Phrynocephalus

Macey, Robert, Schulte, James A., Ananjeva, Natalia B., Van, Erik T., Wang, Yuezhao, Orlov, Nikolai, Shafiei, Soheila, Robinson, Michael D., Dujsebayeva, Tatjana, Freund, Gabriel S., Fischer, Clayton M., Liu, David & Papenfuss, Theodore J., 2018, A molecular phylogenetic hypothesis for the Asian agamid lizard genus Phrynocephalus reveals discrete biogeographic clades implicated by plate tectonics, Zootaxa 4467 (1), pp. 1-81 : 34-38

A. Phrynocephalus Taxonomic Recommendations

Based on phylogenetic analyses, a few taxonomic changes are proposed. The following taxonomic changes are based on the principle of monophyly as well as genetic and geographic differences. In each case, the reason a change is suggested is provided, whether data are genetic, distributional or phylogenetic (ie. non-monophyly).

1). Phrynocephalus maculatus Complex as Previously Recognized

Based on complete regional sampling, this study recognizes all previous subspecies of P. maculatus as distinct species on the minim-basis of non-monophyly. Phrynocephalus maculatus was described from the Iranian Plateau ( Anderson 1872) with the type locality of “Awada, Shiraz, Persia ”; corrected to “Abadeh [31o 10' N, 52o 37' E; Fars Province, Iran], north of Shiraz” by Blanford (1876). Phrynocephalus maculatus longicaudatus was described by Haas (1957), with a type locality of “Doha Dhalum, Saudi Arabia ”. The sole known population of P. maculatus from Turkmenistan was reported by Bogdanov et al. (1974), and subsequently described as a distinct species, Phrynocephalus golubewii with the type locality of “vicinity of the Bami rail-road station” [= 7 km N railway station Bamy], ( Shenbrot & Semenov 1990). It is recommended to recognize P. golubewii , P. longicaudatus , and P. maculatus as distinct species based on phylogenetic results:

a). While Phrynocephalus golubewii ( Shenbrot & Semenov 1990) may look like P. maculatus living in saltbed dry lakes, it is directly-related to the clade containing P. turcomanus and P. helioscopus (mt-DNA, bootstrap 91%, decay index 6; combined data, bootstrap 87%, decay index 7; allozyme presence/absence, bootstrap 78%, decay index 1; allozyme step matrix, bootstrap 75%, decay index 2; non-monophyly in RAG-1); in fact P. turcomanus occurs within meters of P. golubewii and may overlap in distribution.

b). Phrynocephalus maculatus ( Anderson 1872) is restricted to salt-bed dry lakes on the Iranian Plateau where its type locality is located, and adjacent Baluchistan Plateau including the base of the Sulaiman Range of southern Pakistan. P. maculatus is related, yet outside and sister to a clade containing P. arabicus and P. longicaudatus (see below) based on both mitochondrial and nuclear RAG-1 DNA evidence (mt-DNA, bootstrap 100%, decay index 32; RAG-1, bootstrap 100%, decay index 13; combined data, bootstrap 100%, decay index 46; all of these taxa were not sampled for allozyme data).

c). Phrynocephalus longicaudatus ( Haas 1957) New Status is restricted to the Arabian Peninsula. It is the sister taxon to P. arabicus which is also endemic to the Arabian Plate and supported by mitochondrial and nuclear RAG-1 DNA evidence (mt-DNA, bootstrap 100%, decay index 22; RAG-1, bootstrap 100%, decay index 7; combined data, bootstrap 100%, decay index 30; these taxa were not sampled for allozyme data).

These taxa in any combination do not form a monophyletic group.

2). Small Sand Dune Species of Southwest Asia and Caspian Basin

Phrynocephalus interscapularis interscapularis (Lichtenstien 1856; type locality “Bucharei” = Uzbekistan) and Phrynocephalus interscapularis sogdianus ( Chernov 1948; type locality “ Tadjikistan, vicinity of the Pjandzh village”, approximately 37o 14' N, 69o 05' E) are recognized here as full species P. interscapularis and P. sogdianus based on allozyme and DNA differences. The mt-DNA pair-wise uncorrected sequence distance between P. interscapularis and P. sogdianus is 3.2% and these taxa show one fixed allozyme difference at the carboxylic ester hydrolase locus (EST; Table 2 ).

Phrynocephalus ornatus vindumi [ Golubev, 1998; type locality “Iran, Khorasan Prov., 35 km. N of Gonabad on road to Torbat-E. Heydariyeh (ca. 34o 49' N, 58o 47' E), 850 m. elevation”; sampling here within ten kilometers of the type locality, see appendix 1] is elevated to species status as Phrynocephalus vindumi New Status. Phrynocephalus ornatus ( Boulenger 1887; type locality “between Nushki and Helmand, Baluchistan”) and P. vindumi appear in a non-monophyletic arrangement as P. vindumi occurring on the northeastern portion of the Iranian Plateau is separated from P. ornatus that occurs south of the Hindu Kush in Afghanistan and southwestern Pakistan. In fact, P. vindumi is related to small sand dune species in the Caspian Basin to the north: P. interscapularis and P. sogdianus . In the mt-DNA parsimony analysis, P. vindumi appears in a sister position to the clade containing P. interscapularis and P. sogdianus , with those three taxa grouping with high support (bootstrap 98%, decay index 13). Alternatively, P. ornatus is sister to P. clarkourum (bootstrap 99%, decay index 21), and those two taxa are included in a clade with the two P. luteogutattus samples that also receive high support (bootstrap 100%, decay index 26). Allozyme and RAG-1 DNA data did not sample appropriately to test this question.

3). Gravel Dwelling Phrynocephalus Species in the Caspian Basin

Previously, P. reticulatus , P. bannikovi , and P. strauchi were all considered subspecies of P. reticulatus ( Eichwald 1831) : The nominal subspecies Phrynocephalus reticulatus reticulatus [ Eichwald 1831; type locality "Hab. ad Oxum, pristinum amnem, in caspii maris litore orietali", corrected to the Amu-Darja River by Golubev (1991)] and two additional distinct subspecies, Phrynocephalus reticulatus strauchi [ Nikolsky 1899; restricted type locality “Khodzhent, 40o 17' N, 69o 37' E, northern Tajikistan ” by lectotype designated by Dunayev (1995)] and Phrynocephalus reticulatus bannikovi ( Darevsky, Rustamov & Shammakov 1976; type locality “Tuarkyr Mountains, northwestern Turkmenistan ”), (see Bannikov et al. 1977; although P. reticulatus and P. strauchi were originally described as species). Here these gravel ecotypes are recognized as distinct species based on genetic difference and lack of monophyly.

In our sampling, P. bannikovi and P. strauchi do not appear monophyletic with P. strauchi and P. rossikowi grouping with light support (mt-DNA, bootstrap 55%, decay Index 2). P. reticulatus from central Uzbekistan was not sampled in our study or the study of Solovyeva et al. (2014). If P. bannikovi and P. rossikowi were removed from the parsimony analysis performed on the mt-DNA data, the topology of the trees in this study and that of Solovyeva et al. (2014) would be congruent. Based on incomplete sampling, it is recommended that members of the P. reticulatus complex be recognized as separate species Phrynocephalus bannikovi ( Darevsky, Rustamov & Shammakov 1976) New Status, P. reticulatus , and P. strauchi . Solovyeva et al. (2014) appear to have not sampled any species from Turkmenistan, which includes P. raddei raddei and P. rossikowi , and in our phylogenetic analyses are related to the P. reticulatus complex. The inclusion of P. raddei raddei and P. rossikowi in the sampling presented here indicates the traditional P. reticulatus complex is not monophyletic, deserving species recognition for the three former subspecies. In common with this study, Solovyeva et al. (2014) only sampled P. strauchi from the P. reticulatus complex.

In the sampling of this study, P. bannikovi and P. raddei are geographically oriented disjunctive-adjacent to each other in eastern and west-central Turkmenistan, respectively, and are sister taxa receiving high support from mt-DNA (bootstrap 100%, decay index 23), nuclear RAG-1 DNA (bootstrap 91%, decay index 3), and combined data analyses (bootstrap 100%, decay index 27). Solovyeva et al. (2014) did not sample P. bannikovi , P. rossikowi , or P. raddei raddei from Turkmenistan. However, Solovyeva et al. (2014) and this study did sample P. strauchi from the Fergan Valley of eastern Uzbekistan or northern Tadjikistan. Sampling between this study and Solovyeva et al. (2014) differs in the subspecies of P. raddei : This study uses P. raddei raddei , ( Boettger 1888: type locality “Perewalnaja an der transcaspischen Bahn", = Perevalnaja railroad station, southwestern Turkmenistan; near where this study's sample originated from), whereas Solovyeva et al. (2014) uses P. raddei boettgeri ( Nikolsky 1905; type locality “Schirabad, Buchara orient” = Sherabad, approximately 37o 42' N, 67o 04' E, southern Uzbekistan) with a vague locality provided as “Turan Depression” [= Caspian Basin]; but occurs in extreme southern Uzbekistan and southern Tajikistan along the Afghanistan border ( Bannikov et al. 1977). Until a single study samples both subspecies of P. raddei , it will be unclear how they relate to P. bannikovi , P. reticulatus , P. strauchi , and P. rossikowi .

In Solovyeva et al. (2014) no detailed locality information is available, nor is an appendix with this information provided. All GenBank entries do not include locality data, so this information is not archival with NIH GenBank. The Zoological Museum of Moscow University (ZMMU) (http://zmmu.msu.ru/en/about-muzeum/ fonds/amphibians-&-reptiles-collections) in which specimens reported in Solovyeva et al. (2014) are housed, does not offer an obvious means of searching by voucher number on their website. Additionally, this institution does not currently participate in global databases which compile data from institutions across the globe and allow free public access to museum collections involving voucher number relating to locality information. VertNet, a global organization whose participants include 64 institutions spanning 13 countries including Russia, is one prominent database to organize museum collections around the world. The Global Information Biodiversity Facility (GBIF) also offers a forum for researchers to access information on specimens collected in select countries on every continent, which is an added resource. The Zoological Museum of Moscow University is not currently participating in any of these databases, and localities in the manuscript of Solovyeva et al. (2014) are very vague, crossing large geographic regions, and not informative for the difficulties of dealing with Phrynocephalus taxonomy and sampling. Solovyeva et al. (2014) present GenBank sequences as common courtesy in the scientific community, but in those sequences, no locality information is provided so comparing Solovyeva et al. (2014) results with this study is difficult across all issues, starting with taxonomy. Published localities in Solovyeva et al. (2014) such as Central and Middle Asia are not informative, as in the Russian style geography, Middle Asia refers to the Caspian Basin, and Central Asia refers to Tibet, Mongolia and western China. Localities of such gross generality do not provide any information for a study of this detail.

4). Phrynocephalus helioscopus Complex in the Caspian Basin

Phrynocephalus helioscopus turcomanus [ Dunayev, Solovyeva & Poyarkov 2012 (in Solovyeva et al. 2012); type locality 40° 01′ 00′′ N 52° 58′ 00′′ E, current name Turkmenbashi, Turkmenistan] is recognized as a distinct species Phrynocephalus turcomanus [ Dunayev, Solovyeva & Poyarkov 2012 (in Solovyeva et al. 2012)] New Status from P. helioscopus based on a north-south distributional disjunction in the Caspian Basin with significant mt-DNA differences. P. helioscopus [ Pallas (1771) listed the type locality as "in deserti australioris collibus ardentissimis"; restricted (with no lectotype) to the "Inderskija Gory, Gebiet des unteren Uralflusses" by Mertens & Müller (1928), now Inderskije (Inder) Mountains, Atyrau Region, northwestern Kazakhstan] here is sampled from the northern Aral Sea region of Kazakhstan, and P. turcomanus from southwestern Turkmenistan (Appendix 1). While no fixed allozyme differences are detected between populations sampled, there is a high mt-DNA difference. The mt-DNA pairwise uncorrected sequence distance between P. helioscopus and P. turcomanus is 4.5%; therefore significantly diverged and in the range of sister species divergence for this segment of the mitochondrial genome among amphibians and reptiles (for summary see table 6 in Weisrock et al. 2001).

5). Use of Names in the Low Elevation Chinese Desert—Northern Caspian Basin Clade

a). Phrynocephalus salenskyi

The name P. salenskyi Bedriaga 1907 is applied here for populations sampled from the Junggur Depression of northwestern China, and the Zaysan Depression of northeastern Kazakhstan. P. salenskyi is the oldest name available by date and page number for Phrynocephalus populations of the low elevation Chinese desert – northern Caspian Basin clade in this geographic region, see below:

Phrynocephalus Salenskyi Bedriaga 1907: 141 , 213.—Restricted type locality: “ Urungu Fluss ” [= Ulungur River, Xinjiang Uygur Autonomous Region, China].

Phrynocephalus Isseli Bedriaga 1907: 229 .— Type locality: Desert in vicinity of Mt. Ssalburty (in Saur Mountains of the Tarbagatay Range, Xinjiang Uygur Autonomous Region, China).

Phrynocephalus Haeckeli Bedriaga 1907: 237 .— Type locality: Desert in vicinity of Mt. Ssalburty (in Saur Mountains of the Tarbagatay Range ), in the Tarbagataï (= Tacheng County) district, northern Xinjiang Uygur Autonomous Region, China.

Phrynocephalus Arcellazzii Bedriaga 1907: 307 .— Type locality: “ Gaschun, in der Nähe von Gutschen in der Dschungarei ” [= Qitai (44° 00' N, 89° 50' E), Junggar Depression, Xinjiang Uygur Autonomous Region, China].

Phrynocephalus Grum-Grzimailoi Bedriaga 1909: 420 .—Restricted type locality: “Gutschen” [= Qitai (44° 00' N, 89° 50' E), Junggar Depression , Xinjiang Uygur Autonomous Region, China].

Phrynocephalus bedriaga Tsarevsky 1926: 214 .—Restricted type locality: Zaissan Lake region , eastern Kazakhstan.

Phrynocephalus albolineatus K. Zhao 1979: 113 .— Type locality: "45 kilometers south of Tacheng County, Xinjiang Uygur Autonomous Region, China; 584 m elevation".

The Irtysh River flows out of Zaysan Lake linking the Junggar Depression with the Zaysan Depression, which makes a compelling case for sampling presented in this study representing P. salenskyi populations.

b). Phrynocephalus przewalskii Complex in the Gobi Desert

Phrynocephalus Przewalskii Strauch 1876: 10 .— Type locality: “deserto Alaschanico ” [= Tengger Desert, Nei Mongol Autonomous Region, China] ; corrected to eastern Tengger Desert west from 106° E, southwest Nei Mongol Autonomous Region, China in Barabanov & Ananjeva (2007). This taxon is generally a sand dwelling ecotype.

P. frontalis Strauch 1876: 15 .—Type locality: locality “Ordos dicta” [= Ordos Desert, Nei Mongol Autonomous Region, China]. This taxon is generally thought of as a gravel dwelling ecotype, and here considered a synonym of P. przewalskii (see below).

A lack of genetic diversity between populations from these taxa (two populations each), confirm that only the name P. przewalskii is valid, whereas P. frontalis is a synonym of P. przewalskii (following Gozdzik & Fu 2009). This is supported by nuclear encoded allozyme data, and mt-DNA data. Alternative names exist as reviewed in Barabanov & Ananjeva (2007), but without detailed sampling of type localities, the validity of alternate names cannot be addressed.

Across the 25 allozyme loci sampled here, we detected no fixed differences among the four populations of P. przewalskii , indicating a fair amount of gene flow between the populations. The average mt-DNA pair-wise uncorrected sequence distances between population 2 and the clade containing populations 1, 3, and 4 is 3.64% (range 3.41–3.79%). The average pair-wise comparison within the clade containing populations 1, 3 and 4 is 0.57% (range 0.51–0.63%). These results match the phylogenetic topology discovered in the mt-DNA parsimony analysis, with population 2 placed in a sister position to the clade containing populations 1, 3 and 4 (bootstrap 100%, decay index 19). No resolution was discovered among the four P. przewalskii populations in the RAG-1 DNA parsimony analysis.

Urquhart et al. (2009) suggests a vicariant break among P. przewalskii populations roughly corresponding to the Helan Shan (mountains). Data reported here do not corroborate that hypothesis. Sampled are a single population east of the Yellow River and three populations west of the Yellow River, which flows north to south in this region. Population 2 is sampled on the east side of the river, and population 3 is sampled on the west side of the river, on banks directly across from each other in Shapotou, Ningxia Autonomous Region, China. Population 2, the only population on the east side of the Yellow River is the most divergent, with populations 1, 3 and 4 on the west side of the river nearly identical in our sampling. Population 1 (Wujiachuan, Gansu Province, China) is west of the Yellow River and further south than sampling conducted in Urquhart et al. (2009) and Wang & Fu (2004), yet nearly identical to other populations sampled west of the Yellow River. In our sampling, only population 4 lies west of the Helan Shan (Wuwei, Gansu Province, China) in the Tengger Desert, the type locality of P. przewalskii . Sampling in this study suggests the Yellow River could be responsible for a genetic break in P. przewalskii populations, and not the Helan Shan, though more detailed, fine-scale sampling and analysis will need to be done to confirm this hypothesis. In the Zhongwei region of Ningxia Autonomous Region, there is an indication that east and west haplotypes as described above occur in the same locality, which could indicate a point of hybridization, but further detailed work is needed ( Wang & Fu 2004). Sampling conducted in similar areas in Wang & Fu (2004) and Urquhart et al. (2009) did not completely overlap, making conclusions between these studies and this study very difficult for comparative interpretation, further indicating additional work is needed.

In sampling of the P. przewalskii complex in Jin & Brown (2013), populations to the east are grouped, and are referred to as P. frontalis , with populations to the west grouping and referred to as P. przewalskii , not conflicting with results reported here.

6). Use of Names in the Northern Tibetan Plateau

Species distributed on the northern side of the Tibetan Plateau form the P. vlangalii complex and have a complicated taxonomic history. Here, the following names are recognized P. vlangalii , P. roborowskii , and P. hongyuanensis .

Phrynocephalus vlangalii, Strauch 1876 ; type locality “Kuku-noor dictum” [= Qinghai Lake, Qinghai Province, China]. The name P. vlangalii predates P. putjatai Bedriaga 1907 ; type locality “Gui-dui am Hoang-ho” [= Guide County (36o 00' N, 101o 40' E), southeast of Qinghai (Kuku-Nor) Lake, Qinghai Province, China]. Phylogenetic results in Jin & Brown (2013) are consistent with results reported here with a reassignment of names for populations sampled in Jin & Brown (2013). If P. putjatai is a junior synonym of P. vlangalii because their type localities appear to overlap, then populations assigned to that species in Jin & Brown (2013) are actually P. vlangalii . Populations assigned to P. vlangalii in Jin & Brown (2013) are P. roborowskii and P. hongyuanensis .

Phrynocephalus roborowskii Bedriaga “1905 ” 1906 ; restricted type locality “ Provinz Zaidam ” [= Qaidam Depression, Qinghai Province, China].

Phrynocephalus hongyuanensis Zhao, Jiang & Huang 1980 (in Jiang et al. 1980); type locality “Waqên, Hongyuan County, 33o 03' N; 102o 37' E, Sichuan Province, China; 3500 m.” This species was originally described as a subspecies of P. vlangalii (see below).

Phrynocephalus vlangalii hongyuanensis is elevated to species status as P. hongyuanensis based on phylogenetic information reported here. P. hongyuanensis from the northeastern portion of the Tibetan Plateau is either sister to P. roborowskii from the northwestern portion of the Tibetan Plateau (mt-DNA bootstrap 100%, decay index 17; combined analysis bootstrap 100%, decay index 20), or nested inside P. roborowskii in the nuclear RAG-1 DNA analysis with little support (decay index 1). Hence, in all analyses P. vlangalii appears outside of the clade containing P. hongyuanensis and P. roborowskii , while being located in the geographical middle of those species. Jin et al. (2008) sampled only P. vlangalii and P. roborowskii , leaving out P. hongyuanensis in sampling, therefore not discovering this distinction. Pang et al. (2003) showed P. hongyuanensis inside of P. vlangalii , but did not consider P. roborowskii as a taxon or did not sample it [sampling between this study and Pang et al. (2003) is not the same].

a). Sex Determinate Dispersal — In our sampling, Tibetan Plateau species of P. vlangalii , P. roborowskii , P. honguyanensis , and P. theobaldi , as well as P. forsythii , are all viviparous. All are high elevation ecotypes except P. forsythii , which is from the low elevation Taklimakan Desert north of the Tibetan Plateau. There is evidence of sex determinate dispersal favoring males in this clade ( Qi et al. 2013). While collecting during summer months in sandy soil habitats of the Tibetan Plateau, Macey and others would commonly dig up to one meter down to find gravid females, and sometimes females who had just given birth with their young in the burrow. Females lack an evolutionary drive to disperse because once impregnated they must dig deep burrows of a meter or more to carry out their two month incubation period needed before parturition ( Qi et al. 2013), and remain in the burrow for hibernation during winter months possibly with their young ( Wu et al. 2015), who are too small to dig burrows of their own. Males disperse for breeding purposes and have the additional ability to disperse for livelihood purposes.

b). Analogous Natural History with Salamanders of the Genus Ensatina— Mitochondrial DNA is maternally inherited, and nuclear DNA is bisexually inherited. Females of the subspecies Ensatina eschscholtzii platensis brood their eggs, (which means to guard their eggs) and as such males have a greater home range than females, dispersing in wider distances ( Staub et al. 1995). Kutcha et al. (2009) reported a sharp mt-DNA break between northern and southern populations of E. e. platensis at the Stanislaus River, which serves as the border between the California counties of Calaveras and Tuolumne. For reference, in the central Sierra Nevada where the mt-DNA break occurs, sample 190 refers to the northern clade of E. e. platensis taken near Arnold in Calaveras County, while sample 192 refers to the southern clade taken near Alder Spring, central Tuolumne County south of the Stanislaus River and north of the Tuolumne River (refer to figure 3 and appendix S1 for locality information in Kutcha et al. 2009). The two clades of E. e. platensis show an intergradation of breaks at eight nuclear encoded allozyme loci ranging from Wagner Ridge in Mariposa County (sample number 27 in Jackman & Wake 1994) to the south and Panther Creek in Amador County to the north in the region of the Mokelumne River (sample number 22 in Jackman & Wake 1994), with indications of significant allozyme variation within populations of the southern clade of E. e. platensis ( Wake & Schneider 1998). Unpublished data recognize a major break between northern and southern E. e. platensis clades, as well as a hybridization zone with E. e. xanthoptica near the area of Wagner Ridge (David B. Wake pers. comm.). Hence, the Ensatina complex shows alternative patterns in the degree of genetic breaks from maternally inherited mitochondrial DNA and bisexually inherited nuclear encoded allozyme alleles that may be attributed to life history traits.

Noble et al. (2010) reported a possible area east of Qinghai Lake in which two different populations of Phrynocephalus of uncertain taxonomic identity come in contact with nuclear DNA and mitochondrial DNA appearing to have discordant breaks. A possible explanation for this could be females must make deep burrows and sit tight ( Wu et al. 2015), while males move around, and as such mitochondrial DNA remains localized, with bisexually inherited nuclear DNA spreading by males as they look for breeding partners ( Noble et al. 2010). As discussed above, we believe populations on the east side of Qinghai Lake should be referred to as P. vlangalii , and populations on the west side of the lake referred to as P. roborowskii . Maternally inherited mitochondrial DNA and bisexually inherited nuclear DNA should not show a discordance based on the sampling strategy adopted in this study.