Model selection, hummingbird natural history, and biological hypotheses: a response to Sazatornil et al.

Abstract

We have previously suggested that a shift from bee to hummingbird pollination, in concert with floral architecture modifications, occurred at the crown of Salvia subgenus Calosphace in North America ca. 20 mya (Kriebel et al. 2020 and references therein). Sazatornil et al. (2022), using a hidden states model, challenged these assertions, arguing that bees were the ancestral pollinator of subg. Calosphace and claiming that hummingbirds could not have been the ancestral pollinator of subg. Calosphace because hummingbirds were not contemporaneous with crown subg. Calosphace in North America. Here, using a variety of models, we demonstrate that most analyses support hummingbirds as ancestral pollinators of subg. Calosphace and show that Sazatornil et al. (2022) erroneously concluded that hummingbirds were absent from North America ca. 20 mya. We contend that "biological realism" - based on timing and placement of hummingbirds in Mexico ca. 20 mya and the correlative evolution of hummingbird associated floral traits - must be considered when comparing models based on fit and complexity, including hidden states models.