Phylogeographic diversification of the Mesalina olivieri species complex (Squamata: Lacertidae) with the description of a new species and a new subspecies endemic from North West Africa

2.2.1 DNA extraction and amplification

Total genomic DNA was extracted from ethanol-preserved tissue using a proteinase K (10 mg/ml) digestion followed by a standard salt-extraction protocol. Amplifications were performed in 5 μl of 2× MyTaq Mix and 0.4 μM of each primer. The PCR conditions adopted for every primer pair are specified in Table 2. Some samples required minor adjustments to the temperature and time of annealing from the reported conditions. We amplified one fragment from the mitochondrial cytochrome b gene (Cyt-b, 400 bp, Kapli et al., 2015) and four nuclear DNA (nucDNA) gene fragments from the beta fibrinogen intron 7 (β-fib7, 600 bp), melanocortin receptor one gene (MC1R, 630 bp), ornithine decarboxylase gene (OD, 467 bp), and phosphogluconase dehydrogenase intron 7 (PgD7, 414 bp). These markers were selected because they were found to be informative in previous studies (Kapli et al., 2015; Simó-Riudalbas et al., 2019; Sindaco et al., 2018) or during our preliminary analyses. PCR products were cleaned using ExoSAP. Purification, and sequencing were outsourced to GENEWIZ, Leipzig, Germany. Amplified fragments were sequenced for the forward strand only (primers sequences are listed in Table 2). Obtained DNA sequences were checked for errors using Codon-Code Aligner (v. 2.0.6, Codon-Code Corporation). Heterozygote positions of nuclear sequences were coded using IUPAC ambiguity codes. The absence of stop codons was checked with MEGAX (Kumar et al., 2018). DNA sequences were aligned using MAFFT v.7 (Katoh et al., 2019) applying the default parameters (Auto strategy, Gap opening penalty: 1.53, Offset value: 0.0). Homozygous indels were coded with dashes to maintain the overall alignment. The length of each indel was reduced to the minimum to maintain the number of variable positions. All sequences that were newly produced for the present study were deposited in GenBank (Accession numbers from MZ223473 to MZ223857; Table S1).

A total of 80, 72, 74, 71, and 68 new samples were successfully sequenced for Cyt-b (Alignment S1), β-fib7 (Alignment S2), MC1R (Alignment S3), PgD7 (Alignment S4), and OD (Alignment S5), respectively (Table S1). Three datasets were produced: (a) Dataset S1: a mitochondrial dataset (Cyt-b; 285 bp), including outgroup (listed below) and all sequences available from the most recent publications (Kapli et al., 2015; Simó-Riudalbas et al., 2019; Sindaco et al., 2018; Šmíd et al., 2017) for a total of 378 sequences; (b) Dataset S2: a concatenated multilocus nuclear dataset (1700 bp) with a total of 78 sequences (including only samples with at least three genes) of the M. olivieri species complex, a few representative specimens of the other species of the genus and one specimen of Acanthodactylus erythrurus as outgroup; (c) Dataset S3: a concatenated mt + nucDNA dataset with a total of 86 concatenated sequences (1986 bp) of the M. olivieri species complex (including only samples with at least three genes) plus four species used as outgroup: Gallotia atlantica, Psammodromus algirus, Podarcis pityusensis, and Podarcis lilfordi, this dataset was used to calculate the phylogeny of the species complex and the time of divergence. Specific details of each database, as well as their implementations, are resumed in Table S2.

2.2.2 Phylogenetic analyses

For all datasets, Bayesian inferences (BI) were performed with BEAST 1.10.4 (Suchard et al., 2018). The best-fitting model of nucleotide substitutions for each gene was determined with PartitionFinder v.1.1.1 (Lanfear et al., 2012). The program was set as follows: branch lengths unlinked, only models available in BEAST were evaluated, BIC model selection criterion applied, and all partition schemes analyzed. Each gene was tested independently. The partition scheme and models of sequence evolution selected were β-fib7: HKY+G; MC1R: HKY+I; OD: HKY; PgD7: K80+G; and Cyt-b: HKY+G+I. No partition by codon position was selected for any of the genes.

To test whether the genes studied evolve in a clock-like manner (strict clock) a preliminary BI analyses was run in BEAST 1.10.4 (Suchard et al., 2018) using a relaxed clock for all the markers. The results of this preliminary run were verified using TRACER 1.6. (http://tree.bio.ed.ac.uk/software/tracer). The strict-clock model was rejected when the standard deviation of the uncorrelated lognormal relaxed clock parameter (ucld.stdev) and the coefficient of variation were greater than one. We used a Speciation Yule Process model assuming a constant lineage birth rate for each branch in the tree. Three independent MCMC runs of 100 million generations were implemented for each analysis for each dataset, sampling every 10,000 generations, and 10% of the trees were discarded as burn-in.

To root and calibrate the mt + nucDNA tree, four species were used as outgroup: G. atlantica, P. algirus, P. pityusensis, and P. lilfordi. The calibration points (in millions of years ago) and priors applied to the divergence time estimation correspond to those used Simó-Riudalbas et al. (2019): (a) the split between G. atlantica and P. algirus (age of the of the Canary Islands Fuerteventura and Lanzarote; normal distribution, mean 18, SD 2); (b) the split between P. pityusensis and P. lilfordi (end of the Messinian Salinity Crisis; normal distribution, mean 5.32, SD 0.05).

2.2.3 Haplotype networks reconstruction and genetic distances

Haplotype networks were reconstructed for each nuclear marker. Nuclear sequences were initially phased per lineage using the PHASE algorithm, implemented in DNASP 5.10.01 (Librado & Rozas, 2009). The algorithm was run five times, for 10,000 iterations, with a burn-in of 1000. The most probable reconstructed haplotypes were used to create the haplotype networks. Haplotype networks were then reconstructed in TCS (Clement et al., 2000) with a 95% parsimony threshold for the nuclear genes: β-fib7 (186 sequences), MC1R (152 sequences), OD (120 sequences), and PgD7 (124 sequences), and then displayed in tcsBU (Múrias dos Santos et al., 2016).

Computation of sequence divergence (uncorrected p-distances) for the Cyt-b fragment was performed in MEGAX (Kumar et al., 2018) to provide an overview of the genetic divergence among taxa. The grouping referred to the topology of the Cyt-b tree built using a database containing the sequences from this study and those published by Kapli et al. (2015).

2.3.1 Morphological analyses

For the morphological data, we examined voucher specimens housed in the collection of the Biogeography and Ecology of Vertebrates (BEV) in the CEFE lab in Montpellier. A total of 32 morphological variables were measured in 252 specimens: 138 of M. olivieri (including eleven specimens from the AS/AM linages and 18 from the ADR lineage), 37 of M. pasteuri (including the TAG lineage) and 21 of M. simoni (samples localities are given in Table 1 and Table S1). Variables were selected according to their relevance as diagnostic characters of the genus Mesalina in previous sources (e.g., Hosseinian Yousefkhani et al., 2015; Trape et al., 2012) or based on our own examination of specimens of the various lineages. We scored each individual for: (a) five quantitative biometric variables; (b) 11 quantitative pholidotic variables; (c) 10 semi-quantitative (ordinal) chromatic variables; (d) four semi-quantitative (ordinal) variables describing morphological states (Table 3). Scale nomenclature, scale counts, and measurements follow Bons and Geniez (1995). All morphological data were obtained by the same observer (CP).

To identify diagnostic characters and to quantify the amount of morphological differentiation between the ADR and TAG lineages and the other species of the olivieri complex, we conducted multivariate analyses over three morphological datasets: biometry (Table S3), pholidosis (Table S4), and coloration (Table S5). Multivariate analyses were restricted to adult specimens to reduce variation due to the strong ontogenetic modifications in color patterns in the genus Mesalina. Both sexes were treated separately, as preliminary analyses revealed significant sexual dimorphism for many variables in every clade (results not shown). Due to the small sample size, statistical tests were not conducted. An exploratory investigation was first performed with Excel using a heating map to facilitate an immediate visualization of the most relevant interspecific differences. Then, Principal Component Analyses (PCAs) were run on each database for a pairwise comparison of the potential new species with the other species of the complex. The data were normalized to zero mean and unit variance prior to PCAs. All analyses were run on PAST 3.24 Software (Hammer et al., 2001).

2.3.2 Distribution modeling and habitat comparison

An ecological niche model was performed to assess the potential distribution and characterize the realized ecological niche of the ADR and TAG lineages. For this analysis, we aimed to include all the known localities of these two lineages. To do so, we considered an area of approximately 1,979,127 km² (varying between 28°N, 17°W to 14°N, 4°W) that includes southern Morocco, South-western Algeria, the full extent of Mauritania, South-western Mali, and North-eastern Senegal, comprising the whole distribution of the potential new species (known to date). Models were based on the 28 presence records confirmed by genetic or morphological assignment. Morphological identifications of unsequenced specimens were based on pictures of live unvouchered animals or from the direct examination of specimens deposited in museum collections. Only records based on adult specimens exhibiting distinctive characteristics of the ADR lineage were treated as valid. To remove duplicated observations from the same geographic locations, data were thinned reducing to 20 the number of observations to build models. Models were built using a spatial resolution of 1 km.

Variables used for the modeling were: (a) Terrain Roughness Index (TRI) calculated by upscaling a digital elevation model (USGS, 2006) from 90 meters to 1 km; (b) 18 land-cover categories (Table S6) adapted from Campos and Brito (2018) and up-scaled from 30 m to 1 km; (c) four bioclimatic variables, maximum temperature of warmest month (BIO5), minimum temperature of coldest month (BIO6), temperature annual range (BIO7), and annual total precipitation (BIO12) from Hijmans et al. (2005). All variables were uncorrelated (r < 0.75).

The models were developed using the Maximum Entropy approach implemented in Maxent v.3.3 (Phillips et al., 2006) with the following settings: 5000 maximum number of iterations; regularization multiplier equal to 1; 10,000 maximum number of background points; 20 replicates selected by bootstrap. The area under the receiver-operating curve (AUC) of each replicate run was taken as a measure of model accuracy and ensemble models were generated by averaging the 20 model replicates. Response curves and jackknife analyses were performed to assess the importance of each variable in each model replicate (e.g., Vale et al., 2014). Finally, the minimum training presence threshold was applied to the ensemble model given that less restrictive thresholds should be applied for conservation purposes (Liu et al., 2005). The resulting binary map (depicting presence/absence areas) was used to calculate the extent of occurrence and area of occupancy following IUCN guidelines for assessing Red List categories (IUCN, 2017).

The percentage of presences of each lineage of the olivieri complex in each land-cover unit was taken as a measure of the biogeographic affinities of each group (Brito et al., 2009). Selection among land-cover units was quantified from the percentages of training observations using the Standardized Levin's B measure of niche breadth: Bs ¼ B_1/n_1, where B is the Levin's index and n the total number of land-cover units. “B” is given by 1/P(p2), where p is the proportion of observations in each land-cover unit. The standardized index was used because of unbalanced sample size among groups. Eight land-cover units selected were sandy areas (51.1% of the study area), compact soil (6.1%), rocky plateaus (16.8%), bare rocks (6.3%), gravel and sand floodplains (1.9%), grasslands (9.1%), savannah (2.9%), and croplands (5.7%).

2.3.3 Taxonomic ranking

We (the authors) do not necessarily adhere to the same species concept. While some follow the Unified Species Concept (USC) of De Queiroz (2007), some prefer the general framework of the Biological Species Concept (BSC; Mayr, 1970). However, we all recognize reproductive isolation as the primary operational criterion for the delimitation of species and we all agree that, while every species is a lineage, not every lineage is a species. The approach used in this paper can be defined either as following the BSC framework, or as applying a Biological Species Criterion under the USC. We thus treated sympatric or parapatric lineages that do not exchange genes when they are not isolated by extrinsic geographical barriers as species. Moreover, this study adopts the framework of integrative taxonomy based on the assumption that divergences in any of the attributes can provide evidence for the species' existence (Dayrat, 2005; Padial et al., 2010). We have divided our dataset into six categories that can each be regarded as distinct lines of evidence. They can be combined and compared in order to assess the congruence of the putative species limits among the olivieri complex: (a) the mtDNA data can be used to test the criteria of reciprocal monophyly and the presence/absence of barcoding gaps, (b) the multilocus nucDNA data can be used to test, independently from the mtDNA set, the criteria of reciprocal monophyly, notably in coalescence theory framework, (c) the morphological data allows to test the criteria of morphological divergence, and (d) habitat and v) distribution range and vi) land-cover datasets can be used as a complementary approach to test the criteria of isolation and distinct habitat requirements.

3.4.1 Morphological analyses

Measurements of all individuals examined for this work are given in Tables S3–S5. Descriptive statistics of the morphological data are shown in Tables 5–7 and Table S7. Morphological differences were found between the two lineages (ADR and TAG) representing the putative new species from Mauritania, and the remaining species of the complex (Figures S5–S8; Tables 5–7 and Table S7).

Compared with the other species of the M. olivieri complex, individuals of the ADR and TAG lineages have: (a) a narrower and more pointed snout (Figures S5 and S6; Table 6 and Table S7), (b) more protuberant nostrils (Figures S5 and S6; Table 5 and Table S7), (c) more dorsal and gular scales (Figure 7 and Figure S7; Table 6 and Table S7), (d) more femoral pores (Figure S7; Table 6 and Table S7), and (e) a higher number of lamellae beneath the fourth toe (Figure S7; Table 6 and Table S7). Moreover, (f) the shape and disposition of the scales composing the eyelid in the new species resemble those of Mesalina guttulata (Lichtenstein, 1823) (1–2 clearly enlarged, transparent scales in the lower eyelids versus several sub-equal scales in the M. olivieri complex) although most of the specimens of the ADR and TAG lineages lack the well-defined dark lines edging the large eyelid scales found in M. guttulata.

Major differences were also found in the dorsal coloration. The PCAs on these characters including specimens of M. olivieri and the putative new species highlights the presence of pale dorsal spots on the dorsum (PSDDSLL, more obvious in both male and females of the potential new species) and the pattern of dark bands on the dorsum (SDDLB) and on the flanks (DBF) both thicker and better defined in M. olivieri than in the putative new species. Similarly, the dorsal pattern differs between the new species and M. pasteuri: in M. pasteuri the dorsal patterns is mostly made of obvious pale and dark continuous longitudinal stripes, whereas in the putative new species the two dark supra-dorsolateral lines are highly fragmented (forming irregular small patches) and the two pale dorsolateral lines are thin and fragmented (Figures S5 and S8; Table 7).

3.4.2 Distribution modeling and habitat comparison

The 20 replicate ecological models exhibited good predictive accuracy (average AUC = 0.906, Table S8). The most important environmental predictors were terrain ruggedness index (42.6% contribution) and land-cover category (35.4%) (Figure S9; Table S8). The highest probability of occurrence is in the intermediate levels of terrain ruggedness and in bare rock habitats (Figure 6 and Figure S9). These findings fit well with our field observations, as samples of the ADR and TAG lineages were always collected in rocky habitats on slopes or plateaux, while M. pasteuri was only found in the adjacent plains (pers. obs.).

We found significant differences in number of observations in each land-cover category (χ²; p = 0.006; df = 21; Figure 6; Table S9) for the area considered in this study: (a) the lineage representing the putative new species from Mauritania (ADR-TAG) was mostly found (70% of observations) in bare rocks, a restricted land-cover category in the study area (6.3% of the area) but showing similarities with M. guttulata habitat; (b) M. simoni and the olivieri lineages AS and AM were most frequently found in rocky plateaus; (c) M. olivieri occurred in almost all units; and (d) M. pasteuri appeared to be most related to sandy areas. The niche breadth estimation was under 0.5 in all cases indicating that all taxa are specialized in habitat selection, and that the olivieri lineages ADR and TAG are the most specialized ones.

3.4.3 Taxonomic implications

The ADR and TAG lineages form a well-supported, monophyletic clade. In all datasets and all analyses, they share the same morphological features and habitat that distinguish them from the other species of Mesalina from northwest Africa. We thus treat them as one evolutionary unit for the time being, pending further studies and larger sampling of the TAG lineage. Given the observed divergences in genetics, morphology, and ecology from the other species of the genus Mesalina, and the lack of alleles sharing with the sympatric M. pasteuri and parapatric M. simoni (Figure 5 and Figure S10) demonstrating reproductive isolation, we treat the evolutionary unit made of the ADR and TAG lineages as a valid species that we describe here.

Although the current taxonomy of the genus Mesalina assigns populations of the complex from southern Morocco and the Atlantic Sahara to M. olivieri, both mtDNA and nucDNA data unambiguously assign the lineages AS and AM to M. simoni rather than to M. olivieri sensu stricto (s.s.). We, thus, formally assign, here, all the populations of the M. olivieri complex from southern and south-western Morocco to the Atlantic Sahara to M. simoni. We thus restrict M. olivieri to the populations distributed from central and eastern Morocco to Israel (in brown in Figure 4; see also Figures 2, 3 and 5). We suspect that M. olivieri, as restricted here, is itself a species complex and that several species-level units are still merged under this name.

In the concatenated mtDNA + nucDNA trees, there are four reciprocal monophyletic lineages within M. simoni: (a) SOUS, (b) AS, (c) AM, and (d) M. simoni s.s. (Figures 3 and 4). We have sequenced or examined very few specimens of the SOUS and AM lineages and thus refrain from proposing a distinct taxonomic status for the time being. From now on, we will refer to these lineages as M. simoni ssp. “SOUS” and M. simoni ssp. “AM”. On the contrary, the AS lineage exhibits a distinct morphology, with most sub-adult or adult specimens being diagnosable from M. simoni s.s. (Figure S5). However, the four lineages are not distinct on the nuclear networks and specimens from the northern Atlantic Sahara or extreme south-western Morocco appear intermediate morphologically between M.simoni s.s. and AS (pers. obs.). We thus refrain from treating the AS lineage as a distinct species and formally recognize it as a new subspecies of M. simoni.

The nomen simoni was based on seven specimens collected “inter urbes Mogador et Marocco” (= between the cities of Mogador and Marocco) and one specimen from “prope urbem Casablanca” (= not far from the city of Casablanca, see Böttger, 1880). Mogador is the city currently known as Essaouira and Marocco was used at the time for the city of Marrakech. As far as we know, there has been no type locality restriction for this nomen so the type locality of M. simoni remains as “between Essaouira and Marrakech and near Casablanca.” This area is inhabited by the lineage currently designed as M. simoni s.s. All the nomina currently allocated to the synonymy of M. olivieri were based on specimens from the eastern clade of M. olivieri, and we were unable to identify any nomen available for the lineages of the M. olivieri complex inhabiting Mauritania or south-western Morocco. We, thus, need to create new nomina to name the new species from Mauritania (ADR and TAG linages) and the new subspecies from south-western Morocco (AS lineage).

Consequently, we suggest the following nomenclatural and taxonomic actions: (a) designate all the populations of the lineages ADR and TAG as a new species here described as Mesalina adrarensis sp. nov.; (b) designate as subspecies of M. simoni all the M. olivieri present in the Atlantic Sahara region, here described as Mesalina simoni saharae ssp. nov.; (c) allocate all populations of M. simoni SOUS and AM (from north of the High Atlas and west of the Middle Atlas) to M. simoni. The two subspecies Mesalina simoni simoni and M. simoni saharae ssp. nov. are bridged by populations of intermediate morphology and genetic background between the High Atlas and the north of the Atlantic Sahara; specimens of M. simoni ‘SOUS’ (inhabiting the area around Tan-Tan and El Ouatia) are morphologically close to typical saharae but seem already admixed genetically with M. s. simoni (see specimen B9429 Figure 2a and Figure S2). The populations currently treated as M. simoni ‘AM’ could conceivably be included in M. simoni saharae ssp. nov., as they are genetically grouped with this taxon, but we prefer to analyses more specimens before we can conclude.

Mesalina adrarensis sp. nov.

(Figures 1-8; Tables 1, 3–6)

Zoobank registration

http://zoobank.org/NomenclaturalActs/dc6c167b-8138-4017-82a4-513d8244072a

Holotype

Adult female (Figure 7) with code MNHN-RA-2020.0017 preserved in the Muséum national d'Histoire naturelle, in Paris, France. Collected in the Tiris Zemmour (Mauritania) by Philippe Geniez, Olivier Peyre, Pierre-André Crochet and José Carlos Brito on 7^th April 2017.

Type locality

Mauritania, 440 m south-west of the Guelta Oumm el Habâl, 17.4 km east-southeast of F'derick. 22.60636°N/−12.55743°W/372 m a.s.l.

Paratypes

BEV.15163, adult female collected in the Adrar Atar, 44 km before Chinguetti coming from Atar (Mauritania, 20.5537°N/12.6916°W) by F. Martínez-Freiría, J.C. Brito, D.V. Gonçalves, J.C. Campos, Z. Boratyński, C.G. Vale, T.L. Silva, X. Santos, J.M. Pleguezuelos, M. Feriche and A.S. Sow on the 29^th October 2011; BEV.15060, adult male, same locality, collected by J.C. Brito, Z. Boratyński, S. Lopes, J. Marques and F. Martínez-Freiría on the 12^th September 2015; BEV.15061, adult female, BEV.15062, adult male, BEV.15063, adult male, BEV.16064, adult male, all collected in the Adrar Atar, Oumm Lekhterat (Mauritania, 21.1596°N/11.9362°W) by J.C. Brito, Z. Boratyński, S. Lopes, J. Marques and F. Martínez-Freiría on the 13^th September 2015, all preserved in the BEV collection in Montpellier.

Etymology

The species epithet “adrarensis” refers to the Adrar Atar region because the new species was first suspected when seeing specimens from this region.

Diagnosis

A species of the M. olivieri complex characterized by the following combination of characters: (a) low number of eyelid scales (5–6); (b) with two clearly larger scales, like M. guttulata (several sub-equal scales or rarely 1–2 larger scales in the other species of the M. olivieri complex); (c) small black dots on the edge of the eyelids (mostly visible in dead animals preserved in ethanol); (d) more elongated snout with more prominent nostrils than the other species of the olivieri complex; (e) adult coloration in life (Figure 7) different from all other species of the complex (see Figure S5 and comparison below).

Coloration in life (adults)

Dorsum coloration mostly brown, from sandy-brown to dark-brown or rufous-brown, with a dorsal pattern made of two central longitudinal lines of small whitish spots (usually partially edged with dark-brown or black) then two supra-dorsolateral longitudinal rows of black blotches edged externally by narrow dorsolateral fragmented whitish lines, sometimes nearly continuous, sometimes reduced to series of small elongated dots; the flanks show small pale spots partially edged by dark coloration, sometimes fusing in a near complete dark band; on the lower part of the flanks a more continuous longitudinal white line separates the dorsal and ventral parts of the body. The pattern from the center of the back fades on the anterior part of the dorsum, especially on the nape, then disappears on the head. Dorsal pattern more contrasted in females than in some males; in some males, this pattern is reduced to a series of whitish and dark-brown spots aligned along the body. The color of the head is similar to the body coloration with faintly spotted pileus and a dark line running though the sides of the head through the eye, bordered below by a pale line reaching to the eye. Forelimbs coloration uniformly brown, hind legs brownish with whitish ocelli. Uniformly brown tail with a median dark stripe, more or less visible, disappearing at the first third of its length, this median stripe is externally edged on each side by a light stripe resulting from the prolongation of each light dorsolateral stripe. Underparts of the head and the legs pinkish-beige, belly overall similar but paler on central belly, underparts of the tail whitish, sometimes yellowish especially in males. Juveniles present a pale background coloration, striped with wide and continuous pale and dark (sometimes white and pure black) bands along the body, then very much like M. pasteuri (this pattern is probably common to all juveniles of the M. olivieri species complex).

Comparison

Mesalina adrarensis sp. nov. resembles (sometimes strongly) M. guttulata, possibly due to adaptation to similar environmental conditions. In comparison with M. guttulata, M. adrarensis sp. nov. has a browner background coloration and a less densely spotted dorsal pattern that often fades on the neck without exhibiting the remarkable pattern of irregularly arranged ocelli that characterizes M. guttulata. Contrary to M. guttulata, M. adrarensis sp. nov. has small light spots and dark marks organized in longitudinal lines. In both species, the eyelids are composed of 2 (rarely 1) large translucent scales, situated above 0 to 8 smaller ones; while in M. guttulata these scales are always edged with a continuous black stripe, in M. adrarensis sp. nov. this black coloration is mostly absent or made of spots along the edges of the scales. Compared with M. olivieri and M. simoni, M. adrarensis sp. nov. has more prominent nostrils, a longer snout, and more flattened head. Dorsal coloration in juveniles of all species of the M. olivieri species complex is characterized by continuous stripes. Meanwhile M. olivieri and M. pasteuri (but less frequent in M. simoni) maintain this striped coloration in their adult forms, these dorsal stripes in adults of M. adrarensis sp. nov. are fragmented. The eyelids of M. simoni and M. olivieri are made of 6–12 medium and small scales but no or (rarely) 1–2 clearly larger scales (cf. Figure 7i).

Adults of M. adrarensis sp. nov. can be easily distinguished from the sympatric M. pasteuri by the lack of obvious dorsal and dorsolateral stripes that characterize the adults of the latter species, and by the brownish coloration typical of M. adrarensis sp. nov. (M. pasteuri has a typical sand color in accordance with its sandy habitat; Figure S5).

Genetic and phylogenetic remarks

The phylogenetic analyses by Kapli et al. (2015), Simó-Riudalbas et al. (2019) and the phylogenetic and nuclear network analyses performed in this study (Figures 2, 3 and 5 and Figures S2–S4; Table 4) validate the specific status of M. adrarensis sp. nov.. The amount of genetic divergence (p-distance) for the Cyt-b gene between the new species and the other members of the M. olivieri complex ranges from 9% (from its sister species M. simoni) to 14% M. olivieri (Table 4). The network analysis of the nuclear genes indicates that, despite the large number of samples of the M. olivieri species complex included in the analysis (70, 85, 71,and 69 for β-fib7, MC1R, PgD7, and OD, respectively), all haplotypes of M. adrarensis sp. nov. are private (Figure 5).

Description of the holotype

Adult female (Figure 7). Slender body slightly depressed in the caudal part. Long and pointed snout with slightly prominent nostrils; SVL = 44 mm, Tail length (entire, not regenerated) = 86 mm, total length = 130 mm, Head length = 10 mm, Head width = 6.4 mm, Head height = 4.3 mm, midbody dorsal scales = 44, small smooth, and regular temporal and dorsal scales, transverse rows of ventral plates = 32 arranged in eight longitudinal rows, enlarged plates in collar = 7, supralabials = 4, infralabials = 7, gular scales (counted along a line from the infrasupralabials to the collar) = 24, femoral pores 15 + 14, lamellae beneath the 4^th toe = 19 (slightly keeled). Eyelid disks with five barely translucent scales (two large + three small) not edged with black.

Distribution and habitat

The known range of M. adrarensis sp. nov. encompasses the mountain rocky areas of the Adrar Atar in Mauritania, further extending to the north up to the plateau of Zednes (Observation 13758; Table S10), to the Northwest up to Koudiet Laghnem (Morocco), and to the south down to the central Tagant Mountain (Figure 8). The observation 13758 is a sight record that is unsupported by any photo or tissue sample but has been identified in the field by its typical habitus and habitat (J. C. Brito pers. obs.)

The Area of Occupancy (AOO) calculated from the ecological models was of 34,766 km², while the Extent of Occurrence (EOO) was of 175,445 km² (Table S8).

The highest probability of occurrence is at an intermediate levels of terrain ruggedness and in bare rock habitats (Figure 6 and Figure S9). The type locality is a flat and rocky area located on the top of a plateau in the Adrar Atar (Figure 8 and Figure S10b,c). The area is very scarcely vegetated with many stones, mostly low and sparse shrubs and, rarely, Acacia sp. trees (Figure 7j).

Ecology

Specimens were observed active among stones mostly from 11:00 am to 17:00 pm. If threatened it usually seeks shelter under a rock or in spiny bushes. Diet and reproductive features are still unknown.

Conservation status

The AOO calculated from the ecological models was 34,766 km², while the EOO was 175,445 km². Most of the habitats are for the moment devoid of major human impact and we are not aware of any direct threat to the species such as direct destruction of human exploitation. Although we do not have any information on the species' population trends, we do not suspect any marked decline based on the lack of direct or indirect menace. Following IUCN guidelines (2017), we thus propose here that the conservation status of M. adrarensis sp. nov. should be Least Concern (LC), based on the high values for the area of occupancy and extend of occurrence and the lack of observed or forecasted population decline. However, the area of occupancy is highly fragmented due to the ecology of the species (Figure 8) and the amount of knowledge currently available for this species remain limited.

Notes

The specimens BEV.10823 (GenBank code KM411138) and CIBIO2952 (GeneBank code KM411139) correspond to a juvenile (BEV.10823) and to a sub-adult (CIBIO2952) M. adrarensissp. nov., previously identified as M. pasteuri in Kapli et al. (2015) and Simó-Riudalbas et al. (2019). The assignment of these two specimens to M. pasteuri was due to their striped color pattern, typical of M. pasteuri. However, both our, Kapli et al. (2015) and Simó-Riudalbas et al. (2019) genetic results assign these two specimens to M. adrarensis sp. nov.. The difference in coloration between these two specimens and the other specimens of M. adrarensis sp. nov. results from strong ontogenetic variation of coloration in M. adrarensis sp. nov. (contrastingly striped coloration in juveniles).

Mesalina simoni saharae ssp. nov.

(Figures 1-5 and 9; Tables 1, 3–6)

Zoobank registration

http://zoobank.org/NomenclaturalActs/30f33118-ec56-48a6-87b9-1d811dac016e

Holotype

Adult male (Figure 9), with code MNHN-RA-2020.0018 (ex BEV.9114) preserved in the Muséum national d'Histoire naturelle, in Paris, France. Collected by Pierre-André Crochet and Julien Renoult on 10^th September 2006.

Type locality

Morocco, Atlantic Sahara, road N1, 69 km past Boujdour toward Laayoune, 26.4925°N/−13.9198°W/60 m a.s.l.

Paratypes

Adult male BEV.10849, from 6 km E of Sidi Kathari, 26.5298°N/−12.3364°W, collected by Pierre-André Crochet on 21^st March 2010; adult female BEV.10850, from 4 km past Awserd toward Dakhla, 22.5709°N/−14.3544°W, collected by Pierre-André Crochet on the 18^th March 2010, all preserved in the BEV collection in Montpellier.

Etymology

The epithet “saharae” refers to the Atlantic Sahara region where this new subspecies is distributed.

Diagnose

A member of the M. olivieri species complex closely related to the nominotypical M. simoni, but with the following combination of characters: (a) eyelids with 5–6 large transparent scales (b) not edged in black (rarely indistinct spots on their edges), (c) dorsal coloration in life generally sandy, sandy-brown, or sandy gray with (d) a poorly marked dorsal pattern mostly composed of longitudinal rows of whitish spots; spots narrowly edged internally by dark coloration (dark inner edge typically 1–2 or 2–3 scales wide), sometimes bordered by a pale grayish dorsolateral line, (e) a dark continuous or near continuous band from the nostril along the flanks to the hind legs constituting the darkest element in the pattern and often including pale spots and their dark edges, (f) a distinctive delicate habitus with elongated body and neck and flattened head (Figure 9 and Figure S5).

Genetic and phylogenetic remarks

The phylogenetic analyses by Kapli et al. (2015), Simó-Riudalbas et al. (2019) and the phylogenetic and nuclear network analyses performed in this study (Figures 2, 3 and 5 and Figures S2–S4; Table 4) support the hypothesis that the populations of M. simoni saharae ssp. nov. belong to the species M. simoni and not to the species M. olivieri. A network analysis of the nuclear genes indicates that, despite the large number of samples of the M. olivieri species complex included in the analysis (70, 85, 71, and 69 for β-fib7, MC1R, PgD7, and OD, respectively), all haplotypes of M. simoni saharae ssp. nov. are shared with M. s. simoni and not with M. olivieri (Figure 5). The amount of genetic divergence (p-distance) in the Cyt-b gene between the new subspecies and the other members of the M. olivieri complex ranges between 4% from M. simoni and 11% from M. olivieri (Table 4). One specimen from the Souss valley, geographically located between the ranges of both subspecies, is morphologically similar to M. s. simoni but is somewhat intermediate in the nuclear and concatenated trees, suggesting a lack of reproductive isolation between simoni and saharae. Based on this and on their level of genetic divergence, we treat saharae and simoni as conspecific pending more detailed analyses of their interactions in contact zones.

Description of the holotype

An adult male (Figure 9) with well-developed femoral pores. Slender and elongated body and slender head; SVL = 41 mm, regenerated tail, Head length = 9.6 mm, Head width = 5.9 mm, Head height = 4.2 mm, midbody dorsal scales = 34, medium sized, slightly pointed, not shining; transverse rows of ventral = 28, longitudinal rows of ventral = 8, enlarged plates in collar = 6, supralabials = 4 + 5, infralabials = 7 + 9, gular scales (counted along a line from the infrasupralabials to the collar) = 25, femoral pores = 10 + 13, lamellae beneath the 4^th toe = 20. Eyelid disks with five translucent scales not edged with black.

Coloration in life

See diagnose above. In addition, pileus almost uniform, with faded spots, sometimes distinctly spotted with small dark spots. Fore limbs uniformly sandy-brown, hind legs brownish with faded whitish ocelli. Almost uniform sandy/brown tail with indistinct spots disappearing in the second half of its length. Ventral coloration white, sometimes turning to yellow on the undertail, at least in the specimens from the north of the range.

Distribution and habitat

As understood here, this subspecies' distribution encompasses approximately 173,970 Km² across the west of the Sahara, from the Mauritanian border to the Tan–Tan area, and as far inland as Awserd (see Figure 4). The current knowledge on this distribution is clearly constrained by the reduced accessibility of the region and is doubtless more extensive than currently known. In its distribution range, this subspecies mostly dwells semi-deserts and coastal steppes with scattered low bushes on loess, coarse sand, or gravel substratum (Figure 9).

Notes

Individuals from Tan–Tan to the Sous-Massa region (Morocco; 1-SOUS in Figures 2-4 and Table 1), seem to be genetically admixed between M. simoni saharae ssp. nov. and M. s. simoni; morphologically, specimens from the Tan–Tan area are rather typical of M. simoni saharae ssp. nov. though while our specimen from the Souss is rather typical of M. s. simoni. More detailed studies are needed to clarify the extant of gene flow between these two subspecies and the taxonomy of these populations that inhabit the area between the cores of the range of the two subspecies.