Synopsis

In 2015, an agreement was reached by the United Nations Framework Convention on Climate Change (UNFCCC) in Paris to keep global warming “well below 2 degrees” (UNFCCC, 2015). Current emissions are predicted to increase global temperatures by 1.5°C within 15 yr and by 2°C within 35 yr (Howarth, 2015). As well as cuts in CO2 emissions, substantial reductions in short-lived climate pollutants such as methane will be needed to stay below the limit (United Nations Environment Programme [UNep] and World Meteorological Organization [wmO], 2011). Greenhouse gas emissions from livestock (estimated at 8% to 51% of global emissions, Herrero et al., 2011) are therefore receiving increased attention. It is now feasible to include methane measurements in genetic selection programs. Possible measurement techniques include respiration chambers (RC), portable accumulation chambers (paC), short-term breath tests, e.g., GreenFeed Emissions Monitor systems (Gem), and tracer gases such as SF6. In RC, feed is often provided according to live weight or maintenance requirements.

Animals on low or maintenance rations typically eat all the feed provided, so feed intake is positively correlated with live weight. To reduce the dependence on weight, RC measurements are often expressed as methane yield (my), calculated as methane emissions per kilogram of DMI. By contrast, measures of grazing animals are often adjusted for live weight to avoid favoring smaller animals that would be expected to eat less. They, therefore, emit less methane than their larger herd mates (Robinson et al., 2014a). The benefits of genetic improvement are maximized by using selection indexes to breed animals that are best suited to (and therefore generate maximum profits in) the production environment in which they will be raised. To help guide decisions about the utility of different methane measurements in sheep and cattle, estimates of genetic parameters and economic values were compiled from the literature and other available information for use in breeding objective software to determine the consequences of including different methane traits in the index.

Commentary

This limited series of the occasional e-letters are comprised of (2) two articles. They will appear fortnightly and are published during February and March, though they will be accessible through our social media pages.

Analysis

There is no scientific consensus about the best way to measure and include enteric methane in a genetic selection program. Measurements in RC involve a substantially higher cost than PAC or GEM, and protocols often restrict feeding to a fixed proportion of weight or maintenance, thereby removing the opportunity for animals to express genetic variation in feed intake. Most animals eat their entire ration when feed is restricted, so feed intake is positively correlated with live weight and, consequently, methane production because of the strong relationship between methane and feed intake. Measurements are often expressed as MY (i.e., emissions per unit of DFI) to create a trait that is much less strongly related to weight and feed intake.

Breeding Objectives and Selection Criteria

The use of methane as a selection criterion in the breeding objective requires an understanding of the interrelationships between methane emissions, efficiency, feed costs, and the consequences and effects of the different methane measurement systems. Models for commercial breeders need to consider many traits and economic values, including carcass and meat quality, wool or other production traits, growth, feed intake and efficiency of growing animals, and the efficiency and productivity of mature females that produce the next generation. The number of traits in this study was kept to a minimum to focus on changes in key traits correlated with methane production and avoid complications that might make the results more challenging to interpret. The results here should therefore be used as a guide to the expected changes when methane is considered as part of a more comprehensive breeding objective.

Feed costs in this study were calculated as simple averages covering the entire age range.

This is a reasonable approximation; heifers identified as phenotypically superior for feed efficiency were noted to be more efficient than cows on medium-quality pasture and unrestricted pellet feeding, although there was no difference in efficiency during restricted feeding (Herd et al., 2011). This suggests that improvements in efficiency might be overestimated unless breeding females are also measured. Traditional feed efficiency tests rarely measure breeding females or use multiple feeding levels because of the additional cost and expense. By contrast, using methane as a proxy for feed intake could allow measurements to be made over the range of pasture and feed conditions experienced by breeding flocks and herds and enable the most suitable and profitable animals to be identified for the environment in which they will be used. Efficiencies of the female breeding herd under good and poor pasture conditions could be considered as different traits from the efficiency of young, growing animals.

When combined with growth rates and weight and size of mature females, information on DMP of mature females and younger, growing animals might facilitate the selection of fast-growing animals that quickly mature into fertile, productive, efficient females for increased profitability. The fact that selection for increased muscling increased feed efficiency without any detrimental effect on maternal productivity (Cafe et al., 2014, 2015) indicates that improving efficiency will not necessarily be detrimental to maternal productivity. If neither methane nor feed intake is measured, returns from selection are maximized by increasing the main trait for which producers receive payment: growth. The result is increased slaughter weight (or a shorter time to reach the target weight if the market penalizes higher carcass weights), but higher daily feed costs offset this. In this study, the increased feed costs were substantially higher for cattle than for sheep because a proportion of cattle (equivalent to an average of 100 d per animal) was assumed to be grain-finished at a cost of A$300/t DM.

In all cattle scenarios, when methane was not measured, the cost of increased feed intake (an additional A$21.65/animal) was much higher than the cost of the increased methane emissions: zero for scenario A, A$0.97/animal for scenario B, and A$5.25/animal for scenario C. Under the optimal selection strategy, even with no payment for methane, 3 sets of GEM measurements as a proxy for feed intake slowed the increase in feed consumption from 0.37 to 0.10 kg/d and reduced the additional cost of feed from A$21.65 to A$5.61/ animal, so that, despite a somewhat lower increase in growth, overall returns increased by A$6.24/animal. Without methane measurements, emissions increased by 3.06 g CH4 per animal per day. Measuring methane (no payment for reduced emissions) slowed the increase to 1.26 g CH4 per animal per day for 1 wk of GEM measurements and 0.28 g CH4 per animal per day for 3 GEM measurement sessions because of the benefits from reduced feed intake.

Decreased methane emissions were evident only for scenario C with 3 GEM measurement sessions, which resulted in a decrease of 0.97 g per animal per day. The results for sheep were slightly different because the lower feed costs resulted in proportionately greater economic weight on increased slaughter weight. In all cases, the most profitable strategy resulted in some increase in methane emissions per animal. However, for scenario D (same costs as scenario C but no increase in methane emissions per animal), 3 PAC measurements per animal resulted in only slightly less profit ($4.67/animal) than the strategy aiming to maximize profits ($4.72/animal, but with increased CH4 emissions of 0.18 g per animal per day).

Adjustment for Feed Intake or Live Weight

It has been argued that because of the strong relationship between feed intake and methane emissions, selection should be based on methane yield rather than overall methane emissions or emissions adjusted for live weight (Amer and Fennessy, 2012). However, when breeding objectives are used to maximize profits (as might be expected in a commercial breeding system), MPadjWt measurements generate greater reductions in total methane emissions because of the additional economic benefits of reduced feed intake, which increases the economic weight on reduced methane emissions. As an example, the reduction in MY in cattle from strategy M3 (0.11 g/kg in scenario C) is four times greater than the decrease of 0.03 g/kg in MY from actually measuring MY instead of MPadjWt. The main emphasis in strategy MY is on increased slaughter weight because of its direct relationship to profitability.

Strategies M1 and M3 have a greater emphasis on reducing methane because it also helps reduce a significant cost, feed intake. Similarly, for scenario A, with no payment for methane, measuring MY results in a small increase in MY (0.01 g/kg), but measuring MPadjWt decreases MY by 0.06 (1 GEM measurement) or 0.09 (3 GEM measurements) because of the greater economic weight placed on MPadjWt to constrain the increase in feed costs. If the total feed resource is fixed, (and producers adjust their stocking rates according to feed availability), M1 and M3 still produce greater reductions in total methane output because they achieve more significant decreases in MY. If there is considerable year-to-year variation in feed availability (as is often the case) and stocking rates are set according to expected or average feed availability (or according to fixed formulae), the reductions in feed intake per animal as well as greater reductions in MY from strategies M1 and M3 compared to MY will enhance the overall benefits of these strategies in reducing total methane production.

Also help reduce vulnerability to drought or other times when feed is inadequate and restricting growth. The increased profits per head of M1 and M3 might also contribute to the increased viability of the production system. The extent that reducing methane production (mp) or MPadjWt also reduces MY depends on the genetic correlations. This study used an estimated genetic correlation (rg) = 0.30 for MY with unadjusted MP, synthesized from estimates of 0.5 (RC measurements of 1,043 Angus cattle; Donoghue et al., 2015) and much lower estimates for sheep (0.1 for correlations between RC measurements of MY and PAC measurements without adjustment for live weight; S. Dominik, CSIRO Agriculture, Armidale, NSW, Australia). A somewhat higher estimate of rg = 0.35 was used for MY with MPadjWt because MP is more strongly correlated with weight than MY, so removing the phenotypic dependence of MP on weight would be expected to increase the genetic correlation with MY.

Higher estimates of rg = 0.57 for MY with MPadjWt and 0.47 for MY with MP were used in the sensitivity analyses. The genetic parameter estimates compiled in the Supplemental Material show that MP is strongly related to weight and DFI, with somewhat stronger relationships observed for RC measurements (especially under restricted feeding protocols in which the amount of feed offered is a function of live weight) than PAC. The phenotypic relationship between weight and PAC measurements of MP was noted to differ between sites with no consistent overall relationship across all sites (Robinson et al., 2014a), suggesting that MPadjWt might prove a more robust selection criterion.

Optimum Measurement Strategies

More generally, the optimum measurement strategy depends on the repeatability, heritability, measurement costs, existence and number of genotype × environment interactions, and the desired selection strategy or breeding objective. Multistage testing (where preliminary measurements are used to select the most likely candidates; see Robinson, 2009) can reduce the number of tests required (and therefore measurement costs) while still providing similar gains to strategies that test all animals. It is now common for accredited ultrasound scanners to travel to breeding herds to measure muscle area, fat depths, and marbling (Robinson et al., 1993). Scientists in New Zealand have developed a trailer-based PAC system that can measure 72 sheep per day (Animal Selection, Genetics, and Genomics Network, 2013). It is envisaged that the costs of providing a service to measure methane using PAC or GEM would be a similar order of magnitude and, based on the results from this study, therefore both cost-effective and feasible.

Repeatabilities over Time. A notable feature of methane measurements is that short-term repeatabilities are often much higher than long-term repeatabilities. For example, Pinares-Patiño et al. (2013) reported that for RC measurements of sheep, the repeatability of MY was 0.89 on consecutive days but only 0.26 for measurements a fortnight apart. For beef cattle, Donoghue et al. (2016) reported repeatabilities of 0.75 for RCMY measures on consecutive days, 0.59 for measurements up to 60 d apart, 0.21 for measurements 61 to 120 d apart, and 0.19 for measurements separated more than 120 d. The higher repeatability of measurements on consecutive days suggests that measurement errors are less significant for overall accuracy than taking into account the variation in an animal’s emissions over time and under different pasture conditions. As might be expected because of the shorter measurement duration, PAC measurements of MPadjWt have lower repeatability, ranging from 0.31 to 0.62 for measurements a few days apart (Dominik and Oddy, 2015; Robinson et al., 2015).

Similar correlations between tests separated by at least two mo, averaging 0.20 for a series of 6 PAC measurements of MPadjWt on the same animals between 2009 and 2014 at 2 different locations using two different measurement techniques (Robinson et al., 2015) and an average of 0.28 for correlations between PAC measurements of MPadjWt of the same female sheep at 15 months, as lactating ewes at 21 months, and dry ewes at 27 months (Dominik and Oddy, 2015). Feed intake varies with the diurnal cycle (Gregorini, 2012), and there is evidence that methane production starts to increase within about 30 min of eating (Robinson et al., 2010a). Emissions measured in PAC have also been shown to decline steadily with increasing time away from the pasture (Robinson et al., 2015). However, this variation does not seem to dominate the comparison between animals, partly because statistical analyses adjust for such effects. Moreover, methane production also varies with feed eaten over periods as long as 3 d.

This was demonstrated by Robinson et al. (2014b). They showed that RC measurements were most closely related to an index of feed eaten in the RC and previous days (0.51*FIC + 0.34*FIP + 0.15*FIP2, where FIC = feed intake in the chamber, FIP = intake 0 to 24 h before measurement, and FIP2 = feed eaten 24 to 48 h before measurement). Goopy et al. (2014) reported mean rumen particulate retention times of 1.34 d for high-MY sheep and 1.11 d for low-MY sheep. Accounting for feed intake for at least 3 d before weighing as a measure of gut fill was also noted to improve the accuracy of weight gain (Robinson and Oddy, 2001). The fact that measured emissions represent a weighted aggregate of feed eaten over the previous 2 or 3 d helps explain why a single 1-h methane measurement can be used as a proxy for feed intake. For beef cattle housed in an automatic feeder (aF) pen, the diurnal variation in GEM spot measurements was minimal (Velazco et al., 2015), perhaps because animals often have to queue up and wait for access to the feeder.

Other research using the same feeding system and an average of 10 steers per pen showed that compared to steers with continuous access to feed, steers in the AF pens ate less but had similar weight gains and carcass attributes, implying increased efficiency (Robinson et al., 2013). Such results highlight the difficulties of measuring traits such as feed intake, efficiency, and methane emissions because the measurement system can affect the trait being measured. Stress associated with isolation for RC testing was found to reduce DFI of beef cattle, resulting in an estimated increase in MY in fibrous diets (Llonch et al., 2016). Testing in sheep in RC was also noted to reduce DFI (Bickell et al., 2014; Robinson et al., 2014b). Further research will be required to understand the effect of pasture quantity and quality on CH4 emissions. For example, significant differences in methane emissions of cattle (selected as low or high for RFI) were observed when grazing high-quality pasture (0.34 vs. 0.46 g CH4/kg live weight) but not low-quality pasture (0.26 g CH4/kg live weight for both groups; Jones et al., 2011).

Dependence on Assumptions and Input Parameters

The results reported here were derived from the estimates of genetic and phenotypic variances and covariances for the traits of interest. These values used were compiled from all available information, as described in the tables and the Supplemental Material (see the online version of the article at http://journalofanimalscience.org). Additional results from a sensitivity analysis did not change the conclusions. Continued updating as new results become available will help refine the results so that optimum selection strategies can be realized.

Methane as a Proxy for Feed Intake

Unlike some RC test protocols, animals in commercial production systems typically have ad libitum access to pasture or other feed because faster growth rates generally result in higher profits. When animals are able to express genetic variation in feed intake and efficiency, our results suggest that when it is not practical to measure feed intake, methane is a useful proxy. Consequently, including methane in the breeding objective could help increase profitability by enabling production to be improved for smaller increases in DFI. This is particularly important for feedlot-finished cattle (even if there is no payment for reduced methane emissions) because of the much greater cost of feedlot rations (approximately A$300/t DM) compared to the cost of pasture (A$50/t DM). Feed intake has been shown to be highly correlated with methane emissions; for example, a correlation of r = 0.96 was reported for a study of 1,034 animal records of RC methane emissions and feed intake of beef and dairy cattle on forage diets (Charmley et al., 2016).

For sheep, a study of mature ewes of 4 different breeds representative of the UK sheep industry with ad libitum access to a fresh-cut pasture of 3 different types, varying in digestibility, reported a correlation between DFI and CH4 emissions of r = 0.77 with little breed effect (Moorby et al., 2015). In a small study of 91 ewes from 4 sires, sire means for RFI at 11 mo were noted to be strongly related to sire means for MPadjWt at the same age and also sire means for MPadjWt of the same sheep as dry 2-yr-olds after pregnancy and lactation (Robinson et al., 2016), suggesting that reducing MPadjWt for the same level of production will indeed improve efficiency. The utility of an estimate of feed intake on a single day depends on the repeatability of the trait and the population of animals to be measured. One evaluation recommended test lengths of 70 d for feed efficiency and 35 d for feed intake measurements (Archer et al., 1997). Such recommendations are appropriate for traits with a considerable day-to-day variation or large measurement errors.

Especially if the cost of additional days of testing is relatively low compared to the cost of transporting animals to a central test station, managing them, and allowing time for animals to adapt to the test conditions. For measurements within a 30-d period, repeatability of DFI ranged from 0.36 for feeder heifers to 0.40 for mature cows (Basarab et al., 2013), implying that the correlation between DFI measured on a single day and the mean of 30 others is 0.58 for feeder heifers and 0.61 for cows. The repeatability of feed intake measurements of ewes aged 11 mo was high (0.76) before adjustment for weight and weight gain but much lower when adjusted for weight (0.51) and lower still (0.42) after adjustment for weight and weight gain (Robinson et al., 2016). It may seem somewhat surprising that a single PAC measurement or a few spot measurements per day over a week can represent a useful alternative to measuring feed intake for 30 d. However, a different cost structure applies when PAC or GEM are transported to breeding herds or flocks to measure emissions of animals at pasture.

A cheaper, less accurate measurement on all animals will often generate better returns than an accurate but costly measurement on a small number of animals. This can be illustrated by a simple example of phenotypic measurements on a single trait to select the best ten animals from a stud of 200. The cost of measuring all 200 animals for 3 d is likely to be similar to measuring 20 animals for 30 d. Consider a trait with day-to-day repeatability of 0.3 (slightly lower than the repeatability noted above for DFI). Assume the more accurate 30-d measurement has a moderate heritability of 0.4 and a phenotypic variance of 10. The mean of 3 d of measures has a more significant measurement error and, therefore, lower heritability (0.24), but selecting 10 out of 200 has a higher intensity (i = 2.063) than selecting 10 out of 20 (i = 0.798). Overall, when all 200 animals have three 1-d tests, the mean of selected animals (i*h*GSD, where GSD = the genetic SD) of 1.16 is about two times higher than the standard of the best 10 of 20 animals chosen at random for a 30-d test (0.57).

Conclusion

The economic analyses presented here suggest that methane emissions measured for 40 to 60 min in PAC or over 1 wk using the GEM system are useful traits to consider as selection criteria to improve the breeding objective. Depending on costs and benefits, it could also be worthwhile to repeat the measurements, ideally after an interval of at least 2 wk or at different times of the year. There are obvious benefits in measuring feed intake for research purposes and to improve the accuracy of estimated genetic and phenotypic covariance matrices. However, when it is not practical or cost-effective to measure feed intake, methane emissions can be used as a proxy for feed eaten over the previous 1 to 3 d. Even at the highest reasonable cost of methane emissions (A$2,580/t, calculated using methane’s 20-yr GWP of 86 and CO2 equivalent cost of A$30/t), the economic benefits achieved by improved feed efficiency are greater than those from reducing methane emissions.

We want to proceed further, more in-depth on this controversial subject. Therefore, please follow us on social media and join us on the (15^th) fifteenth of the month for Part (2!) two to learn more about the environmental impact of beef production.  Thereafter, please join us on the (1^st) first and (15^th) fifteenth of each month for our fortnightly delivery of insightful, informative must-reads from some of the world’s scientific thinkers. Selected by our editors is a collection of current topics with a profound ability for beneficial improvements, guidelines, and process practices. Thank you for reading our publication entirely; please share it with others who also care. We look forward to your comments and having you with us again fortnightly; we will be thrilled in having you with us; thus, we will take your trust in us with great honor and appreciation.