Four different habitats in a spring-fed forested wetland (Clear Springs Wetland, Panola County, Mississippi, USA) varying in hydrologic regime were examined for methane and carbon dioxide fluxes from soils over 15 and 9 months, respectively. There was an increasing gradient of CH 4 flux rates from an unflooded upper-elevation forest site to an occasionally flooded bottomland forest site to a shallow permanently flooded site, and then to a deeper-water permanently flooded site. Depending on the time of year, all sites were sources of methane but only at the upper-elevation forest site, when gravimetric soil moisture content fell below 54%, was atmospheric methane consumed. On average, summer CH 4 emission rates were higher than those in other seasons. A multiple regression model with soil temperature and soil redox potential as independent variables could explain 65% of the variation in CH 4 flux rates. In the flooded zone, variation in CH 4 flux rates was correlated with aboveground plant biomass and stem density of emergent vascular plants, and plant-mediated CH 4 transport depended on plant type. The efflux of CH 4 to plant biomass (Eff:B) ratio was generally lower in Hydrocotyle umbellata compared to Festuca obtusa. Compared to several other freshwater forested wetlands in the southeastern USA, this spring-fed forested wetland ecosystem was a strong source of atmospheric CH 4 , likely due to a long hydroperiod and high soil organic matter content. Carbon dioxide fluxes show a reverse spatial pattern than CH 4 fluxes with highest CO 2 emissions in the non-flooded zone at all times of the year, indicating the dominance of aerobic soil respiration. A multiple regression model also revealed a strong dependency of CO 2 fluxes (r 2 = 0.73) on soil temperature and soil redox potential.

Forest canopy water storage (<i>S</i>), direct throughfall fraction (<i>p</i>) and mean evaporation rate to mean rainfall intensity ratio (<i>E</i>/<i>R</i>) vary between storms and seasonally. Typically, researchers only quantify the mean growing and dormant season values of <i>S</i>, <i>p</i> and <i>E</i>/<i>R</i> for deciduous forests, thereby ignoring seasonal changes <i>S</i>, <i>p</i> and <i>E</i>/<i>R</i> .Past researchers adapted the mean method, which is usually used to estimate <i>S</i>, <i>p</i> and <i>E</i>/<i>R</i> on an annual or seasonal basis, to estimate the same canopy variables on a per storm basis (individual storm (IS) method). The disadvantage of the IS method is that it requires more expensive equipment and the calculation of the canopy variables is more labor intensive relative to the mean method. The goal of this study was to explore the use of the IS method for northern hardwood forests and to determine whether estimates of <i>S</i>, <i>p</i> and <i>E</i>/<i>R</i> derived by the IS method produce more accurate estimates of rainfall interception loss (I_n) using the Gash model relative to estimates derived by the mean method. The IS method estimated that S increased from approximately 0.11 mm in the early spring to 1.2 mm in the summer and then declined to 0.24 mm after fall senescence. Direct throughfall decreased from 0.4 in the early spring to 0.11 in the summer, and then increased to 0.4 after leaf senescence. When measurement period estimates of <i>p</i>, <i>S</i> and <i>E</i>/<i>R</i> derived by the IS and mean methods were applied to the Gash model, the modeled estimates of I_n differed from the measured values by 14.0 mm and 1.3 mm, respectively. Therefore, because the mean method provided more accurate estimates of I_n, the extra effort and expense required by the IS method is not advantageous for studies in northern hardwood forests that only need to model annual or seasonal estimates of I_n

No abstract