Response of carbon fluxes to drought in a coastal plain loblolly pine forest

Full accounting of ecosystem carbon (C) pools and fluxes in coastal plain ecosystems remains less studied compared with upland systems, even though the C stocks in these systems may be up to an order of magnitude higher, making them a potentially important component in regional C cycle. Here, we report C pools and CO₂ exchange rates during three hydrologically contrasting years (i.e. 2005–2007) in a coastal plain loblolly pine plantation in North Carolina, USA. The daily temperatures were similar among the study years and to the long‐term (1971–2000) average, whereas the amount and timing of precipitation differed significantly. Precipitation was the largest in 2005 (147 mm above normal), intermediate in 2006 (48 mm below) and lowest in 2007 (486 mm below normal). The forest was a strong C sink during all years, sequestering 361 ± 67 (2005), 835 ± 55 (2006) and 724 ± 55 (2007) g C m^?2 yr^?1 according to eddy covariance measurements of net ecosystem CO₂ exchange (NEE). The interannual differences in NEE were traced to drought‐induced declines in canopy and whole tree hydraulic conductances, which declined with growing precipitation deficit and decreasing soil volumetric water content (VWC). In contrast, the interannual differences were small in gross ecosystem productivity (GEP) and ecosystem respiration (ER), both seemingly insensitive to drought. However, the drought sensitivity of GEP was masked by higher leaf area index and higher photosynthetically active radiation during the dry year. Normalizing GEP by these factors enhanced interannual differences, but there were no signs of suppressed GEP at low VWC during any given year. Although ER was very consistent across the 3 years, and not suppressed by low VWC, the total respiratory cost as a fraction of net primary production increased with annual precipitation and the contribution of heterotrophic respiration (R_h) was significantly higher during the wettest year, exceeding new litter inputs by 58%. Although the difference was smaller during the other 2 years (R_h : litterfall ratio was 1.05 in 2006 and 1.10 in 2007), the soils lost about 109 g C m^?2 yr^?1, outlining their potential vulnerability to decomposition, and pointing to potential management considerations to protect existing soil C stocks.