All newly reported specimens of CVL Typoderus–like beetles are adults obtained from fifteen forest litter samples (Table 1) in wet primary forests of Mt. Oku, Mt. Cameroon and Mt. Kupe in Cameroon using standard sifting methods (Grebennikov 2017, 2019c). This paper is a sequel to the recent work focused on Tanzanian Typoderus (Grebennikov 2019c) and the sister clade of the genus (Grebennikov 2017), therefore it re-uses already reported DNA data and re-employs the same laboratory and analytical procedures. In brief, about a hundred newly sampled CVL Typoderus-like weevils were preliminary sorted into five geographically- and visually-coherent morphospecies. External diagnostic morphological characters for these morphospecies are given in Table 4 and were clear-cut and diagnostic for all individual specimens examined (except for two characters with intraspecific variation). Of them 20 specimens representing all five morphospecies were DNA barcoded (=sequenced for 658 bp of the 5′ end of cytochrome c oxidase subunit I, Hebert et al. 2003a, b; implemented in the Canadian Centre for DNA Barcoding, CCDB, University of Guelph, Canada, http://www.ccdb.ca) in support of analysis A1, A2 and A4 (see below); their images and DNA barcode sequences are available online at BOLD dataset dx.doi.org/10.5883/DS-VGDS006. GenBank accession numbers of 20 new DNA barcodes are MH917885–902, MH981205–6 (Table 3). Additionally, 14 of them were sequenced for two more markers, in support of analysis A3 (see below; Table 3).

To test hypotheses H1–H5, the total of four DNA analyses was implemented.

First analysis (A1) designed to assess hypothesis H1 (five morphospecies represent five biological species) by testing whether DNA barcode clusters match morphological and geographical grouping of specimens. By doing so, consistent signal was sought from different and independent sources of evidence in an attempt to delimit independent evolutionary units herein described as new species. For this purpose, 20 newly generated DNA barcodes 531–658 bp in length were analysed using the Neighbour Joining (NJ) method, Barcode Index Number cluster identification algorithm (BIN; Ratnasingham and Hebert 2013) and Kimura 2 parameter, as implemented in the online engine of the Barcode of Life online database (=BOLD, Ratnasingham and Hebert 2007, http://www.boldsystems.org).

To estimate evolutionary divergence over sequence pairs for groups recognized in section 4.2 below as five new Cameroonian Typoderus species, 14 terminals representing these species were analysed separately using MEGA5 (Tamura et al. 2011). The 2136 position matrix was reduced to that of a total of 1599 positions, by removing positions containing gaps and missing data. Two pairs of values were calculated: the number of base differences per site from averaging over all sequence pairs within and between species, as well as uncorrected p-distance.

Second analysis (A2): designed to test hypothesis H2 (CVL weevils taxonomically belong to the genus Typoderus). Considering (1.) diagnostic morphological characteristics of CVL specimens strongly in support of this hypothesis, (2.) lack of a clearly identified sister-group of the clade formed by Typoderus and its species-poor sister group of the genus Lupangus; (3.) multiple non-monophyly of Molytinae (Shin et al. 2017) and (4.) the lack of new significant DNA data on Molytinae (as compared to those recently released, Grebennikov 2017), a full-sized phylogenetic analysis with multiple non-Typoderus outgroups was not deemed feasible or, indeed, necessary. Instead, all 20 newly generated DNA barcodes were identified by using phenetic methods such as the BLAST algorithm implemented in GenBank (Altschul et al. 1997) or through the Neighbour Joining (NJ) clustering method (as in Analysis A1). This comparison was done against all Curculionidae records currently in BOLD (5,117 species with barcodes, 55,115 specimens with barcodes; as on November 25, 2020).

Third (topological) analysis (A3): designed to shed light on all five hypotheses. Fourteen CVL specimens representing all five morphospecies (as recovered on the NJ tree in analysis A1) were additionally sequenced for two nuclear ribosomal loci: internal ribosomal spacer 2 (ITS2) and 28S rDNA (Table 2). All corresponding laboratory work was done in CCDB using protocols and primers described earlier (Grebennikov 2017). The three-marker dataset was formed by 14 newly sequenced CVL terminals (the ingroup) merged with the 71 terminal dataset from Grebennikov (2019b); the latter containing 70 Tanzanian Typoderus (the outgroup) and a single Lupangus terminal to root the topology. Images of all 85 sequenced specimens and their DNA data are available online at BOLD dataset dx.doi.org/10.5883/DS-VGDS005. Alignment of the ITS2 and 28S sequences was made using the MAFFT 7 online platform (http://mafft.cbrc.jp/alignment/server) and the Q-INS-i algorithm utilising the secondary structure information (Kuraku et al. 2013; Katoh et al. 2017). To correct for visually detected alignment errors, 34 positions at the 5 end of four relatively short ITS2 sequences were manually re-aligned (specimens 2130: 10 positions; 2154: eight positions; 3094: three positions and 7164: 11 positions). No other modifications to the alignments were made and no parts of the alignments were excluded from the analysis. The resulting matrix of 85 terminals had 2,136 aligned positions (Table 2) and 23% missing data (mainly due to insertions and deletions in ITS2). GenBank accession numbers of new DNA sequences from 14 CVL weevils are in Table 3. The concatenated matrix was partitioned into three fragments (Table 2) and an independent CAT approximation (Stamataikis et al. 2008) of the GTR+G model was applied independently to each data partition (Abadi et al. 2019). Phylogenetic analysis was conducted on the CIPRES Science Gateway online platform (Miller et al. 2010; http://www.phylo.org) using the Maximum Likelihood (ML) method and RAxML 8 tool (Stamatakis 2014). Support values were obtained with 1000 bootstrap replicates (Stamatakis et al. 2008) and topology was visualized in FigTree v1.4.0 (Rambaut 2019).

Fourth (temporal) analysis (A4) was used to partially test hypotheses H3–H5, each of them requiring dates on species divergences. Considering monophyly of CVL Typoderus–like weevils (see Results) and the lack of relevant calibrating points, a fixed-rate temporal analysis was the only available option. For this purpose the matrix from the analysis A1 consisting of 20 DNA barcodes of CVL weevils was re-analysed and no outgroup used. To estimate divergence times, a fixed molecular clock rate of 0.018 nucleotide substitutions per site per million years per lineage (subs/s/MY/l) was applied. This value is consistent with those obtained in other beetles (Papadopoulou et al. 2010; Andújar et al. 2012), other insects (Brower 1994) and other arthropods (Bauzà-Ribot et al. 2012), and was used for dating evolutionary events in Tanzanian Typoderus (Grebennikov 2019b). Bayesian phylogenetic analysis in BEAST 1.8 (Drummond et al. 2012) was used to simultaneously estimate an ultrametric phylogenetic tree and rates of diversification. The HKY+G evolutionary model (estimated in MEGA 7, Kumar et al. 2016) was applied and the MCMC chains ran for 10 million generations. Convergence of all parameters was checked in TRACER (Drummond et al. 2012) and consensus trees were estimated with TreeAnnotator (Drummond et al. 2012) discarding 25% initial trees as a burn-in fraction. No a priori topological constrains (such as rooting) were added in the analysis.

Various species description workflows (Riedel et al. 2013; Meierotto et al. 2019) converged on importance of providing (1.) DNA barcode, (2.) precise locality data and (3.) image of the holotype. Here I follow the new species description workflow adopted for the Typoderus + Lupangus clade (Grebennikov 2017, 2019b) by presenting species-level diagnostic morphological characters in a table (Table 4). These characters consistently list all easily observed differences of five new species, including those of male genitalia. The latter were dissected only for holotype male specimens (distinguished from females by slightly depressed abdominal ventrites 1 and 2).

No attempt was made to compare male genitalia within the newly described species, nor to dissect and study female genitalia. Considering that Typoderus form narrow-range morphologically and genetically distinct clades in at least relatively well-sampled Tanzania (Grebennikov 2019b), no consistent attempt was made to compare morphology of five herein described new Cameroonian species with that of six Typoderus from Tanzania (Grebennikov 2019b). As before, the holotype body length (measured in dorsal view between anterior margin of pronotum and apex of elytra) is the only reported measurement; all other measurements and ratios can be obtained from images (dx.doi.org/10.5883/DS-VGDS006). Overall, taxonomic procedures implemented in this paper were designed to consistently generate valid species names with the least effort, with focus on diagnosis, not descriptions (Wheeler et al. 2012; Renner 2016; Miralles et al. 2020; Vences 2020). This approach termed “turbo-taxonomy” (Butcher et al. 2012; Summers et al. 2014) or fast-track taxonomy (Riedel et al. 2013) was designed to expedite naming of new species in particularly understudied and species-rich clades, which the genus Typoderus appears to be. Other examples include Neotropical Gracillariidae leaf-mining moths (Leed et al. 2014) or Cecidomyiidae gall midges; the latter cosmopolitan family of some 6–7 thousand species was estimated to contain up to 1,800,000 species, a bewildering number exceeding that of the total of all scientifically named animal species (Hebert et al. 2017).

Interpretation of the herein presented results led to conclude that:

Hypothesis 1 (five preliminary recognized and geographically coherent morphospecies of CVL Typoderus-like weevils are biological species) is strongly supported, since three independent lines of evidence (morphology, geography, DNA) groups specimens in identical clusters herein assumed to be discrete biological species. The sympatric occurrence of two strongly genetically divergent pairs (at Mt. Oku and at Mt. Kupe) offers an additional argument for species-level distinctness of these lineages. Moreover, the herein defined species (including both sympatric pairs) exhibit constant differences not only in COI (Fig. 2, a mitochondrial marker), but also in comparably rapidly evolving ITS2 (while having almost no differences in more conservative 28S; both are nuclear markers). Such concordance of unlinked markers (mtDNA vs. nucDNA) offers further evidence for species-level distinctness. The species to be named T. iphitus in section 4.2 might consist of two cryptic species due to the deep barcode divergence in the same order as that between the other species, but this needs to be investigated in more detail and is beyond the scope of this paper

Hypothesis 2 (all analysed CVL weevils belong to the genus Typoderus) is strongly supported, since phenetic similarities in DNA sequences (analysis A2) reinforce morphological identifications and assign all specimens to monophyletic Typoderus. Consequently, these CVL weevils are consistently referred to as Typoderus using five new species-group names introduced below.

Hypothesis 3 (at least one of the newly discovered CVL Typoderus is a paleoendemic) is weakly rejected, since none of CVL Typoderus forms a sister-group to a well-supported clade containing non-CVL species and reliably well-known and sufficiently old divergence date (although the cut-off date can hardly be precisely defined).

Hypothesis 4 (at least one divergence between allopatric CVL Typoderus species might be attributed to simple vicariance via habitat isolation of CVL sky islands during the post-Miocene climatic fluctuations) is weakly rejected, because by the beginning of the Pliocene at 5.3 MYA all five species of CVL Typoderus have already diverged (Fig. 4, but see below on the limitations of the temporal analysis).

Hypothesis 5 (at least one case of sympatry of CVL Typoderus might be attributed to a secondarily meeting of recently spectated populations through temporary habitat reconnection; the “species-pump” hypothesis) is weakly supported. In both cases of CVL sympatry (T. amphion sp. nov. and T. canthus sp. nov. co-occurring on Mt. Oku and T. iphitus sp. nov. and T. telamon sp. nov. co-occurring on Mt. Kupe) sister clades of each of four species occur on a different, although a relatively nearby CVL sky-island, which might have served as a source of secondary colonisation.

As a word of caution, the herein undertaken attempt to study spatial and temporal aspects of CVL Typoderus evolution is subject to four significant limitations. Firstly, the genus remains acutely undersampled. No DNA data on Typoderus are available from the >3,000 km gap separating sequenced populations in Tanzania and Cameroon, and even these two best studied countries likely have just a fraction of their evolutionary distinct Typoderus lineages samples and studied. It seems unlikely that Typoderus is truly absent from most of the Congo Basin lowland rainforest (Fig. 1), since focused litter sampling in nearby Tanzania detected this genus in all but two targeted forest blocks (Grebennikov and Heiss 2018; those two are geologically recent volcanoes Mt. Hanang and Mt. Meru likely not yet colonized by the genus). When available, new data are expected to modify, or perhaps even negate, some of the herein presented conclusions (for example, detection of non-monophyly of CVL Typoderus or perhaps significantly different divergence dates).

Secondly, very few other CVL animal clades were studied in sufficient detail to permit meaningful comparison. Similar to Typoderus weevils, all CVL laminate-toothed Otomys rats are monophyletic (two species diverging in the Pleistocene) and represent the western-most records of the genus, which is absent throughout most of the lowland Congo Basin (fig. 1 in Taylor et al. 2014). Phrynobatrachus puddle frogs widely distributed in sub-Saharan Africa are also represented in CVL by an endemic clade only and, similar to Typoderus, the clade’s divergences took place in Middle to Late Miocene (Zimkus and Gvoždík 2013; the analysis was perhaps biased to older dates by using suboptimal priors, email correspondence with V. Gvoždík on August 12, 2019). These two papers are the only ones providing phylogeographic results comparable with other of CVL Typoderus.

Thirdly, and perhaps most significantly, the herein implemented temporal analysis utilizes a fixed-rate molecular clock of a single maternally-inherited protein-coding marker. This methodological oversimplification is unavoidable, since no other calibrating methods are presently available for this understudied clade. This substitution rate agrees with those of other Arthropoda (see four references in Material and Methods), however its application to CVL Typoderus might, or might not, be correct, because at least a four times greater rate has been detected in the same DNA fragment among similarly wingless Trigonopterus Fauvel, 1862 weevils from the Sunda Arc (0.0793 subs/s/MY/l, analysis 2 in Tänzler et al. 2016, versus 0.018 subs/s/MY/l implemented herein). If the CVL Typoderus rate is at least twice higher than the one implemented in analysis A4, then the origin of the CVL clade’s crown group and its diversification would be estimated at the beginning of the Pliocene (Fig. 4). Such a result would reverse the present rejection of hypothesis H4, and suggest a link between CVL Typoderus evolution and post-Miocene cyclic climatic fluctuations.

Fourthly, the herein implementer temporal analysis possibly overestimated the ages, necessarily using (largely saturated) mitochondrial sequences to estimate comparatively old ages (as demonstrated Near et al. 2017 for ray-finned fishes). Without relevant fossil evidence and improved phylogenetic resolution within the Molytinae, it is not possible to fine-tune the herein implemented temporal analysis A4 and, therefore, these results should be taken with caution.

Notwithstanding these limitations, the most significant phylogeographic result of this study is that all Typoderus weevils currently known west of the Congo basin form a clade (Fig. 3). This, together with the fact that the sister group of Typoderus is strictly Tanzanian in distribution, seemingly suggests an East African origin of Typoderus and single colonization of Cameroon. However, with such little DNA data available for the, likely, many dozens (if not hundreds or perhaps thousands) of other Typoderus species, this cannot be concluded with sufficient confidence.