Date of Award


Degree Type


Degree Name

Master of Science (MS)



First Advisor

Dr. James Watling


Aim To assess the relative impact of different landscape variables on species richness and to determine whether species richness declines more rapidly below an extinction threshold of remaining habitat. The results of this study will help to better inform future conservation strategies. Location Global Time period 1997 – 2013 Major taxa studied Amphibians, birds, invertebrates, mammals, and reptiles Methods Data from 71 studies published in the global BioFrag database were used to determine species richness across multiple landscapes and biomes. The Hansen dataset was used to collect data on habitat amount (forest area), fragmentation (patch density), and matrix quality (mean % tree cover in the matrix) within the local landscape of each plot. Multi-model inference and meta-analysis were used to compare the relative impacts of standardized predictor variables on species richness. Break point and linear regression models of percent forest cover and species richness were used to test for the presence of extinction thresholds. Results Of the 29 studies included in multi-model inference, habitat amount had a greater regression coefficient than patch density in 15 studies and matrix quality in 21 studies. Patch density had a greater regression coefficient than habitat amount in 4 studies and matrix quality in 16 studies. The meta-analysis found habitat amount to have the greatest effect size with a |Fisher’s z-score| ~1.7x greater than that of patch density and ~2.6x greater than matrix quality. The breakpoint regression model was significant and outperformed the linear regression model in 7 out of 71 studies. Main conclusions Habitat amount had the greatest relative impact on species richness followed by patch density and matrix quality. We did not find support for the extinction threshold hypothesis.

Included in

Biology Commons