Predictors of avian and mammalian translocation success: reanalysis with phylogenetically independent contrasts

We use the phylogenetically based statistical method of independent contrasts to reanalyze the Wolf et al., 1996translocation data set for 181 programs involving 17 mammalian and 28 avian species. Although still novel in conservation and wildlife biology, the incorporation of phylogenetic information into analyses of interspecific comparative data is widely accepted and routinely used in several fields. To facilitate application of independent contrasts, we converted the dichotomous (success/failure) dependent variable ( Wolf et al. 1996, Griffith et al. 1989. Translocations as a species conservation tool: status and strategy. Science 245, 477–480) into a more descriptive, continuous variable with the incorporation of persistence of the translocated population beyond the last release year, relative to the species' longevity. For comparison, we present three models: nonphylogenetic multiple logistic regression with the dichotomous dependent variable (the method used by Wolf et al. 1996and Griffith et al. 1989), nonphylogenetic multiple regression with the continuous dependent variable, and multiple regression using phylogenetically independent contrasts with the continuous dependent variable. Results of the phylogenetically based multiple regression analysis indicate statistical significance of three independent variables: habitat quality of the release area, range of the release site relative to the historical distribution of the translocated species, and number of individuals released. Evidence that omnivorous species are more successful than either herbivores or carnivores is also presented. The results of our reanalysis support several of the more important conclusions of the Wolf et al. (1996)and Griffith et al. (1989)studies and increase our confidence that the foregoing variables should be considered carefully when designing a translocation program. However, the phylogenetically based analysis does not support either the Wolf et al. (1996)or Griffith et al. (1989)findings with respect to the statistical significance of taxonomic class (bird vs mammal) and status (game vs threatened, endangered, or sensitive), or the Griffith et al. (1989)findings with respect to the significance of reproductive potential of the species and program length.