Effects of anthropogenic forest stresses on the distribution and abundance of the Olympic Torrent Salamander (Rhyacotriton olympicus) in the Olympic National Forest and Olympic National Park

Salamanders worldwide are faced with habitat loss, and much of the remaining habitat is under the constant pressure of degradation. The forests of the North American Pacific Northwest are no exception. The primary anthropogenic forces impacting the stability of lotic salamander populations on the Olympic peninsula are commercial timber harvest and culverts necessitated by roads crossing streams to facilitate the removal of timber from these forests. In this study, I conducted stream surveys on 139 headwater stream reaches in 77 streams in mature and recently harvested forests both above and below culverts on forest roads in Washington’s Olympic National Forest and Olympic National Park, collecting environmental data and counting Olympic Torrent Salamanders (Rhyacotriton olympicus). I used an information theoretic approach to model selection to evaluate sets of candidate models for both occupancy and abundance of the salamander in streams. Occupancy model selection showed support for models including Gradient, Turbidity, Forest Stage, and Harvest Distance as important predictors of R. olympicus presence at the stream-reach level. I conducted further tests on all models with a ΔAICc score of less than four to determine the relative impact of individual predictor variables. The abundance analysis failed on a goodness of fit test for the global model as the result of a high degree of overdispersion. Because of this failure I was unable to conduct further model selection analyses with the candidate model set. I instead conducted simple post hoc analyses to explore variables not used in the initial candidate model set. The variables that drive occupancy all point to stream gradient as the most important factor in whether a stream reach is suitable for R. olympicus occupancy. Neither the candidate models nor most of the variables explored independently show a strong relationship with salamander abundance. The presence of fish and Tailed Frogs (Ascaphus truei) were both significant predictors of variation in salamander abundance, as were elevation and stream flow. The lack of robust results in the abundance analysis highlights the need for further research using a different framework for questioning, possibly at a different spatial scale like Welsh and Lind (1995, 1996) or even shifting the priority from environmental factors to interspecific interactions. This study’s results provide a direction for future species management. It is clear that preserving suitable Olympic Torrent Salamander habitat requires the protection of high gradient stream reaches and the surrounding forests. The results also found no significant effect of proximity to recently harvested forests (forest age ≤ 30 years) on probability of detection, though associations with forest age may be obscured by combining all forests ages greater than 30 years. However, because occupancy analysis highlights the minimum suitable habitat needs and the abundance analysis relied on post hoc analyses, the need to understand the drivers of abundance in order to create a comprehensive species management plan persists