Interseasonal comparison of CO2 concentrations, isotopic composition, and carbon dynamics in an Amazonian rainforest (French Guiana)

Canopy CO₂ concentrations in a tropical rainforest in French Guiana were measured continuously for 5 days during the 1994 dry season and the 1995 wet season. Carbon dioxide concentrations (CO₂]) throughout the canopy (0.02–38 m) showed a distinct daily pattern, were well-stratified and decreased with increasing height into the canopy. During both seasons, daytime CO₂] in the upper and middle canopy decreased on average 7–10 μmol mol⁻¹ below tropospheric baseline values measured at Barbados. Within the main part of the canopy (≥ 0.7 m), CO₂] did not differ between the wet and dry seasons. In contrast, CO₂] below 0.7 m were generally higher during the dry season, resulting in larger CO₂] gradients. Supporting this observation, soil CO₂ efflux was on average higher during the dry season than during the wet season, either due to diffusive limitations and/or to oxygen deficiency of root and microbial respiration. Soil respiration rates decreased by 40% after strong rain events, resulting in a rapid decrease in canopy CO₂] immediately above the forest floor of about 50␣μmol mol⁻¹. Temporal and spatial variations in CO₂]_canopy were reflected in changes of δ¹³C_canopy and δ¹⁸O_canopy values. Tight relationships were observed between δ¹³C and δ¹⁸O of canopy CO₂ during both seasons (r ² > 0.86). The most depleted δ¹³C_canopy and δ¹⁸O_canopy values were measured immediately above the forest floor (δ¹³C = −16.4‰; δ¹⁸O = 39.1‰ SMOW). Gradients in the isotope ratios of CO₂ between the top of the canopy and the forest floor ranged between 2.0‰ and 6.3‰ for δ¹³C, and between 1.0‰ and 3.5‰ for δ¹⁸O. The δ¹³C_leaf and calculated c _i/c _a of foliage at three different positions were similar for the dry and wet seasons indicating that the canopy maintained a constant ratio of photosynthesis to stomatal conductance. About 20% of the differences in δ¹³C_leaf within the canopy was accounted for by source air effects, the remaining 80% must be due to changes in c _i/c _a. Plotting 1/CO₂] vs. the corresponding δ¹³C ratios resulted in very tight, linear relationships (r ² = 0.99), with no significant differences between the two seasons, suggesting negligible seasonal variability in turbulent mixing relative to ecosystem gas exchange. The intercepts of these relationships that should be indicative of the δ¹³C of respired sources were close to the measured δ¹³C of soil respired CO₂ and to the δ¹³C of litter and soil organic matter. Estimates of carbon isotope discrimination of the entire ecosystem, Δ_e, were calculated as 20.3‰ during the dry season and as 20.5‰ during the wet season. Received: 3 March 1996 / Accepted: 19 October 1996