Can inverse calibration help improving process-explicit species distribution models?

Process-explicit models (PEMs) are expected to provide reliable projections of species range shifts because they explicitly model the biological mechanisms that drive species responses to climate. However, their application is often limited by the need for diverse and detailed datasets, which are only available for a limited number of species. Inverse calibration has been identified as an avenue to help calibrate PEMs for many species, but it is still unclear whether it can provide biologically meaningful parameter estimates. Here, we investigated the potential of inverse calibration techniques to improve the accuracy of PEMs. We examined the discrepancies in parameter estimates obtained by classical and inverse calibration approaches. We evaluated two inverse calibration strategies: (i) calibrating all parameters simultaneously and (ii) focusing only on critical parameters. We assessed the realism of the obtained parameter estimates and the simulated processes by comparing them with measurements and observations across Europe. We show that when the entire model is calibrated at once, the model structure alone may not sufficiently constrain parameter estimation, leading to unrealistic parameter values. However, selective application of the inverse calibration approach – focusing on critical parameters – can improve model performance while still simulating realistic biological mechanisms.