An evaluation of prior influence on the predictive ability of Bayesian model averaging.

Model averaging is gaining popularity among ecologists for making inference and predictions. Methods for combining models include Bayesian model averaging (BMA) and Akaike's Information Criterion (AIC) model averaging. BMA can be implemented with different prior model weights, including the Kullback-Leibler prior associated with AIC model averaging, but it is unclear how the prior model weight affects model results in a predictive context. Here, we implemented BMA using the Bayesian Information Criterion (BIC) approximation to Bayes factors for building predictive models of bird abundance and occurrence in the Chihuahuan Desert of New Mexico. We examined how model predictive ability differed across four prior model weights, and how averaged coefficient estimates, standard errors and coefficients' posterior probabilities varied for 16 bird species. We also compared the predictive ability of BMA models to a best single-model approach. Overall, Occam's prior of parsimony provided the best predictive models. In general, the Kullback-Leibler prior, however, favored complex models of lower predictive ability. BMA performed better than a best single-model approach independently of the prior model weight for 6 out of 16 species. For 6 other species, the choice of the prior model weight affected whether BMA was better than the best single-model approach. Our results demonstrate that parsimonious priors may be favorable over priors that favor complexity for making predictions. The approach we present has direct applications in ecology for better predicting patterns of species' abundance and occurrence.

EVALUATING THE THRESHOLD DENSITY HYPOTHESIS FOR MOOSE (ALCES ALCES), WHITE-TAILED DEER (ODOCOILEUS VIRGINIANUS), AND PARELAPHOSTRONGYLUS TENUIS.

Despite the importance of the Parelaphostrongylus tenuis infection for moose (Alces alces) and white-tailed deer (Odocoileus virginianus) management, only one peer-reviewed study has evaluated the relationship between deer and moose densities and the potential for parasite-mediated competition between the species. Using polynomial-regression modeling, that study identified a deer-density threshold above which moose populations declined; however, the nature of the data and apparent outliers suggests the approach used to develop that threshold may not have been appropriate. We used the data from the original study to test whether alternative models, including linear models and negative binomial models would be less sensitive to outliers and could better explain that relationship. We found no evidence that moose density decreases as deer density increases. We concluded that, although the proposed moose-deer-P. tenuis relationship could be partially density dependent, additional factors, such as frequency dependence of disease transmission, gastropod abundance, and shared use of resources by moose and deer should also be considered.

Population reduction by hunting helps control human-wildlife conflicts for a species that is a conservation success story.

Among the world's large Carnivores, American black bears (Ursus americanus) are the foremost conservation success story. Populations have been expanding across North America because the species is adaptable and tolerant of living near people, and because management agencies in the U.S. and Canada controlled hunting and other human-sources of mortality. As a result, human-black bear conflicts (damage to property, general nuisance, threat to human safety) have dramatically increased in some areas, making it urgently important to develop and deploy a variety of mitigation tools. Previous studies claimed that legal hunting did not directly reduce conflicts, but they did not evaluate whether hunting controlled conflicts via management of population size. Here, we compared temporal patterns of phoned-in complaints about black bears (total ~63,500) in Minnesota, USA, over 4 decades to corresponding bear population estimates: both doubled during the first decade. We also quantified natural bear foods, and found that large year-to-year fluctuations affected numbers of complaints; however, since this variation is due largely to weather, this factor cannot be managed. Complaints fell sharply when the management agency (1) shifted more responsibility for preventing and mitigating conflicts to the public; and (2) increased hunting pressure to reduce the bear population. This population reduction was more extreme than intended, however, and after hunting pressure was curtailed, population regrowth was slower than anticipated; consequently both population size and complaints remained at relatively low levels statewide for 2 decades (although with local hotspots). These long-term data indicated that conflicts can be kept in tolerable bounds by managing population size through hunting; but due to the bluntness of this instrument and deficiencies and uncertainties in monitoring and manipulating populations, it is wiser to maintain a population at a level where conflicts are socially-acceptable than try to reduce it once it is well beyond that point.

Modeling habitat suitability for Greater Rheas based on satellite image texture.

Many wild species are affected by human activities occurring at broad spatial scales. For instance, in South America, habitat loss threatens Greater Rhea (Rhea americana) populations, making it important to model and map their habitat to better target conservation efforts. Spatially explicit habitat modeling is a powerful approach to understand and predict species occurrence and abundance. One problem with this approach is that commonly used land cover classifications do not capture the variability within a given land cover class that might constitute important habitat attribute information. Texture measures derived from remote sensing images quantify the variability in habitat features among and within habitat types; hence they are potentially a powerful tool to assess species-habitat relationships. Our goal was to explore the utility of texture measures for habitat modeling and to develop a habitat suitability map for Greater Rheas at the home range level in grasslands of Argentina. Greater Rhea group size obtained from aerial surveys was regressed against distance to roads, houses, and water, and land cover class abundance (dicotyledons, crops, grassland, forest, and bare soil), normalized difference vegetation index (NDVI), and selected first- and second-order texture measures derived from Landsat Thematic Mapper (TM) imagery. Among univariate models, Rhea group size was most strongly positively correlated with texture variables derived from near infrared reflectance measurement (TM band 4). The best multiple regression models explained 78% of the variability in Greater Rhea group size. Our results suggest that texture variables captured habitat heterogeneity that the conventional land cover classification did not detect. We used Greater Rhea group size as an indicator of habitat suitability; we categorized model output into different habitat quality classes. Only 16% of the study area represented high-quality habitat for Greater Rheas (group size > or =15). Our results stress the potential of image texture to capture within-habitat variability in habitat assessments, and the necessity to preserve the remaining natural habitat for Greater Rheas.

Spatio-temporal variation in prevalence and intensity of trematodes responsible for waterfowl die-offs in faucet snail-infested waterbodies of Minnesota, USA.

Several non-native trematodes hosted by the invasive Eurasian faucet snail, Bithynia tentaculata, have been causing die-offs of waterfowl in the Midwestern United States and Canada for several decades. Because of the potential implications of these die-offs on waterfowl in non-native settings, it is necessary to better understand the trematodes that cause the die-offs. Here, we studied the spatio-temporal dynamics of two trematodes, Cyathocotyle bushiensis and Sphaeridiotrema spp., known to infect waterfowl in northern Minnesota, USA, via their intermediate host, the faucet snail (Bithynia tentaculata). We studied prevalence (% of snails infected within a sample) and intensity (mean number of parasites per infected snail within a sample) of faucet snail infection with these two trematodes in small lakes, large lakes, ponds, and rivers in northern Minnesota in the spring, summer, and fall of 2011-2013. We tested whether parasite prevalence and infection intensity could be explained spatially (as a function of the abundance of faucet snails, average snail size, water depth, and proximity to known waterfowl groups) and temporally (across years and seasons) using generalized estimating equation models. The spatial and temporal patterns we observed varied within and among waterbodies. For both parasite species, parasite prevalence and intensity of infection were consistently higher in samples with larger snails and in deeper portions of the waterbodies. In Lake Winnibigoshish, prevalence was lower farther from the large waterfowl groups we observed, but the abundance of snails in a sample had no effect on prevalence or intensity of infection. Our findings help improve understanding of this multi-species system, but also illustrate the complexity of modeling the spatial and temporal dynamics of infections in waterbodies that are so variable in size, shape, waterfowl use, and function.

Modelling avian biodiversity using raw, unclassified satellite imagery.

Applications of remote sensing for biodiversity conservation typically rely on image classifications that do not capture variability within coarse land cover classes. Here, we compare two measures derived from unclassified remotely sensed data, a measure of habitat heterogeneity and a measure of habitat composition, for explaining bird species richness and the spatial distribution of 10 species in a semi-arid landscape of New Mexico. We surveyed bird abundance from 1996 to 1998 at 42 plots located in the McGregor Range of Fort Bliss Army Reserve. Normalized Difference Vegetation Index values of two May 1997 Landsat scenes were the basis for among-pixel habitat heterogeneity (image texture), and we used the raw imagery to decompose each pixel into different habitat components (spectral mixture analysis). We used model averaging to relate measures of avian biodiversity to measures of image texture and spectral mixture analysis fractions. Measures of habitat heterogeneity, particularly angular second moment and standard deviation, provide higher explanatory power for bird species richness and the abundance of most species than measures of habitat composition. Using image texture, alone or in combination with other classified imagery-based approaches, for monitoring statuses and trends in biological diversity can greatly improve conservation efforts and habitat management.