Using regional patterns for predicting local temporal change: a test by natural experiment in the Great Lakes bioregion, Ontario, Canada

Abstract

AbstractAimIt has been found that species richness can be a peaked function of natural area, meaning that conversion of natural cover can increase species numbers. Here, we test whether regional richness–environment relationships can predict local change in species richness.LocationThe greater park ecosystem of Thousand Islands National Park (TINP ecosystem), Ontario, Canada.MethodsWe evaluated change in bird richness in 85 100‐km2 sites, censused from 1981 to 1985 and from 2001 to 2005. We related change in richness over 20 years to change in natural land cover for sites having initially less or greater than half natural area to determine whether richness changes in the direction predicted. We also assessed the extent to which local effects (site‐level extinction and colonization based on species‐specific habitat amount and neighbourhood occupancy by conspecifics) affected the predictability of species richness response to environmental change.ResultsLocal effects predicted bird richness changes through time far better than natural area change (adjusted r2 = 0.56 vs. adjusted r2 = 0.13). Species richness did not consistently respond to change in natural area in the directions predicted by the broader spatial richness–environment relationship, tending to increase with decrease in natural area regardless of the initial amount of natural area. The observed natural area changes (+0.7 km2 to −4.4 km2) were small relative to the changes in richness (+60 species to −37 species), likely impacting our ability to detect a response.Main conclusionsThe importance of local effects observed here in determining site‐level species presence and consequently species richness has implications for species monitoring and the use of species richness as a measure of the avian response to land cover modification in Ontario. Detecting richness responses to environmental changes may often be challenged by low ‘signal‐to‐noise’ ratios and highlights the benefits of long‐term species monitoring.