Desert ants, Cataglyphis, are known to rely on path integration [1, 2, 5] and visual landmarks [3, 4] during homing. In the present study, we ask whether ants associate their nest entrance with environmental odours. By collecting and analysing volatiles from different positions within the salt-pan habitat of C. fortis, we show that the environment provides the ants with potential odour-landmark information, i.e. with different place-specific blends (Figure 1). The blend components that were identified induced EAG responses in these ants. Can the ants make use of such information? Yes, they can. Homing ants that were trained to the nest marked by an odour focused their nest search on this odour in a test situation, but they did not search at the solvent control (Figure 3, ). Hence, Cataglyphis ants can learn the association between the nest entrance and an environmental odour and use this information for homing.
The salt-pan habitat is not an odour-free space, interrupted by single odour peaks but, rather, is loaded with a variety of odours (Figure 1). Hence, when using olfactory information as landmarks, ants must be able to distinguish a learned odour from other odours. As the ability to discriminate among odours relates to the dissimilarity among molecules, bees frequently confuse odours that share functional groups or have similar chain lengths . The most similar among the four odours tested in the present account were the aldehydes nonanal and decanal; they have the same functional group and chain lengths that differ by only one carbon atom, whereas methyl salicylate and indole have different chemical structures (Figure 3A). Even so, the ants were able to discriminate among all of them (Figure 3). Ants that were tested with a non-trained odour did not avoid this odour during the nest search as naïve ants seemed to do (Additional file 1). The training on one odour resulted in an odour-specific response by homing ants, which is one prerequisite for odour-landmark navigation.
The use of odour landmarks requires a further skill: the olfactory background of a stimulus might change dramatically when for example the wind direction changes and an odour source suddenly appears upwind of the landmark. In order to use an olfactory landmark, an ant must be able to identify the learned odour against a changing background of odours. Cataglyphis ants fulfilled this demand when trained on indole and tested with a blend of four odours including indole. Ants that were tested with the blend showed a less focused search than ants that were tested with the trained odour only, but still a more focused search than ant tested with the solvent control (Figure 4).
Hence, the ants can use olfactory information for homing. Trail following is the predominant means of orientation in a large number of ant species . These trails always consist of ant-derived trail pheromones. Unlike orientation that is guided by pheromones, orientation that uses non-pheromonal chemical cues is less well investigated. Carpenter ants can be trained to search for food on specific species of trees. Trained ants decide for the right tree species even when tactile cues are experimentally excluded. Therefore, the ants seem to use tree-derived chemical cues . Bees have been shown to associate different odours with different feeding places. Blowing a learned odour into the hive triggered the trained bees to visit the corresponding feeder, i.e. navigational memory can be evoked just by providing the learned odour . Finally, it has been shown that around cities, spatial gradients of atmospheric volatiles exist  and seem to be used by pigeons to pinpoint their loft . Unlike bees and pigeons, the desert ant Cataglyphis has so far been a model only for visually guided orientation.
We are amazed to discover that while keeping track of the path integrator and learning visual landmarks, these ants can also collect information about the olfactory world. In future experiments, we hope to clarify how visual and olfactory landmarks interact to provide accurate information regarding, for example, nest location.