Skip to main content

Tree functional traits as predictors of microburst-associated treefalls in tropical wet forests

Cite this dataset

Rader, Alana et al. (2020). Tree functional traits as predictors of microburst-associated treefalls in tropical wet forests [Dataset]. Dryad.


On 19 May 2018 a microburst caused 600 isolated forest gaps in a Costa Rican tropical forest. We surveyed fallen and standing trees within gaps to determine if certain variables are associated with treefalls. Our results highlight considerations for future research to understand the impacts of microbursts in tropical forests. Our results show that at the scale and locality of our study, treefall vulnerability to microbursts and characteristics of fall events are independent of the biomechanical and biophysical measurements that we took. We recommend that future studies on microbursts build from the measurements we took in this study to include observations at the individual tree, forest community, and forest landscape scales.


Treefall observations were sampeled within 20 m of trails at La Selva Biological Staion for safe access. We sampled damaged trees greater than or equal to 30 cm diameter at breast height (DBH), and only surveyed fallen trees that had abundant green leaves in their canopies, to ensure that trees studied were alive at the time of the microburst. For each treefall event, we noted whether it involved one or multiple trees. For multiple treefalls, we measured the largest tree among the fallen. Each treefall was identified as uprooted (main root system visible) or snapped (snapped trunk with root system in the ground). Because stem size is often a predictor of damage probability in strong winds (Everham & Brokaw 1996), each fallen tree was matched to the closest standing tree within 20 m, with an equivalent DBH (± 5 cm). DBH was measured at standard height or immediately above the buttresses. Trees were identified to the species level and assessed for liana presence and buttress size. Lianas were measured as either absent or present in the crowns of both fallen and standing trees. Buttresses were categorized ordinally from 0 to 2, where 0 represented absence of buttresses, 1 indicating buttresses less than or equal to the width of the main trunk on each side, and 2 indicating buttresses wider than the main trunk on each side. We determined wood density of each tree species using the Tree Functional Attributes and Ecological Database (Harja et al. n.d.).  

We evaluated how liana presence/absence impacts tree fate using Fisher’s Exact Test, which calculates an exact P-value for a 2 x 2 table. We evaluated if buttress size impacts treefall using a Cochran-Mantel-Haenszel Test (CMH test), appropriate for comparing ordinal vs. categorical variables. The relationships of wood density and tree diameter to tree fate were examined using paired t-tests. Two logistic generalized linear models within the subset of fallen trees analyzed 1) how the same four predictor variables may influence treefall type, and 2) whether treefalls were single or multiple events.  Continuous data was standardized and z-scores were calculated for wood density (g/cm3) and tree diameter (cm) based on their mean and standard deviation. All analyses were performed in Program R version 3.5.1 (R Core Team, 2018).