Meta-analysis of the effects of canopy removal on terrestrial salamander populations in North America

Terrestrial salamander abundance in North American forests is closely associated with forest characteristics, and salamander populations typically decline following timber removal. However, salamander responses to timber harvest vary considerably from one study to the next – some studies have shown limited or no effects whereas others have found harvest-related declines of up to 95%. To estimate overall effect sizes of timber harvest on salamander populations and to examine factors that contribute to variation in salamander responses to harvest, we carried out a meta-analysis of published studies of the effects of timber harvesting on terrestrial salamander abundance. Based on 108 salamander species/treatment combinations from 24 studies, we found mean harvest-related declines ranged from 29% (95% CI = −2% to 51%) for long-term responses to partial canopy removal to 62% (95% CI = 29% to 80%) for short-term responses to clearcutting. Multivariate mixed models suggested that clearcutting had a modestly greater impact on salamander populations than partial harvest and that decreases following harvest were more pronounced at sites with warmer summer high temperatures. In addition, active sampling of salamanders (e.g., surface counts) led to greater apparent effects of timber harvest than passive sampling (e.g., cover boards or pitfall traps), suggesting that sampling methodology influenced the perceived effects of harvest. In the studies evaluated, salamander numbers almost always declined following timber removal, but salamanders were never extirpated and populations typically increased as forests regenerated.

► Effect sizes for overall impacts of canopy removal on terrestrial salamanders.
► Clearcuts produce modestly greater salamander declines than does partial canopy removal.
► Greater reductions in salamander abundance are associated with warmer sites.
► Active sampling techniques lead to larger apparent reductions than does passive sampling.