4.2 Paleo-distributions and habitat stability
The nature of the intervening habitat surrounding rainforest refugia during the Pleistocene has been widely debated. Some authors have argued that much of the Central African rainforest was replaced by savannas (DeMenocal, 2004; Maley, 1996; Maley & Brenac, 1998), while others have emphasized the possibility of more subtle shifts in forest composition (i.e., from wet to dry forest; Colinvaux et al., 1996, 2001; White, 1981). Toxicodryas species are generally characterized as arboreal across rainforest and woodland habitats and the two species exhibit widely overlapping distributions in West and Central Africa (Chippaux & Jackson, 2019). Our paleo-distribution modeling suggested that no substantial contraction of suitable climate occurred for either species during the LGM (Fig. 7a), and our habitat stability mapping suggested that core ranges of both species have remained stable for the past 22,000 years (Fig. 7c). The greatest potential for habitat expansion in this species appears to be to the south into today’s northern Angola and the southern DRC (Fig. 7).
Similar paleo-distribution studies on frogs have suggested substantial habitat contraction in Central Africa during the Pleistocene (Leaché et al., 2019; Portik et al., 2017). In contrast, our inferred widespread habitat stability in Toxicodryas may be due to the relatively reduced dependence of arboreal snakes on moist habitats, as reflected by their distribution in both woodland and rainforest. The stability ofToxicodryas habitat through the Pleistocene supports the hypothesis that rainforest composition shifted to dryer woodlands surrounding rainforest refugia, instead of a more dramatic shift to strict savannah habitat. Southward shifts in species suitability may correspond with predicted forest distribution shifts of White (1981), suggesting a replacement of lowland rainforest with montane forest habitat.