Comparing disturbance gradients and bird-based indices of biotic integrity for ranking the ecological integrity of Great Lakes coastal wetlands

摘要

We compared different methods for generating indices of biotic integrity (IBIs) for Great Lakes coastal wetlands using bird community data collected by participants in Bird Studies Canada's Great Lakes Marsh Monitoring Program (GLMMP) including: rank sum and multivariate approaches for defining landscape disturbance gradients; and generalist-specialist (IBI-1), multimetric (IBI-2), and probabilistic (IBI-3) approaches for calculating IBIs. Scores from the multivariate disturbance gradient, IBI-1, and IBI-3 increased rapidly at the impaired and unimpaired ends of the impaired-to-unimpaired spectrum, whereas scores from the rank sum disturbance gradient and IBI-2 increased rapidly only at the unimpaired end. IBIs with metrics that both increased and decreased along the landscape disturbance gradient were more sensitive for identifying especially impaired and unimpaired sites (i.e., IBI-1 and IBI-3) compared to IBIs with metrics that only increased (i.e., IBI-2). Scores from all but one of the IBIs were significantly correlated with scores of at least one of the landscape disturbance gradients and scores from all three of the IBIs were significantly moderately correlated with each other (r(s) = 0.3-0.7). Site ranks arranged from impaired to unimpaired differed by 25-50 positions out of 142 possible positions depending on the pair of IBIs chosen. Much of the variation that we observed could be explained by differences among IBIs in the metrics that contributed most to impaired and unimpaired sites. Thus, we recommend the following not only for IBI users assessing the integrity of Great Lakes coastal wetlands, but also any other ecosystem where multiple landscape disturbance gradients and IBIs are available for use: (1) use multivariate instead of rank sum approaches for defining landscape disturbance gradients; (2) use IBIs with metrics that both increase and decrease along the landscape disturbance gradient instead of IBIs with metrics that only increase or only decrease; and (3) ensure that site-level species lists are reasonably complete, particularly for species that disproportionately contribute to especially impaired and unimpaired scores. Following these guidelines will increase the sensitivity and accuracy of IBIs for identifying especially impaired and unimpaired sites and ultimately result in better conservation. (C) 2015 Elsevier Ltd. All rights reserved.