Studies on African Agama VI. Taxonomic status of the West African Agama (Sauria: Agamidae) with prominent tail crests: Agama boulengeri Lataste 1886, Agama insularis Chabanaud, 1918 and Agama cristata Mocquard, 1905

2009·8.1 MB·English

by  Philipp Wagner

Bonnerzoologische Beiträge Band 36 Heil 4 Seiten 239-253 Bonn, November 2009
Studies on African Agama VI.
Taxonomic status of the West African Agama (Sauria: Agamidae)
with prominent tail crests: Agama boulengeri Lataste 1886,
Agama insularis Chabanaud, 1918 and Agama cristata Mocquard, 1905
Philipp Wagner'*, Ivan Ineich-, Adam D. Leaché\ Thomas M. Wilms-*, Sébastien Trapeé
Wolfgang Böhme' & Andreas Schmitz^
' Zoologisches Forschungsmuseum A. Koenig, Adenauerallee 160, D-53113 Bonn, Gennany
[email protected]; [email protected]
2Museum national d'Histoire naturelle, Departement de Systematique et Evolution (Section Reptiles) UMR
7205 CNRS "Origine, Stmcture et Evolution de la Biodiversite", CP n° 30-25 rue Cuvier,
F-75231 Paris Cedex 05, France.
3Genome Center & Department ofEvolution and Ecology, University ofCalifornia, Davis, CA 95616 USA.
4Zoologischer Garten Frankfurt, Bemhard-Grzimek-Allee 1, D-60316 Frankfurt, Germany
5Université de Montpellier II, UMR 5119 Ecolag, IRD-CNRS-UM2-IFREMER, cc093,
place Eugene Bataillon, F-34095 Montpellier, France.
^Muséum d'histoire naturelle, 1 route de Malagnou, CH-1208 Geneva, Switzerland
corresponding author
Abstract.TiiispublicationreviewsthetaxonomyofthreeWestAfricanAgama species,A. boulengeri,A. cristata, and
A. insularis, each characterizedby aprominent tail crest in adult males. Followingtheresults frommorphological and
genetic analyses,Agama insularis isrecognizedasa synonym ofthe revalidatedAganwcristata, whereasthis species
isclearlydistinct fromAgama boulengeri. Wepresentadetaileddistribution map fortheseAgama species, aswell as
forA. weidholzi. Followingtheresults ofrecentpublications,Agama atraknobeliisherein regardedasa full species.
Key words.Africa, Guinea, Los Islands, lie de Roumé; Sauria,Agamidae,Agama cristata,A. insularis,A. boulengeri,
A. knobeli,Agamaweidholzi.
Introduction
Despite an increasing amount ofwork on the taxonomy materialcoupledwith inadequatedescriptionsoftheadult
and phylogeny ofAfrican lizards in the genus Agama coloration in life prevents formal taxonomic evaluations.
(BÖHME et al. 2005, Padial 2005, Wagner2007, Wagn-
er et al. 2008a. Wagner et al. 2008b. Wagner et al. ThreespeciesinWestAfricaarecharacterizedbypossess-
2009a, Wagner et al. 2009b) several species are only ing a prominent tail crest in adult males: Agama boulen-
poorlyknown,andthevalidityofsometaxaremainsques- geri,A. insularisandA. cristata. Thetaxonomic statusof
tionable (e.g. A. bocourti Rochebmne, 1884; A. bottegi these taxa is the focus ofthis study.
Boulenger, 1897; A. cornii Scortecci, 1928; A. insularis
Chabanaud, 1918 andA. agamaboensisMonard, 1940 [re- Agamainsularis,wasdescribedbyChabanaud(1918)on
garded as a synonym of A. weidholzi by Grandison thebasis ofajuvenile anda subadult female collected by
1969]).Manyspeciesareonlydiagnosablebasedonadult Dr. G. Bouet in 1914 on lie Roumé, which is part ofthe
male coloration (Loveridge 1933, Thys van den Aude- lies de Los archipelago located in theAtlantic Oceanjust
naerde 1963, McLachlan 1981, Böhme et al. 2005, several kilometres (ca. 8 km) offshore from Conaki-y.
Wagner 2007, W.agner et al. 2008a, Wagner et al. Guinea. In his description of A. insularis. Chabanaud
2008b), and the lack of appropriate comparative (1918) only coinparedhis new specieswithAgama kirkii
240 Philipp Wagner et al.: Studies onAfricanAgama VI. Taxonomic status oftheWestAfrican^gawa
Fig. 1. The two syntypes oíAgama insnlaris Chabanaud, 1918 (MNHN 1918.041-042).
1
Bonnerzoologische Beiträge 56 241
Boulenger, 1884 from southernAfrica and stated both as tionable, sinceitwasonly known atthistime from itstype
closely related. The only moiphological differences not- locality. Joger (1979) reasserted the validity ofA. insu-
ed were related to (1) head shapes (more elongate in A. laris; however. Moody (1980)placed^, insularis in syn-
insularis) and (2) body scalation (A. insiilaris has small- onymyofA. boulengeri,butwithoutgivingareason. Nev-
er scales, 120-130 scale rows around midbody versus ertheless, BÖHME (1985) followedJoger (1979) and stat-
about90inA. kirkii. and^. insularishaddorsalscalesthat ed that A. insularis and A. boulengeri are both valid
are smaller compared to ventral scales. species and probably closely related. Finally, Brygoo
(1988) andUlber& Barts (1997) treatedA. insularisas
Based on a larger series ofspecimens collected from the avalidspecies, althoughwithoutofferingadditional com-
type locality, which included adult specimens, Parker ments regarding this decision.
(1939) reassessedthe validity oíA. insularis (material of
the IRSNB, Appendix 1). In this publication Parker Interestingly, anotherspeciesofAgama fromWestAfrica
(1939) also compared A. insularis with A. boulengeri possessing a prominent tail crest has remained mostly
Lataste, 1886, a morphologically similar species distrib- overlooked since its description. Agama cristata was de-
utedinMauritaniaandMali. Hedistinguished^, insularis scribedbyMocquard(1905) fromasinglespecimencol-
fromA. boulengeri by its more strongly carínate scales, lected from Bomanesco (Sankaran) in Guinea by M. A.
especially on the occiput and lower surfaces ofthe tail; Chevalier, andlaterdonatedtotheMuséumnational d'his-
largernasal, separated from the rostral by a single elon- toirenaturelleinParisbyM. MauricedeRothschild. Moc-
gate scale; larger number oflabial scales, 8-9 versus 1 quard (1905) did not compare his new species with any
and the colouration ofadult males. Nevertheless, Park- otherspeciesandcharacterizeditmainlybythe largecrest
er (1939) only described the adult colouration from pre- proceedingfromthenecktothetail. Later,A. cristatawas
servedspecimens.Adultmales are brownishblackabove apparentlyregardedasasynonymoíA. sankaranicaCha-
with light specking, especially on the vertebral region; banaud, 1918byGuibé(1954)andWermuth(1967),pre-
lowersurfaces ofthe head, chest and abdomen are black- sumably because both taxa share the same type locality
ish-grey; middle ofthe guiar region is jet black. Joger (A. sankaranica: Moussaia, Sankaran, Guinea; A. crista-
(1979) described the adult male colouration ofliving A. ta: Bomanesco, Sankaran,Guinea). Later,Brygoo(1988)
boulengerias follows: (1) groundcolourpale gray-brown confinnedthe validityoíA. cristata inhistype catalogue
withtransverserowsofwhitespots; (2)tliroatdirty white presumablybecauseofthetail crest,butagainhewasnot
with longitudinal stripes, parts ofthe head gray-blue; (3) followed by Ulmer & Barts (1997).
tail gray-blue and (4) a characteristic half-moon shaped
blackbarontheanteriormarginofthe shoulder. Lambert Despite the presence of a prominent tail crest in adult
& MuLLiÉ (1998) refer, beside the tail crestofthe males, males sharedbetweenAgama cristata, A. boulengeriand
to the nuptial colouration offemales as the most striking the island endeinic species, Agama insularis, a coinpari-
difference between the sexes ofA. boulengeri. They de- son ofthese species remains to be conducted. In contrast
scribe the males as drab with uniform purple-grey to the previous three species,A. sankaranica anáAgama
colouredscalesdorsally,whereasthe females haveabril- weidholzi Wettstein, 1932 are small and solitary-living
liantyellowmid-dorsalbandwiththreeapproximatelybat- species. The latter was included in the comparison, be-
shaped transverse bands on a bright orange background. cause it is endemic to this region and so far only known
from Senegal (e.g. Wettstein 1932), Gambia (Böhme
Laurent(1947)reportedaseriesofAgamainsularisfrom 2005), Mali (e.g. Grandison 1969) and Guinea-Bissau
liesRouméandaneighboring island, lie Kassa, andmen- (Monard 1940). Additionally, Monard (1940) included
tioned that the characters given by Parker (1939) were specimensoíA. weidholziand.4. sankaranica inthetype
in agreement with his specimens. However, Laurent seriesoíA. boensis. (Grandison 1969,Böhme2005)and
(1947)describedadditionalsourcesofmoiphologicalvari- forthatreasonaclose relationtoA. sankaranica couldbe
ation, including 10to 13 precloacal pores, 8 to 10 supral- possible.
abialand7to 10infralabialscales. Becauseofdifferences
inthenumberofscalerowsaroundmidbody(lie Roumé: Ouraim isthereforetocoinparethespeciesofAgamawith
115 to 125; ile Kassa: 143 to 147) he suggested that the proininent tail crests in adult males (Agama cristata, A.
lie Kassapopulation deserved subspecific status. boulengeri and the island endemic species, Agama insu-
laris) to clarify their taxonomic status, distribution, and
The taxonomic status of Agama insularis has been phylogenetic relationships. Wealso includeotherspecies
changed since these early studies. Guibé (1954) listed/Í. ofWest African Agama in our study that lack prominent
insularis as valid in the type catalogue ofthe collection tail crests in adult males, but that are hypothesized to be
ofthe Muséum national d'histoire naturelle in Paris, al- close relatives because they occur in the same areas.
though Wermuth (1967) regarded this species as ques-
242 Philipp Wagner etal.: Studies onAfricanAgama VI. Taxonomic status ofthe WestAfricanAgaina
Fig. 2. The liolotypes ofa)Agamasankaranica (MNHN 1901.0395) andb)Agama cristata (MNHN 1901.0394).
Bonnerzoologische Beiträge 56 243
Material & Methods to compute the uncorrected pairwise distances for all se-
quences.
Thevoucherspecimensexaminedinthis studyareacces-
sioned in the following natural history collections: Col- The followingmeasurements and scale counts were used
lection ofJean Francois Trape, deposited in Institut de tocomparethedifferent species in statistical analysis(for
Recherchepourle Développement (IRD) Dakar, Senegal selectedcharacters seetable 1): Snout-vent length (SVL):
(TR); Institut royal des Sciences naturelles de Belgique measured from mental scale to cloaca; tail length (TL):
(IRSNB); Museum national d'histoire naturelle de Paris, measuredfromcloacatotipoftail;Tailcrestlength(TcL):
France (MNHN; Museum für Naturkunde, Berlin, Ger- length ofthe tail crest from midpoint ofpelvic region to
many (ZMB) and Zoologisches Forschungsmuseum tip ofcrest; head length (HL): measured from jugalc to
Alexander Koenig, Bonn, Germany. The type specimens rostral scale; headheight(HH): measuredatthejugal-pos-
oíAgama cristata and A. insiilaris were included, how- torbitalregion; headwidth(HW): measuredacrosstheju-
ever, the type specimens oíA. insiilaris arejuveniles and gal-postorbital regionjust anteriorto the external audito-
therefore topotypical material was also used forthis tax- 17 meatus; Midbodyscales(MS): scale rowsaroundmid-
on. body; tail crest scales (TcS): number oftail crest scales,
countedfrommidpointofpelvicregiontotipofcrest; dor-
Since several studieshave now shown that DNAbarcod- sal scales (DS): number of midline longitudinal dorsal
ing, especially when using the mitochondrial 16S rRNA scales, counted from midpoint ofpectoral region to mid-
gene, is a reliable tool in reptile or amphibian taxonomy point ofpelvic region; cloacal pores (CP).
(e.g. Vences et al. 2005; Bwong et al. 2009) molecular
datawere collected to calculate a simple neighbour-join- Excel2000and SPSS (10.0) softwarepackageswereused
ing phylogeny (fig. 4) and to analyse the sequence vari- to run statistical analyses. Hierarchical Cluster analysis
ationbetweenspecies.Aportionofthemitochondrial 16S and Principal Component Analysis (PCA) were used to
rRNAgene oíAgama insiilaris (ZFMK 88247; GenBank evaluate the moiphological data and to explore the phe-
accession number: GUI33326) from the type locality, as netic relationships between the taxa examined.
wellasAgamacristata(TR555; GenBankaccessionnum-
ber: GUI33325) from Guinea, Agama boulengeri
(MNHN; Gen-Bankaccessionnumber: GUI33324) from Results & Disussion
Mauritania, Agama weidholzi (ZFMK 75001; GenBank
accession number: GUI33328) from Gambia, Agama Morphology. Significantdifferenceswerefoundwhenthe
sankaranica(ZFMK84992; GenBankaccessionnumber: effect ofbody sizes was removed from analyses (see fig.
GUI33327) from an unknown locality and Agama aga- 3), butthere is still ahigh level ofoverlapamong species
ma (ZFMK 15222 [neotype]; GenBank accession num- inall PC's. Bestresults were found in PC 1 and2 (PC 1=
ber: GUI33323)weresequenced.Acanthocerciisatricol- 41.379%ofvariance; PC 2=25.825%ofvariance; PC 3=
lis (ZFMK 41748; GenBank accession number: 24.098%ofvariance; PC4= 8.699% ofvariance) andare
GU133322) was used as outgroup. visualized in fig. 3. Agama insiilaris has a significantly
larger relative head height, head width and head length
DNAwas extracted using QuiAmp tissue extraction kits than A. boulengeri, but lower average ofmid-body scale
(Quiagen) or a modified Chelex-Protocol (Walsh et al. rows and significantly smaller relative head length than
1991, Schmitz 2003). The primers 16sar-L (light chain; A. cristata.Agamacristata differs significantlyinahigh-
5'-CGC CTGTTTATCAAAAACAT-3') and 16sbr- eraverageofmid-bodyscalerowsandlargerrelativehead
H (heavy chain; 5' - CCG GTC TGA ACT CAG ATC length from both A. insiilaris and A. boulengeri. Agama
ACG T - 3') ofPalumbi et al. (1991) were used to am- boulengeri differs significantly in smaller relative head
plify aportion ofthe mitochondrial 16S ribosomal RNA height,headwidth andhead lengthfrombothA. insiilaris
gene. PCR cycling procedure was as described in andA. cristata. Males have significantly lowertail-crest-
Schmitz et al. (2005). PCR products were purifiedusing scales and tail-crest-length than eitherA. insiilaris orA.
Quiaquickpurificationkits(Quiagen). Sequenceswereob- cristata. DifferencesinpholidosisbetweenA. boulengeri
tainedusinganautomaticsequencer(ABl 377). Sequences andA. cristata are low and mostly overlapping (see table
were aligned using ClustalX (Thompson et al. 1997; de- 1), which isunsurprisingwithinthe genusAgama. But in
fault parameters) and manually checked using the origi- average, A. cristata has a higher count of scale rows
nal Chromatograph data in the program BioEdit (Hall aroundmidbodythan^. boulengeri. Also the differences
1999). Apreliminary phylogenetic tree was calculated in in colouration between A. boulengeri and A. cristata are
the program Paup* 4.0b10 (Swofford 2002) using the low. Small differences in colourationwere foundbetween
Neighbor-joiningalgorithm(NJ)and20000pseudo-repli- AgamaboulengeriononehandandA. cristataontheoth-
catestoestimatenodesupport.Thisprogi'amwasalsoused er hand. Especially the colouration ofthe throat in adult
244 Philipp Wagner et al.: Studies onAfricanAgama VI. Taxonomic status oftiieWestAfricanAgama
Agamaboulengeriisdistinctinpholidosis,colourationand
genetics from^. insularis and^. cristata. But no signif-
icant differences in moiphology andgenetics were found
betweenA. cristataandA. insularis. Itisimportanttonote
that this result is based on an examination oftype speci-
mens, topotypical material, and freshly collected materi-
c al (seeappendix). Thereforeweconsider^ga/z/a insularis
c Chabanaud, 1918 as ajunior synonym ofthe revalidated
o
Q. Agama cristata Mocquard, 1905.
E
o
_
-l.DOOOO-
(0
Q.
Oc Agama cristata Mocquard, 1905
1905Agamacristata Mocquard, Bull. Mus. Hist.Nat. 11:
Principal component2 288.
1918Agama insularis Chabanaud, Bull. Mus. Hist. Nat.,
Fig. 3. Projectionoftlic firstand secondprincipal component 3: 2, 3. Terratypica: "ile Rooma(groupedesilesde
fromaPCAmnon 16individualsassignedto =Agamahou- Las)" (=Roumé Island, Los Islands), Guinea; Syn-
lengeri, =Agama insiilaris =Agama cristata. types. MHNP 1918.41, 1918.42
1939 Agama insularis - Parker, Mém. Mus. Roy. Hist.
males isdistinct. InA. hoiilengerithethroat isdittywhite Nat. Belg., Sér. 2, 15: 89.
with longitudinal stripes, whereas inA. cristata the basal 1947 Agama insularis - Laurent, Bull. Mus. Roy. Hist.
paits ofthe throat are dark bluish-black. Nat. Belg., 23 (16): 5.
1954 Agama insularis - Guibé, Catalogue des Types de
Genetics. As obvious from fig. 4, Agama boulengeri is Lézards du Muséum national d'Histoire naturelle:
very distinct from the fwo other known species with tail 26.
crests (A. insiilaris;A. cristata), and it is found in a basal 1954Agamacristata-Guibé.CataloguedesTypesdeLé-
position to all other/ige//;;« species included in this study zards du Muséum national d'Histoire naturelle: 26.
(NJ: 70). In contrast, the two other species, A. insiilaris 1967 Agama cristata - Wermuth, Das Tierreich, 86: 11.
and A. cristata form a maximally supported clade (NJ: 1967Agama insularis-Wermuth, DasTierreich, 86: 16.
100) that is the sisterclade of A. weidholzi(NJ: 84) with 1979Agama insularis - Joger, Salamandra 15: 36.
this subclade significantly separated from its sister clade 1982Agamainsularis-Welch, HerpetologyofAfrica: 48.
containing A. agama and A. sankaranica as sister taxa, 1985 Agama insularis - Böhme, Proc. Intl. Syinp. Afr.
though this latter subclade receives hardly any support Vertebr., Bonn, 1985: 471.
(NJ: 56). This low suppoil comes as no surprise as other 1988Agamainsularis-Brygoo, Bull. Mus.Nat. Hist.Na-
studiesshowclearlythatA. agamaandA. sankaranicaare turelle 10, Supplement 3, 1—56.
not very closely related (Leaché et al. 2009). 1997Agama insularis-Ulber& Barts, Herprint Interna-
tional, Breden, SouthAfrica, 418 pp.
Table 2 summarizes the uncorrected p-distances for 496 2008 Agama maria nom. nov. Barabanov, Russ. J. Herp.
bp ofthe 16S rRNA gene. Agama boulengeri shows the 15: 206.
highest genetic differences to all other species with val- 2009Agama cristata -Wagner& Böhine, Russ. J. Herp.
uesrangingfrom 12.1%-15.9%thus showingthatitisnot 16: 161-162.
closelyrelatedtoanyoftheotherincluded.^gawaspecies,
MNHN
andisfarremovedfromthetwoothertaxawithtail crests Holotype. 1901.0394.
(A. insiilaris: \5.5%A. cristata: 14.1%). Thus the genet-
ic data clearly confirm the specific distinctness of A. Terra typica. Bomanesco, Sankaran, Guinea.
cristata/A. insiilaris from A. boulengeri.
Commenton Nomenclature. Barabanov(2008)recent-
Although genetic interspecific differences range at mod- ly proposed a nomen novum for Agama cristata Moc-
erately highvaluesrangingfrom 8.0%-!1.9%, thegenet- quard, 1905 becauseherecognizedthistaxonasaprimary
ic distance betweenA. insiilaris andA. cristata is low at juniorhomonymoíAgamacristataMeirem, 1820,recent-
onlyca. 0.7%(con^espondingtoonlydifferingbasepairs). ly known as the iguanid lizard Coiytophanes cristatiis.
The low level ofsequence divergence observed between However, Wagner & Böhme (2009) did not accept this
A. cristata and.4. insularis is consistent with the hypoth- suggestion because in accordance to article 23.9.5 ofthe
esis that these taxa represent a single species. ICZN(1999)theauthorfailedtoreferthecasetotheCoin-
Bonnerzoologische Beiträge 56 245
Acanthocercusatricollis
Agamaboulengeri
Agamaagama-Neotype
100
56
Agamasankaranica
Agamaweidhoizi
84
Agamacristata
100
Agamainsularis
0.1
Fig. 4. Neighbor-joiningtreebased on 496 bp ofthe mitochondrial 16S rRNAgene. Values above the nodes representbootstrap
(20000 pseudo-replicates) values in percent for the neighbor-joining analysis. Significantly supported values are in bold. Values
below 50% are not shown.
missionofZoologicalNomenclaturewhichwouldbenec- The species differs from nearly all otherAgama species
essary because both taxa are not congeneric since 1827, in having a large tail crest, usually reachingthe last third
whereBoie(in Schlegel 1827)connectsMerrem'scrista- ofthetail. Onlythreeotherspecies,A. boulengeri,A. kirkii
ta as type species with his newly described genus Con- andA. knobeli, have such a prominent tail crest.
thophanes.
Agama cristata differs in detail from:
Diagnosis.AlargespeciesofAgama(total lengthofadult
maleupto 340mm),which ischaracterizedby itsbluish- (\) A. boulengeri in (a) higher numbers of scales rows
black colouration on the throat and the large nuchal and around midbody (97 to 123 in A. boulengeri and 1 1 1 to
tail crests ofadult males. 147 [latter value fide Laurent 1947] in A. cristata); in
246
Table 1.
Agama boulengeri Agama cristata Agama weidholñ^ Agamagracilimembris*
average (min-max), n average (min-max), n
SVL 88.4 (62.5-129.8), 22 92.0 (69.0-123.0), 16 54-65 44-57
TL 185.0 (119.2-236.2), 12 149.3 (98.8-215.0), 12
HH
11.0 (8.1-16.3), 22 12.3 (8.6-17.6), 16
HW
16.1 (12.0-24.0), 22 17.5 (11.1-23.8), 16
HL 23.8 (17.0-33.6), 22 25.3 (20.1-33.3), 16
SAM 112.9 (97-123), 22 121.4 (111-136), 13 - 68-82 70-85
TCS 71.3 (62-78), 10 77.5 (75-80), 6
PP 11.3 (8-14), 12 11.4 (10-13), 7
SVL= snout-vent length; TL= tail length; HH= head height; HW= head width; HL=head length; SAM= scale
rows around midbody; TCS= tail crest scales; PP= precloacal pores. All measurements in mm;.*= after Grandison
1969
Table 2. Uncorrected p-distances for496 bp ofthe mitochondrial 16SrRNAsequences used in this study.
Taxa 1 2 3 4 5 6 7
1 Acanthocerciis atricoUis ZFMK 41748
2 Agama boulengeri MNHN 0.2135
3 Agama agama ZFMK 15222 0.2029 0.1348
4 Agama cristata TR555 0.1937 0.1405 0.1040
5 Agama insiilaris ZFMK 88247 0.2022 0.1553 0.1123 0.0072
6 Agama saiúaranica ZFMK 84992 0.1929 0.1209 0.0803 0.1036 0.1192
7 Agama weicUwhi ZFMK 75001 0.2117 0.1590 0.1069 0.0892 0.1002 0.1068 -
(b) having smaller relative head height, head width and However, A. knobeli differs from A. cristata in (a) hav-
head length; and (c) males oíA. boulengeri have a low- ing larger body scales, (b) dorsal scales in same size as
ernumberoftail-crest-scalesandalowerlengthofthetail- ventral scales and (c) in having a pale vertebral stripe in
crest, (d) male coloration (largerocelli, lowerparts ofthe nuptial coloured adult males, and (d) a geographic distri-
throat and guiar fold deep bluish-black). bution restricted to Nainibia.
(2)A. kirkii in (a) having smallerbody scales; (b) dorsal Colouration in alcohol.TheholotypeoíA. cristataisuni-
scales are smaller than ventral scales; and (c) in having form brown. Syntypes of,4. insularis are grey to brown
nonan^owblueandwhitebandedtail(d)ageographicdis- in different shades, but typical ocelli ofAgama females
tribution confined to northern parts of southern Africa andjuveniles are visible. One subadult female specimen
(Malawi, Zambia, Zimbabwe, Mozambique, Botswana). (TR 2353) from the island lie de Roumé (Los islands) is
uniform brown with scattered orange stripes and bars on
0)A. laiobeli is hereinrecognizedas avalid species,be- the body and shows dark stripes on a white throat. Even
cause we recognized striking dissimilarities in moiphol- juveniles from the same locality show dark framed ocel-
ogybetweenA. knobeliandthe forniernominate fonnA. li on the body and white dots on the head. In bothjuve-
atra: Agama atra is lacking the prominent large tail crest niles (MNHN2008.0023,TR2352)alateralorangesüipe
andhas more spinose scales asA. knobeli, but furtherre- is obvious.
searchon therelationshipsofthesetwospecies is inneed.
Bonnerzoologische Beiträge 56 247
Fig. 5. Images ofthe followingAgama species (in life):
A= 'Agama insularis' from island Roumé (photo by Dr. Guy Kremer). B=Agama ciistata, male, from Kindia, Pastoría, Guinea
(photobyIvan Ineich). C=Agama cristata, male, fromChutesde Kinkon, Guinea (photoby Ivan Ineich). D=Agama cristata, fe-
male, from Chutes de Kinkon, Guinea (photo by Ivan Ineich). E=Agama boiilengeri, male, from Mauritania (photo by Hemmo
Nickel). F=Agamahoulengeri, female, from Mauritania(photobyHemmoNickel). G=Agamaboiilengeri, male, from Chutesde
Félou, Mali (photo by Ulrich Joger). H= Agama weidholzi, male, from Mali (photo by Ulrich Joger). 1=Agama cristata, male,
from Kindia, Pastoría, Guinea (tail crest in detail) (photo by Ivan Ineich).
248 Philipp Wagner etal.: Studies onAfricanAgama VI. Taxonomic status ofthe WestAfricanAgama
Mauritania
10
,13
i16 -17
18.
•19 •21
22
Senegal
24
1•-, %25 'o Mali
Gambia m
40
°
°41
Guiniéá-BissauJ",
28 Guinea
29,
27
Fig. 6. Distribution map oíAgama boiilengeri,A. cristata andA. weiclhulzi.
Agamahoiilengeri: Mauritania: I=Choüm; 2=BenAmira;3=Aggui;4=GueitaMolomhar;5=Chinguetti; 6=GueltaHamdoun;
7= Kanoai; 8= Terjit; 9= Zerga Mountains; 10= Tintäne; 11= Iriji; 12= Tidjikja; 13= Tichit; 14= Guelta Fanar; 14= Guelta Mat-
mata; 15= Achram; 16= Ai'n El Ghaire; 17= Bou Blei'ine; 18= Kifa; 19= Bougari; 19a= Guelta Oumm Lebare; 19b= Guelta Me-
traucha; 19c= Oumm El Khez; 20= AyoQn El Atroiis; 21= between Timbcdgha and Ayoün El Atroüs; 22= Mbout; 23= Sélibabi
(fordetails see Padial 2006). Mali: 24= Medine (type locality of.4. hoiilengeri): 25= Chútes du Félou (Joger 1981).
Agamacristata: Guinea: 26= Moussaia, Sankaran (type localityoíA. cristata): 27= lie Rooma, liesde Los(typelocalityof^. in-
siilaris); 28=Chutes deKinkon near Pita; 29= Kindia. Agama weidholzi: The Gambia: 30= Bwiam. Senegal: 31=Boughari, Ca-
samance; 32=betweenTiaraand Mantiankani; 33= 13 km SouthwestofKolda; 34=Tabadienke, 30km South ofDiallakoto; 35=
20 km S ofMedina Gounas; 36= 12 km West ofKounkané; 37= 5 km East ofDarsalam (Niokolo-KobaNational Park). Guinea-
Bissau:38=MadinadoBoé,Pitche. Mali:39=9kmNorthofFatao(14°24'N,9°29'W);40=20kmWestofKita(13°03'N,9°42'W);
41= 5 km EastofKita (13°03'N, 9°25'W);42=betweenNégalaandKassaro(12°55'N, 8°40"W); 43=Négala(12°52"N, 8°27'W)
Colouration in life (see fig. 5). Males. Body brownish,
scattered with dark framedwhiteto creamyocelli, usual-
lybiggerocelli fomidistinctrows. Lipscreamytobluish-
creamy. Head and neck brownish with a pale stripe un-
derneath the eye. Lowerparts ofthe throat and guiarfold
deep bluish-black with strips running to the chin. Poste-
rior part ofthe neck, body, parts ofthe hindlimb and tail
sometimes speckled with white and dark scales. Some-
times apalevertebral bandbetween the limbs is obvious.
Tail at the base pale speckled black, downwards brown-
ish. Belly and underside of the limbs creamish to dark
grey.
Females. Body brownish, scattered with dark framed
white to creamy ocelli, when pregnant with dark orange
to red coloured bands on the lateral body sides between
the limbs. Sometimes lateral paits ofthe body white fol-
lowed by a darker band and pale brownish on vertebral
parts. Head and neck brownish with pale to yellow dots
on the upper head and a pale stripe underneath the eye.
Fig. 7. Mitepocket-like structure (indicated by an an-ow) at a
Lips creamy to white. specimen oíAgama cristata from Kindia, Pastoría in Guinea.