Towards an annual carbon balance of biological soil crusts: parametric equations and neural networks to model gas exchange and net primary productivity

Abstract. Biological soil crusts (biocrusts) are communities of photoautotrophic and heterotrophic organisms forming a common ecological feature in dryland areas across the globe. Their ability to fix atmospheric carbon may coin them as an important factor for carbon balance and cycling, yet, the quantification of the net primary production of various biocrust types in natural environments remains largely uncertain. Therefore, the physiological response of biocrusts, as related to CO₂ gas exchange is a key area of investigation using both laboratory and modelling approaches. We present two methods to model the physiological response, specifically CO₂ gas exchange rates of biocrusts as a function of soil moisture, temperature and light intensity. The models are a parametric equation with optimized fitting parameters, and an artificial neural network model. Both methods can be applied to any specific biocrust type and, in this study, are applied to two types of biocrusts, a cyanobacteria- and a lichen-dominated biocrust, using laboratory measurements of CO₂ gas exchange rates as training data. Our models achieve very good agreement with independent test data and permit detailed insights into the physiological response of biocrusts to environmental conditions. As the models are not mechanistic, they can easily be applied to other organisms or environmental parameters in a similar fashion. We also demonstrate how such models can be used alongside field measurements of micrometeorological conditions in order to calculate the net primary productivity of biocrusts in specific locations.