Does canopy mean nitrogen concentration explain variation in canopy light use efficiency across 14 contrasting forest sites?

The maximum light use efficiency (LUE = gross primary production (GPP)/absorbed photosynthetic photon flux density (aPPFD)) of plant canopies has been reported to vary spatially and some of this variation has previously been attributed to plant species differences. The canopy nitrogen concentration [N] can potentially explain some of this spatial variation. However, the current paradigm of the N-effect on photosynthesis is largely based on the relationship between photosynthetic capacity (A(max)) and [N], i.e., the effects of [N] on photosynthesis rates appear under high PPFD. A maximum LUE-[N] relationship, if it existed, would influence photosynthesis in the whole range of PPFD. We estimated maximum LUE for 14 eddy-covariance forest sites, examined its [N] dependency and investigated how the [N]-maximum LUE dependency could be incorporated into a GPP model. In the model, maximum LUE corresponds to LUE under optimal environmental conditions before light saturation takes place (the slope of GPP vs. PPFD under low PPFD). Maximum LUE was higher in deciduous/mixed than in coniferous sites, and correlated significantly with canopy mean [N]. Correlations between maximum LUE and canopy [N] existed regardless of daily PPFD, although we expected the correlation to disappear under low PPFD when LUE was also highest. Despite these correlations, including [N] in the model of GPP only marginally decreased the root mean squared error. Our results suggest that maximum LUE correlates linearly with canopy [N], but that a larger body of data is required before we can include this relationship into a GPP model. Gross primary production will therefore positively correlate with [N] already at low PPFD, and not only at high PPFD as is suggested by the prevailing paradigm of leaf-level A(max)-[N] relationships. This finding has consequences for modelling GPP driven by temporal changes or spatial variation in canopy [N]