Darwinian Decision Making: Putting the Adaptive into Adaptive Management

The best management decisions are based on the best science, or so scientists are taught. In this issue Seddon et al. (2007) review reintroduction of science and note that much of it has been ad hoc and not designed to be experimental. Moreover, managers are not benefiting as much as they could from population viability analysis and geographic information systems. This is surprising because of the general emphasis in the literature on “active adaptive management” (Walters & Holling 1990). Adaptive management plans are modified on the basis of the results of well-designed experiments that collect data on factors or variables that are demonstrably important for conservation or management (e.g., Ministry of Forest and Range 2001). As Seddon et al. (2007) discuss, the results of properly designed experiments can be revealing. Comparison of proper controls with formal treatments is an essential part of such experiments because it helps isolate the effect of a particular manipulation (Underwood 1992). If, for instance, one manipulation is done in one year and another manipulation is done in another year, the difference between years may not be a result of the manipulations, but rather some other factor that varied across years. Thus, managers could make spurious conclusions and management decisions may not be scientifically sound. Nonetheless, managers may resist designing studies with control groups ( Johnson 1999; Lee 1999) for several reasons. First, there may be too few animals with which to conduct a proper study. For instance, in lieu of an experimental approach, managers charged with recovering the Po’ouli (Melamprosops phaeosoma)—a critically endangered Hawaiian honeyeater—opted for a probabilistic decision-tree analysis (VaderWerf et al. 2006). Second, there may be many factors that have to be manipulated simultaneously. In this case it is rare with an en-