19 Reduced Impact Logging and Its Effects on the Pollination of Amazonian Trees Márcia Motta Maués Embrapa Amazônia Oriental, Lab. Entomologia, CP 48, Belém, PA, 66.017-970 - Brasil (marcia@cpatu.embrapa.br) Paulo Eugênio Oliveira Deptº Biociências, Universidade Federal de Uberlândia CP 593, Uberlândia, MG, 38.400-902 - Brasil Milton Kanashiro Embrapa Amazônia Oriental, Lab. Entomologia, CP 48, Belém, PA, 66.017-970 - Brasil In the last three decades, the Brazilian Amazon (4.9 3 10 –1 km 2 ) has lost approximately 17% of its forest cover due to the expansion of logging activities, cattle ranching, and agricultural systems [1]. Habitat loss and forest fragmentation affect pollen flow because they reduce the number of pollen donors and possibly the amount of compatible pollen, which may lead to low fruit set and genetic drift [2, 3]. These processes may also result in the decline of pollinator populations, threatening the connectivity of remnant trees in fragments [4]. We studied the effect of reduced impact logging (RIL) policies on the pollination efficiency and pollinator groups of Jacaranda copaia (Aubl.) D. Don (Bignoniaceae), Dipteryx odorata (Aubl.) Willd. (Leguminosae–Papilionoideae), Carapa guianensis Aubl. (Meliaceae), Symphonia globulifera L.f. (Clusiaceae), and Bagassa guianensis Aubl. (Moraceae) (Figure 1). These species are hardwood trees that (except J. copaia) are used in the forestry industry for timber and/or plywood. Our research was conducted in the Western Amazon (Tapajós National Forest) under two systems of forest management: non-logged forest (NLF) and logged forest (LF) subject to RIL. For each study species, we determined the pollen deposition rate (PDR = average number of pollen grains deposited per sample stigma) and percentage of pistil fertilization (PPF = % of sample pistils fertilized). These data were compared between logging treatments using ANOVA. Pollinator activity (species, behavior, and visitation rate) was monitored during the main flowering period of J. copaia, D. odorata, and S. globulifera in six-hour schedules, totaling at least 48 hours per species. The pollinators were grouped into nine groups: 1) Small- to middle-sized bees (> 2.5 cm); 2) Large bees (≥ 2.5 cm); 3) Wasps; 4) Dipterans; 5) Lepidoterans; 6) Beetles; 7) Passeriformes; 8) Hummingbirds; 9) Other birds. Pollen deposition rate was different between the two logging treatments when all the species were analyzed together. Separately, the PDR of D. odorata was significantly higher in the logged forest (F 1, 11 = 4.96; p = 0.05), whereas S. globulifera showed the opposite response with significantly lower PDR at the logged site (F 1, 13 = 4.59; p = 0.05). For the remaining species, no significant difference was detected (Figure 2). Concerning PPF, over all species it was significantly lower in the LF than the NLF (F 1, 4 = 5.74; p = 0.018), but single species analyses were not significant. In terms of pollinator activity, J. copaia was visited by small bees (e.g., Centris, Euglossa), D. odorata by large bees (e.g., Epicharis, Bombus, and Eulaema), and S. globulifera mainly by perching birds. Also, C. guianensis was visited by microlepidopterans and stingless bees, while B. guianensis presented strong evidence of wind pollination. J. copaia, S. globulifera, and especially D. odorata, showed qualitative differences in the frequency and composition of pollinator groups between logging treatments, which may affect pollination efficiency. In LF, secondary pollinators