Litter-fall patterns were examined for 3 years within a floodplain forest on the upper coastal plain of South Carolina. Average annual leaf fall and total litter fall ranged from 415-538 g-m-2.year-1 and 570-667 g-m-2-year-1, respectively, across the floodplain habitat. These values were similar to those of other alluvial habitats and exceeded the litter-fall totals reported for nonwetland, warm-temperate, deciduous forests. Leaf fall increased across the floodplain from stream bank to upland forest. Leaf fall varied from 73-84%o of total litter fall which exceeds the worldwide average of 70%o. Woody litter fall decreased across the floodplain whereas reproductive inputs remained similar. Nutrient return (Ca> N > K> Mg) in leaf fall was relatively high, particularly for the less mobile elements (Ca, Mg). Calcium and Mg inputs in leaf fall increased from bank to upland terrace. Leaf fall inputs of N and K were reduced on the sandy upland where soil leaching potential is greatest. Translocation may minimize nutrient return of N and K on the upland, whereas fluvial subsidies of these nutrients would compensate for potential losses within the alluvial floodplain habitat. Our results support the contention that leaf fall and nutrient return in litter are higher in forested wetlands receiving fluvial subsidies than in most nonwetland, deciduous forests. INTRODUCTION The available information on litter-fall inputs to terrestrial ecosystems indicates geographic patterns which are correlated with particular environmental factors (Bray and Gorham, 1964; Jordan, 1971; Jordan and Murphy, 1978; Meentemeyer et al., 1982). Litter production rates generally increase latitudinally from Arctic to tropics in conjunction with total light availability during the growing season (Jordan and Murphy, 1978) and total actual evapotranspiration (Meentemeyer et al., 1982). Leaf fall accounts for approximately 70% of total litter fall in different ecosystems worldwide (Meentemeyer et al., 1982). These overall patterns have usually been derived from a single year's estimates and therefore preclude considerations of annual variations in litter-fall rates and composition (Sykes and Bunce, 1970). Litter-fall patterns in wetland forests are usually not included in geographical determinations. However, recent evidence indicates that litter production in forested wetlands often exceeds corresponding values for nonwetland habitats (Bell et al., 1978; Brinson et al., 1980; Gomez and Day, 1982). Many wetlands receive fluvial inputs of water, oxygen and nutrients which can increase nutrient cycling and productivity (Brinson et al., 1980). Primary productivity and litter production are often higher in flowingwater than still-water wetlands which lack these fluvial inputs (Schlesinger, 1978; Brinson et al., 1980; Gomez and Day, 1982; but see Brown and Peterson, 1983). The extent and timing of flooding may thus affect wetland litter production and the import-export dynamics (Post and de la Cruz, 1977) of allochthonous litter sources within watershed units. The present study was conducted to determine litter-fall patterns in a floodplain forest surrounding a blackwater stream on the coastal plain of South Carolina. We were interested in assessing spatial and temporal changes in litter-fall components across the floodplain from stream bank to adjacent upland terrace forest. The study was continued for several years to determine the annual variations in litter-fall production at different points along the flood frequency gradient. METHODS The study was conducted in a floodplain along Lower Three Runs Creek (LTRC)