Flawed analysis and unconvincing interpretation: a comment on Chapron and Treves 2016.

Abstract

Chapron & Treves [1] (hereafter C&T) believe that quantifying poaching is ‘one of the most crucial questions for the conservation of large carnivores’ (p. 2). We agree that evaluating poaching is important and merits rigorous attention. Yet, we argue that C&T's claim, ‘allowing culling increases poaching’, is not supported by their data. We assert that C&T is based on flawed analysis and unconvincing interpretation of scientific literature. C&T claimed to ‘present the first quantitative evaluation of the hypothesis that culling will reduce poaching’. However, Olson et al . [2] used empirical data (fates of wolves) to demonstrate that illegal killing decreases with increasing availability of lethal depredation management (hereafter, LDM). C&T claimed to ‘show that allowing wolf [ Canis lupus ] culling was substantially more likely to increase poaching than to reduce it’. However, C&T produced no empirical evidence of increased poaching, but only showed a marginal association between policy change allowing LDM and reduction in expected wolf population growth in Wisconsin and Michigan (USA). Additionally, C&T later reported a misalignment in their dataset between wolf population size, number of wolves culled and policy change [3]. C&T claim that the conclusion of their ‘paper is still supported by the correct results’ (p. 1) [3]. However, the lack of a significant change in results following the correction of their data suggests either important design flaws or a phenomena largely uncoupled from their putative ‘policy signals’. C&T also claimed ‘replicated quasi-experimental’ (p. 2) design because changes in policy led to variation in LDM authority [1]. C&T compared ‘treatment’ periods (periods with LDM) with ‘control’ periods (when wolves were federally protected). C&T's replication claim implies independence among treatments with respect to effect of policy signals [1, p. 3], something most-certainly untrue. …