Open Access

Heike B. Feldhaar

• One of the most species-rich ant-plant mutualisms worldwide is the palaeotropical Crematogaster-Macaranga system. The pioneer-tree genus Macaranga (Euphorbiaceae) is mainly inhabited by at least nine specific species of Crematogaster (Myrmicinae), of which eight belong to the subgenus Decacrema, as well as several species of Camponotus (Formicinae). Ant species are not randomly distributed among the Macaranga host plants but distinct patterns of associations have been found (Fiala et al., 1999 and references cited therein). The specificity of the associations is maintained in spite of common sympatric distribution of several host-plant species. Associations are, however, usually not species-specific and especially the Decacrema ants, that are the focus of this study, usually colonize several host plant species each. In this study I used a combined approach of ecological data as well as phylogenetic data based on mitochondrial DNA sequences in order to elucidate the factors determining the patterns found in the associations and the evolution of this mutualistic system between the specific Decacrema ant partners and their Macaranga host plants. Life history traits of seven different morphospecies found on the most common Macaranga host plants were compared and colony development was followed from colony founding on saplings to adult trees. Temporal variability of the associations between Decacrema ants and their respective host plants was also examined. Associations between Crematogaster ants of the subgenus Decacrema and their Macaranga host plants were found to be stable over periods of time, long enough to enable reproduction of the ant colony and (in most cases) the host plants, too. Life-expectancy of the ant colony seems to be shorter than that of the host plant in general. All adult trees still provide nesting space as well as food for the ants. Colonies from different morphospecies differed in longevity, the onset of alate production, queen number and mode of colony founding. The examined Decacrema species could be placed into two groups according to their life-history traits as well as on morphological grounds: The decamera-group and the captiosa-group, each named after one species that could be synonymized with one morphospecies included in the group. Members of the captiosa-group have larger colonies, presumably with a longer life-span, and a later onset of reproduction compared to the decamera-group. Additionally, queens of the captiosa-group found colonies on saplings as well as in the crown region of bigger trees, whereas queens of the decamera-group found colonies on saplings and small treelets only. Queens belonging to the captiosa-group are brown with relatively large eyes (= 1/3 of the head length), whereas queens from the decamera-group are smaller in size, are dark brown to black in colour and have smaller eyes (< 1/3 of the head length). On some of the host plants examined in this study lifespan of the host plant and their specific ant partners seemed to be well matched whereas on others an ontogenetic succession of specific Decacrema partner ants was found, when host plants were abandoned due to the death of comparatively short-lived ant colonies, usually from species belonging to the decamera-group. Ant-partners of saplings or young plants often differed from specific partner ants found on bigger trees. Only species belonging to the captiosa-group were found to re-colonize the crown region of adult trees, thus facilitating a change of ant species, when longlived host plant species were colonized by relatively short-lived species from the decamera-group first. When long -lived host plants were colonized by long-lived species from the captiosa-group associations were stabler: I did not find any temporal variation in ant-inhabitants then. Life-span of the ant colony as well colony founding behaviour of the different partner ant species therefore play an important role for these ontogenetic changes and the specificity of the associations over time. For the host plant the ontogenetic changes have a strong impact as uninhabited host plants that are not patrolled by workers of specific ant partners suffer higher herbivore damage. Uninhabited host plants may also be colonized by unspecific arboreal ants that only make use of the nesting space and/ or food offered by the plant but do not confer protection against herbivores. Stable associations with a specific ant partner are therefore most beneficial for the host plants. Usually ant colonies are monogynous, but changes in the colony structure were found locally in two Decacrema species. I found colonies that turned secondarily polygynous, possibly after the death of the original founding queen. Secondary polygyny therefore can prolong the life-span of the antcolony on its host plant, leading to a parallel life-history and stable association as it was the case in Macaranga bancana-Crematogaster captiosa. However, in the other association (Macaranga hypoleuca-Crematogaster cf. decamera) life-expectancy of the ant-colony is still much shorter than that of its host plant species, leading to a change in the specific ant partner at a later stage. Pleometrotic foundress associations that directly led to polygynous colonies in one species were also found locally, a phenomenon hardly ever reported from ants in general. Foundress associations were found to be more successful in establishing colonies than single queens. I found indications that this change in colony founding behaviour might be due to interspecific competition for the same host plant species with another Decacrema species specific to Macaranga. For the phylogenetic analysis partial mitochondrial cytochrome oxidase I and II were sequenced and Neighbor-Joining, Maximum Parsimony, Maximum Likelihood as well as Bayesian analyses were performed. The four different analyses yielded phenetic as well as phylogenetic trees that all had a similar topology. Ants of the subgenus Decacrema formed a monophyletic clade, indicating a single colonization event at the beginning of the Macaranga-Decacrema symbiotic system. In the phylogenetic analysis the decamera-group as well as the captiosa-group were confirmed and clearly separated from each other. However, two species that would have been placed into the decamera-group, due to morphological as well as life-history traits, formed a third separate clade within the Decacrema. These two species (msp. 7- group) as well as the decamera-group came out as the basal groups in the phylogenetic analysis. Thus, life -history traits of these two groups (relatively small colonies, early onset of alate production, colony founding in ground region only) would be the ancestral state for Macarangaassociated ants of the subgenus Decacrema. Changes in colony structure, like secondary polygyny, were found in the captiosa- as well as the decamera-group and are therefore independent of the affiliation within the phylogeny. I did not find evidence for strict cocladogenesis between the subgenus Decacrema and their Macaranga host-plants, although ecological interactions between the two partner groups are close and associations can be rather specific. The phylogenies presented here, along with the known association patterns indicate that host-shifting of the ants is common in some of the species, opening the possibility of sympatric speciation as a result of increased host usage. Additionally, the considerable geographic substructuring found in the phylogenetic trees suggests that allopatric speciation has played a major role in diversification of the Decacrema ants.