James A. Schaefer

Caribou movement as a correlated random walk

Abstract Movement is a primary mechanism coupling animals to their environment, yet there exists little empirical analysis to test our theoretical knowledge of this basic process. We used correlated random walk (CRW) models and satellite telemetry to investigate long-distance movements of caribou, the most vagile, non-volant terrestrial vertebrate in the world. Individual paths of migratory and sedentary female caribou were quantified using measures of mean move length and angle, and net squared displacements at each successive move were compared to predictions from the models. Movements were modelled at two temporal scales. For paths recorded through one annual cycle, the CRW model overpredicted net displacement of caribou through time. For paths recorded over shorter intervals delineated by seasonal behavioural changes of caribou, there was excellent correspondence between model predictions and observations for most periods for both migratory and sedentary caribou. On the smallest temporal scale, a CRW model significantly overpredicted displacements of migratory caribou during 3 months following calving; this was also the case for sedentary caribou in late summer, and in late winter. In all cases of overprediction there was significant positive autocorrelation in turn direction, indicating that movements were more tortuous than expected. In one case of underprediction, significant negative autocorrelation of sequential turn direction was evident, indicating that migratory caribou moved in straightened paths during spring migration to calving grounds. Results are discussed in light of known migration patterns and possible limiting factors for caribou, and indicate the applicability of CRW models to animal movement at vast spatial and temporal scales, thus assisting in future development of more sophisticated models of population spread and redistribution for vertebrates.

Habitat selection at multiple scales 1

Abstract: Habitat selection is the disproportionate use of available conditions and resources, and involves responses in space and time to perceived risks and rewards. It frequently depends on the scale of measurement, often in non-linear ways that preclude simple extrapolation across scales. More critically, animals often select different habitat components at different scales, and species vary in their scales of selection. Although multi-scaled research on habitat selection has proliferated, synthesis of this work has been impeded by imprecise terminology and arbitrarily defined analytical scales. Here, we review key concepts and findings and evaluate future prospects opened up in part by new technologies that enable novel and more efficient data collection. Innovative measurement, combined with novel analytical approaches, permits habitat selection to be investigated across a broad continuum of scales. By linking habitat selection to fitness as a function of scale, use-of-habitat assessments can be more effective. The fitness costs and benefits of habitat selection change with scale; the scales of habitat selection may inform us of limiting factors. We outline how reward/risk ratios can be used to assess the fitness implications of habitat selection across scales.

Woodland Caribou Extirpation and Anthropogenic Landscape Disturbance in Ontario

Abstract The decline of woodland caribou (Rangifer tarandus caribou) has been attributed to anthropogenic landscape disturbances, but critical distance thresholds and time lags between disturbance and extirpation are unknown. Using a database of caribou presence and extirpation for northern Ontario, Canada, geo-coded to 10 × 10-km cells, we constructed logistic regression models to predict caribou extirpation based on distance to the nearest of each of 9 disturbance types: forest cutovers, fires, roads, utility corridors, mines, pits and quarries, lakes, trails, and rail lines. We used Akaikes Information Criterion to select parsimonious models and Receiver-Operating Characteristic curves to derive optimal thresholds. To deal with the effects of spatial autocorrelation on estimates of model significance, we used subsampling and restricted randomizations. Forest cutovers were the best predictor of caribou occupancy, with a tolerance threshold of 13 km to nearest cutover and a time lag of 2 decades between disturbance by cutting and caribou extirpation. Management of woodland caribou should incorporate buffers around habitat and requires long-term monitoring of range occupancy.

Hydroelectric development and the disruption of migration in caribou

Abstract We investigated the effects of hydroelectric development on the movements and space-use of caribou ( Rangifer tarandus caribou ) in west-central Newfoundland, Canada. We compared patterns of range use, site fidelity, and timing of migration before, during, and after project construction. Coincidental with the first year of project construction, caribou were less likely to be found within 3 km of the site; this persisted at least 2 years after construction was completed. Relative timing of migration was individual-specific; the rank order of spring arrival on, and autumn departure from, the calving and summer grounds tended to be consistent year-to-year. This is the first report of such individual-specific consistency in migration for a non-avian species. This predictability disappeared during development: the year-to-year consistency of fall and spring migration among individuals was apparent before and after construction, but not during construction. Variation in calving site fidelity was correlated to year-to-year differences in snowfall. We conclude that the development caused a disruption of migrational timing during construction and longer-term diminished use of the range surrounding the project site. Long-term studies of individually marked animals can aid in environmental assessments for migratory animals.

LONG-TERM CHANGES IN DEMOGRAPHY AND MIGRATION OF NEWFOUNDLAND CARIBOU

Abstract Long-term studies of demography and movements are rare but critical to understanding long-lived mobile mammals like caribou (Rangifer tarandus). We studied changes in the abundance, vital rates, body size, and timing of migration of the Buchans Plateau caribou herd in Newfoundland, Canada. From the early 1960s to 2000, the population grew at 6.5%/year, although survival and recruitment indicated a declining growth rate (1.4%) by the late 1990s. The numerical increase was negatively associated with other population attributes. Rates of parturition, survival to 6 months of age, and recruitment diminished significantly. Adults exhibited substantial decreases in body size. Spring migration was significantly later and autumn migration significantly earlier. We surmise that these responses may reflect heightened density-dependent competition for summer forage.

Demography of decline of the Red Wine Mountains Caribou Herd

The causes of decline of woodland caribou (Rangifer tarandus caribou) populations remain incompletely understood. We compared population characteristics of woodland caribou of the Red Wine Mountains Herd (RWMH) in central Labrador before (1981-88) and during a population decline (1993-97). During the 1980s, population estimates were 751 (no error estimation) animals in 1981, 736 ± 172 (x ± SE) in 1983, 610 ± 9 in 1987, and 741 ± 165 in 1989. By 1997, the herd declined to 151 animals (95% CI = 65-251). The decline was not associated with changes in parturition rate or in mean age of >1-year-old females, but the decline was associated with significantly lower recruitment, a greater proportion of females in the >1-year-old population, increased mortality of >1-year-old females, and emigration to the parapatric George River Caribon Herd. Throughout the study. predation by gray wolves (Cauis lupus) remained the most frequent cause of mortality of >1-year-old caribou. We hypothesize that wolves may mediate the population dynamics of sed entary woodland caribon when associated with high densities of moose (Alces alces) and migratory caribon, but that the management implications of such a triad remain unclear.

Effects of Progressive Clearcut Logging on Newfoundland Caribou

Abstract Logging has often been implicated in the decline of caribou (Rangifer tarandus), but its effects are incompletely understood. We used a distance-based approach to assess the effects of progressive clearcut logging on the summer (28 May to 15 Sep) range of caribou in Newfoundland, Canada. We compared distances of random locations and of caribou, from 9 years of radiotelemetry, to landcover types across 3 spatial scales: population range, individual ranges, and radiolocations. We tested for incremental avoidance of cutovers and mature softwood forests, the preferred type for caribou and forest harvesting, while controlling for the confounding effects of each. At the individual range, females selected for hardwood and softwood forests, bogs, and barrens, and they avoided open water. Patterns for males were similar, although they avoided bogs and barrens at both scales. The sexes differed in their response to forest harvesting. Females progressively avoided cutovers, both pre- and postharvest, likely due to their spatial proximity. Females maintained an average of 9.2 km from active cutovers. Cutover avoidance was evident even if we controlled statistically for distances to other habitats, and it accounted for heightened disuse of softwood forests. Compared with females, males occurred in proximity, with no incremental response to clearcutting. These results imply deleterious effects of timber harvesting on female caribou. Long-term investigations will enhance our capacity to evaluate such anthropogenic habitat changes.

SPECTRUM OF SELECTION: NEW APPROACHES TO DETECTING THE SCALE-DEPENDENT RESPONSE TO HABITAT

Detecting habitat selection depends on the spatial scale of analysis, but multi-scale studies have been limited by the use of a few, spatially variable, hierarchical levels. We developed spatially explicit approaches to quantify selection along a continuum of scales using spatial (coarse-graining) and geostatistical (variogram) pattern analyses at multiple levels of habitat use (seasonal range, travel routes, feeding areas, and microsites). We illustrate these continuum-based approaches by applying them to winter habitat selection by woodland caribou (Rangifer tarandus caribou) using two key habitat components, Cladina lichens and snow depth. We quantified selection as the reduction in variance in used relative to available sites, thus avoiding reliance on correlations between organism and habitat, for which interpretation can be impeded by cross-scale correlations. By consistently selecting favorable habitat features, caribou experienced reduced variance in these features. The degree to which selection was accounted for by the travel route, feeding area, or microsite levels varied across the scale continuum. Caribou selected for Cladina within a 13-km scale domain and selected shallower snow at all scales. Caribou responded most strongly at the dominant scales of patchiness, implicating habitat heterogeneity as an underlying cause of multi-scale habitat selection. These novel approaches enable a spatial understanding of resource selection behavior.

Fuzzy structure and spatial dynamics of a declining woodland caribou population

Examining both spatial and temporal variation can provide insights into population limiting factors. We investigated the relative spatial and temporal changes in range use and mortality within the Red Wine Mountains caribou herd, a population that declined by approximately 75% from the 1980s to the 1990s. To extract the spatial structure of the population, we applied fuzzy cluster analysis, a method which assigns graded group membership, to space use of radio-tracked adult females, and compared these results to a hard classification based on sums-of-squares agglomerative clustering. Both approaches revealed four subpopulations. Based on the subpopulation assignments, we apportioned the number of animals, radio-days, calving events and mortalities across subpopulations before and after the decline. The results indicated that, as the herd declined, subpopulations were disproportionately affected. In general, subpopulations with the greatest range overlap with migratory caribou from the George River herd experienced comparative reductions in activity and increased mortality. The subpopulation with the least overlap exhibited the converse pattern. The infra-population imbalances were more pronounced when hard clustering was employed. Our results reiterate that refugia from other ungulates may be important in the persistence of taiga-dwelling caribou. We propose that changes across time and space are valuable assays of localised demographic change, especially where individuals exhibit spatial hyperdispersion and site fidelity.

Predicting local and non‐local effects of resources on animal space use using a mechanistic step selection model

Predicting space use patterns of animals from their interactions with the environment is fundamental for understanding the effect of habitat changes on ecosystem functioning. Recent attempts to address this problem have sought to unify resource selection analysis, where animal space use is derived from available habitat quality, and mechanistic movement models, where detailed movement processes of an animal are used to predict its emergent utilization distribution. Such models bias the animals movement towards patches that are easily available and resource-rich, and the result is a predicted probability density at a given position being a function of the habitat quality at that position. However, in reality, the probability that an animal will use a patch of the terrain tends to be a function of the resource quality in both that patch and the surrounding habitat. We propose a mechanistic model where this non-local effect of resources naturally emerges from the local movement processes, by taking into account the relative utility of both the habitat where the animal currently resides and that of where it is moving. We give statistical techniques to parametrize the model from location data and demonstrate application of these techniques to GPS location data of caribou (Rangifer tarandus) in Newfoundland. Steady-state animal probability distributions arising from the model have complex patterns that cannot be expressed simply as a function of the local quality of the habitat. In particular, large areas of good habitat are used more intensively than smaller patches of equal quality habitat, whereas isolated patches are used less frequently. Both of these are real aspects of animal space use missing from previous mechanistic resource selection models. Whilst we focus on habitats in this study, our modelling framework can be readily used with any environmental covariates and therefore represents a unification of mechanistic modelling and step selection approaches to understanding animal space use.