j = the fixed effect of contemporary group (j = 1,…, 856), DamWT_k = covariate representing the weight of dam k who is the dam of calf i and β_DamWT = the associated regression coefficient, and Age_i = the covariate of age of calf i, in days, when weighed relative to a set value of 200 d, β_Age = the associated regression coefficient, and e_ijk = random residual term.

For the derivation of DIFFdam the following model was initially fitted:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{D}\mathrm{a}\mathrm{m}\mathrm{W}{\mathrm{T}}_i=\mathrm{\mu }+\mathrm{C}{\mathrm{G}}_j+{\mathrm{\beta }}_{\mathrm{C}\mathrm{a}\mathrm{l}\mathrm{f}\mathrm{W}\mathrm{T}}\beta \mathrm{C}\mathrm{a}\mathrm{l}\mathrm{f}\mathrm{W}{\mathrm{T}}_k+{\mathrm{\beta }}_{\mathrm{D}\mathrm{a}\mathrm{y}\mathrm{s}}\beta \mathrm{D}\mathrm{a}\mathrm{y}{\mathrm{s}}_i+e_{ijk}}\end{array}}$$ (2)

where DamWT_i = weight of dam i, µ = the population mean, CG_j = the fixed effect of contemporary group (j = 1,…, 856), CalfWT_k = covariate representing the weight of calf k who was the calf of dam i, β_CalfWT = the associated regression coefficient, Days_i = the covariate of days postpartum for dam i relative to a set value of 200 d, β_Days = the associated regression coefficient, and e_ijk = random residual term.

DIFFcalf was subsequently derived using solutions for equation 1 as follows:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{D}\mathrm{I}\mathrm{F}\mathrm{F}\mathrm{c}\mathrm{a}\mathrm{l}{\mathrm{f}}_i=\mathrm{C}\mathrm{a}\mathrm{l}\mathrm{f}\mathrm{W}{\mathrm{T}}_i-[\mu +({\beta }_{\mathrm{D}\mathrm{a}\mathrm{m}\mathrm{W}\mathrm{T}}\beta \mathrm{D}\mathrm{a}\mathrm{m}\mathrm{W}{\mathrm{T}}_k)+({\beta }_{\mathrm{A}\mathrm{g}\mathrm{e}}\beta \mathrm{A}\mathrm{g}{\mathrm{e}}_i)]}\end{array}}$$ (3)

while DIFFdam was calculated using solutions for equation 2 as follows:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{D}\mathrm{I}\mathrm{F}\mathrm{F}\mathrm{d}\mathrm{a}{\mathrm{m}}_i=\mathrm{D}\mathrm{a}\mathrm{m}\mathrm{W}{\mathrm{T}}_i-[\mu +({\beta }_{\mathrm{C}\mathrm{a}\mathrm{l}\mathrm{f}\mathrm{W}\mathrm{T}}\beta \mathrm{C}\mathrm{a}\mathrm{l}\mathrm{f}\mathrm{W}{\mathrm{T}}_k)+({\beta }_{\mathrm{D}\mathrm{a}\mathrm{y}\mathrm{s}}\beta \mathrm{D}\mathrm{a}\mathrm{y}{\mathrm{s}}_i)]}\end{array}}$$ (4)

Three additional traits were also generated:

• (a) DIFFcalf expressed relative to the weight of the dam as follows:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{D}\mathrm{I}\mathrm{F}\mathrm{F}\mathrm{c}\mathrm{a}\mathrm{l}\mathrm{f}\mathrm{\_}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}{\mathrm{o}}_i=\frac{\mathrm{C}\mathrm{a}\mathrm{l}\mathrm{f}\mathrm{W}{\mathrm{T}}_i-[\mu +({\beta }_{\mathrm{D}\mathrm{a}\mathrm{m}\mathrm{W}\mathrm{T}}\beta \mathrm{D}\mathrm{a}\mathrm{m}\mathrm{W}{\mathrm{T}}_k)+({\beta }_{\mathrm{A}\mathrm{g}\mathrm{e}}\beta \mathrm{A}\mathrm{g}{\mathrm{e}}_i)]}{\mathrm{D}\mathrm{a}\mathrm{m}\mathrm{W}{\mathrm{T}}_k-({\beta }_{\mathrm{D}\mathrm{a}\mathrm{y}\mathrm{s}}\beta \mathrm{D}\mathrm{a}\mathrm{y}{\mathrm{s}}_k)}}\end{array}}$$

where DIFFcalf_ratio_i = the deviation in calf weight as a ratio of dam weight for calf i, CalfWT_i = the phenotypic calf weight, µ = the population mean for calf weight, β_DamWT*DamWT_k = the estimated model regression coefficient on DamWT times the respective DamWT of dam k (from equation 1), β_Age*Age_i = the age model solution times the respective age of calf i relative to 200 d (from equation 1), DamWT_k = the phenotypic weight of dam k (i.e., dam of calf i), and β_Days*Days_k = the days postpartum model solution times the respective days relative to 200 d postpartum (from equation 2).

• (b) DIFFdam expressed relative to the weight of the dam as follows:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{D}\mathrm{I}\mathrm{F}\mathrm{F}\mathrm{d}\mathrm{a}\mathrm{m}\mathrm{\_}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}{\mathrm{o}}_i=\frac{\mathrm{D}\mathrm{a}\mathrm{m}\mathrm{W}{\mathrm{T}}_{\mathrm{i}}-[\mu +({\beta }_{\mathrm{C}\mathrm{a}\mathrm{l}\mathrm{f}\mathrm{W}\mathrm{T}}\beta \mathrm{C}\mathrm{a}\mathrm{l}\mathrm{f}\mathrm{W}{\mathrm{T}}_k)+({\beta }_{\mathrm{D}\mathrm{a}\mathrm{y}\mathrm{s}}\beta \mathrm{D}\mathrm{a}\mathrm{y}{\mathrm{s}}_i)]}{\mathrm{D}\mathrm{a}\mathrm{m}\mathrm{W}{\mathrm{T}}_i-({\beta }_{\mathrm{D}\mathrm{a}\mathrm{y}\mathrm{s}}\beta \mathrm{D}\mathrm{a}\mathrm{y}{\mathrm{s}}_i)}}\end{array}}$$

where DIFFdam_ratio_i = the deviation in dam weight (i.e., DIFFdam) as a ratio of dam weight, DamWT_i = the phenotypic weight of dam i, µ = the population mean for dam weight, β_CalfWT*CalfWT_k = the estimated model regression coefficient for CalfWT times the respective CalfWT (from equation 2), and β_Days*Days_i = the days postpartum model solution times the respective days relative to 200 d (from equation 2).

• (c) The standardized difference between the deviation in calf weight and dam weight defined as follows:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{D}\mathrm{I}\mathrm{F}\mathrm{F}\mathrm{c}\mathrm{a}\mathrm{l}\mathrm{f}\mathrm{d}\mathrm{a}{\mathrm{m}}_i=\frac{\mathrm{D}\mathrm{I}\mathrm{F}\mathrm{F}\mathrm{c}\mathrm{a}\mathrm{l}{\mathrm{f}}_i}{\mathrm{S}\mathrm{D}(\mathrm{D}\mathrm{I}\mathrm{F}\mathrm{F}\mathrm{c}\mathrm{a}\mathrm{l}\mathrm{f})}-\frac{\mathrm{D}\mathrm{I}\mathrm{F}\mathrm{F}\mathrm{d}\mathrm{a}{\mathrm{m}}_i}{\mathrm{S}\mathrm{D}(\mathrm{D}\mathrm{I}\mathrm{F}\mathrm{F}\mathrm{d}\mathrm{a}\mathrm{m})}}\end{array}}$$

where DIFFcalfdam_i = the standardized difference between the deviation in calf weight and dam weight, DIFFcalf_i = the deviation in calf weight (from equation 3), SD(DIFFcalf) = is the SD of DIFFcalf in the entire population, DIFFdam_i = the deviation in dam weight (from equation 4) and SD(DIFFdam) = is the SD of DIFFdam in the entire population, and

• (d) The ratio of calf to dam weight (RATIOcalfdam) defined as follows:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{R}\mathrm{A}\mathrm{T}\mathrm{I}\mathrm{O}\mathrm{c}\mathrm{a}\mathrm{l}\mathrm{f}\mathrm{d}\mathrm{a}{\mathrm{m}}_i=\frac{\mathrm{C}\mathrm{a}\mathrm{l}\mathrm{f}\mathrm{W}{\mathrm{T}}_i-({\beta }_{\mathrm{A}\mathrm{g}\mathrm{e}}\beta \mathrm{A}\mathrm{g}{\mathrm{e}}_i)}{\mathrm{D}\mathrm{a}\mathrm{m}\mathrm{W}{\mathrm{T}}_k-({\beta }_{\mathrm{D}\mathrm{a}\mathrm{y}\mathrm{s}}\beta \mathrm{D}\mathrm{a}\mathrm{y}{\mathrm{s}}_k)}}\end{array}}$$

where RATIOcalfdam_i = the ratio of calf to dam weight, CalfWT_i = the phenotypic calf weight, β_Age*Age_i = the age model solution times the respective Age relative to 200 d (from equation 1), DamWT_k = the phenotypic dam weight, and β_Days*Days_k = the days postpartum model solution times the respective days relative to 200 d (from equation 2). For comparison purposes, the ratio of raw calf weight relative to raw dam weight was also calculated. A modification to the deviation in the weight of the dam relative to its calf’s weight (i.e., DIFFdam) and expressed relative to the weight of the dam (DIFFdam_ratio) and the standardized difference between the deviation in calf weight and dam weight (i.e., DIFFcalfdam) was also investigated whereby cows that failed to produce a calf were also included in the analyses. In this scenario, cows with a known weight that failed to produce a calf for a given parity across the 551 herds retained in the complete analyses were included in the data set; the corresponding weight of the calf was assumed to be zero. The contemporary groups of herd-year-week of weighing were reformed for the entire data set to include the cows that failed to produce a calf. The final data set consisted of 14,350 dam–calf weights. Unless otherwise stated, all results pertain to the variables, which only included the cows that produced a live calf.

Statistical Analyses

Genetic and residual variance components for the calf traits were estimated in Asreml (Gilmour et al., 2009) using an animal–dam repeatability linear mixed model represented as follows:

$${\displaystyle \begin{array}{ll}{\displaystyle Y_{abcijklmno}=}& \mathrm{C}{\mathrm{G}}_i+\mathrm{P}\mathrm{a}\mathrm{r}\mathrm{i}\mathrm{t}{\mathrm{y}}_j+\mathrm{H}\mathrm{e}{\mathrm{t}}_k+\mathrm{R}\mathrm{e}{\mathrm{c}}_l+\mathrm{D}\mathrm{a}\mathrm{m}\mathrm{\_}\mathrm{H}\mathrm{e}{\mathrm{t}}_m+\mathrm{D}\mathrm{a}\mathrm{m}\mathrm{\_}\mathrm{R}\mathrm{e}{\mathrm{c}}_n\\ & +\mathrm{G}\mathrm{e}\mathrm{n}\mathrm{d}\mathrm{e}{\mathrm{r}}_o+\mathrm{D}\mathrm{i}\mathrm{r}\mathrm{e}\mathrm{c}{\mathrm{t}}_a+\mathrm{M}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{n}\mathrm{a}{\mathrm{l}}_b+\mathrm{M}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{n}\mathrm{a}\mathrm{l}\mathrm{P}{\mathrm{E}}_c+e_{abcijklmno}\end{array}}$$

where Y_ijklmnoprq = dependent variable of calf weight, DIFFcalf, DIFFcalf_ratio, DIFFcalfdam, and RATIOcalfdam, CG_i = the fixed class effect of contemporary group (i = 1,…, 856), Parity_j = the fixed class effect of dam parity (j = 1, 2, 3, 4, ≥5), Het_k = the fixed class effect of heterosis coefficient of the calf (k = <50%, ≥50% to ≤99%, and 100%), Rec_l = class effect of recombination loss coefficient of the calf (l = <25% and ≥25% to 50%), Dam_Het_m = the fixed class effect of heterosis coefficient of the dam (m = <50%, ≥50% to ≤99%, and 100%), Rec_n = class effect of recombination loss coefficient of the dam (n = <25% and ≥25% to 50%), Gender_o = the fixed class effect of the calf gender (o = Male or Female), Direct_a = the random calf direct additive genetic effect (a = 11,254); (N(0, A${\sigma }_a^2$)), Maternal_b = the random dam maternal genetic effect (b = 9,367); (N(0, A${\sigma }_b^2$)), MaternalPE_c = the random maternal dam permanent environmental effect (c = 9,367); (N(0, I${\sigma }_c^2$)), and e_ijklmnopqr = random residual term (N(0, I${\sigma }_e^2$)) where A is the numerator relationship matrix, I is the identity matrix, ${\sigma }_a^2$ is the direct additive genetic variance, ${\sigma }_b^2$ is the maternal genetic variance, and ${\sigma }_c^2$ is the maternal permanent environmental effect.

The genetic and residual components for the dam traits (i.e., dam weight, DIFFdam, and DIFFdam_ratio) were also estimated using the model already described except that the direct calf effect was removed from the model and the maternal dam effect was included as the direct effect and the maternal permanent environmental effect was included as the permanent environmental effect.

The pedigree of each animal was traced back to the founder population; breed groups were fitted via the pedigree. Breeding values for all deviation traits were estimated in Mix99 (Lidauer et al., 2015) using the statistical model already described and the variance components estimated within.

Results

The proportion of male and female calves was 0.502 and 0.498, respectively. The most common breeds represented in the study (in descending order) were Limousin, Charolais, Simmental, Holstein, Aberdeen Angus, and Hereford. Approximately 70% of the calves in the present study were composed entirely of beef bloodlines. Of the remaining animals, dairy breeds (Holstein and Friesian mainly) accounted for, on average, 25% of the breed fraction. The percentage of dams in parity 1, 2, 3, 4, and ≥5 was 24%, 19%, 16%, 13%, and 28%, respectively. The percentage of calves with a heterosis level of <50%, 50% to 99%, and 100% was 65%, 11%, and 24%, respectively; the corresponding values for the 2 recombination loss classes (<25% and 25% to 50%) was 71% and 29%. The proportion of cows that actually never had a calf for a given parity was 20%.

Deviation in Calf Weight

Average calf weight in the data set was 252 (SD = 72) kg with an average age at weighing of 182 (SD = 54) d; the average calf weight adjusted to 200 d was 272 kg (Fig. 2). The correlation between dam weight and calf weight was 0.28 (Table 1). The proportion of phenotypic variation in calf weight explained by dam weight, contemporary group, and calf age was 80%. The proportion of phenotypic variation in calf weight explained by just dam weight and calf age was 66%. The regression coefficient of calf weight on calf age was 1.09 ± 0.01 kg; the regression coefficient of calf weight on dam weight was 0.09 ± 0.004 kg. The phenotypic SD of DIFFcalf (i.e., the deviation in calf weight relative to its dam’s weight) was 42.31 kg. The phenotypic SD for DIFFcalf, when expressed as a ratio of dam weight (i.e., DIFFcalf_ratio), was 0.07. The phenotypic correlation between calf weight and DIFFcalf was 0.55, whereas the corresponding correlation between calf weight and DIFFcalf_ratio was 0.36 (Table 1).

Figure 2.

Mean daily dam and calf weight recorded from 90 to 300 d post-calving or birth.

Table 1.

Phenotypic correlations between calf weight, the deviation in calf weight (DIFFcalf), deviation in calf weight as a ratio of dam weight (DIFFcalf_ratio), dam weight, deviation in dam weight (DIFFdam), the deviation in dam weight as a ratio of dam weight (DIFFdam_ratio), the difference between the deviation in calf weight and dam weight (DIFFcalfdam), and finally the ratio of calf to dam weight (RATIOcalfdam)

	Calf weight	DIFFcalf	DIFFcalf_ratio	Dam weight	DIFFdam	DIFFdam_ratio	DIFFcalfdam
DIFFcalf	0.55
DIFFcalf_ratio	0.36	0.91
Dam weight	0.28	0.13	−0.24
DIFFdam	0.12	−0.07	−0.42	0.98
DIFFdam_ratio	0.00	−0.12	−0.46	0.93	0.97
DIFFcalfdam	0.11	0.46	0.73	−0.83	−0.92	−0.91
RATIOcalfdam	0.12	0.72	0.94	−0.55	−0.68	−0.71	0.89

Relative to first parity dams, fourth parity dams produced the heaviest DIFFcalf at 13.94 ± 1.13 kg (Table 2); female calves were, on average, 23.24 ± 0.60 kg and 0.03 ± 0.001 lighter (P < 0.001) compared with male calves for DIFFcalf and DIFFcalf_ratio, respectively. DIFFcalf and DIFFcalf_ratio differed by the heterosis class of the calf; calves with a heterosis class of 100% were, on average, 5.09 ± 1.20 kg and 0.01 ± 0.002 heavier (P < 0.001), for DIFFcalf and DIFFcalf_ratio, respectively, compared with calves with a heterosis class of <50%. Similarly, DIFFcalf and DIFFcalf_ratio differed by heterosis class of the dam; dams with a heterosis class of 100% produced calves that were, on average, 6.93 ± 1.60 kg and 0.01 ± 0.003 heavier (P < 0.001) for DIFFcalf and DIFFcalf_ratio, respectively, compared with dams with a heterosis class of <50%.

Table 2.

Dam parity regression coefficients (SE in parenthesis) for calf weight, deviation in calf weight (DIFFcalf), deviation in calf weight as a ratio of dam weight (DIFFcalf_ratio), dam weight (cow parity regression coefficients), deviation in dam weight (DIFFdam; cow parity regression coefficients), the deviation in dam weight as a ratio of dam weight (DIFFdam_ratio; cow parity regression coefficients), the difference between the deviation in calf weight and dam weight (DIFFcalf_dam), and finally the ratio of calf to dam weight (RATIOcalfdam)

Paritya	Calf weight, kg	DIFFcalf, kg	DIFFcalf_ratio	Dam weight, kg	DIFFdam, kg	DIFFdam_ratio	DIFFcalfdam, kg	RATIOcalfdam
2	8.11 (1.65)	10.60 (0.99)	0.004 (0.002)	53.85 (1.63)	46.64 (1.68)	0.06 (0.002)	−0.24 (0.03)	−0.01 (0.002)
3	13.92 (1.67)	13.54 (1.06)	−0.001 (0.002)	94.71 (1.80)	84.77 (1.85)	0.10 (0.002)	−0.56 (0.04)	−0.03 (0.002)
4	16.63 (1.70)	13.94 (1.13)	−0.004 (0.002)	113.70 (1.98)	102.60 (2.03)	0.12 (0.002)	−0.73 (0.04)	−0.03 (0.002)
≥5	15.15 (1.97)	13.10 (1.02)	−0.007 (0.002)	122.20 (1.95)	111.00 (1.97)	0.12 (0.002)	−0.82 (0.04)	−0.04 (0.002)

^aParity 1 was the reference category against which all other parities are compared.

The direct and maternal heritability estimated for DIFFcalf was 0.28 (SE = 0.04) and 0.11 (SE = 0.02), respectively (Table 3). For DIFFcalf_ratio, the direct and maternal heritability of 0.24 (SE = 0.04) and 0.17 (SE = 0.03), respectively. The dam repeatability estimated for both traits ranged from 0.20 to 0.30.

Table 3.

Phenotypic mean (µ) SD and range, direct and maternal genetic SD (σ_g) in kg, residual SD (σ_res) in kg, direct and maternal heritability (h²; SE in parentheses), and maternal repeatability (t_dam; SE in parentheses) for calf weight, deviation in calf weight (DIFFcalf), deviation in calf weight as a ratio of dam weight (DIFFcalf_ratio), dam weight, deviation in dam weight (DIFFdam), the deviation in dam weight as a ratio of dam weight (DIFFdam_ratio), the difference between the deviation in calf weight and dam weight (DIFFcalfdam) and finally the ratio of calf to dam weight (RATIOcalfdam)

Trait	µ (SD)	Range	σ g		σ res	h 2		t dam
			Direct	Maternal		Direct	Maternal
Calf weight, kg	252.18 (72.03)	80 to 500	17.66	9.73	21.86	0.32 (0.04)	0.10 (0.02)	0.19 (0.02)
DIFFcalf, kg	103.84 (42.31)	−93 to 339	16.66	10.29	22.64	0.28 (0.04)	0.11 (0.02)	0.20 (0.02)
DIFFcalf_ratio	0.16 (0.07)	−0.16 to 0.64	0.02	0.02	0.03	0.24 (0.04)	0.17 (0.03)	0.30 (0.02)
Dam weight, kg	680.42 (101.89)	338 to 996	48.23		30.77	0.52 (0.03)		0.78 (0.01)
DIFFdam, kg	198.73 (97.73)	−107 to 518	47.09		33.13	0.50 (0.03)		0.75 (0.01)
DIFFdam_ratio	0.28 (0.11)	−0.34 to 0.57	0.05		0.04	0.51 (0.03)		0.72 (0.01)
DIFFcalfdam, kg	0.40 (1.46)	−6.89 to 8.17	0.48	0.58	0.73	0.17 (0.03)	0.24 (0.03)	0.43 (0.02)
RATIOcalfdam	0.41 (0.08)	0.10 to 0.93	0.03	0.03	0.04	0.24 (0.04)	0.24 (0.03)	0.41 (0.02)

The distribution in estimated breeding values (EBVs) for DIFFcalf for sires of the 6 main breeds each with >5 calves is shown in Fig. 3. The mean DIFFcalf EBV for the sire with >5 calves for Aberdeen Angus (n = 106), Belgian Blue (n = 19), Charolais (n = 100), Hereford (n = 37), Limousin (n = 198), and Simmental (n = 55) was 33.68, 46.96, 59.04, 34.34, 42.81, and 55.68 kg, respectively. The average within breed correlation between DIFFcalf EBV and DIFFcalf_ratio EBV for the sire with >5 calves ranged from 0.87 (Simmental) to 0.95 (Hereford).

Figure 3.

Box and whisker plots depicting the estimated breeding values (EBVs) for (a) deviation in calf weight relative to its dam’s weight (DIFFcalf; direct and maternal EBVs; kg) and (b) deviation in the weight of the dam relative to the calf weight (DIFFdam; direct EBVs only; kg) for Aberdeen Angus (AA), Belgian Blue (BB), Charolais (CH), Hereford (HE), Limousin (LM), and Simmental (SI) sires.

Deviation in Dam Weight

Average dam weight in the edited data set was 680 (SD = 102) kg (Fig. 2). The proportion of phenotypic variation in dam weight explained by calf weight, days post-calving, and contemporary group was 42%; the proportion of phenotypic variation in dam weight explained by just calf weight and days post-calving was 11%. The regression coefficient of dam weight on calf weight was 0.47 ± 0.02 kg; the regression coefficient of dam weight on days postpartum −0.40 ± 0.03 kg. The phenotypic mean SD for DIFFdam (i.e., the deviation in dam weight) was 198.73 and 97 kg, respectively; when the contemporary group solution was included in the calculation of DIFFdam the corresponding phenotypic mean was 0 kg. The phenotypic SD of DIFFdam, when expressed as a ratio of dam weight (DIFFdam_ratio) was 0.11 (Table 3). Near unity phenotypic correlations existed among all of dam weight, DIFFdam, and DIFFdam_ratio (Table 1).

Relative to first parity dams, the greatest DIFFdam and DIFFdam_ratio were recorded in fifth parity dams at 111.00 ± 1.97 kg and 0.12 ± 0.002, respectively. Greater DIFFdam and DIFFdam_ratio were observed in female progeny (12.85 ± 1.08 kg and 0.02 ± 0.001, respectively) relative to male progeny. DIFFdam and DIFFdam_ratio differed by dam heterosis class with the greatest DIFFdam and DIFFdam_ratio observed in dams with a heterosis class of <50%. DIFFdam and DIFFdam_ratio also differed by dam recombination class, with greater weights associated with greater levels of dam recombination (P < 0.001).

The genetic SD and direct heritability for dam weight was 48.23 kg and 0.52 (SE = 0.03), respectively (Table 3). The direct heritability for DIFFdam was 0.50 (SE = 0.03), whereas the dam repeatability was 0.75 (SE = 0.01). The direct heritability calculated for the deviation in DIFFdam as a ratio of dam weight (DIFFdam_ratio) 0.51 (SE = 0.03; Table 3). The genetic SD for DIFFdam and DIFFdam_ratio when cows that failed to produce a calf were included in the analyses was 46.18 and 45.56 kg, respectively; the corresponding direct heritability for DIFFdam and DIFFdam_ratio was 0.44 (SE = 0.03) and 0.47 (SE = 0.03), respectively.

The breed average EBVs for DIFFdam for sires with >5 calves ranged from 12.55 kg (Limousin) to 96.42 kg (Charolais; Fig. 3). The average within breed correlation between DIFFdam EBV and DIFFdam_ratio EBV for sires with >5 calves ranged from 0.94 (Charolais) to 0.98 (Simmental).

Deviation in Calf–Dam Weight

Summary statistics including estimates of genetic parameters for traits related to the deviation in calf–dam weight are in Table 3. The average ratio of calf to dam weight (i.e., RATIOcalfdam) was 0.41; the phenotypic SD for RATIOcalfdam was 0.08 (Table 3). The correlation between RATIOcalfdam as defined in the present study (i.e., with adjustment for differences in age at measure) and the simple ratio of the calf weight to dam weight was 0.41. The phenotypic SD for the standardized difference between the deviation in calf weight and dam weight, (i.e., DIFFcalfdam) was 0.40 kg. The greatest DIFFcalfdam and RATIOcalfdam were recorded for first parity dams; relative to a first parity dam, the DIFFcalfdam and RATIOcalfdam recorded for fifth parity dams was −0.82 ± 0.04 kg and −0.04 ± 0.002, respectively (Table 2). Male calves had, on average, a 0.68 ± 0.02 kg greater DIFFcalfdam and 0.04 ± 0.001 heavier RATIOcalfdam than their female counterparts.

DIFFcalfdam and RATIOcalfdam differed by the heterosis class of the calf; calves with a heterosis class of 100% were, on average, 0.24 ± 0.04 kg and 0.01 ± 0.002 heavier (P < 0.001), respectively, compared with calves with a heterosis class of <50%. DIFFcalfdam and RATIOcalfdam also differed by heterosis class of the dam; relative to dams with a heterosis class of <50%, dams with a heterosis class of 100% were, on average, 0.21 ± 0.06 kg and 0.01 ± 0.003 heavier (P < 0.001), respectively. Calves with greater levels of recombination had greater DIFFcalfdam and RATIOcalfdam (P < 0.001).

The direct and maternal heritability of RATIOcalfdam were 0.24 (SE = 0.04) and 0.24 (SE = 0.03), respectively (Table 3). The direct genetic SD for DIFFcalfdam was 0.48 kg. The direct and maternal heritability estimates for DIFFcalfdam was 0.17 (SE = 0.03) and 0.24 (SE = 0.03), respectively. The direct and maternal heritability for DIFFcalfdam when cows that failed to produce a calf were included in the analyses was 0.42 (SE = 0.02) and 0.16 (SE = 0.02), respectively.

The average DIFFcalfdam EBV for sires with >5 calves for Aberdeen Angus, Belgian Blue, Charolais, Hereford, Limousin, and Simmental was −143.97, −153.76, −138.03, −165.55, −141.43, and −133.52 kg, respectively. The average within breed correlation between DIFFcalfdam EBV and RATIOcalfdam EBV for sires with >5 calves ranged from 0.83 (Simmental) to 0.88 (Hereford).

Discussion

Gains in performance are a function of the coevolution of genetic gain and animal management (Berry, 2018). Although the relative contribution of genetic and management factors to performance differs per trait, both disciplines warrant consideration to sustainably accelerate gains. The novel traits described in the present study have the potential to be used for the monitoring of both genetic and management trends. Moreover, decomposing the variability of such metrics into genetic and nongenetic factors helps to inform the necessary emphasis that should be placed on different strategies for improving efficiency.

Practical Implementation and Usefulness for Intra- and Interherd Benchmarking of the Novel Efficiency Measures in Global Beef Production Systems

Motivation for improvement is often prompted by knowledge of performance relative to peers; this form of benchmarking requires a metric from which the benchmark is based on. For such benchmarking to be successful, however, 1) the metric must be relevant, and ideally relatively easily measured and calculated, 2) knowledge must exist to help explain why an individual is achieving suboptimal performance, but also 3) the tools must exist to facilitate continuous improvement. Although the contribution of the maintenance requirements of the mature beef herd to the production cost of the entire beef sector is well established (Montañdo-Bermudez et al., 1990), any tool to improve efficiency of the mature herd must be measurable and implementable; such a prerequisite would therefore preclude individual mature animal feed intake as a constituent of such a metric, at least for the foreseeable future.

Although measuring individual cow liveweight can be resource intensive, all beef cows and calves generally need to be herded at least once annually for weaning. When separating the calves from the cows, it may not be overly cumbersome to concurrently weigh all animals; simple low-cost genotyping technologies (Boichard et al., 2012; Judge et al., 2016) could be used to assign parentage if not already available. Although differences in age of the calves are likely to exist on a set calendar date, assuming all bulls were mob-mated at the same time, mean differences in calf age per sire (which should be random) are expected to be minimal. If all cows were equally exposed to the bulls, then any big differences in calf age per cow may largely reflect differences in the inherent reproductive performance of the cow which in itself is a contributing factor to the overall efficiency of the cow; slight differences in calf age are nonetheless inevitable owing to the stage of the estrus cycle the cow was at when joined with the bull(s).

One modification to the metrics proposed in the present study included a penalty for cows that did not produce a calf. In this scenario, cows with a known liveweight but failed to produce a calf were also included in the analyses; the corresponding weight of the calf was set to zero. In reality, however, producers are unlikely to retain cows that are not in calf or did not produce a live calf, thus limiting the usefulness of such a metric phenotypically. In genetic evaluations, arguably a more efficient approach to penalize such cows would be through a multitrait breeding objective that, as well as considering the traits included in the present study, also include a binary trait to reflect whether or not the cow produced a calf; this approach is analogous to how days open or calving interval is treated in genetic evaluations with cows that failed to establish pregnancy being penalized through some measure of pregnant or not. The contribution of differences in BCS to differences in liveweight in beef cattle has been previously documented (Drennan and Berry, 2006), as has the association between lower BCS and compromised reproductive performance (Drennan and Berry, 2006). Ideally, the BCS of the cows should be assessed during weighing and adjusted for in the derivation of the different metrics; no BCS data were available on the animals included in the present study.

Once all data are compiled, the mean performance of the herd can be compared with similar farms; model solutions derived in the present study for the different associated factors (e.g., mean cow parity) could be used to help explain differences among farms. For example, some herds may have lighter calves and some of this could be due to a younger herd (inferred from the results of McHugh et al., 2010); a younger herd may be due to that herd going through expansion or due to a greater replacement rate. Differences in mean calf age, approximated from the date of bull joining, could also explain differences in mean herd performance.

Importantly, the value of each kg liveweight of the calf can be used to derive a monetary value of differences in DIFFcalf (i.e., the deviation in calf weight relative to its expectation based on it dam’s weight) among farms but for the same expected cost of feeding the mature cow herd; in direct contrast, the cost of feed can be used to ascribe a monetary value to differences in DIFFdam (i.e., the deviation in the weight of the dam relative to its expectation based on its calf’s weight) per herd for the same calf weight. The extent of inter-animal variability within farm, as observed in the present study, provides concrete evidence of the potential that exists, even within farm, to improve the mean herd statistic, and the value of any improvement can be readily quantified. Assuming the phenotypic SD of 42.31 kg for DIFFcalf estimated in the present study is reflective of the within-herd variability, then the difference in DIFFcalf between the average of the herd and mean of the top decile is expected to be 74.26 kg (i.e., 42.31 kg times 1.7551); hence, considerable scope exists for improvement. The fact that this is heritable (Table 3) implies that this metric could be improved on generation-by-generation with an appropriate breeding program.

Practical Implementation and Usefulness of the Novel Efficiency Measures in Beef Breeding Programs

The notion of using residual-based traits as a strategy to improve performance or efficiency in cattle is not novel (Berry and Crowley, 2013); arguably the most familiar of such traits is residual feed intake (Koch et al., 1963). The merits and demerits of actually using such traits in a breeding program have been discussed at length (Berry and Crowley, 2013). Nonetheless, the traits in themselves are informative, especially for research purposes, but also can have potential practical uses.

First, knowledge of the actual genetic variability in these traits demonstrates the potential, or lack thereof, for genetic improvement, whatever the eventual strategy pursued. The genetic SD of 16.66 kg estimated in the present study for DIFFcalf clearly reveals the considerable exploitable genetic variability that exists. The difference in mean genetic merit between the top and bottom 10% of individual is expected to be 58.5 kg.

Second, the trait itself could be useful as a stand-alone trait with an associated expected progeny difference (EPD) for each animal. Although including both cow weight and calf weight in an index could be mathematically equivalent to including DIFFcalf (Kennedy et al., 1993), it is not always easy for the end user to compare two (or more animals) purely on efficiency using such an approach; this is because such traits are a function of the relative difference in the EPD for calf weight and for cow weight. Although independent culling levels are known to be generally less efficient that selection on a selection index (Hazel and Lush, 1942), individual breeders and producers have their own preferences; a stand-alone deviation trait, as described in the present study, can arguably be more easily interpreted, especially if an EPD of zero is chosen to represent the average. Although DIFFcalf identifies animals that produce heavier calves for the same cow liveweight (i.e., no expected reduction in input required for the mature herd), and DIFFdam favors lighter cows for the same calf weight (i.e., expected benefit in input required for the mature herd but with no expected difference in calf liveweight), DIFFcalfdam exploits the favorable attributes of both metrics; DIFFcalfdam favors lighter cows with heavier calves and is analogous to residual intake and gain defined by Berry and Crowley (2012) who combined the benefits of residual feed intake with residual daily gain in growing cattle. Plotting genetic gain of such traits over time can also be useful in informing the impact of prevailing breeding programs on efficiency.

Although the traits proposed in the present study were derived using phenotypic regression, genetic regression using available EPDs is also possible. Not alone will this ensure genetic independence between, for example, DIFFcalf with cow liveweight, or between DIFFdam with calf liveweight, but it could also facilitate easier implementation. Phenotypic derivation of the traits requires that both cow liveweight and calf liveweight be recorded in relatively close proximity in time; moreover, as alluded to previously, ideally BCS should also be assessed to avoid any unfair advantage to lower BCS cows that, in turn, are, on average, actually less efficient in the long term. Genetic evaluation models generally account for differences in systematic environmental effects (e.g., calf age, days post-calving) when estimating EPDs, implying that the measurement of cow and calf liveweight do not need to be so synchronized; moreover, by exploiting genetic relationships among animals, the cows with weight records need not necessarily be the dams of the calves with weight records. Furthermore, the inclusion of traits associated with reproductive performance or animal health in an overall breeding objective could mitigate the effect of such a selection strategy on the health and fertility of the cow through a reduction of BCS. Some breeding programs (e.g., Crowley et al., 2011; McHugh et al., 2011) have information on carcass fat of animals, including cull cows, and incorporation of the EPDs of such traits could also be used in the multiple regression for deriving the deviation traits.

In conclusion, a novel selection of traits has been proposed. Such metrics can be useful to compare animals within a herd or even compare contemporary herds. Although inclusion of the individual traits themselves in an overall breeding objective is mathematically equivalent to including the proposed individual traits, the proposed traits still have merit as stand-alone traits to more easily judge the efficiency of each animal or herd.