Leaf respiration in two tropical rainforests: constraints on physiology by phosphorus, nitrogen and temperature

1. Leaf respiration is a major component of the terrestrial carbon cycle, but is poorly quantified for tropical forests. We measured dark respiration, R, and nutrient concentration (nitrogen, N and phosphorus, P) of leaves within two forest canopies: in Reserva Jaru, south-west Brazil; and Mbalmayo Reserve, central Cameroon. The data, expressed on a leaf area basis (R(a), N(a), P(a)) and a leaf mass basis (R(m), N(m), P(m)), were used to quantify the temperature sensitivity of R and to test the hypothesis that leaf metabolism is more strongly constrained by phosphorus than by nitrogen in these lowland rainforests. 2. Leaf respiration rate (R(a), at 25 degreesC) at Jaru was nearly half that at Mbalmayo (the range in R(a) from near the ground to the upper canopy was 0.11-0.78 mu mol m(-2) s(-1) at Jaru versus 0.22-1.19 mu mol m(-2) s(-1) at Mbalmayo), and the mean Q(10) for respiration at each site was 2.3 +/- 0.9 (1 SD) and 2.0 +/- 0.5 (1 SD), respectively. There were significant differences (P < 0.01) between sites in leaf phosphorus concentration, but not in leaf nitrogen concentration: P(m) was very low at Jaru (0.2-0.7 mg g(-1)) but higher at Mbalmayo (0.5-2.4 mg g(-1)), whilst N(m) was similar at both forests (10-45 mg g(-1)). 3. R(m) was not significantly associated with canopy position or specific leaf area (SLA, m(2) g(-1)) in either forest, but a significant relationship between SLA and N(m) was found for both sites (P < 0.05), consistent with existing data. At Jaru, R(m) was strongly related to P(m) (P < 0.001) and less strongly related to N(m) (P < 0.05), but at Mbalmayo, R(m) was not significantly related to either P(m) or N(m). 4. R(a) was linearly related to N(a) and P(a) at both sites (P < 0.01), principally because of changes in leaf mass per area (LMA, g m(-2)) associated with canopy position. At Mbalmayo, LMA explained 70% of the variation in R(a), but only 20% at Jaru. For Jaru, the strongest relationship with R(a) was obtained by combining LMA with P(m) in a multiple regression (r(2) = 0.53); further inclusion of N(m) did not improve the regression. At Mbalmayo neither N(m) or P(m) improved the regression of R(a) on LMA. 5. These results indicate a strong influence of LMA on the relative rates of R(a) within the vertical gradient of each canopy. They also suggest that at Jaru P(m) constrains respiration more strongly than N(m), and further, that the very low P(m) at Jaru may explain the lower absolute values of respiration there relative to Mbalmayo, where P(m) was higher. The leaves at both sites are typical of lowland tropical rainforests in not having particularly low N(m), and consistent with this, N(m) was a weaker predictor of respiration than P(m) or LMA.