全球变化知识资源中心

Land use in the GS.

To analyse the precise assumptions of at least one prominent analysis, our identification of the wet savannahs of Africa tracks the map used by the World Bank to identify the GS. That generally identifies lands with a potential crop-growing season based on adequate soil moisture between 150 and 239 days per year (roughly areas with >600 mm rainfall per year that are not dense forests). In reality, this definition extends the area labelled GS, which properly refers to a particular ecosystem in West Africa, to a wide range of savannahs, shrublands and woodlands. We calculate that 51% of the 718 Mha area has canopy cover of 10–30%, 33% has canopy cover of 30–50%, and 3% has canopy cover over 50%. Wetlands cover 47 Mha (6%), and protected areas cover 106 Mha (Fig. 1), but croplands already cover 82 Mha (11%). On the basis of an existing database²¹, 260 Mha of the GS was used for pasture in 2000, but the densities vary greatly with 3.5% of pasture in excess of 50 tropical livestock units (TLU) km⁻²; 33.5% of pasture with 10–50 TLU km⁻² and 63% of pasture with 0–10 TLU km⁻². Some of the studies we cite exclude some more managed grazing lands from their estimates of potentially suitable lands.

Potential to be a low-carbon source of staple crops.

We analyse the potential of additional cropland in the GS to be a low-carbon source of staple crops first by estimating the carbon conversion efficiencies of maize or soybeans on existing global cropland. (The World Bank found that maize and soybean are the optimal staple crops for 88% of the suitable potential new cropland in SSA (ref. 2, and see Supplementary Information).) Studies of agricultural conversion costs typically focus on carbon releases per hectare²², but if crop yields are low, using land with little carbon can result in more hectares of conversion and more overall release of carbon. Here, we focus instead on the carbon releases from land conversion per ton of crop because that precisely measures land’s ability to contribute to food needs while minimizing total carbon releases. Following ref. 23, we calculate these efficiencies as the carbon lost by the conversion of native ecosystems to cropland divided by the current annual yields of those croplands, so the lower the number the more efficient. Unlike ref. 23, however, we analyse these ratios for individual crops rather than aggregate crops. The different yields of different crops will lead to different carbon loss/yield ratios. The ratios must be compared for the same crops to properly reflect the differences in land characteristics alone.

Using methods described below, we find that lands used for maize globally have experienced a mean per hectare average carbon loss of 20.8 tons per ton of annual crop yield (tC tY⁻¹), and a median of 14 tC tY⁻¹. For soybeans, the mean ratio (carbon conversion efficiencies) is 44.5 tC tY⁻¹ and the median ratio is 41 tC tY⁻¹.

We compare these global conversion efficiencies with estimates of carbon release per potential ton of rain-fed maize and soybeans on wet savannahs in the GS while excluding existing cropland and protected areas. To provide sensitivities for our analysis, we base our spatial estimates of carbon first on soil and vegetation carbon maps, and alternatively on the same vegetation and soil model used for the global analysis (LPJmL; Supplementary Information).

For yields, our first method estimates potential yields optimistically assuming high inputs and absence of major crop diseases for the whole GS using a crop model (DSSAT; Supplementary Information; Supplementary Figs 1–4 map and show distributions for this approach). We alternatively estimate potential yields using a yield gap analysis that estimates the 90th percentile of reported yields within zones of comparable climate. The two methods generate quite similar patterns of yield estimates overall, although the DSSAT yields are modestly lower (Supplementary Table 1), and there are spatial differences (Supplementary Fig. 6). The two estimates for potential yields and the two estimates of carbon losses generate four different estimates overall of carbon conversion efficiencies.

According to these four estimates, 18.3–19.2% of the GS has potential to be converted to maize while releasing less carbon per ton of crop than the global mean. The potential areas for soybeans are 30.8–32.9%. Conversion of only 6.3–7.8% (maize) and 12.1–16.2% (soybeans) would release at least one-third less carbon than the global mean (Supplementary Figs 7a and 8), which we consider a modest standard for ‘low’ carbon costs per ton.

As a minimum level of yield and acceptable yield variability are needed to justify high inputs, we calculated areas in the GS that also meet three modest practicability tests for high inputs, leading to four total suitability criteria for potentially low-carbon cropland: carbon loss/yield ratios for maize or soybeans under our optimistic assumptions of potential yields would be at least one-third lower than the world mean; potential yields would reach at least 4 t ha⁻¹ yr⁻¹ for maize or 1.5 t ha⁻¹ yr⁻¹ for soybeans, roughly half the yields of high-exporting countries; the yield coefficient of variation (CV) would be less than 30%; the cropland season failure rate would be less than 10%. Using our crop-modelled yields, 2–2.2% meet criteria for maize and 9.5–11.5% for soybeans. Figure 2 shows lands that pass these tests using the DSSAT/database carbon method. The Supplementary Information includes statistics for different combinations of criteria (Supplementary Tables 2 and 3). Analyses that use attainable yields from the yield gap study²⁴, which are methodologically restricted to the first two ‘suitability’ criteria, produced estimates of 1.2–3.6% (maize) and 10.5–12% soybeans (Supplementary Tables 4 and 5). These similar estimates suggest limited practical capacity to generate crops with low carbon costs.

1. Alexandratos, N. & Bruinsma, J. World Agriculture Towards 2030/2050: The 2012 Revision (ESA Working paper Rome, FAO, 2012); http://go.nature.com/fM9kF2
2. Deininger, K. & Byerlee, D. Rising Global Interest in Farmland: Can it Yield Equitable and Sustainable Results? (World Bank, 2011).
3. Lambin, E. F. & Meyfroidt, P. Global land use change, economic globalization, and the looming land scarcity. Proc. Natl Acad. Sci. USA 108, 3465–3472 (2011).
   • PubMed
   • Article
4. Hoogwijk, M., Faaij, A., Eickhout, B., de Vries, B. & Turkenburg, W. Potential of biomass energy out to 2100, for four IPCC SRES land-use scenarios. Biomass Bioenergy 29, 225–257 (2005).
   • ISI
   • Article
5. van Vuuren, D. P., van Vliet, J. & Stehfest, E. Future bio-energy potential under various natural constraints. Energy Policy 37, 4220–4230 (2009).
   • ISI
   • Article
6. Cai, X., Zhang, X. & Wang, D. Land availability for biofuel production. Environ. Sci. Technol. 45, 334–339 (2011).
   • CAS
   • ISI
   • PubMed
   • Article
7. Chum, H. et al. in IPCC Spec. Rep. Renew. Energy Sources Clim. Change Mitig. (eds Edenhofer, O. et al.) (International Panel on Climate Change, 2011).
8. Bauen, M. Bioenergy-A Sustainable and Reliable Energy Source. A Review of Status and Prospects (International Energy Agency, 2009); http://go.nature.com/QX3o1m