Prospects for predicting changes to coastal wetland bird populations due to accelerated sea level rise. Issue 12 (21st December 2015)

• Record Type: Journal Article
• Authors: Nuse, Bryan L.; Cooper, Robert J.; Hunter, Elizabeth A.
• Is Part Of: Ecosphere. Volume 6:Issue 12(2015)
• Page Start: 1
• Page End: 23
• Publication Date: 2015-12-21
• Subjects: Altamaha River estuary, Georgia, USA -- Bayesian occupancy model -- Clapper Rail (Rallus longirostris) -- coastal wetlands -- habitat niche -- k-fold cross-validation -- Least Bittern (Ixobrychus exilis) -- marsh birds -- model predictive accuracy -- sea level rise -- Seaside Sparrow (Ammodramus maritimus) -- separation in logistic regression; Ecology -- Periodicals; Ecology; Periodicals; 577.05
• DOI: 10.1890/ES15-00385.1
• Languages: English
• ISSNs: 2150-8925

Abstract

Abstract : Accelerating sea level rise (SLR) is likely to cause considerable changes to estuarine and other coastal wetlands. Efforts to forecast the effects of SLR on coastal wetland vegetation communities should be useful in making predictions for individual species that depend upon those communities. However, considerable uncertainty exists when predicting a chain of events that passes from the global climate to local effects to implications for a single species. One component of this uncertainty is the classification resolution used by SLR landscape change models such as the Sea Level Affects Marshes Model (SLAMM). To isolate and assess the effects of this kind of uncertainty on species‐level SLR prediction, we analyzed surveys of birds and plants in the lower Altamaha River and its estuary in Georgia, USA. For 19 marsh and forest bird species, we tested the predictive value of three classes of covariates of site occupancy: (1) field‐measured habitat variables and spatial information, (2) information available from a SLAMM map, including the spatial configuration of the SLAMM habitat classes, and (3) SLAMM habitat class alone. We found that the predictive ability of occupancy models built from these three kinds of information varies widely among species. We therefore suggest criteria for classifying species according to the amount of detail necessary to describe their habitat niche, and thus to maximize the accuracy of predictive models. We point out that for species with habitat requirements that can be represented well by SLAMM classes, such as the Clapper Rail, forecasts of SLR‐induced population change are probably feasible. For species with more narrow habitat needs, however, such as the Seaside Sparrow, reasonable predictions of SLR effects may not be possible without further refinement of SLR landscape change models. We suggest that improved thematic resolution of such models should be a priority, if the implications of SLR models for individual species are to be ascertained fully.