The influence of soils on heterotrophic respiration exerts a strong control on net ecosystem productivity in seasonally dry Amazonian forests

Net ecosystem productivity of carbon (NEP) in seasonally dry forests of the Amazon varies greatly between sites with similar precipitation patterns. Correctly modeling the NEP seasonality with terrestrial ecosystem models has proven difficult. Previous modelling studies have mostly advocated for incorporating processes that act to reduce water stress on gross primary productivity (GPP) during the dry season, such as deep soils and roots, plant-mediated hydraulic redistribution of soil moisture, and increased dry season leaf litter generation which reduces leaf age and thus increases photosynthetic capacity. Recent observations, however, indicate that seasonality in heterotrophic respiration also contributes to the observed seasonal cycle of NEP. Here, we use the dynamic vegetation model CLASS-CTEM (Canadian Land Surface Scheme–Canadian Terrestrial Ecosystem Model) – without deep soils or roots, hydraulic redistribution of soil moisture, or increased dry season litter generation – at two Large-Scale Biosphere–Atmosphere Experiment (LBA) sites (Tapajós km 83 and Jarú Reserve). These LBA sites exhibit opposite seasonal NEP cycles despite reasonably similar meteorological conditions. Our simulations are able to reproduce the observed NEP seasonality at both sites. Simulated GPP, heterotrophic respiration, latent and sensible heat fluxes, litter fall rate, soil moisture and temperature, and basic vegetation state are also compared with available observation-based estimates which provide confidence that overall the model behaves realistically at the two sites. Our results indicate that representing the effect of soil moisture on heterotrophic respiration in terms of soil matric potential and constraining heterotrophic respiration when absolute soil matric potential is both low (wetter soils) and high (drier soils), with optimum conditions in between, allows to correctly simulate NEP seasonality.