Zoologica Scripta

Molecular phylogeny and systematics of Dipodoidea: a test of morphology-based hypotheses

VLADIMIR S. LEBEDEV,ANNA A. BANNIKOVA,MARIE PAGE` S,JULIE PISANO,JOHAN R. MICHAUX & GEORGY I. SHENBROT

Submitted: 5 February 2012 Lebedev, V.S., Bannikova, A.A., Page`s, M., Pisano, J., Michaux, J.R. & Shenbrot, G.I. Accepted: 18 October 2012 (2012). Molecular phylogeny and systematics of Dipodoidea: a test of morphology-based doi:10.1111/zsc.12002 hypotheses. —Zoologica Scripta, 00, 000–000. The superfamily Dipodoidea (Rodentia, Myomorpha) in its current interpretation contains a single family subdivided into six subfamilies. Four of them include morphologically spe- cialized bipedal arid-dwelling jerboas (Dipodinae – three-toed jerboas, Allactaginae – five- toed jerboas, Cardiocraniinae – pygmy jerboas and Euchoreutinae – long-eared jerboas), the other two are represented by more generalized quadrupedal taxa (Zapodinae – jumping mice and Sminthinae – birch mice). Despite considerable effort from morphologists, the taxonomy as well as the phylogeny of the Dipodoidea remains controversial. Strikingly, molecular approach has never been envisaged to investigate these questions. In this study, the phylogenetic relationships among the main dipodoid lineages were reconstructed for the first time using DNA sequence data from four nuclear genes (IRBP, GHR, BRCA1, RAG1). No evidence of conflict among genes was revealed. The same robustly supported tree topology was inferred from the concatenated alignment whatever the phylogenetic methods used (maximum parsimony, maximum-likelihood and Bayesian phylogenetic methods). Sminthinae branches basally within the dipodoids followed by Zapodinae. Monophyletic Cardiocraniinae is sister to all other jerboas. Within the latter, the mono- phyly of both Dipodinae and Allactaginae is highly supported. The relationships between Dipodinae, Allactaginae and Euchoreutinae should be regarded as unresolved trichotomy. Morphological hypotheses were confronted to findings based on the presented molecular data. As a result, previously proposed sister group relationships between Euchoreutes and Sicista, Paradipus and Cardiocraniinae as well as the monophyly of Cardiocaniinae + Dipo- dinae were rejected. However, the latter association is consistently supported by most mor- phological analyses. The basis of the obvious conflict between genes and morphology remains unclear. Suggested modifications to the taxonomy of Dipodoidea imply recogni- tion of three families: Sminthidae, Zapodidae and Dipodidae, the latter including Cardioc- raniinae, Euchoreutinae, Allactaginae and Dipodinae as subfamilies. Corresponding author: Vladimir S. Lebedev, Zoological Museum, Moscow State University, B.Nikitskaya 6, 125009 Moscow. E-mail: [email protected]. Anna A. Bannikova, Lomonosov Moscow State University, Vorobievy Gory, 119992 Moscow, Russia. E-mail: [email protected] Marie Page`s, INRA, UMR CBGP (INRA ⁄ IRD ⁄ Cirad ⁄ Montpellier SupAgro), Campus Interna- tional de Baillarguet, CS 30016, 34988 Montferrier-sur-Lez cedex, France; Laboratoire de ge´ne´- tique des microorganismes, Institut de Botanique, Universite´ de Lie`ge, 4000 Lie`ge (Sart Tilman), Belgique. E-mail: [email protected] Julie Pisano, INRA, UMR CBGP (INRA ⁄ IRD ⁄ Cirad ⁄ Montpellier SupAgro), Campus Interna- tional de Baillarguet, CS 30016, 34988 Montferrier-sur-Lez cedex, France; Laboratoire de ge´ne´- tique des microorganismes, Institut de Botanique, Universite´ de Lie`ge, 4000 Lie`ge (Sart Tilman), Belgique. E-mail: [email protected] Johan R. Michaux, INRA, UMR CBGP (INRA ⁄ IRD ⁄ Cirad ⁄ Montpellier SupAgro), Campus International de Baillarguet, CS 30016, 34988 Montferrier-sur-Lez cedex, France; Laboratoire de ge´ne´tique des microorganismes, Institut de Botanique, Universite´ de Lie`ge, 4000 Lie`ge (Sart Tilman), Belgique. E-mail: [email protected]

ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters 1 Molecular phylogeny of Dipodoidea d V. S. Lebedev et al.

Georgy I. Shenbrot, Mitrani Department of Desert Ecology, Jacob Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev, 84990 Midreshet Ben-Gurion, Israel. E-mail: [email protected]

Introduction should be placed in a single family (Vinogradov 1930, The superfamily of Dipodoidea is the sister group of 1937; Ellerman 1940; Ognev 1948; Klingener 1984; Hol- Muroidea based either on current molecular (e.g. Michaux den & Musser 2005). The rationale for this approach rests & Catzeflis 2000; Huchon et al. 2002; DeBry 2003; Adkins on the observation that bipedal Euchoreutinae and Cardi- et al. 2003; Montgelard et al. 2008; Blanga-Kanfi et al. ocraniinae share with Zapodinae and Sminthinae a number 2009; Churakov et al. 2010) or on morphological (e.g. of unspecialized skeletal features (Vinogradov 1930, 1937). Klingener 1964) data. Compared with the latter taxon, Based on equivocal and possibly intermediate position of which is the most diverse group of mammals – no less Euchoreutinae and Cardiocraniinae, it was concluded that than 6 families, 310 genera and more than 1518 species simple dichotomy between bipedal and non-bipedal taxa is (Musser & Carleton 2005) – the diversity of dipodoids inadequate to accommodate significant morphological var- appears relatively modest – just 16 genera and 51 species iation within the superfamily. The point is illustrated by (Holden & Musser 2005). However, the Dipodoidea dem- an informal morphological phylogenetic tree of Vinogra- onstrate diverse ecological and morphological adaptations dov (1937), which supported a basal position of Zapodinae ranging from the forest- and meadow-dwelling mouse-like (here including also Sicista) with Cardiocraniinae and Eu- birch mice and jumping mice to the arid-dwelling saltato- choreutes branching deeper than Allactaginae and Dipodi- rial jerboas, which exhibit the most extreme specializations nae (Fig. 1). for bipedal locomotion among rodents (Fokin 1978). A different taxonomic arrangement (Dipodidae, Zapodi- Although the superfamily has received much taxonomic dae, Sminthidae as separate families) was proposed based attention (see bellow), its classification remains controver- on chromosome data (Vorontsov et al. 1971). Zapodidae sial. (Fig. 1). Most researchers (e.g. Holden & Musser were considered karyotypically primitive as compared to 2005) agree that dipodoids can be arranged into six major jerboas and birch mice. No evident karyological similarity groups, which are usually assigned to subfamilial rank. between any two of these well-defined groups could be These groups are as follows: Dipodinae (three-toed jer- traced. boas – five genera), Allactaginae (five-toed jerboas – three The first explicitly phylogenetic system of the superfam- genera), Cardiocraniinae (pygmy jerboas – two genera), ily (Stein 1990) recognized only two families, Sicistidae Euchoreutinae (long-eared jerboas – one genus), Zapodi- (=Sminthidae) and Dipodidae, the latter including also nae (jumping mice – three genera) and Sminthinae (syno- Zapodinae. Contrary to Vinogradov’s pattern, Cardiocra- nym Sicistinae, birch mice – one genus). niinae is placed sister to the Dipodinae. This study was In contrast, family-level classification has long been a primarily based on cladistic analysis of limb myology. matter of debate, the number of recognized families rang- To a large extent, the above morphology-based systems ing from one to five. This lack of consensus on dipodoid reflect the evolution of locomotory adaptations, with sub- taxonomy is, to a large extent, accounted for by the fact families (or families) corresponding to grades of evolution- that phylogenetic relationships among the main lineages ary development from primitive quadrupedal to specialized have not been unambiguously established yet. bipedal locomotion. The least advanced stage is repre- Some of the early precladistic classifications contrasted sented by birch mice (strictly quadrupedal, non-leaping) the bipedal arid-dwelling true jerboas (Dipodinae, Allacta- followed by jumping mice (predominantly quadrupedal ginae, Cardiocraniinae and Euchoreutinae) to the forest but capable of making long leaps; hind foot elongated). All and grassland quadrupedal birch mice (Sminthinae) and strictly bipedal forms (jerboas) show important modifica- jumping mice (Zapodinae) considering these two groups as tions of hindlimb skeletal and muscular morphology (see separate families: the Dipodidae and the Zapodidae (Miller Fokin 1978 for a review). Among jerboa sufamilies, the & Gidley 1918; Vinogradov 1925; Simpson 1945). Obvi- highest degree of morphological specializations is achieved ously, the latter family was based upon plesiomorphic sim- in fast-running allactagines and dipodines (maximum speed ilarity between the two quadrupedal lineages and was approximately 5.1–11.9 m ⁄ s), whereas the lowest is found never formally supported on phylogenetic grounds. in cardiocraniines (maximum speed <2.5 m ⁄ s; antipredator According to the alternative view, all dipodoid groups response is freezing rather than fleeing) (Fokin 1978). The

2 ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters V. S. Lebedev et al. d Molecular phylogeny of Dipodoidea

Fig. 1 Graphic representation of previously suggested taxonomic systems and phylogenetic patterns in Dipodoidea.

ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters 3 Molecular phylogeny of Dipodoidea d V. S. Lebedev et al. problem is that, given the adaptive nature of morphologi- 2004; Montgelard et al. 2008; Jansa et al. 2009; Fan et al. cal change and limitations on the number of pathways 2009). However, important taxa such as Cardiocraniinae along which it can occur, one may expect that the evolu- and Euchoreutes remain unstudied. tionary transformation of the locomotory system in dipod- Given all these contradictions and gaps in our knowl- oids could be the subject of massive parallelisms, which edge of dipodoid phylogeny, it is not surprising that many are rather hard to differentiate from true synapomorphies. taxonomists accept a conservative classification retaining a Therefore, it seems reasonable to focus a phylogenetic single family with six subfamilies. Actually, this uninforma- study upon traits that are not directly associated with loco- tive system was first introduced by Ognev (1948) and now motion. This approach was accepted in a cladistic analysis is reproduced in the third edition of Mammal Species of based on characters of dentition, male reproductive system the World (MSW3, Holden & Musser 2005). In the pres- and auditory bulla (Shenbrot 1992). The analysis did not ent paper, we also follow the Holden and Musser’s classifi- reveal any synapomorphies to support the monophyly of cation, with minor nomenclatural changes. In particular, the bipedal taxa. On the contrary, a non-orthodox group- we accept that Sminthinae Brandt, 1855 has priority over ing inferred in this study (Fig. 1) suggested that long- Sicistinae Allen, 1901 (see Data S1). eared jerboa (Euchoreutes) could be more related to Sicista The goal of the present study is to examine the relation- rather than to Allactaginae or Dipodinae. This pattern ships between major branches of Dipodoidea based on would then imply at least two independent events of molecular data (four nuclear exons), an approach that acquisition of bipedal locomotion in lineages leading to proved its effectiveness in similar studies in Muroidea (e.g. the long-eared jerboa and to Dipodinae – Allactaginae. In Michaux et al. 2001; Steppan et al. 2004; Jansa et al. 2009). addition, this study is incongruent with traditional views Available morphological hypotheses will be tested against in supporting a close relationship between Cardiocraniinae genetic evidence. Specifically, we will address the follow- and Paradipus, which had been otherwise regarded as one ing questions: (i) what is the basal branching order within of the dipodine genera. Shenbrot’s arrangement concurs Dipodoidea, (ii) what is the position of phylogenetically with that of Stein (1990) in placing Dipodinae and Cardi- controversial taxa (Cardiocraniinae, Euchoreutes and Paradi- ocraniinae as closest relatives; at the same time, the rela- pus), (iii) what are the intergeneric relationships within tionships among Zapodidae, Sminthidae, Allactagidae and Allactaginae and Dipodinae, (iv) is there significant discor- Dipodidae remain unresolved. The resulting phylogeny dance between molecular and morphological data and (v) suggested important changes in the dipodoid taxonomy. do molecular data support independent origin of bipedal Some of Vinogradov‘ subfamilies were elevated to the locomotion in several jerboa lineages. Finally, the taxon- familial rank dividing jerboas into four families: Allactagi- omy will be revised. dae, Dipodidae (including Paradipodinae and Cardiocra- niinae), Sminthidae (with Euchoreutinae) and Zapodidae. Material and methods Paleontological data as overviewed in Zazhigin & Lopa- Specimens examined tin (2000a,b, 2001) support this last classification in some The original material consists of 26 specimens of 15 spe- aspects. Their system lists Zapodidae (containing Sicisti- cies of jerboas, two species of jumping mice (Zapodinae) nae and Zapodinae), Allactagidae (containing Allactaginae and three birch mice (Sminthinae). Most of voucher speci- and Euchoreutinae) and Dipodidae (which includes Cardi- mens of animals used in this study are deposited in the ocraniinae, Dipodinae, and the extinct Lophocricetinae) Zoological Museum of Moscow Lomonosov State Univer- (Zazhigin & Lopatin (2000b). The evolutionary scenario sity (ZMMU). For phylogenetic analysis, 12 sequences of proposed therein is consistent with the independent origin different genes of Dipodoidea were retrieved from Gen- of the two more specialized jerboa lineages suggesting that Bank (Data S1). The total matrix contains 33 specimens of the roots of Allactagidae and Dipodidae are related to dif- Dipodoidea representing all six subfamilies, 15 genera (out ferent groups of sminthine-like ancestors. The last two of recognized 16) and 20 species (out of recognized 51). classifications are based on the view of dipodoid evolution The data set includes 20 outgroup OTUs (members of as a complex process involving independent (and often Spalacidae, Muridae, Nesomyidae, Cricetidae, Calomysci- parallel) locomotory, trophic and substrate adaptations. dae, Geomyidae, Heteromyidae, Castoridae, Anomaluri- Although far from complete, available molecular data dae, Pedetidae, Gliridae and Sciuridae (in total 68 consistently places Sicista as the sister group to other di- sequences of 44 species, some of which are combined into podoids with zapodines branching off next, while the composite taxa). Information on specimens used including three-toed and the five-toed jerboas (represented by single the list of species, collecting sites and museum catalogue exemplars of Dipus or Jaculus and Allactaga respectively) numbers is given in Table 1, the GenBank Accession num- appear as sister taxa (Jansa & Weksler 2004; Steppan et al. bers are presented in Data S1.

4 ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters V. S. Lebedev et al. d Molecular phylogeny of Dipodoidea

Table 1. Characterization of the original material

Specimen code Museum catalogue number, Species (Figs 1 and 2 ref.) tissue or field code Collecting locality

Allactaga major 1 ZMMU 01 ⁄ 08_32 Russia, Dagestan, Chakanny 2 t.c. Ama Locality unknown DNA donated by D. Kramerov Allactaga bullata 1 ZMMU S179572 Mongolia, Dundgovi aymag, Luus-somon 10 km W 2 ZMMU S188089 Mongolia, O¨ mno¨ govi aymag, Manlai–Mandakh road Allactaga sibirica 2 ZMMU S181016 Mongolia, Khovd aymag, S shore Durgen-nur Lake Allactaga elater Uncatalogized (coll. L.Khlyap) Russia, Dagestan, Chakanny Allactodipus bobrinskii t.c. Abo Turkmenistan, Chardzhou region, left bank of Amur-Daryi, 6 km W Lebap vill. DNA donated by D. Kramerov Pygeretmus pumilio 1 t.c. Apy Kazakstan, Kzyl-Ordy region, N Kyzylkum, 80 km WSW Kzyl-Ordy DNA donated by D. Kramerov 2 ZMMU S181007 Mongolia, Govi-Altai aymag, W shore Beger-nur Lake Eremodipus lichtensteini t.c. Eli Kazakstan, Kzyl-Ordy region, N Kyzylkum, 80 km WSW Kzyl-Ordy DNA donated by D. Kramerov Dipus sagitta 1 ZMMU S179627 Mongolia, Bayankhongor aymag, S shore Bon-Tzagan-nur Lake 2 ZMMU S188090 Mongolia, O¨ mno¨ govi aymag, Manlai – Mandakh road Stylodipus telum Uncatalogized (coll. L. Savinetskaya) Russia, Kalmykia, Chernye zemli Jaculus jaculus Moscow Zoo Locality unknown Jaculus blanfordi San-Diego, Uzbekistan, C Qyzylqum, Mynbulak 40 km SW, Djarakuduk t.c. 2006 ZIN 96242 ibid Paradipus ctenodactylys ZMMU S143250 Turkmenistan, Kazandjik 25 km NE ZMMU S112338 (D3049) Uzbekistan, W Qyzylqum, Takhtakupyr 100 km E, Kempir-Tyube Cardiocranius paradoxus 1 ZMMU S188830 China, Inner Mongolia 2 ZMMU S187344 Mongolia, Govi-Altai aymag, Buutsagaan 50km SW Salpingotus kozlovi ZMMU S183051 Mongolia, Govi-Altai aymag, Nogon-Davon Euchoreutes naso ZMMU S179636 Mongolia, Bayankhongor aymag, Ekhiin-gol 47 km NE Eozapus setchuanus D3040 China, Sichuan donated by J.-P. Que´re´ Napaeozapus insignis ZMMU S183269 Canada, Ontario, vic. of Deepriver Sicista ex. gr. subtilis YuMK 1559 Russia, Volgograd region, Kamyshin 20 km S Sicista ex. gr. caucasica D764 Russia, NW Caucasus donated by F. Catzeflis

ZMMU, Zoological Museum of Moscow State University, Moscow; ZIN, Zoological Institute RAS, St.-Petersburg.

DNA isolation, PCR amplification and sequencing Data S1. For amplification of the exon 10 of GHR,in Total DNA was extracted from ethanol preserved tissues addition to the original primers, ghrEXON10 and from liver, kidney or from dried muscles using standard ghrEND_R (Steppan et al. 2004) were used. Double- protocol of proteinase K digestion, phenol-chloroform stranded polymerase chain reaction (PCR) usually entailed deproteinization and isopropanol precipitation (Sambrook 30–35 thermal cycles as follows: 30 s denaturation at et al. 1989). The DNA of Paradipus was purified 94 C, 1 min annealing at 55–65 C and 1 min extension directly from the bone tissue of museum ethanol and at 72 C. PCR products were visualized on 1% agarose dry specimens using silica-based spin columns of MinE- gel and then purified using DEAE Watman (UK) or lute PCR Purification Kit (QIAquicky; Qiagen, Hilden, NH4EtOH. Approximately 10–40 ng of the purified PCR Germany) and according to recommendation of Yang product was used for sequencing with each primer by au- et al. (1998). tosequencing system ABI 3100-Avant using ABI The regions of four nuclear genes (GHR, IRBP, RAG1, PRISMBigDye Terminator v. 3.1 (Foster, CA, USA). BRCA1) were amplified and sequenced in 26 animals using external forward ⁄ reverse primer combinations as well as Phylogenetic analysis: molecular data internal primers. The sequences of the original primers Base frequency homogeneity was tested at third codon for amplification and sequencing are presented in the positions separately for each locus based on values of

ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters 5 Molecular phylogeny of Dipodoidea d V. S. Lebedev et al. disparity index (Kumar & Gadagkar 2001) calculated in gamma distribution or proportion of invariant sites. To Mega version 4.0 (Tamura et al. 2007). correct for potential bias in branch length estimation Phylogenetic reconstructions were conducted on the branch length prior were adjusted following recommenda- four loci independently and also on concatenated data. tions given in Brown et al. (2010). Transversion data were Primary analyses were conducted using information on all analysed using restriction model (0 ⁄ 1 recoding) assuming substitution types. To check for potential biases due to gamma distribution of rates. Each analysis included two GG-content heterogeneity (see below), a second round of independent runs of four chains (one cold plus three tree inference was performed for the combined data based heated following default settings). Chain length was set at on transversions only. 15 million generations for the original data and at 7.5 mil- For phylogenetic reconstructions, maximum parsimony lion generations for the transversion data with sampling (MP), maximum-likelihood (ML) and Bayesian methods each 1000th generation. With these settings, the effective (BI) were used. Unweighted parsimony analysis was per- sample size exceeded 200 for all estimated parameters. formed using PAUP* 4.0b10 (Swofford 2003). The follow- Tracer 1.5 software (Rambaut & Drummond 2005) was ing options were invoked: random addition sequence with used to check for convergence and determine the neces- 20 replicates, no limit for number of optimal trees, TBR sary burnin fraction, which was 500 000 generations in branch swapping. Clade stability was assessed using 1000 both cases. bootstrap replicates using the same tree search parameters. Levels of genetic divergence among taxa were estimated Incongruence length difference (ILD) test (Farris et al. based on patristic distances calculated from the ML tree 1995) was implemented to check for significant heteroge- branch lengths (concatenated data, all substitution types). neity among genes. Hypothesis testing was conducted in Treefinder using Maximum-likelihood was performed in Treefinder (Jobb AU test. Eight a priori hypotheses on the relationships 2008). The data were partitioned into gene·codon posi- among lineages of Dipodoidea were tested against our best tions and separate models were used for each of the 12 molecular tree. These hypotheses were as follows: partitions. This partitioning strategy was chosen based on 1. Dipodinae and Cardiocraniinae are sister taxa as the results of a preliminary analysis comparing five alter- inferred from morphological data (Stein 1990; Shenbrot native partitioning schemes (see Data S1). Best models 1992) and suggested based on fossil record (Zazhigin & were chosen based on BIC criterion from the set of mod- Lopatin 2000b). els available in Treefinder and using the routine imple- 2. Paradipus is sister to Cardiocraniinae and not a part of mented in it. Rate heterogeneity was modelled using Dipodinae (Shenbrot 1992). discrete gamma with four categories. Tree search was 3. Within Dipodinae, Dipus descends from the basal-most employed with the following options: parameter optimiza- node, while Paradipus is sister to Eremodipus + Jaculus as tion simultaneous with tree search, optimized partition proposed by Vinogradov (1937). rates, proportional branch lengths for all partitions, maxi- 4. Euchoreutes is sister to Cardiocraniinae + Dipodi- mum search depth, 10 random starting trees generated as nae + Allactaginae as follows from Vinogradov (1937) recommended in program manual (using default settings and Shenbrot (1992). for number of steps and tries, with the centre tree 5. Euchoreutes is sister to Allactaginae as argued by Zazhi- obtained under the HKY model for the unpartitioned gin & Lopatin (2000b). data). Bootstrap analysis (1000 pseudoreplicates) was per- 6. Euchoreutes is sister to Sminthinae as inferred by Shenb- formed using model parameters and rate values optimized rot (1992). for the ML tree. Transversion-based analysis was con- 7. Birch mice together with Zapodinae constitute a mono- ducted with GTR2 (+G) models implemented in Treefind- phyletic group as accepted by Vinogradov (1937), er. Gene tree concordance was tested in Treefinder using Simpson (1945) etc. approximately-unbiased (AU) tests (Shimodaira 2002) con- 8. Allactaga is monophyletic relative to other five-toed jer- trasting the ML gene-specific topology with the tree in boas (Pygeretmus and Allactodipus) as routinely accepted which relationships within Dipodoidea were modified to in taxonomic accounts (e.g. Holden & Musser 2005). correspond to that of the ML tree for concatenated data. Bayesian methods was performed in Mrbayes 3.1.2 Phylogenetic analysis: morphological data (Ronquist & Huelsenbeck 2003). Model complexity was The parsimony analysis of morphological data was based assessed based on BIC criterion using Modeltest version on a matrix, which closely follows the one used in Shenb- 3.7 (Posada & Crandall 1998). Given its result, models rot (1992) with some modifications. The matrix includes with either two or six rate matrix parameters were chosen. 31 OTUs representing all genera and major species groups Among-site rate heterogeneity was modelled using either of Dipodoidea. Four morphological character partitions

6 ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters V. S. Lebedev et al. d Molecular phylogeny of Dipodoidea were included in the analysis: dentition (17 characters), Felsenstein (1985), the more conservative two-sided ver- auditory bulla (five characters), glans penis (seven) and sion of the test was used. accessory reproductive glands (six). The detailed descrip- To identify which morphological characters were pre- tion of character states is given in Data S1. As in Shenbrot dominantly responsible for the conflict with molecular evi- (1992), the primary (ordered = O) analysis was based on a dence, we assessed the level of conflict between each of priori assumptions on polarity, character state ordering the four morphological partitions and the molecular-con- implying irreversibility of certain transformations. All strained tree. Conflict was quantified as the average information on allowed character change was coded via decrease of single-character consistency indices (number of step-matrices (Data S1). To test the sensitivity of the phy- states minus one ⁄ observed number of steps) for constraint logenetic results to the above a priori assumptions, we also trees versus unconstrained trees. Significance was tested conducted unordered analyses (U), which impose mini- using nonparametric Kruskal–Wallis test as implemented mum constraints on character evolution. In ordered analy- in Statistica 6.0 (StatSoft, Inc. 2001). Also, partitioned ses, the position of the root was inferred based on Bremer support (Baker & DeSalle 1997) was estimated for optimization in irreversible characters, while in the unor- selected nodes. dered analyses, trees were rooted using the explicitly speci- The key morphological transformations including those fied hypothetical ancestor. Taking into account that the involved in the evolution of bipedalism were mapped onto number of recognized character states was highly variable both morphological and molecular trees using MPRSets across characters (from two to nine) and across partitions command in PAUP* (ordered data). (ranging from 15 in total for bulla to 58 for dentition), we also performed weighted analyses to achieve a more bal- Results anced contribution of different traits to phylogenetic Alignment and partitioning reconstruction. Two methods of character weighting were The complete alignment of IRBP consisted of 1107 bp used. The first method (Wc) estimated the weight of each (636 variable, 458 parsimony-informative) for 26 speci- individual character as inversely proportional to its mini- mens of Dipodoidea and 20 outgroups. The alignment of mum possible number of steps (equal to the number of GHR, exon 10 consisted of 927 bp (577 variable, 403 par- states in a character minus one). The second one (Wp) simony-informative) for 25 specimens of Dipodoidea and assigned the weights on a partition-wise basis as reciprocal 20 outgroups. The complete alignment of BRCA, exon 11 to the minimum possible amount of change for the parti- comprised 840 bp (628 variable, 418 parsimony-informa- tion as a whole (equal to the total number of states for all tive) for 22 specimens of Dipodoidea and 15 outgroups. characters within a partition minus the number of charac- For RAG1, the alignment included 1143 bp (429 variable, ters). In both cases, weighting was expected to reduce the 285 parsimony-informative) for 22 specimens of Dipodoi- impact of dental traits compared with unweighted analysis. dea and 13 outgroups. The best-fit substitution models In total, six analyses employing different combinations of employed for each of the 12 partitions are given in character ordering (O vs. U) and weighting (W0, Wc, Table S1. Wp) were performed. All reconstructions were conducted in PAUP*. Starting topology was obtained via step-wise Base composition addition with random order (20 replicates). The search for Base composition at the 3rd codon position is presented in optimal trees was conducted using TBR branch swapping the Table S2. Individual genes demonstrated significant with MULPARS option on. To sum up the output, which departure from homogeneity, which is hardly a result of consists of several equally parsimonious trees, majority- multiple comparisons given highly significant (P < 0.001) rule consensus trees were reconstructed. To test clade values of disparity index. In IRBP data set, CG content stability, bootstrap analysis was performed with 1000 pseu- was much higher in Dipodinae (except Paradipus) and Sal- doreplicates using the same options of tree search as above pingotus (but not Cardiocranius) than in other groups; in but with MULPARS set to off. To check for possible con- RAG1 data set, CG proportion was notably higher in flict among partitions, ILD test was used as implemented Zapodinae and Sicista. There was no common trend of CG in PAUP* with 1000 replicates. variation across genes; thus, in Dipodinae, the CG content To test whether morphological data are compatible with was relatively high in IRBP but rather low in RAG1. the molecular topology, all reconstructions were repeated For data recoded into purines and pyrimidines, neither imposing molecular-derived constraints. The trees thereby test was significant at P < 0.01 and just five tests (four for obtained were compared with the unconstrained ones IRBP and 1 for GHR) were significant at P = 0.01. The using nonparametric Templeton (Wilcoxon signed-ranks) latter might be attributed to a multiple comparison arte- test (Templeton 1983). Following recommendations of fact. As the R ⁄ Y proportion appeared relatively stable,

ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters 7 Molecular phylogeny of Dipodoidea d V. S. Lebedev et al. the RY analysis is not expected to produce biased tree (original or R ⁄ Y-recoded). This tree (Fig. 2, see also topologies. Fig. S1) was well resolved with most suprageneric clades receiving high or absolute support (i.e. bootstrap support Phylogenetic results (BS) of 100% and posterior probability (PP) of 1.0). Sicista The results of combined analysis (original data and RY appears as the first group to diverge followed by the Zapo- recoded data) as well as those obtained based on each dinae (BS = 100%, PP = 1). Among the Zapodinae, Eoza- independent gene are presented in Figs 2 and 3. It appears pus is placed as the sister group to the American genera that although the best topologies inferred from individual Zapus and Napaeozapus (BS = 100%, PP = 1). Cardiocranii- genes were not completely identical, the four alignments nae is recovered as a monophyletic group (BS = 100%, were combinable: in neither case, the AU test rejected the PP = 1) branching as sister to all other jerboas H0 (P = 0.29, 0.27, 0.26, 0.28 for IRBP, GHR, BRCA1, (BS = 100%, PP = 1). Within the latter, the five-toed and RAG1, correspondingly) indicating that for each gene the three-toed jerboas (Allactaginae and Dipodinae) constitute fit of its individual ML tree was not significantly better a clade closely related to Euchoreutes. However, the posi- than that of the ML tree generated from the combined tion of Euchoreutes is just weakly supported (BS = 65–68%, data. Moreover, in neither case, a group recovered in an PP = 0.74), and hence, based on these data, the relation- individual gene analysis but incompatible with the com- ship between Euchoreutinae, Allactaginae and Dipodinae bined topology received bootstrap support higher than should be regarded rather as an unresolved trichotomy. 61%. The results of the ILD test as well indicated the lack Within the Dipodinae, the genus Paradipus descends from of conflict among the four partitions (P = 0.155). There- the basal-most split (BS = 97–100%, PP = 1), Dipus is next fore, we will focus on the topology inferred from the to branch off, while Stylodipus is recovered as the sister combined analysis. (For details of single-gene analyses see group to a clade composed of Eremodipus and Jaculus. Data S1). Within the Allactaginae, the genus Allactaga appears as pa- The same optimal tree topology was inferred from the raphyletic in respect to both Pygeretmus and Allactodipus. concatenated alignment for Dipodoidea whatever the Indeed, A. elater was most closely related to Pygeretmus methods considered (MP, ML, BI) or the data coding used rather than to any other Allactaga representatives

Fig. 2. The maximum-likelihood (ML) phylogeny of Dipodoidea as inferred from a concatenated alignment of four nuclear genes (outgroups not shown, for the complete version see Fig. S1). Values above the branches are the Bayesian posterior probabilities (BPP) and bootstrap support (1000 pseudoreplicates) in maximum-likelihood (ML) and maximum parsimony (MP) analyses, respectively. Values below the branches refer to corresponding values of support obtained in transversion-based analyses. ‘*’ denotes bootstrap support of 100% and BPP of 1.0; ‘-’ indicates support values of <50%. Specimen codes are as in Table 1.

8 ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters V. S. Lebedev et al. d Molecular phylogeny of Dipodoidea

AB

CD

Fig. 3. Maximum-likelihood (ML) gene trees for Dipodoidea based on separate analysis of —A. IRBP, —B. GHR, —C. BRCA1 and —D. RAG1. Values above the branches correspond to bootstrap support (1000 pseudoreplicates) in the ML analyses. Specimen codes are as in Table 1.

(BS = 83–95%, PP = 1), while A. bobrinskii was placed as Table 2. Results of the approximately-unbiased (AU) test of a sister to A. major (BS = 84–96%, PP = 1). priori morphology based hypotheses on Dipodoidea phylogeny The analysis based on the RY recoded alignment pro- duced essentially the same results. Most nodes retained The results of the AU-test: high support with the exception of the position of Stylodi- P – probability that H0 is pus. Thus, data recoding does not result in a significant correct given molecular data loss of phylogenetic signal. Also, it can be concluded that H0: All the observed base-compositional heterogeneity, whatever morphology-based hypothesis on dipodid substitution Transversions pronounced, does not introduce any bias into phylogenetic phylogenetic relationships types only reconstructions based on the original combined data set. Dipodinae + Cardiocraniinae is <0.01 <0.01 According to the results of the AU tests (Table 2), all monophyletic but one a priori advanced hypothesis are rejected at least Paradipus is sister to Cardiocraniinae <0.01 <0.01 Paradipus is sister to Eremodipus + Jaculus <0.01 0.017 at P < 5%. Euchoreutes is basal relative to the three <0.01 <0.01 A comparison of the levels of intragroup divergence as other groups of jerboas assessed from genetic distance at the basal nodes of the Euchoreutes is sister to Allactaginae 0.19 0.18 main jerboa clades demonstrates that the highest differen- Euchoreutes is sister to Sminthinae <0.01 <0.01 tiation is observed within the Cardiocraniinae (5.7% Sminthinae + Zapodinae is monophyletic <0.01 0.029 Allactaga is monophyletic relative to <0.01 <0.01 between Cardiocranius and Salpingotus). The clade of Pygeretmus and Allactodipus five-toed jerboas is relatively shallow (1.7–2.2% between

ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters 9 Molecular phylogeny of Dipodoidea d V. S. Lebedev et al.

Pygeretmus + A. elater and the rest of allactagines), whereas blage appeared to be paraphyletic in regards to Cardi- three-toed jerboas are significantly more divergent (3.9– ocraniinae in all weighted analyses. 4.6% between Paradipus and other dipodine genera, 3.3– 5. The support for the group of true jerboas (ignoring the 3.8% between Dipus and Eremodipus + Jaculus). position of Euchoreutes) should be considered as unstable. The clade was retrieved in half of the analyses Parsimony analysis of morphological data with some bootstrap support (59–63%) in Wp only. The data set for morphological analysis included charac- 6. Eozapus showed just insignificant tendency to group ters of dentition, auditory bulla, glans penis and accessory with other zapodines – in all analyses other than OW0 reproductive glands. The ILD tests suggest significant dis- jumping mice were paraphyletic. cordance among these character partitions (P < 0.01 for all 7. Basal radiation appeared poorly resolved: if the basal variants of ordering and weighting). However, this test is node was not reduced to a multifurcation, basal posi- known to be too liberal due to its sensitivity to various tion was occupied either by zapodines (OW0) or by confounding factors (see e.g. Barker & Lutzoni 2002); Sicista (UWp&c). therefore, we consider the positive result as a preliminary indication of heterogeneity within morphological data. Constrained analyses The nature of this incongruence requires further explora- In the constrained analyses, the following well-supported tion. At the same time, the fact that H0 is rejected does features of the molecular tree were used as constraints: all not necessarily imply that combined approach is inade- subfamilies are monophyletic; Sicista is sister to all other quate (Hipp et al. 2004); besides, some of our partitions dipodoids; jumping mice is the sister group to the jerboa include too few characters for their separate analysis, so clade; Zapus + Napaeozapus is monophyletic; Cardiocranii- here we focus on the results for the combined data. nae is sister to a trichotomy of Dipodinae, Allactaginae Data re-analysis with original settings (unequal transfor- and Euchoreutinae; Paradipus is sister to other dipodines. mation costs with equal character weights and irreversible If all constraints were enforced simultaneously, the trees evolution postulated for many traits – O-W0) reproduced inferred under all regimes other than OWp were signifi- a tree very similar to the one previously obtained by cantly less parsimonious than corresponding unconstrained Shenbrot (1992) with the use of the WISS algorithm. trees (P < 0.05, Templeton test, see Table S3). However, other combinations of character ordering and The four morphological partitions differed in the level weighting produced topologies deviating from the former of incongruence with the molecular-based tree (P < 0.05, in several aspects (see Fig. 4). The following points should Kruskal–Wallis test; for details see Table S4). The be noticed: decrease in the mean consistency index for the constrained 1. The position of Euchoreutes was sensitive to changes in topologies was more pronounced in dental characters as character weighting – it was sister to Sicista with un- compared to other partitions. In contrast, traits of genital weighted data (as in Shenbrot 1992), while down- morphology fit nearly equally well on both molecular and weighting of dental traits resulted in its association with morphological trees. However, insufficient character sam- Allactaginae (O-Wp&c, U-Wp&c); no bootstrap sup- pling precluded more rigorous testing of the hypothesis port for any of these relationships was obtained with the that dental partition is responsible for most of the conflict exception of O-W0 analysis where Euchoreutes + Sicista between morphological and molecular evidence. clade received bootstrap support of just 57%. Individual constraints demonstrated unequal contribution 2. The Cardiocraniinae and Dipodinae constitute a stable to the increase in tree length (see Table S5). Most con- clade, which was retrieved in all analyses; it received straints required few (1–2) extra steps; however, topologies moderate to high bootstrap support (78–93%) with the constrained for the monophyly of Euchoreutinae + Dipodi- exception of analyses using Wp weighting. nae + Allactaginae involved 12 and eight additional steps in 3. Paradipus was associated with Cardiocraniinae in all ordered and unordered analysis, respectively. The cost of variants other than U-W0 analysis in which it is placed enforcing all other constraints together was lower (five steps at the very top of the dipodine clade as a sister group in both cases). This result is consistent with the fact that our of Eremodipus; however, neither of its positions received morphological data robustly support the monophyly of Di- significant bootstrap support. podinae + Cardiocraniinae and, hence, contradict the Eu- 4. In most cases, Allactaginae and Cardiocraniinae choreutinae + Dipodinae + Allactaginae clade. emerged as monophyletic groups with at least some support. In contrast, both Dipodinae sensu lato and Di- Character mapping podinae sensu stricto (without Paradipus) never received Character mapping was performed using both molecular any support above 50%. Moreover, the latter assem- and modified morphological tree (Figure. 5). The only dif-

10 ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters V. S. Lebedev et al. d Molecular phylogeny of Dipodoidea

Fig. 4. Phylogenetic relationships among Dipodoidea species as follows from parsimony analysis of morphological data by Shenbrot (1992, with modifications). Six analyses were performed: data were treated as either ordered (O) or unordered (U), characters were either given equal weights (W0) or weighted according to one of the two schemes (Wc or Wp, see text for details). Topologies correspond to majority-rule consensus trees inferred from all equally parsimonious trees. Compatible nodes present in <50% trees are shown as well. Numbers above branches designate percentage of most parsimonious trees supporting corresponding splits, and numbers below branches indicate bootstrap support. ‘*’ stand for 100%; ‘-’ designate bootstrap support of <50%. Numbers of equally parsimonious trees, values of consistency index (CI) and retention index (RI) are given above each dendrogram.

ference between the two topologies lies in the placement Inspection of character state changes showed that there of Dipodinae as sister to Cardiocraniinae in the morpho- are no apparent synapomorphies for Euchoreutinae + Di- logical tree. This choice of morphological topology is podinae + Allactaginae in our data set. justified by the fact that Cardiocraniinae + Dipodinae is In contrast, Dipodinae + Cardiocraniinae was supported the only morphologically well-supported grouping which by the following derived character states (all dental) is incompatible with the molecular phylogeny. regardless of ordering scheme:

ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters 11 Molecular phylogeny of Dipodoidea d V. S. Lebedev et al.

A B

Fig. 5. —A. Parsimony character mapping on molecular and —B. modified morphological trees. Dental and bulla traits are ordered (see information on characters A5, A7, B1 in the Data S1). Arrows and notes on branches indicate inferred character transformations (unique or homoplastic) and specify the derived states. Most of transformations that are mapped identically on both trees are omitted from the morphological tree. See legend for character state designations.

1. partial to complete reduction of anteroloph; support for Dipodinae + Cardiocraniinae (see also data on 2. reduction of mesoloph, (with a parallelism in Pygeret- partitioned Bremer support, Table S6). Regardless of the mus) (Fig. 5); topology used, the development of highly inflated 3. paracone (not protocone) connected via entololoph multi-chambered bulla in Cardiocraniinae, Euchoreutes and with hypocone (or conditions derived from this), (with Dipodinae is explained by multiple parallelisms; at the a parallelism in Zapus + Napaeozapus) (Fig. 5); same time, this character showed less homoplasy on the 4. reduction of mesolophids on both m1 and m2; morphological tree. 5. entoconid (not hypoconid) connected via longitudinal Traits associated with advanced bipedalism (not crest with paraconid (or conditions derived from this), included in our dataset) such as cylindrical shape of caput (with a parallelism in Zapus + Napaeozapus); femoris and fused metatarsus showed no homoplasy on 6. presence of labial crest on lower molars (or conditions the molecular tree and support Euchoreutinae + Dipodi- derived from this). nae + Allactaginae; in contrast, morphological tree Taken together, these changes illustrate transition to required parallel acquisition of derived states in at least shorter and broader molar crown with opposite cusps, two groups of jerboas. reduced mesostyles(-ids) and diagonally positioned longi- tudinal crests. Correspondingly, the molecular tree sug- Discussion gested that this advanced molar pattern evolved Phylogenetic relationships within Dipodoidea independently in three-toed and pygmy jerboas. Our results show that internal consistency among the Cardiocraniinae and most of Dipodinae share specific molecular data and its information content are high bulla morphology characterized by extremely inflated mas- enough to produce a well-supported molecular hypothesis toid; however, bulla traits do not provide unambiguous on phylogenetic relationships among the main dipodoid

12 ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters V. S. Lebedev et al. d Molecular phylogeny of Dipodoidea lineages. Below, we address the main issues emerging from between specialized jerboas (Dipodinae and Allactaginae) the confrontation of the molecular and the morphological and sminthines (namely, generalized zygoma, less special- data focusing on potential taxonomic implications. ized pelvis and hind limb). The same pattern was after- wards recovered in the cladistic analysis by Stein (1990) Basal branching order in Dipodoidea. Our data support but with poor bootstrap support (<60%, our re-analysis). previous morphological (Stein 1990; Potapova 1976) and According to an alternative view (Shenbrot 1992), Euchore- molecular (Jansa & Weksler 2004; Montgelard et al. 2008; utes is considered as an independent derivative from ances- Jansa et al. 2009) findings confirming that the basal split is tral dipodoid stock, which shares true synapomorphies between Sicista and other dipodoids. The Zapodinae only with Sicista. This implies that similarity between appears as the sister group of the monophyletic jerboas. Euchoreutes and other jerboas should be treated as a result Accordingly, this pattern contradicts the view that the of parallel (or convergent) evolution. However, our nonbipedal birch mice (Sminthinae) and jumping mice re-analysis of morphological data indicates that such a (Zapodinae) constitute a monophyletic group as proposed position of the long-eared jerboa is neither well supported by Vinogradov (1937) and subsequently advocated by nor insensitive to variations in character weights. In most some morphologists (e.g. Gambaryan et al. 1980) and pale- of weighted analyses, Euchoreutes demonstrates affinity to ontologists (Shevyreva 1983; Zazhigin & Lopatin 2000b). Allactaginae, but with insignificant bootstrap support. The Most paleontological scenarios agree with a divergence molecular data conflict both of the above morphology- among major lineages in Late Eocene – Oligocene when based scenarios, supporting instead the monophyly of the diversity of sminthine-like dipodoids was high (Shevyr- Euchoreutinae + Dipodinae + Allactaginae. This result is eva 1983; Zazhigin & Lopatin 2000b). However, the rela- in line with the opinion of Potapova (2000a) that dental tionships among fossil genera as well as the ancestry of traits shared by the long-eared jerboa, Euchoreutes and Sici- modern groups appear controversial (compare Wang & sta should be regarded not as synapomorphies but as a Qiu 2000; Zazhigin & Lopatin 2000b; Lo¨pez-Antonanzas result of parallel adaptation to insectivory. The hypothesis & Sen 2006). Zazhigin & Lopatin (2000b) identified two that Euchoreutes is related to five-toed jerboas as argued by Oligocene lineages, Heosminthus -Plesiosminthus and Sinos- Zazhigin & Lopatin (2000b, 2005) receives no corrobora- minthus -Parasminthus, as the ancestral groups for Zapodi- tion but cannot be formally rejected. Our data do not con- nae-Sminthinae and Allactaginae-Dipodinae, respectively. tain sufficient information to resolve the trichotomy However, our topological tests clearly rejected the mono- among allactagines, dipodines and Euchoreutes; at the same phyly of the former association, thus suggesting that fossil time, one may hypothesize that the three lineages diverged data should be revised to determine which of the Paleo- in rapid succession within a relatively short time span. gene taxa represent stem-groups of the three – but not The earliest fossils attributed to the Euchoreutinae are two – main dipodoid clades. found in Early Miocene (MN 3–4, 16–20 Mya) what is not significantly later than the earliest record for allacta- Monophyly of jerboas. One of the most important results gines (Zazhigin & Lopatin 2005). Therefore, the split stemming from our study is the monophyly of bipedal between the three jerboa lineages must have occurred not Dipodoidea including Euchoreutinae and Cardiocraniinae later than the Early Miocene. in congruence with many previous authors (Vinogradov 1937; Vorontsov et al. 1971; Stein 1990). Although this The position of Cardiocraniinae and Dipodinae. The molec- group comprises highly divergent lineages (either at a ular data reject genealogical affinity between Dipodinae morphological point of view or a genetic one), its cohe- and Cardiocraniinae, which was supposed based on mor- siveness is supported by an unprecedentedly stable karyo- phology (Shenbrot 1985, 1992; Stein 1990; Zazhigin & type (2n = 48) shared by all genera except Salpingotus and Lopatin 2000b, 2001) and suggest that the line leading to Stylodipus (Vorontsov et al. 1971; Orlov & Yatsenko 1985). the pygmy jerboas (Cardiocraniinae) was the first to split from the common ancestor of bipedal taxa. According to Phylogenetic position of Euchoreutes. Previously, the mono- the hypothetical paleontological scenario (Zazhigin & phyly of bipedal dipodoids was questioned, primarily due Lopatin 2000b), all extant jerboas descended from two to the controversial phylogenetic position of the long- ancestral lineages, which had separated from each other by eared jerboa, Euchoreutes naso. It has been commonly the end of Oligocene and evolved later into Allactaginae agreed that this species is a single survivor of some ancient and Lophocricetinae, respectively. Now-extinct Lophocri- lineage. Vinogradov (1930, 1937) suggested that Euchore- cetinae is regarded as the immediate ancestor of the Car- utes is sister to all other bipedal taxa based on the observa- diocraniinae (Shenbrot et al. 1995) and, indirectly, of the tion that this species retains some characters intermediate Dipodinae. The latter taxon is believed to have originated,

ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters 13 Molecular phylogeny of Dipodoidea d V. S. Lebedev et al. in its turn, from some early Cardiocraniinae (Shenbrot aga relative to morphologically distinct Pygeretmus and 1992; Zazhigin & Lopatin 2001). Apparently, the molecu- Allactodipus. In our combined tree, Pygeretmus form a clade lar position of the Cardiocraniinae (but not of the Dipodi- with Allactaga elater, while Allactodipus is grouped with All- nae) is consistent with the hypothesis of its origin from actaga major. Notwithstanding the fact that this result the Lophocricetinae. It remains unclear which fossil appears highly supported (see results of the AU test), we taxa might be a likely candidate for the ancestry of the recommend that it should be treated with caution taking Dipodinae. into account relatively low number of informative sites available at this level of divergence what could mask Monophyly of Cardiocraniinae. A cladistic analysis of geni- potential conflict among genes. At the same time, from a tal morphology suggested that Cardiocraniinae is not morphological viewpoint, we cannot reject a hypothesis monophyletic indicating substantial differentiation between that Allactaga is a paraphyletic assemblage ancestral to two modern branches represented by Cardiocranius and Sal- both of the other allactagine genera. Among the morpho- pingotus (Pavlinov & Shenbrot 1983). The same idea was logical characters considered, only the shape of glans penis independently proposed based on cytogenetic data (Orlov can be counted as a potential synapomorphy for Allactaga & Yatsenko 1985). The molecular evidence also demon- sensu lato. Paleontological record indicates that the first strates that the two genera are highly divergent, but at the unequivocal allactagines appear in the Lower Miocene of same time, it consistently supports the monophyly of car- (MN 3–4, 16–20 Mya), while the earliest fossils attributed diocraniines, thus, confirming morphological results to the genus Allactaga (subgenus Paralactaga) are found in obtained with more comprehensive data sets (Shenbrot MN11 (7.5–8.7 Mya) (Zazhigin & Lopatin 2000a). The 1992). The first representatives of Cardiocraniinae are genera Protalactaga, Proalactaga and Paralactaga in their found in the earliest Late Miocene (MN 9, 9.7–11.1 interpretation by most paleontologists are rather the stages Mya – Li & Zheng 2005) or even late Middle Miocene of evolutionary development of molar crown patterns (MN 7 + 8, 11.1–12.5 Mya – Zazhigin & Lopatin 2005). (grades) than real phylogenetic clades. Therefore, the roots of different lineages within the genus Allactaga sensu lato Relationships among the three-toed jerboas, the could be significantly deeper than it is usually believed. It Dipodinae. Considering phylogenetic relationships among is commonly accepted that Pygeretmus is a rather recent dipodines, morphologists generally agree in placing Dipus (Latest Miocene) derivative of Allactaga, which has evolved basal to Stylodipus, Jaculus and Eremodipus and in treating towards more specialized herbivore diet (Shenbrot 1984; the latter two genera as sister taxa based on their advanced Zazhigin & Lopatin 2000a). In contrast to that, Allactodi- molar morphology (Vinogradov 1937; Heptner 1984; pus was hypothesized to be an early offshoot of the ances- Shenbrot 1992). Our molecular tree is fully congruent tral allactagine stock, thus, representing a separate with this pattern. The major point of disagreement evolutionary lineage (Zazhigin & Lopatin 2000a). Our between different morphology-based phylogenies is the data strongly contradict this view indicating instead that position of Paradipus – a monotypic genus, which is char- Allactodipus lineage originated at approximately the same acterized by a unique combination of autopomorphic and time as major clades within Allactaga. In his review of shared derived features. According to Vinogradov (1937), postcranial skeletal morphology in Dipodoidea, Fomin Paradipus is a crown dipodine close to Eremodipus + Jaculus (2006) mentioned that Allactodipus shares some similarity clade. Alternatively, it is regarded as a separate lineage just in pelvis shape with A. major (and A. severtzovi). Although distantly related to other Dipodinae (Heptner 1984) or this finding is in line with the pattern inferred from even as a sister group to Cardiocraniinae, thus, indicating genetic data, the phylogenetic value of this condition extensive parallelism between Paradipus and other three- remains unclear. toed jerboas (Shenbrot 1992; Shenbrot et al. 1995; Potap- ova 1998). Molecular evidence rejects both the first and Incongruence between molecules and morphology and the last hypotheses and supports Paradipus as the most evolution of bipedalism divergent lineage within Dipodinae. The re-analysis of DNA versus morphology: is the conflict real? Our re-analysis morphological matrix demonstrated that the tendency for of morphological data demonstrated that some non-trivial Paradipus to group with Cardiocraniinae is dependent on groupings retrieved in the previous study (i.e. Sicista + Eu- initial assumptions (character ordering and weighting). choreutes, Paradipus + Cardiocraniinae) are, in fact, poorly supported and unstable. Consequently, most of the appar- Phylogenetic relationships within the five-toed jerboas, the ent discrepancies between molecular phylogeny and mor- Allactaginae. One of the most striking results of our molec- phology should be attributed to topological instability in ular analysis is the lack of monophyly of the genus Allact- morphological reconstructions. This would be due to

14 ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters V. S. Lebedev et al. d Molecular phylogeny of Dipodoidea insufficient information content and higher homoplasy rate from non-ricochetal Sicista via ricochetal but quadrupedal typical for the latter type of data (see Springer et al. 2007, zapodines to primitively bipedal cardiocraniines and, 2008 for a discussion). However, concerning the position finally, to highly specialized Allactaginae and Dipodinae. of Cardiocraniinae, this does not seem to be the case. Our Thus, it appears that each stage of progressive specializa- current morphological analyses recapitulate Dipodi- tion might have been attained just once. nae + Cardiocraniinae clade with bootstrap support of up However, although the available genetic data do not to 93%. Supposed synapomorphies for this association are support the hypothesis of independent acquisition of key presented mostly by dental features (Shenbrot 1992; our morphological adaptations to saltatory locomotion in dif- analysis). The phylogenetic signal of the dental partition ferent branches of Dipodoidea, adaptive parallelisms could may have been distorted by character non-independence have played certain role in its locomotory evolution. Thus, or higher homoplasy rate (see Naylor & Adams 2001); the finding of a Middle Miocene Protalactaga with incom- however, even severe downweighting of dental characters pletely fused metatarsus (Zazhigin & Lopatin 2000a) sug- (Wp variant) yielded the same overall topology. gests that the formation of the advanced state (complete Furthermore, Dipodinae + Cardiocraniinae clade was fusion) observed in modern allactagines and dipodines sig- recovered with bootstrap support of 75–79% (our re-anal- nificantly postdates the split between them. ysis) in the cladistic analysis of hind and forelimb myology by Stein (1990). The two groups also share a number of Possible interactions between adaptive trends in jerboa derived postcranial features such as fused cervical vertebrae evolution. The source of the conflict between molecular (Vinogradov 1937) and specific structure of scapulae and morphological partitions (as well as within the lat- (Fomin 2006). ter) remains unclear. Adaptive parallelisms are often Thus, different types of morphological characters (den- regarded as the main reason of incongruence between tary, myological, osteological) support the phylogenetic morphological and molecular phylogenies (see Wiens position of Cardiocraninae as sister to Dipodinae what is et al. 2003 for a discussion). To provide a non-contro- definitely rejected by molecular data; therefore, we have to versial phylogenetic history of Dipodoidea based on the conclude that the conflict between the two kinds of data is molecular hypothesis, one needs to identify the factors significant. responsible for morphological parallelisms between Car- diocraniinae and Dipodinae and ⁄ or for the lack of such Molecular phylogeny and locomotory evolution: apparent parallelisms between other taxa (such as Dipodinae and concordance. At the same time, some morphological features Allactaginae). agree with molecular evidence (Fig. 5). The fact that Car- The pattern of morphological diversity in Dipodoidea is diocraniinae retain primitive condition in several traits of a result of an interplay among several adaptive trends. hindlimb morphology (i.e. spherical and not cylindrical Trophic adaptations range from omnivory (Sicista, most al- shape of caput femoris and lack of metatarsal fusion) is lactagines) to specialized insectivory (Euchoreutes), grani- consistent with their early separation from the stem of vory (most dipodines, Cardicraniinae) or folivory (a trend bipedal taxa (Vinogradov 1937). The same hypothesis was characteristic for allactagines and dipodines, with Pygeret- advocated by Fokin (1978, 1983) based on the comparative mus and Paradipus as the most specialized forms). Even study of pelvis and hindlimb myology. The latter work more important is the evolution of substrate adaptations. demonstrated that pygmy jerboas represent a mosaic of The Dipodinae and Allactaginae represent alternative both primitive and uniquely derived traits, suggesting trends: dipodines are predominantly psammophilous, while independent development of efficient saltatory locomotion. allactagines are better adapted to hard substrate (Shenbrot Another important question is whether bipedality was et al. 1995). Euchoreutinae and Cardiocraniinae are less acquired only once or evolved independently is several di- specialized showing more affinity for soft substrate. Allac- podoid lineages. Available paleontological data do not pro- tagines demonstrate a number of unique habitat-related vide direct support for either of these hypotheses (for locomotory adaptations including specific morphology of information on the earliest fossils of bipedal dipodoids see pelvis and hindlimb (Fokin 1978). In contrast to scratch- Zazhigin & Lopatin 2000a,b, 2005). The most parsimoni- digging cardiocraniines and dipodines, allactagines are ous scenario, given the molecular tree, suggests that the highly specialized to chisel-tooth digging, their mandible last common ancestor of extant jerboas was bipedal. More- being distinctively shallow and elongate, incisors procum- over, the molecular phylogeny, while at odds with most of bent (Potapova 2000b). At the same time, Allactaginae is morphological characters, is compatible with the general the only jerboa subfamily which shows no tendency to trajectory of locomotory evolution in dipodoids leading develop hyperinflated multi-chambered bulla. In other

ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters 15 Molecular phylogeny of Dipodoidea d V. S. Lebedev et al. lineages, the latter condition evolved independently at be expected based on dental similarity, it will be confirmed least four times, namely, in Euchoreutinae, Cardiocranii- that they belong to the same clade as Microallactaga then nae, Paradipus and Eremodipus + Jaculus. Also, five-toed jer- the three lineages should be recognized at subgeneric level boas retain rather primitive molar morphology (narrow and retained in a single genus (under the name Scarturus). crown with alternating cusps, mesostyle(-id) present) con- The proposed system for all taxa above the species rank trasting advanced molar pattern (broad crown with oppo- follows below. site cusps), which is observed (probably, as a parallelism) in the Cardiocraniinae and Dipodinae. This differentiation Dipodoidea between the lineages can hardly be explained exclusively Sminthidae Brandt, 1855 by difference in diet. Instead, one may hypothesize that Sicista Gray, 1827 substrate specialization in allactaginae (and ⁄ or other groups) may have narrowed the spectrum of adaptive options in the evolution of their skull and dentition. Zapodidae Coues, 1875 However, the exact nature of limitations imposed by spe- Zapodini cialized structure of jaw or bulla upon evolutionary change Zapus Coues, 1875. of the molar crown is yet unclear. Whether this effect Napaeozapus Preble, 1899. could significantly distort the phylogenetic signal of the Eozapodini Zazhigin and Lopatin, 2000 morphological data remains to be elucidated as well. Eozapus Preble, 1899.

Taxonomic implications Dipodidae Fischer de Waldheim, 1817 According to the taxonomy proposed herein Dipodoidea is Cardiocraniinae Vinogradov, 1925 divided into three families Sminthidae, Zapodidae and Di- Cardiocraniini podidae. The same pattern was previously suggested only Cardiocranius Satunin, 1903. by Vorontsov et al. (1971); yet, the latter classification was Salpingotini Vinogradov, 1925 based on similarity ⁄ dissimilarity between karyotypes and Salpingotus Vinogradov, 1922 (with subgenera hence cannot be regarded as strictly phylogenetic. In con- Salpingotus s.str., Salpingotulus Pavlinov, 1980, trast to that, our system explicitly adheres to the principle Prosalpingotus Vorontsov and Shenbrot, 1984; of monophyly. In previous classifications the four main Anguistodontus Vorontsov and Shenbrot 1984) lineages of jerboa were assigned either familial or subfami- Euchoreutinae Lyon, 1901. lial rank (Fig. 1). We prefer to treat them rather as sub- Euchoreutes Sclater, 1890. families within more comprehensive Dipodidae. This Dipodinae solution was chosen to emphasize the monophyly of all Dipodini bipedal taxa within Dipodoidea. Dipus Zimmermann, 1780 It should be mentioned that the system of Allactaginae Stylodipus G. M. Allen, 1925 remains controversial. The situation is complicated by the Eremodipus Vinogradov, 1930 position of Pygeretmus and Allactodipus as sister to different Jaculus Erxleben, 1777 (with subgenera Jaculus s. str. branches of Allactaga, thus making the latter paraphyletic. and Haltomys Brandt, 1844) Although the relationships within the five-toed jerboas Paradipodini Pavlinov and Schenbrot, 1983 appear well resolved and consistently supported by the Paradipus Vinogradov, 1930 current data, we believe that this group requires additional Allactaginae Vinogradov, 1925. examination based on an extended sample of species and Allactaga F. Cuvier, 1836 (with A. major and A. se- genes. If the phylogenetic pattern inferred here proves to vertzovi) be correct we will face a dilemma that we should either Orientallactaga Shenbrot, 1984 (with O. sibirica, include both Pygeretmus and Allactodipus into Allactaga or O. bullata and O. balikunica) elevate several subgenera within the latter (Orientallactaga, ? Scarturus Gloger, 1841 (with S. tetradactylus) Microallactaga) to full genera. Morphologically, Pygeretmus ? Paralactaga Young, 1927 (with P. euphratica, P. wil- and Allactodipus are well differentiated from Allactaga. Fol- liamsi, ?P. hotsoni) lowing this argument, we provisionally accept the second ? Microallactaga Shenbrot, 1974 (with M. elater and concept (elevation of some subgenera of Allactaga to full M. vinogradovi) generic rank). The taxonomic position and status of cer- Allactodipus Kolesnikov, 1937 tain taxa for which molecular data is so far unavailable Pygeretmus Gloger, 1841 (with subgenera Pygeretmus (Scarturus and Paralactaga) remain problematic. If, as might s.str. and Alactagulus Nehring, 1897)

16 ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters V. S. Lebedev et al. d Molecular phylogeny of Dipodoidea

Acknowledgements Fomin, S. V. (2006). Comparative morphological analysis of the DNA samples of Eremodipus lichtenstein, Pygeretmus pumilio, postcranial skeleton of Dipodoidea (Rodentia). PhD Thesis. Allactodipus bobrinskii and Allactaga major were kindly Moscow: Moscow State University, Sotsvetie krasok Press (in Russian). provided by Dr. D. Kramerov, tissue samples of Jaculus Gambaryan, P. P. (1983). Suprafamilial groupings in the order blanfordi were presented by A. Abramov, Eozapus setchuanus Rodentia. In I. M. Gromov (Ed.) Materials of the 6th All- and Sicista caucasica were presented by Jean-Pierre Que´re´ Union Conference on Rodents. (pp. 74–76). Leningrad: Nauka and Franc¸ois Catzeflis correspondingly. We are grateful (in Russian). to A.A. Surov, Yu.M. Kovalskya, L.A. Khlyap and Gambaryan, P. P., Potapova, E. G. & Fokin, I. M. (1980). I. Tikhonov for help in collecting material. We also thank Morphofunctional analysis of the myology of the jerboa head. I.Ya. Pavlinov, E.G. Potapova, S.V. Fomin, K.A. Rogovin Trudy Zoologicheskovo Instituta, Akademiya Nauk SSSR, 91, 3–51 (in Russian). and two anonymous reviewers for useful comments. The Heptner, V. G. (1984). Contributions to phylogeny and work was supported by RFBR 11-04-00020a. classification of jerboas (Dipodidae) of the fauna of the USSR. Archives of Zoological Museum Moscow State University, 22, 37–60 References (in Russian). Adkins, R. M., Walton, A. H. & Honeycutt, R. L. (2003). Hipp, A. L., Hall, J. C. & Sytsma, K. J. (2004). Congruence Higher-level systematics of rodents and divergence time versus phylogenetic accuracy: revisiting the incongruence length estimates based on two congruent nuclear genes. Molecular difference test. Systematic Biology, 53, 81–89. Phylogenetics and Evolution, 26, 409–420. Holden, M. E. & Musser, G. G. (2005). Superfamily Dipodoidea. Baker, R. H. & DeSalle, R. (1997). Multiple sources of molecular In D. E. Wilson & D. M. Reeder (Eds) Mammal Species of the characters and the phylogeny of Hawaiian drosophilids. World, 3nd edn, vol. 2 (pp. 871–893). Baltimore: Johns Hopkins Systematic Biology, 46, 654–673. University Press. Barker, F. K. & Lutzoni, F. M. (2002). The utility of the Huchon, D., Madsen, O., Mark, J. J., Sibbald, B., Ament, K., incongruence length difference test. Systematic Biology, 51, 625– Stanhope, M. J., Catzeflis, F., de Jong, W. W. & Douzery, 637. E. J. P. (2002). Rodent phylogeny and a timescale for the Blanga-Kanfi, S., Miranda, H., Penn, O., Pupko, T., DeBry, evolution of glires: evidence from an extensive taxon sampling R. W. & Huchon, D. (2009). Rodent phylogeny revised: using three nuclear genes. Molecular Biology and Evolution, 19, analysis of six nuclear genes from all major rodent clades. BMC 1053–1065. Evolutionary Biology, 9, 71. doi:10.1186/1471-2148-9-71. Jansa, S. & Weksler, M. (2004). Phylogeny of muroid rodents: Brown, J. M., Hedtke, S. M., Lemmon, A. R. & Lemmon, E. M. relationships within and among major lineages as determined (2010). When trees grow too long: investigating the causes of by IRBP gene sequences. Molecular Phylogenetics and Evolution, highly inaccurate Bayesian branch-length estimates. Systematic 31, 256–276. Biology, 59, 145–161. Jansa, S., Glarla, T. C. & Lim, B. K. (2009). The phylogenetic Churakov, G., Sadasivuni, M. K., Rosenbloom, K. R., Huchon, position of the rodent genus Typhlomys and the geographic D., Brosius, J. & Schmitz, J. (2010). Rodent evolution: back to origin of Muroidea. Journal of Mammalogy, 90, 1083–1094. the root. Molecular Biology and Evolution, 27, 1315–1326. Jobb, G. (2008). TREEFINDER version of October 2008. DeBry, R. W. (2003). Identifying conflicting signal in a Munich, Germany. Available via http://www.treefinder.de. multigene analysis reveals a highly resolved tree: the Klingener, D. (1964). The comparative myology of four dipodoid phylogeny of Rodentia (Mammalia). Systematic Biology, 52, rodents (genera Zapus, Napaeozapus, Sicista, and Jaculus). 604–617. Miscellaneous Publications, Museum of Zoology, University of Ellerman, J. R. (1940). The Families and Genera of Living Rodents. Michigan, 124, 1–100. Vol. 1. Rodents other than Muridae. London: Trustees of the Klingener, D. (1984). Gliroid and dipodoid rodents. In S. British Museum (Natural History). Anderson & J. K. Jones Jr (Eds) Orders and Families of Recent Fan, Z., Liu, S., Liu, Y., Zeng, B., Zhang, X., Guo, C. & Yue, B. Mammals of the World (pp. 381–388). New York: John Wiley & (2009). Molecular phylogeny and taxonomic reconsideration of Sons. the subfamily Zapodinae (Rodentia: Dipodidae), with an Kumar, S. & Gadagkar, S. R. (2001). Disparity Index: a simple emphasis on Chinese species. Molecular Phylogenetics and statistic to measure and test the homogeneity of substitution Evolution, 51, 447–453. patterns between molecular sequences. Genetics, 158, 1321– Farris, J. S., Ka¨llersjo¨, M., Kluge, A. G. & Bult, C. (1995). 1327. Testing significance of incongruence. Cladistics, 10, 315–319. Li, Q. & Zheng, S. H. (2005). Note on four species of dipodids Felsenstein, J. (1985). Confidence limits on phylogenies with a (Dipodidae, Rodentia) from the Late Miocene Bahe Formation, molecular clock. Systematic Zoology, 34, 152–161. Lantian, Shaanxi. Vertebrata Palasiatica, 43, 283–296. Fokin, I. M. (1978). The Locomotion and Morphology of the Locomotory Lo¨pez-Antonanzas, R. & Sen, S. (2006). New Saudi Arabian Organs in Jerboas. Leningrad: Nauka. 1–118 (in Russian). Miocene jumping mouse (Zapodidae): systematics and Fokin, I. M. (1983). Specific features of muscular apparatus in phylogeny. Journal of Vertebrate Paleontology, 26, 170–181. dwarf jerboas. In I. M. Gromov (Ed.) Materials of the 6th All- McKenna, M. C. & Bell, S. K. (1997). Classification of Mammals Union Conference on Rodents. (pp. 202–204). Leningrad: Nauka above the Species Level. New York: Columbia University (in Russian). Press.

ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters 17 Molecular phylogeny of Dipodoidea d V. S. Lebedev et al.

Michaux, J. & Catzeflis, F. (2000). The bushlike radiation of Shenbrot, G. I. (1984). Dental morphology and phylogeny of five- muroid rodents is exemplified by the molecular phylogeny of toed jerboas of the subfamily Allactaginae. Archives of Zoological the LCAT nuclear gene. Molecular Phylogenetics and Evolution, Museum Moscow State University, 22, 61–92 (in Russian). 17, 280–293. Shenbrot, G. I. (1985). Current status and aspects of the Michaux, J., Reyes, A. & Catzeflis, F. (2001). Evolutionary history phylogenetic study of jerboas. In V. E. Sokolov & V. V. of the most speciose mammals: molecular phylogeny of muroid Kucheruk (Eds) Jerboas of USSR (pp. 21–25). Moscow (in rodents. Molecular Biology and Evolution, 18, 2017–2031. Russian). Miller, G. S., Jr & Gidley, J. W. (1918). Synopsis of the Shenbrot, G. I. (1992). Cladistic approach to the analysis of supergeneric groups of rodents. Journal of the Washington phylogenetic relationships among dipodoid rodents (Rodentia, Academy of Sciences, 8, 431–448. Dipodoidea). Sbornik Trudov Zoologicheskovo Muzeya MGU, 29, Montgelard, C., Forty, E., Arnal, V. & Matthee, C. A. (2008). 176–201 (in Russian). Suprafamilial relationships among Rodentia and the Shenbrot, G. I., Sokolov, V. E., Heptner, V. G. & Koval’skaya, phylogenetic effect of removing fast-evolving nucleotides in Yu. M. (1995). Mammals of the Fauna of Russia and Contiguous mitochondrial, exon and intron fragments. BMC. Evolutionary Countries. Dipodoid Rodents. Moscow: Nauka Publishers (in Biology, 8, 321. doi: 10.1186/1471-2148-8-321. Russian); English translation (2008). Jerboas. Mammals of Russia Musser, G. G. & Carleton, M. D. (2005). Superfamily Muroidea. and Adjacent Regions.; scientific editors: R. S. Hoffmann & D. E. In D. E. Wilson & D. M. Reeder (Eds) Mammal Species of the Wilson. Enfield, New Hampshire: Science Publishers. World, 3nd edn, (2, pp. 894–1531). Baltimore: Johns Hopkins Shevyreva, N. S. (1983). Eurasian and North African Neogene University Press. rodents (Rodentia, Mammalia): the evolutionary basis for the Naylor, G. J. P. & Adams, D. C. (2001). Are the fossil data really Pleistocene and Recent Palearctic rodent fauna. In I. M. at odds with the molecular data? Morphological evidence for Gromov (Ed.) History and Evolution of the Recent Rodent Fauna Cetartiodactyla phylogeny reexamined. Systematic Biology, 50, (Neogene to Modern Times), I (pp. 9–145). Moscow: Nauka (in 444–453. Russian). Ognev, S. I. (1948). Mammals of the USSR and adjacent countries: Shimodaira, H. (2002). An approximately unbiased test of Rodents (continued). Mammals of eastern Europe and northern Asia. phylogenetic tree selection. Systematic Biology, 51, 492–508. Akademiya Nauk SSSR, 6, 1–559 (in Russian).). Simpson, G. G. (1945). The principles of classification and a Orlov, V. N. & Yatsenko, V. N. (1985). Comparative cytogenetics classification of mammals. Bulletin of the American Museum of and phylogeny of Dipodidae. In V. E. Sokolov & V. V. Kucheruk Natural History, 85, 1–350. (Eds) Jerboas of USSR (pp. 20–21). Moscow (in Russian). Springer, M. S., Burk-Herrick, A., Meredith, R., Eizirik, E., Pavlinov, I. Ya. & Shenbrot, G. I. (1983). Male genital structure Teeling, E., O’Brien, S. J. & Murphy, W. J. (2007). The and supraspecific taxonomy of Dipodidae. Trudy Zoologicheskovo adequacy of morphology for reconstructing the early history of Instituta, Akademiya Nauk SSSR, Leningrad, 119, 67–88 placental mammals. Systematic Biology, 56, 673–684. (inRussian). Springer, M. S., Burk-Herrick, A., Meredith, R., Eizirik, E., Posada, D. & Crandall, K. A. (1998). Modeltest: testing the Teeling, E., O’Brien, S. J. & Murphy, W. J. (2008). model of DNA substitution. Bioinformatics, 14, 817–818. Morphology and placental mammal phylogeny. Systematic Potapova, E. G. (1976). Structure of the tongue in some Biology, 57, 499–503. Dipodoidea. Zoologicheskii Journal, 55, 1383–1389 (in Russian, StatSoft, Inc. (2001). STATISTICA (data analysis software English abstract). system), Version 6. Available via http://www.statsoft.com. Potapova, E. G. (1998). Evolutionary development of osseus Stein, B. R. (1990). Limb myology and phylogenetic relationships middle ear in Rodentia, Dipodoidea. (Rodentia, Dipodoidea). in the superfamily Dipodoidea (birch mice, jumping mice, and Zoologicheskii Journal, 77, 80–87 (in Russian, English abstract). jerboas). Zeitschrift fu¨r Zoologische Systematik und Potapova, E. G. (2000a). Systematic position of Euchoreutes within Evolutionsforschung, 28, 299–314. Dipodoidea (Rodentia). In A. K. Agadzhanyan & V. N. Orlov, Steppan, S. J., Adkins, R. M. & Anderson, J. (2004). Phylogeny (Eds) Systematics and Phylogeny of the Rodents and Lagomorphs (P. and divergence-date estimates of rapid radiations in muroid 135–137). Moscow: KMK (in Russian). rodents based on multiple nuclear genes. Systematic Biology, 53, Potapova, E. G. (2000b). Pathways of transforming mandible 533–553. morphology in Dipodoidea (Rodentia) and the role of various Swofford, D. L. (2003). PAUP* – Phylogenetic Analysis Using adaptations in their specific features. Zoologicheskii Journal, 79, Parsimony (*and Other Methods), Version 4. Sunderland, 1445–1456 (in Russian, English abstract). Massachusetts: Sinauer Associates. Rambaut, A. & Drummond, A. (2005). Tracer version Tamura, K., Dudley, J., Nei, M. & Kumar, S. (2007). MEGA4: 1.5.Computer program distributed by the authors. Department molecular Evolutionary Genetics Analysis (MEGA) software of Zoology, University of Oxford, UK. Available via http:// version 4.0. Molecular Biology and Evolution, 24, 1596–1599. http://evolve.zoo.ox.ac.uk/software.html. Templeton, A. R. (1983). Phylogenetic inference from restriction Ronquist, F. & Huelsenbeck, J. P. (2003). MRBAYES 3: bayesian endonuclease cleavage site maps with particular reference to the phylogenetic inference under mixed models. Bioinformatics, 19, evolution of humans and the apes. Evolution, 37, 221–244. 1572–1574. Vinogradov, B. S. (1925). On the structure of the external Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecular genitalia in Dipodidae and Zapodidae (Rodentia) as a Cloning: A Laboratory Manual. New York: Cold Spring Harbor classificatory character. Proceedings of the Zoological Society of Lab. Press. London, 1925, 572–585.

18 ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters V. S. Lebedev et al. d Molecular phylogeny of Dipodoidea

Vinogradov, B. S. (1930). On the classification of Dipodidae Bulletin of Moscow Society of Naturalists, 110, 69–71 (in Russian, (Rodentia). I. Cranial and dental characters. Izvestiya Akademii English abstract). Nauk SSSR, 1930, 331–350 (in Russian). Vinogradov, B. S. (1937). Fauna of the USSR; Mammals, vol. 3, pt. Supporting information 4. Jerboas. Moscow-Leningrad: Academy Nauk SSSR Publishing Additional Supporting Information may be found in the House (in Russian). online version of this article: Vorontsov, N. N., Malygina, N. A. & Radjabli, S. I. (1971). Chromosomes of the jerboas (Rodentia, Dipodidae). Zoological Fig. S1 The maximum likelihood (ML) phylogeny of Journal, 50, 1853–1860 (in Russian). Dipodoidea and outgroup taxa as inferred from a concate- Wang, B.-Y. & Qiu, Z.-X. (2000). Dipodidae (Rodentia, nated alignment of four nuclear genes (designations as in Mammalia) from the lower member of Xianshuihe Formation Fig. 2). in Lanzhou Basin, Gansu, China. Vertebrata PalAsiatica, 38, 10– Fig. S2 Relative rates for different codon positions in 35 (in English with Chinese summary). four nuclear genes in Dipodoidea as estimated in ML Wiens, J. J., Chippindale, P. T. & Hillis, D. M. (2003). When are analysis. phylogenetic analyses misled by convergence? A case study in Texas cave salamanders. Systematic Biology, 52, 501–514. Table S1. Models for the three codon positions of the Yang, D. Y., Eng, B., Waye, J. S., Dudar, J. C. & Saunders, S. R. four nuclear genes employed in maximum likelihood (ML) (1998). Technical note: improved DNA extraction from ancient and Bayesian (BI) analyses (ML ⁄ BI). bones using silica-based spin columns. American Journal of Table S2. CG-content and proportion of purines in the Physical Anthropology, 105, 539–543. 3rd codon positions (%) of the four nuclear exons in Zazhigin, V. S. & Lopatin, A. V. (2000a). The history of the Dipodoidea. The values given in bold denote the most sig- Dipodoidea (Rodentia, Mammalia) in the Miocene of Asia: 3. nificant deviations from the rest of the sample. Allactaginae. Paleontological Journal, 34, 553–565 (translated from Paleontologicheskii Zhurnal, 5, 82–94). Table S3. The results of comparisons between molecu- Zazhigin, V. S. & Lopatin, A. V. (2000b). Evolution, phylogeny, lar-constrained and unconstrained trees (Templeton test). and classification of Dipodoidea. In A. K. Agadzhanyan & V. Table S4. The level of incongruence of four morpho- N. Orlov, (Eds) Systematics and Phylogeny of the Rodents and logical partitions with the molecular-constrained tree Lagomorphs (pp. 50–52). Moscow: KMK (in Russian). (Kruskal–Wallis test). Zazhigin, V. S. & Lopatin, A. V. (2001). The history of the Table S5. The impact of molecular-derived constraints Dipodoidea (Rodentia, Mammalia) in the Miocene of Asia: 4. on the length of MP trees. Dipodinae at the Miocene-Pliocene transition. Paleontological Journal, 35, 60–74 (translated from Paleontologicheskii Zhurnal, 1, Table S6. Partitioned Bremer support for the two con- 61–75). flicting nodes Cardiocraniinae + Dipodinae and Euchore- Zazhigin, V. S. & Lopatin, A. V. (2005). The history of utinae + Dipodinae + Allactaginae. Dipodoidea and the aridization of Asia during the Neogen. Data S1. Sminthinae ⁄ Sicistinae controversy.

ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters 19