Simulating gross primary production across a chronosequence of coastal Douglas-fir forest stands with a production efficiency model

Eddy-covariance (EC) measurements in three different-aged Douglas-fir (Pseudotsuga menziesii [Mirb.] Franco var. menziesii) stands were used to calibrate a production efficiency model (PEM) and explore the sources of error in simulated annual gross primary production (Pg). Parameters were derived on a daily time scale, assessing absorbed photosynthetically active radiation (Qa), maximum gross photosynthetic efficiency (ɛg max), and functions of environmental stress. Despite similar climate, ɛg max varied between sites in correspondence with measurements of site index derived from forest inventory, suggesting that landscape variation of ɛg max is controlled mainly by non-climatic factors and ranges approximately between 1.28 and 4.42 g C MJ−1. Within stands, daily variation of Pg was most strongly controlled by decreasing ɛg with increasing Qa. We therefore devised a method of incorporating the nonlinear light response (NLR) that is apparent within stands into the model, while preserving the linearity in the relationship between annual Pg and Qa across stands that is assumed in conventional PEMs. The ability to match observed seasonal and inter-annual variability of Pg improved by taking into account antecedent effects of cumulative heat on plant development. An ecosystem-specific model (i.e., fitted collectively to all stands) explained 81, 95, and 97% of the monthly variation of Pg in regenerating, juvenile, and mature stands, respectively. The model was able to collectively explain 96% of variability of annual total Pg with a root mean squared error 130 g C m−2 yr−1, constituting 6% of the mean and 113% of the standard deviation at the mature site. The capacity to predict inter-annual variability of Pg was strongly limited by discrepancies that persisted for days-to-weeks at a time, which implies that poor model skill was caused, to uncertain degrees, by inadequate representation of acclimation to environmental stress, and discrepancies between measurement and footprint-weighted conditions.