US-IALE 2018 has ended
Back To Schedule
Wednesday, April 11 • 10:15am - 10:30am
HABITAT FRAGMENTATION/CONNECTIVITY: Understanding Land-cover Change and Its Impact on Biodiversity in an Ecuadorian Fragmented Landscape

Sign up or log in to save this to your schedule, view media, leave feedback and see who's attending!

AUTHORS: Xavier Haro-Carrión*, Bette Loiselle, Jane Southworth – University of Florida

ABSTRACT: We aimed to investigate how land-cover affects tree diversity in a fragmented landscape ( 19,000 ha) of tropical dry forest in coastal Ecuador. Our objectives were to understand the extent of land-cover types and to evaluate the impact of land-cover and forest fragmentation on tree species diversity. Landsat 2013 (dry season) and 2015 (wet season) images were used to classify the area into old-growth forest, plantation forest (mostly teak), secondary forest, pastureland, water, and built areas. We used a Random Forest Classifier (RF) with 94 predictor variables that included Landsat bands, band ratios, vegetation indices, and image transformations for each season. The proportion of correctly classified RF training points was 95%, and Kappa index was 94%, indicating high accuracy. Dry season variables were predominantly used to classify pastureland and forests. About equal numbers of dry and wet season variables were used to classify forestry plantations. We used the generated map and tree species inventories of old growth forest (N=8) and secondary forest (N=13) to evaluate the impact of land-cover and forest fragmentation on tree `diversity. First, we use RF to identify key patch, landscape composition, and plot variables that contribute to tree diversity as indicated by Hills Numbers. Results indicate that land management variables (e.g. proximity to plantations) are important indicators of species diversity numbers in old growth and secondary forests. We also tested the effect of patch, landscape, plot and species traits on endemic tree species occurrence using a generalized linear mixed model (GLMM). The GLMM used wood density as the only significant predictor. However, distance to road and land-cover successfully helped predict endemic tree species occurrence contingent to species, indicating a differential response of species to human disturbance.

Wednesday April 11, 2018 10:15am - 10:30am CDT
Hancock Parlor