Chris J. Johnson

Resource Selection Functions Based on Use–Availability Data: Theoretical Motivation and Evaluation Methods

Abstract Applications of logistic regression in a used–unused design in wildlife habitat studies often suffer from asymmetry of errors: used resource units (landscape locations) are known with certainty, whereas unused resource units might be observed to be used with greater sampling intensity. More appropriate might be to use logistic regression to estimate a resource selection function (RSF) tied to a use–availability design based on independent samples drawn from used and available resource units. We review the theoretical motivation for RSFs and show that sample “contamination” and the exponential form commonly assumed for the RSF are not concerns, contrary to recent statements by Keating and Cherry (2004; Use and interpretation of logistic regression in habitat-selection studies. Journal of Wildlife Management 68:774–789). To do this, we re-derive the use–availability likelihood and show that it can be maximized by logistic regression software. We then consider 2 case studies that illustrate our findings. For our first case study, we fit both RSFs and resource selection probability functions (RSPF) to point count data for 4 bird species with varying levels of occurrence among sample blocks. Drawing on our new derivation of the likelihood, we sample available resource units with replacement and assume overlapping distributions of used and available resource units. Irrespective of overlap, we observed approximate proportionality between predictions of a RSF and RSPF. For our second case study, we evaluate the classic use-availability design suggested by Manly et al. (2002), where availability is sampled without replacement, and we systematically introduce contamination to a sample of available units applied to RSFs for woodland caribou (Rangifer tarandus caribou). Although contamination appeared to reduce the magnitude of one RSF beta coefficient, change in magnitude exceeded sampling variation only when >20% of the available units were confirmed caribou use locations (i.e., contaminated). These empirically based simulations suggest that previously recommended sampling designs are robust to contamination. We conclude with a new validation method for evaluating predictive performance of a RSF and for assessing if the model deviates from being proportional to the probability of use of a resource unit.

Foraging across a variable landscape: behavioral decisions made by woodland caribou at multiple spatial scales

We examined the foraging behavior of woodland caribou (Rangifer tarandus caribou) relative to the spatial and temporal heterogeneity of their environment. We assessed (1) whether caribou altered their behavior over time while making trade-offs between forage abundance and accessibility; and (2) whether foraging decisions were consistent across spatial scales (i.e., as scale increased, similar decision criteria were used at each scale). We discuss whether caribou adjusted their behavior to take advantage of changing forage availability through time and space. At the scale of the feeding site (as revealed by discriminant function analyses), caribou in both forested and alpine (above tree-line) environments selected sites where the biomass of particular lichen species was greatest and snow the least deep. Caribou did not select those species with the highest nutritional value (i.e., digestible protein and energy) in either area. Where snow depth, density, and hardness limited access to terrestrial lichens in the forest, caribou foraged instead at those trees with the greatest amount of arboreal lichen. Selection of lichen species and the influence of snow differed across time, indicating that in this system the abundance or accessibility of forage temporally influenced foraging behavior. A path analysis of forest data and multiple regression analysis of alpine data were used to test the hypothesis that variables important at the scale of the feeding site explained foraging effort at the scale of the patch. For forest patches, our hypothesized model reliably explained foraging effort, but not all variables that were statistically important at the scale of the feeding site were significant predictors at the scale of the patch. For alpine patches, our hypothesized model did not explain a statistically significant portion of the variation in the number of feeding sites within the patch, and none of the individual variables from the feeding site remained statistically significant at the patch scale. The incongruity between those variables important at the scale of the feeding site and those important at the patch showed that spatial scale affects the foraging decisions of woodland caribou. At the scale of the landscape, there was a trade-off between forage abundance and accessibility. Relative to the alpine environment, caribou in the forest foraged at feeding sites and patches with greater amounts of less variably distributed lichen, but deeper less variable snow depths. Considering the behavioral plasticity of woodland caribou, there may be no distinct advantage to foraging in one landscape over the other.

A MULTISCALE BEHAVIORAL APPROACH TO UNDERSTANDING THE MOVEMENTS OF WOODLAND CARIBOU

We assessed the response of woodland caribou (Rangifer tarandus caribou) to land-cover type, predation risk, energetic costs of movement, and patch configuration at multiple spatial scales. We applied a nonlinear model to frequent locations collected with Global Positioning System (GPS) collars to identify discontinuities in the scales of movement by caribou found in forested and alpine (above tree line) habitats. We differ- entiated intra- from interpatch movements and identified collections of patches (multiple- patch scale) where caribou concentrated intrapatch movements. On average, intra- and interpatch movements were 450.7 and 1268.8 m, respectively, and multiple-patch move- ments occurred over an area of 182 ha. Intrapatch movements were highly correlated, indicative of a strong relationship between behavior and place. Caribou in the forest selected patches of Pine terrace, whereas caribou in the alpine selected patches of Alpine-little vegetative cover. Predation risk was not a factor influencing movements of caribou at the intrapatch scale. Selection of cover types was more variable during interpatch movements. At that scale, caribou selected patches of Pine terrace, Lakes/rivers, Alpine-little vegetative cover, and Alpine-grass. The routes selected by caribou had lower energetic costs relative to surrounding terrain, and during some winters, caribou were subjected to higher levels of predation risk during those movements. At the multiple-patch scale, selection was more specific and encompassed patches of Alpine-little vegetative cover, Alpine-grass, and Pine terrace. Predation risk was relatively unimportant at the multi-patch scale, but animals that moved from forested to alpine habitats reduced their relative risk of predation. Patch con- figuration was a poor predictor of those areas where caribou concentrated intrapatch move- ments. There was some evidence of caribou selecting patches of Pine terrace within a matrix of Wetlands and Pine-black spruce/black spruce patches. Caribou in the alpine avoided patches of Alpine-little vegetative cover adjacent to forest types. Our results indicate that forest managers should maintain widely distributed patches of Pine terrace and implement silvicultural regimes that do not stimulate predator populations across areas used for in- terpatch movements.

An evaluation of mapped species distribution models used for conservation planning

SUMMARY The widespread use of spatial planning tools in conjunction with increases in the availability of geographic information systems and associated data has led to the rapid growth in the exploration and application of species distribution models. Conservation professionals can choose from a considerable number of modelling techniques, but there has been relatively little evaluation of predictive performance, data requirements, or type of inference of these models. Empirical data for woodland caribou Rangifer tarandus caribou was used to examine four species distribution models, namely a qualitative habitat suitability index and quantitative resource selection function, Mahalanobis distance and ecological niche models. Models for three sets of independent variables were developed and then a temporally independent set of caribou locations evaluated predictive performance. The similarity of species distribution maps among the four modelling approaches was also quantified. All of the quantitative species distribution models were good predictors of the validation data set, but the spatial distribution of mapped habitats differed considerably among models. These results suggest that choice of model and variable set could influence the identification of areas for conservation emphasis. Model choice may be limited by the type of species locations or desired inference. Conservation professionals should choose a model and variable set based on the question, the ecology of the species and the availability of requisite data.

Temperature and Flow Effects on Migration Timing of Chinook Salmon Smolts

Abstract Physiological and behavioral changes occur in the spring when juvenile Pacific salmon Oncorhynchus spp. undergo smolting. Survival is maximized if the timing of these changes coincides with migration from fresh to marine environments. Therefore, understanding how environmental conditions influence the onset, duration, and termination of smolting can have substantial management implications, particularly for flow-controlled rivers. We used an information-theoretic model comparison analysis to investigate the roles of daily mean temperature, temperature experience (accumulated thermal units [ATU]), photoperiod, and flow on the timing of the downstream migration of Chinook salmon O. tshawytscha smolts from the Nechako River in central British Columbia. Both binary (migration or not) and count (the total number of migrants) models were developed that predicted the downstream migration of Chinook salmon based on data collected from fish captured at rotary-screw traps from 1992 to 2004. The analyses id...

Toward rigorous use of expert knowledge in ecological research

Practicing ecologists who excel at their work (“experts”) hold a wealth of knowledge. This knowledge offers a wide range of opportunities for application in ecological research and natural resource decision-making. While experts are often consulted ad-hoc, their contributions are not widely acknowledged. These informal applications of expert knowledge lead to concerns about a lack of transparency and repeatability, causing distrust of this knowledge source in the scientific community. Here, we address these concerns with an exploration of the diversity of expert knowledge and of rigorous methods in its use. The effective use of expert knowledge hinges on an awareness of the spectrum of experts and their expertise, which varies by breadth of perspective and critical assessment. Also, experts express their knowledge in different forms depending on the degree of contextualization with other information. Careful matching of experts to application is therefore essential and has to go beyond a simple fitting of the expert to the knowledge domain. The standards for the collection and use of expert knowledge should be as rigorous as for empirical data. This involves knowing when it is appropriate to use expert knowledge and how to identify and select suitable experts. Further, it requires a careful plan for the collection, analysis and validation of the knowledge. The knowledge held by expert practitioners is too valuable to be ignored. But only when thorough methods are applied, can the application of expert knowledge be as valid as the use of empirical data. The responsibility for the effective and rigorous use of expert knowledge lies with the researchers.

Displacement of Mountain Caribou From Winter Habitat by Snowmobiles

Abstract Mountain caribou are an ecotype of woodland caribou (Rangifer tarandus caribou) that live in subalpine forests in south-eastern British Columbia, Canada, extending into northern Idaho and Washington, USA. These caribou are listed as Threatened in Canada, Endangered in the United States, and are the subject of recovery planning efforts in both countries. Many areas of mountain caribou winter habitat experience intensive use by recreational snowmobilers. During 4 surveys, we recorded caribou on all 4 census blocks with little or no snowmobile activity (x̄ density = 0.41 caribou/km2), but during 3 of 4 years, we observed no caribou on the census block with intensive snowmobile activity. The year we observed caribou on the snowmobile block, most were using areas inaccessible to snowmobiles. We used a Resource Selection Function (RSF) based on radiotelemetry data for the area to compare habitat quality among the different census blocks. The absence of caribou from the intensive snowmobile area during most years could not be explained by differences in habitat quality. The RSF predicted that the intensive snowmobile area could support 53–96 caribou (95% CI). We conclude that intensive snowmobiling has displaced caribou from an area of suitable habitat. We recommend that snowmobile activity be restricted from all or most high-quality mountain caribou habitat as part of the recovery planning process.

Expectations and Realities of GPS Animal Location Collars: Results of Three Years in the Field

GPS collars have the potential to automatically collect large numbers of relatively accurate animal relocations. Collar costs, levels of accuracy, and satellite signal reception have been reported by other studies, but there has been little discussion of long-term performance under field conditions. Between March 1996 and April 1999, we placed 11 GPS collars on 23 individual woodland caribou Rangifer tarandus caribou for a total of 26 collar deployments. Reliability was highly variable; some collar deployments operated normally for their expected period of time, other deployments functioned for less than half of their expected lives. Collars attempted 41,822 locations and collected 15,247 3-D and 10,411 2-D locations, for an average acquisition rate of 59%. We recommend that researchers carefully consider project objectives, budget constraints, and available options such as differential correction and remote collar communication, before purchasing GPS collars.

Maintaining or restoring connectivity of modified landscapes: evaluating the least-cost path model with multiple sources of ecological information

Habitat connectivity is an important element of functioning landscapes for mobile organisms. Maintenance or creation of movement corridors is one conservation strategy for reducing the negative effects of habitat fragmentation. Numerous spatial models exist to predict the location of movement corridors. Few studies, however, have investigated the effectiveness of these methods for predicting actual movement paths. We used an expert-based model and a resource selection function (RSF) to predict least-cost paths of woodland caribou. Using independent data for model evaluation, we found that the expert-based model was a poor predictor of long-distance animal movements; in comparison, the RSF model was effective at predicting habitat selection by caribou. We used the Path Deviation Index (PDI), cumulative path cost, and sinuosity to quantitatively compare the spatial differences between inferred caribou movement paths and predicted least-cost paths, and quasi-random null models of directional movement. Predicted movement paths were on average straighter than inferred movement paths for collared caribou. The PDI indicated that the least-cost paths were no better at predicting the inferred paths than either of two null models—straight line paths and randomly generated paths. We found statistically significant differences in cumulative cost scores for the main effects of model and path type; however, post-hoc comparisons were non-significant suggesting no difference among inferred, random, and predicted least cost paths. Paths generated from an expert based cost surface were more sinuous than those premised on the RSF model, but neither differed from the inferred path. Although our results are specific to one species, they highlight the importance of model evaluation when planning for habitat connectivity. We recommend that conservation planners adopt similar techniques when validating the effectiveness of movement corridors for other populations and species.

Relationship between resource selection, distribution, and abundance: a test with implications to theory and conservation

Much of applied and theoretical ecology is concerned with the interactions of habitat quality, animal distribution, and population abundance. We tested a technique that uses resource selection functions (RSF) to scale animal density to the relative probability of selecting a patch of habitat. Following an accurate survey of a reference block, the habitat-based density estimator can be used to predict population abundance for other areas with no or unreliable survey data. We parameterized and tested the technique using multiple years of radiotelemetry locations and survey data collected for woodland caribou across four landscape-level survey blocks. The habitat-based density estimator performed poorly. Predictions were no better than those of a simple area estimator and in some cases deviated from the observed by a factor of 10. We developed a simulation model to investigate factors that might influence prediction success. We experimentally manipulated population density, caribou distribution, ability of animals to track carrying capacity, and precision of the estimation equation. Our simulations suggested that interactions between population density, the size of the reference block, and the pattern of distribution can lead to large discrepancies between observed and predicted population numbers. Over- or undermatching patch carrying capacity and precision of the estimator can influence predictions, but the effect is much less extreme. Although there is some empirical and theoretical evidence to support a relationship between animal abundance and resource selection, our study suggests that a number of factors can seriously confound these relationships. Habitat-based density estimators might be effective where a stable, isolated population at equilibrium is used to generate predictions for areas with similar population parameters and ecological conditions.