The multitasking lesser mealworm (Alphitobius diaperinus) is a special beetle known as a pest in poultry, a resource for waste degradation and an alternative for protein production. This study compares the predictive accuracy of correlative species distribution models (SDMs) with a risk index derived from a mechanistic model. The study derives the mechanistic-based risk index from the ordinary differential equation that describes the population dynamics of A. diaperinus using the temperature-dependent bio-demographic rates, while the ensemble SDM is derived using well-known algorithms such as maximum entropy, random forest and so forth. We finally propose a hybrid model combining both approaches using a weighted average approach. When overlaid on occurrence data, the predictive accuracy of the mechanistic model globally varied across temporal scales, with the highest performance observed in the October-December quarter (27% of occurrences were predicted correctly). The comparison across geographic regions model had the best performance in Asia (94.4% accuracy), outperforming the two scenario SDMs (78.3%). In contrast, the correlative ensemble SDM performed better in Europe (93%), where we have most of the data, but was very sensitive to data gaps, especially in Africa. Finally, the proposed hybrid model outperforms both individual models in the global scenario (86.5% accuracy). These findings highlight the strengths and limitations of both modelling approaches and provide critical insights to optimise pest management strategies, sustainable utilisation and ecological forecasting by refining SDM through the integration of biological realism and empirical data.
Gwokyalya et al. (Wed,) studied this question.