Open Access

Eva Kaufmann

• For palaeotropical regions, only a few anecdotal reports had been published on the existence of 'ant-gardens' before this study started. As opposed to this, 'ant-house epiphytes' (i.e. domatiabearing epiphytes) were reported to be highly abundant in Southeast Asia and were presumed to be a second type of ant-epiphyte interaction. In the much better studied neotropical regions the situation seemed to be the reverse: Many reports on AGs in contrast to very few reports on anthouse epiphytes. In this study, I have presented extensive data which may help towards a better understanding of the 'Southeast Asian part' of this 'ant-epiphyte puzzle'. In Peninsular Malaysia, Borneo, Java, and Southern Thailand, a great variety of formerly unknown AG systems were discovered. 18 ant species (from 5 genera, 4 subfamilies) were identified as true AG ants, i.e. these ants actively retrieved seeds of certain epiphyte species into their carton nests. Another 49 ant species inhabited AGs as secondary, opportunistic settlers. On the epiphyte side, 84 plant species were found growing on AGs, 51 (19 genera, 12 families) of which were probably true AG epiphytes, i.e. ants retrieved the seeds to their arboreal carton nests, on which the epiphytes were then cultivated. Most of the epiphyte flora of lowland forests in Peninsular Malaysia (except for ferns, orchids and facultative epiphytes) seemed to be totally dependent on ants for their establishment in the canopy. Together with the high number of opportunistic AG inhabitants (ants, epiphytes, and many arthropod guests), these facts suggest that AGs function as pioneers in the canopy of Southeast Asian rain forests. Moreover, AG-associations might even have accounted for the unusual species richness in the epiphyte genera Dischidia, Hoya (Asclepiadaceae), Myrmecodia, and Hydnophytum (Rubiaceae). The definition of the term ant-garden only describes the basic interactions. In the ant-garden associations investigated in this study, interactions going beyond these basic ones varied depending on ant and epiphyte species. Ant-gardens initiated by Diacamma spKfmA111 were regarded as the 'most primitive' type, because this ponerine was totally dependent on preformed cavities for nest establishment, did not tend any trophobionts, and was the least selective in its seed-retrieving behavior. On the other end of the scale, Crematogaster spKfmA18 and Camponotus spKfmA9 were rated as 'most advanced' because both lived in free (i.e. cavityindependent) AGs, tended trophobionts underneath their nests, were associated with a couple of other organisms, and were highly selective in their seed-retrieving behavior. Moreover, Camponotus spKfmA9 occurred preferentially with one single epiphyte species, Hoya elliptica (Asclepiadaceae), and Crematogaster spKfmA18 was specialized on some species of giant bamboo as phorophyte. Philidris spKfmA160, which occupied a medium position in relation to the other AGs was particularly interesting for several reasons. This ant species was mainly associated with ant- house epiphytes and occurred in the heath forests of Borneo. However, the major part of the colonies, including the queen, was located underneath carton structures near the surface of the host tree and not inside the domatia of the associated plants. Moreover, very young Philidris spKfmA160 colonies had only small seedlings growing on their carton nests. The ant workers actively retrieved the seeds of their epiphyte partners into the nests. These results indicate that associations with ant-house epiphytes must be regarded as a special case of ant-gardens. I therefore suggest using the term 'ant-house' only to describe the epiphytes, but not to describe the association, and to include this type of association in the group of AGs. Strict species-specificity never occurred, but some epiphytes showed great preference for growing on the nests of certain ant species, while others occurred over a wider range. Vice versa, most ant species had several epiphytes growing on their nests, while others were mostly found with one or very few epiphyte species. These patterns were shown to be the effect of different factors, including common microclimatic preferences of ants and epiphytes, interspecific competition of epiphytes, and selective seed retrieval of AG ants. The main behavioral trait responsible for the establishment of AGs was the selectivity shown by the ants in the epiphyte seeds they carried. However, details of the mechanisms, i.e. what characteristics of the seeds are important and what motivates the ants to retrieve them, varied widely. In many cases, seed compounds located on the surface triggered carrying behavior. Detailed experimental investigations combined with literature data from the two other known 'myrmecochory systems', terricolous myrmecochores and neotropical AGs, suggested that myrmecochory is frequently triggered by a two-stage system. One relatively unspecific compound (or a combination of such compounds) constitutes the basic attractiveness for a number of ant species. Other seed characteristics (elaiosomes, mechanical properties, other surface-compounds) modulate this basic signal, accounting for species-specific preferences of ants towards certain plant species. A comparison of AGs in Southeast Asia and the neotropics shows that the numbers of AG ant and epiphyte species in each case are almost equal. Southeast Asian AG epiphytes might even turn out to outnumber the neotropical ones. Thus, not only was it possible to break down the distinction between ant-house and AG associations, but also to show that AGs in Southeast Asia are present in such high diversity and abundance as to diminish the apparent contrast between the two biogeographical regions yet further. These data help to solve at least the Southeast Asian part of the 'ant-epiphyte puzzle'.