4.3 Integrative species diversification studies
Prior to the availability of genomic data, dated phylogenies, paleo-distribution modeling, and statistical analyses of historical demography (Carstens & Richards, 2007; Ellegren, 2014; Knowles, 2009; Knowles & Madison, 2002; Luikart, England, Tallmon, Jordan, & Taberlet, 2003), biogeographers commonly observed patterns of species distributions from which they attempted to infer mechanisms of diversification (e.g. Avise, 2000; Templeton, 2001). The majority of rainforest refugial studies were conducted in this descriptive, pattern-based manner, usually employing only single-locus mitochondrial DNA data sets (e.g. Fjeldså & Lovett, 1997; Haffer, 1969; Mayr & O’Hara, 1986). The pitfalls and lack of power in such approaches have been discussed elsewhere (e.g., Knowles & Madison, 2002; Provan & Bennett, 2008; Stewart, Lister, Barnes, & Dalén, 2010), and are not the focus of this study. However, with respect to hypotheses of tropical African species diversification, it is noteworthy that so many of the concepts and hypotheses surrounding refugia have only been evaluated by single-locus data sets during the era of mitochondrial gene phylogeography (but see more recent, multilocus examples in Leaché et al., 2019; Portik et al., 2017).
The complexity of geographic barriers in West and Central Africa, and the association of refugia with areas of high surface relief or riparian zones (Hofer et al., 1999; 2000; Fig. 1), makes it extremely difficult to untangle the relative importance of different diversification mechanisms with distribution data alone (Leaché et al., 2019; Portik et al., 2017). This difficulty is particularly salient in our study system, where distribution data may have suggested the association of populations with hypothesized refugia around the Congo River, Gabon, and in West Africa (refugia 9, 5–8, and 1–3 respectively, Figs. 1, 2). Yet, our dated phylogenies and paleo-distribution models reject the Pleistocene population age and habitat contraction predictions of the refugial hypotheses in favor of the river barrier hypothesis. These results highlight the importance of using an integrative, multidisciplinary approach to statistically distinguish among competing hypotheses to explain high levels of geographically concentrated species biodiversity. Moving beyond pure pattern-based inference, a deeper and more nuanced understanding of the production, partitioning, and maintenance of diversity in complex landscapes may lead to inference of environmental and evolutionary processes that accumulate terrestrial biodiversity in tropical areas, which coincide in many cases with Global Biodiversity Conservation Hotspots (Hrdina, & Romportl, 2017; Mittermeier, Myers, & Mittermeier, 2000; Mittermeier, Turner, Larsen, Brooks, & Gascon, 2011; Myers, 1988) and other imperiled ecosystems of Earth.