OPS 2-12
Combining population-level and individual-level studies to assess salamander responses to disturbances

Monday, August 10, 2015
Exhibit Hall, Baltimore Convention Center
Katherine M. O'Donnell, Division of Biological Sciences, University of Missouri, Columbia, MO
Frank R. Thompson III, Northern Research Station, USDA Forest Service, Columbia, MO
Raymond D. Semlitsch, Division of Biological Sciences, University of Missouri, Columbia, MO

Prescribed fire and timber harvest are anthropogenic disturbances that modify resource availability and ecosystem structure, and can affect wildlife both directly and indirectly. Individual animals may respond to habitat alteration in physiological, behavioral, or ecological ways that subsequently influence population-level dynamics such as survival rate and spatial distribution. Thus, to fully evaluate and predict wildlife responses to disturbances, both population-level studies and individual-level mechanistic studies are necessary. Terrestrial salamanders are an ideal study system, since their high abundance and environmental sensitivity make them effective indicators of ecosystem health. We surveyed for southern red-backed salamanders from 2010–2014 in a Missouri Ozark (USA) forest, and used binomial-mixture models to estimate abundance in a replicated, large-scale Before-After, Control-Impact (BACI) experiment. We compared abundance, surface activity, detectability, and microhabitat use among four treatments – prescribed burn, shelterwood harvest, midstory herbicide, and untreated controls. We also used passive integrated transponders (“PIT-telemetry”) to investigate responses of individual western slimy salamanders to prescribed fire. Using a portable reader/antenna system, we could locate salamanders implanted with uniquely coded PIT-tags ≤ 30 cm belowground. We compared salamander movement (daily displacement [DD], minimum convex polygons [MCP]) and surface activity before and after prescribed fire using a randomized block design.


Salamander abundance decreased following treatments; abundance estimates in midstory and burn units rebounded in the second post-treatment year, but further declined in shelterwood harvest units. Surface activity also decreased, relative to controls, in shelterwood (58.4%), prescribed burn (49.0%), and midstory herbicide (36.7%) units. Additionally, salamander microhabitat use differed following prescribed burns; cover object captures (73.8%)  were more frequent than leaf litter captures in burned units (26.3%), but leaf litter capture frequency remained higher in all other units (mean = 56.5%). In our PIT-telemetry study, we recorded 918 detections of 142 individuals. Over 75% of detections were of belowground salamanders. Following prescribed burns, the surface-active proportion of re-detected salamanders was nearly 7 times higher in control than burned areas. We found no difference between treatments in either total re-detections or the proportion of salamanders known alive, but DD and mean MCP size were greater in burned plots. Lower levels of surface activity may indicate a behavioral avoidance of increased physiological stress. If salamander surface activity is depressed for long enough, populations may decline due to reductions in foraging and breeding opportunities. These indirect consequences of disturbance could take time to manifest into detectable changes in abundance.