Geographic patterns of speciation in West and Central Africa: The role of rivers, mountains, and refugia

Abstract

The tropical rainforests of Africa make up one of the largest rainforest blocks in the world, second only to the Amazon Rainforest of South America. This area hosts an incredible diversity of plants and animals and contains three of the world’s biodiversity hotspots, indicative of both its high levels of endemism and its high conservation priority. Despite this, the study of biodiversity in the African rainforest lags far behind that of almost every other major rainforest in the world. This is widely due to logistical issues driven by infrastructure, political instability, and fragmented political boundaries. The high levels of diversity in the African rainforest have been shaped by a variety of geological factors including the formation of two mountain ranges, four islands, and six major rivers, as well as continent wide climate events such as the Miocene aridification and the Pleistocene glaciation cycles. However, the roles of these factors in driving diversification in Central Africa are generally poorly understood. A major sampling gap exists for many taxa in the Congo Basin, and pollen core and fossil data tend to be poorly preserved in the tropics. In addition, many climatic and geologic events overlap spatially making it difficult to tease apart their combined/confounding effects. This dissertation addresses these difficulties by filling major sampling gaps in Central Africa and by using a combination of phylogenomic, population genomic, demographic and niche modeling analyses to gain a better understanding of the responses of West and Central African taxa to the major climatic and geographic events shaping African rainforest biota. Chapter 1 examines the role that the Pleistocene glaciation cycles played in the distribution of mountain endemic taxa across the Afromontane archipelago. Since the 1800s biogeographers have noted the high floral and faunal similarity between Afromontane regions when compared to the intervening lowlands. The prevailing hypothesis is that this similarity results from the spread of montane habitat into lowland areas during the glacial cycles of the Pleistocene. However, a lack of pollen core data from the Central African lowlands has made it difficult to determine if habitat shifts were extensive enough to allow biotic exchange between montane regions. In this study I use ecological niche modeling for sixteen Afromontane endemic bird, tree, and shrub species to infer the extent and most likely geographic positions of suitable habitat for Afromontane taxa during the present, the mid-Holocene and the Last Glacial Maximum (LGM). The results suggest widespread climatic suitability across lowland Central Africa for all taxa in our study during the LGM and, to a lesser extent, during the Mid-Holocene. Suitable areas connected all Afromontane regions in sub-Saharan African except for the Ethiopian Highlands, supporting the hypothesis that faunal and floral similarity between Afromontane regions is a result of species distributional changes during recent periods of global glaciation. Chapter 2 takes a modern, integrative approach to testing classical hypotheses of rainforest speciation in lowland West and Central Africa. This chapter tests predictions derived from the river, refuge, and river-refuge hypotheses of tropical species diversification proposed for the Amazon rainforest in the 1960s. Evolutionary and biogeographic patterns inferred from these hypotheses were originally based solely on species distribution data and suggest that most tropical speciation events are driven by allopatry associated with the formation of rivers, Pleistocene refugia, or a combination of the two. However, application of these hypotheses to the African rainforest was limited. In this study, I use phylogenomic and population genomic methods, combined with ecological niche model projections to the LGM and mid-Holocene, to assess the predictions of these classical hypotheses in the arboreal sub-Saharan snake genus Toxicodryas. Rivers were found to represent strong barriers to gene flow among populations of this genus, and no support was found for a major contraction of suitable habitat during the LGM. These results allow the rejection of both the refuge and river-refuge hypotheses in favor of the river hypothesis for diversification in this genus and have led to the description of two new species of Toxicodryas from the Congo Basin. Chapter 3 takes a deeper look at population genetic patterns associated with African Pleistocene refugia. Based on limited pollen core records, Pleistocene refugia have been proposed for several localities across West and Central Africa. Many of these refugia are associated with areas of high surface relief, such as montane regions, but also include a fluvial refuge in gallery forests around the Congo River. However, the degree to which these refugia have driven species diversification patterns in the African rainforest has been widely debated, often because of conflicting phylogeographic or species distribution data. In this study I test the predictions of the Pleistocene refugial diversification hypothesis in a widespread tropical African skink species, Trachylepis maculilabris, using a combination of dated phylogeographic analysis, population demography, and ecological niche modeling with projections to the LGM and mid-Holocene. I found five distinct genetic clusters with divergence times in the Pliocene and Pleistocene. Demographic analyses suggest population expansion as the earth warmed after the LGM and allowed for the determination of the most likely locations of Pleistocene refugia. These results suggest that the Congo River fluvial refuge played a role in species diversification patterns in T. maculilabris, and that mountain ranges played complex roles, with the Albertine Rift acting as a refugium and the Cameroon Volcanic Line acting as both a population barrier and a refugium. Chapter 4 examines the colonization history of two species in the Trachylepis maculilabris complex in the Gulf of Guinea islands. These islands are an offshore extension of the Cameroon Volcanic Line and consist of one continental and three oceanic islands of varying sizes and distances from the mainland. The oceanic islands have unusually high levels of biodiversity, resulting from dispersal from the mainland combined with in-situ diversification. Dispersal to the islands likely occurs through rafting via three major river systems and two oceanic currents that result in complex colonization patterns from both West and Central Africa. In this study I use dated phylogenomic, historical demographic, and ancestral geographic range reconstruction analyses to determine the most likely colonization history for this complex in the Gulf of Guinea islands. I found that the island species were monophyletic, indicating a single colonization event on São Tomé Island from the eastern Congo Basin around 14 million years ago. Colonization of the islands then likely proceeded sequentially from São Tomé to Príncipe to Tinhosa Grande, with the colonization of Tinhosa Grande occurring via a land bridge from Príncipe during the Pleistocene glaciation cycles. Chapter 5 focuses on the role that polyploidization plays in climatic niche evolution in the sub-Saharan African frog genus Xenopus. There are 29 species of Xenopus in two sub-genera. One of these species is diploid, 16 are tetraploid, seven are octoploid, four are dodecaploid and one is unknown. There are clear geographic patterns to the distributions of different ploidy levels within the genus, such that octoploids and dodecaploids are found mainly in and around the Cameroon Volcanic Line and the Albertine Rift. Polyploid plant literature has suggested a relationship between polyploidization and adaptation to new or extreme environments, potentially explaining these geographic patterns. In order to examine the relationship between polyploidization and niche evolution in Xenopus, I conducted ancestral state reconstructions of each species’ realized niche on a reticulate phylogeny for the genus. I found that niche expansion and contraction around the ancestral niche was common in this genus, while niche shift and niche novelty were relatively rare. While niche expansion and contraction were equally distributed throughout the phylogeny, niche novelty was more likely to be associated with higher polyploidy species.