Date of Award

Spring 1-1-2014

Document Type


Degree Name

Doctor of Philosophy (PhD)


Ecology & Evolutionary Biology

First Advisor

M. Deane Bowers

Second Advisor

Robert P. Guralnick

Third Advisor

Kendi Davies

Fourth Advisor

Brett Melbourne

Fifth Advisor

Cesar Nufio


Landscape modification is leaving an irrevocable scar on the planet, most notably through habitat fragmentation. Fragmented landscapes are often unable to support communities that once inhabited them, leading to unprecedented rates of global biodiversity loss. As a result, substantial research effort focuses on investigating the drivers of species' responses to habitat fragmentation, usually for one or a few species at select locations. This dissertation expands upon previous research in order to broaden understanding of the determinants of diversity patterns in fragmented landscapes. I modeled variation in among fragment butterfly diversity for entire communities, using both environmental attributes and species traits as predictors. I then compared models across three, widely separated fragmented landscapes. I found that patch area and water availability had consistent influences on butterfly diversity patterns; these factors may warrant inclusion into management policies for fragmented landscapes worldwide. Other predictors, e.g., butterfly wing length, had variable influences on diversity patterns, although results revealed similarities between certain study areas. For example, habitat heterogeneity influenced diversity patterns similarly in two study areas, possibly due to similarities in ecological and/or climatic characteristics (e.g., drought-prone summers). Furthermore, species traits played important, albeit inconsistent, roles in driving butterfly diversity patterns; this pattern also potentially driven by among location ecological and/or climatic conditions. In all, this integrative data reuse analysis demonstrated patterns that may provide crucial information for better understanding wide-spread species responses to habitat fragmentation. The final component of this dissertation was an exploration of questions that arose from the data reuse strategy employed: how different are models constructed from datasets obtained via disparate levels of survey effort, and what implications does this have for data reuse analyses. I constructed a new model using data collected via 2/3 of the full sampling effort for one dataset. The model was almost identical to that constructed from the full dataset, and the use of this `reduced sampling effort' dataset would thus have had negligible impact on previous results. This work provides insight into the sensitivity of downstream analyses to variation in survey methods, and substantiates the validity of analyses reusing datasets collected by different researchers.