Introduction

Palaeogeographic and palaeoclimatic events are regarded as important factors that, through the fragmentation of distributional range of species, promote the allopatric isolation and subsequently the genetic divergence among populations (see Avise, 1994; Hewitt, 1996). During the last 10 million years (Ma), North Africa and the Mediterranean area have gone through a series of tectonic and climatic events that favoured intraspecific divergence, leading ultimately to speciation (see Kornilios et al., 2010 and references therein). In this context, terrestrial and freshwater reptiles, due to the limited dispersal capacities and temperature dependence, represent sensitive indicators of palaeoclimatic and palaeogeographic phenomena and the study of their genomic markers substantially contributes to the understanding of biogeographical processes (Lenk et al., 1999).

Map showing Chalcides mertensi distributional range (in grey), as reported in IUCN Red List of Threatened Species (Miras et al., 2006). Sampling locations are indicated by solid circles and capital letters. ASO: Ain Soltane; KAS: Kasserine; SIB: Sidi Bouzid; TAB: Tabarka.

Citation: Amphibia-Reptilia 34, 3 (2013) ; 10.1163/15685381-00002901

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Map showing Chalcides mertensi distributional range (in grey), as reported in IUCN Red List of Threatened Species (Miras et al., 2006). Sampling locations are indicated by solid circles and capital letters. ASO: Ain Soltane; KAS: Kasserine; SIB: Sidi Bouzid; TAB: Tabarka.

Citation: Amphibia-Reptilia 34, 3 (2013) ; 10.1163/15685381-00002901

• Download Figure
• Download figure as PowerPoint slide

Map showing Chalcides mertensi distributional range (in grey), as reported in IUCN Red List of Threatened Species (Miras et al., 2006). Sampling locations are indicated by solid circles and capital letters. ASO: Ain Soltane; KAS: Kasserine; SIB: Sidi Bouzid; TAB: Tabarka.

Citation: Amphibia-Reptilia 34, 3 (2013) ; 10.1163/15685381-00002901

• Download Figure
• Download figure as PowerPoint slide

Chalcides is a genus of scincid lizards comprising 30 currently accepted species distributed in the Macaronesian and Mediterranean areas and from central and eastern Africa to Pakistan, with the majority of species diversity in Morocco and surrounding areas (see Carranza et al., 2008). The species of this group are viviparous and differ in the degree of limb reduction and body elongation (Caputo, Lanza and Palmieri, 1995; Greer et al., 1998).

The Algerian three-toed skink (Chalcides mertensi Klausewitz, 1954) belongs to the “grass-swimming” clade of Chalcides (sensu Carranza et al., 2008). This clade is almost exclusively comprised of elongated forms living in mesic habitats (with the exception of C. mauritanicus), and includes seven species (Carranza et al., 2008). The diversification of “grass swimming” Chalcides began around 9.9 Ma ago, soon after the initial break-up of this genus (Carranza et al., 2008). Chalcides mertensi is distributed in northern Algeria and western Tunisia (Sindaco and Jeremčenko, 2008) (fig. 1).

The aim of this paper is to evaluate the extent of genetic variability in Tunisian populations of C. mertensi to test the influence of palaeoclimatic and palaeogeographic history of North Africa on this species. This was done by means of the Restriction Fragment Length Polymorphism (RFLP) analysis of two fragments of mitochondrial DNA (mtDNA) comprising NADH dehydrogenase subunits 1 and 2 (ND-1/2) and NADH dehydrogenase subunits 3, 4 and 4L (ND-3/4), and by sequencing segments of the mitochondrial genes encoding cytochrome b (Cyt b) and 12S rRNA (12S).

Materials and methods

A total of 56 individuals of Chalcides mertensi were collected in the following sampling locations of Tunisia: Tabarka (20), Ain Soltane (21), Kasserine (13) and Sidi Bouzid (2) (fig. 1). Tail tips were removed and preserved in 99% ethanol until DNA extraction. Total genomic DNA was extracted from ethanol preserved tissues following a standard phenol-chloroform protocol (Sambrook, Fritsch and Maniatis, 1989).

For the study of the genetic variability of C. mertensi populations, RFLP analysis was carried out on 56 individuals. Two segments of mtDNA were PCR-amplified: a segment of approximately 2.7 kb encompassing NADH dehydrogenase subunits 1 and 2 (ND-1/2) was amplified using primers L3827 and H6313 (Sorenson et al., 1999); a segment of about 2.4 kb encompassing NADH dehydrogenase subunits 3, 4L and 4 (ND-3/4) was amplified with primers ND 3/4 F and ND 3/4 R designed by Nielsen, Hansen and Mensberg (1998). Both ND-1/2 and ND-3/4 were amplified with the same PCR protocol which included an initial denaturation step at 95°C for 5 min, followed by 35 cycles of 94°C/45 s denaturation, 52°C/1 min annealing and 72°C/1.5 min extension, then a final extension at 72°C for 10 min. The amplified fragments were then digested with restriction endonucleases. Nine endonucleases were used for ND-1/2 (AluI, ApaI, AvaII, DraI, HinfI, HpaII, RsaI, ScaI, TaqI) and seven for ND-3/4 (AvaII, DraI, HinfI, HpaII, RsaI, ScaI, TaqI). Restriction digestions were carried out in 20 μl volumes using 3-5 μl of PCR product, 10 Units of the enzymes and buffers according to the manufacturer’s instructions (Fermentas). Restriction fragments were electrophoresed on 2% agarose gels (Bio-Rad Laboratories) using ethidium bromide staining, alongside a 100 base pairs (bp) ladder and visualized and photographed on an ultraviolet transilluminator. Fragments shorter than 100 bp were not generally detectable and some hypothetical fragments were assumed in order to explain all the mutational steps (see Jaarola and Tegelstrom, 1996).

Restriction patterns generated by each enzyme were designated by capital letters which were then combined to define composite mtDNA haplotypes. These patterns were subsequently used to infer the presence or absence of restriction sites in the different haplotypes. Restriction site data were analyzed with REAP 4.0 (McElroy et al., 1991) to calculate, with the program DA, haplotype diversity (h) and nucleotide diversity (π) of the sampled populations. The population genetic structure was determined by estimating the molecular variance by AMOVA analysis (Excoffier, Smouse and Quattro, 1992) using ARLEQUIN 2.000 (Schneider, Roessli and Excoffier, 2000). The analysis is performed at three hierarchical levels: among groups, among populations within groups and within populations. AMOVA analysis was carried out based on the pair-wise differences between haplotypes using 1000 permutations. The statistical significance of haplotype frequency differentiation among populations was tested by Monte Carlo simulation with 10 000 replications, as described by Roff and Bentzen (1989), using the program MONTE of the REAP package.

Table 1.

Taxa and corresponding sequence haplotypes (12S + Cyt b) used in the phylogenetic analyses. Sampling location, GenBank accession numbers and reference are reported for each haplotype.

View Table

For the phylogenetic analysis, two fragments of mtDNA were amplified and sequenced from ten individuals of northern and eight of southern populations. The individuals chosen were representative of the RFLP haplotypes. A 450 bp segment of the mtDNA including the 5′ end of the Cyt b gene was PCR-amplified using the universal primers L14724 (Palumbi, 1996) and H15149 (Kocher et al., 1989), and a segment of 450 bp encompassing a portion of the 12S gene was amplified using the universal primer L1091 and H1478 (Kocher et al., 1989). For both genes, the PCR profile included an initial denaturation step at 95°C for 2 min, followed by 35 cycles of 1 min 95°C/1 min denaturation, 52°C/45 s annealing and 72°C/1 min extension, then a step of final elongation at 72°C for 7 min. PCR products were sequenced on an ABI PRISM 3730XL (Applied Biosystems) automatic sequencer, using the above mentioned primers. In order to infer the phylogenetic relationships of the C. mertensi haplotypes detected in the present study with taxa of the “grass swimming clade” (sensu Carranza et al., 2008), haplotype sequences (12S and Cyt b) of C. chalcides chalcides, C. chalcides vittatus, C. guentheri, C. mauritanicus, C. mertensi, C. minutus, C. parallelus, C. pseudostriatus, C. striatus, C. lanzai and C. parallelus (Carranza et al., 2008) (table 1), were retrieved from GenBank, combined and aligned using Clustal W (Larkin et al., 2007) with default parameters. The combined data set (12S + Cyt b) from 25 haplotypes of the above species was used for the phylogenetic analyses that were carried with Maximum Likelihood (ML) and Bayesian Inference (BI) methods. The best fit models of nucleotide substitution for each gene and for the combined dataset were selected among the 88 models available in jModeltest 2.1.3 (Darriba et al., 2012) using the Akaike Information Criterion (AIC). The most appropriate model was GTR + I + G for the combined data set, TPM2uf + G for 12S and HKY + I for Cyt b. ML analysis was carried out using PhyML 3.0 (Guindon et al., 2010) using the NNI method, with the model parameters fitted to the data by likelihood maximization. Statistical reliability of the ML trees was assessed by 1000 bootstrap replicates (Felsenstein, 1985). Bayesian inference was carried out on the combined data set with MrBayes v3.2 (Ronquist et al., 2012). The data set was partitioned into the two genes and the appropriate model of nucleotide substitution was used for each of them, and the respective parameters re-estimated. For 12S, because the model TPM2uf + G is not implemented in MrBayes, a similar model (GTR + G + I) was used. The BI analysis was run with four incrementally heated Markov chains for 2 × 10⁶ generations in two independent runs with samplings at intervals of 100 generations that produced 20 000 trees. Once the stationarity had been reached, both in terms of likelihood scores and parameter estimation, the first 5 × 10³ trees (25% ‘burn-in’) were discarded in both runs and a majority-rule consensus tree was generated from the 15 000 remaining (post burn-in) trees. The posterior probability (pp) was calculated as the percentage of samples recovering any particular clade (Huelsenbeck and Ronquist, 2001), with pp ≥ 0.95 indicating a statistically significant support (Wilcox et al., 2002). Both ML and BI trees were displayed with FigTree v1.4.0 (http://tree.bio.ed.ac.uk/software/figtree/).

The divergence time between lineages was estimated using BEAST v1.7.5 (Drummond et al., 2012) with the combined data set (12S + Cyt b). The input file for BEAST was generated with BEAUti v1.7.5 (Drummond et al., 2012). Groups of taxa were defined on the basis of the phylogenetic trees obtained to calculate their divergence time. The data set was partitioned and the most appropriate substitution model was used for each partition: HKY for Cyt b, and GTR for 12S. In order to estimate dates of divergence, a substitution rate of 1.35%, previously estimated for this data set in skinks by Carranza et al. (2008) was fixed and the strict clock option was chosen for the molecular clock model. MCMC chains were run in BEAST for 25 × 10⁶ generations. The output of BEAST was explored with TRACER v1.5 (Rambaut and Drummond, 2009).

The nucleotide divergence in pair-wise comparisons of haplotypes was calculated using Tamura-Nei (Tamura and Nei, 1993) method as implemented in MEGA version 5.1 (Tamura et al., 2011). MEGA was also used to estimate nucleotide composition and number of substitutions of the sequences analysed.

Results

PCR-RFLP analysis of the ND-1/2 and ND-3/4 regions of 56 C. mertensi individuals from 4 different sampling locations revealed 4 composite haplotypes. Ain Soltane: H1 (21), Tabarka: H1 (19), H2 (1); Kasserine: H3 (1), H4 (12); Sidi Bouzid: H4 (2). In C. mertensi, the 16 endonucleases recognized an average of 53 restriction sites per composite haplotype, corresponding to an average of 228 bp.

The populations from Ain Soltane and Sidi Bouzid were fixed for haplotypes H1 and H4, respectively. Tabarka (H1, H2) and Kasserine (H3, H4), characterized by the presence of two haplotypes each, showed values of h equal to 0.2468 ± 0.1075 and 0.3429 ± 0.1278, respectively. Nucleotide diversity was 0.0030 in Tabarka and 0.0021 in Kasserine.

The genetic structure of populations, inferred by AMOVA analysis, was carried out on 4 populations and two groups of populations: Ain Soltane and Tabarka (northern Tunisia) vs Kasserine and Sidi Bouzid (southern Tunisia). The groups were formed on the basis of the clusters of haplotypes recovered by the phylogenetic analysis (see below). The test, conducted with the pair-wise difference method, showed that the majority of total molecular variance (99.32%, ${\mathrm{\Phi }}_{\mathrm{CT}}=0.99319$, $P<0.001$) was distributed among groups suggesting a pronounced genetic separation between populations from northern and southern Tunisia. The other components of the total molecular variance were very low, with percentages of −0.03% (${\mathrm{\Phi }}_{\mathrm{SC}}=-0.03911$, non-significant) and 0.71% (${\mathrm{\Phi }}_{\mathrm{ST}}=0.99293$, $P<0.001$) for the comparisons among populations within groups and within populations, respectively. These results can be explained by the fact that populations and groups of populations are quite homogenous while the two groups do not share any haplotypes.

The extent of heterogeneity in the frequency distribution of haplotypes among the four sampled populations of C. mertensi, tested by Monte Carlo simulation, showed significant heterogeneity ($P<0.001$) in the pair-wise comparisons involving populations that did not share any haplotypes: Ain Soltane vs Kasserine ($P<0.001$), Ain Soltane vs Sidi Bouzid ($P<0.001$), Kasserine vs Tabarka ($P<0.001$), Sidi Bouzid vs Tabarka ($P<0.01$). Non-significant heterogeneity was recorded in the pair-wise comparisons involving populations sharing common haplotypes: Ain Soltane vs Tabarka and Kasserine vs Sidi Bouzid.

The length of the 18 C. mertensi sequences (individuals representing the 4 RFLP composite haplotypes) here produced was 396 bp for Cyt b, and 386 bp for 12S. The alignment revealed three C. mertensi sequence haplotypes: Ain Soltane (CME1 = H1), Tabarka (CME1 = H1, CME2 = H2), Kasserine (CME3 = H3,4), Sidi Bouzid (CME3 = H3,4). Each of these three haplotypes is distinguished from the others for both 12S and Cyt b sequences. The sequences of 12S and Cyt b of CME1 haplotype are identical to the sequences of the three individuals of C. mertensi from Ain Soltane analysed by Carranza et al. (2008).

Maximum Likelihood tree depicting the phylogenetic relationships among the 25 sequence haplotypes based on the combined data set 12S + Cyt b. Bootstrap values higher than 50 and pp values higher than 0.95 are indicated at nodes. Below the branch, left of the node, the age of the most recent ancestor for some of the haplotype lineages is indicated. The date is reported as a mean and expressed in millions of years (Ma), followed by standard error, and 95% Highest Posterior Density (HPD) intervals in parentheses. Haplotype codes as in table 1.

Citation: Amphibia-Reptilia 34, 3 (2013) ; 10.1163/15685381-00002901

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Maximum Likelihood tree depicting the phylogenetic relationships among the 25 sequence haplotypes based on the combined data set 12S + Cyt b. Bootstrap values higher than 50 and pp values higher than 0.95 are indicated at nodes. Below the branch, left of the node, the age of the most recent ancestor for some of the haplotype lineages is indicated. The date is reported as a mean and expressed in millions of years (Ma), followed by standard error, and 95% Highest Posterior Density (HPD) intervals in parentheses. Haplotype codes as in table 1.

Citation: Amphibia-Reptilia 34, 3 (2013) ; 10.1163/15685381-00002901

• Download Figure
• Download figure as PowerPoint slide

Maximum Likelihood tree depicting the phylogenetic relationships among the 25 sequence haplotypes based on the combined data set 12S + Cyt b. Bootstrap values higher than 50 and pp values higher than 0.95 are indicated at nodes. Below the branch, left of the node, the age of the most recent ancestor for some of the haplotype lineages is indicated. The date is reported as a mean and expressed in millions of years (Ma), followed by standard error, and 95% Highest Posterior Density (HPD) intervals in parentheses. Haplotype codes as in table 1.

Citation: Amphibia-Reptilia 34, 3 (2013) ; 10.1163/15685381-00002901

• Download Figure
• Download figure as PowerPoint slide

The alignment of 783 bp (387 from 12S and 396 from Cyt b) from 25 haplotypes revealed: i) 77 variable sites and one indel in the 12S; 131 substitutions with 16 amino acid changes in the Cyt b sequence. The average nucleotide composition was A: 33.5; C: 26.7; G: 20.5; T: 19.3 for 12S, and A: 27.1; C: 29.9; G: 15.5; T: 27.5. The transition/transversion ratio was 4.64 for 12S and 4.62 for Cyt b.

For the phylogentic analyses, a combined data set comprising 12S and Cyt b sequences from 25 sequence haplotypes (22 from Carranza et al., 2008; 3 from the present study) were used (table 1), for a total of 783 characters aligned. The analyses with ML and BI methods of this data set produced very similar topologies, and, for this reason, only one tree is presented in fig. 2. Both methods clearly supported the polyphyly of the individuals here analysed, currently referred to as C. mertensi. In fact, the two haplotypes (CME1 and CME2) from northern Tunisia are included in a monophyletic clade (pp = 1.0; bootstrap = 99) in which two subclades could be recognised: i) one formed by two C. minutus (CMI2, CMI3) haplotypes; ii) and the other including northern individuals of C. mertensi (CME1, CME2). This relationship between CMI2 and CMI3 and C. mertensi from northern Tunisia had been already shown by Carranza et al. (2008). As for the haplotype representative of southern Tunisia (CME3), the phylogenetic analyses performed here, recovered it as a sister taxon of C. chalcides (CCC + CCV). However, the monophyly of the clade (CCC + CCV + CME3) is supported only by ML analysis (pp = 0.84; bootstrap = 93) (fig. 2).

The estimates of divergence times between lineages indicated that the split between the lineages that led to the northern and southern populations of Tunisian C. mertensi occurred approximately 4.1 Ma ago (Early Pliocene). Southern populations diverged from the lineage leading to C. chalcides approximately 3.2 Ma ago, while northern C. mertensi and C. minutus diverged from their common ancestor approximately 2.1 Ma ago (fig. 2).

The nucleotide divergence in pair-wise comparisons of sequences of “grass swimming” Chalcides, calculated under the Tamura-Nei (Tamura and Nei, 1993) model of evolution, ranged from 0.1% to 9.9% (table 2).

Table 2.

Estimates of genetic distances between haplotypes of “grass swimming” Chalcides. Genetic distances (expressed as percentage) are shown below the diagonal, and standard error estimates are shown above the diagonal. Analyses were conducted using the Tamura-Nei model (Tamura and Nei, 1993). All positions containing gaps and missing data were eliminated.

View Table

Discussion

The Moroccan region probably represents the centre of origin of the genus Chalcides. In fact, 16 out of the 30 currently recognized species in this genus and all its main lineages occur in this area (Caputo, 2004; Carranza et al., 2008). This high degree of diversification is probably related to the complex palaeogeographic and palaeoclimatic history of this region. The radiation of this genus is mainly related to the orogenic folding that occurred over the past 20 Ma and produced the Atlas and Rif Mountains and led to the fragmentation of the areas of light soil and vegetation where Chalcides often occurs, giving opportunity for diversification (see Brown, Suárez and Pestano, 2002; Carranza et al., 2008).

The results of the present study indicate that the populations of C. mertensi investigated represent two distinct evolutionary lineages. This conclusion is supported by phylogenetic analyses that clearly indicated the polyphyly of the C. mertensi individuals investigated, with southern populations showing a sister taxon relationship with C. chalcides, and the northern Tunisia individuals being the sister taxon of C. minutus (individuals from middle Atlas), as also demonstrated by Carranza et al. (2008) for three individuals of C. mertensi from northern Tunisia (Ain Soltane). In addition, the genetic distances calculated in pair-wise comparisons of sequences showed that values between haplotypes from southern and northern Tunisia are of the same order of magnitude as those recorded in comparisons with other taxa regarded as full species, or even higher (table 2). In addition, AMOVA analysis and Monte Carlo simulation further supported the pronounced genetic divergence between populations form northern and southern Tunisia. In this context, it seems to be clear that these populations, so far assigned to a single species (C. mertensi), should be split into two different evolutionary units. The name C. mertensi should be retained for the southern populations. Indeed, the type locality of C. mertensi (Biskra, eastern Algeria, see Klausewitz, 1954) is geographically closer to the southern populations than to the northern ones. In addition, Biskra, Kasserine and Sidi Bouzid, lie on the southern side of the Aurès mountains (the easternmost portion of the Atlas mountain range). According to this hypothesis, another specific name should be assigned to the northern Tunisia populations. However, it cannot be excluded that both northern and southern Tunisia populations here investigated may represent two new taxonomical entities (both different from C. mertensi). This taxonomic issue will be sorted out only by analysing specimens from Algeria, including also individuals representing typical C. mertensi, and re-evaluating all the Maghreb populations attributed to the “grass swimming” clade of Chalcides.

The estimate of divergence times suggests a split between C. mertensi populations from northern and southern Tunisia, that can be dated back to approximately 4.1 Ma ago. This estimation places the diversification of these two evolutionary lineages within a period (early Pliocene) that witnessed the diversification of most Chalcides taxa (see Carranza et al., 2008; Kornilios et al., 2010), and characterized by several simultaneous diversification events in both reptiles and amphibians (see Carranza et al., 2008 and references therein). The high degree of diversification found in reptiles and amphibians of northwestern Africa in this period may have a plausible explanation in the palaeogeography of this area between the end of Messinian salinity crisis (5.33 Ma ago) and Pliocene (approximately 3 Ma ago), when the catastrophic refilling of the Mediterranean basin (approximately 5.33 Ma ago) led to the formation of smaller or larger islands (“fossil islands”) (Steininger and Rögl, 1984), thus causing the long lasting isolation of faunas throughout Morocco, Algeria and Tunisia, that could explain the phylogeographic patterns recorded in numerous reptiles (Kornilios et al., 2010). The fragmentation of the distributional range of northern African reptiles could be also the result of climatic oscillations that affected this area during the last 10 Ma. These oscillations produced repeated changes in habitat, from heavily vegetated land to desert and back again (Prentice and Jolly, 2000; Douady et al., 2003; Schuster et al., 2006; Carranza et al., 2008), that might have provoked the fragmentation of the habitat of “grass swimming” Chalcides.

Another aspect that it is worth being discussed is the variability, in terms of haplotype and nucleotide diversity, of the two evolutionary lineages here investigated, and currently ascribed to C. mertensi. The values of these parameters, ranging from 0.0000 to 0.3429 for haplotype diversity (average 0.1966), and from 0.0000 to 0.0030 (average 0.0017) for nucleotide diversity, are low if compared to the values recorded for Tunisian populations of C. chalcides by Giovannotti et al. (2007), which exhibited values of haplotype diversity between 0.3680 and 0.8165 (average 0.6278), and nucleotide diversity between 0.0019 and 0.0063 (average 0.0044). This difference cannot be explained on the basis of ecological characteristics of these taxa that are very similar (Schleich, Kästle and Kabisch, 1995). On the contrary, a possible explanation could be related to the position of these populations within the range of each taxa. In fact, it is likely that the samples investigated in the present study are located at the eastern margin of the geographical range of the respective taxonomic units. As for C. chalcides, Tunisia can be regarded as the centre of origin and diversity for this species, with Peninsular Italy, Sicily and Sardinia being colonized from North Africa in more recent times (Caputo, 1993; Giovannotti et al., 2007). The samples here investigated could represent populations occupying a very peripheral position in the range of the respective species, with southern Tunisia samples probably representing peripheral populations of C. mertensi. The northern samples could represent peripheral populations of another lineage, mainly distributed westward. This hypothesis seems to be supported by the close relatedness of northern populations to C. minutus from the middle Atlas. Indeed, a growing body of evidence indicates a decrease in within-population genetic diversity in peripheral populations of both plants and animals (Eckert, Samis and Lougheed, 2008). Another explanation could be represented by palaeoclimatic changes during the last 10 Ma that might have had drastic effects on the genetic variability of these taxonomic entities through the shrinking of suitable habitats (moist grasslands) within the range of this species. Indeed, it is well known that North Africa has been subject to enormous landscape changes for both natural reasons (climate warming) (Carrington, Gallimore and Kutzbach, 2001) and human activities during Holocene (Charco, 1999). Presently, it is estimated that only 4.7% of the original Mediterranean forests survive (Cuttelod et al., 2008). Past climatic events would then have produced a strong reduction of the genetic variability in this “grass-swimming” lizard. These effects on the genetic variability could have been more drastic on the populations located at the margin of the distributional range of the species, than on those of C. chalcides, that has its centre of origin and diversification in Tunisia. The low genetic variability observed in both northern and southern Tunisian populations could be related to the colonization process that took place with the climatic improvement, and that probably implied successive bottlenecks, according to a leptokurtic model of expansion (see Hewitt, 1996, 2000).

In conclusion, the findings here presented of a new evolutionary unit in the grass swimming Chalcides shows once more that northwestern Africa represents an hot spot for the biodiversity of this genus, and the crucial role of palaeogeographic and palaeoclimatic events for the diversification of reptiles in this area. However, in order to evaluate conclusively the taxonomic status of these two entities and clarify the question of their low genetic variability, a more exhaustive analysis aimed at exploring the biodiversity of skinks in the area extending to the West (including Algeria, and Morocco up to the middle Atlas) of the populations here investigated will be necessary.