Plant invasion reduces density-dependent pollination but not florivory

Area of Science:

• Ecology
• Conservation Biology
• Deep Learning Applications

Background:

• Plant invasions disrupt native ecosystems and require effective monitoring.
• Deep learning offers automated methods for assessing ecological interactions and ecosystem functioning.

Purpose of the Study:

• To assess the impact of dog-strangling vine (Vincetoxicum rossicum) invasion on native plant communities using drone technology and deep learning.
• To quantify changes in floral density, pollinator activity, and herbivory in response to invasion.

Main Methods:

• Utilized a consumer-grade drone and deep learning algorithms to measure floral density in invaded meadows.
• Conducted field observations of pollination and herbivory on the native plant Symphyotrichum novae-angliae.
• Analyzed the relationship between invasion levels, floral density, pollinator abundance, and herbivory rates.

Main Results:

• Increased invasion correlated with decreased floral density, reduced pollinator abundance, and suppressed pollination services.
• Herbivory impacted plant reproduction independently of plant density, primarily through indirect effects on host abundance.
• Drone-based deep learning detected reduced pollinator habitat quality along the invasion gradient.

Conclusions:

• Invasive dog-strangling vine impairs native plant reproduction through mechanisms both dependent and independent of pollination.
• The invasion degrades pollinator habitat quality and negatively affects the reproductive success of late-season flowering plants.
• Integrated drone and deep learning approaches provide efficient tools for ecological monitoring of invasive species impacts.