IN AND TACHIGALI 533

171 worker - hostplant(s) behavior inhabitant?

T. natalensis timid no n/a P. gracilis timid no n/a P. termitarius timid no n/a

P. kuenckeli aggressive no n/a P. insuavis aggressive yes Tachigali P. concolor (s.l.) aggressive yes Tachigali P. malignus aggressive yes Tachigali P. tachigaliae aggressive yes Tachigali P. vitabilis unknown unknown unknown P. viduus aggressive yes various, incl. Triplaris P. triplaridis aggressive yes Triplaris P. ultrix aggressive yes Triplaris P. mordax aggressive yes Triplaris P. triplarinus aggressive yes Triplaris P. dendroicus aggressive yes Triplaris

172 P. vitabilis unknown P. viduus Triplaris weigeltiana, T. melaenodendron (and other plant genera) P. triplaridis Triplaris weigeltiana, T. dugandii P. ultrix Triplaris dugandii P. mordax , T. cumingiana P. triplarinus Triplaris americana P. dendroicus Triplaris americana

Figures 171, 172. Fig. 171. Distribution of worker behaviour (timid/aggressive) and ant-plant associations in the Pseudomyrmex viduus group. The first three taxa are outgroups. Fig. 172. The species of Triplaris recorded serving as host in the Pseudomyrmex triplarinus subgroup, mapped on the phylogeny of the ants.

Although P. kuenckeli appears to nest primarily in dead branches, these can be large in size and have substantial contact with live plant tissue (Ward, pers. obs.), suggesting the possibility that the nest cavities extend into live branches, where scale insects are tended. Davidson (1997) argues that the carbohydrate-rich diet of homopteran- tending ants facilitates their adoption of ‘high tempo’ foraging activity and aggressive behaviour. In this way, the common ancestor of P. kuenckeli and the other P. viduus group species could have been primed for a plant-protective role. The ancestral host plant for the ant-plant mutualists of the P. viduus group (i.e. for the P. triplarinus subgroup plus P. concolor subgroup) cannot be inferred with certainty, since each subgroup is more or less restricted to its own host plant (Fig. 171). Of the two clades, however, the P. concolor subgroup appears to have more apomorphic workers and queens, at least with respect to morphology (Fig. 169), suggesting that association with Tachigali may be a derived condition relative to association with Triplaris. How species-specific are the associations between the ants and their host plants? The data are insufficient to provide a detailed answer, since many of the plant records are at the level of only. The best information comes from the Pseudomyrmex triplarinus subgroup, and here there is an indication of some degree of