Résumé
Tropical forests host globally significant carbon stocks, which are relied on to help mitigate the effects of human-induced climate change. Characterizing the uncertainty around carbon stock estimates is thus essential to inform the carbon budgets needed to safely limit global warming, with implications for policy and decision-making worldwide. We used a large forest inventory dataset from the Australian tropical rainforest (7208 stems) to evaluate height:diameter (H:DBH) allometric variation and quantified within-species variation in wood density (WD) and woody tissue carbon ([C]) content (98 trees). Together, H:DBH, wood density, and woody tissue C were used to estimate forest carbon stocks (i.e. the carbon content of woody biomass: CAGB). Using simulations, we then provide estimates and recommendations on the uncertainty around CAGB. H measurement errors contributed more random error (4.5%) to CAGB than the best performing (site-level) H:DBH allometric models (+/- 2.4%) and led to under-estimation of CAGB by roughly 15%. Comparison of H:DBH allometric models that were developed at different biogeographic scales shows that widely used pantropical models substantially over-estimated tree H, and thus tropical forest CAGB, at some Australian sites by close to 100%. By contrast, wood trait variation generated just 3% uncertainty in CAGB, which is reassuring since global wood trait datasets often contain a single record or a few records for tropical species. Our findings reinforce the promise of laser scanning technologies to improve biomass estimation via more accurate measurements of canopy height and nondestructive development of local-to-regional allometric models, which provide clear advantage over pantropical equations.