Parameterization and testing of a coupled photosynthesis-stomatal conductance model for boreal trees.

Abstract

A coupled photosynthesis-stomatal conductance model was parameterized and tested with branches of black spruce (Picea mariana (Mill.) B.S.P.) and jack pine (Pinus banksiana Lamb.) trees growing in the Northern Study Area of the Boreal Ecosystem-Atmosphere Study (BOREAS) in Manitoba, Canada. Branch samples containing foliage of all age-classes were harvested from a lowland old black spruce (OBS) and an old jack pine (OJP) stand and the responses of photosynthesis (A(n)) and stomatal conductance (g(s)) to temperature, CO(2), light, and leaf-to-air vapor pressure difference (VPD) were determined under controlled laboratory conditions at the beginning, middle, and end of the growing season (Intensive Field Campaigns (IFC) 1, 2, and 3, respectively). The parameterized model was then tested against in situ field gas-exchange measurements in a young jack pine (YJP) and an upland black spruce (UBS) stand as well as in the OBS and OJP stands. Parameterization showed that Rubisco capacity (V(max)), apparent quantum yield (alpha') and Q(10) for sink limitation were the most crucial parameters for the photosynthesis sub-model and that V(max) varied among different measurement series in the laboratory. Verification of the model against the data used to parameterize it yielded correlation coefficients (r) of 0.97 and 0.93 for black spruce and jack pine, respectively, when IFC-specific parameters were used, and 0.77 and 0.87 when IFC-2 parameters were applied to all IFCs. For both measured and modeled g(s), the stomatal conductance sub-model, which linearly relates g(s) to (A(n)h(s))/c(s) (where h(s) and c(s) are relative humidity and CO(2) mole fraction at the leaf surface, respectively), had significantly steeper slopes and higher r values when only the VPD response data were used for parameterization than when all of the response data were used for parameterization. Testing the photosynthesis sub-model against upper canopy field data yielded poor results when laboratory estimates of V(max) were used. Use of the mean V(max) estimated for all upper canopy branches measured on a given day improved model performance for jack pine (from a nonsignificant correlation between measured and modeled A(n) to r = 0.45), but not for black spruce (r = 0.45 for both cases). However, when V(max) was estimated for each branch sample individually, the model accurately predicted the 23 to 137% diurnal variation in A(n) for all stands for both the upper and lower canopy. This was true both when all of the other parameters were IFC-specific (r = 0.93 and 0.92 for black spruce and jack pine, respectively) and when only mid-growing season (IFC-2) values were used (r = 0.92 for both species). Branch-specific V(max) estimates also permitted accurate prediction of field g(s) (r = 0.75 and 0.89 for black spruce and jack pine, respectively), although parameterization with all of the response data overestimated g(s) in the field, whereas parameterization with only the VPD response data provided unbiased predictions. Thus, after parameterization with the laboratory data, accurately modeling the range of A(n) and g(s) encountered in the field for both black spruce and jack pine was reduced to a single unknown parameter, V(max).