Four Louisea species were collected from four different locations in southwestern Cameroon between 2015 and 2021 (Fig. 1). The species were identified by following Cumberlidge (1994, 1999) and Mvogo Ndongo et al. (2019). Eight specimens of L. balssi were collected from 1,958 m a.s.l., Mount Manengouba; 30 specimens of L. edeaensis from 90 m a.s.l., Bedimet Island, Lake Ossa; 50 specimens of L. nkongsamba from 1000–1400 m a.s.l., Mount Nlonako; and 35 specimens of L. yabassi from up to 300 m a.s.l., the Ebo Forest near Yabassi. Specimens of L. nkongsamba and L. yabassi were studied to clarify intraspecific morphological variation within each species. Specimens were measured; their gender and life stage (juvenile, subadult, and adult) recorded; and their habitat preferences noted. Most of the crabs were released into their natural habitat after recording all relevant morphological data. Only a few whole adult specimens (males and females), as well as one of the walking legs was removed from each of the other selected specimens were preserved in ethanol for further morphological descriptions and molecular analyses. The newly collected specimens were deposited either in the Museum für Naturkunde, Berlin, Germany (ZMB), or in the Unity of Taxonomy, Production and Sustainable Management of Aquatic Animals, Department of Management of Aquatic Ecosystems, Institut des Sciences Halieutiques, University of Douala, Cameroon (LABO-PASMAT).

Figure 1. 

Map of Cameroon showing collection sites of Louisea species. Louisea nkongsamba: type locality (green triangle), new localities (green circles); Louisea yabassi: type locality (purple hexagon), new localities (purple circles).

Descriptive morphometrics of L. edeaensis and L. balssi specimens are given in Mvogo Ndongo et al. (2019: tables 2 and 3, respectively). Measurements (in mm) of the carapace of all the specimens were made with digital callipers. Characters of the carapace, thoracic sternum, chelipeds, and mandibles were examined in detail. The terminology used follows Cumberlidge (1999), and the classification by Cumberlidge and Daniels (2022). Images of the body parts were taken using a Leica microscope (model Z16A POA), and the LAS v. 4 and Helicon Focus v. 6.7.1 software. Post processing was undertaken using Adobe Photoshop CC5. Specimens were sorted according to their stage of development into juveniles, subadults, and adults. Furthermore, the maturity of adults was deciphered by identifying specimens that had undergone the pubertal moult from subadult to adult. The pubertal moult was determined by examining the degree of development of the pleon of a series of juvenile, subadult and adult females. The pleon of juvenile females is undeveloped and resembles the slim pleon of juvenile males; the pleon of subadults is significantly widened and partially covers the thoracic sternum. In comparison, the pleon of adult females is conspicuously enlarged and rounded such that its lateral margins overlap the coxae of the pereiopods, and the telson covers thoracic sternites 1 and 2. The lower limit of the range for the pubertal moult was judged as the CW of the largest non-adult female, while the upper limit of the pubertal moult was the CW of the smallest adult female.

Genomic DNA was extracted from a tissue sample of up to 25 mg cut from the pereiopod muscle of 70% ethanol-preserved specimens using the Qiagen DNeasy Blood & Tissue kit following the manufacturer’s instructions. Polymerase chain reaction (PCR) was used to amplify three mitochondrial gene fragments: a ~638 bp region of the 16S ribosomal RNA gene (16S rRNA) using primers 16L29 and 16HLeu (Schubart 2009); a ~594 bp region of the 12S ribosomal RNA gene (12S rRNA) using primers 12L4 and 12H2 (Schubart et al. 2006); and a 648 bp region of the protein-coding mitochondrial gene, cytochrome oxidase subunit I gene (COI) using primers LCO-1490 and HCO-2198 (Folmer et al. 1994). PCR was performed in 25 μl volumes containing 1× Taq buffer, 1.5 mM MgCl₂, 200 μM each dNTP, 1 U Taq polymerase, ~50–100 mg DNA and ddH₂O up to volume. After an initial denaturation step of 4 min at 94 °C, cycling conditions were 35 cycles at 94 °C for 30 s, 45 °C for 60 s, and 72 °C for 90 s, with a final elongation step of 5 min at 72 °C. The same primers were used in PCR and sequencing. PCR products were sent to Macrogen Europe for purification and cycle sequencing of both strands of each gene. The sequences obtained were proofread manually using Chromas Lite (v. 2.1.1) (Technelysium Pty Ltd, Queensland, Australia) and aligned with ClustalW (Thompson et al. 1994) implemented in BioEdit 7.0.5 (Hall 1999). New sequences were submitted to the National Center for Biotechnology Information and are available from GenBank under the accession numbers in Table 1. Results from these genes were concatenated into a single alignment, which was then converted into a Nexus file with FaBox (Villesen 2007).

The COI mitochondrial gene employed here is relatively variable and is commonly used for population genetics, and more recently also for faunal species identification using the barcoding approach (Hebert et al. 2003). This was useful for the examination of the population structure of L. nkongsamba, which provides evidence for genetic substructure among the sampling sites in Nlonako Ecological Reserve. These data are critical for the investigation of the historical connectivity among populations of Louisea species and are useful for the implementation of the future management of genetic diversity.

Maximum parsimony genotype networks (Templeton et al. 1992) were built with the software PopArt (Leigh and Bryant 2015) in order to graphically depict the genetic distances between mitochondrial genotypes. Haplotype and nucleotide diversities were used to compare genetic diversities among the sampling sites in terms of the number of haplotypes and the genetic distances of these haplotypes. Phylogeographic investigations have been successfully used by several researchers to determine connectivity among populations of other endemic crab species, e.g., Sesarma fossarum Schubart, Reimer, Diesel & Türkay, 1997, from the Cockpit Country, Jamaica (see Stemmer and Schubart 2016).

The mitochondrial genes (COI, 12S rRNA, 16S rRNA) were used to identify the species boundaries, to examine the evolutionary origins and the relationships within Louisea species, and to determine whether morphological and ecological similarities between species are based on convergence or common ancestry. Here two methods of phylogenetic inference were applied to the data set: maximum likelihood (ML) using the software PAUP*, and Bayesian inference (BI) as implemented in MrBayes (v. 3.3; Huelsenbeck and Ronquist 2001) (see Mvogo Ndongo et al. 2017b, 2017c; Fratini et al. 2005). The best evolutionary model was determined with jModeltest v. 2.1.7 (Darriba et al. 2012) based on the Akaike Information Criterion (Posada and Buckley 2004) and resulted in the GTR+I+G (COI), GTR+G (16S rRNA) and HKY+G (12S rRNA) models. ML tree was obtained for each alignment with 1000 bootstrap pseudoreplicates. BI was performed to infer phylogeny by using MrBayes v. 3.2.2 (Huelsenbeck and Ronquist 2001). The Markov Chain Monte Carlo was run with four independent chains for 10,000,000 generations, samplefreq = 500, and burnin = 10,001. Analyses were conducted separately to test for topology congruence.

A total of 138 DNA sequences were obtained, 46 sequences each of COI, 16S rRNA, and 12S rRNA (Table 1). ML and BI trees were constructed for individual gene. The relative tree presented here for ML topology has been reconstructed from the concatenation of the three partial loci (COI, 16S rRNA, and 12S rRNA) into a single alignment, which was then converted into a Nexus file with FaBox. This tree includes L. balssi, L. edeaensis, L. nkongsamba, and L. yabassi as the in-group, and Potamonemus man Mvogo Ndongo, von Rintelen & Cumberlidge, 2021a and Buea mundemba Mvogo Ndongo, von Rintelen & Cumberlidge in Mvogo Ndongo, von Rintelen, Tomedi-Tabi & Cumberlidge, 2020 as the out-group species.

To estimate clade divergence times based on the COI gene, a Bayesian analysis with the software BEAST v. 2.6.2 (Bouckaert et al. 2019) was conducted using a strict clock model (Yule Model) with a rate of evolution for the COI of 2.33% per million years (my) (10% SD) (following Schubart et al. 1998). Markov chains for 10 million generations were undertaken, sampling every 1000^th iteration and discarding the first 25% as burn-in. Overall, 7500 trees were obtained, and these trees were used to calculate the maximum clade credibility tree in TreeAnnotator v. 1.6.1 (part of the BEAST package). The uncorrected p-distances (%) was calculated in MEGA 7 (Kumar et al. 2016).

Morphometric measurements of L. yabassi and L. nkongsamba populations are provided in Table 2. The adult size range of L. yabassi, based on male and female specimens from the two populations, was determined to be between CW 16.5 and CW 24.0 mm. Subadults of L. yabassi ranged from CW 11.0 mm to CW 15.5 mm, whereas juveniles of this species were CW 10.0 mm or less. The adult size range of L. nkongsamba, based on male and female specimens from four of the six sites, was between CW 15.8 mm and CW 20.0 mm. Subadults of L. nkongsamba ranged from CW 11.5 mm to CW 14.4 mm (two populations, PAMN 02.12.19 and PAMN 10.12.19, consisted entirely of subadults), whereas juveniles of this species measured CW 10.0 mm or less. No major differences were found between the carapace proportions (CW/FW, CL/FW, and CH/FW) of any of the populations of these two species, and these proportions were virtually identical in all cases (Table 2). The difference between the adult size range of L. yabassi and L. nkongsamba is minor, with the former species growing up to CW 24 mm and the latter species reaching only CW 20 mm.

Differences in certain morphological characters of the specimens of L. yabassi from two populations in the Ebo Forest are noteworthy (Table 3). Like L. yabassi, L. nkongsamba also showed differences in several morphological characters among the specimens from six sites, which are organised here into morphotype 1 (Nlonako Enguegue NO. 1_1462) and morphotype 2 (Nlonako Eyimba, Ngaltongue, Enguegue NO. 2_1382 m, Nguegue) (Table 4). These morphological differences between the two populations/morphotypes of L. yabassi and L. nkongsamba are also illustrated (Figs 2, 3). Despite those morphological differences, there is no genetic support for recognising these differences as indicating different genetic lineages that would warrant formal taxonomic recognition (Figs 4–6).

Figure 2. 

Louisea yabassi from Ebo Forest, Cameroon, adult male (CW 20.2 mm) from site no. 1 (A, C, E, G), adult male (CW 21.3 mm) from site no. 2 (B, D, F, H). Louisea nkongsamba from Nlonako, Cameroon, adult male (CW 18.2 mm) from Eyimba (I, K, M, O), subadult male (CW 12.0 mm) from Enguegue (site no. 1) (J, L, M, P). A, B, I, J dorsal view of cephalothorax C, D, K, L frontal view of cephalothorax E, F, M, N ventral view of thoracic sternum G, H, O, P frontal view of left mandible. Scale bars: 8 mm (A, C, E); 9 mm (B, D, F); 1 mm (G, H); 4 mm (I, K); 12 mm (J, M); 3 mm (L); 8 mm (N); 2 mm (O, P).

Figure 3. 

Louisea nkongsamba from Nlonako, Cameroon, subadult male (CW 12.0 mm) from Enguegue (site no. 1) (A, D, I), adult male (CW 18.2 mm) from Eyimba (B, E, J). Louisea yabassi from Ebo Forest, Cameroon, adult male (CW 21.3 mm) from site no. 2 (C, H), adult male (CW 20.2 mm) from site no. 1 (F, G). A, B, F, H frontal view of chela C, D, E, G cheliped carpus I, J cheliped merus. Scale bars: 5 mm (A–J).

The pubertal moult estimates indicate that the largest Louisea species is L. yabassi (CW 24 mm); the smallest species is L. balssi (CW 16.2 mm); while the size ranges of L. edeaensis and L. nkongsamba overlap with each other (~CW 20 mm) in between those of L. balssi and L. yabassi (Table 2). Louisea balssi is a high-altitude species that dwells at 1958 m a.s.l.; L. nkongsamba is a submontane species found between 938 and 1462 m a.s.l.; while both L. edeaensis and L. yabassi are low-altitude crabs, occurring at 90 m a.s.l. and 300 m a.s.l., respectively (see Mvogo Ndongo et al. 2017a, 2017c, 2019, 2021b; Table 2).

The present molecular analyses support recognition of three lineages (as population 1, 2, and 3) of L. nkongsamba from six sites on Mount Nlonako (Figs 4–6). These distinct lineages, however, do not correlate with the two morphotypes recognised herein for L. nkongsamba (Table 4). Population 1 of L. nkongsamba included specimens that were collected from all six localities of Mount Nlonako (Tables 1, 5; Fig. 4); population 2 of L. nkongsamba comprised specimens that were collected from five out of six sites of Mount Nlonako (Tables 1, 5; Fig. 4); and population 3 of L. nkongsamba comprised specimens that were collected from four out of six sites of Mount Nlonako (Tables 1, 5; Fig. 4). Both morphotypes of L. nkongsamba are represented in one or the other population (Table 1).

Figure 4. 

ML tree topology for Louisea species of Cameroon, derived from mtDNA sequences corresponding to three mtDNA loci (partial 12S rRNA, 16S rRNA, and COI genes). BI and ML statistical values (%) on the nodes indicate posterior probabilities and bootstrap support, respectively.

The uncorrected p-distance between Louisea species pairs reveal that each is well isolated from other taxa assigned to this genus (Table 6). Louisea nkongsamba is sister species to L. yabassi with relatively low p-distance (3.97%) (Table 6); both are sister to L. edeaensis. Louisea balssi is isolated from L. edeaensis, with a sequence divergence of 11.04% (12S rRNA), 10.15% (COI), and 7.77% (16S rRNA) (Table 6); from L. yabassi, with a sequence divergence of 12.94% (12S rRNA), 7.32% (COI), and 5.36% (16S rRNA); and from L. nkongsamba, with a sequence divergence of 12.42% (12S rRNA), 7.98% (COI), and 5.04% (16S rRNA) (Table 6). The uncorrected p-distances between the three genetic populations of L. nkongsamba are given in Table 7. Population 1 of L. nkongsamba is sister to population 2, both populations are sister to population 3.

The phylogenetic analysis indicates that L. balssi from Mount Manengouba is the ancestral species, while L. edeaensis from Lake Ossa is the sister species of the clade that includes L. yabassi and L. nkongsamba (Fig. 4). Divergence time calculations of Louisea species (Fig. 5) showed that the early divergence within the genus occurred during the late Miocene, i.e., L. balssi diverged from other species at about 5.6 myr. Louisea yabassi and L. nkongsamba diverged from L. edeaensis at about 4.1 myr, and L. yabassi separated from L. nkongsamba at about 2.48 myr.

Figure 5. 

BI tree topology for Louisea species of Cameroon, derived from COI mtDNA sequences. Statistical values on the nodes indicate dates in millions of years.

The haplotype network recovered eight haplotypes for L. nkongsamba with maximum four mutation steps between the specimens of this species (Fig. 6) and distinguishes between the four Louisea species (Fig. 6).

Figure 6. 

Maximum parsimony genotype networks for Louisea species of Cameroon, derived from COI mtDNA sequences. Hatch marks stand for mutation steps.

The four Louisea species recovered here each has a monophyletic clade (Fig. 4) with strong topological statistical support, and high pairwise uncorrected p-distance values between species pairs (Table 6). This study, therefore, supports the continued recognition of all four Louisea species that are endemic to the southwest Cameroon rainforests. The earliest divergence for Louisea species happened at ~5.6 myr (late Miocene; Fig. 5), which corresponds to the dates for cladogenesis within genera provided by Daniels et al. (2015). In contrast, the latest Louisea divergence between L. nkongsamba and L. yabassi seems to have occurred during the late Pliocene (2.48 myr). Similar divergence times were recovered for another Central African freshwater crab species pair, Sudanonautes aubryi (H. Milne Edwards, 1853) and S. floweri (De Man, 1901) (see Daniels et al. 2015: fig. 3). Even the two morphologically variable Louisea species, L. nkongsamba and L. yabassi, were found in the molecular analyses to have low uncorrected p-distance values (Table 6), but both are recognised as distinct (see Mvogo Ndongo et al. 2019).

Louisea species are found in different habitats within the rainforest zone: L. balssi in montane forest streams; L. nkongsamba in submontane forest streams; L. edeaensis on the islands of a freshwater lake; and L. yabassi in lowland forest streams. Louisea nkongsamba specimens from cool mountain streams draining the submontane forests of Mt. Nlonako (938–1462 m a.s.l.) are small-bodied with adult males measuring CWs 16–20 mm. Louisea balssi adult males from the cool high-altitude streams (1,958 m a.s.l.) draining into the caldera of Mount Manengouba are also noticeably small-bodied (CWs 13.0–16.2 mm). This agrees with the findings of Daniels et al. (2016) who reported that genetic differentiation tends to be somewhat limited in small-bodied montane species of freshwater crabs. Only a limited genetic variation, however, was found in the lowland forest species, L. edeaensis. In comparison, the moist tropical rainforests surrounding Mount Manengouba receive a high annual rainfall that has maintained a stable forest ecosystem, even during drier periods in the past when rain forests were replaced by savannas in other parts of Central Africa (Brown and Ab’Saber 1979; Diamond and Hamilton 1980; Mayr and O’Hara 1986; Grubb 1992; Zimkus 2009). Consequently, in such high rainfall areas, L. balssi would be sheltered from the harsher effects of rainforest disruption arising from prolonged dry periods in the past, making the Cameroon Highlands a Pleistocene forest refuge for freshwater crab species. Over time, Louisea dispersed from its original location around Mount Manengouba into the surrounding forests of southwest Cameroon, including Mount Nlonako. There L. nkongsamba evolved and continued to disperse into the forested lowlands around Yabassi and Lake Ossa, where L. yabassi and L. edeaensis evolved.

The two L. yabassi populations from localities ~2–3 km apart in the Ebo Forest genetically form a single clade with little lineage differentiation (Fig. 4; populations 1 and 2), and these individuals show relatively low levels of morphological variation (Table 3). Despite this, two L. yabassi morphotypes could be identified (Table 3). Similarly, the six sampled localities around Mount Nlonako, where L. nkongsamba is found, are 4–10 km apart (Tables 2, 5). These individuals of L. nkongsamba fall into three genetically recognisable populations (Fig. 4; populations 1–3), which in turn have two distinct morphotypes (Table 4). Populations 1 and 3 consisted of individuals that all belong to morphotype 1, while population 2 included individuals of both morphotypes (Table 1). The high carapace (CH/FW = 1.3) and narrow front width (CW/FW = 2.9) of both L. yabassi and L. nkongsamba are associated with a semi-terrestrial, air-breathing lifestyle (Cumberlidge 1999). Populations of both species prefer temporary water bodies such as puddles near small permanent streams, as well as damp environments under small stones or in forest floor leaf litter adjacent to streams (Mvogo Ndongo et al. 2017a, 2018, 2019, 2021b). Freshwater crabs have limited dispersal abilities due to the absence of a free-swimming larval phase and their direct development resulting in crab hatchlings; the limited dispersal abilities of the crabs and the restricted movements of the adults in combination with the isolated and fragmentary nature of their wetland habitats might be at least partly responsible for their rich diversity and high endemism (Cumberlidge et al. 2009; Mvogo Ndongo et al. 2021b). The intraspecific morphological and genetic variations observed within L. yabassi and L. nkongsamba are crucial for adaptation by natural selection, not least because low levels of variation are associated with the extirpation of populations and an increased risk of species extinction (Bolnick et al. 2011; Scheiner and Holt 2012; Forsman 2014).