Net ecosystem productivity of a boreal aspen forest during and following drought

Continuous eddy correlation measurements of CO₂ and water vapour have made above a boreal aspen forest in the Boreal Ecosystem Research and Monitoring Sites (BERMS) study area in Saskatchewan, Canada for 11 years (1994-2005, excluding 1995) including a three year long drought (2001-2003). We examined the interannual and seasonal variability of net ecosystem production (F_NEP), evapotranspiration (E), gross ecosystem photosynthesis (P) and ecosystem respiration (R),and their relationship to environmental factors to explain changes observed in CO₂ exchange at the onset of, during, and following the drought. The forest has been a weak to moderate carbon (C) sink over the entire period (F_NEP = 144 ± 102 g C m^-2 y^-1), including the peak drought years of 2002 and 2003. In the first year of drought (2001), the depletion of water near the soil surface likely reduced heterotrophic soil respiration while water remaining deep in the root zone maintained P above the pre-drought mean, resulting in above-average F_NEP. In 2002 and 2003, the forest remained a C sink even though P was much less than average because R was also much less than the average—a likely consequence of the influence of low soil water content in deep and shallow soil layers on both autotrophic and heterotrophic respiration. Low spring temperatures and the persistent low soil water content at depth through the spring of 2004 continued to depress annual P. In June 2004, the recharge of soil water in shallow soil layers allowed R to recover to seasonal average values before P, producing the lowest annual F_NEP of the 11-year record (close to zero g C m^-2 y^-1), a result that was mirrored by the lowest stem growth and LAI values of the 11-year record. In 2005, a warm wet year, both the annual values and seasonal variationsof F_NEP, P and R returned to those of pre-drought years; the partial recovery of LAI to pre-drought values suggests that aspen P was able to adapt to this restriction on C assimilation. Analysis showed that ecosystem P was slower to respond to the onset and end of drought conditions than ecosystem R, an effect largely attributable to the control of P by the water content deep in the root zone and the control of R largely by the near surface soil water content. Analysis showed that growing season average dry surface conductance, Priestley-Taylor alpha and light use efficiency reached their lowest values in 2003 and became similar to pre-drought years in 2004-2005. Water use efficiency (WUE = P/E) was highest in 2003 and remained above average in 2004 and 2005. Analysis of the canopy intercellular CO₂ concentration suggested that, with few exceptions, P was increasingly supply limited by decreased stomatal conductance following the onset of drought. We found a decreasing trend in E over the years excluding the drought years, which was accompanied by an increasing trend in WUE. Non-drought a values remained fairly constant; suggesting that the reduction in E is most likely resulted from a reduction in available energy flux. At the ecosystem scale, our results suggest that the above-average gains made in C sequestration in the first year of the drought were significantly offset by below-average stand F_NEP in the final two years of the drought, and in the year following the drought.