Quantifying canopy traits is central to understanding the influence of tree species diversity on forest productivity. Recent results from field experiments manipulating tropical tree diversity suggest positive effects on productivity, but results from temperate forests are lacking. Studies in managed, mature forests suggest increased canopy packing with increasing diversity is an important mechanism. However, experimental tests for relationships between productivity, canopy packing, and LAI are rare. Furthermore, despite promising results for UAV remote sensing of canopy traits from agricultural fields, applications in biodiversity experiments are limited. To address these knowledge gaps, we explore three related questions: 1) Does productivity increase with increasing tree species diversity? 2) Are canopy packing and LAI related to differences in productivity? 3) Can key canopy traits be predicted by UAV remote sensing?
We studied three experiments (645 plots, 22 species) with planted gradients in tree diversity. In 2017, tree crown dimensions were measured to calculate plot-level canopy packing. Plot LAI was estimated from LAI2000 and litterfall. Biomass and ANPP were calculated from annual height and diameter data. UAV-based hyperspectral imagery was collected and 3D hyperspectral orthomosaics generated, permitting calculation of plot vegetation indices and Partial Least Squares Regression (PLSR) models of plot responses.
Results/Conclusions
Preliminary results suggest productivity increased with tree diversity. Crown packing significantly increased with diversity, with a ~70% correlation between the degree of crown packing and productivity. However, community LAI estimated from LAI2000 measurements did not significantly increase with diversity and were not positively related to canopy packing as expected. Canopy traits were better predicted by PLSR than by vegetation indices as expected. PLSR outperformed random forest regression, even when applying pre-modeling dimensionality reduction. Initial results show that full bandwidth PLSR models produced cross validated R2 values of ~50% for canopy nitrogen, ~65% for standing biomass, and ~60% for LAI.
We conclude that 1) tree diversity likely plays an important role in influencing forest productivity and that crown complementary is likely important for understanding diversity-productivity relationships, 2) UAV remote sensing shows promise for scale-appropriate estimation of canopy traits and plot responses that are central to understanding tree diversity-productivity relationships.