Ecosystem models such as ANAFORE are models that dynamically simulate pools and fluxes of carbon, water, nitrogen and other nutrients through an ecosystem. These models originally served as tools to predict plant production and were therefore highly plant-oriented. Increasingly, the recognition of other ecosystem services has created the demand for more complex process-based soil models. Only models that account for all key interactions between climate, plants and soil can become versatile tools to predict the effect of different management or future global changes on ecosystem services. Increasingly, the active role of microbiota has been recognized in models as an important driver for many soil processes. Furthermore, implementing soil fauna effects such as bioturbation, aggregation, fragmentation, biopore formation, and soil foodweb interactions into ecosystems has been suggested as the next step. The KEYLINK model steps up to this challenge and simulates soil fauna effects both on soil organic matter dynamics and hydrology, with the final goal of serving as a soil module for the next generation of ecosystem models that allow simulating effects of climate change and management on ecosystems, including its soil functions. However, the development of KEYLINK and other similar models has been impeded by the lack of empirical data available to inform such models. The overall objective of ANAFAUNA was to carry out empirical studies that contribute to the development of KEYLINK. Such studies allow i) to test specific assumptions made in first version of KEYLINK and ii) to validate the current version of the model as a whole.
This dedicated effort was carried out in two different systems: (i) a field mesocosm experiment (REGIMESHIFT) simulating climate change towards more persistent precipitation regimes, (ii) a laboratory mesocosm experiment (EARTHWORM) testing the effect of earthworms on soil structure and C cycling. These matched each other well because the former system contained plants but not earthworms while the latter contained earthworms but not plant roots allowing to disentangle these two sources of macroporosity. Each system was used to test selected assumptions made in the current KEYLINK model.
While several assumptions of the model were supported by the experimental results, some others were not. The results highlight that soil structure is dynamic but cannot be simply estimated from other soil properties such as total or microbial C, roots, fungal or earthworm biomass. Instead, cumulative C inputs or earthworm activity over time might be more important and could be a way forward in modelling aggregation or burrowing in KEYLINK. Furthermore, some interesting and potentially important patterns were observed that will be implemented in KEYLINK, namely the changes in soil water repellency and its impact on infiltration. Altogether, these findings will result in an improved version of KEYLINK. Furthermore, the project brought important evidence of the detrimental effect of persistent precipitation regimes on soil.
