Background/Question/Methods Climate change may result in a wide range of potential impacts to forest ecosystems, including altered precipitation regimes, more intense storm events, and an increase in invasive insects.  Understanding mechanisms underlying recovery following disturbance is essential for the accurate prediction of carbon (C) dynamics of forest ecosystems.  However, few studies have quantified net CO₂ exchange (NEE) pre- and immediately post disturbance, making it difficult to generalize about processes controlling the recovery of NEE.  We used eddy covariance and biometric measurements to track changes in net CO₂ exchange and C pools before and after complete defoliation of Oak and Pine-dominated stands by Gypsy moth (Lymantria dispar L.) in the New Jersey Pine Barrens.  We also measured soil CO₂ flux, and estimated ecosystem respiration (R_eco) and gross ecosystem productivity (GEP) from eddy flux measurements. 
Results/Conclusions Following complete defoliation in early summer of 2007, maximum seasonal leaf area (LAI) following a second leaf flush was 2.2 ± 0.3.  In 2008 and 2009, LAI had recovered to 3.4 ± 0.7 and 4.5 ± 0.8, respectively.  Over the five years measured, annual GEP was a linear function of leaf area (r² = 0.932, n = 5), increasing from 726 g C m^-2 yr^-1 to 1532 g C m^-2 yr^-1 from 2007 to 2009.   Complete defoliation of the Oak-Pine stand in 2007 resulted in considerable mortality, and by late 2009, 15% of the individuals and 25% of oak basal area were dead.  Annual R_eco was depressed in 2007 and 2008 relative to pre-defoliation values, totaling 972 and 1066 g C m^-2 yr^-1, respectively.  By 2009, R_eco totaled 1523 g C m^-2 yr^-1, 14% greater than pre-defoliation values, reflecting the abundance of detrital material on the forest floor and belowground mortality.  As a result, the pattern of annual NEE was complex; values were +246, -77, and -9 g C m^-2 yr^-1 in 2007, 2008 and 2009, respectively.  Our data indicate that enhanced ecosystem respiration due to mortality delayed recovery from this disturbance.  Quantifying these complex relationships assists in the development of more accurate models of forest C dynamics during disturbance and subsequent recovery.