Understanding carbon cycling in tropical forest is more important than ever due to the expansion of secondary tropical forest and growing efforts to make global carbon models more accurate. Few studies have measured productivity in dry forests and especially secondary tropical dry forests (TDF) regenerating following agriculture. This study measures aboveground net primary productivity (ANPP) in secondary TDF in Guanacaste, Costa Rica, to determine how productivity varies across gradients of stand age, forest composition and season. This area receives an average of 1600 mm of precipitation per year and has a 5-6 month dry season, which is on the wet end of the TDF spectrum. We set up 18, 20 × 50 meter plots in sites ranging in recovery time from 7 to 60 years. In each plot we installed four 0.25 m² litter traps that are emptied monthly, and band dendrometers on all trees above ten centimeters (~900 trees) that are measured seasonally. Aboveground biomass increment was calculated by subtracting initial biomass from final biomass. Biomass was estimated using an allometric equation based on trunk diameter and wood density. Litterfall and biomass increment were summed to estimate ANPP.  
Results/Conclusions

Our results from the first year, which was an unusually dry year, show that 85% of productivity occurred in the wet season, while 15% of productivity occurred in the dry season. Sites with more than 20 years of recovery time had 22% of their productivity in the dry season which is significantly higher than the 8% found in younger sites (P=0.005). Among sites ANPP ranged from 4-14 Mg/ha•yr. Annual ANPP is poorly correlated with time since recovery, with a mean of 7.4 Mg/ha•yr for younger forests and 9.6 of older sites. Similar values for ANPP have been reported for forests drier than our study site. This may be a result of unusually low precipitation of 2009. Litterfall was greater than biomass increment at most sites with an average of 60% of ANPP from litter. However, in three sites biomass increment exceeded litterfall. These data show that patterns of seasonal productivity in tropical dry forest change with recovery from disturbance. As forests regenerate, productivity increases and a greater fraction of annual productivity occurs in the dry season. We speculate that this result may be due to a change in functional diversity between younger and older sites.