HERPETOLOGICAL JOURNAL, Vol. 5, pp. 271-280 (1995)

CONTRIBUTION TO THE SYSTEMATICS OF THE LIZARD ACANTHODACTYLUS ER YTHR UR US (SAURIA, LACERTIDAE) IN MOROCCO

JACQUES BONS AND PHILIPPE GENIEZ

Laboratoire de Biogeographie et Ecologie des Ve rtebres, Ecole Pratique des Ha utes Etudes, Un iversite Montpellier II, 34095 Mo ntpellier Cedex 05, France

We have analysed several scalation characters and the geographic distribution of lizards of the Acanthodactylus erythrurus group to verify the validity of these criteria. These data are collated with biogeography to demonstrate the existence of two distinct species within what are known as common fringe-toed lizards: Acanthodactylus erythrurus, consisting of three subspecies, and Acanthodactylus lineomaculatus, monotypic and endemic to Morocco. Hypotheses concerning the population history of these animals are proposed.

INTRODUCTION coast. However, it does not appear to be known from Tunisia. At the beginning of the last century common fringe­ Morphologically, it is characterised by the presence toed lizards were described from Europe (Lacerta of three complete rows of scales round the fingers and erythrura Schinz 1833, and Acanthodactylus vulgaris small dorsal scales, either keeled or unkeeled, on the Dumeril & Bibron 1839) and from North Africa back, with the underside of the tail bright red in juve­ (Acanthodactylus lineomaculatus Dumeril & Bibron niles and subadults. The combination of these three 1839). Since then, they have been the subject of many characters distinguishes Acanthodactylus erythrurus investigations. Boulenger' s revision in 1878 was the within the genus. Three subspecies have been recog­ first attempt to introduce some order to the genus nised in Morocco (Pasteur & Bons, 1960; Bons & Acanthodactylus, often considered as one of the most Girot, 1962): difficult in the Palearctic. The increase in the number - ssp. belli Gray 1845, in the Rif, on the coast and of specimens collected, and the use of often contradic­ plateaux east of the Atlases, the southern slopes of the tory terminology by different authors, led Salvador High Atlas and the far west of the Anti-Atlas (this sub­ (1982) and Arnold (1983) to make a revision of the species is the only representative of Acanthodactylus numerous species that this genus comprises and to erythrurus in Algeria; Salvador, 1982). clarify the phylogenesis. Other authors have confined - ssp. atlanticus Boulenger 1918, in the Middle their research to a single group, e.g. Squalli-Houssaini Atlas, the northern slopes of the High Atlas and the (1991) on the fringe-toed I izards of the erythrurus plains situated north of the High Atlas (endemic to group. Morocco). Since we had access to abundant material from - ssp. lineomaculatus Dumeril & Bibron 1839 on North Africa and precise data on these animals' ecol­ the Atlantic coast from Tangiers to Essaouira (ende ic ogy and distribution in Morocco, our intention was to � to Morocco). readdress the problem, restricting our analysis to cer­ The nominate subspecies is confined to the Iberian tain scalation characters that, in our opinion, are peninsula. discriminative. These characters were then collated The two recent revisers of the genus, Salvador with the ecological and biogeographical characteristics ( 1982) and Arnold( 1983 ), do not retain the subspecies of the populations being studied. We restricted our study to the different forms of the common fringe-toed atlanticus. They consider it as intermediate between A. erythrurus belli and A. erythrurus lineomaculatus. lizard in Morocco because it is the country where the . Whereas Squalli Houssaini (1991) , without adopting a widest diversity of forms and habits is found. definite position, considers that the Moroccan subspe­ The common fringe-toed lizard, Acanthodactylus cies have little taxonomic value and are only a erythrurus (Schinz 1833), is the only representative of reflection of their distribution, and that the Iberian the genus occuring in Europe, where it is confined to fringe-toed lizards are sufficiently differentiated from the southern two thirds of the Iberian Peninsula ex­ their Moroccan counterparts to merit a distinct specific tending its range as far north as Gerona (Barbadillo status. Escriva, 1987), Zaragoza, Burgos (old record) and Leon (old record) (Salvador in Bohme, 1981 ). It is also MATERIALS AND METHODS the only fringe-toed lizard to occupy the whole of Mo­ We examined 496 Moroccan individuals from 22 rocco, north and west of the Atlases. This distribution localities, or groups of localities, throughout the coun­ stretches eastwards along a large part of the Algerian try. For comparative purposes, we added 11 individuals 272 J. BONS AND P. GENIEZ

,• t . SPAIN I + I

ALGERIA e Mamora Forest e Zaen; Forest (franc e e Region of Lakes • Azrou Tarmilate, Oulm�s e • Tioumliline, Agdal Plateau e Doukkala

MO ROCCO

e Haouz, Marrakech Essaouira

• • Jbel Saghro Jbel Siroua e 40 km after Tazenakht - Taroudannt

0 100 200 km. Oued Noun

FIG. I. Location of populations studied. from Andalusia (Spain) (Fig. 1). The samples varied in Jbel Siroua (2 individuals) size depending on the localities, ranging from 1 to 69 Oued Noun ( 1 individual) specimens (average : 22.55). The examined samples are as follows : SPAIN Benidorm (1 individual) MOROCCO Almeria (2 individuals) Larache (47 individuals) Hues car (2 individuals) Mehdiya, Kenitra (36 individuals) Torre de la Higuera (3 individuals) Mamora Forest ( 41 individuals) unknown locality (3 individuals) Rabat-Agdal (55 individuals) Temara, Sables d'or (14 individuals) Five scale characters were monitored: (1) dorsal Zaers Forest (47 individuals) scales strongly keeled or not; (2) position of the Casablanca (47 individuals) subocular in relation to the edge of the upper lip; (3) Doukkala (4 individuals) number of scales and granules arising from the frag­ Essaouira (6 individuals) mentation of the fi rst supraocular, right and left; ( 4) Region of Lakes ( 5 individuals) number of interprefrontal granules; and (5) internasal Ifrane (3 4 individuals) divided or not. Azrou (17 individuals) We then compared the percentage of individuals Tioumliline, Agdal Plateau (9 individuals) within each population that presented different con­ Tarmilate, Oulmes (69 individuals) figurations of each character. Haouz, Marrakech (3 individuals) The specimens examined come from the collection Rif (18 individuals) of the Laboratoire de Biogeographie et Ecologie des Eastern Morocco (13 individuals) Vertebres de l'E.P.H.E., Montpellier, France. The re­ Eastern High Atlas ( 13 individuals) sults of these examinations were collated with Jbel Saghro ( 4 individuals) numerous observations made in the field in Morocco 40 km W. of Tazenakht (11 individuals) and the Iberian Peninsula. SYSTEMATICS OF ACANTHODACTYLUS 273

( SPAIN e SPAIN �

ALGERIA \ .�o�\ ALGERIA •�-�JI e -,-.ii0 MoRocco �- -•I .- -� MOROCCO • 0 ·�· j . �j= • .. ' ...• •• 00 o •

FIG. 2. Texture of dorsal scales, expressed in percentages of individuals per site.

RESULTS The results of the pholidotic analyses are ordered according to scale characters.

DORSAL SCALATION (FIG. 2) 2 An examination of the dorsal scales reveals a parti­ tion of our samples into two groups: (I) lizards from FIG. 3. Position of subocular scale, expressed in percentages the Atlantic coastal area, between Tangiers and of individuals per site. Essaouira, possess strongly and sharply keeled dorsal scales, starting on the back and sides of the neck; (2) Position 4 : The subocular is in wide or narrow contact lizards from all other localities (Morocco and Spain) with the upper lip. This subocular/edge of lip contact is possess smooth scales on the anterior part of the back. characteristic of the form be/Ii. In some populations they become tectiform or weakly The analysis of our samples shows that only animals keeled at the rear of the back. from the Rif, the southern slopes of the High Atlas and the We did not observe any individuals presenting in­ east of Morocco are 100% consistent with position 4. termediate characters between these two types of Position 3 is characteristic of animals from the Mid­ scalation, nor any mixed populations. Consequently, dle Atlas and the Haouz plain, in percentages varying strongly keeled dorsal scales make it possible to distin­ from 87 to 100%. The other situations are infrequent in guish with certainty Atlantic coast animals fromall the these regions: 0 to 8.7% of position 1, 0 to 20% of posi­ other common fringe-toed lizards examined from Mo­ tion 4. rocco and Spain. A later examination of a sample (c. The coastal animals are distributed along a band 30 individuals) from the Aures in Algeria, put at our running north-east/south-west. In localities north of disposal by Laurent Chirio, also confirmed our find­ Rabat they mostly (75 to 95%) possess a small ings for Algeria. trapezoidal upper labial (position 2) that is usually ab­ sent (57.1 to 95%) in specimens from south of Rabat POSITION OF THE SUBOCULAR (FIG. 3) (position 1 ). Four positions of this scale in relation to the upper Consequently the traditional criterion of the posi­ lip were observed. This is the most commonly used tion of the subocular used to distinguish subspecies of character for the recog1.1ition and distinction of the Acanthodactylus erythrurus should be used with care. North African subspecies (Bons & Girot, 1962). Position 4 certainly enables animals from the Rif, east­ Position 1: The subocular is wedged between the 4th ern Morocco, the southern slopes of the High Atlas and and 5th upper labials, but does not touch the lip (labials the far west of the Anti-Atlas, (i.e. belonging to the 4 and 5 are still in contact). This is characteristic of the subspecies be/Ii), to be identified unequivocally. How­ form lineomaculatus. ever, positions 1 and 3, that were traditionally used to Position 2: The subocular is separated from the lip by a identify respectively the lineomaculatus and atlanticus small trapezoidal scale that appears to result from the forms, are not absolute and cannot be used for the de­ fragmentation of the 4th supralabial. termination of all individuals : Position 3: The subocular is separated from the lip by a Position 3 (small rectangular supralabial) is peculiar to small "independent" rectangular scale. This position is the form atlanticus, but a small percentage of individuals characteristic of the form atlanticus. from the Middle Atlas present situations 1 and 4. 274 J. BONS AND P. GENIEZ

FIG. 4. Number of fragments of the 1 st supraocular scale, FIG. 5. Presence or absence of interprefrontal granules, expressed in percentages of individuals per site. expressed in percentages of individuals per site.

Position I, characteristic of lineomaculatus, can be On the other hand, the absence of interprefrontal used to identify a large proportion of animals from the granules is the most common situation in animals from Atlantic coastal area south of Rabat, but becomes infre­ the rest of Morocco and Andalusia (80 to 100% of indi­ quent north of Rabat. It is replaced by position 2 viduals, depending on localities, with values always (fragmented supralabial), a configuration that may higher than in the Atlantic localities). Some animals sometimes be confusedwith position 3. from the Rif, the Middle Atlas, the East and Spain have one granule (5.9 to 20% of individuals depending SUPRAOCULAR SHIELDS (FIG. 4) on localities), while it is absent in all specimens exam­ The genus Acanthodactylus is characterized by the ined fromthe High Atlas, Haouz and the far west of the presence of four large supraocular shields that have a Anti-Atlas. tendency to fragmentation. In common fringe-toed liz­ The presence of one or two interprefrontal granules, ards, only shields 2 and 3 remain entire, while the 4th, characteristic of the form lineomaculatus, is therefore and to a lesser extent the I st, are fragmented into gran­ predominant in animals from the coast south of Rabat ules or small scales. and infrequent in the north. It is rare or exceptional in The Atlantic samples are the most affected by this animals from other areas. fragmentation, since between 85. 7% and I 00% of DIVISION OF THE INTERNASAL (FIG. 6) specimens, depending on localities, have the 1 st supraocular fragmented into over fivescales and gran­ The division of the intemasal is a recognition char­ ules on at least one side. (From 50 to 100% of acter of the subspecies lineomaculatus. individuals possess over five scales in the place of the With the exception of those from the extreme north 1 st supraocular on both sides). (Larache and Mehdiya), the majority of individuals In samples from the Rif, the Atlases, western Mo­ from theAtlantic coast have a divided intemasal (66.7 rocco and Iberia, the 1 st supraocular is nearly always to 85.7%, depending on localities), whereas the other fragmented into fewer than six scales on both sides populations show few cases of intemasal division (from 76.9 to 100% of individuals, depending on lo­ (Middle Atlas, 2.9 to 5.9%, depending on localities), calities). or none (Anti-Atlas, High Atlas, Haouz, Rif, East and The division of the I st supraocular into more than Spain). five fragments is an identifying feature for individuals All the fringe-toed lizards from the Atlantic coast from the Atlantic coast. However, this situation is also between Tangiers and Essaouira, i.e. those that always found, though in very low percentages, in Andalusia, present strongly keeled scales, when compared to the the Rif, the Middle Atlas and the East. other forms, also have a much more pointed snout, a more slender head, a sharper angle between the pileus EXISTENCE OF INTERPREFONTAL GRANULES (FIG. 5) and the cheeks, more angular lateral edges to the fron­ In some fringe-toed lizards, one or several granules tal, which is markedly concave, and variably keeled are intercalated between the prefrontal scales. These temporal scales. Squalli-Houssaini (1991) adds other granules exist in 75 to 95.8% of individuals from the distinctive characters such as the number of rows of Atlantic coast, except those from the extreme north supraciliary granules (two for lineomaculatus, one for (Mehdiya, 47.2%; Larache, 25%). the other tax ons of the group), and the frequentpres- SYSTEMATICS OF ACANTHODACTYLUS 275

claims of Salvador (1982) and Arnold (1983). Acan­ thodacty/us erythrurus at/anticus is therefore a valid subspecies perfectly consistent with the definition of a subspecies. We then investigated the nature of contacts and re­ lations between these different subspecies. A large number of specimens, observed in many localities (cf. Fig. 2), enabled us to make the followingclarifications. Contact zones between the forms atlanticus and be/Ii exist in the northern foothills of the Middle Atlas. The transition from one subspecies to the other occurs via a mosaic of small populations in which all the indi­ viduals possess either the labial character of atlanticus (position 3) or of bel/i (position 4). Thus, 5 km north of Dayet Ifrah (region of lakes, north of the Middle At­ las), all individuals of Acanthodactylus erythrurus possess the headscale pattern of be/Ii, while the sur­ FIG. 6. Division ofinternasal scale, expressed in percentages of rounding populations belong to the subspecies individuals per site. atlanticus. We have no knowledge of contacts between ence ( 41%) of occipital granules in the former and atlanticus and belli along the ridge of the High Atlas, their rarity in the others. These characters were not even though the two subspecies are separated by only a monitored during our analyses but they have been few kilometres. The very high altitude of the High At­ largely confirmedby our studies in the field. las mountains may mean that the supposed contacts do DISCUSSION not exist, the two subspecies being allopatric in the southern part of their range. It is therefore established that the head scalation cri­ We know several areas where populations of teria used to identify the subspecies of Acanthodactylus lineomacu/atus and be/Ii or atlanticus live in immedi­ erythrurus should not be taken into account if used in ate proximity. No morphologically intermediate isolation or on a single specimen. On the other hand, if animals have been observed. This apparent absence of applied together, they are very interesting and easy to intermediates between the two forms,together with di­ use. They also make it possible to show up three geo­ agnostic morphological differences (dorsal scalation, graphic regions, which are occupied by three head shape) and other important differences (position morphologically different types. These correspond to of subocular, divided intemasal, shape of frontal) indi­ the three traditionally accepted subspecies of Morocco: cate that we are dealing with two distinct species: ssp. /ineomaculatus, ssp. be/Ii and ssp. atlanticus (Pas­ Acanthodacty/us lineomaculatus on the one hand, and teur & Bons, 1960; Bons & Girot, 1962). In spite of the Acanthodactylus erythrurus, represented in Maghreb total confirmation of these characters and percentages by the subspecies be/Ii and atlanticus, on the other. by Squalli-Houssaini (1991), this author could not Within the range of Acanthodactylus lineomaculatus, adopt the same conclusions as ours because of the low disparity in several characters is observed between ani­ number of populations studied. mals from south of Rabat and those from the north: Concerning the form atlanticus, it should be noted divided intemasal, number of interprefrontal granules, that it is not recognised by Salvador (1982) and Arnold fragmented 4th supralabial. However these variations (1983). They consider it an intermediate between A. are of a clinal nature, with the transition consisting of a erythrurus belli and A. erythrurus lineomaculatus. Yet progressive inversion of frequency, so the town of the homogeny of its scalation characters, distributed Rabat is only a reference mark. Consequently, it is im­ over a well defined geographical area (the Middle At­ possible to establish the existence of two subspecies for las, northern slopes of the High Atlas and Haouz), A. /ineomaculatus. Besides, other characters are in­ make it quite different from the other subspecies. Fur­ variable throughout the whole of this species range thermore this form, usually considered as mountainous (keeled dorsal scales, slim pointed snout, concave fron­ (Pasteur & Bons, 1960; Sons & Girot, 1962 ), is found tal, strongly fragmented lst supraocular), providing in plains at Haouz of Marrakech. The hypothesis that supplementary evidence forA. lineomaculatus specific the form atlanticus could be a mountain ecotype of status. Acanthodactylus erythrurus can therefore not be re­ Of course, there are no contacts between the nomi­ tained. Lastly, the absence of intermediate animals nal subspecies, confined to the Iberian peninsula, and between the subspecies lineomaculatus and be/Ii, as the other forms that are peculiar to the Maghreb. How­ well as the absence of intermediate morphological ever, from a strictly morphological point of view, the characters between these two forms, is contrary to the European animals are very similar to A. erythrurus 276 J. BONS AND P. GENIEZ belli and at/anticus, to such an extent that the Spanish from the lip by the 4th and 5th labials that are joined in animals, whose subocular is in contact with the lip, are this case). practically indistinguishable, apart from slight Distribution: The southern two-thirds of the Iberian colouration variations, fromA. erythrurus belli. They Peninsula, extending northwards along the Mediterra­ have a rounded snout, low headscale fragmentation nean coast as far as the environs of Gerona. and smooth dorsal scales. So A. erythrurus belli and A. Acanthodacty/us erythrurus be/Ii Gray, 1845 erythrurus at/anticus should be considered as conspecific with A. erythrurus erythrurus, in spite of Diagnosis: Smooth dorsal scales (or weakly keeled on the electrophoretic differences observed by Squalli rear of back), rounded snout, barely concave frontal, Houssaini (1991) between animals from the Iberian entire internasal, usually no interprefrontal granules peninsula and those fromMorocco. (exceptionally one), 1 st supraocular generally frag­ In fact, this author has found a DNei genetic distance mented into fewer than six scales on both sides of 0.327 between his Spanish samples (Alicante) and (exceptionally six on one or both sides), subocular in those from Morocco. This distance is higher than that contact with lip. observed between his oceanic coast population and Distribution (Fig. 7): Rif mountains and foothills,east­ ern Morocco, southern slopes of the High Atlas and far both Rifain and Middle Atlas samples (DNei = 0.081). In Busack's (1986) study of Acanthodacty/us west of the Anti-Atlas (Foum Assaka, Ifni), as well as erythrurus, the genetic distance between different Mediterranean Algeria. samples from Tingitane Peninsula (north of Morocco) Acanthodacty/us erythrurus at/anticus Boulenger, was no different from that between Tingitane and An­ 1918 dalusia (DNei = 0.10 and 0.09 respectively). The genetic similarity between Spanish and Moroccan populations Diagnosis: Smooth dorsal scales (or weakly keeled on found by Busack (1986) is in partial contradiction of rear of back), rounded snout, barely concave frontal, those of Squalli-Houssaini (1991) but agrees with our internasal nearly always whole (exceptionally di­ morphologicalresults: we observe a closer phenotypic vided), no interprefrontal granules (sometimes one, similarity between Acanthodacty/us ythruruser exceptionally two), 1 st supraocular generally frag­ erythrurus (Spain), A. e. be/Ii (northern, eastern and mented into fewer than six scales on both sides southern Morocco) and A. e. at/anticus (central Mo­ (exceptionally six on one or both sides), subocular gen­ rocco) than between these and A. /ineomacu/atus. erally separated from the lip by a small rectangular or Moreover, the weaker electrophoretic polymor­ oval scale (exceptionally this small scale is absent and phism in Spanish populations (six diagnostic alleles the subocular may or may not be in contact with the lip). for 10 individuals, DNei = 0.01) compared to one of the Moroccan populations (12 diagnostic alleles for9 indi­ Distribution (Fig. 7): Morocco only; Middle Atlas in­

= cluding the Central Plateau, northern slopes of the viduals, DNei 0.10) (Busack, 1986), suggest the Iberian peninsula was colonized from NorthAfrica by High Atlas, plains to the north and west of the Atlases, a small number of individuals. This led to a 'founder with the exception of an Atlantic coastal fringe. effect' characterized by the large genetic homogeneity Acanthodacty/us /ineomacu/atus Dumeril & Bibron, of Spanish populations. This homogeneity approaches 1839 the phenotypic uniformity observed in Andalusia. Diagnosis: Strongly keeled dorsal scales, slim pointed SYSTEMATIC REVIEW OF THE FRINGE-TOED LIZARDS snout, concave frontal (with sharply angled lateral OF THE GROUP ER YTHR UR US edges), internasal usually divided, except north of Common fringe-toed lizards are distributed in the Rabat, usually one or two granules, except north of Iberian peninsula and the Maghreb, except Tunisia. Rabat where they are generally absent, 1 st supraocular They are represented by two geographically parapatric usually fragmented into more than six scales or gran­ species, one of which is monotypical and the other ules, subocular usually separated from the lip by the comprising three subspecies. They are all character­ 4th and 5th supralabials, the 4th being usually frag­ ized by three entire rows of scales round the fingers, mented at the rear by a small trapezoidal scale in small dorsal scales, and the bright red underside of the localities north of Rabat. tail in juveniles and subadults. Distribution (Fig. 7): Moroccan endemic along the At­ lantic coast from Tangiers to Essaouira. Acanthodacty/us erythrurus erythrurus (Schinz, 1833) Acanthodacty/us (erythrurus) /ineomacu/atus has been Diagnosis: Smooth dorsal scales (or weakly keeled on frequently reported from Agadir and the plain of the rear of back), rounded snout, barely concave frontal, Oued Souss. However, all fringe-toed lizards of this entire internasal, usually no interprefrontal granules area that we examined in collections or in the field, be­ (exceptionally one), 1 st supraocular generally frag­ long to the parda/is group, especially Acanthodacty/us mented into fewer than six scales on both sides busacki Salvador, 1982. Raxworthy et al. (1984) also (sometimes into six scales on both sides), subocular mentions the problem in identifying Cap Rhir' s ani­ usually in contact with lip (sometimes it is separated mals (northen Agadir) and suggests that only SYSTEMATICS OF ACANTHODA CTYLUS 277

.. t •• i....II' • . ,..i. 11 lz. � 1 _.. n L " . 'j •111 91-..' " I • • .� .� .·,. .. .llllJ.i lll� •• II"' • I ··.J I j • ..1 . :0.1"" I/ ••I • \., iJ • . 1 • • • • i 1 .. • I•...... �' I r...... I• • I • •• • w', ...... w • • � .1:1 tt T u Vl L.ti� I/ 1• ... D �n .a '5j "e o tJ ' � ... o •• u • 1. ul' 0 o be :'"' IM n'�- t1 ,. 141( ' !h i... 'l• 11 hi " ... •le -- p ..0 � ... . v w o" ' I . t . " 0 • /Al ... ( .. 101 "'"' ... 0 ..: 0 p • .. 0 0 0 In• t . 20 2< Q • ., " - fJ �· "' <.,_ m 2 1--·· - • 1r>' ;1) • • ,. Jo N lnh I I • J ,I 0 .. too ... In I I I �· JI I b • • . ' • ,,< • "� ,. " In ,__j.4i_ J "' i,.vwxv "v • Acanthodactyluseryt hrurusbellii J< r. M �·-- lio ll n • v l:n 0 Acanthodactylus erythrurusatl anticus e �1 v ., ... � IJJ ... m 114 Acanthodactylus lineomaculatus c 'o�·r'F G H I J K L M N ·a p Q R s r u ,,.

FIG. 7. Distribution of common fringe-toed lizards in Morocco (from Bons & Geniez,in press).

Acanthodacty/uspardalis s.I. was present in this area. basis of electrophoresis with the exception of So we must accept that A. lineomacu/atus reaches its lineomaculatus, which presents a more developed geographical limit 15 km south of Essaouira and that enzymatic polymorphism with a higher number of rare furthersouth it is replaced by another species, A. busacki. alleles. This taxon is foundin the plain of Oued Souss and the We have also observed that animals from the Medi­ Atlantic coast from Tamri, in the North, to Boujdour, terranean coast of Morocco, and those from both the in the South. Mediterranean and Atlantic coast of Iberia, did not display any of the morphological characters of CONCLUSION Acanthodactylus lineomacu/atus. This suggests that On the basis of several scalation characters, we have these lizards have been isolated on the Atlantic coast been able to demonstrate that there is a larger disparity for a long time and enables us to reject the hypothesis between animals from the Atlantic coast of Morocco of a single ecotypical adaptation to sandy habitats, and those from the rest of the country, than exists be­ where thermal variations are mitigated by the proxim­ tween the latter and the fringe-toed lizards from the ity of the Atlantic. This also suggests that Spain was Iberian Peninsula. This disparity is largely confirmed colonised by a small group of individuals from Mo­ by the morphological analysis of Squalli Houssaini rocco. This idea is reinforced by Busack ( 1986) who (1991) but is contrary to her genetic findings. Thisau­ obtained an insignificant genetic distance (DNei = 0.01) thor used electrophoresis, to show that the Iberian between his different Iberian samples and found for animals she studied differ from those fromMorocco in Spanish samples only six diagnostic alleles for 10 indi­ two diagnostic loci (FUM and ME. I) and by a fixed al­ viduals, as opposed to 12 diagnostic alleles for nine lele [LDH-2(90)] that is not shared by their Moroccan individuals in northern Morocco. This type of biogeo­ counterparts. On the other hand, the three Moroccan graphical scenario is found in other species, for forms are barely distinguishable from each other on the example Podarcis hispanica, Lacerta lepida and 278 J. BONS AND P. GENIEZ

Ma croprotodon cucullatus (cf. Busack, 1986; Table in the East of the Atlas (and in Algeria), where there 3). are cold or cool winters (sensus Brignon & Sauvage, The presence of Acanthodactylus erythrurus 1962). The same species group also includes an en­ atlanticus in the plains to the north-west of the Atlases demic Moroccan species, Acanthodacty/us busacki, suggests the recent colonisation of this low altitude which is restricted to the oceanic coast from Tamri to habitat by a form that could have been differentiatied Boujdour and the plain of Oued Souss, an area with by long isolation in high mountains and selective pres­ hot winters. sures associated with these extreme conditions. It is Within the scutellatus group, Acanthodactylus conceivable that this colonisation is still taking place dumerili is linked to the Saraha's sand dunes with tem­ and that this form could come into contact with A. perate winters. On the other hand, Acanthodacty/us /ineomaculatus. Would this develop into competition aureus, also in the scutellus group, occurs from the Sa­ between the two forms, or even lead to the elimination hara's Atlantic littoral sands (from Agadir to Senegal), of the coastal form with its more demanding ecological where winters are hot. requirements? Alternatively, will the present situation These patterns are also observable in other reptile stabilise, with the sandy substrate and the presence of a genera (in particular Saurodactylus, Chalcides and well-adapted species to this environment proving an Sp henops; cf. Table 1 ), as well as some mammals insuperable barrier for A. erythrurus atlanticus? (Aulagnier & Thevenot, 1986), members of the Another hypothesis can be put forward to explain Gerbillidae belonging to the genus Gerbil/us and the presence of the form atlanticus in the plains. This Gerbil/us pyramidum group. They are represented in subspecies could have been much more widespread Morocco by a Saharan and sand-loving species, during the pluvial periods and the last glaciations than Gerbil/us pyramidum, located in all the Sahara, and at present. The general warming and aridification that fourendemic species in sand fromboth the Moroccan have led to the present climate could have reduced the oceanic coast and western Sahara (from the north to range of atlanticus to small isolated populations, in the south, Gerbil/us hesperinus, G. hoogstral/i, G. which case this form would be retreating in the plains occiduus and G. riggenbachi) where there are hot win­ to the north-west of the High Atlas. ters. Patterns of interspecific diversity within other spe­ Caputo et al. (1993) used Sue's works (1984, 1989) cies groups which are comparable to Acanthodactylus to explain the history of the specific distinction be­ erythrurus are known from Morocco. In the tween Sp henops sepsoides and S. sphenopsiformis. Acanthodactylus genus, two species groups are known These authors state that "these two species may have (Salvador, 1982; Bons & Geniez, in press). The diverged as a result of allopatry during the drastic cli­ pardalis group contains one species, Acanthodacty/us matic fluctuations of the Plio-Pleistocene. During the maculatus, which has a wide extension in arid steppes most mesic climatic periods, the pluvial phases in the

TABLE 1. Bioclimatic characteristics of ranges of several vicariant species. end., endemic to Morocco; temp., substage with

temperate winter; * * , species well represented in substage; • , marginal species in substage.

Atlantic sides of Eastern and southern Morocco & W. Sahara sides of Morocco

Species endemic? hot temp. cool cold hot temp. cool cold

A. erythrurus N • • • •• •• A. lineomaculatus y •• •• A. maculatus N • •• •• A. busacki y •• • A. dumerili N •• •• A. aureus N •• Sp h. bou/engeri N •• Sp h. sphenopsif. N •• Saur. mauritanicus N •• •• Saur. brosseti y •• •• • Ch. ocel/atus N •• •• •• Ch. polylepis y •• •• •• * Ch. minutus ? •• •• • • Ch. pseudostriatus y • •• • • • SYSTEMATICS OF ACANTHODACTYL US 279

Saharan region (corresponding to high-latitude Nicolas Privat for the map treatment, and Veronique glaciations) would have caused the contraction of the Jallageas forthe re-reading of the English manuscript. once more-or-less continuous desert into separate arid Translated from the French original by Elizabeth refuges". The explanation given by these authors may Guillosson. also apply to taxa listed in Table 1. REFERENCES These examples illustrate the biological originality of the north-west African Atlantic littoral region. This Arnold, E. N. (1983). Osteology, genitalia and originality is reinforcedby the existence of a cortege of relationships of Acanthodactylus (Reptilia: species endemic to this area. Pelobates varaldii Lacertidae). Bull. Brit. Mus. Nat. Hist. (Zo o/.) 44, (Amphibia, Anura, Pelobatidae), Geckonia chazaliae 291-339. (Reptilia, Sauria, Gekkonidae), Chalcides mionecton Axelrod, D. I. (1978). Evolution and biogeography of (Reptilia, Sauria, Scincidae) and Crocidura tarfayensis madrean-tethyan sclerophyll vegetation. Ann. (Mammalia, Insectivora, Soricidae) are some exam­ Missouti Bot. Gard. 62, 280-334 ples of this cortege. These last species do not present Aulagnier, S. & Thevenot, M. ( 1986). Catalogue des oriental or non coastal vicariance. Mammiferes sauvages du Maroc. Trav. Inst. Sci. In the Miocene, the climate of palearctic Morocco Rabat.. ser. zoo/. 41, 163 was arid, with vegetation consisting of mainly Barbadillo-Escriva, L. J. (1987). La guia de Incafo de /os sclerophyll forests (Axelrod, 1978). This period coin­ anfibios y reptiles de la peninsula Iberica, is/as cided with the start of orogenic movement leading to Baleares y Canarias. Incafo ed., Madrid, 694 p. the formation of the Atlas mountains. We believe that Bernardi, G. (1971). L'espece et ses subdivisions du these two events played a major role in the differentia­ point de vue de la taxonomie evolutive. Proc. XIII0 tion of endemic species. This differentiation was int. Congr. Ent. Moscou 1, 112. accelerated by alternate pluvial and dry periods which Bons, J. & Girot, B. ( 1962). Cle illustree des Reptiles du . isolated populations in either mountains or plains. Maroc. Trav. Inst. Sci. Cherif , ser. Zool. 26, 1-64. Moreover, Morocco is the only north-west African Boulenger, G. A. ( 1878). Sur Jes especes country which has both Mediterranean and Atlantic d'Acanthodactyles des bords de la Mediterranee. coasts. The arid depression of Oued Moulouya and the Bull. Soc. Zoo/. Fr. 3, 179-1 97. Sahara Desert provide further barriers to the east and Brignon, C. & Sauvage, Ch. (1962-1963). Carte des south respectively. Populations tend to be separated by etages bioclimatiques in Atlas du Maroc. Comite de these geographical partitions. In addition, Morocco it­ Geographie du Maroc, Rabat, pl. 6B. self can be divided into nine distinct geographical Busack, S. D. (1986). Biogeographic analysis of the units: the Rif; the Oued Souss valley; the Middle Atlas; herpetofauna separated by the formation of the strait the Mediterranean coastal fringe between Melilla and of Gibraltar. National Geogr. Res. 2, 1, 17-36. Oran; the High Atlas; the Hauts Plateaux; the Anti-At­ Caputo, V. ( 1993 ). Taxonomy and evolution of the las; the Sahara; and the Atlantic plains. Chalcides chalcides complex (reptilia, Scincidae) This geographical partitioning has induced an ex­ with description of two new species. Boll. Mus. reg. ceptional diversification of the Moroccan herpetofauna Sci. nat. Torino, 11, 1, 47-120. (I 05 species of amphibians and reptiles, of which 22 Caputo, V., Odierna, G. & Aprea, G. (1993). are endemics (Bons & Geniez, in prep.). The fringe­ Karyological comparison of Sp henops sepsoides, toed lizards of the erythrurus group follow the same Chalcides chalcides and C. ocellatus (Reptilia: pattern,with one endemic species on the oceanic coast Scincidae): Taxonomic implications. Copeia 4, 1181- (A canthodactylus lineomaculatus) and another one 1184 living in the rest ofpalearctic Morocco (A . erythrurus). Mayr, E. (1974). Populations, especes et evolution. The latter is split into two subspecies: one, an endemic Hermann, Paris, 496 pp. (subsp. atlanticus) in the mountains and the hills of the Monroe jr., B. L. ( 1982). A modern concept of the Atlantic side, the other one found everywhere else in subspecies. Auk 99, 608-609. the Moroccan mountains. With nine species, including Pasteur, G. & Bons, J. (1960). Catalogue des reptiles two endemics in Mor9cco (Salvador, 1982), the actuels du Maroc. Trav. Inst. Sci. Cherif., ser. zoo/. Acanthodactylus genus is less diversified than the 21, 1-132. Chalcides genus which has no less than 11 species, Raxworthy, C. J., Rice, S., Smith, D. & Claudius, F. eight of which are endemic to Morocco (Mateo et al. , 1983, 1984. A study of the Reptile Fauna at Cap Rhir, in prep.). Morocco. University of London & Natural History Society , 77 pp. ACKNOWLEDGEMENTS Salvador, A. ( 1981 ). Acanthodactylus erythrurus (Schinz The authors would like to express their thanks to 1833). Europaischer Fransenfinger. In Bohme, W., Philippe Roux and Laurent Chirio for their original Handbuch der Reptilien und Amphibien Europas, 1 data and the samples they kindly provided, as well as (Echsen I), 3 76-3 88. Wiesbaden: Akademische Claude P. Guillaume, Stephane Boissinot and Tristan Verlagsgesellschaft . Guillosson for their advice on evolutionary systematics, 280 J. BONS AND P. GENIEZ

Salvador, A. (1982). A revision of the lizards of the genus Sue, J. P. ( 1989). Distribution latitudinale et etagement Acanthodactylus (Sauria: Lacertidae). Bonn. Zoo/. des associations vegetales au CenozoYque superieur Monogr. 16, 167 p. dans l'aire ouest-mediterraneenne. Bull. Soc. geol. Squalli-Houssaini, H. (1991). Systematique et biogeographie Fr, 5, 541-550 evolutive du complexe Acanthodactylus erythrurus (Reptilia, Lacertidae). These de doctorat, Marseille, 191 pp. Sue, J. P. (1984). Origin and evolution of the Mediterranean vegetation and climate in Europe. Nature 307, 429-432 Accepted: 9. 7.94