Numerous ant taxa naturally inhabit stems of live and dead Guadua bamboo (Bambusoidea, Poaceae) in western Amazonia. In an experiment at the onset of the wet season in Peru's Manu National Park, we augmented potential nest sites in stems of live bamboo, dead bamboo and dead cana brava (Gynerium sagittatum, another woody grass) at five stations within each of ten bamboo patches and ten control areas outside those patches. Each experimental stem possessed three vacant and available internodes, pre-drilled with, respectively, large, small and linear holes, mimicking the range of forms of surveyed natural entrances. After 24 days, approximately 13% of 798 available internodes had been colonized, the majority by fragments of existing colonies. Ignoring entrance type, which did not affect colonization for any species or species group, and censoring non-independent internodes of the same stem, we used individual stems as independent sample units in other tests. One specialist in live bamboo (Camponotus longipilis), and a likely specialist in dead bamboo (Camponotus depressus), were identified based on overrepresentation in bamboo habitat and disproportionate occurrence in live or dead bamboo stems. A third species, Camponotus (Pseudocolobopsis sp.) was more abundant in bamboo areas but colonized both dead bamboo and dead cana. Relatively high abundance of standing dead stems in Guadua forests may account for the presence of a dead stem specialist. The experiment missed detecting specialization in one live culm specialist (Camponotus mirabilis), likely due to its failure to simulate conditions required for the species' unique modes of colony establishment and spread into new culms. Most opportunistic stem nesters colonized dead bamboo at significantly greater rates than dead cana, but were either equally well represented in bamboo and control areas, or underrepresented in bamboo habitat. Given low colonization rates overall, underrepresentation in bamboo cannot be attributed to competition from bamboo specialists for nesting space. Rather, it may be due to combined effects of seasonal flooding of bamboo habitat, and greater importance of food limitation, relative to nest site limitation, in that habitat.