Creating virtual forests to understand fragmentation in tropical ecosystems

The study of forest fragmentation, the break-up of forests into smaller patches, has become increasingly important due to increases in human-induced forest clearance, with 12 million hectares of forest being lost per year and 32% of this loss being tropical. There is substantial evidence showing that edge effects can alter the ecology structure and vertical profile of remaining forests, even hundreds of meters from the forest edge.  However, implementing empirical experiments (for example in the framework of the Biological Dynamics of Forest Fragments Project (BDFFP)) to understand the effects of fragmentation on forest structural traits can be logistically and scientifically challenging and limited to smaller areas.  The use of forest models may help overcome these limitations, as they are able to quickly reproduce long-term data, as well as simulate a broad range of geographical conditions. This study aimed to reproduce the vertical distribution of plants in Amazonian forests affected by fragmentation using the forest model FORMIND. To achieve this, we optimized parameters driving plant demography and mortality, as well as their response to edge effects. FORMIND is an individual and process-based gap model suited for species rich vegetation communities, with the option of a fragmentation module. We modified processes and parameters in FORMIND to mimic the dynamics observed in the BDFFP experiment. Forest structural traits extracted from the FORMIND model output were compared with those obtained from terrestrial laser scans of the BDFFP fragments. The resulting simulations demonstrated that, after 40 years of edge effects, the optimized model was able to reproduce similar results to those observed using the terrestrial LiDAR system. Total plant area index (PAI), and PAI at varying height intervals (PAI 0-10m, PAI 10-20m, PAI 20-30m), showed consistent responses from edge effects, thus resulting in an adequate vertical plant distribution. Results demonstrate that, forest models such as FORMIND have strong potential to study the mechanisms and the impact of environmental changes on forests. Models can also expand the possibilities of in-situ studies, which are limited in time and space, when calibrated carefully with suitable in-situ data, here delivered by terrestrial LiDAR.