Abstract/Summary

Increasing plant diversity is seen as an important method for improving agricultural soil health, with benefits extending to increased soil carbon content and improved ecosystem health. However, including multi-species interactions into models is challenging. In this study, we represented multi-species interactions by producing different cash crop parameter values depending on the presence of undersown species. For this purpose, we used the STICS soil-crop model with observations of yield, green area index, and net ecosystem exchange chamber measurements from a number of plots in an extensive field experiment in southern Finland, where barley was cultivated with 0–8 undersown species. Calibration with the four-dimensional ensemble variational data assimilation method was found to be effective with up to three parameters of interest, beyond which issues of equifinality were present. Calibration had a positive effect on the performance of projected yields and other measurements, demonstrating the potential of this approach to capture unsimulated interactions. However, we found that the dominant parameter values change interannually to the degree that the calibration improvements did not translate, limiting the effectiveness of this approach to similar conditions as the calibration data. Additionally, large variance in secondary species abundance remains a challenge for modelling this style of intercropping.