Habitat structure is a key factor for explaining biodiversity patterns at a local scale. In forests, heterogeneity of the vegetation structure has been shown to have important effects on the distribution and abundance of many animal taxa. Airborne light detection and ranging (LiDAR) allow researchers to study forest structure at a much finer detail than vegetation measurements in the field. Most structure-animal diversity work focused on structural metrics derived from variations of the canopy height, despite the ability of LiDAR to capture 3D structure beneath the top of the canopy. Here, we built LiDAR metrics based on the Leaf Area Density (LAD) at each height layer, and used them to study how the different aspects of forest structural heterogeneity explain bird richness variability. Our goals were to test:1) whether LAD-based metrics explained better the bird richness than metrics based on the top of the canopy; and 2) if different aspects of the structural heterogeneity had diverse effects on bird richness. We used LiDAR data and 61 bird point counts provided by the National Ecological Observation Network (NEON). We then used LiDAR metrics as predictors in random forest models of bird species richness and analyzed the shape of the responses against each predictor.
Results/Conclusions
Random forest models based on LAD metrics had better explanatory power of bird richness (43% of variance explained) than models using top of the canopy metrics (32%). Models that combined both type of metrics (for which feature selection was applied) explained 45% of variance, with the LAD metrics being the highest ranked in explanatory power. The most important metric captured the vertical variation of the vegetation density (measured with a Shannon entropy index) at the forest plot level, followed by a metric that measured the horizontal variation of the mean LAD distribution across 10m x 10m grids within the plot. We showed that structural heterogeneity-richness relationships are generally non-linear and that complex interactions between different aspects of heterogeneity exist. Our key findings were that vertical structural heterogeneity had a negative effect on bird richness, while horizontal heterogeneity had a positive effect.This contradicts previous studies that did not consider all canopy layers or horizontal variations of the vegetation vertical distribution. The increasing capabilities of LiDAR will allow researchers to characterize forest structure with higher detail. Our findings highlight the need for structure-animal diversity studies to incorporate metrics able to capture the different aspects of the forest 3D heterogeneity.