The mango tree - blossom gall midge system: toward in silico assessment of management practices

Mango (Mangifera indica L.), a major fruit production in tropical and subtropical regions, is facing many production constraints. Mango yield is irregular across years, fruit quality is heterogeneous at harvest, and mango tree exhibits phenological asynchronisms within and between trees that result in long periods with phenological stages susceptible to pests and diseases. Among them, the mango blossom gall midge (BGM, Procontarinia mangiferae Felt) is a major pest of mango which can cause significant yield losses by damaging mango inflorescences. A mango-BGM model is currently developed from experimental data. Its final objective will be in silico assessment of BGM management levers (e.g., manipulation of mango phenology to synchronize flowering and soil mulching during flowering, used as physical barrier to break the BGM life-cycle). The mango-BGM model simulates the dynamics of inflorescence and BGM populations of an orchard at a daily time-step during the period of mango flowering. The orchard is structured into three zones (or patches) according to applied mulching treatments. In a first approach, the model is defined at the patch scale and considers: i) age-structured inflorescence population dynamics within each patch, accounting for natural development and BGM-induced mortality of inflorescences; ii) stage-structured BGM population dynamics within each patch, differentiating the effect of mulching treatments on the BGM life-cycle for each patch; and iii) orchard colonization and movements of BGM adults between patches, both driven by inflorescence abundance in each patch. The model was parameterized from existing knowledge and experimental data collected in Reunion Island on ‘Cogshall’ cultivar. In a further step, virtual experiments will be performed with the model to assess the effects of management levers on mango flowering and subsequent fruit yield, according to exogenous pest pressure. The on-going modeling approach and preliminary results are presented and discussed.