Trapping and netting of small mammals on a land-bridge island (Leyte) and on an oceanic island (Negros) revealed similar patterns of elevational change in abundance and species richness. Fruit bats (Pteropodidae) reached their greatest local densities in agricultural areas, and were least abundant in montane mossy forest. The fruit bats found to be common in agri-cultural areas are widespread in Southeast Asia: in contrast, species that were most common in forested areas are Philippine endemics. Fruit bat abundance was greater on the oceanic island than on the land-bridge island. Trappable small mammals (families Soricidae and Muridae) showed no change in species richness with increasing elevation, but did show a gradual increase in overall abundance. Even though the oceanic island was depauperate of non-volant mammal species, it had the higher abundance of non-volant mammal individuals; however, this may have been due primarily to differences associated with the elevation of sampling sites. Murid rodents at the higher elevations on Leyte tend to be those that are members of the old endemic group of Philippine murids, and those at lower elevations tend to be members of more recently arrived groups. Most non-volant small mammals at all elevations on the oceanic Negros, and in agri-cultural areas on Leyte, are non-native species.

Traits characterizing those proteas adapted for polhnation by nonflying mammals include: bowlshaped heads bearing fleshy bracts, these borne on short, flexible peduncles, often at or near ground level (geoflorous) and hidden beneath dense overlying foliage (cryptic), and producing copious nectar (ca. 1.8 ml/head, standing crop); individual florets with wir>', yet flexible styles and a nectar-stigma distance of 10 mm; a distinctive yeastlike odor; nocturnal anthesis; sucrose-rich nectar with a high total carbohydrate content (ca. 36%) and a relatively low proportion of amino acids. Evidence of small-mammal visitation to protea flowering heads includes: the presence of pollen on the rostra (carried in a position to effect pollination when foraging for nectar); the transport of fluorescing powders to flowering heads both within and between plants; the accumulation of small-mammal feces in flowering heads, and the destruction of exciosure bags containing nectar-rich heads. The period of greatest small-mammal activity (1800 hr.) coincides with maximum flower opening. T maze experiments showed that small mammals, when given a choice between typically bird-pollinated proteas and those having characteristics of flowers pollinated by nonflying mammals, always foraged on the latter. That small mammals can effect pollination is indicated by their foraging behavior on flowering heads while in captivity, the morphological "fit" between individual florets and the rostra of small mammals, and by selective exciosure experiments that reduced seed set (50% and 95%) when small mammals were excluded and visitation was limited to insects (mostly honey bees). The nectar produced ' The study was supported in part by NSF grant (DEB 78-11624). We thank B.

Ant communities were surveyed along an elevational gradient in the Philippines extending from lowland dipterocarp forest (250 m elevation) to mossy forest (1750 m). Standardized pitfall trapping in arboreal and terrestrial microhabitats at seven sites yielded 51 species. Collecting by hand at five of the sites yielded 48 species. The two methods produced substantially different assemblages, with only 22 species (29%) taken in common. Only a fraction of the total ant community appeared to be sampled at most of the sites. Measures of species richness and relative abundance peaked at mid‐elevations and declined sharply with increasing elevation. Ants were extremely rare above 1500 m elevation. Arboreal ants were trapped much less frequently than terrestrial ants at all sites. Ant species that were abundant had broader elevational distributions than those that were less common, but most species were rare and occurred at only one or two sites. The elevational patterns for ants are largely the inverse of those documented for Philippine small mammals which reach their greatest diversity and abundance at high elevations where ants are rare. This suggests that the two groups may interact competitively. Some of the patterns observed or inferred from this study may apply to tropical ant communities in general, and are presented as series of testable hypotheses as a guide and stimulus for future research.