Spatial reorientation by geometry in bumblebees

Abstract

Human and non-human animals are capable of using basic geometric information to reorient in an environment. Geometric information includes metric properties associated with spatial surfaces (e.g., short vs. long wall) and left-right directionality or 'sense' (e.g. a long wall to the left of a short wall). However, it remains unclear whether geometric information is encoded by explicitly computing the layout of surface geometry or by matching images of the environment. View-based spatial encoding is generally thought to hold for insect navigation and, very recently, evidence for navigation by geometry has been reported in ants but only in a condition which does not allow the animals to use features located far from the goal. In this study we tested the spatial reorientation abilities of bumblebees (Bombus terrestris). After spatial disorientation, by passive rotation both clockwise and anticlockwise, bumblebees had to find one of the four exit holes located in the corners of a rectangular enclosure. Bumblebees systematically confused geometrically equivalent exit corners (i.e. corners with the same geometric arrangement of metric properties and sense, for example a short wall to the left of a long wall). However, when one wall of the enclosure was a different colour, bumblebees appeared to combine this featural information (either near or far from the goal) with geometric information to find the correct exit corner. Our results show that bumblebees are able to use both geometric and featural information to reorient themselves, even when features are located far from the goal.

Figure 1. Schematic view of the test apparatus.

Corners are conventionally identified as C = Correct, R = Rotation (geometrically equivalent corner), F = Far, N = Near. Only the external hole of the correct corner was opened allowing bumblebees to exit.

Figure 2. Top panels: the mean number of first choices made to each corner by the group of bumblebees (Exp. 1; groups means with SEM are shown) per session.

Bottom panels: the mean number of times (choice frequency) the same bees (group means with SEM are shown) visited each corner per session.

Figure 3. (A) The mean number of first choices made to each corner by the group of bumblebees (Exp. 2; groups means with SEM are shown) per session.

Top panels: Near-feature: the feature (white wall indicated by the horizontal grey stripe, all other walls are green) is located at the target corner C; Bottom panels: Far-feature: the feature (white wall indicated by the horizontal grey stripe, all other walls are green) is located away from the target corner C. (B) The mean number of times (choice frequency) bumblebees visited each corner (Exp. 2; groups means with SEM are shown) per session. Top panels: Near-feature: the feature (white wall indicated by the horizontal grey stripe, all other walls are green) is located at the target corner C; Bottom panels: Far-feature: the feature (white wall indicated by the horizontal grey stripe, all other walls are green) is located away from the target corner C.

Figure 4. The mean number of times (choice frequency) bumblebees visited each corner (group means with SEM are shown).

Data for the two testing sessions were pooled because there was no significant interaction in the general ANOVA between distance from the feature and session (Exp. 2). Left panel: Near-feature: the feature (white wall indicated by the horizontal grey stripe, all other walls are green) is located at the target corner C. Right panel: Far-feature: the feature (white wall indicated by the horizontal grey stripe, all other walls are green) is located away from the target corner C.