-The community structure of cavities-nesting solitary bees species was systematically investigated over two years, from May/97 to April/99, in a fragment of tropical sand dune, Salvador, Bahia (12º56'S; 38º21'W), Northeast of Brazil. Trap nests were placed randomly in the area, in blocks with 16 nests, with variable diameter at a height of 1.5 m. Traps were inspected each 15 days. Seven bee species established 62 nests. Two species were predominant in the area, Centris (Hemisiella) tarsata Smith (58% of the total of established nests) followed by Euplusia musitans Fabricius (31%). In general, bees' abundance showed seasonal fluctuations. The total frequencies of nesting varied between the two years of sampling. The two predominant species presented different seasonal patterns. There were always trap nests available in the field for the bees. Just 14% of the available nests were occupied.

ECOLOGY OF XYLOCOPA (NEOXYLOCOPA) CEARENSIS (HYMENOPTERA, ANTHOPHORIDAE) IN ABAETÉ SAND DUNES, SALVADOR, BAHIA. The study was carried out in a 8.2 ha area in an environmental protection area of tropical sea coastal sand dune with "restinga" vegetation in Salvador, Bahia, northeastern Brazil. A total of 1760 bees of Xylocopa (Neoxylocopa) cearensis Ducke, 1910 were netted during the whole year on flowers of 43 plant species belonging to 26 botanic families. The majority of the individuals (79%) concentrated their foraging activity in five plant species. Individuals foraged all day long being the greatest activity between 8 h and 14 h. Similar proportions of young and old bees were sampled over the year. The density of substrates used for nesting was 4.56/ha. In total, 94% of the nests were found in branches of Agaristha revoluta (Spr.) DC. (Ericaceae). The great occurrence (68.7%) of old perforations indicates that the nests were used twice or more times by bees.

We investigated changes in the communities of trap-nesting Hymenoptera in forests in relation to forest loss on a landscape scale and understory conditions on a local habitat scale. Two specific questions were addressed. (1) Do the communities change with degrees of forest loss? (2) Do the communities change with varying local environmental conditions of understory habitats? The study was made in a landscape characterized by distributed forest patches within intensively managed agricultural surroundings. We deployed trap-nests at eight randomly selected sites in forests in summer. To quantify forest loss, the amount of forest coverage was calculated using GIS. To indicate local habitat conditions, the species richness of understory flowering plants was used. All together, 12 species of wasps and no bees were captured. Regression analyses showed that both abundance and species richness of the wasps were not significantly related to forest coverage. However, abundance of trap-nesting wasps was significantly related to species richness of understory plants, but species richness of the wasps was not significantly related to the plants. These results suggest that communities of trap-nesting wasps in forests are influenced more by the local habitat conditions than by forest loss.