Model-based agricultural landscape assessments: a review

Abstract

Agricultural landscapes are multifunctional and closely connected to the much wider food system. In our review, we explore three specific aspects of modelling approaches for agricultural landscape assessments: (a) how multifunctionality is commonly analysed to support decision-making for sustainable agricultural land management; (b) how the modelling approaches relate to the wider food systems; and (c) how gaps in the existing modelling approaches might be addressed. For this, we identified key elements of agricultural landscape assessments, including ecosystem services (ESS) provided, driving factors, and linkages between crop and livestock production, and to the wider food system. We reviewed 238 publications with respect to these elements. While biodiversity and the ESS 'water conditions' and 'atmospheric composition/conditions' are widely covered, they are rarely used in combination. Other ESS, such as 'pest and disease control', are largely missing. Our results further indicate strong differences between individual approaches regarding model parameterisation and consideration of abiotic, biotic, and management driving factors. Our analysis also shows that the interconnectedness of crop and livestock production is rarely considered and that return flows from the food system are not considered. Furthermore, impacts from the production of external inputs and off-site effects are not considered. Consequently, existing models might overlook trade-offs and synergies between landscape functions. Failure to consider variations in relevant driving factors and food system linkages likely results in studying incorrect levers for change and failing to show decision-makers the full scope of available action. We thus suggest adopting more encompassing modelling approaches to ensure coverage of the full scope of potential actions, whilst hedging against overly costly data requirements by, e.g. employing well-designed sensitivity analyses. In this way, the most relevant components and thus the most important leverage points for interventions can be identified.