Deforestation may increase soil carbon but it is unlikely to be continuous or unlimited

Identifying land uses and management practices that maintain or enhance soil carbon storage are important for sequestering carbon from the atmosphere and improving soil ecosystem services (Herrero et al., 2016). There is debate about how much additional carbon can be stored annually in soil, and for how long, following change in land use or management (Smith, 2014) and resolving this question is important. This is particularly relevant now following the aspirational goal established in the 4 per mille initiative (4p1000, 2017). In a paper titled: “Continuous soil carbon storage of old permanent pastures in Amazonia”, Stahl et al. (2017) purported to show increases in soil carbon when tropical forests are converted to grazed pasture of about 5.3 t C ha−1 year−1 through stock measurements across a chronosequence of sites. They state in the abstract: “… our results show that if sustainable management is applied in tropical pastures coming from deforestation… they can ensure a continuous C storage”, in the discussion: “This result supports the idea that certain ecosystems such as permanent pasture continuously accumulate soil carbon in the long-term (ie several decades).” and later in the conclusions: “Moreover, the humified C stored in the deeper soil layers is likely to provide unlimited C accumulation and conservation of C in the long-term.” (Italics added). Continuous and unlimited accumulation of soil carbon would certainly be of great interest to researchers, policy makers and the broader public as a large contributor to mitigating greenhouse gas emissions, particularly at rates as high as 5 t C ha−1 year−1. However, continuous and unlimited carbon accumulation is not supported by the author's data or the general literature. Using measurements of soil carbon stocks in a chronosequence of sites converted from forest to pasture, they reported no change in soil carbon for the first 24 years followed by increases in soil carbon between 24 and 36 years at 5 t C ha−1 year−1. From these data, it is not reasonable to extrapolate these increases for 12 years as continuous or unlimited without further justification. In the broader literature, it has been widely observed that carbon reaches a plateau in roughly 30 to 50 years or perhaps 100 years after system perturbation as new equilibrium is reached, and that pastures are not perpetual sinks of carbon (Smith, 2014). Soil carbon accumulation rates of greater than 1 t C ha−1 year−1 are also rare (Conant, Cerri, Osborne, & Paustian, 2017). At face value, accumulation of 5 t ha−1 year−1 would also require a high proportion of the gross primary production (GPP) captured as soil carbon rather than lost as carbon dioxide. Assuming a GPP of about 22 t C ha−1 for a high productivity pasture (e.g., Rutledge et al., 2015) would mean that 20%–25% of GPP was stabilized in soil every year for ~12 years. Additionally, by the end of the chronosequence the soil had a C:N ratio of 20 and was gaining of 5 t C ha−1 year−1 which would require a net nitrogen storage in soil organic matter (not in plants) of 250 kg N ha−1 year−1. In the absence of nitrogen fertilizer inputs, gross biological nitrogen fixation rates would need to be in excess of 250 kg N ha−1 year−1 from a reported legume cover of <10%. So what could be an alternative explanation for the trends observed in the data of the authors? Parsimoniously, the chronosequence sites may not have been derived from the same starting point as they were situated some 200 km apart and matching sites across such a wide geographical space is difficult (Walker, Wardle, Bardgett, & Clarkson, 2010). The parent materials of soils were matched by examining geological outcrops on roads. This is a rather limited approach to ensuring the landscape started at the same carbon stocks, particularly as the chronosequence length is only 36 years. Indeed, inspection of the soil texture data in the author's supplemental material had % sand of between 11% and 73% (average 48%, SD 15%), which does not support the contention that these soils were initially the same. Alternatively, these pastures may not have reached a new steady state within the time frame of the chronosequence and rates of accumulation will decrease as this steady state is reached, consistent with all other long-term experiments in the literature (Smith, 2014). Certainly, 12 years is not long enough to determine whether a new equilibrium has been reached. The data presented provide evidence of increased soil carbon stocks when Amazonian forest is converted to pasture, confirming a trend shown by some (Neill et al., 1997) but not all (Don, Schumacher, & Freibauer, 2011). However, labelling any gains observed as a continuous sink of carbon could lead to false hope by land owners and incorrectly inform policy makers. It should further be noted that only about 10% of the total ecosystem carbon lost after deforestation (due to tree removal, burning etc.) can be recovered through soil carbon sequestration (Fearnside, 1997; Neill et al., 1997). It remains critical that we determine rates of long-term carbon accumulation and how long they are sustained if we are to make progress towards the 4 per mille aspirational goal but methodological approaches require careful consideration. Miko Kirschbaum is thanked for providing helpful comments on a draft. The input of PS contributes to RCUK funded projects N-Circle (BB/N013484/1), U-GRASS (NE/M016900/1), GREENHOUSE (NE/K002589/1) and Soils-R-GGREAT (NE/P019455/1).