The Proceedings of the National Academy of Sciences published in Nov. 2013 a comprehensive and quantitative analysis of anthropogenic (i.e., human-made) methane emissions in the United States (Miller et al., 2013). The analysis used measurements of regional atmospheric methane concentrations to estimate and characterize total methane emissions for the U.S. One of the conclusions of the study, directly related to animal agriculture, was that "…emissions due to ruminants and manure are up to twice the magnitude of existing [i.e., US EPA] inventories," This conclusion created a justifiable concern among animal scientists and deserves a closer look.
We contacted the authors and in a series of conversations concluded that their complex and important analysis had one significant drawback - the method used to differentiate methane emissions from "ruminants" (the term used by the authors also includes manure emissions and emissions from non-ruminant farm animals and wild ruminants) from other anthropogenic sources. In a nutshell, the study assumed that methane emissions over states with large animal industries, but fewer other sources of methane, are emissions from ruminants. There are apparently large uncertainties in this kind of approach. By the authors' own words "The uncertainties on the sector-based [methane] budget estimates are large…" and as a result, "…the atmospheric methane…estimates by source sector [for example, ruminants vs. other methane sources] often have larger confidence intervals [i.e., low reliability of the prediction]".
We evaluated the validity of the conclusions regarding methane emissions from ruminants by Miller et al. by utilizing a relatively simple "bottom-up" method based on current livestock inventories and enteric or manure methane emission factors (see Hristov et al., 2014). Our approach took into consideration the number of animals in each of several cattle categories (beef cows, dry and lactating dairy cows, beef and dairy heifers, steers and heifers on feed, bulls, and calves) and feed consumption and methane production rates per unit of dry matter intake for each animal category.
As with every indirect method, this "bottom-up" approach has to use a series of estimates (with their own uncertainties), which are, however, based on research data from animal measurements. Animal scientists have generated large datasets of enteric methane emission rates per unit of feed or energy intake and although there is animal-to-animal variability, average methane production rates are representative for the majority of ruminant production systems. Since methanogenesis in the rumen is substrate-dependent, methane production data derived from studies utilizing respiratory chambers (or other techniques) are representative of field emissions. In other words, a ruminant animal cannot produce more methane from ruminal fermentation (lower gut fermentation is a minor contributor to methane emissions) than what fermentation of the ingested feed will allow.
A methane dataset we have generated for a recent FAO report (Food and Agriculture Organization of the United Nations) and similar meta-analyses by others have yielded a relatively close range of methane production rates per unit of feed intake. For cattle fed forage/concentrate rations (most dairy cows in the U.S.), for example, these rates are on average around 20 g of methane/kg dry matter intake. Cattle fed diets with a greater proportion of concentrate feeds, such as grain-finished feedlot cattle, will produce less methane because starch yields less fermentation gases than fiber.
A recent chamber study by Hales et al. (2013; J. Anim. Sci., 91:819-828) reported emission rates of around 7.8 to 12.8 g/kg dry matter intake for cattle fed 90% concentrate/10% forage diets. Based on these resources, methane production rates were assumed at 8 to 13 (cattle on feed) or 20 (all other categories) g/kg feed dry matter intake. Minimum and maximum emissions were estimated with a range of ± one standard deviation (for most animal categories: ± 4 g/kg). Contributions to methane emissions by other ruminants or non-ruminant herbivores (sheep, goats, wild ruminants, horses, etc.) are small in the U.S. and were not included in our analysis.
In addition to emission rates, another important factor in our analysis was feed intake. For most animal categories feed dry matter intake was based on the beef (2000) and dairy (2001) National Research Council (NRC) requirements and ranged from 3.8 (calves < 500 lbs live weight), to 9-10 (cattle on feed or other steers and heifers > 500 lbs), 11 (beef cows), and 22 kg/d (dairy cows).
For animal inventories, we used the USDA-National Agricultural Statistics Service livestock inventory estimates for 2013. Total cattle inventories in 2013 were 89,299,600 head (including 29,295,200 beef cows, 9,219,900 dairy cows, and 13,351,700 cattle on feed, among other categories). This is down about 7-8% from the 97,003,000 to 96,035,000 total cattle inventories in 2007-2008, the years the Miller et al. analysis was based on.
Using USDA-NASS cattle inventories and the above estimates for dry matter intake and emission factors, total U.S. methane emissions from enteric fermentation were estimated at approximately 6.2 tera grams (Tg; i.e., 1012)/yr (min = 5.0 and max = 7.5). In the most recent US EPA publication (DRAFT Inventory of U.S. Greenhouse Gas 6 Emissions and Sinks: 1990 - 2012) methane emissions from enteric fermentation, i.e. rumen and large intestinal microbial fermentation, were 6.71 Tg in 2012 and were slightly higher, 7.00 Tg in 2008. Of the 2012 emissions, 71.3% were from beef cattle, 24.8% from dairy cattle, and the remaining 3.9% from swine, horses, sheep, goats, bison, and mules and asses. Thus, ruminant methane emission estimates calculated using our "bottom-up" approach were comparable to the current US EPA estimates. Our estimates were also independently verified by our collaborators using enteric methane prediction equations proposed by Moraes et al. (2013).
The Miller et al. analysis also included manure methane emissions. We used USDA-NASS inventories for cattle, swine (59,387,000 market swine and 5,834,000 breeding swine), and poultry (a total of 8.562 billion birds) and IPCC (2006; Intergovernmental Panel on Climate Change, Chapter 10. Emissions from livestock and manure management. Guidelines for National Greenhouse Inventories. Vol. 4. Agriculture, Forestry and Other Land Use) manure methane emission factors [from 0.02 (most poultry categories), to 1 (beef cattle), and 53 (dairy cows) kg/hd/yr] to estimate emissions from manure management. Using this approach, manure emissions in the U.S. were estimated at 1.60 Tg/yr, which is lower than the current, 2012, US EPA estimate of 2.52 Tg/yr (2.5 Tg in 2008). We have to point out that the US EPA figure is perhaps more realistic and representative of manure systems in the U.S. than our estimates based in IPCC emission factors.
Based on our analysis, the conclusions by Miller et al. that US EPA estimates for livestock methane emissions are grossly underestimated appears to be unsubstantiated. Methane production in the rumen must come from fermentation of dietary substrate (mostly carbohydrates) and emission rates are established and relatively well-studied by the animal science community. Feed intake can also be reliably predicted based on animal requirements and national cattle inventories are accurate.
Thus, we are confident that enteric methane emission estimates derived using our "bottom-up" approach, which are similar to current US EPA data, are accurate and more representative of actual methane emissions from livestock than the Miller et al. estimates, which are based on uncertain assumptions regarding methane source differentiation. There needs to be more research to more closely link "top-down" and "bottom-up" inventories. Both techniques are important and using them together can assist in reducing the error associated with estimates of greenhouse gas emission by ruminants. We also find a need for a more detailed inventory of manure systems for all farm animal species and categories, which will help to more accurately estimate greenhouse gas (and ammonia) emissions from animal manure in the U.S.