Allometric models to estimate leaf area for Tropical African broadleaved forests

Direct and semidirect estimations of leaf area (LA) and leaf area index (LAI) are scarce in dense tropical forests despite their importance in calibrating remote sensing products, forest dynamics, and biogeochemical models. We destructively sampled 61 trees belonging to 13 most abundant species in a semideciduous forest in southeastern Cameroon. For each tree, all leaves were weighed, and for a subsample of branches, leaves were counted and the LA measured. Allometric models were calibrated to allow semidirect estimation of LAI at tree and stand levels based on forest inventory data (R-2 = 0.7, bias = 21.2%, error = 39.5%) and on predictors that could be extracted from very high resolution remote sensing data (R-2 = 0.63, bias = 35.1%, error = 58.73). Using twenty-one 1-ha forest plots, stand level estimations of LAI ranged from 4.42-13.99. These values are higher than previous estimates generally obtained using indirect methods. These results may have important consequences on ecosystem exchanges and the role of tropical forest in global cycles. Plain Language Summary Leaf area (LA) and leaf area index (LAI) are useful parameters characterizing the plant-atmosphere interface where matter and energy are exchanged. However, direct or semidirect estimations are not common in dense tropical forests. In this study, we used a destructive data set of trees of varied species and sizes from the semideciduous forest of southeastern Cameroon to predict total tree LA. Based on this data, we developed operational allometric models to allow for semidirect estimation of LA and LAI at tree and stand levels. These models would be of considerable use for climate-vegetation modeling and remote sensing communities