Concerns regarding the use of amphibians as metrics of critical biological thresholds: a comment on Welsh & Hodgson (2008)

Welsh & Hodgson (2008) argued that stream-associated amphibians (Ascaphus truei, Dicamptodon tenebrosus and Rhyacotriton variegatus) are reliable indicators of ecosystem health in the Pacific Northwest (PNW), U.S.A. We demonstrate that their assertions rely on circular reasoning as well as logical and empirical assumptions that have not received rigorous evaluation. Welsh & Hodgson (2008) based their arguments on data collected in northwestern California, U.S.A. However, all three taxa occur across a wide range of environmental gradients in the PNW, and other research results suggest that the taxa respond differentially to environmental perturbations across their geographical distribution. Hence, differences in abundance are not always reliable indicators of the state of an ecosystem, as population numbers can be influenced by factors that may not be associated with habitat quality (e.g. predation and disease). Welsh & Hodgson (2008) present indices (rather than estimates) of abundance, which contain an unquantifiable amount of uncertainty not reliably attributable to either environmental or management factors. Evaluating parameters such as reproductive success and survival are more likely to provide strong inference about factors influencing population size than correlative studies that associate some measure of relative abundance with individual habitat features. We argue that direct measurement of physical features (water temperature, sediment and wood), for which Welsh & Hodgson (2008) suggest using amphibians as surrogates, is more accurate, efficient and cost effective compared with monitoring cryptic species with methods that are unable to quantify spatial and temporal uncertainty in biological responses. Headwater streams are critical components of catchments in the PNW, and all three amphibian taxa show strong associations with low-order streams. However, we think research and monitoring of headwater streams should be guided by hypotheses that evaluate causal mechanisms, rather than relying on descriptive patterns that are limited in geographical scope and which support only weak inference about ecosystem processes. Other studies provide direct links between physical features and responses of target populations (i.e. fish). Adding a second correlative link (i.e. amphibians) provides no added benefit, and increases uncertainty in the association between management actions and changes in population size.