Background/Question/Methods The recent availability of Light Detection and Ranging (LiDAR) technologies has allowed researchers to quantify canopy structural metrics at unprecedented high resolutions. Novel work has found canopy structural complexity (CSC) to be a predictor of important forest functions like carbon sequestration and biodiversity. However, we still lack an understanding of the determinants of CSC itself, especially in understudied areas like tropical forests and at larger spatial scales. Understanding ecosystem characteristics that underpin CSC is therefore key to shedding light on mechanisms of carbon assimilation, diversity maintenance, and disturbance response. Here, we use USGS LiDAR measurements in Puerto Rico to ask, what are the determinants of three CSC metrics - canopy height, canopy rugosity, and canopy gap fraction - in tropical forests? To address the different scales of potential drivers - stand, landscape, and climatic - we randomly sampled 16,000 forested areas across soil types, stand ages, topographic and climatic gradients in Puerto Rico. We used PRISM and Global Wind Atlas 3.0 to quantify mean annual precipitation and wind exposure, a detailed land-use map to quantify forest type, and a DEM to calculate elevation and slope. To test the relative importance of each driver, we ran a Random Forest model for each CSC metric as a response variable with 7 predictors: stand age, forest type, soil type, elevation, slope, mean annual precipitation, and chronic wind exposure, accounting for interactions using Friedman’s H factor. Results/Conclusions Our models explained 40% of the variance in canopy height and 38% in canopy rugosity, but 11% in gap fraction across the forests of Puerto Rico. Precipitation, forest age, chronic wind exposure and elevation were the most important predictors of canopy height and rugosity. Canopy height and soil type were the most important determinants of gap fraction or openness. Our combined results demonstrate that canopy height and rugosity increase with precipitation and plateau after 2,000mm-yr and decrease linearly with elevation; older forests are taller, less open, and less rugose; and there is a strong interaction between forest age, elevation, and precipitation. Finally, chronic wind exposure reduces canopy height and homogenizes canopy structure, an effect that is consistent after accounting for other effects. Our results utilize novel LiDAR and remote-sensing data to shed light on drivers of CSC at an island-wide scale, with important implications for carbon cycling in the tropics. However, future work is necessary to disentangle the many interaction effects elucidated by our models.
