Date Approved

12-2014

Graduate Degree Type

Thesis

Degree Name

Biology (M.S.)

Degree Program

Biology

First Advisor

Gary Greer

Second Advisor

Shaily Menon

Third Advisor

Todd Aschenbach

Abstract

Approximately 94% of Puerto Rico’s forests were converted into agricultural systems by 1950. Since then, extensive abandonment of agricultural land has resulted in a considerable amount of forest regeneration throughout the main island. Ferns are a major non-woody component of oceanic, tropical island forests comprising up to seventy percent of the flora. Consequently, the composition and community structure of ferns may be indicative of the relative richness of these secondary forests. I used Maximum Entropy (Maxent), a widely-used mathematical tool for distinguishing suitable versus unsuitable fern niche space, along with ENMTools, a tool that assists Maxent with proper model selection, for accurately predicting 29 common, rare, terrestrial, and epiphytic tropical fern species’ distributions. Model discrimination was assessed via area under the receiver operating characteristic curve values, a common metric for model evaluation. Akaike information criteria were utilized for assessing model complexity and in selecting the most parsimonious model for each species. I highlight the importance of modeling with proper model complexity and emphasize the use of information criteria to accurately infer AUC values. Field testing of model predictions also reinforced that these models are successful at identifying suitable habitat for ferns in Puerto Rico and conservation recommendations are explored.

Included in

Biology Commons

Share

COinS