Predictors of avian and mammalian translocation success: reanalysis with phylogenetically independent contrasts
Introduction
Numerous methodological, environmental, species-specific, and population-level factors influence whether the intentional release of wild-caught or captive-reared animals into new locations will result in the successful establishment, re-establishment, or augmentation of a wild population. Because specific causal factors and their relative importance vary widely among such release programs (herein referred to as translocations, following Griffith et al., 1989; see also Wolf et al., 1996), it is difficult to identify general trends associated with success. Nevertheless, both theoretical considerations and empirical evidence suggest that some methodological and biological factors are of general importance. For example, such methodological details as raise and release procedures, number and composition of animals released, and choice of source stock for the released animals were shown to influence translocation outcomes in single-species studies (Beck et al., 1991, Allen et al., 1993, Reed et al., 1993, Bright and Morris, 1994, Veltman et al., 1996). Environmental factors perceived as important to success include general habitat quality and climatic conditions (Lindenmayer, 1994, Veitch, 1994), as well as the absence of predators or competitors (Crawley, 1986, Short et al., 1992). Some species-specific and population characteristics considered favorable for successful invasion of a new location include a relatively high reproductive potential, an omnivorous diet, small body mass, and high genetic diversity (Mayr, 1965Laycock, 1966Berger, 1972Smith et al., 1976Crawley, 1986Ehrlich, 1986O'Connor, 1986). The question remains: how universal is the influence of such methodological and biological factors? Also, in light of substantial case-by-case variations, are generalizations across a range of taxa possible?
Griffith et al. (1989)(see also Griffith et al., 1990, Griffith et al., 1993) used a comparative approach to test for general patterns underlying the success vs failure of translocations among species of birds and mammals. They conducted surveys of translocation programs throughout North America, Australia, and New Zealand; coded the outcomes as either a success (reported establishment of a self-sustaining population), a failure, or incomplete; and used multiple logistic regression to identify seven statistically significant predictors of success: (1) taxonomic class (bird vs mammal), (2) legal status of the translocated species (native game vs threatened, endangered, or sensitive species), (3) habitat quality of the release area (excellent, good, or fair/poor), (4) location of the release area relative to the historical range of the species (core vs periphery or outside), (5) number of animals released (log10 transformed), (6) program length (number of years over which releases occurred), and (7) potential productivity of the translocated species (high vs low). Wolf et al. (1996)conducted a follow-up survey, in which they up-dated the status of the translocations in the Griffith et al. study, increased the number of programs available for multiple regression analyses from 155 to 181, and tested additional variables as predictors of success. Their results, using comparable analyses, were largely consistent with the findings of Griffith et al. (1989). Wolf et al. (1996)produced a model which contained the first five of the Griffith et al. variables (as listed earlier) plus adult diet of the species in the wild (herbivorous vs omnivorous vs carnivorous).
As was abundantly demonstrated in the last decade, interspecific comparisons are potentially compromised by statistical non-independence of species values (Felsenstein, 1985, Harvey and Pagel, 1991, Garland et al., 1993, Martins and Hansen, 1996). A simplified consensus view of the problem can be summarized as follows. Species are related to greater or lesser degrees, as indicated by their phylogenetic (evolutionary) relationships. Closely related species possess many characteristics, and sometimes much of their selective regime, that were inherited from common ancestors. As a result of such inheritance, data for a series of species may contain hierarchical resemblances (e.g. snakes look like snakes, elephants look like elephants) and cannot be assumed to represent independent data points, a key assumption of both traditional parametric and nonparametric statistical methods. Detailed discussions of why closely related species tend to be similar are available elsewhere (Grafen, 1989, Brooks and McLennan, 1991, Harvey and Pagel, 1991, Eggleton and Vane-Wright, 1994, Martins, 1996a). In general, common effects of violating the assumption of independence include an inflation of Type I error rates when hypothesis testing (claiming statistical significance when none actually exists) and poor estimation of relationships among variables (and hence diminished predictive accuracy). The reality of these adverse statistical effects was demonstrated both analytically and by computer simulation studies (Felsenstein, 1985, Grafen, 1989, Martins and Garland, 1991, Garland et al., 1992, Garland et al., 1993, Pagel, 1993, Purvis et al., 1994, Dı́az-Uriarte and Garland, 1996Dı́az-Uriarte and Garland, in pressMartins, 1996b).
Inclusion of phylogenetic information in statistical analyses is now routine in such fields as behavioral and physiological ecology (Miles and Dunham, 1993, Garland and Carter, 1994, Losos and Miles, 1994, Dı́az et al., 1996, Martin and Clobert, 1996, Martins, 1996a, Reynolds and Lee, 1996, Ricklefs and Starck, 1996, Williams, 1996, Abouheif and Fairbairn, 1997, Bauwens and Dı́az-Uriarte, 1997, Price, 1997Clobert et al., in press). Although an historical, evolutionary approach may not initially seem germane to questions pertaining to contemporary wildlife translocation success, all interspecific (and many interpopulation) comparisons are potentially subject to phylogenetic influence in statistical tests. To illustrate, one might expect a translocation of bighorn sheep in Nevada to share inherent similarities with a translocation of bighorn sheep conducted elsewhere. Part of this similarity would be caused by general capture and release methodology, whereas part of this similarity would also be caused by biological characteristics of bighorn sheep (e.g. reproductive potential, social systems, disease susceptibility). Likewise, one might expect similarities between translocations conducted with closely related species (e.g. Prairie Chicken and Sharp-tailed Grouse), as compared with translocations involving more distantly related species (e.g. Canada Goose).
Although comparative biologists have long recognized that species should not be treated as independent data points, early proposed solutions to this problem, such as averaging within genera or choosing only one species per genus, are inadequate because they ignore potentially informative variation within phylogenetic lineages (clades), ignore hierarchical relationships among clades, and diminish statistical power (review in Harvey and Pagel, 1991). State-of-the-art phylogenetically based analyses allow incorporation of phylogenetic information without loss of statistical power (Martins and Garland, 1991Garland and Adolph, 1994Purvis et al., 1994Martins, 1996b).
The present study, therefore, employs phylogenetically based statistical methods to reanalyze the Wolf et al. (1996)data set (104 avian and 77 mammalian translocation programs, representing 28 avian and 17 mammalian species). Felsenstein, (1985)method of phylogenetically independent contrasts, the first fully phylogenetic method to be proposed, is the best understood of available methods (Grafen, 1989, Martins and Garland, 1991, Garland et al., 1992, Pagel, 1993, Purvis and Garland, 1993, Purvis et al., 1994, Dı́az-Uriarte and Garland, 1996Dı́az-Uriarte and Garland, in pressMartins, 1996b, Martins and Hansen, 1996). Moreover, several free computer programs are available which implement the method (e.g. Joe Felsenstein's PHYLIP package, the PDAP package (Garland et al., 1993), CAIC (Purvis and Rambaut, 1995), COMPARE (Martins and Hansen, 1996)). In brief, the method of independent contrasts uses phylogenetic information (topology and branch lengths) to transform interspecific data (i.e. estimates of mean values for a series of species) into values (standardized independent contrasts) that, in principle, are independent and identically distributed, and hence can be analyzed with standard statistical methods (see Methods section for more details).
To facilitate application of phylogenetically independent contrasts, we used a more detailed indicator of translocation outcome as a dependent variable, rather than the dichotomous measure (success vs failure) used previously (Griffith et al., 1989, Wolf et al., 1996). We developed a continuous outcome variable that was a composite (see Methods section) of the following information: (1) the translocated population persistence (in years) in the field; (2) the classification of the population as self-sustaining, declining, or gone; and (3) the maximum potential life span (years) of each species as a scaling factor. The continuous dependent variable better meets assumptions of phylogenetically independent contrasts, and the inclusion of population persistence information provides a more quantitative measure of the population's ability to persist through time. In principle, this more inclusive dependent variable should increase statistical power to detect significant predictors of `success.'
Section snippets
The data
The up-dated avian and mammalian translocation data used by Wolf et al. (1996)in the multiple logistic regression models were also used in this study. Of the 421 targeted avian and mammalian translocations throughout North America, Australia, and New Zealand, 181 programs were used by Wolf et al. (1996)in the logistic analyses. Within the sample of 181 translocations, 122 were classified as successful and 59 as unsuccessful; 104 involved translocations of birds and 77 of mammals; and 24
Results
After converting the dichotomous dependent variable (success vs failure) into a composite, continuous variable (see Methods section), the nonphylogenetic ordinary stepwise regression produced a model with five main-effect variables and one first-order interaction term: habitat quality (B=0.252, p=0.0176, where B equals the partial regression coefficient), migratory behavior (B=−0.504, p=0.0021), status (B=0.551, p=0.0002), range (B=2.332, p<0.0001), number of animals (B=0.954, p<0.0001), and
Discussion
Complicated biological questions and associated data sets can be analyzed in many different ways. Often, little consensus exists as to the `best' way to analyze a particular type of data. The application of different analytical methods to a single data set is, therefore, prudent. To the extent that different analytical techniques lead to similar conclusions, then we may have enhanced confidence in those conclusions.
With respect to interspecific comparative studies, a flurry of research activity
Conclusion
Although routinely employed in other disciplines, phylogenetically based statistical analyses are virtually nonexistent within such fields as conservation biology and wildlife management. Yet conservation and wildlife biologists often make interspecific comparisons. The same resemblances among closely related species that are often exploited by wildlife biologists for guidance in developing methodological protocols can confound traditional correlational and regression analyses of multi-species
Acknowledgements
The authors thank M.R. Dohm for assistance with statistical analyses and comments on manuscript drafts; R.E. Bleiweiss, J.A.W. Kirsch, and S. Moore helped with derivation of the phylogeny. Also greatly appreciated were C.M.W's. graduate committee members, P. Arcese, T.C. Moermond, and S.A. Temple. This study was funded with grants from the Zoological Society of Milwaukee County, the Chicago Zoological Society, and the Idaho Department of Fish and Game; T.G. is supported by National Science
References (79)
- et al.
DNA hybridization-based phylogeny for higher nonpasserines: reevaluating a key portion of the Avian family tree
Molecular Phylogenetics and Evolution
(1994) Seeking the evolutionary regression coefficient: an analysis of what comparative methods measure
Journal of Theoretical Biology
(1993)- et al.
Truth or consequences: effects of phylogenetic accuracy on two comparative methods
Journal of Theoretical Biology
(1994) - et al.
Reintroduction of macropods (Marsupialia: Macropodoidea) in Australia — a review
Biological Conservation
(1992) - et al.
A comparative analysis of allometry for sexual size dimorphism: assessing Rensch's rule
American Naturalist
(1997) - et al.
Efficacy of translocation strategies for red-cockaded woodpeckers
Wildlife Society Bulletin
(1993) - et al.
New Zealand translocations: theory and practice
Pacific Conservation Biology
(1995) - et al.
Covariation of life-history traits in lacertid lizards: a comparative study
American Naturalist
(1997) - et al.
Losses and reproduction in reintroduced golden lion tamarins Leontopithecus rosalia
Dodo
(1991) - Berger, A.J., 1972. Hawaiian Birdlife. University of Hawaii Press, Honolulu,...
Confirmation of a portion of the Sibley–Ahlquist “tapestry”
Auk
Animal translocation for conservation: performance of dormice in relation to release methods, origin and season
Journal of Applied Ecology
The population biology of invaders
Philosophical Transactions of the Royal Society of London
A comparative study of the relation between heating rates and ambient temperatures in lacertid lizards
Physiological Zoology
Testing hypotheses of correlated evolution using phylogenetically independent contrasts: sensitivity to deviations from Brownian motion
Systematic Biology
Relocation, repatriation, and translocation of amphibians and reptiles: are they conservation strategies that work?
Herpetologica
The pattern of mammalian evolution and the relative rate of molecular evolution
Genetics
Phylogenies and the comparative method
American Naturalist
Rate tests for phenotypic evolution using phylogenetically independent contrasts
American Naturalist
Why not to do two-species comparative studies: limitations on inferring adaptation
Physiological Zoology
Evolutionary physiology
Annual Review of Physiology
Procedures for the analysis of comparative data using phylogenetically independent contrasts
Systematic Biology
Phylogenetic analysis of covariance by computer simulation
Systematic Biology
The phylogenetic regression
Philosophical Transactions of the Royal Society of London
Translocations as a species conservation tool: status and strategy
Science
Translocations of captive-reared terrestrial vertebrates, 1973–1986
Endangered Species UPDATE
Animal translocations and potential disease transmission
Journal of Zoo and Wildlife Medicine
The marsupial mitochondrial genome and the evolution of placental mammals
Genetics
Cited by (214)
A major myna problem; invasive predator removal benefits female survival and population growth of a translocated island endemic
2023, Global Ecology and ConservationSpecies Distribution Models predict abundance and its temporal variation in a steppe bird population
2023, Global Ecology and ConservationLong-term demography of a reintroduced population of endangered falcons
2022, Global Ecology and ConservationFactors affecting success of conservation translocations of terrestrial vertebrates: A global systematic review
2021, Global Ecology and Conservation
- 1
Present address: Alaska Cooperative Fish and Wildlife Research Unit, 216 Irving I Building, University of Alaska, Fairbanks, AK 99775.