Baseline soil carbon, time since conversion and climate, not system characteristics, shape soil carbon responses in Neotropical agroforestry. A meta-analysis

Abstract Agroforestry is promoted as a nature-based climate solution. Yet, the size of soil organic carbon gains and the factors controlling them remain uncertain, particularly in the tropics, where much of the evidence lies in grey and non-English literature. Here we analysed 322 response ratios and 277 storage rates from 54 Neotropical field studies with three-level meta-analyses, meta-regressions and gradient-boosted trees. Agroforestry increased soil organic carbon stocks by 6.8% (95% CI: 1.7−12.2%), or 0.25 (0.05−0.45) Mg C ha - ¹ yr - ¹. The large heterogeneity in the response was mostly explained by biophysical conditions (16 to 23%), whereas agroforestry type, previous land use, number of tree species and crop contributed little. Relative gains declined with increasing initial soil carbon, from 19% at 20 Mg C ha - ¹ to 1.3% at 90 Mg C ha - ¹, and storage rates followed the same pattern. The absolute amount of carbon added, 2.9 Mg C ha - ¹ (0.3−5.5), did not vary with initial stock. The decline in relative gains was more pronounced at warmer sites. Time since conversion was the second strongest driver: relative gains increased by 0.4% per year, whereas annual storage rates fell from 1.2 Mg C ha - ¹ yr - ¹ at 2 years to 0.1 at 10. Results were the same in English and Spanish publications, and in peer-reviewed and grey literature, and no single study drove the relative gains. Only seven studies went beyond 20 years, and deeper layers were sampled only in young systems. Here we show for the first time that the relative soil carbon response of Neotropical agroforestry is governed by the initial state of the soil and by climate, not by the system typologies used in carbon accounting. Targeting carbon-depleted soils maximises the relative gain, not the amount stored, which increased with system age.