this system have suffered from the lack of a precise Despite the prominence of the Azteca-Cecropia inphylogeny. Although the extensive morphological and teraction as the focus of extensive ecological investigabehavioral investigations have produced reliable taxtion, a reliable phylogeny of the Azteca ants has been onomies of the Azteca (Longino, 1989; 1991a,b; Shatlacking, primarily because many of the morphological tuck, 1994), they have not yielded a reliable phylogeny, and behavioral characters are phylogenetically uninprimarily because many character states are homoplasformative or conflicting. A phylogenetic analysis of a ious and conflicting (Harada and Benson, 1988; Lonselect set of Azteca ants, including six Cecropia inhabgino, 1991b).
itants and two non-Cecropia inhabitants, plus an out- Benson (1985) and Longino (1991b) have proposed that group taxon, is presented on the basis of mitochondrial at least two independent lines of Azteca have colonized DNA sequences. The evolutionary relationships de-Cecropia: The alfari complex, comprising Azteca alfari duced from the molecular data are analyzed with refand A. ovaticeps, share the following characters: hairless erence to ecological and morphological studies, speappendages, small queen size, deep ventral lobe on the cifically addressing the phylogenetic relationship of queen petiole, polydomous nests, and the lack of standing structurally and behaviorally ambiguous taxa, species setae on the tibiae and scapes (Longino, 1989). The muelcomplex groupings, and the colonization of Cecropia leri complex, comprising A. aragua, A. australis, A. isthtrees. According to the molecular phylogeny, the Cemica, A. merida, A. muelleri, A. petalocephala, A. salti, cropia-inhabiting Azteca do not form a monophyletic and A. xanthocroa share mottled orange to pure orange clade, indicating multiple independent colonization or queens, large subpyramidal petioles, densely pilose tibae, abandonment of Cecropia trees by the Azteca. © 1996 and the formation of a central carton in the tree bole (Lon-Academic Press, Inc.
gino, 1991b). The two complexes also differ incolonization behavior (Yu and Davidson, submitted for publication). Species belonging to the alfari complex most likely INTRODUCTION evolved from an ancestor that nested in live stems. Alfari The Azteca ants (subfamily Dolichoderinae) comprise species exhibit strong habitat specialization but weak approximately 150 described species, all of which are host species specialization, exclusively colonizing linear found exclusively in the Neotropics (Wheeler, 1942; and contiguous riverine environments. Yu and Davidson Kempf, 1972; Shattuck, 1994). Because of their obligate hypothesized that the streamlined thoraces of the queens association with myrmecophytic trees, the Azteca have in this group are adaptive for entering small holes (such received considerable attention from evolutionary ecolas live stems) but render queens unable to fly the long ogists (Bequaert, 1922;Wheeler, 1942;Buckley, 1982; distances needed to colonize patchily distributed forest Davidson and Fisher, 1991; Yu and Davidson, submitgaps where many Cecropia species establish. In contrast, ted for publication). Many Azteca species have specialmuelleri complex species most likely evolved from an anized to live in particular plant lineages; 13 known specestor that nested in cartonnests. Membersof this species cies have evolved an obligate relationship with groupexhibit weak habitat specialization, colonizing both Cecropia trees (Urticaceae) (Ho ¨lldobler and Wilson, riverine and forest gap habitats, but strong host species 1990; Longino, 1991b). Despite the prominence of the specialization, preferring Cecropia species with high lev-Azteca-Cecropia interaction as the subject of much ecoels of pearl body production. The placement of carton logical research, evolutionary and ecological studies of nests on host trees producing some form of ant reward is a common behavior in many carton-nesting ants.