Ecological Applications, 25(8), 2015, pp. 2240–2254 Ó 2015 by the Ecological Society of America Mining-caused changes to habitat structure affect amphibian and reptile population ecology more than metal pollution KIYOSHI SASAKI, 1 DAVID LESBARRE ` RES, 1 GLEN WATSON, 2 AND JACQUELINE LITZGUS 1,3 1 Department of Biology, Laurentian University, 935 Ramsey Lake Road, Sudbury, Ontario P3E 2C6 Canada 2 Decommissioning and Reclamation, Vale Canada Ltd., 337 Power Street, Copper Cliff, Ontario P0M 1N0 Canada Abstract. Emissions from smelting not only contaminate water and soil with metals, but also induce extensive forest dieback and changes in resource availability and microclimate. The relative effects of such co-occurring stressors are often unknown, but this information is imperative in developing targeted restoration strategies. We assessed the role and relative effects of structural alterations of terrestrial habitat and metal pollution caused by century- long smelting operations on amphibian and reptile communities by collecting environmental and time- and area-standardized multivariate abundance data along three spatially replicated impact gradients. Overall, species richness, diversity, and abundance declined progressively with increasing levels of metals (As, Cu, and Ni ) and soil temperature ( T s ) and decreasing canopy cover, amount of coarse woody debris (CWD), and relative humidity (RH). The composite habitat variable (which included canopy cover, CWD, T s , and RH) was more strongly associated with most response metrics than the composite metal variable (As, Cu, and Ni ), and canopy cover alone explained 19–74% of the variance. Moreover, species that use terrestrial habitat for specific behaviors (e.g., hibernation, dispersal), especially forest- dependent species, were more severely affected than largely aquatic species. These results suggest that structural alterations of terrestrial habitat and concomitant changes in the resource availability and microclimate have stronger effects than metal pollution per se. Furthermore, much of the variation in response metrics was explained by the joint action of several environmental variables, implying synergistic effects (e.g., exacerbation of metal toxicity by elevated temperatures in sites with reduced canopy cover). We thus argue that the restoration of terrestrial habitat conditions is a key to successful recovery of herpetofauna communities in smelting-altered landscapes. Key words: abundance; amphibian and reptile communities; diversity; habitat restoration; multiple stressors; smelting; species richness; synergistic effects. INTRODUCTION Given that wildlife habitats are almost always simultaneously subjected to multiple anthropogenic stressors, an understanding of relative effects of co- occurring stressors is essential for developing effective conservation and restoration strategies (Blaustein and Kiesecker 2002, Brook et al. 2008). However, determin- ing the (relative) effects of co-occurring factors is not straightforward because of their potential interactive effects, correlations among the stressors, and the spatial scale-dependency of most response metrics. Effects of co-occurring stressors are not always additive; combined effects can become larger (synergistic) or smaller (antagonistic) than the simple algebraic sum of individ- ual effects alone (Crain et al. 2008, Darling and C ˆ ote´ 2008). In addition, spatial correlations among stressors (e.g., increased pollution with increasing magnitude of habitat change) pose an analytical challenge. The relative importance of stressors are commonly assessed by conventional methods, such as multiple regression, but these methods can distort inference about the relative importance of predictor variables even when correlations among variables are small (Mac Nally 2002, Murray and Conner 2009). Finally, the values of most biodiversity metrics (e.g., species richness, Shannon diversity entropy) vary with spatial scales (Chase and Knight 2013) and the relationships between diversity and environmental factors are also scale dependent (Wiens 1989, Scheiner et al. 2000). That is, the scale at which data are analyzed may affect the results of a given analysis (e.g., the strength and nature of species– environmental relationships) and, hence, our ability to detect and interpret impacts and underlying causes of the impacts (Sandel and Corbin 2012). Metal smelting can bring about extreme and long- lasting modifications to wildlife habitats (Kozlov and Zvereva 2007). Smelting emits large quantities of sulphur and metals, and often causes large-scale forest dieback in fallout zones, leaving an extensive barren landscape with little vegetation and highly acidic, metal- contaminated water and soil (Kozlov and Zvereva Manuscript received 23 July 2014; revised 8 December 2014; accepted 2 March 2015. Corresponding Editor: T. W. J. Garner. 3 Corresponding author. E-mail: jlitzgus@laurentian.ca 2240