Two newly recognized species of Hemidactylus (Squamata, Gekkonidae) from the Arabian Peninsula and Sinai, Egypt

Abstract A recent molecular phylogeny of the Arid clade of the genus Hemidactylus revealed that the recently described H. saba and two unnamed Hemidactylus species from Sinai, Saudi Arabia and Yemen form a well-supported monophyletic group within the Arabian radiation of the genus. The name ‘Hemidactylus saba species group’ is suggested for this clade. According to the results of morphological comparisons and the molecular analyses using two mitochondrial (12S and cytb) and four nuclear (cmos, mc1r, rag1, rag2) genes, the name Hemidactylus granosus Heyden, 1827 is resurrected from the synonymy of H. turcicus for the Sinai and Saudi Arabian species. The third species of this group from Yemen is described formally as a new species H. ulii sp. n. The phylogenetic relationships of the members of ‘Hemidactylus saba species group’ are evaluated and the distribution and ecology of individual species are discussed.


Introduction
The genus Hemidactylus Oken, 1817, the second most species-rich genus of Gekkonidae (122 currently valid species; Uetz 2013), has been witnessing a species-description boom within the last decade. Eighteen species have been described within the last two years, most of them from the Arabian Peninsula and surroundings areas where 13 new species and a new subspecies have been discovered (Busais and Joger 2011a;Moravec et al. 2011;Torki et al. 2011;Carranza and Arnold 2012). Despite the large number of taxa added recently to the Arid clade of Hemidactylus [sensu Carranza and Arnold (2006)], it has been shown that the real diversity of Hemidactylus in Arabia and northeast Africa is still underestimated, with at least seven species remaining to be described (Busais and Joger 2011b;Moravec et al. 2011;Šmíd et al. 2013). A recent study (Šmíd et al. 2013) revealed that two of these newly recognized but still unnamed species, one from Sinai [labelled in accordance to previous works (Moravec et al. 2011;Šmíd et al. 2013) as Hemidactylus sp. 1] and one from Yemen (Hemidactylus sp. 4), clustered with the recently described Yemeni endemic H. saba Busais & Joger, 2011. They form a very well supported clade within the Arabian radiation of the genus (Fig. 1). Although the phylogenetic relationships among these three species were not resolved satisfactorily, it was inferred that they began to diversify approximately 7 million years ago (95% highest posterior density interval 4.3-10), what was followed by a subsequent dispersal of the Sinai species from southern Arabia to the north (Šmíd et al. 2013).
The discovery of a monophyletic species group consisting of one recently described and two newly recognized species calls upon a more thorough study of the nomenclatural status, evolutionary relationships, taxonomy and distribution of its members based on further genetic and morphological data. The present study focuses on this task.

Material for phylogenetic analyses
In order to resolve the phylogenetic relationships between the two newly recognized Hemidactylus species and H. saba based on genetic data, a dataset containing only representatives of these three species was assembled. Apart from the data used by Šmíd et al. (2013), additional sequences of the following specimens were produced (Table 1): the holotype and two paratypes of H. saba (the only known existing material), 21 individuals from Sinai and Saudi Arabia belonging to H. sp. 1 (Šmíd et al. 2013), and five individuals of the undescribed species from Yemen (H. sp. 4; Šmíd et al. 2013), one of which was included in the study by Busais and Joger (2011a) (labelled as 'OTU 7' therein). Total genomic DNA was extracted using DNeasy Blood & Tissue Kit (Qiagen). Subsequently, sequences for up to two mitochondrial (12SrRNA [12S] -ca. 400 bp and cytochrome b [cytb] -307 bp) and four nuclear (cmos -402 bp, mc1r -666 bp, rag1 -1023 bp, rag2 -408 bp) were produced using primers and PCR conditions described in details elsewhere (Šmíd et al. 2013). Chromatograms of all newly obtained sequences were checked by eye and assembled in Geneious 5.6.5 (Biomatters, http:// www.geneious.com/ ). All genes were aligned individually using MAFFT (Katoh and Toh 2008) with the iterative refinement algorithm with 1000 iterations. Poorly aligned positions in the alignment of 12S were eliminated with Gblocks (Castresana 2000) under low stringency options (Talavera and Castresana 2007), producing a final 12S alignment of 386 bp. Alignments of all coding genes were trimmed so that all started by the first codon position and no stop codons were revealed when translated into amino acids with the appropriate genetic codes.

Phylogenetic analyses and haplotype networks construction
The final dataset consisted of 36 ingroup individuals. Specimen numbers, localities, and GenBank accession numbers of all genes sequenced are presented in Table 1. The alignment of all concatenated genes was 4012 bp long. The software jModelTest 2.1.1 (Guindon and Gascuel 2003;Darriba et al. 2012) was used to assess the best-fitting model of nucleotide substitution for each gene separately under the Akaike information criterion [AIC, Akaike (1973)]. The best-fitting models were selected as follows: 12S -GTR+G; cytb -GTR+I+G; cmos -HKY+I; mc1r -TIM2+I; rag1 -HKY+I; rag2 -TrN+I). Phylogenetic analyses were performed using maximum likelihood (ML) and Bayesian inference (BI) methods. In order to detect the potential effect of the nuclear genes on the tree topology and nodal support, independent analyses were run on two datasets: (1) a dataset containing mtDNA genes only (12S, cytb), and (2) a concatenated dataset of all mtDNA and nDNA genes. Sequences of nuclear genes were not phased; heterozygous positions were coded according to the IUPAC ambiguity codes. Gaps were treated as missing data. Three specimens of H. flaviviridis and one of H. angulatus, representatives of two different clades of Hemidactylus (Carranza and Arnold 2006), were used to root the trees. Uncorrected genetic distances (p distances) were calculated in MEGA 5 (Tamura et al. 2011). Almost complete cytb sequences (1127 bp) of the new species from Yemen deposited in GenBank (Šmíd et al. 2013) were used to calculate p distances within this species, whereas an alignment of 307 bp was used to obtain intraspecific p distances within H. saba and the new species from Saudi Arabia and Sinai, and also interspecific p distances between these three species.
Maximum likelihood analyses of both datasets were performed in RAxML 7.0.3 (Stamatakis 2006) using raxmlGUI (Silvestro and Michalak 2012) graphical extension with parameters estimated independently for each partition, GTR+I+G model of nucleotide evolution and a heuristic search with 100 random addition replicates. Support of the tree nodes was assessed by bootstrap analysis with 1000 pseudoreplications (Felsenstein 1985).
The BI analyses were run in MrBayes 3.2.1 (Ronquist et al. 2012). Appropriate equivalents of the best-fitting models were specified to each partition (gene) and all parameters were unlinked across partitions. Analyses were performed with two runs and four chains for each run for 10 7 generations, with sampling interval of 1000 generations. Appropriate sampling was confirmed by examining the stationarity of log likelihood (lnL) values and the value of average standard deviations of the split frequencies. Convergence between two simultaneous runs was confirmed by the PSRF (potential scale reduction factor) value. From 10 4 sampled trees, 25% were discarded as a burn-in and a majority-rule consensus tree was produced from the remaining ones, with posterior probabilities (pp) of each clade embedded. Nodes with ML bootstrap values ≥ 70% and pp values ≥ 0.95 were considered highly supported (Huelsenbeck and Rannala 2004).
Heterozygous positions in nuclear genes were identified based on the presence of double peaks in chromatograms and using the Heterozygote Plugin in Geneious. For the purpose of haplotype network construction, haplotypes from sequences with more than one heterozygous position were resolved in PHASE 2.1.1 (Stephens et al. 2001). Input data for PHASE were prepared in SeqPHASE (Flot 2010). In order to include as much data as possible, sequences of all Hemidactylus species from the Arid clade used in our previous study (Šmíd et al. 2013) were combined with the newly produced sequences and phased together (data not shown). In the case of rag1, the original alignment was trimmed to 846 bp, the length at which sequences of all individuals did not contain any N ends that would give misleading results in the allele reconstruction (Joly et al. 2007). PHASE was run under default settings except the probability threshold, which was set to 0.7. Haplotype networks of the four nuclear markers (cmos, mc1r, rag1, rag2) were drawn using TCS 1.21 (Clement et al. 2000) with 95% connection limit.

Material for morphological analyses
Material for morphological comparison included 225 specimens of 8 Hemidactylus species and one subspecies (Appendix) and was obtained from the following collec-

Morphological characters
The following measurements were taken with Powerfix digital calliper to the nearest 0.1 mm: snout-vent length (SVL), measured from tip of snout to vent; head length (HL), measured from tip of snout to retroarticular process of jaw; head width (HW), taken at the widest part of the head; head depth (HD), maximum depth of head; left eye diameter (E), measured horizontally; axilla-groin distance (AG), measured from posterior end of front limb insertion to anterior end of hind limb insertion; tail length (TL), measured from vent to tip of original tail. In addition to these metric characters, the following meristic characters were examined using a dissecting microscope: number of upper and lower labials (left/right); contact of nasals; number of infralabials in contact with first postmentals; mutual position of first postmentals; number of longitudinal rows of enlarged dorsal tubercles; number of lamellae under the first and fourth toe including unpaired proximal ones; and number of preanal pores in males. Terminology and diagnostic characters follow Moravec and Böhme (1997) and Moravec et al. (2011).

Results
Phylogenetic analyses of both datasets resulted in trees presented in Fig. 2. Tree topology remains congruent with that showed in Šmíd et al. (2013). The three species form a well-supported monophyletic group (mtDNA: ML bootstrap 85/ Bayesian pp 1; mtD-NA + nDNA: 100/1) to which we will refer to as the 'Hemidactylus saba species group' [support of individual species: H. saba (100/1; 100/1), Hemidactylus sp. 1 from Sinai and Saudi Arabia (100/1; 100/1), Hemidactylus sp. 4 from Yemen (83/1; 100/1)]. The performed analyses did not resolve the topology within this species group despite the inclusion of more individuals and additional genetic data in comparison with previous works (Moravec et al. 2011;Šmíd et al. 2013). Therefore, with the current knowledge, this group remains polytomic. There is no genetic variability within H. saba (all three specimens analyzed originate from the same locality) in both of the studied mtDNA genes and a very little variability in nDNA (mc1r and rag1 only) (Fig. 3). The species from Sinai and Saudi Arabia also shows very little variation in mtDNA (intraspecific p distance max. 1.3% in both 12S and cytb), but it varies in sequences of all the nDNA genes studied (Fig. 3). On the other hand, the unnamed Hemidactylus from Yemen exhibits relatively deep intraspecific differentiation into three well supported lineages. Uncorrected genetic distances between these lineages are up to 6.3% in cytb and up to 4.2% in 12S (Fig. 2). Moreover, the nDNA genes show a high level of genetic differentiation (Fig. 3). Intra-and interspecific genetic distances in both mtDNA genes analyzed between all three species are shown in Fig. 2. The results of the nuclear networks indicate that all alleles for all four independent loci are specific for each species.
The results of the molecular analyses, together with a unique combination of morphological features (see below) confirm the earlier conclusion that the newly recognized Hemidactylus sp. 1 and Hemidactylus sp. 4 represent two separate species, whose taxonomy and nomenclature need to be resolved.  (cmos, mc1r, rag1, rag2). Circle sizes are proportional to the number of alleles. Small white circles represent mutational steps. Position of alleles BJ09a and BJ09b in the mc1r network is indicated by dashed lines because the sequence of the sample BJ09 (voucher NHM-BS N41916) was 108 bp shorter than the rest of the alignment and haplotype network reconstructions based on both 666 bp and 558 bp alignments linked these alleles to JS32b and JS32a, respectively. Status and nomenclature. Heyden (1827) described Hemidactylus granosus as a new species occurring in Egypt, Arabia and Abyssinia (Ethiopia and Eritrea). Although not explicitly mentioned by the author, the description was apparently based on four specimens collected by Rüppell currently deposited in the Senckenberg Naturmuseum Frankfurt (collection numbers SMF 8723-8726). Heyden did not diagnose the new species against H. turcicus (Linnaeus, 1758) and in respect to our today's knowledge on the morphological variation in Hemidactylus the description of H. granosus is very general. Traditionally, H. turcicus has been considered a common species widely distributed across the Mediterranean and the Middle East. As the general diagnostic characters of H. granosus given by Heyden (1827) were also applicable to H. turcicus at that time, the name Hemidactylus granosus Heyden, 1827 was considered its junior synonym (e.g. Boulenger 1885, Loveridge 1947, Mertens and Wermuth 1960, Mertens 1967, Salvador 1981, Baha El Din 2006.

Redescription of Hemidactylus granosus
Recent examination (by JŠ) of four specimens collected by Rüppell (SMF 8723-8726) has shown that one of them [SMF 8723 designated by Mertens (1967) as lectotype of H. granosus; for description see below] corresponds morphologically to Hemidactylus sp. 1 from Sinai. The other three specimens from this series morphologically correspond to H. robustus Heyden, 1827 (SMF 8725, 8726) and H. cf. granosus (SMF 8724), an animal superficially resembling H. granosus but differing from the members of the 'H. saba species group' in several important characters (see below). These findings lead to the conclusion that Hemidactylus granosus Heyden, 1827 is a valid taxon and needs to be resurrected from the synonymy of H. turcicus. In the light of current knowledge, the range of H. turcicus does not include a large part of Egypt, being restricted mostly to northern Egypt including Sinai and its Red Sea coast. The species is also missing in Arabia (sensu lato) and Ethiopia (Carranza and Arnold 2006;Moravec et al. 2011;Rato et al. 2011;Šmíd et al. 2013).
Description of the lectotype. SMF 8723, adult male [erroneously determined as female by Mertens (1967)]. Head and body moderately depressed (Fig. 4). Upper labials (10/10), lower labials (8/7). Nostril between rostral, three subequal nasals and in punctual contact with first upper labial. Uppermost nasals separated by a small inserted scale. Mental triangular, as long as wide. Anterior postmentals long, in a broad contact with each other, both in contact with the 1 st and 2 nd lower labial reaching in about one fourth of the width of the 2 nd labial. Second postmentals almost round, touching only the 2 nd lower labial (Fig. 5). Two enlarged scales behind each second postmental, the lateral ones in contact with the 3 rd lower labial. Eye moderate (E/HL=0.26). Head long, distinctly separated from body by a slender neck. Crescent-shaped ear opening. Interorbital region, crown of head and temporal area above the level of ear opening covered by round smooth tubercles. Dorsal region of the specimen is slightly scarred so it is not possible to count the enlarged tubercles on both sides precisely, but there are seven longitudinal rows of large, keeled and caudally pointed tubercles on the left side from which we infer there were originally 14 rows on both sides together. Lower arms, thighs and lower legs with prominent tubercles without keels. Tail original with 6 segments bearing 6 pointed tubercles, broken into three pieces, subcaudals enlarged from just after the hemipenial bulges. Lamellae under the 1 st toe 7/7, lamellae under the 4 th toe 11/11. Four preanal pores in a continuous row. No femoral pores or enlarged femoral scales. Colour (in alcohol) faded due to long fixation.
Paralectotype SMF 8724 differs from other individuals of H. granosus in having relatively high head (HD 50% of HL), lower number of lower labials (6), uppermost nasals in wide contact, first postmentals in contact with 1 st lower labials, and 2 preanal pores.
Comparison. Hemidactylus granosus can be distinguished from other member of the 'Hemidactylus saba species group' and from other congeners distributed in Sinai and the Red Sea coast by the following set of characters (see also Table 2).
From H. flaviviridis by its smaller size (max. SVL 53.2 mm in males and 53.3 mm in females vs. up to 90 mm [Anderson (1999); sexes not distinguished]), by the presence of enlarged dorsal tubercles, and the absence of femoral pores in males.
From H. turcicus by its higher number of upper labials (9-11 vs. 7-10), in having anterior postmentals more frequently in contact with 2 nd lower labial (100% vs. 12.1%), in having anterior postmentals in wide mutual contact behind the mental scale (contact punctual in 67% specimens of H. turcicus), and by the lower number of preanal pores in males (4-7 vs. 6-10).
Variation. Specimens with intact tail vary in number of tail segments bearing 6 pointed tubercles (7-8). The original portion of the tail of the female NMP6V 70163/4 is very wide at the base, separated from cloacal region by a basal constriction. One specimen (IBES10212) is the only animal with 15 longitudinal rows of enlarged tubercles. Another one (IBES10284) has uppermost nasals in wide contact. Most striking is the variation in the number of preanal pores in males. Whereas the lectotype and the only male from Sinai (NMP6V 70163/2) have both 4 pores, all males from Saudi Arabia have 6-7 pores. There seems to be clinal variability in this character, males from NW of the known range (Fig. 6) possess only 4 preanal pores, all animals from the eastern Red Sea coast in Saudi Arabia have 6 pores and a single individual from the southern limit of the range has 7 pores.
There is a very low variation in mtDNA between specimens from Sinai and Saudi Arabia (max. 1.3% in both 12S and cytb). All animals from Sinai share the same haplotypes in 12S and also cytb gene. All four nuclear loci studied show some degree of intraspecific variation (Fig. 3).
Distribution and ecology. Eduard Rüppell collected the original series in 1827 when he began his marine biological studies of the Red Sea and travelled from Egypt to Eritrea. There is no specific information that he went to Arabia as well (Rüppell 1826(Rüppell -1828Klausewitz 2002;Wagner 2008); therefore the original distribution of H. granosus described as "Egypt, Arabia, and Abyssinia [Ethiopia and Eritrea]" by Heyden (1827) was probably too general and incorrect. Because there were no other specimens assignable with certainty to H. granosus apart from the four individuals collected in Sinai (SMF 8723-8726, for their current status see 'Status and nomenclature' section) (Boettger 1893), one of which became the lectotype after Mertens' (1967) designation, Sinai could  Table 1. be considered the only reliable locality for H. granosus. Here, H. granosus is also confirmed from two coastal localities in south and west Sinai and from coastal and inland regions in western and central Saudi Arabia (Fig. 6). Nevertheless, a wider distribution of the species along the Red Sea coast can be expected. According to Baha El Din (2005), Hemidactylus geckos inhabiting the interior lowland of Sinai and the Eastern Desert in Egypt stand out in having notably coarse scalation. Interestingly, the areas with occurrence of animals with coarse scalation correspond with the presence of individuals with low numbers of preanal pores (Baha El Din 2005), which is typical for the Sinai populations of H. granosus.
In 1996, when the NMP specimens were collected, the locality in Sharm el-Sheikh was formed by a crop field supplied with drain water from nearby habitations. Geckos were found during the day under unused empty barrels and also inside buildings. Other species syntopic with H. granosus in Sharm el-Sheikh were: Hemidactylus turcicus, Chalcides ocellatus (Forskål, 1775), Stenodactylus sthenodactylus (Lichtenstein, 1823), and Ptyodactylus hasselquistii (Donndorff, 1798) (R. Víta in litt, 2013). However, when visited again in 2010, the locality had changed dramatically (R. Víta in litt, 2013). The whole area was under heavy development and the irrigation channels had disappeared. The current conditions at the place are unknown to us. In 2011 JM surveyed a neighbouring urban area east of this locality. It was covered by a mosaic of tourist resorts and abandoned ruderal plots. In dry anthropogenic habitats (e.g. rubbish dumps, road ditches, old walls and buildings, abandoned construction sites, natural but heavily disturbed open areas, etc.) dominated two very abundant gecko species. Ptyodactylus hasselquistii occupied primarily various vertical surfaces whereas Cyrtopodion scabrum (Heyden, 1827) prevailed on the ground. Tropiocolotes nattereri Steindachner, 1901 was found in dry and relatively well-preserved natural places. Hemidactylus turcicus was occasionally encountered in more humid artificial habitats in parks and hotel gardens. Specimens from Saudi Arabia were mostly collected during the day inside concrete tunnels under roads. In some of the tunnels they were syntopic with Ptyodactylus hasselquistii. One specimen was also collected on the walls of the Taif National Wildlife Research Centre, where it was also syntopic with Ptyodactylus hasselquistii. Diagnosis. A small species of the 'Hemidactylus saba species group' within the Arabian radiation of the Arid clade of Hemidactylus, as evidenced by the mtDNA and nDNA analyses. The new species is characterized by the following combination of molecular and morphological characters: (1) Uncorrected genetic distances from H. saba: 9.9-10.7% in 12S, 13.5-14.9% in cytb; from H. granosus: 10.2-12.3% in 12S, 11.2-13.5% in cytb; (2) small size with a maximum recorded SVL 40.7 mm (36.8-40.4 mm in males, 39.4-40.7 mm in females); (3) moderately robust head, head length 28-30% of SVL, head width 70-75% of head length, head depth 37-46% of head length; (4) tail length 116% of SVL (only 1 specimen with intact tail); (5) uppermost nasals separated by a small shield (60% specimens) or in wide contact (40%); (6) large anterior postmentals in wide mutual contact in 90% of individuals, and in contact with the 1 st and 2 nd lower labial (scarcely and unilaterally with the 1 st lower labial only); (7) 8-10 upper labials; (8) 7-9 lower labials; (9) dorsum with 12-16 longitudinal rows of enlarged, slightly keeled, conical tubercles; (10) 5-6 lamellae under the 1 st toe and 8-9 lamellae under the 4 th toe; (11) ca. 6-8 tail segments bearing 6 tubercles; (12) 8 preanal pores in one continuous row in males; (13) subcaudals enlarged; (14) in alcohol dorsum brownish grey with a pattern of more or less conspicuous dark transverse bands starting on the nape, tail with 9 dark brown transverse bands. Comparison. Hemidactylus ulii sp. n. can be distinguished from the other members of the 'Hemidactylus saba species group' and from all other congeners distributed in the region by the following combination of characters (see also  . 4-7), and lower number of lamellae under the 1 st (5-6 vs. 7-8) and 4 th (8-9 vs. 10-13) toe.
From H. flaviviridis by its smaller size (maximum SVL 40.4 mm in males, 40.7 mm in females vs. up to 90 mm [Anderson (1999); sexes not distinguished]), the presence of enlarged dorsal tubercles, and the absence of femoral pores in males.
From H. sinaitus by the presence of enlarged tile-like subcaudals and in having separated uppermost nasals (60% vs. 9% of specimens).
Description of holotype. NMP6V 74833/2, adult male. Body slightly depressed to cylindrical (Fig. 8). Upper labials 8/8, lower labials 7/7. Nostril between rostral, three nasals and in punctual contact with the first upper labial. Uppermost nasals separated by a small inserted shield. Mental almost triangular. Anterior postmentals large and very long, in wide mutual contact behind mental, in contact with the 1 st lower labial (left) and the 1 st and 2 nd lower labials (right) (Fig. 5). Posterior postmentals smaller, in contact with the 1 st and 2 nd (left) and the 2 nd (right) lower labial. Eye moderate (E/HL=0.24). Supraciliar granules with prominent projections, which form a comb-like structure above the eyes. Parietal and temporal region covered with round pointed regularly distributed tubercles. Ear opening oval. Dorsum with 14 longitudinal rows of enlarged, prominent, caudally pointed tubercles bearing distinct longitudinal keels. Thighs and lower legs with scattered enlarged tubercles. Tail partially regenerated from about half of its original length (estimate), original part relatively thick without basal constriction. Conical and keeled tail tubercles on tail segments forming regular whorls. Each whorl separated from the next one by four small scales. Subcaudals enlarged, tile-like. Regenerated part of the tail with small uniform scales without tubercles. Lamellae under the 1 st toe 6/6, lamellae under the 4 th toe 8/8. Eight preanal pores, no femoral pores or enlarged femoral scales.
Measurements (in mm): SVL 40.4, HL 11.5, HW 8.6, HD 5.2, E 2.8, AG 16.2. Coloration of holotype in preservative. Overall dorsal coloration brownish grey. An indistinct dark horizontal stripe in loreal and temporal area. Seven dark brown transverse bands across the nape and body, the one in scapular region being the most conspicuous. Dark brown bands also on the original part of the tail. Belly whitish.
As already mentioned (Results), the level of genetic variability within H. ulii sp. n. is very high. The species is divided into three well supported sublineages which reflect the geographic origin of the samples. Although there is a certain geographic separation corresponding with these sublineages, the exact limits are not distinct and also morphological variation among paratypes is not congruent with geography.
Etymology. The species epithet "ulii" is a patronym for Prof. Ulrich Joger, a German herpetologist known as Uli among friends, in recognition of his important contribution to the knowledge of the herpetofauna of the Western Palearctic.
Distribution and ecology. Hemidactylus ulii sp. n. is known from inland midaltitude areas (292-1182 m) of southwestern Yemen (Fig. 6). Most specimens were collected in open dry wadis with scattered rocks and boulders, in stony deserts and also in the vicinity of villages in gardens and irrigated cropland fields.
The following reptile species were found to occur in sympatry with H. ulii: Bunopus spatalurus Anderson, 1901

Discussion
Previous phylogenetic studies of the Arid clade of Hemidactylus disclosed an extraordinarily rich diversity within this genus in the Arabian Peninsula (Moravec et al. 2011;Carranza and Arnold 2012;Šmíd et al. 2013). The latter work, besides of showing the phylogenetic relationships among individual species of the Arid clade, highlighted the high level of genetic differentiation and existence of several yet undescribed taxa within this genus. The 'Hemidactylus saba species group' as defined herein represents one of the monophyletic groups within the Arabian radiation. All three species forming this group -H. granosus, H. saba, and H. ulii sp. n. -are well defined and distinguishable both genetically and morphologically from each other, as well as from other Hemidactylus species that occur in the same area. Geographically, H. saba and H. ulii sp. n. are confined to the foothills and submontane areas of southwestern Yemen, where they occupy mid-altitude elevations (292-1182 m in H. ulii sp. n., 1180 m in H. saba). In comparison, H. granosus has a much wider distribution, spanning from northeastern Egypt to central Saudi Arabia. It was found from the sea-level up to almost 1600 m in the Asir Mountains, which stretch along the eastern Red Sea coast of the Arabian Peninsula. Its occurrence in eastern Egypt is also likely based on observations of Baha El Din (2005,2006), who reported morphologically variable populations of H. turcicus (sensu lato) in these regions attribut-able to H. granosus (see Distribution and ecology). The distribution of H. granosus in the coastal Sinai and Saudi Arabia near important marine junctions together with the genetic uniformity of this species indicates extensive gene flow between these populations. It may be the result of recent colonization event(s), their inadvertent human-mediated transportation or perpetual contact of populations in a continuous range. The continuous range of H. granosus along the Hijaz and Asir Mountains in western Arabia confirms that these mountain ranges can serve as a corridor providing connection between the eastern Mediterranean and southern Arabia (Scott 1942;Gvoždík et al. 2010).
The highlands of southwestern Saudi Arabia and Yemen are known to host a high number of endemic taxa (Balletto et al. 1985;Arnold 1986;Gasperetti 1988;Harrison and Bates 1991;Gasperetti et al. 1993). The genus Hemidactylus also shows a high rate of speciation and endemicity in the area. Currently, there are eight species and one subspecies known from the Yemen highlands, which makes Hemidactylus one of the most specious reptile genera in the area (Fritz and Schütte 1987;Busais and Joger 2011b;Šmíd et al. 2013;Uetz 2013). As new genetic and morphological data are becoming available from Arabia even more new species are to be expected (Moravec et al. 2011;Šmíd et al. 2013), thus fulfilling the prognosis of Baha El Din (2005) and the models of Ficetola et al. (2013) which suggested that the Red Sea region is likely to contribute significantly to the diversity of Hemidactylus.