Phylogenetic estimation of diversity-dependent biogeographic rates using deep learning

Area of Science:

• Ecology
• Evolutionary Biology
• Biogeography

Background:

• Ecological theory suggests local species richness impacts speciation, extinction, and dispersal rates.
• High species density may lead to a carrying capacity for local species richness due to increased competition and extinction.

Purpose of the Study:

• Introduce DDGeoSSE, a phylogenetic diversification model incorporating diversity-dependent effects on biogeographic rates.
• Test alternative diversification scenarios, including positive, negative, and neutral species interactions.
• Investigate the role of local species richness in shaping diversification dynamics.

Main Methods:

• Developed a fully generative, event-based phylogenetic diversification model (DDGeoSSE).
• Derived mathematical and statistical properties of the model, including carrying capacity at equilibrium.
• Employed deep learning with phyddle for parameter inference and model selection due to complex likelihood functions.

Main Results:

• Validated the model's carrying capacity predictions through simulation.
• Applied DDGeoSSE to Caribbean Anolis lizards and cloud forest Viburnum plants.
• Found significant evidence for local species richness influencing diversification dynamics in both studied clades.

Conclusions:

• Local species richness is a significant factor shaping clade diversification.
• The DDGeoSSE model provides a robust framework for studying diversity-dependent speciation and extinction.
• Deep learning facilitates the analysis of complex phylogenetic diversification models.