Windscapes and olfactory foraging in a large carnivore

Abstract

The theoretical optimal olfactory search strategy is to move cross-wind. Empirical evidence supporting wind-associated directionality among carnivores, however, is sparse. We examined satellite-linked telemetry movement data of adult female polar bears (Ursus maritimus) from Hudson Bay, Canada, in relation to modelled winds, in an effort to understand olfactory search for prey. In our results, the predicted cross-wind movement occurred most frequently at night during winter, the time when most hunting occurs, while downwind movement dominated during fast winds, which impede olfaction. Migration during sea ice freeze-up and break-up was also correlated with wind. A lack of orientation during summer, a period with few food resources, likely reflected reduced cross-wind search. Our findings represent the first quantitative description of anemotaxis, orientation to wind, for cross-wind search in a large carnivore. The methods are widely applicable to olfactory predators and their prey. We suggest windscapes be included as a habitat feature in habitat selection models for olfactory animals when evaluating what is considered available habitat.

Figure 1. Study area in Hudson Bay, Canada.

Shaded area represents the western Hudson Bay (WH) polar bear subpopulation management boundary. Map was created using QGIS version 2.14 ESSEN (http://www.qgis.org/en/site/).

Figure 2. Movement relative to north (0°).

Frequency of polar bear orientation during (a) summer, (b) autumn, (c) freeze-up, (d) winter, and (e) break-up. Curves represent probability density functions based on maximum likelihood of a mixture of two (for a, d and e) and a single (for b and c) von Mises-Fisher distributions.

Figure 3. Movement relative to wind on land.

Frequency of polar bear orientation during (a) summer and (b) autumn while wind speed was <36 km/h and polar bear speed was <2 km/h. Curves represent probability density function based on maximum likelihood of a mixture of two von Mises-Fisher distributions.

Figure 4. Movement relative to wind during freeze-up.

Frequency of polar bear orientation relative to wind while (a) polar bear speed was <2 km/h or wind speed was >21.6 km/h and (b) polar bear speed was >2 km/h and wind speed was <21.6 km/h. Curves represent probability density functions based on maximum likelihood of a single (for a) and a mixture of two (for b) von Mises-Fisher distributions.

Figure 5. Movement relative to wind during winter.

Frequency of polar bear orientation relative to wind while polar bear speed was <2 km/h or wind speed was >36 km/h (a and c - representing two collar types, see below), and while polar bear speed was >2 km/h and wind speed was <36 km/h (b and d - representing two collar types, see below). (a and b) represent collars that had 4-hour fix intervals while (c and d) represent collars that had 30-minute fix intervals. (b) is subset into day (light grey) and night (dark grey). Curves represent probability density functions based on maximum likelihood of a single (for a and c) and a mixture of two (for b and d) von Mises-Fisher distributions.

Figure 6. Movement relative to wind during sea ice break-up.

Frequency of polar bear orientation relative to wind bearings. Curve represents probability density function based on maximum likelihood of a von Mises-Fisher distribution.