Unreliable amphibian population metrics may obfuscate more than they reveal

Welsh et al. [Welsh, H.H., Jr. Pope, K.L., Wheeler, C.A., 2008. Using multiple metrics to assess the effects of forest succession on population status: a comparative study of two terrestrial salamanders in the US Pacific Northwest. Biological Conservation 141, pp. 1149-1160] argued for the use of four metrics to monitor population status of two terrestrial amphibians, the Siskiyou Mountains (Plethodon stormi) and Del Norte (Plethodon elongatus) salamanders. We show that inherent problems exist with Welsh et al.'s application of all four of these metrics, and that their inferences about population status are likely to be uncertain as a consequence and potentially misleading to managers. Welsh et al. (2008) used data from different populations and different years to estimate detection probabilities for both P. elongatus and P. stormi, a decision which assumes that detection probabilities do not differ across sites and populations are closed. In addition, Welsh et al. (2008) present count data for both salamanders, a metric that assumes capture probability does not vary by age, gender, size, other individual characteristics, and most importantly in their application, habitat characteristics. Welsh et al. (2008) estimated survival based upon age ratios. Age ratios are known to be biased when immigration and emigration are unequal Conn et al. [Conn, P.B., Doherty, P.F., Jr. Nichols, J.D., 2005. Comparative demography of New World populations of thrushes (Turdus spp.): comment. Ecology 86, pp. 2536-2541]. In source-sink systems, immigration and emigration are expected to be unequal. Thus, the use of this estimation technique to characterize survival in source-sink systems is invalid unless it can be shown that immigration and emigration are equal, a rare occurrence in source-sink systems. The final metric used by Welsh et al. (2008) was an index of body condition, i.e., the residuals of a least squares regression of mass on the length. Several recent articles have outlined a number of potential problems with the technique and a theoretically more robust and efficient alternative has been published. Unfortunately, Welsh et al. (2008) did not provide the details (model diagnostic statistics, tests of assumptions) needed to assess whether or not their analyses of body condition may be accurate. We think that evaluation of specific hypotheses, well-designed sampling programs, and methods such as mark-recapture and ratio sampling are more likely to provide reliable inference than re-analysis of old data sets that were collected for other purposes and obsolete methods (i.e., counting individuals on single occasions) that are known to be flawed. We agree that potential forest management impacts to sensitive taxa require evaluation and monitoring, and that changes in occupancy and abundance are useful metrics for this task (while recognizing that, when feasible to collect, information about reproductive success and survival is superior). However, reliable inference about potential impacts can only be made if reliable methods are employed and if critical assumptions receive empirical evaluation. We offer several suggestions to strengthen inference about management treatments.