Modelling vegetation primary production during HAPEX-Sahel using production efficiency and canopy conductance model formulations

Abstract

Canopy scale vegetation net production of four canopies of the HAPEX-Sahel west central supersite was modelled using two contrasting models of photosynthesis. The vegetation canopies included the Guiera senegalensis shrubs of the shrub fallow site, the herb layers of the shrub fallow and grass fallow sites and the millet crop. The two models were based on (i) the production efficiency approach, which assumes that light is the primary limiting factor, and (ii) the CO 2 supply function approach, which assumes that the rate of CO 2 influx to the leaves is the main limiting factor. The models were generalised to apply across vegetation types and through time by allowing important parameters to vary according to the proportion of C 3 plants in the canopy, by addition of maintenance and growth respiration terms and by addition of an empirical term related to leaf age. The models were driven using ground measurements of the biophysical variables (light interception, leaf area index, stomatal and canopy conductances) summed or averaged to the ten-day temporal scale and fitted to harvest estimates of ten-day net production. Each term of the overall models was tested for statistical significance during the model fitting procedure. Despite their opposing assumptions, both models were able to explain a large proportion (>80%) of the total variance in ten-day net production in the four canopies during the growing season. This was attributed in part to the fact that both models are dependent on the assessment of canopy amount (represented by light interception and LAI, respectively) which are correlated, and because the functions describing the effect of environmental variables (soil moisture, vapour pressure deficit) on photosynthesis were also correlated. Inclusion of the maintenance respiration term was statistically significant for both modelling approaches. Leaf age was significant in most cases but this may be related to the covariance of the day of the year with environmental parameters. In most cases significant differences in the average values of maximum PAR conversion efficiency ( ϵ ∗ ) and CO 2 concentration gradient (Δ) were found between C 3 and C 4 species. However, the values of ϵ ∗ and Δ do not agree with expected or literature estimates of these values. This suggests that the assumptions of neither model were entirely correct and implies that a calibration step is required for their application.