Search

Two species of Australian desert ants, one inhabiting a visually rich environment (Melophorus bagoti) and one inhabiting a visually barren environment (as yet unnamed and dubbed Melophorus sp.), were tested on path integration and subsequent search. For each species, prominent landmarks were placed near the nest, and ants were trained to visit a feeder. After training over 2 days, an ant was captured at the feeder when it had grabbed some food and tested, just once individually, in homing with the training landmarks either present or absent. Their subsequent search was also recorded on gridded paper according to a grid marked at the test site. Both species headed initially in the feeder-nest direction, but directional scatter was larger when training landmarks were absent. Melophorus bagoti ran a shorter distance before starting to search on tests with the landmarks absent, but Melophorus sp. ran a similar distance in both conditions. In both species, the variance in distance run was larger when landmarks were absent. In searching, both species expanded their search pattern more when landmarks were absent than when landmarks were present. Contrary to predictions, the distribution of the lengths of search segments was best described as a single exponential function on tests with landmarks absent and as a double exponential function with landmarks present. Two species of Australian desert ants of the same genus but occupying visually different habitats were studied in similar experiments. One inhabits a barren saltpan while the other inhabits a semi-arid environment filled with grass, shrubs and trees. The ants were compared in their return home after a short outbound trip (2 m) to a feeder and the subsequent search for the nest. When the ants were trained in an environment with some prominent artificial landmarks which were subsequently removed in tests, ants living in the visually cluttered habitat cut short their initial run, starting to engage in search earlier than ants living on the salt pan. Search characteristics were, on the other hand, similar in the two species, both expanding their search loops over time, more so when the familiar scene was changed. Visual ecology drives some aspects of navigation, while common task requirements drive other aspects.

Honeybees (Apis mellifera) are regularly faced with the task of navigating back to their hives from remote food sources. They have evolved several methods to do this, including compass-directed “vector” flights and the use of landmarks. If these hive-centered mechanisms are disrupted, bees revert to searching for the hive, using an optimal Levy flight searching strategy. The same strategy is adopted when a food source at a known location ceases to be available. Here, we show that the programming for this Levy strategy does not need to be very sophisticated or clever on the bee’s part, as Levy flight patterns can be derived from the Weber–Fechner law in a bee’s odometer. Odometry errors of a different kind occur in desert ants (Cataglyphis spp., Melophorus bagoti). The searching behaviors of these ants are very similar in overall structure to that of honeybees but do not display any Levy flight characteristics. We suggest that errors in the estimation of distance can be implicitly involved in shaping the structure of systematic search behavior and should not be regarded as merely deficiencies in the odometer.