CANCELLED - Capturing vertical structural dynamics with lidar remote sensing from chronosequences of coastal Florida shrubland and forest communities recovering from prescribed fire

Background/Question/Methods

Disturbance profoundly affects vegetation in many terrestrial ecosystems and numerous studies have documented changes in the vertical structure of vegetation recovering from fire and other types of disturbance. However, most of these studies relied on field methods for data collection, limiting their spatial scale to only a few plots. In this study, we used discrete-return lidar remote sensing to characterize changes in the vertical structure of vegetation in a mosaic of oak scrub and wetland hardwood forest ecosystems recovering from prescribed fire. Our study area consisted of two independent sites, separated by over 1.5 km, in the Kennedy Space Center/Merritt Island National Wildlife Refuge complex on the east-central coast of Florida. Each site was ≈5 km²  and comprised of patches with time since fire (TSF) values of 1, 8, 14, 22, or >27 years. We used  non-ground lidar returns to calculate median height, shrub layer density (returns < 3 m high), canopy layer density (returns ≥ 3 m high), and mean cover in each patch and then averaged these metrics to the site level. We then performed linear, quadratic, and cubic regressions of all metrics against TSF to identify the best-fitting models for each vegetation type.

Results/Conclusions

For both oak scrub and wetland hardwood forest vegetation, we found significant relationships between TSF and all four metrics of vertical structure. Models with quadratic terms for TSF explained the most variation in oak scrub median height and mean cover (R²_adj = 0.916 and 0.659, respectively); models with cubic terms explained the most variation in shrub layer and canopy layer return density (R²_adj = 0.903 and 0.937, respectively). For wetland hardwood forest vegetation, linear models explained the most variation in median height and shrub layer density (R²_adj = 0.726 and 0.591, respectively); cubic models explained the most variation in canopy layer density and mean cover (R²_adj = 0.970 and 0.876, respectively). Our results suggest that the vertical structure of oak scrub undergoes non-linear changes when recovering from prescribed fire, while wetland hardwood forest exhibits more linear responses. This may be attributable to the different post-fire dynamics of these two ecosystems, because oak scrub is adapted to a fire return interval of 5-15 years while that of wetland hardwood forest exceeds the range of TSF in the study area. We believe that discrete-return lidar holds enormous potential for characterizing the vertical structural dynamics of vegetation recovering from disturbance at landscape scales.