Assessment of two biomass estimation methods for aquatic vegetation growing on the Amazon Floodplain

Abstract

Studies of macrophyte productivity in the Amazon region are limited by accessibility and costs; hence, they may suffer from reduced sample size and representation. The present study compares a phenometric (indirect) method and a subsampling (direct) method in terms of accuracy and applicability to estimation of aquatic macrophyte biomass in the Amazon. The results show that phenometric models were not as effective as selective subsampling for the estimation of macrophyte biomass under the studied conditions. Phenometric models performed more acceptably for predicting emergent biomass, and less for submerged and total biomass (r2 = 0.77, p < 0.05, RMSE = 200–600 g/m2 dry mass). Improvements in r2 by using species-specific phenometric models were mostly not significant. Phenotypic variation across the studied region was large enough to preclude the generalization of phenometric relationships into accurate numeric models, while the direct subsampling method was able to account for this variation (RMSE < 500 g/m2 dry mass). Subsampling also allowed a significant reduction on the physical effort of biomass sampling, which directly translated into wider and more complete sampling. We suggest that direct subsampling presents the best trade-off between accuracy and coverage for macrophyte biomass measurement in the Amazon floodplain.