The Systematics and Evolution of Neotropical Bellflowers (Campanulaceae: Lobelioideae)

Abstract

The Andean mountain chains of South America are home to exceptionally high levels of biodiversity, including one-sixth of all plant species. Among them, the 550 species in the Neotropical bellflower clade (Campanulaceae: Lobelioideae) represent one of the single largest endemic radiations. In this dissertation, I perform a series of studies, ranging from alpha taxonomy to phylogenetic comparative biology, to understand this group’s evolution.

In Chapter 1, I resolve phylogenetic relationships of the Neotropical bellflowers from five plastid DNA regions using maximum likelihood and Bayesian inference. This phylogeny greatly improves the resolution within this group and highlights the need for taxonomic revisions. My results demonstrate that the majority of species of Centropogon, Burmeistera, and Siphocampylus together form a primarily mainland Neotropical clade, collectively termed the centropogonids. Within the centropogonids, I identify high support for the monophyly of Burmeistera and the polyphyly of Centropogon and mainland Siphocampylus. Caribbean Siphocampylus, however, group with other Caribbean lobelioid species. Ancestral character state reconstruction identifies a dynamic pattern of fruit evolution within the centropogonids, which underscores the difficulty of diagnosing broad taxonomic groups on the basis of fruit type. Finally, we identify that the centropogonids form a Pan-Andean radiation with broad habitat diversity, which sets the stage for a subsequent chapter in my thesis (Chapter 4).

In Chapters 2 and 3, I describe four species of centropogonids as new to science. In Chapter 2, two species of Burmeistera from the Cordillera de Talamanca are described: B. serratifolia Lagom. & D. Santam., endemic to Panama, and B. monroi D. Santam. & Lagom., from both Panama and Costa Rica. These species are included in a dichotomous key to all Central American species of Burmeistera. In Chapter 3, I describe two species of Siphocampylus from the Central Andes of Peru and Bolivia: S. antonellii Lagom. & D. Santam., endemic to high elevation grasslands of Calca, Peru, and S. siberiensis Lagom. & D. Santam., endemic to cloud forests of Cochabamba, Bolivia. Both species are robust shrubs that produce tubular pink flowers that are likely pollinated by hummingbirds.

In Chapter 4, I use the Neotropical bellflowers as a model system to understand the triggers of rapid diversification in the Andes. I explore the interaction of abiotic and biotic factors by applying a series of diversification models that incorporate orogeny, climate, and biological trait evolution to a time-calibrated phylogeny of the group. Here, I demonstrate that speciation rates rose with increasing Andean elevation, while extinction decreased during global cooling. Of these, climate appears to have played the more important role, which I attribute to preadaptation to cool climates. Biotic traits that facilitate plant-animal mutualisms, specifically fruit type and pollination syndrome, additionally enhanced diversification rates. These results suggest that, while the Andes acted as a species pump for this lineage, a synergistic interaction of abiotic and biotic factors underlies its rapid diversification, which culminated in one of the fastest plant radiations documented to date: the centropogonids. This study underscores the complex interplay of ecological and historical determinants in generating the world’s biodiversity.

Finally, in Chapter 5, I use comparative methods to explore shifts between vertebrate pollination syndromes in the centropogonid clade. Such shifts, which are often associated with convergent floral phenotypes, are invoked as key factors underlying floral diversification in angiosperms. Using phylogenetic principal components analyses, I show that, despite morphological similarities between these pollination syndromes, centropogonid flowers adapted to pollination by bats and by hummingbirds fall into separate regions of morphospace. I further document the repeated evolution of pollination syndromes: the clade is ancestrally hummingbird pollinated, and bat pollination has evolved independently fourteen times, with nine reversions to hummingbird pollination. An obligate relationship with sicklebill hummingbirds, characterized by extremely curved bills, evolved only once. While the evolution of floral morphology has been dynamic in this clade, there is no difference in diversification rate in hummingbird and bat pollinated linages. These results suggest that specialized pollination syndromes are not evolutionary dead-ends. Instead, shifts between vertebrate pollinators may have greatly facilitated the maintenance of an ancestrally high diversification rate among centropogonid species.