Answers are grounded in this paper and cite their sources.
Species distribution models depend on modeling choices. We used MaxEnt for five pest insects in the contiguous United States, four introduced and one native: the brown marmorated stink bug (Halyomorpha halys), corn earworm (Helicoverpa zea, native), spongy moth (Lymantria dispar dispar), Diaprepes root weevil (Diaprepes abbreviatus) and spotted lanternfly (Lycorma delicatula). Models trained on US records and the 1970–2000 climate were projected to 2021–2040 and 2041–2060 under four emission pathways with five climate models, mainly ACCESS-CM2. We varied m, the share of background points drawn near the records over the whole country and inside species-specific accessible areas. No single m was best in either extent. The Boyce index separated the settings, while the area under the receiver operating characteristic curve (AUC) changed little for m≤0.75. Removing out-of-range records made a uniform background clearly best for the spongy moth. Averaged over climate models, the suitable area decreased by 2041–2060 for four species and increased for the root weevil. Its size, and for two species its sign, depended on the climate model; the corn earworm loss and root weevil gain also depended on the calibration area. Permutation-importance rankings changed between background draws. The maps show climatic suitability, not forecasts of occupation.
No takes yet. Share an insight, caveat, or question.
Chouhan et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: