Material for : Benefits of including methane measurements in selection strategies

1. Correlations with weight at slaughter (SltWt) SltWt, DFI: For cattle, weighted mean estimates from the 6 studies cited by Arthur and Herd (2008) and Crowley et al. (2010) are: rg = 0.72, rp = 0.61, for weight (during a feed efficiency test) with DFI measured over at least 50 days. Genetic correlations of weight with DFI in lambs are similar (rg = 0.85, François et al., 2007; rg = 0.71 ± 0.11, Snowder and Van Vleck, 2003, rg = 0.71, rp = 0.29, Lee et al., 2002) but lower for adults (rg = 0.34 ± 0.22, rp = 0.35 ± 0.03, Lee et al., 2002; rg = 0.20 ± 0.09, rp = 0.12 ± 0.03 for digestible DFI in Merino ewes, measured by chromium sesquioxide capsules and expressed as a ratio of the estimate for each ewe to the mean of the contemporary group, Fogarty et al., 2009; rg = 0.23 ± 0.10, rp = 0.15 ± 0.02 for correlations of same trait with post-weaning weight in crossbred ewes, Fogarty et al., 2006). In this evaluation, slaughter weight was considered a proxy for meat production, so correlations for non-mature animals were considered the most relevant. After ‘bending’ to make the correlation matrices positive definite, the values used in the evaluation were: rg = 0.63, rp = 0.59. SltWt, DMP: The table shows estimates of rg and rp from studies in Australia using RC and PAC. The RC protocol of restricted feeding based on a function of liveweight is likely to result in higher estimates of the correlation between weight and DMP than expected under commercial conditions when animals have ad lib access to feed. Therefore, the estimates from PAC of rg = 0.67 and rp = 0.47 were considered the most appropriate. After ‘bending’ to make the correlation matrices positive definite, a slightly lower value of rg = 0.64 was used in the final analysis. SLtWt, MPadgWt: rp should be 0 because of the adjustment for weight. There is no evidence that rg differs from 0, so the values used were: rp = 0, rg = 0. SLtWt, MY: In studies that report them, genetic correlations were low, e.g. rg = -0.10 ± 0.18 for test weight, and 0.05 ± 0.17 for final weight in cattle (Donoghue et al., 2015); 0.06 ± 0.12, 0.06 ± 0.13 for weaning and 8-month weight in sheep (Pinares-Patiño et al., 2013). Given the relatively large SE, the best estimate is rg = 0. Phenotypic correlations with liveweight had lower SE and were are all positive, rp = .04 ± 0.04 for test weight, rp = .10 ± .04 for final weight in cattle (Donoghue et al., 2015), rp = 0.01 ± 0.02, 0.03 ± 0.03 for weaning and 8-month weight in sheep (Pinares-Patiño et al., 2013). The pooled estimate (used in Table 5) is rp = 0.03 ± 0.01. SLtWt, RFI: rp should be 0 because RFI is a measure of feed intake adjusted for weight and weight gain. The weighted average of estimates from the studies cited in Arthur and Herd (2008) and Crowley et al. (2010) was very low: rg = -.06 ± 0.05. Although this is close to zero, the pooled value was considered preferable to simply rounding the average to zero because of the possibility that the genetic correlation differs according to environment. In one study (Morris et al., 2014), low-RFI heifers (progeny of 4 low-RFI sires, range –0.82 to -1.16 kg/day) had faster weight gains than the progeny of 4 high-RFI sires (range 1.0 to 1.14 kg/day), suggesting a possible negative correlation between RFI and weight. However, this particular study did not provide information on EBVs for 400day or final weight, so the results could simply reflect the differences in sire EBV for weight. After bending Table. Literature estimates of genetic and phenotypic correlations between DMP and weight.