Fuente: PubMed "pollination" Oecologia. 2026 Apr 16;208(5):59. doi: 10.1007/s00442-026-05881-x.ABSTRACTUnderstanding the impact of plant invasion on multitrophic community dynamics and coexistence requires widespread and frequent monitoring. Deep learning can be used to automate the measurement of indicators of ecological interactions and ecosystem functioning. In this study, we used a consumer-grade drone paired with deep learning to assess floral density in meadows invaded by the dog-strangling vine Vincetoxicum rossicum (Kleopow) Barbar. (Gentianales: Apocynaceae) at the Rouge National Urban Park in the Greater Toronto Area, Ontario, Canada. Alongside these measurements, observations of pollination and herbivory was completed on Symphyotrichum novae-angliae (L.) G.L.Nesom (Asterales: Asteraceae), a self-incompatible, pollinator-dependent native plant that experiences herbivory by a widespread specialist weevil, Anthonomus rufipes LeConte (Coleoptera: Curculionidae). Our results suggest that as invasion progresses, pollination services are reduced due to the decrease in floral density which suppresses pollinator abundance and activity. Conversely, while herbivory had a strong effect on plant reproduction, it was density independent and thus unaffected by direct effects of invasion, but rather indirect through reduced host abundance. By pairing deep learning with drone technology, we detected patterns consistent with a reduction of pollinator habitat quality along the invasion gradient. Furthermore, we find that invasion appears to suppress plant reproduction by means of separate processes that are either independent of or dependent on pollination. Overall, the results suggest that invasion reduces pollinator habitat quality while simultaneously resulting in ecological conditions consistent with the reproductive impairment of late-season flowering resident plant species.PMID:41989621 | DOI:10.1007/s00442-026-05881-x