J. Dairy Sci. 2007. 90:4835-4845. doi:10.3168/jds.2007-0116
© 2007 American Dairy Science Association ®
Genetics of Grass Dry Matter Intake, Energy Balance, and Digestibility in Grazing Irish Dairy Cows
D. P. Berry1,
B. Horan,
M. O’Donovan,
F. Buckley,
E. Kennedy,
M. McEvoy and
P. Dillon
Teagasc, Moorepark Dairy Production Research Centre, Fermoy, Co. Cork, Ireland
1 Corresponding author: donagh.berry{at}teagasc.ie
 |
ABSTRACT
|
|---|
The objective of this study was to estimate genetic parameters for grass dry matter intake (DMI), energy balance (EB), and cow internal digestibility (IDG) in grazing Holstein-Friesian dairy cows. Grass DMI was estimated up to 4 times per lactation on 1,588 lactations from 755 cows on 2 research farms in southern Ireland. Simultaneously measured milk production and BW records were used to calculate EB. Cow IDG, measured as the ratio of feed and fecal concentrations of the natural odd carbon-chain n-alkane pentatriacontane, was available on 583 lactations from 238 cows. Random regression and multitrait animal models were used to estimate residual, additive genetic and permanent environmental (co)variances across lactations. Results were similar for both models. Heritability for DMI, EB, and IDG across lactation varied from 0.10 [8 days in milk (DIM)] to 0.30 (169 DIM), from 0.06 (29 DIM) to 0.29 (305 DIM), and from 0.08 (50 DIM) to 0.45 (305 DIM), respectively, when estimated using the random regression model. Genetic correlations within each trait tended to decrease as the interval between periods compared increased for DMI and EB, whereas the correlations with IDG in early lactation were weakest when measured midlactation. The lowest correlation between any 2 periods was 0.10, –0.36, and –0.04 for DMI, EB, and IDG, respectively, suggesting the effect of different genes at different stages of lactations. Eigenvalues and associated eigenfunctions of the additive genetic covariance matrix revealed considerable genetic variation among animals in the shape of the lactation profiles for DMI, EB, and IDG. Genetic parameters presented are the first estimates from dairy cows fed predominantly grazed grass and imply that genetic improvement in DMI, EB, and IDG in Holstein-Friesian cows fed predominantly grazed grass is possible.
Key Words: dry matter intake • energy balance • genetic parameter • grass
|
INTRODUCTION
|
|---|
Most modern breeding objectives in dairy cattle include traits related to milk production, fertility, health, survival, calving performance, conformation, and beef merit (Miglior et al., 2005). However, despite its large contribution to the variable costs of milk production (Dillon et al., 2007), feed intake is not directly included in international dairy breeding objectives, although genetic responses in feed intake are expected through the inclusion of correlated traits such as milk production (van Arendonk et al., 1991) or conformation (Veerkamp and Brotherstone, 1997) in breeding objectives. Applying a positive weighting to DMI, or a correlated trait, in a breeding objective will enhance the rate of genetic response in DMI (Veerkamp, 1998). However, to derive optimal selection indices, accurate genetic parameters (as well as economic weights) are required.
Previous studies have reported heritability estimates ranging from 0.13 to 0.54 for DMI (Koenen and Veerkamp, 1998; Veerkamp, 1998; Veerkamp and Thompson, 1999), although genetic variation and heritability in DMI has been shown to vary with stage of lactation (Koenen and Veerkamp, 1998; Veerkamp and Thompson, 1999; Coffey et al., 2001). This suggests that if measurable, or correlated to an easily measured trait, genetic selection for increased DMI will be worthwhile. However, to date, all estimates of genetic parameters have originated from cows fed indoors predominantly on TMR or ensiled forages, and no study has attempted to estimate genetic parameters for grass DMI in grazing dairy cows.
Energy balance (EB), of which DMI is a component trait, has been associated with impaired fertility both phenotypically and genetically (Veerkamp et al., 2000). However, there are only a few heritability estimates for EB (Svendsen et al., 1994; Veerkamp et al., 2000); Veerkamp et al. (2000) reported a heritability estimate of 0.33 in primiparous Holstein dairy cows, whereas Banos et al. (2006) reported heritability estimates for a related trait, body energy content, ranging from 0.46 to 0.88. A related trait, residual feed intake, has also been shown to exhibit genetic variation, with heritability estimates ranging from 0.14 to 0.38 (van Arendonk et al., 1991; Veerkamp et al., 1995). Additionally, because of the change in genetic variation throughout lactation in milk yield, BW, and DMI (Koenen and Veerkamp, 1998; Veerkamp and Thompson, 1999; Berry et al., 2003), the main components of EB, change in genetic variation in EB across lactation is also expected. However, no estimates of genetic variation in EB have been reported from cows fed predominantly grazed grass, although significant differences in EB among genotypes of Holstein-Friesians on basal grass-based diets have been reported (Horan et al., 2006).
There is a lack of consensus of whether genetic variation among dairy cows exists in their ability to digest a given diet. Grieve et al. (1976) and Davey et al. (1983) both failed to identify any significant difference in digestive ability between groups of Holstein and Friesian cows, respectively, genetically selected for divergent milk yield. This was further substantiated by Kennedy et al. (2003) comparing contrasting genotypes of Holstein-Friesian dairy cows fed predominantly grazed grass. In direct contrast, Trigg and Parr (1981) using calorimetric data from 6 Jersey cows of contrasting genetic merit, found a greater partitioning of gross energy to digested energy in early lactation in high genetic merit animals, although the effect was not significant in midlactation.
The objective, therefore, of this study was to quantify the degree and change in genetic (co)variance across lactation in grass DMI, EB, and internal digestibility (IDG) in grazing Holstein-Friesian cows. Results from this study will be useful in determining the feasibility of identifying animals of divergent genetic potential for DMI and EB on a predominantly grass-based diet as well as potentially identifying animals that have a greater genetic ability to digest feed.
|
MATERIALS AND METHODS
|
|---|
The data used in the present study were collated from several studies (Buckley et al., 2000; Kennedy et al., 2003; O’Donovan and Delaby, 2005; Horan et al., 2006; Kennedy et al., 2006; McCarthy et al., 2007; McEvoy et al., 2007) that compared either alternative genotypes of Holstein-Friesian dairy cattle on different production systems or alternative grazing strategies or grass varieties. All studies were undertaken during an 11-yr period from 1996 through 2006 on 2 nearby research farms, namely, Curtins Research Farm and Moorepark Research Farm, both in southern Ireland (latitude 52°9', longitude 8°16'). A total of 1,609 lactations from 765 Holstein-Friesian cows were available for inclusion in the analyses.
Research Farms
Perennial ryegrass (Lolium perenne) was the predominant pasture species at both research farms. At Curtins Research Farm, alternative strains of Holstein-Friesians were evaluated on contrasting grass-based production systems (Buckley et al., 2000; Kennedy et al., 2003; Horan et al., 2006; McCarthy et al., 2007). Animals within strain, at the start of lactation, were randomly assigned to feed systems differing in stocking rate, concentrate input, or both. Annual concentrate feeding level across studies varied from 325 to 1,452 kg per cow.
A rotational grazing management system was operated at Curtins Research Farm and was facilitated by weekly monitoring of farm grass cover. Residency time in each paddock was determined by the achievement of predetermined pasture allowances within a target postgrazing sward surface height. Target postgrazing sward surface heights were reached in residency times that generally ranged from 1.5 to 2.5 d/subpaddock over the experimental period.
Studies included in the present study from Moorepark Research Farm involved investigation of the effect of alternative grass varieties and grazing management strategies on overall animal performance. O’Donovan and Delaby (2005) compared animal and sward performance in a 2 x 2 x 2 factorial study design of grass cultivars (diploid or tetraploid), heading dates (intermediate or late), and stocking rate (high or low). Total amount of concentrates offered annually ranged from 113 to 164 kg/cow. Kennedy et al. (2006) compared the effect of initial spring grazing date and stocking rate on grazing management and animal performance, whereas McEvoy et al. (2007) compared the effect of concentrate feeding level and daily herbage allowance on animal performance, both using a randomized block design. Animals on the latter 2 experiments were offered up to 6 kg of concentrate/day during the study period.
Animal Measures
Calving assistance for all lactations was scored on a scale of 1 to 4, for normal calving, some assistance, serious calving difficulty, and veterinary assisted, respectively. Individual milk yields were recorded daily, whereas milk fat, protein, and lactose concentrations were determined from successive evening and morning milk samples once per week using a Fos-let instrument (AS/N Foss Electric, Hillerød, Denmark). Body weight of each animal was recorded weekly using portable weighing scales and the Winweigh software package (Tru-Test Ltd., Auckland, New Zealand). The scales were calibrated weekly against known weights. Body condition score was recorded approximately every 3 wk during the lactation on a 1 to 5 scale (1 = emaciated, 5 = extremely fat) in increments of 0.25 as outlined by Edmonson et al. (1989).
Individual animal DMI was measured while at pasture when the diet consisted of exclusively pasture or pasture plus concentrate depending on feed system. Individual animal grass DMI was estimated using the n-alkane technique (Mayes et al., 1986) as modified by Dillon (1993). During each intake period, the cows were dosed twice daily (after milking) for a 12-d period with paper filters or bungs (Carl Roth GmbH and Co. KG, Karlesruhe, Germany) containing approximately 500 mg of C32 (dotriacontane) each.
Fecal grab samples were collected twice daily from each cow immediately before or after milking in the last 6 d. The fecal samples from each cow for each 6-d period were bulked for analysis. Herbage samples were collected manually to represent herbage grazed (following close observation of the grazing animal) after both the morning and evening milking on d 6 to 11 of each measurement period. The ratio of herbage C33 (tritria-contane) to dosed C32 was used to estimate DMI. The n-alkane concentration of the dosed pellets, feces, herbage, and concentrate was determined as described by Dillon (1993).
Estimates of grass DMI were calculated as follows:
where Fi, Ci, and Hi = the concentrations (mg/kg of DM) of the natural odd-chain n-alkanes in feces, concentrate, and pasture, respectively; Fj, Cj, and Hj = the concentrations (mg/kg of DM) of the even-chain n-alkane in feces, concentrate, and pasture, respectively; Dj = the dose rate (mg/d) of the even-chain n-alkane; and Ic = the daily concentrate intake (kg of DM/d).
Estimates of diet IDG were calculated, for a subset of the data set, from the ratio of feed and fecal concentrations of the natural odd-chain n-alkane C35 as follows:
where H and F = the concentrations (mg/kg of DM) of the natural C35 alkane in feed and feces, respectively, and Ri = the recovery rate of C35 from feces. The recovery of C35 was assumed to be 0.90, based on the results reported for dairy cows fed pasture (Dillon, 1993).
Data Editing
A total of 5,195 DMI test-day observations were available for inclusion in the analyses. Only intake measures in the first 305 d of lactation were retained, after which 5,118 records remained. Average milk yield and composition during the week of each intake measure were retained. The BCS and BW record nearest to the week of intake measure (but within 2 wk) was retained. Reproductive information was used to determine the stage of gestation at intake measurement for each animal for a subset of the animals where the data were available. Energy balance for each test day was calculated in accordance with the net energy system outlined by O’Mara (2000); the equations are also provided by Berry et al. (2006).
The variable herd test day was generated by concatenating the feed system the animal was exposed to at the time of intake measurement and the date of intake measurement. Age was nested within parity, and animals calving greater than 200 d from the median age within parity were discarded (n = 7). Adjacent weeks of the year were merged, and the period of the year at calving for each lactation was determined when January 1 of each year was assumed to be the beginning of the first week; animals calving after the 20th week (i.e., May 21) were removed. Furthermore, only animals that had a known sire or dam were retained. Following all edits, 5,050 records for grass DMI and 5,017 records for EB, from 1,588 lactations on 755 cows, were available for inclusion in the analyses. A total of 2,044 records for IDG from 583 lactations on 238 cows were also available. No records were available before 8 DIM. Pedigree information 4 generations deep was collated. A total of 3,191 nonfounder animals were included in the pedigree file.
Data Analysis
Grass DMI, EB, and IDG were normally distributed. The data were analyzed using both multitrait linear animal models and random regression animal models (RRM) across DIM. The lactation was divided into 6 stages: 8 to 50 DIM, 51 to 100 DIM, 101 to 150 DIM, 151 to 200 DIM, 201 to 250 DIM, and 251 to 305 DIM. Difficulty arose in obtaining positive definite (co)variance matrixes with a 6 x 6 multitrait analysis, and therefore, additive genetic, permanent environmental, and residual (co)variances between the different stages of lactation were estimated using a series of bivariate analyses in ASReml (Gilmour et al., 2006). Fixed effects included in the models of analysis were herd test date, period of the year at calving, parity, DIM, and age at calving nested within parity. A linear regression on concentrate feeding level at the respective test day was also included when the dependent variable was DMI and IDG. The significance of the deviation of the genetic correlation between lactation stages from unity was determined using a log-likelihood ratio test, comparing an unconstrained model with a model in which the genetic correlation was fixed at 1.
Fixed effects included in the RRM were the same as those for the bivariate models with the inclusion of a fixed polynomial regression on days postcalving at the time of measurement. The order of the fixed regression for grass DMI and EB were both cubic based on the lactation profiles reported by Berry et al. (2006); a linear regression was fitted to the IDG data. Furthermore, the fixed regressions were included in a 2-way interaction with parity to account for significantly different lactation profiles for different parity animals as identified by Berry et al. (2006) for DMI and EB. Random regressions, using Legrende polynomials, were used to model the additive genetic variance and within-lactation permanent environmental variance. A permanent environmental component was also fitted across lactations, and residual variances were estimated within the stages of lactation described for the bivariate analyses. Residual variances within stage were therefore assumed to be homogenous, whereas heterogeneity in residual variances was possible across stages of lactation. No residual covariance was assumed among stages of lactation.
The most parsimonious random regression was determined by progressively increasing the order of the random regression from zero for both the additive genetic component and permanent environmental component within lactation. The ratio of the log likelihood of nested models, assuming a 
2 distribution and appropriate degrees of freedom, was the main statistical test to determine the most parsimonious model. However, if the variance of the higher-order regression coefficients was bound at zero, then the immediately lower-order polynomial was chosen.
Eigenvalues and eigenvectors were calculated from the additive genetic covariance matrix, and eigenfunctions were subsequently calculated from the product of the eigenvectors and Legrende polynomial coefficients. To investigate the feasibility of genetically altering the lactation profiles for DMI, EB, and IDG, the methodology presented by Togashi and Lin (2006) for improving milk yield persistency during lactation was used. Togashi and Lin (2006) defined the genetic gain (
Gi) attributable to the ith eigenvectors index as:
where 
= the matrix of Legrende regression coefficients; K = the additive genetic covariance matrix estimated from the RRM; ei = the ith normalized eigenvector; 
= the selection intensity (assumed to be 1 in the present study); and 
i = the eigenvalue associated with the ith eigenvector.
|
RESULTS
|
|---|
A frequency distribution of the number of records available for DMI by week of lactation is illustrated in Figure 1
. Means and standard deviations for grass DMI, EB, and IDG by stage of lactation are summarized in Table 1. Mean DMI and EB increased with lactation stage, although a decline in DMI occurred in late lactation. No obvious trend in IDG was evident across stages of lactation.
View this table:
[in this window]
[in a new window]
|
Table 1. Number of records, raw means, and standard deviations for grass DMI, energy balance, and internal digestibility across stages of lactation
|
|
Bivariate Analyses
Estimated residual variances were largest in early lactation and decreased almost consistently as the lactation progressed (Table 2). In contrast, the trend in genetic variance estimated from the bivariate analyses increased almost consistently with DIM, the exceptions being EB when genetic variance declined in the final lactation stage and IDG when the genetic variance was also large in the first stage of lactation (Figures 2, 3, and 4). No permanent environmental variance for IDG existed within stage across lactations. The effect of the changing variances resulted in the phenotypic variance declining almost consistently with DIM for DMI, whereas the phenotypic variance for EB and IDG was largest at both the start and end of lactation (Table 1
). Heritability estimates varied from 0.03 to 0.38 (SE = 0.07 to 0.10 except for the final stage where SE = 0.22) for DMI, from 0.05 to 0.18 (SE = 0.04 to 0.09 except for the final stage where SE = 0.19) for EB, and from 0.10 to 0.38 (SE = 0.05 to 0.15) for IDG. Repeatability estimates varied from 0.18 to 0.57 (SE = 0.04 to 0.14) for DMI and from 0.04 to 0.41 (SE = 0.04 to 0.16) for EB.
View this table:
[in this window]
[in a new window]
|
Table 2. Residual variance estimated from the bivariate analyses for DMI, energy balance (EB), and internal digestibility (IDG) for different stages of lactation
|
|
View larger version (9K):
[in this window]
[in a new window]
|
Figure 2. Genetic variance (diamonds) and heritability (squares) for grass DMI estimated from bivariate (shaded) and random regression (unshaded) models. One standard error for heritability estimated from the bivariate models is shown with vertical bars.
|
|
View larger version (9K):
[in this window]
[in a new window]
|
Figure 3. Genetic variance (diamonds) and heritability (squares) for energy balance estimated from bivariate (shaded) and random regression (unshaded) models. One standard error for heritability estimated from the bivariate models is shown with vertical bars. UFL = Unité Fourragère Lait (the net energy of lactation equivalent to 1 kg of standard air-dry barley).
|
|
View larger version (10K):
[in this window]
[in a new window]
|
Figure 4. Genetic variance (diamonds) and heritability (squares) for diet digestibility estimated from bivariate (shaded) and random regression (unshaded) models. One standard error for heritability estimated from the bivariate models is shown with vertical bars.
|
|
Genetic correlations between DMI and EB at different stages of lactation varied from –0.33 to 1.00 and from 0.03 to 1.00, respectively (Table 3). However, only the genetic correlation between DMI in the second (i.e., 51 to 100 DIM) and last (i.e., 251 to 305 DIM) lactation stage differed significantly from unity. Nonetheless, there was a tendency for the correlation to weaken as the interval between the stages being compared increased. Genetic correlations among IDG at different stages of lactation (Table 4) varied significantly from unity for the first stage (8 to 50 DIM) with both the fifth (201 to 250 DIM) and sixth (251 to 305 DIM) stages.
View this table:
[in this window]
[in a new window]
|
Table 3. Genetic correlations among different stages of lactation for DMI (above diagonal) and energy balance (below diagonal)
|
|
RRM
A quadratic random regression describing both the additive genetic and permanent environmental component of DMI and EB fit the data significantly better than lower-order functions, whereas zero variance existed for higher-order coefficients. The ratio of each eigenvalue to the sum of the eigenvalues was 0.88, 0.11, and 0.001 for DMI and 0.67, 0.30, and 0.03 for EB. This suggests that, despite its statistical significance, there was only a small improvement in the genetic variance explained by the quadratic term, especially for DMI. No variance existed in a linear regression coefficient fit to the permanent environmental effect of IDG, and therefore, a constant permanent environmental effect was assumed across lactation. Although the quadratic random regression fit to the additive genetic component did not fit the IDG data significantly (P = 0.17) better than a linear random regression, the genetic parameters estimated from the quadratic random regression were more consistent with the results from the bivariate analyses. Furthermore, the eigenvalue associated with the highest order suggested that 8% of the total variation was explained by the quadratic term, and thus, the quadratic random regression was used in all further analyses. The 2 larger eigenvalues explained 76 and 16% of the additive genetic variation in IDG.
Variance components for DMI, EB, and IDG estimated with a RRM were, in most cases, similar to those estimated from the bivariate analyses (Figures 2, 3, and 4). The estimated genetic variance for DMI increased with DIM after calving up to 183 DIM, after which it began to decline (Figure 2). A similar trend was observed for estimated heritability across lactation, although the increase in heritability in early lactation was not as steep, resulting from an opposite trend in residual variance. Heritability for DMI varied from 0.10 (8 DIM) to 0.30 (169 DIM) when estimated using the RRM. Genetic correlations between DMI at different DIM varied from 0.10 to 1.00, the strength of the genetic correlation being inversely related to the interval between compared DIM (Figure 5); the lowest genetic correlation was observed between DMI at 8 DIM and 305 DIM. The eigenfunction associated with the largest eigenvalue was strongly curvilinear and was negative in early and late lactation but positive in midlactation. The eigenfunction associated with the eigenvalue accounting for 11% of the genetic variation closely resembled a declining straight line, being positive in the first half of lactation and negative in the second half of lactation (Figure 6).
Genetic variance for EB was largest at the beginning and end of lactation (Figure 3), which is opposite to the trend in genetic variance observed from the bivariate analyses. Nonetheless, genetic variances estimated by both approaches were similar in midlactation. Heritability for EB (Figure 3) increased with DIM from 0.06 (29 DIM) to 0.29 (305 DIM). Genetic correlations across DIM varied from –0.36 to 1.00 (Figure 5), with the correlations weakening as the distance between time intervals compared increased. The lowest genetic correlation was observed between EB at 8 DIM and 305 DIM. The eigenfunction associated with the largest eigenvalue was almost linear and positive across all DIM (Figure 6). The response to selection on indexes derived from the individual eigenvectors is illustrated in Figure 7 and shows that ample variation exists to genetically alter the shape of the EB lactation profile in dairy cows.
Additive genetic variance for IDG varied from 0.27 to 2.25 units2, with the higher genetic variances observed at the start and end of lactation (Figure 4). Nonetheless, the genetic variances estimated with the RRM closely resembled the genetic variances estimated from the bivariate analyses (Figure 4). Estimates of heritability from the RRM varied from 0.08 (50 DIM) to 0.45 (305 DIM) across lactation (Figure 4). Genetic correlations ranged from –0.04 to 1.00 (Figure 5). The lowest genetic correlation was observed between 8 and 190 DIM. The eigenfunction associated with the largest eigenvalue was negative in early lactation and positive in late lactation (Figure 6).
|
DISCUSSION
|
|---|
The objective of this study was to quantify the degree of genetic variation in DMI, EB, and IDG present in dairy cows fed predominantly grazed grass and to compare the estimated genetic parameters to previously published estimates from cows fed TMR or conserved forages. Results from this study, in which pasture DMI was estimated using the n-alkane technique (Mayes et al., 1986; Dillon, 1993), clearly reveal genetic variation in grass DMI and EB, similar to estimates from TMR diets (Koenen and Veerkamp, 1998; Veerkamp and Thompson, 1999; Veerkamp et al., 2000). Analysis of the eigenvalues and eigenfunction of the additive genetic covariance matrices suggests a large opportunity to alter the shape of lactation profiles for DMI, EB, and IDG.
Alternative methods have been used across studies to determine the most parsimonious order of random regression that adequately describes the change in genetic (co)variances over time. In the present study, the ratio of log likelihood between nested models was used and, in doing so, revealed that a quadratic polynomial was most appropriate for DMI and EB, which is consistent with previous literature (Koenen and Veerkamp, 1998; Veerkamp and Thompson, 1999). Although increasing the order of the random regression fit to the genetic component from a linear to a quadratic regression did not significantly improve the fit to the data for IDG, examination of the eigenvalues of the additive covariance matrix as well as comparison with results obtained from the bivariate analyses revealed that the quadratic random regression was also most appropriate for modeling the additive genetic variance in IDG.
The larger residual variance in early lactation for all 3 traits is consistent with previous studies on DMI (Veerkamp and Thompson, 1999; Coffey et al., 2002) and other performance traits (Veerkamp and Thompson, 1999; Coffey et al., 2002). This suggests that unsystematic environmental variation has a large influence on observed phenotypic performance in early lactation, thereby indicating that the accuracy of measurement of these traits may be compromised in early lactation or that factors associated with the dependent variable may not be fully accounted for in the model of analysis. For example, live weight postcalving used in the calculation of EB is a function of many factors, including gut fill and weight of the involuting uterus, that vary among animals (Sakaguchi et al., 2004) but are not accounted for in the mixed model. This was somewhat confirmed by Coffey et al. (2002), who reported a considerably larger residual variance for live weight in Holstein cows in the first week postcalving compared with later lactation. Although calving dystocia, which may be a contributing factor to increased random variation, especially in early lactation, was not significant as a fixed effect in the model, other possibly subclinical diseases that were not recorded in the present study may also lead to increased residual variance in early lactation observations.
Grass DMI
The trend of increasing genetic variances to midlactation followed by a slow decline closely resembles the phenotypic lactation profile of DMI on grass-based (Table 1; Berry et al., 2006) and TMR (Coffey et al., 2001) diets and is generally consistent with previously reported trends in genetic variances for cows fed TMR (Coffey et al., 2001). Although Koenen and Veerkamp (1998) and Veerkamp and Thompson (1999) both reported a similar trend in different datasets, genetic variance in their study peaked earlier than the present study at approximately 84 to 119 DIM. Furthermore, the generally larger genetic variances estimated from the RRM compared with the bivariate analyses in the present study are consistent with previous studies on DMI (Veerkamp and Thompson, 1999). Estimates of genetic variance reported by Koenen and Veerkamp (1998) and Veerkamp and Thompson (1999) were higher than those presented in Figure 2; the differences may be attributable to greater diversity in the genotypes, feed systems, or both, investigated by Koenen and Veerkamp (1998) as well as the anatomical constraints of grazed grass (Gill et al., 1988) limiting the full expression of genetic variation between animals. This may also suggest the existence of a genotype x environment interaction for DMI across environments with a lower genetic variance observed in a grass-based system for milk production.
Heritability estimates of DMI in the present study are consistent with estimates from TMR (Koenen and Veerkamp, 1998; Veerkamp and Thompson, 1999) and are similar to heritability estimates for milk yield, BCS, and BW on pasture-fed cows (Koenen and Veerkamp, 1998; Berry et al., 2003). Although no heritability estimates of DMI in grazing dairy cows is available, heritability estimates of intake using a sire model in grazing sheep, predicted using the n-alkane method, range from 0.08 to 0.20 (Lee et al., 2002); the corresponding standard errors ranged from 0.05 to 0.08. Coupled with the ample genetic variation in DMI observed in the present study, the moderate heritability estimates for DMI suggest genetic improvement in this trait is possible. Nonetheless, estimating DMI on large daughter progeny groups is expensive, although alternatives exist, such as using predictor traits (e.g., linear-type traits), nucleus testing of bull dams, performance testing of young bulls, as well as marker-assisted or genomic selection. Nieuwhof et al. (1992) reported a strong genetic correlation between feed intake of bulls and cows, implying the possibility of performance testing young bulls as an indication of feed intake in their daughters. Furthermore, the large heritability and repeatability estimates for DMI in the present study suggest that across diverse genotypes and feed systems, the n-alkane method of predicting grass DMI is consistent across animals. Although repeatability statistics do not indicate accuracy of prediction, the validation of n-alkanes as an accurate predictor of herbage intake has been discussed elsewhere (Dove and Mayes, 1991).
The inverse relationship between the strength of the genetic correlation and the interval between periods compared agrees with previous studies using RRM on DMI (Koenen and Veerkamp, 1998; Veerkamp and Thompson, 1999) or BW, BCS, or milk yield (Koenen and Veerkamp, 1998; Veerkamp and Thompson, 1999; Berry et al., 2003). Although Coffey et al. (2001) reported a decrease in genetic correlations of DMI measured in the first week postpartum, with DMI measured later in lactation up to 127 DIM, correlations thereafter strengthened. Most of the correlations estimated using the bivariate analyses between stages of lactation were not significantly different from unity, suggesting a small loss of information by using a repeatability model within lactation to estimate breeding values for DMI. However, the change in variance across lactation suggests that the heterogeneity of variances across lactation stages should be accounted for if using a repeatability model. The low genetic correlation between DMI in early and late lactation, estimated with the bivariate and RRM analyses, coupled with a similarly low correlation reported by others (Koenen and Veerkamp, 1998; Veerkamp and Thompson, 1999), suggests the effect of different genes for DMI at different stages of lactation.
Eigenvalues and eigenfunctions of a covariance matrix provide an insight into how the lactation profile is likely to change due to selection (Kirkpatrick and Heckman, 1989). A large eigenvalue indicates that the change depicted in the associated eigenfunction will happen rapidly, whereas the change will be slower as the eigenvalue decreases. The sign of the eigenfunction is irrelevant; it is the change in the eigenfunction over time that is important. The main eigenfunction for DMI changed sign at 80 DIM but reverted back at 230 DIM. This indicates that the response to genetic selection for higher (lower) DMI in early lactation diminishes with DIM up to 80 DIM, after which it reverses, thereby decreasing (increasing) DMI in midlactation but increasing (decreasing) DMI in late lactation. The change in sign of the main eigenfunction corroborates the low genetic correlations between distant periods observed in the present study. The almost linearly decreasing eigenfunction associated with the second largest eigenvalue suggests that selection for greater DMI in early to midlactation will increase DMI in early lactation faster than in midlactation but will decrease DMI in late lactation. Although Koenen and Veerkamp (1998) reported a similar (albeit inverted) curvilinear main eigenfunction, it retained the same sign through the first 25 wk of lactation, which was the duration of the study period.
EB
Heritability estimates for EB and the related trait, residual feed intake, from cows fed indoors on TMR or conserved forages range from 0.08 to 0.43 (Svendsen et al., 1994; van Arendonk et al., 1991; Veerkamp et al., 1995, 2000), agreeing with heritability estimates reported in the present study for cows fed predominantly grazed grass. Heritability estimates for EB across all studies are, however, similar, albeit generally slightly lower, to its component traits of milk yield (Veerkamp and Thompson, 1999; Berry et al., 2003) and BW (Koenen and Veerkamp, 1998; Veerkamp and Thompson, 1999; Berry et al., 2003;), although it is within the range of heritability estimates reported for BCS (Berry et al., 2003), which is used as an indicator of EB (Veerkamp, 1998). Additionally, the genetic correlations between EB at different stages of lactation follow a similar pattern to those observed in the present study for DMI, which is expected given the part-whole relationship between the 2 traits. Nevertheless, the weak and sometimes negative correlations between EB in mid and late lactation with EB at the start of lactation has implications for breeding strategies, because these correlations indicate that measurements of EB in mid to late lactation have poor predictive ability when the goal is to improve EB in early lactation.
The eigenfunction pertaining to the largest eigenvalue was positive and relatively constant throughout lactation, suggesting that the majority of genetic variance in EB is explained by a principal component acting equally throughout lactation. The main eigenfunction associated with the additive genetic component for milk yield (Van der Werf et al., 1998) and BW (Koenen and Veerkamp, 1998), the contributing traits of EB, have also been shown to be relatively constant throughout lactation. Considerably less genetic variation was associated with the eigenfunction that was negative in early lactation and positive in late lactation. This therefore implies that greater weight would have to be imposed on the second eigenfunction to alter the shape of the lactation profile for EB. Further substantiating this, the heritability for EB change from 8 DIM to 70 DIM was 0.02 when calculated from the genetic and phenotypic variances at each time point and the covariance between them.
Figure 7 illustrates the response to selection on the individual eigenvectors of the genetic covariance matrix for EB. Although the shape of the genetic response is similar to that of the eigenfunctions (Figure 6), the relative difference between them differs, because the response is also a function of the associated eigenvalues. Furthermore, because the calculated eigenvectors are independent and orthogonal, the overall response to selection is the sum of the individual responses. In the present study, equal weights were placed on the individual eigenvectors, and thus the shape of the overall response to selection can be further altered by manipulating the relative weights. However, genetic correlations with longitudinal traits have been shown to vary throughout lactation (Veerkamp and Thompson, 1999; Berry et al., 2003), and thus the inclusion of such indexes in a broader breeding objective including traits such as milk production or fertility may also alter the genetic response in EB across lactation.
Diet IDG
Because of the demand in resources required to measure the ability of a cow to digest a given feed, few studies have attempted to estimate genetic parameters for IDG or to determine the effect of cow genotype on IDG. From a review of the literature, Veerkamp et al. (1995) suggested that there is probably no large genetic variation in the ability to digest a given feed among dairy cows fed the same level. However, Trigg and Parr (1981) reported a significantly (P < 0.05) greater apparent digestibility of cut pasture in cows of higher genetic merit in early lactation, the significance of which disappeared in midlactation. Results from the present study provide evidence of significant genetic variation in IDG in late lactation; estimates in early lactation were not significantly different from zero. Furthermore, the heritability estimates of 0.08 to 0.45 reported in the present study encompass the range of 0.13 to 0.17 (SE = 0.06 to 0.08) reported by Lee et al. (2002) in grazing sheep, estimated using a sire model with no numerator relationship matrix. It should, however, be noted that IDG, as measured in the present study, may not be a true reflection of IDG but may include some variance associated with the ability of a cow to identify and ingest pasture of higher digestibility relative to other cows in the paddock. Furthermore, only a constant permanent environmental variance for IDG could be modeled in the present study, and thus a portion of this variance, if truly different within lactation, may have entered the genetic (and residual) component.
The main eigenfunction of the genetic covariance matrix was negative in the first half of lactation and positive in the second half, implying that genetic forces governing IDG in early lactation have an opposite effect in late lactation, albeit the effect was small. The leveling off of the main eigenfunction in late lactation agrees with the strengthening of the genetic correlation in late lactation with IDG in early lactation (Figure 5). This phenomenon does, nevertheless, suggest a large potential to improve cow IDG in early lactation with minimal effect on IDG in late lactation. This is especially true when combined with the eigenfunction corresponding to the second largest eigenvalue, which was positive but increased with DIM, thereby partially negating the negative effect of the main eigenfunction on IDG in late lactation.
|
CONCLUSIONS
|
|---|
This is the first study to quantify the genetic variation in DMI, EBI, and IDG within the Holstein-Friesian breed fed predominantly grazed grass. Estimates of heritability and genetic variation for all 3 traits were within the ranges reported for other production traits such as milk yield, BW, and BCS, thereby suggesting that genetic gain in these traits is possible. The benefit of genetically improving feed intake is arguably greater in grass-based systems of milk production in which the botanical structure of grass (Gill et al., 1988) limits DMI. Greater genetic potential to eat more implies that supplementation of lactating animals with concentrates can be partly replaced with grazed grass. However, weak genetic correlations were observed within each trait at different stages of lactation, suggesting that genes influencing each trait may differ across stages of lactation, which has repercussions for genetic selection. Analysis of the eigenvalues and associated eigenfunctions of the additive genetic covariance matrix also revealed that considerable genetic variability exists among animals in the shape of the lactation profile for DMI, which may be exploited in breeding programs.
|
ACKNOWLEDGEMENTS
|
|---|
We gratefully acknowledge the help of farm staff and technicians during the periods of intake measurements. This project was partly financially supported by the Research Stimulus Fund (RSF-06-353).
Received for publication February 15, 2007.
Accepted for publication June 4, 2007.
|
REFERENCES
|
|---|
Banos, G., M. P. Coffey, E. Wall, and S. Brotherstone. 2006. Genetic relationship between first-lactation body energy and later life udder health in dairy cattle. J. Dairy Sci. 89:2222–2232.[Abstract/Free Full Text]
Berry, D. P., F. Buckley, P. Dillon, R. D. Evans, and R. F. Veerkamp. 2003. Genetic parameters for body condition score, body weight, milk yield, and fertility estimated using random regression models. J. Dairy Sci. 86:3704–3717.[Abstract/Free Full Text]
Berry, D. P., R. F. Veerkamp, and P. Dillon. 2006. Phenotypic profiles for body weight, body condition score, energy intake, and energy balance across different parities and concentrate feeding levels. Livest. Sci. 104:1–12.[CrossRef]
Buckley, F., P. Dillon, M. Rath, and R. F. Veerkamp. 2000. The relationship between genetic merit for yield and live weight, condition score and energy balance of spring calving Holstein-Friesian dairy cows on grass based systems of milk production. J. Dairy Sci. 83:1878–1886.[Abstract]
Coffey, M. P., G. C. Emmans, and S. Brotherstone. 2001. Genetic evaluation of dairy bulls for energy balance traits using random regression. Anim. Sci. 73:29–40.
Coffey, M. P., G. Simm, and S. Brotherstone. 2002. Energy balance profiles for the first three lactations of dairy cows estimated using random regression models. J. Dairy Sci. 85:2669–2678.[Abstract/Free Full Text]
Davey, A. W. F., C. Grainger, D. D. S. MacKenzie, D. S. Flux, G. F. Wilson, I. M. Brookes, and C. W. Holmes. 1983. Nutritional and physiological studies of differences between Friesian cows of high and low genetic merit. Proc. N. Z. Soc. Anim. Prod. 43:67–70.
Dillon, P. 1993. The use of n-alkanes as markers to determine intake, botanical composition of available or consumed herbage in studies of digesta kinetics with dairy cows. PhD Thesis. Natl. Univ. Ireland, Dublin.
Dillon, P., T. Hennessy, L. Shalloo, F. Thorne, and B. Horan. 2007. Future outlook for the Irish dairy industry: A study of international competitiveness, influence of international trade reform and requirement for change. Int. J. Dairy Technol. (accepted)
Dove, H., and R. W. Mayes. 1991. The use of plant wax alkanes as marker substances in studies of the nutrition of herbivores: A review. Aust. J. Agric. Res. 42:913–952.[CrossRef]
Edmonson, A. J., I. J. Lean, L. D. Weaver, T. Farver, and G. Webster. 1989. A body condition scoring chart for Holstein dairy cows. J. Dairy Sci. 72:68–78.[Abstract/Free Full Text]
Gill, M., A. J. Rook, and L. R. S. Thiago. 1988. Factors affecting the voluntary intake of roughages by the dairy cow. Pages 262–279 in Nutrition and Lactation in the Dairy Cow. Butterworths, London, UK.
Gilmour, A. R., B. R. Cullis, S. J. Welham, and R. Thompson. 2006. ASReml Reference Manual. New South Wales Agric., Orange Agric. Inst., Orange, New South Wales, Australia.
Grieve, D. G., C. K. Macleod, T. R. Batra, E. D. Burnside, and J. B. Stone. 1976. Relationship of food intake and ration digestibility to estimate transmitting ability, body weight, and efficiency in first lactation. J. Dairy Sci. 59:1312–1318.[Abstract/Free Full Text]
Horan, B., P. Faverdin, L. Delaby, M. Rath, and P. Dillon. 2006. The effect of strain of Holstein-Friesian dairy cows and pasture-based system on grass intake and milk production. Anim. Sci. 82:435–444.[CrossRef]
Kennedy, J., P. Dillon, P. Faverdin, L. Delaby, G. Stakelum, and M. Rath. 2003. Effect of genetic merit and concentrate supplementation on grass intake and milk production with Holstein-Friesian dairy cows. J. Dairy Sci. 86:610–621.[Abstract/Free Full Text]
Kennedy, E., M. O’Donovan, J. P. Murphy, F. P. O’Mara, and L. Delaby. 2006. The effect of initial grazing date and subsequent stocking rate on the grazing management, grass dry matter intake and milk production of dairy cows in summer. Grass Forage Sci. 61:375–384.[CrossRef]
Kirkpatrick, M., and N. Heckman. 1989. A quantitative genetic model for growth, shape and other infinite-dimensional characters. J. Math. Biol. 27:429–450.[CrossRef][Medline]
Koenen, E. P. C., and R. F. Veerkamp. 1998. Genetic covariance functions for live weight, condition score, and dry-matter intake measured at different lactation stages of Holstein-Friesian heifers. Livest. Prod. Sci. 57:67–77.[CrossRef]
Lee, G. J., K. D. Atkins, and A. A. Swan. 2002. Pasture intake and digestibility by young and non-breeding adult sheep: The extent of genetic variation and relationships with productivity. Livest. Prod. Sci. 73:185–198.[CrossRef]
Mayes, R. W., C. S. Lamb, and P. M. Colgrove. 1986. The use of dosed and herbage n-alkanes as markers for the determination of herbage intake. J. Agric. Sci. Camb. 107:161–170.
McCarthy, S., D. P. Berry, P. Dillon, M. Rath, and B. Horan. 2007. Effect of strain of Holstein-Friesian and feed system on udder health and milking characteristics. Livest. Sci. 107:1–28.[CrossRef]
McEvoy, M., M. O’Donovan, J. P. Murphy, F. O’Mara, M. Rath, and L. Delaby. 2007. Effect of concentrate supplementation and herbage allowance on milk production performance of spring calving dairy cows in early lactation. Page 46 in Proc. Irish Agric. Res. Forum, Tullamore, Ireland.
Miglior, F., B. L. Muir, and B. J. Van Doormaal. 2005. Selection indices in Holstein cattle of various countries. J. Dairy Sci. 88:1255–1263[Abstract/Free Full Text]
Nieuwhof, G. J., J. A. M. van Arendonk, H. Vos, and S. Korver. 1992. Genetic relationships between feed intake, efficiency and production traits in growing bulls, growing heifers and lactating heifers. Livest. Prod. Sci. 32:189–202.
O’Donovan, M., and L. Delaby. 2005. A comparison of perennial rye-grass cultivars differing in heading date and grass ploidy with spring calving dairy cows grazed at two different stocking rates. Anim. Res. 54:337–350.[CrossRef]
O’Mara, F. 2000. A net energy system for cattle and sheep. Version 1.2. Univ. Coll. Dublin, Ireland.
Sakaguchi, M., Y. Sasamoto, T. Suzuki, Y. Takahashi, and Y. Yamada. 2004. Postpartum ovarian follicular dynamics and estrous activity in lactating dairy cows. J. Dairy Sci. 87:2114–2121.[Abstract/Free Full Text]
Svendsen, M., P. Skipenes, and I. L. Mao. 1994. Genetic correlations in the feed conversion complex of primiparous cows at a recommended and reduced plane of nutrition. J. Anim. Sci. 72:1441–1449.[Abstract]
Togashi, K., and C. Y. Lin. 2006. Selection for milk production and persistency using eigenvectors of the random regression coefficient matrix. J. Dairy Sci. 89:4866–4873.[Abstract/Free Full Text]
Trigg, T. E., and C. W. Parr. 1981. Aspects of energy metabolism of Jersey cows differing in breeding index. Proc. N. Z. Soc. Anim. Prod. 41:44–47.
van Arendonk, J. A. M., G. J. Nieuwhof, H. Vos, and S. Korver. 1991. Genetic aspects of feed intake in lactating dairy heifers. Livest. Prod. Sci. 29:263–275.[CrossRef]
Van der Werf, J. H. J., M. E. Goddard, and K. Meyer. 1998. The use of covariance functions and random regressions for genetic evaluation of milk production based on test-day records. J. Dairy Sci. 81:3300–3308.[Abstract]
Veerkamp, R. F. 1998. Selection for economic efficiency of dairy cattle using information on live weight and feed intake: A review. J. Dairy Sci. 81:1109–1119.[Abstract]
Veerkamp, R. F., and S. Brotherstone. 1997. Genetic correlations between linear type traits, feed intake, live weight and condition score in Holstein Friesian dairy cattle. Anim. Sci. 64:385–392.
Veerkamp, R. F., G. C. Emmans, A. R. Cromie, and G. Simm. 1995. Variance components for residual feed intake in dairy cows. Livest. Prod. Sci. 41:111–120.[CrossRef]
Veerkamp, R. F., J. K. Oldenbroek, H. J. Van der Gaast, and J. H. J. Van der Werf. 2000. Genetic correlations between days until start of luteal activity and milk yield, energy balance, and live weights. J. Dairy Sci. 83:577–583.[Abstract]
Veerkamp, R. F., and R. Thompson. 1999. A covariance function for feed intake, live weight, and milk yield estimated using a random regression model. J. Dairy Sci. 82:1565–1573.[Abstract]
TOP
ABSTRACT
INTRODUCTION
), energy balance (
), and internal digestibility (•).
