In Benin, a series of sampling campaigns was organized between 2015 and 2019 in the large Ouémé watershed. A collection of isopod specimens was obtained from a sampling of 169 stations in the Ouémé/ Yéwa basin, 98 of which contained isopods (Fig. 1). Samples were taken in traditional hand-dug, as well as in modern wells lined with casing (as described in BURGÉAP 1981) by means of a modified Cvetkov phreatobiological net (funnel 200 µm mesh size, 150 µm below valve) (Cvetkov 1968; Boutin et al. 2011). For each sampled well, traps baited with a piece of beef were used to collect bottom dwelling animals (Lagnika et al. 2014). Traps were deposited at the bottom of the well for about 24 hours (depending on field constraints). Faunal samples were fixed in ethanol 95% on the spot and were later sorted out at the laboratory of the Zoology Department of the Abomey-Calavi University. Isopod specimens from the same sample were stored in the same vial, preserved in 95% ethanol, and kept in the fridge at 5 °C. Samples were later transferred to the Royal Belgian Institute of Natural Sciences for further processing and were kept at -20 °C in a deep freezer.

Figure 1. 

Location of stations sampled in Benin between 2015 and 2019.

Figure 2. 

Location of the 19 stations in the Ouémé basin (Benin) from which 57 Metastenasellus specimens were selected for DNA taxonomy. The numbers connected to the station symbols correspond to the numbering of the different MOTUs identified in this study.

In Cameroon, a total of about 150 stations were sampled during field campaigns organized during 2009 and 2019 (Fig. 3). The collection of fauna from the wells was done following the same protocol as in Benin (modified phreatobiological net and baited traps). Baited traps were in place for a period of 10 to 18 h, following the recommendations of Boutin and Boulanouar (1983). The stations were distributed in three major watersheds, namely the Wouri (northwestern Coastal rivers basin), the Sanaga and the Nyong (southern Coastal rivers basin). Specimens of the genus Metastenasellus were found at 44 stations (Fig. 4), preserved in 95% ethanol, and stored in a freezer at – 4 °C for further molecular processing.

Figure 3. 

Location of stations sampled in Cameroon between 2009 and 2019.

Figure 4. 

Location of the 44 stations where Metastenasellus specimens were found in Cameroon and studied for DNA taxonomy (white dots). The numbers connected to the station symbols correspond to the numbering of the different MOTUs identified in this study.

Both ASAP or GMYC delineated 23 similar MOTUs that corresponded to singletons or strongly supported clades in the ML tree (BV: 99–100) (Fig. 5). ASAP analyses consistently suggested the same partitioning into 23 different MOTUs, regardless of how the distances were estimated (p-distances, JC69, K80), 11 MOTUs in Benin and 12 MOTUs in Cameroon.

Figure 5. 

Molecular phylogeny constructed using the maximum likelihood method and COI gene fragment of Metastenasellus specimens from Benin and Cameroon. Partitions at the right side of the figure represent the results of the species delimitation analyses with single-locus methods (ASAP, GMYC, bPTP). Numbers at nodes are ultrafast bootstrap values (BV). Nodes were considered as supported if BVs were higher or equal to 90 (Hoang et al. 2018). For the sake of clarity, BVs are not shown within MOTUs delimited by ASAP and GMYC.

mPTP delineated 22 MOTUs, the same as ASAP and GMYC except for MOTUs 13 and 14 which were lumped, albeit separated by long branches on the tree (in red; Fig. 5), and uncorrected pairwise distances as high as 19.2% (Table 2; see below). However, this delimitation was the only one to receive no support (0.2), unlike all the others which received support values ranging from 0.64 to 1.00.

In contrast, bPTP delineated 26 MOTUs. Compared to others approaches, MOTU 1 was split into two singleton MOTUs, and MOTU 12 into two singleton MOTUs (RK_19.88), (RK_19.14) and a duo MOTU (KY623775, RK_19.01). Unlike mPTP, MOTUs 13 and 14 remained distinct.

Uncorrected pairwise distances between specimens ranged between 0.0 and 28.6%. Considering the 23 MOTUs defined according to the results of the species delimitation analyses performed with ASAP and GMYC (Fig. 5), the maximal distances within MOTUs varied between 0.2% (MOTU 9, MOTU 10) and 3.5% (MOTU 12, MOTU 22) while the mean distances between MOTUs varied between 17.1% (M5, M7) and 27.5% (M8, M18; M9, M23; M10, M22; M12, MM18) (Table 2).

The main objective of this study was to provide a first insight into the species diversity within the genus Metastenasellus in Benin and Cameroon. For this, an accurate delimitation of species is not yet required, although it will be desirable in the future. In this respect, the use of a single-locus approach as a first step in a species delimitation is justified, despite its well-known weaknesses (Leliaert et al. 2014). Even if a single locus may not follow the history of the species, due to introgression and incomplete lineage sorting (Puillandre et al. 2021), it nevertheless provides a first overview on the species-level diversity within a group.

Whether ASAP, GMYC or PTP, all these methods provide congruent results in suggesting about 23 highly divergent lineages, once the probably misleading lumping of MOTUs M13 and M14 in the mPTP analysis has been excluded (MOTUs M13 and M14 are separated by p-distances as high as 19.1% and appear to be joined by long branches in the ML tree, making their lumping into one hypothetical species questionable). In contrast, the bPTP approach provided some species hypotheses that are highly unlikely, e.g. splitting MOTUs with p-distances values as low as 2.1% (M1) and 3.5% (M12).

The performance of each method is variable and subject to its own errors, resulting in either oversplitting or overlumping (Dellicour and Flot 2018), and the GMYC approach is known to belong to the first category (Puillandre et al. 2021). In this respect, it is reassuring to see that ASAP, GMYC and, to a lesser extent PTP, all yielded congruent results. This congruence can be interpreted as compelling evidence for the reliability of the outcomes obtained. With regard to the PTP approach, it is interesting to note that mPTP produces results closer to those of ASAP and GMYC than bPTP. This is consistent with the observation that mPTP is superior to PTP in producing delimitations more congruent with taxonomy (Kapli et al. 2017).

Although it is not advisable to consider MOTUs as distinct species on the basis of the mitochondrial COI alone, without at least including a nuclear marker (Dellicour and Flot 2018), several evidences suggest that most, if not all MOTUs identified by the congruence of the different methods used herein corresponds to a valid species.

The first piece of evidence is the particularly high mean p-distances between MOTUs (17.1 to 27.5%). Based on a dataset including a wide taxonomic coverage of Crustacea, Lefébure et al. (2006) have suggested that two monophyletic groups divergent by more than 0.16 substitution per site in the COI gene, as measured by patristic distances, have a strong probability to belong to different species. Patristic distances are defined as the amount of divergence since two taxa shared a common ancestor, i.e., the path-length distance between the two taxa along a phylogenetic tree. Later, Morvan et al. (2013) showed that the threshold method of Lefébure et al. (2006) applied to Aselloidea remained relevant. In this study, mean p-distances between Metastenasellus MOTUs are well above this threshold. This observation makes even more sense given that patristic distances are necessarily higher than p-distances because, unlike the latter, they take account of multiple substitutions. It should also be noted that the separation between the lineages must probably be ancient, as suggested by the particularly large distances between MOTUs, which would make the biases usually associated with the use of a single locus for species delimitation (such as introgression or incomplete sorting of lineages), all the more unlikely. It should be noted, however, that Raupach et al. (2022) have recently shown the existence of surprisingly high genetic divergences in the DNA barcode fragment (COI) within some woodlouse species, which it seems difficult to attribute to the existence of cryptic species, an observation that requires, however, future confirmation with nuclear markers.

Second evidence is the observation that two morphologically distinct species, M. boutini (M15) and M. camerounensis (M23), are separated by p-distances of 24.5%, i.e., a mean interspecific distance of the same order of magnitude as the distances between most other MOTUs (Table 2). An ongoing morphological study also shows that specimens from the same faunistic samples as MOTUs 9, 10 and 11 have clearly distinct male pleopods 2, confirming that these MOTUs correspond to distinct species (Lagnika, pers. comm.). These three MOTUs are separated by p-distances ranging from 20.4 to 24.8% (Table 2).

Third piece of evidence is the coexistence of two MOTUs in the same station, at the same time (BEN072: M6, M7; IBT: M1, M11) indicating that these MOTUs are separately evolving lineages, in other words potential distinct species according to the de Queiroz’s species concept (de Queiroz 2007). However, it is important to keep in mind that divergent mitogenomes found in sympatry are not always associated with divergent nuclear sequences (Martinsson et al. 2020) and may coexist in a single species (Giska et al. 2015).

A high level of species diversity in the genus Metastenasellus in Benin and Cameroon, with about 23 potential distinct species, is all the more remarkable given that, although this is the first significant sampling effort in these two countries, it is still very limited given the geographic area of these countries. However, it is not really a surprise given the recent realisation that the aquatic groundwater environment harbours rich macrobiological diversity, with a high level of endemism and numerous relict species (Gibert et al. 1994; Dole-Olivier et al. 2005; Gibert et al. 2009; Halse et al. 2014; Borko et al. 2021).

Although this study focused on a limited number of Metastenasellus stations, the distribution maps (Figs 2, 4) show a narrow geographical distribution of all MOTUs, suggesting high levels of endemism. The coexistence of two MOTUs in the same station remains the exception, but this may result from the limited sampling effort. In groundwater habitats, exceptional levels of endemism are generally assumed to be caused by strong hydrogeographical isolation, resulting in vicariance, and low dispersal abilities of their inhabitants (Gibert et al. 2009; Iepure et al. 2021). Following an exhaustive analysis of the literature on Stenasellidae in Africa, Pountougnigni et al. (2021) had already observed that the known distribution of this isopod family on this continent was very patchy, with most species known only from their type localities. It is generally accepted that hyporheic habitats of rivers can act as dispersal corridors for subterranean aquatic animals (Stanford and Ward 1993; Malard et al. 2023). In Benin, although the Ouémé river could have played this role, no MOTU from the upper part of the Ouémé basin was observed in the lower part of the basin, which suggests significant faunal isolation.

In Europe, Trontelj et al. (2009) suggested that macro-stygobiotic species showing range sizes over 200 km were most likely an assemblage of cryptic species with much smaller geographic ranges. Our data do not contradict this observation for this region of Africa as well. If so, we can expect future studies to reveal levels of diversity of stygobiotic isopods as high as those documented in other parts of the world. In Western Europe alone, Morvan et al. (2013) identified about 150 species and 12 genera of obligate groundwater Aselloidea (values corrected excluding groundwater species from USA, Japan, Lebanon and Mexico). This observation alone shows the extent to which the diversity of Aselloidea in Africa is probably unknown and remains to be discovered.