Designing integrated management scenarios using simulation-based and multi-objective optimization : application to the peach tree-Myzus persicae aphid system

Summary

The design of crop-pest management systems in the context of integrated fruit production (IFP) is a complex decision-making process for which model-based design approaches are of growing interest. In this study, such an approach was developed to enhance the management of the Prunus persica (L.) Batsch-Myzus persicae (Sulzer) (Homoptera: Aphididiae) system. Firstly, we developed a process-based model and defined a set of performance criteria in line with IFP objectives. The model describes the interacting peach-aphid dynamics, fruit production and fruit quality, and their control by (i) cultural practices, (ii) releases of the biological control agent Harmonia axyridis Pallas (Coleptera: Coccinellidae) and (iii) insecticide applications. Secondly, model simulations of pre-defined management scenarios were performed. Scenarios were based on theoretical pest control strategies (i.e., 'no treatment', 'conventional', 'organic farming' and 'integrated') and combined different levels of cultural and pest control management variables. The relationships between management and performance variables were studied. Thirdly, the simulation model was coupled with a multi-objective optimization module to design new management scenarios for three virtual farmer production profiles that varied in terms of the relative importance the farmer attaches to each one of the performance criterion. Our results showed that agronomic performances were largely explained by cultural practices, while aphid pressure was largely explained by pest control strategies. Under the 'no-treatment' strategy, aphid pressure displayed a bell-shaped response to nitrogen levels and increased with pruning intensity. These responses were attenuated or even reversed under the other strategies. The scenarios using the 'conventional' (insecticide-based) pest control strategy were the best, regardless of the farmer profile. However, scenarios using the 'no-treatment' strategy had good performances as well, as long as the long-term risk of aphid infestation was not considered. Under the 'no treatment' and 'integrated' pest control strategies, optimal values of cultural management variables displayed a high variability between the farmer production profiles, while they were independent of the farmer profile under the other two strategies. This study demonstrates that multi-criteria crop-pest simulation models associated with multi-objective optimization approaches provide a framework suitable for designing production systems, while taking farmers' expectations into account.