Measuring the effects of forest extent and road traffic on anuran populations.

摘要

Landscape ecologists have made large strides in the last 15 years in understanding the effect that landscape structure has on the abundance and persistence of organisms. However, conducting research at the landscape scale brings with it a number of methodological challenges. In this thesis, I use careful site selection and novel analytical techniques to examine the effects of forest extent and road traffic on anuran populations, and, more generally, examine how methodological issues affect landscape-scale inference. I address four questions: (1) What is the effect of experimental design on landscape-scale inference? (2) What are the relative effects of road traffic and forest extent on anuran populations? (3) What are the combined effects of a major highway and the forest extent near it on anuran populations? (4) What is the road effect zone of a major intercity highway for anurans?;I showed that sub-optimal study design at the landscape scale -- particularly incomplete sampling across the possible range of values in a predictor variable and correlations between landscape-scale predictors - is likely to have a major impact on landscape-scale inference. I then showed that the negative effect of road traffic is three times as great as the landscape-scale effect of deforestation on anuran species richness in eastern Ontario when these two predictor variables are considered independently. Having established that road traffic has an important negative effect on anurans, I then focused on better understanding how roads affect wildlife. First, I showed that a novel measure of the combined effects of roads and habitat loss on wildlife - accessible habitat - was a better predictor of anuran species richness near a major highway (Highway 401) than considering distance to the highway and habitat amount alone. I then showed that the negative effect of Highway 401 on anuran populations extends at least 500 m into the landscape. Overall, my results suggest that sub-optimal study design has likely led to an underestimate of the true effect of roads on wildlife. More generally, careful study design is necessary for landscape-scale studies to continue to provide meaningful guidelines for conservation.