Relationships between live weight and herd-test traits in a Saanen goat herd in New Zealand

Abstract

Genetic relationships among doe live weight and herd-test milk traits were investigated in a large herd of Saanen goats, managed with seasonal (spring) kidding and twice-daily milking. There were 34,183 herd-test records of milk solids yield (FPL) for 20 milking seasons to 2010–2011, with individual fat (F), protein (P) and lactose (L) yields in later seasons, somatic cell count (SCC) from 1999–2000 onwards (26,575 records), and weights recorded on four dates between June 2008 and November 2010 (4358 records). FPL, log₁₀SCC and live weight were analysed using restricted maximum likelihood, with a repeated-animal model and full relationship matrix, yielding heritability estimates of 0.35±0.02, 0.20±0.02 and 0.64±0.04 respectively and repeatabilities of 0.48±0.01, 0.37±0.01 and 0.80±0.01 respectively. Genetic correlation estimates were 0.41±0.06 (FPL, live weight), −0.07±0.06 (FPL, log₁₀SCC) and 0.24±0.08 (live weight, log₁₀SCC). Expected correlated changes in doe live weight will depend on the selection index weights applied to doe weights and the other traits.

Keywords:

• Saanen
• goats
• milk
• somatic cell count
• live weight

Introduction

Producing milk from small ruminants has become a relatively recent enterprise in New Zealand, providing specialist milks from goats and sheep for human consumption. This paper estimates genetic parameters for a large Saanen goat herd in Northland, New Zealand, where data have been collected and stored over the last 20 years.

Genetic parameters for the first recorded traits from the herd (milk yield and its components) were described by Morris et al. (1997); a description of the genetics of reproductive parameters in the herd was produced later (Morris et al. 2006). Since 2008, doe live weights have been added to the recorded traits, with a possible endpoint of including live weight into a multiple-trait breeding objective, along with milk production, reproduction and somatic cell count (SCC). This would provide the opportunity to move closer to an objective that describes milk production efficiency rather than just milk production output in order to balance the costs of possible extra food for maintenance and production with the extra returns from heavier does. The present report estimates the genetic parameters for doe live weight and SCC, along with updates on milk production parameters and the relationships among the traits.

Method

Ethics

The data in this study were obtained from a commercial property using routine recording procedures. No experimental treatments were involved and no ethics approval was required.

Herd description and management

The herd was established in 1978, at Hikurangi, Northland, with the purchase of 30 in-kid Saanen does from a number of small producers. Expansion and management practices since then have been described by Morris et al. (1997, 2006). Briefly, another 40 does were purchased in 1979, again from a variety of sources. Two more purchases, totalling 180 Saanen does, were made in 1991 and 1992, bringing the herd to a total of 300 does. Doe numbers have subsequently risen almost exclusively through home-bred replacements. Since 2006, around 1300 does have been herd-tested each year. With the exception of three Nubian and one Toggenburg, all bucks used in the herd were Saanen. Since 1991, 73 (86%) of the 85 bucks used at the property were home-bred (∼4 new bucks per year).

The herd is managed as a spring-kidding enterprise, with does milked twice daily. Kids not selected as replacements are culled. Replacement decisions for bucks and does have been based mainly on milk production data (from dams, progeny or self). Additionally, since 2005, there has been selection for earlier kidding date, with breeding values for kidding date and milk solids yield combined using a selection index.

From the 1996–1997 season, 80% of the does were managed in covered yards. Since 2001, this has been increased to 100%. Feed is transported to the herd daily as a combination of grass, maize silage and concentrates. Yearlings are managed separately from the mixed-age does.

Milk production recording

Herd-testing was carried out 2–4 times per year from the 1991–1992 season to the 2010–2011 season, with the following exceptions: in 2009–2010, only one herd-test was carried out and, in 2010–2011, only a single herd-test had been completed at the time of this analysis. Herd-testing dates ranged from September, early in the season, to April, using the standard dairy cow p.m./a.m. herd-test procedure. At each herd-test, milk yields for each doe were recorded. The primary herd-test production records were milk yield (litres) on herd-test day and the concentration of milk solids, consisting of fat (F), protein (P) plus lactose (L), as determined for individual does by the Livestock Improvement Corporation from pooled p.m./a.m. samples, giving a daily yield of milk solids (FPL). Where lactose data were not included in FPL, a multiplicative factor of 1.747 was used to convert data on F and P to FPL, being the conversion factor provided by the Dairy Goat Co-operative. In later years, interest widened to storing individual components for F and P from each herd-test on individual does. Electronic copies of the individual milk components had not been stored for past years; F and P data for each doe were available for this study from 1998–1999 and L records for each doe were available from 2008–2009. SCCs were obtained from each doe from the 1999–2000 season onwards.

Live weight recording

Live weights were recorded on the does on four occasions between 2008 and 2010, each at different times of the year, namely June 2008, March 2009, January 2010, and, for the 2009-born does, November 2010.

Traits analysed

Seven traits were analysed in this study: live weight, herd-test milk production data (daily milk yield [litres], component daily yields for F, P, and L, milk solids yield [the sum of F, P, and L]) and SCC.

Statistical methods

There were 34,183 herd-test records available for analysis after the records from does with less than 21 days in milk had been excluded. Linear model methods (SAS 1995) were applied to each trait to find the most appropriate fixed-effects models. For milk production traits, the significant fixed effects were sire, contemporary group (kidding year, management group and relevant age of doe combinations) and number of kids born (NKB), with days in milk (DIM) within contemporary group as a covariate. For SCC, transformed to base-10 logarithms, the fitted effects were as for milk production, plus DIM² within contemporary group. For live weight, fixed effects were contemporary group (weigh date), sire, NKB and age of doe, with kidding date deviation within contemporary group and, for the 1-year-olds, birth date deviation as a covariate.

Breeding values (BVs) for each trait were estimated for all animals using restricted maximum likelihood (REML) procedures (Gilmour et al. 2009), with a repeated-animal model and a full relationship matrix, tracing pedigrees back to the foundation stock. Fixed effects determined in the analyses above were included, except for sire (which was accounted for in the relationship matrix). Genetic and phenotypic correlations among live weight, each milk production trait and log₁₀SCC were estimated, using a series of two-trait animal REML models, including repeatability terms for each trait.

Results

Live weight

A total of 4358 live weight records were included in the analyses, comprising between 1215 and 1529 records at each of the first three recording dates and 319 records at the last recording date, when only the 1-year-old does were weighed. A total of 2236 does, most weighed more than once, were recorded in the study. Table 1 shows the overall mean, phenotypic standard deviation and repeatability estimate for live weight. Within-animal repeatability for live weight was high (0.80). Table 2 summarises raw live weight means by doe age and weigh date. Yearlings were about 20 kg lighter than their 2-year-old herd-mates.

Table 1  Means, phenotypic standard deviations and repeatabilities for herd-test yields, somatic cell count (SCC) and live weight data in Saanen does.

Trait	No. of records	Mean	Phenotypic SD	Repeatability
Live weight (kg)	4358	64.4	9.7	0.80±0.01
Milk (litres/day)	34180	2.53	0.72	0.52±0.01
Solids (kg/day)	34180	0.288	0.080	0.48±0.01
Fat (kg/day)	28535	0.098	0.030	0.48±0.01
Protein (kg/day)	28535	0.081	0.021	0.54±0.01
Lactose (kg/day)	4509	0.122	0.036	0.50±0.02
Log10SCC^1	26575	5.76	0.436	0.37±0.01
^1A log10 transformation was carried out; back-transformed (geometric) mean SCC = 581,000 cells/ml.

Table 2  Raw live weight means by doe age (in years) and weigh date. The number of records is shown in brackets.

Age of doe kidding 2008	Pre-kidding weight (kg) Jun. 2008	Autumn weight (kg) Mar. 2009	Age of doe kidding 2009	Pre-mating weight (kg) Jan. 2010	Age of doe kidding 2010	Spring weight (kg) Nov. 2010
1	43 (479)	52 (414)	1	48 (375)	1	44 (319)
2	68 (306)	65 (256)	2	64 (303)
3	78 (293)	74 (254)	3	69 (171)
4+	88 (451)	79 (371)	4+	75 (366)

SCC and milk production traits

Measurements of SCC were available from 1999–2000 onwards; there were 26,575 records overall from 3957 does. From over 34,000 herd-tests for litres and total solids (FPL) from 1991–1992 to 2010–2011, there were 4870 separate does recorded at least once for milk production. On average, there were seven test days per animal, comprising repeated herd-tests within lactation and an average of 2.9 lactations per doe. Overall means, standard deviations and repeatability estimates for the milk traits are shown in Table 1. Within-animal repeatability for log₁₀SCC was 0.37; this was lower than for the milk production traits, which all had repeatabilities around 0.5.

Heritabilities and correlations

Heritabilities of, and pairwise genetic and phenotypic correlations between, the traits are summarised in Table 3. Live weight was highly heritable (0.64), milk production traits moderately so, and log₁₀SCC had a lower (0.20) but still significant heritability. Live weight was positively correlated with all the milk production traits, but was also positively correlated with log₁₀SCC (0.24). The milk production traits displayed no significant genetic correlations with log₁₀SCC (e.g. −0.07±0.06 between log₁₀SCC and milk solids yield), though the phenotypic correlations were significant and negative (favourable).

Table 3  Genetic parameters for doe weight, milk production traits and somatic cell count (SCC). Heritabilities are shown in bold on the diagonal, genetic correlations are below the diagonal and phenotypic correlations above the diagonal.

	Live weight	Litres	Solids yield	Fat yield	Protein yield	Lactose yield	log10SCC
Live weight	0.64±0.04	0.25±0.03	0.27±0.03	0.25±0.03	0.32±0.03	0.18±0.04	0.25±0.04
Litres	0.38±0.06	0.39±0.02					−0.21±0.01
Solids yield	0.41±0.06		0.35±0.02				−0.21±0.01
Fat yield	0.39±0.06			0.36±0.02			−0.24±0.01
Protein yield	0.47±0.06				0.41±0.02		−0.18±0.01
Lactose yield	0.35±0.08					0.35±0.04	−0.29±0.02
log10SCC	0.24±0.08	−0.01±0.06	−0.07±0.06	−0.12±0.06	−0.05±0.06	−0.00±0.10	0.20±0.02

Discussion

Live weight parameters

There seem to be few published estimates of genetic parameters for goats, including milk yields and live weights. Live weight in lactating Saanen does was shown in this study to be a heritable trait. This was consistent with a finding reported for doe kids in France, where heritability estimates of weights between ages of 2 and 7 months averaged 0.50 (Ricordeau et al. 1972). Genetic and phenotypic correlations in the French doe kid data between weights at 5 and 7 months were found to be 0.95 and 0.85 respectively; the repeatability for live weight in mature does in the current study was 0.80±0.01, consistent with the phenotypic correlation of 0.85. Other parameters were reviewed by Iloeje & van Vleck (1978) who reported a variety of estimates across the papers reviewed. Constantinou (1989) reported estimates of 0.20±0.11 and 0.65±0.03 respectively for the heritability and repeatability of adult doe body weight, and a genetic correlation estimate of 0.03±0.29 for body weight with total milk yield; this appears to be low given the feeding regime applied. The phenotypic and genetic correlations between live weights and milk production traits in the present herd were all positive in sign and significantly greater than zero (P<0.05 or better). Since the live weights were also moderately repeatable, then it is probably only useful to record this trait two or three times in each animal's lifetime. For example, a spring weight on the new crop of lactating yearlings and a pre-kidding weight on all does in alternate years may well be sufficient to incorporate live weight into some form of selection index for higher milk production, subject to live weight constraints or a negative selection pressure on live weight (as in the New Zealand dairy cattle industry's breeding objective (AEU 2011)).

SCC and milk production parameters

The heritability and repeatability estimates from herd-test data on milk production traits were at least moderately heritable, and repeatable. These findings are broadly consistent with previously published estimates for milk production parameters from the same herd (Morris et al., 2006). For log₁₀SCC, the heritability (0.20±0.02) and repeatability were lower than for milk production traits, but not as low as for SCC in dairy cows; for example, the literature review of Mrode & Swanson (1996) reports an overall heritability estimate of 0.11±0.04.

In asking what other aspects of overall production might be relevant to herd productivity, kidding date (KD), a proxy for fertility, would need to be included. In this study, KD had a heritability of 0.10±0.02 (repeatability of 0.19±0.02). Latest genetic correlation estimates, not tabulated (with phenotypic correlation in brackets) between KD and litres were −0.18±0.11 (−0.04±0.03); F, −0.19±0.12 (−0.07±0.03); P, −0.18±0.11 (−0.08±0.03); L, −0.07±0.15 (−0.03±0.04); log₁₀SCC, 0.04±0.13 (−0.09±0.03); live weight, −0.05±0.12 (−0.09±0.04). Thus, on present data, KD was unlikely to show any change as a response to indirect selection, although, with the limited data on KD so far, a reassessment of the correlations would be useful after further years of data collection.

In order to improve biological efficiency, restrictions may need to be applied to live weight for doe replacements and this would have a negative impact on rates of genetic change for milk production traits. It is likely that the heaviest does would be penalised (assuming that live weight and maintenance food intake are related) because the genetic correlations between live weight and milk production traits were considerably less than unity (0.35–0.47). The genetic correlation between live weight and log₁₀SCC was positive. Having all the relevant parameters available for Saanen goats in New Zealand now provides opportunities for appropriate index construction.

Conclusions

For herd-test data from a large Saanen herd in Northland, all the recorded traits obtained (milk yield and its components [F, P, L], log₁₀SCC and live weight) are heritable and repeatable. In the case of live weight, the genetic correlations with milk yield or milk component yields are positive. Thus, selection to reduce live weight (for improved biological efficiency) at the same time as increasing milk production traits would be slow.

Acknowledgements

The authors thank other members of the Foote family for supplying data for this study. The authors were funded by AgResearch in carrying out the analyses and reporting on the results.