Models are essential to assess the socio-economic credentials of new agroforestry systems. In this study, we showcase robust optimisation as a tool to evaluate agroforestry’s potential to meet farmers’ multiple goals. Our modelling approach has three parts. First, we use a discrete land-use model to evaluate two agroforestry systems (alley cropping and silvopasture) and conventional land uses against five socio-economic objectives, focusing on the forest frontier in eastern Panama. Next, we couple the land-use model with robust optimisation, to determine the mix of land uses (farm portfolio) that minimises trade-offs between the five objectives. Here we consider uncertainty to simulate the land-use decisions of a risk-averse farmer. Finally, we assess how the type and amount of agroforestry included in the optimal land-use portfolio changes under different environmental, socio-economic and political scenarios, to explore the conditions that may make agroforestry more attractive for farmers. We identify silvopasture as a promising land use for meeting farmers’ goals, especially for farms with less productive soils. The additional labour demand compared to conventional pasture, however, may prove an important barrier to adoption for farms facing acute labour shortages. The selection of agroforestry responded strongly to changes in investment costs and timber prices, suggesting that cost-sharing arrangements and tax incentives could be effective strategies to enhance adoption. We found alley cropping to be less compatible with farmers’ risk aversion, but this agroforestry system may still be a desirable complement to the land-use portfolio, especially for farmers who are more profit-oriented and tolerant of risk.
Agroforestry has been promoted as a key forest landscape restoration (FLR) option to restore ecosystem services in degraded tropical landscapes. We investigated the share and type of agroforestry selected in an optimized landscape, accounting for a mosaic of alternative forest landscape restoration options (reforestation and natural succession) and forest and common agricultural land-uses. We extend previous studies on multi-objective robust optimization and the analytic hierarchy process by a systematic sensitivity analysis to assess the influence of incorporating agroforestry into a landscape. This approach accounts for multiple objectives concurrently, yet data and computational requirements are relatively low. Our results show that experts from different backgrounds perceive agroforestry (i.e., alley cropping and silvopasture) very positively. Inclusion of large shares of agroforestry (41% share of landscape) in the FLR mix enhanced simulated ecosystem service provision. Our results demonstrate that landscapes with high shares of agroforestry may also comprise of high shares of natural forest. However, landscapes dominated by single agroforestry systems showed lower landscape multifunctionality than heterogeneous landscapes. In the ongoing effort to create sustainable landscapes, our approach contributes to an understanding of interrelations between land-covers and uncertain provisions of ecosystem services in circumstances with scarce data.