We examine the effect of red wood ants on the spatial distribution of carabids (picture) and describe mechanisms of the ant-beetle interactions at the individual level. Measurements of running speed, duration of stops, and individual trajectories have shown significant alterations in the behaviour of beetles in the ant-controlled territory. It was demonstrated that red wood ants strongly affect the spatial distribution of carabids and change their behavioural patterns at the level of individual direct interaction. Carabids can apply species-specific sets of behavioural tactics to avoid contact with the ants. The species that are comparable with red wood ants in body size and running speed (such as P. magus and P. oblongopunctatus) possess the most effective stereotyped tactics in combination with a flexible behaviour. Such properties allow these species to penetrate ant foraging territory and partly avoid interference competition.

Background for this study is that a spatial texture of habitats helps to support diverse species assemblages in a landscape. The presence of environmental "grains" generates a patchy habitat structure, increasing regional species richness. Habitat patches may include mountaintops, areas of forest of certain age, forest gaps, or even individual trees or anthills. Engineering species, such as beavers and ants, can increase landscape-level heterogeneity, creating patches with a peculiar combination of environmental conditions. On the other hand, within their large feeding territories the ants may create "black holes" in the habitat, i.e. areas that are highly dangerous for other species, where intruders can be killed or at least injured. Carabids seem to be appropriate group for studying the interference competition between ants and other arthropods. Both ants and carabids are generalist predators of comparable size and are very abundant in forest habitats. To examine the ability of carabids to avoid clashes with ants, simple Y-shaped mazes were used. One section was empty, while the other contained an active ant tied up by a thin thread. It had been shown in preliminary experiments that tethered ants attack and bite approaching beetles ( picture, picture1, picture2, picture3, picture4, picture5, picture7). Meeting of insects rarely ended with ant's murder.

In the main series of trials with Y-shaped mazes a beetle was placed at the entrance of the labyrinth, and subsequent events were recorded. To avoid the possible influence of smell tracking, the paper on the floor of the labyrinths was changed after each test. To exclude possible influence of preference for the left or the right section, control tests for all individuals were conducted in empty labyrinths, lacking any stimuli.

The ability of the beetles to avoid collision with ants was estimated as a ratio between errors and 'correct actions'. We defined the 'correct action' as a modification of behaviour that allowed the beetle to avoid a clash successfully. Such modifications included both avoidance of the dangerous section of the labyrinth and use of specific behavioural tactics such as (1) attempted to round the ant ( picture1, picture2, picture3, picture4, picture5, picture6); (2) turned away after touching the ant with antennae; (3) turned away at a short distance (but not less then 1 cm) from the ant ( picture); (4) avoided the section with the ant; (5) stopped near the ant with legs and antennae hidden. Actions that resulted in clashes between insects were considered as errors.

Laboratory experiments showed that different carabid species used species-specific sets of behavioural tactics for avoiding conflict with ants (see: Reznikova, Dorosheva, 2004).

Currently we use model of beetles in order to examine what stimuli release aggression in ants and how exactly and at what age after hatching from cocoons ants can distinguish "images of enemies". We compare ants' reaction for alive beetles in small arena ( picture) and for models of different sizes, shapes, colour, that possess or lack legs ( picture, picture). Preliminary results have shown that ants aggressively react to dark models and that presence of legs serve as one of main releasers already at early age of the ant.