Understanding the controls on floodplain carbon (C) cycling is important for assessing greenhouse gas emissions and the potential for C sequestration in river‐floodplain ecosystems. We hypothesized that greater hydrologic connectivity would increase C inputs to floodplains that would not only stimulate soil C gas emissions but also sequester more C in soils. In an urban Piedmont river (151 km²watershed) with a floodplain that is dry most of the year, we quantified soil CO₂, CH₄, and N₂O net emissions along gradients of floodplain hydrologic connectivity, identified controls on soil aerobic and anaerobic respiration, and developed a floodplain soil C budget. Sites were chosen along a longitudinal river gradient and across lateral floodplain geomorphic units (levee, backswamp, and toe slope). CO₂emissions decreased downstream in backswamps and toe slopes and were high on the levees. CH₄and N₂O fluxes were near zero; however, CH₄emissions were highest in the backswamp. Annual CO₂emissions correlated negatively with soil water‐filled pore space and positively with variables related to drier, coarser soil. Conversely, annual CH₄emissions had the opposite pattern of CO₂. Spatial variation in aerobic and anaerobic respiration was thus controlled by oxygen availability but was not related to C inputs from sedimentation or vegetation. The annual mean soil CO₂emission rate was 1091 g C m⁻² yr⁻¹, the net sedimentation rate was 111 g C m⁻² yr⁻¹, and the vegetation production rate was 240 g C m⁻² yr⁻¹, with a soil C balance (loss) of −338 g C m⁻² yr⁻¹. This floodplain is losing C likely due to long‐term drying from watershed urbanization.

[1] Watershed Best Management Practices (BMPs) are often designed to reduce loading from particle-borne contaminants, but the temporal lag between BMP implementation and improvement in receiving water quality is difficult to assess because particles are only moved downstream episodically, resting for long periods in storage between transport events. A theory is developed that describes the downstream movement of suspended sediment particles accounting for the time particles spend in storage given sediment budget data (by grain size fraction) and information on particle transit times through storage reservoirs. The theory is used to define a suspended sediment transport length scale that describes how far particles are carried during transport events, and to estimate a downstream particle velocity that includes time spent in storage. At 5 upland watersheds of the mid-Atlantic region, transport length scales for silt-clay range from 4 to 60 km, while those for sand range from 0.4 to 113 km. Mean sediment velocities for silt-clay range from 0.0072 km/yr to 0.12 km/yr, while those for sand range from 0.0008 km/yr to 0.20 km/yr, 4–6 orders of magnitude slower than the velocity of water in the channel. These results suggest lag times of 100–1000 years between BMP implementation and effectiveness in receiving waters such as the Chesapeake Bay (where BMPs are located upstream of the characteristic transport length scale). Many particles likely travel much faster than these average values, so further research is needed to determine the complete distribution of suspended sediment velocities in real watersheds.