An Analysis of “Livestock and Climate Change”

In 2006, the FAO published Livestock’s Long Shadow (LLS), which estimated that the livestock sector accounted for 18 percent of anthropogenic GHG emissions. In 2009, the World Watch Institute released Livestock and Climate Change (LCC), authored by Robert Goodland and Jeff Anhang, which claimed that “livestock and their byproducts actually account for at least 32,564 million tons of CO2e per year, or 51 percent of annual worldwide GHG emissions.” (pg. 11) The following page provides a critical assessment of the analysis of Goodland and Anhang. Four issues are outlined below. These issues have been identified previously by Stephen Walsh. I provide additional details that add to Walsh’s critique.

Here is a brief summary of the critique:

  1. Contrary to IPCC guidelines, Goodland and Anhang add animal respiration to GHG inventories, without subtracting CO2 sinks due to photosynthesis of pastures and feed crops. They also quietly neglect to mention that the FAO estimated animal respiration, instead opting to misuse a back-of-the-envelope calculation by Alan Calverd to obtain a figure nearly three times as high as the FAO estimate.
  2. They recalculate methane emissions using the 20-year global warming potential (GWP) instead of the lower 100-year GWP to convert methane emissions into tons of carbon dioxide equivalent (CO2e), but only do so for the 37 percent of anthropogenic methane emissions that can be attributed to livestock.
  3. They adjust  emissions for an increase in the tonnage of livestock products that occurred between 2002 and 2009. However, since their base emission figures are for the year 2000, this is an inappropriate adjustment, given that they do not also account for increases in emissions over the same period from other sources (i.e. fossil fuels).
  4. They allege that LLS undercounts the livestock population, with the FAO’s own statistics showing higher numbers than those mentioned in LLS. Based on this, Goodland and Anhang increase their emissions figures for livestock by a further 10 percent. However, this claim is not supported by FAO statistics, and appears to arise because the authors have conflated estimates of the population of animals alive at a point in time with estimates of the number of animals slaughtered in a year.

To motivate what follows, it is worth summarizing what the earlier FAO analysis is attempting to do. This is summarized in the more recent FAO publication entitled Tackling Climate Change Through Livestock, which notes “The 2006 assessment compared its estimate (based on a 2001 to 2004 reference period) with total CH4, N20, and CO2 athropogenic emissions estimated provided by the World Resource Institute (WRI) for the year 2000.” (pg. 15) The FAO is estimating the “total GHG emissions from livestock supply chains. (ibid)” and comparing it to total anthropogenic emissions in a nearby year. The FAO study is answering a well-posed research question: Of the net anthropogenic GHG emissions emitted in the year 2000, what percentage are attributable to the livestock sector?

Goodland and Anhang revisit the FAO analysis, claiming in the process that “We show that 25,048 million tons of CO2e attributable to livestock have been undercounted or overlooked; of that subtotal, 3,000 million tons are misallocated and 22,048 million tons are entirely uncounted. When uncounted tons are added to the global inventory of atmospheric GHGs, that inventory rises from 41,755 million tons to 63,803 million tons.” (pg. 11) A key question for what follows is what the initial number of 41,755 million tons represents. Goodland and Anhang cite a World Resource Institute Publication entitled “World Greenhouse Gas Emisssions in 2005.” However, this document reports emissions of 44,153 million tons CO2e. The figure of 41,755 million tons can be found in another WRI publication, which clearly indicates (note to Figure 1.3) that the estimate is for the year 2000.

Goodland and Anhang do not attempt to replicate the FAO analysis. Instead they take the 18 percent figure reported in LLS and apply it to the WRI estimate, obtaining a number of 7,516 million tons CO2e. They then proceed to make a number of adjustments. The major ones include (1) adding livestock respiration, which they estimate at 8,769 million tons, (2) recalculating methane emissions from livestock using the 20-year global warming potential instead of the 100-year global warming potentials used in LLS, which results in an emissions figure 5,047 million tons CO2e higher than the value obtained using the 100-year GWP, and (iii) adjusting all three numbers to account for an increase in the tonnage of livestock products that occurred between 2002 and 2009, which results in an additional 2,560 million tons. An additional 3,000 million tons are added for a series of adjustments that include a 10 percent increase for alleged undercounting of livestock in official statistics. In what follows, I will outline conceptual problems with each of these adjustments.

1. According to IPCC guidelines on GHG inventory accounting methods ,”CO2 emissions from livestock are not estimated because annual net CO2 emissions are assumed to be zero – the CO2 photosynthesized by plants is returned to the atmosphere as respired CO2.” (pg. 10.7)

Goodland and Anhang add animal respiration to the WRI estimates. As a justification, Goodland and Anhang note: “The FAO asserts that livestock respiration is not listed as a recognized source of GHGs under the Kyoto Protocol, although in fact the Protocol does list CO2 with no exception, and “other” is included as a catchall category.” This claim misunderstands the Kyoto Protocol; Annex I countries are required to use IPCC Guidelines in constructing their reported inventories. While the FAO may have clumsily explained why they didn’t add their own estimate of respiration to the WRI figures, they are correct in not doing so. As the inventories are measuring *net* emissions, a proper accounting requires measuring both sources (respiration) and sinks (photosynthesis of feed crops and pasture). The convention adopted by the IPCC assumes these are equal, and consequently inventories can be estimated without individually establishing all the sources and sinks related to photosynthesis of crops and subsequent respiration of this material. Adding respiration, as Goodland and Anhang do, either amounts to double-counting respiration or omitting some CO2 sinks.

“Livestock’s Long Shadow” provides an estimate of animal respiration of 3,161 million tons (pg.97) Goodland and Anhang apparently missed this; instead they misuse a “back-of-the-envelope” estimate from a one-page editorial by physicist Alan Calverd to obtain an estimate of 8,769 Mt, which is nearly three times as large as the FAO’s estimate.

Calverd estimates that 21 percent of the *CO2* emissions  from the combination of livestock respiration, human respiration, and fossil fuel consumption (not all GHGs, as mistakenly reported and used by Goodland and Anhang) are from livestock. In particular, although Calverd provides no sources for his numbers, he uses the following numbers (i) per capita energy consumption of 1500W,(ii)  an average human mass of 75 kg and (iii) average per capita meat consumption of 75 kg, which he argues requires three time the mass of standing livestock (225 kg) to sustain. Using a metabolic rate of 2 W/kg, the combined power consumption driving emissions from the energy sector, human respiration and animal metabolism is 2100 W. Assuming emissions are proportionate to power consumption, animal respiration accounts for 21 percent (or 450/2100) of the combined CO2 emissions from power consumption and human and livestock respiration.

From Table 3.1 of the WRI’s “Navigating the Numbers,” 58.8 percent of emissions are CO2 emissions from energy. Indeed, in the document itself, it is noted that “energy related emissions account for about 60 percent of the world total” of GHG emissions. Based on the value of 41, 755 Mt, this is 24,552 Mt. Using the Calverd approximation, livestock respiration would be 450/1500, or 30 percent of this number, which would be 7,366 million tons. The world human population in 2000 was 6.1 billion. Calverd’s estimate implies that the standing population of livestock in 2000 had a mass of 1372 million tons. In constrast, the FAO’s estimate of the standing biomass of livestock in 2002 was 699 million tons (FAO 2006, pg. 95).

Despite working with a respiration estimate based on a biomass tonnage estimate that is nearly twice as high as the FAO’s own estimate, Goodland and Anhang still increase their own estimate to account for subsequent (out of period) increases in livestock tonnage and alleged undercounting of animals by the  FAO (more on this below).

Eventually Goodland and Anhang, in response to queries about how 22 Gton of emissions were missed in previous inventories, indicate in a follow-up that “If respired GHGs are counted as a proxy for foregone carbon absorption, then most of the 22 billion tons of emissions that we claim were previously not counted can be understood as a potential carbon sink rather than an actual carbon source.” However, as noted above, LLS is attempting an attribution of actual emissions (i.e. what percent of estimated net emissions in a given year are attributable to livestock). If Goodland and Anhang wish to reinterpret their results in this fashion (their original article already uses a bio-fuel scenario as a conservative estimate of sequestration potential), then a more accurate presentation of their findings would be that livestock accounts for a certain percentage of actual emissions, while *potential* exists through dietary changes to offset some of the remaining emissions by allowing land used previously to feed livestock to revert to natural vegetation (e.g. forest). However, much of this potential involves reversing past emissions due to land-use change. Furthermore, in the WRI GHG inventories, such sequestration would appear as negative emissions at the time the sequestration is realized. In any case, unrealized sequestrations are not treated as emissions in the accounting exercises attached to climate treaties, and doing so amounts to double-counting.

2. By their own admission (pg.14), Goodland and Anhang only “recalibrate” emissions from the 37 percent of anthopogenic methane emissions from livestock using the higher 20-year global warming potential, noting “Further work is needed to recalebrate methane emissions other than those attributable to livestock products using a 20-year timeframe (pg.14).” Given that this “recalibration” involves a trivial calculation involving numbers readily available to Goodland and Anhang, this will result in under-counting of emissions from other sectors, inflating the share of emissions attribute to livestock. The adjustment is trivial…if applying the 20 year GWP to methane from livestock results in an additional 5,047 Mt CO2e, then applying the same conversion to the 63 percent of methane from sectors other than livestock will further increase emissions by 5,047 * (63/37) = 8,594 Mt CO2e. By itself, applying the 20 year GWP of methane to the FAO estimate will increase the share of GHGs attributable to livestock to ((7,516+5,047)/(41,755 + 5,047 + 8,594))*100% = 22.7 percent of the total.

In a follow-up article, Goodland and Anhang attempt to explain this omission, noting:

“Because we questioned many aspects of the FAO’s work, we were reluctant to use their figures for methane, but did so anyway for livestock methane because we couldn’t find a more reliable figure. Yet we remained uncomfortable using the FAO’s figure for non-livestock methane.” (pg. 8)

Recall, however, that the FAO isn’t estimating total emissions (or the subtotals of the various GHGs involved). They are using estimates from the WRI. Non-livestock methane emissions are simply the difference between total emissions of methane and the amount attributed to the livestock sector. Goodland and Anhang never identify their issues with these estimates.

In the same follow-up, Goodland and Anhang try to justify this omission, noting:

“We discovered after publication that the FAO’s own statistical division reported 56 billion livestock worldwide in 2007. This is many more than are counted in our article and doubtless outweighs whatever the increase would be in non-livestock methane.” (pg.8)

“Livestock’s Long Shadow” uses estimates of live animals (i.e. the number of animals alive at a point in time), and Goodland and Anhang apparently understand this, given that they cite the appropriate FAO statistics amid their sources. However, 56 billion refers to the number of animals raised and slaughtered in the course of the year. The discrepancy between the two figures is largely related to the brief lives of broilers.  The IPCC considers this issue explicitly, noting in GHG inventory guidelines that

“Broiler chickens are typically grown approximately 60 days before slaughter. Estimating the average annual population as the number of birds grown and slaughtered over the course of a year would greatly overestimate the population, as it would assume each bird lived the equivalent of 365 days. Instead, one should estimate the average annual population as the number of animals grown divided by the number of growing cycles per year.” (pg. 10.8)

3. Goodland and Anhang adjust emissions to account for increases due to livestock tonnage between 2002 and 2009 (pg.14). This would be a reasonable thing to do, *if* they were either doing an attribution of the emissions for 2009 or updating *all* emissions to reflect more current values. However, as noted previously, their emission figure of 41,755 million tons is for the year 2000. According to the WRI, energy related emissions increased by 21.7 percent between 2000 and 2009. Adjusting only for increases in livestock will inflate the share of emissions attributable to the sector. Furthermore, ignoring the inappropriateness of including respiration, the estimate obtained using Calverd’s method is not based on the FAO’s livestock estimates, so there is little reason to increase this number accordingly.

4. Goodland and Anhang further increase their numbers by 10 percent , claiming that livestock populations are larger than those reported in the FAO study.  In particular, Goodland and Anhang claim, “The report states that 21.7 billion head of livestock were raised worldwide in 2002, while many nongovernmental organizations report that about 50 billion head of livestock were raised each year in the early 2000s.”(pg. 14) When pressed on the issue of why the 50 billion figure from nongovernmental organizations is plausible, Goodland and Anhang reply, “After our article was submitted for publication, we discovered that the FAO’s own statistical division says that there were 56 billion livestock in 2007.” Once again, we find them conflating animals slaughtered in year with live animal estimates at a point in time.

The 21.7 billion figure comes from Table 3.6 of LLS, which reports the following populations for the year 2002 (FAO 2006, pg.96):

million head
Cattle and buffalo 1 496
small ruminants 1 786
Camels 19
Horses 55
Pigs 933
Poulty 17 437
Total 21 726

From the discussion surrounding the table, these numbers are being used to estimate the standing biomass of livestock, and are estimates of the population alive at a point in time, not annual production figures. These can be confirmed directly from FAOSTAT (accessed on January 21, 2015) — the reported figures under Production/live animals for “world + (total)”, “stocks” for the years 2002 and 2007 are:

2002 (million head) 2007 (million head)
Cattle and Buffaloes 1 496 1 611
Sheep and Goats 1 826 2 066
Camels 21 25
Horses 57 60
Pigs 864 919
Poultry 17 760 20 312
Total 22 690 25 964

The numbers available in FAOSTAT may differ from those reported in LLS due to subsequent revision of estimates. Including the remaining reported categories of animals (excluding beehives), the total population for 2002 (the year used by the FAO in LLS) is 22 690 million heads, which is similar to the figure of 21 700 million heads reported in LLS. As the IPCC guidelines above indicate, these are the appropriate figures for GHG inventory analysis. For large ruminants (Cattle and Buffalo), which are the major emitters of methane, the figures are identical. Goodland and Anhang “only” increase their estimate by 10 percent for “undercounted” animals  However, as the above tables indicate, there was no significant undercounting of animals (the totals differ by less than 5 percent), so either Goodland and Anhang have again conflated estimates of live animals with estimates of animals slaughtered, or they are using numbers for a different year. Furthermore, the estimates for 2007 are nowhere near 56 billion.

The issue of conflating figures was raised by Herrero et. al.. in a commentary in Animal Feed Science and Technology. In their response, Goodland and Anhang state “whenever we thought that we thought that we might not have been able to interpret correctly FAO numbers (FAO, 2006; FAO, 2007a; FAO, 2007b), we built in an allowance for the lower possible numbers to apply (pg. 7).” However, the above numbers indicate that the 10 percent increase that they made for “uncounted animals” cannot be justified based on numbers from the FAO, and Goodland and Anhang exercise no such care when defending their decision to not adjust methane emissions from human sources outside the livestock sector using the 20-year instead of the 100-year GWP.


Last edited on October 15, 2015 – added revision of FAO estimate using the 20 year GWP for methane.

Tags: Cowspiracy, Goodland and Anhang

2 thoughts on “An Analysis of “Livestock and Climate Change”

  1. Seems Calverd’s estimate was based on some inflated values. According to wikipedia, the average human now weighs ~62 kgs, not 75kg. According to ourworldindata, the average amount of meat consumed per year per capita in 2014 was 43kgs, not 75kgs, and was certainly less than that in 2000. So, even if Calverd’s assumption that standing livestock biomass is 3x greater than annual consumption were true, that would mean a standing biomass of ~780 million tons, not 1,372 million tons. And if the remainder of his analysis is correct, then the consequent total for livestock respiration would be ~5 gigatons, not 8.8 gigatons. The actual figure for the year 2000 would be less than 5 gigatons because meat consumption increased from 2000 to 2014. Moreover, the revised estimate would be much closer to my own off-the-cuff estimate, undertaken in an entirely different fashion. To wit, humans exhale about 6 liters of air per minute, of which roughly 4.5% is CO2. Livestock biomass is roughly twice that of human biomass and respires at about the same rate. If so, we have the following calculation string…

    6 liters/min x O.012 kg of air/liter x 4.5% CO2 x 44grams/mole of CO2/~29 grams/mole of air 524k mins/year x 6.1b people in year 2000 x 2 kgs livestock per kg of human = ~3.2 gigatons CO2 emissions by livestock per year, which, for ballparking, is pretty damn close to the FAO estimate of ~3.0 gigatons!


    • Hi Ray. Thanks for your comment. Your figures and calculations seem reasonable to me. I did come across another back-of-the envelope estimate that’s closer to the one that Goodland and Anhang achieved by (mis-)using Calverd’s method on the Skeptical Science blog (see comment #58 by “saileshrao”):

      I view the exact estimate of respiration to be tangential in the context of the accounting exercise that’s being attempted by Goodland and Anhang, in so far as that it’s only one half of the net flux.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s