Low-severity winds reduce tropical forest structural complexity regardless of climate, topography or forest age

Forests are often exposed to regular, non-severe winds (chronic wind exposure), yet the effect of such winds on canopy structure in tropical forests remains understudied. The height and structural complexity of a forest canopy are strongly and positively correlated with biodiversity and carbon accumulation. Understanding the drivers of canopy structural complexity across broad environmental gradients can therefore improve the mapping and modeling of diversity and carbon dynamics. Here we predict the height and structural complexity of forests in the heterogeneous island of Puerto Rico, with a particular focus on the impacts of chronic wind exposure. To do so, we used remote sensing to randomly sample ~20,000, 0.28 ha forested sites stratified by forest age, and used airborne LiDAR data from 2016 to quantify canopy height and a key metric of structural complexity, rugosity – the standard deviation in canopy height. We then ran random forest models to predict canopy height and rugosity ba..., , We measured canopy height and rugosity for ~20,000 30 m-radius forested sites in Puerto Rico using airborne LiDAR data from a 2016 flight collected by the USGS (Carswell Jr. 2016) with aggregate nominal pulse spacing of ≤0.35 m and pulse density of ≥8.0 pulses/m2. We then used 16 years of wind direction and speed data from 9 weather stations to calculate a map of chronic wind exposure using a 30 m DEM from the USGS. We then ran random forest models to predict drivers of canopy height and rugosity based on chronic wind exposure, mean annual precipitation, forest age class, elevation, slope, soil type, soil available water storage, and exposure to two previous hurricanes (Hugo in 1989 and Georges in 1998)., # Drivers of tropical forest height and canopy structural complexity across heterogeneous landscapes:

---

This dataset was used for analyses and figures in Ankori-Karlinsky et al., 2023 in *Ecosystems*.

The dataset contains 20,660 30 m-radius forested sites in Puerto Rico stratified by forest age class (see Martinuzzi et al., 2020 for details).

For each site, we quantified canopy height and rugosity using 2016 airborne USGS LiDAR data.

The dataset was used to predict both metrics based on forest age, exposure to chronic winds, precipitation, topography, soil properties, and exposure to previous hurricanes.

## Metadata

Each row is a 30 m-radius site.
Below is information on each column:
*x* - longitude in NAD83 (EPSG 4269)
*y* - latitude in NAD83
*Forest_Age* - Forest age class (1-5):
* 5-16 years
* 17-25 years
* 26-39 years
* 40-65 years
* 66+ years
*CHM* - Mean height (m) from canopy height model
*Max_Height* - Maximum height (m)
*Rugosity* - Stdev in height from 15m moving...