2018 ESA Annual Meeting (August 5 -- 10)

PS 40-54 - Topography mediates the effects of a severe frost in a tropical dry forest

Thursday, August 9, 2018
ESA Exhibit Hall, New Orleans Ernest N. Morial Convention Center
Adrian Bojorquez1, Juan C. Alvarez-Yepiz2, Angelina Martinez-Yrizar1 and Alberto Burquez1, (1)Instituto de Ecologia, Universidad Nacional Autonoma de Mexico, Hermosillo, Mexico, (2)Instituto Tecnologico de Sonora, Mexico
Background/Question/Methods

Extreme climatic and weather events are increasing in frequency in many regions of the world and often causing widespread tree mortality in forest ecosystems. However, little is known about the local environmental factors modulating the effects of such events. In February 2011, temperatures dropped below freezing for at least three consecutive days in northwestern Mexico. This freezing event triggered tree mortality and forest die-off primarily in the lowland tropical dry forest (TDF), where the hilly landscape is composed of mature and secondary forests. Our objective was to analyze the local environmental drivers of frost-induced TDF vegetation damage. Based on density of dead woody trees and branches we selected forest sites with and without visible frost damage (D, binary variable) in our 28 x 28 km study area. Environmental variables included elevation (A), slope steepness (S), northness (N), eastness (E) and forest type (FT= mature, secondary). After accounting for spatial autocorrelation, 232 sites were retained for further analysis. We fitted logistic regression models to assess the effect of topography and FT on the presence/absence of frost-induced vegetation damage. The best statistical models (selected with AIC) were incorporated into a GIS to predict the probability of frost-induced TDF damage in our entire study area

Results/Conclusions

The best model included A and E as explanatory variables of frost damage. The second and third models included N and FT in addition to A and E, respectively. R2 was similar for all three models (0.44) and area under the ROC curve (AUC = 0.78) indicated that all models discriminate well between sites with presence/absence of damage. Forest tree mortality was strongly influenced by elevation. Our modeling identified an elevation threshold of 430 m below which vegetation damage was most likely to occur. The three best models predicted a higher probability of frost damage in forests at low elevation. This is likely related to the surface movement of the arctic cold front that entered from the northeastern mountains and drained into the lowlands, causing the severe frost. Our analysis shows that topography and particularly elevation can mediate the effects of severe frosts on the lowland TDF. This information can be useful to design better forest management practices and recovery plans under a future scenario of more frequent extreme climatic and weather events.