Chrysochloridae

Our analyses of extant taxa (Figs 10, 11) strongly support the placement of Calcochloris obtusirostris as the sister taxon of the two species of Chrysospalax, Chrysospalax trevelyani and Chrysospalax villosus Smith, 1833 . They also support Carpitalpa arendsi as the sister taxon to Neamblysomus, a clade which, in turn, forms the sister taxon to Amblysomus . The species stuhlmanni, long regarded as part of the genus Chrysochloris (Bronner and Jenkins 2005), appears instead as sister taxon to a Chrysochloris asiatica – Cryptochloris clade. We therefore elevate Kilimatalpa (Lundholm 1954) from subgeneric to generic status, leaving both generic names monotypic for extant species, in reference to Kilimatalpa stuhlmanni and Chrysochloris asiatica .

Based on these results, we define the Amblysominae as the group encompassing Amblysomus and its sister taxon Neamblysomus – Carpitalpa arendsi (Figs 10, 11). Our data also support the intrageneric relationships of Amblysomus as articulated by Mynhardt et al. (2015), in particular that the species marleyi and meesteri Bronner, 2000, previously regarded as subspecies of A. ‘ hottentotus ’, comprise the sister clade to all other species of Amblysomus . This is also consistent with the treatment of marleyi as a separate species of Amblysomus by Bronner (1995a, 2000). As in the study by Mynhardt et al. (2015), our data place Amblysomus corriae Thomas, 1905, the only species of Amblysomus broadly distributed throughout the Western Cape province of South Africa, as the sister taxon to the remaining non- meesteri / marleyi species of Amblysomus . Our data do not address the phylogeography of A. ‘ hottentotus ’ populations outside of the Eastern Cape, but are also consistent with Mynhardt et al. (2015) by placing A. septentrionalis – A. robustus as the mostnested clade within the genus, sister to A. h. hottentotus sensu stricto (i.e., populations near Grahamstown and King Williams Town in the Eastern Cape).

The two information-theoretic criteria (BIC vs. AICc; see Table 5) we used to derive models of sequence evolution resulted in mutually consistent topologies except for the chrysochlorid root node. The BIC topology based on DNA–indels–morphology showed a trichotomy (Fig. 11A); the AICc topology placed a Kilimatalpa – Chrysochloris – Cryptochloris group (pp = 1.0) diverging from a clade containing the remaining extant chrysochlorids (pp =.56). Using only DNA–indels, BIC favoured a divergence of Eremitalpa – Huetia (pp =.67) from a clade containing other extant chrysochlorids (pp =.59; Fig. 10A), whereas AICc showed a trichotomy. Neither BIC nor AICc Bayesian analyses including our fossil taxa resolved the root of crown Chrysochloridae, but both reconstructed Namachloris and Prochrysochloris on the chrysochlorid stem (Fig. 12A; Supporting Information, Fig. S1).

Besides the ambiguity surrounding the extant chrysochlorid root, and using either BIC or AICc for model selection, adding morphology to our DNA–indel alignment led to one topological difference: the placement of Chlorotalpa as sister to amblysomines (pp =.92; Fig. 11A) rather than Chrysospalax – Calcochloris (pp =.87; Fig. 10A). Parsimony with implied weights favoured the latter ( Chlorotalpa, ( Chrysospalax, Calcochloris)), with the further complication that without morphology, Huetia was also drawn into a clade with Chlorotalpa (Fig. 10B), albeit with bootstrap support <50%. In contrast, the optimal parsimony topology using implied weights with DNA–indels–morphology reconstructed Huetia near the base of the chrysochlorid crown radiation, one node crownward from Eremitalpa with parsimony bootstrap support of 61% (Fig. 11B).

While our dataset does not resolve the chrysochlorid root node, it does narrow down the possibilities to a subset of the species previously included in the ‘Chrysochlorinae’ (Bronner and Jenkins 2005), rendering that taxon paraphyletic. We therefore restrict the taxon Chrysochlorinae to the smaller clade of Kilimatalpa as sister taxon to Chrysochloris asiatica – Cryptochloris (Bayesian pp = 1.0). We propose the designation Chrysospalacinae for Calcochloris obtusirostris – Chrysospalax (Figs 10, 11). The affinities of the remaining three chrysochlorid genera, Chlorotalpa (restricted here to Chlorotalpa sclateri and Chlorotalpa duthieae), Eremitalpa and Huetia, are not yet resolved with sufficient support to merit high-level taxonomic designations.