Net ecosystem production in a Spanish black pine forest after a low burn-severity fire: Significance of different modelling approaches for estimating gross primary production

Abstract Many studies have examined post-fire net ecosystem production (NEP), however, the CO 2 dynamics in Mediterranean mountain ecosystems after low burn-severity fires are poorly understood. To fill this gap, we used biometric and flux chamber-based methods to assess the NEP in a mature Spanish black pine ( Pinus nigra Arn. ssp. salzmannii ) forest at an unburnt (UB) and a low burn-severity (LS) sites during the early (1.5–4.5 years) post-fire stages. In this study, we also compared the significance of two different laborious modelling approaches for estimating the gross primary production (GPP) in order to deduce the NEP estimates: GPP C and GPP M . The former is based on a carbon (C)-mass balance approach which is calculated as the sum of the aboveground net primary production, total belowground C flux, and aboveground autotrophic respiration. The latter is based on a whole-canopy photosynthesis-modelling approach and is obtained by combining an environmental-dependent non-rectangular hyperbolic light-response model applied to different pine needle age-cohorts and coupled to a two-leaf scaling-up strategy. Our results indicate that both sites were a consistent C sink between 2011 and 2013, with a mean respective annual NEP C of 2.43 and 2.09 Mg C ha −1  year −1 at the UB and LS sites, and a mean annual NEP M of 2.04 and 1.82 Mg C ha −1  year −1 at the UB and LS sites, respectively; all these figures are comparable in magnitude to other European forests. The difference between the NEP C and NEP M estimates within a given year varied by 6–25%, which, based on the fact that we used a single-ecosystem respiration (R eco ) data set, indicates a strong correlation between the GPP C and GPP M estimates. Our findings indicated that the low burn-severity fire did not substantially alter the annual GPP, R eco , or NEP fluxes in this particular disturbed-forest ecosystem. Moreover, this work provides evidence for the suitability of either modelling approach to effectively simulate the GPP in future stand-scale C-cycling studies. Thus, our results are relevant not in that they directly evaluate and compare the approaches themselves, but also because they identify method-specific uncertainties that deserve exhaustive evaluation, which should be an important component in the future work of environmental researchers, especially those interested in ecological modelling.