This study aims to analyse the consequences of elevated tree mortality on forest dynamics and to provide a methodology that can reduce uncertainties in estimating biomass mortality rates at local and country level. We achieved this by linking benefits of individual-based forest model-ling, statistical linear regression, and remote sensing. We applied an individual-based forest model to investigate the impact of varying disturbance regimes on the succession dynamic of a humid Terra Firma forest at the Paracou study site in French Guiana. By simulating increased tree mortality rates, we were able to investigate their influence on several forest attributes, namely biomass, leaf area index, forest height, gross primary production, net primary production, and biomass mortality. Based on simulations of leaf area index and forest height, we developed a linear multivariate regression model to project biomass mortality.

Our findings demonstrate that severe disturbances altered the succession pattern of the forests in favour of fast-growing species, which changed gross primary production, but net primary production remained stable. We also observed a strong influence on biomass mortality rates as well as observed complex relationships between these rates and single forest attributes (leaf area index, forest height, and biomass). By combining leaf area index and forest height we obtained relationships that allow an estimation of the biomass mortality. Based on these findings, we mapped the biomass mortality for whole French Guiana. We found a nation-wide biomass mortality of 3 % per year (standard deviation = 1.4 % per year).

The approach we describe here, provides a novel methodology for quantifying the spatial-temporal distribution of biomass loss, which has recently been identified as particularly critical for monitoring mortality hot spots. Quantifying biomass mortality rates may help reducing uncertainties in the terrestrial component of the global carbon cycle.