Abstract. 1. The division of macrohabitats, seasons and microhabitats among mosquitoes which inhabit water-filled treeholes was investigated in a wet/dry seasonal environment near the coast of Kenya. Twenty-two species from four genera (Aedes fourteen, Culex three, Eretmapodites three, Toxorhynchites two) were identified from treeholes and bamboo traps. These included five which were more abundant in other microhabitats: leaf axils, fruit husks, or abandoned snail shells.
2. Trapping in different vegetation zones or heights in forests revealed that all seventeen species examined showed opposition preferences for one or more macrohabitats. Segregation of related species by macrohabitat was particularly apparent among members of the subgenus Stegomyia of Aedes which accounted for 57% of individuals identified from one locality.
3. Seasonal overlap was low between Aedes, which tracked rainfall patterns, and Culex and Eretmapodites. Aedes bromeliae differed from other Aedes found in treeholes in that maximal larval numbers did not follow heavy rainfall.
4. Negative associations between Aedes fulgens and two congeners which co-occur in the same forest and season were attributed to preferences for tree-holes of different sizes. Positive associations between Aedes aegypti and several other Aedes were reduced when root buttresses, a subclass of larger treeholes containing more species, were removed.
5. Separation of species by growth stages may be accomplished by two schemes of egg hatching which desynchronize larval cohorts in the same tree-hole: the early pattern of Stegomyia whereby 50–95% of all eggs hatch at first immersion, and the delayed pattern of other subgenera (Diceromyia, Finlaya, Pseudarmigeres) in which less than one-third of all eggs hatch at first immersion.
6. The number of wet treeholes changed seasonally. During one year five or six periods occurred when 50% or more of the censused holes contained sufficient water for larval survival. The number of individuals and species of mosquitoes in treeholes were positively correlated with treehole volume. In a comparison of treeholes and their mosquito fauna from two forests, treehole volume and mean number of species per hole differed significantly, but mosquito density did not.
7. Ecotonal trapping sites of mixed grasses, trees, and scrub yielded the highest mean numbers of species per sample, and traps placed outside a village, the lowest. Specific dominance was significantly less in the ecotone than elsewhere. The predatory larvae of Toxorhynchites brevipalpis in ecotonal and forested traps were not correlated with species richness.