Cheeked Ground Squirrels (Spermophilus 9 10 10 11 Erythrogenys Sensu Lato) Suggest Taxonomic Delineation 11 12 12 13 13 14 Vera A

Integrative Zoology 2019; 14: 341–353 doi: 10.1111/1749-4877.12383

1 ORIGINAL ARTICLE 1 2 2 3 3 4 4 5 5 6 6 7 Phylogenetic relationship and variation of alarm call traits of 7 8 8 9 populations of red-cheeked ground squirrels (Spermophilus 9 10 10 11 erythrogenys sensu lato) suggest taxonomic delineation 11 12 12 13 13 14 Vera A. MATROSOVA,1 Anastasia D. IVANOVA,1,2 Elena V. VOLODINA,3 Ilya A. VOLODIN,2,3 14 15 4 2 5 15 16 Dmitry Yu. ALEXANDROV, Olga V. SIBIRYAKOVA and Oleg A. ERMAKOV 16 17 1Engelhardt Institute of Molecular Biology, RAS, Moscow, Russia, 2Lomonosov Moscow State University, Moscow, Russia, 17 18 3Moscow Zoo, Moscow, Russia, 4Severtsov Institute of Ecology and Evolution, RAS, Moscow, Russia and 5Department of Zoology 18 19 19 and Ecology, Penza State University, Penza, Russia 20 20 21 21 22 22 23 Abstract 23 24 24 Distribution area and taxonomic borders within the species complex Spermophilus erythrogenys sensu lato re- 25 25 main questionable. Early evidence suggests that red-cheeked ground squirrels of Southeast Kazakhstan are re- 26 26 markably different in terms of the acoustic structure of their alarm calls from the red-cheeked ground squirrels 27 27 of the Kurgan region in Russia. In this study, we analyzed the differences in the acoustic structure of the alarm 28 28 call and mitochondrial DNA (complete control region, 1005–1006 bp and complete cytochrome b gene, 1140 29 29 bp) in 3 populations of red-cheeked ground squirrels (Tara, Altyn-Emel and Balkhash), all located within ar- 30 30 eas isolated by geographical barriers in Southeast Kazakhstan. We found that the alarm call variables were sim- 31 31 ilar between the 3 study populations and differed by the maximum fundamental frequency (8.46 ± 0.75 kHz) 32 32 from the values (5.62 ± 0.06 kHz) reported for the red-cheeked ground squirrels from the Kurgan region of Rus- 33 33 sia. Variation in mtDNA control region was only 3% and variation in cytochrome b gene was only 2.5%. Phylo- 34 34 genetic trees based on cytochrome b gene polymorphism of 44 individuals from the study area and adjacent ter- 35 35 ritories indicated 3 clades with high (98–100%) bootstrap support: “intermedius,” “brevicauda” and “iliensis”). 36 36 We conclude that the 3 study populations in Southeast Kazakhstan belong to the clade intermedius and suggest 37 37 a taxonomical revision of the species complex Spermophilus erythrogenys sensu lato, including analyses of nu- 38 38 clear DNA and alarm calls for populations of the brevicauda and iliensis clades. 39 39 40 Key words: alarm call, control region, cytochrome b, mitochondrial DNA, vocal communication 40 41 41 42 42 43 43 44 44 45 INTRODUCTION 45 46 46 Correspondence: Vera A. Matrosova, Russian Academy of 47 Red-cheeked ground squirrels of Southeast Kazakh- 47 48 Sciences, Division Department of Structural and Functional stan, commonly considered within taxon Spermophi- 48 49 Genomics, Engelhardt Institute of Molecular Biology, Vavilova lus erythrogenys sensu lato, represent an excellent mod- 49 50 str., 32, Moscow 119991, Russia. el for studying biogeographic speciation (Nikol’skii & 50 51 Email: [email protected] Rumyantsev 2004). Southeast Kazakhstan is a geo-mor- 51

1 phologically heterogeneous region with deep rivers, tions of ground squirrels from this area include both 1 2 mountains and lowlands providing numerous eco-geo- bioacoustical analyses (Nikol’skii & Rumyantsev 2004) 2 3 graphical barriers for ground squirrels, including the ar- and genetic analyses (Ermakov et al. 2015). 3 4 eas that are unfit for habitation although seem surmount- Considering the acoustic structure of the alarm call 4 5 able (Fig. 1). Studying populations of red-cheeked (primarily based on frequency modulation pattern of the 5 6 ground squirrels from this region might cast new light alarm call notes), Nikol’skii and Rumyantsev (2004) 6 7 on their geographical divergence (Harrison et al. 2003; subdivided the alarm calls recorded in 13 populations of 7 8 Herron et al. 2004; Thorington & Hoffmann 2005; Hel- ground squirrels (distributed over the large area shown 8 9 gen et al. 2009) and raise new questions about the by the dotted line in Fig. 1) into 4 clusters, potentially 9 10 mechanisms by which populations diverged (Zachos corresponding to the 4 species (S. erythrogenys, Sper- 10 11 2018). mophilus brevicauda, Spermophilus carruthersi and 11 12 Geographical boundaries between ranges of differ- Spermophilus major) with non-overlapping distribution 12 13 ent species and forms of ground squirrels from South- areas; the boundaries between the clusters corresponded 13 14 east Kazakhstan are presently unknown. Approaches to large zoogeographic barriers (Fig. 1). Among them, 14 15 for identifying taxonomical ranks of different popula- most distinctive were 4 individual red-cheeked ground 15 16 squirrels from the Altyn-Emel area: their alarm calls 16 17 were unusually high-frequency, with the maximum fun- 17 18 damental frequency of over 8 kHz compared to values 18 19 below 7 kHz in ground squirrels in other studied popu- 19 20 lations. The authors hypothesized that animals from the 20 21 Altyn-Emel area belong to the form “carruthersi” with- 21 22 in S. erythrogenys sensu lato; however, unequal sam- 22 23 ples of animals from different populations (from 1 to 9) 23 24 and small numbers of alarm calls available for analyses 24 25 (1 per individual) along with their unknown genetic sta- 25 26 tus prevented more decisive conclusions about their tax- 26 27 onomical position (Nikol’skii & Rumyantsev 2004). To 27 28 date, the alarm call acoustics of ground squirrels from 28 29 the Altyn-Emel area have not been studied in detail. 29 30 30 A recent barcoding study (Ermakov et al. 2015), 31 31 based on polymorphism of cytochrome c oxidase sub- 32 32 unit 1 (COI) of all Eurasian species of ground squir- 33 33 rels, showed a paraphyly of S. erythrogenys, S. brevi- 34 34 cauda and Spermophilus pallidicauda. Moreover, on the 35 35 basis of mtDNA polymorphism, 5 individuals of sup- 36 36 posed S. erythrogenys turned out to be 3 genetically dis- 37 37 tant forms, for which a putative species level rank was 38 38 proposed: (i) S. erythrogenys (right bank of the Irtysh 39 39 River); (ii) another, previously unknown, form from the 40 40 right bank of the Ob’ River (Kuznetsk Depression); and 41 41 (iii) S. carruthersi (Zaysan Depression, the Dzungar- 42 42 ian Alatau). Thus, although it has been suggested that 43 43 ground squirrels inhabiting the area of the Dzungarian 44 44 Figure 1 Map showing the study area (a) and data collection Alatau including the Altyn-Emel area should be consid- 45 45 localities (b). Sites of matched alarm call and DNA sampling ered as S. carruthersi, the available genetic and acoustic 46 46 are marked in white; sites of DNA only sampling are marked in data are not sufficient to support this suggestion. 47 black (see Table 1 for details). Stars designate type localities: 47 Formerly, taxon S. erythrogenys sensu lato was inter- 48 “bre” – S. brevicauda Brandt, 1843; “int” – S. intermedius 48 preted as a single wide-range polymorphic species com- 49 Brandt, 1843; “car” – S. carruthersi (Thomas, 1912); “ili” – S. 49 plex involving all ground squirrels from Kazakhstan 50 e. iliensis (Belyaev, 1945). The dotted line delineates the study 50 51 area by Nikol`skii & Rumyantsev (2004). to Central Mongolia from west to east and all ground 51

1 squirrels from Southwest Siberia to Tien Shan from into the phylogeography of the ground squirrels inhabit- 1 2 north to south (Belyaev 1955; Gromov et al. 1965; Vasi- ing this area. 2 3 lyeva 1968; Kryštufek & Vohralík 2013). From anoth- 3 4 er point of view, S. erythrogenys sensu lato includes a MATERIALS AND METHODS 4 5 few morphologically different forms considered as sepa- 5 6 rate species (Afanas’ev et al. 1953; Ognev 1963; Sluds- Study sites, animals and dates 6 7 kiy et al. 1969). Recently, these forms with unclear tax- 7 8 onomic status were separated into 3 species based on Alarm call and DNA samples of red-cheeked ground 8 9 cytochrome b gene (cyt b) polymorphism: S. erythroge- squirrels were collected in Southeast Kazakhstan from 9 10 nys, inhabiting the northern part of the area; S. brevicau- 10 to 24 June 2015 in the 3 natural colonies which are 10 11 da, inhabiting the southern part of the area and S. pallid- representative of the 3 study populations, “Tara,” “Al- 11 12 icauda, inhabiting the eastern part of the area (Harrison tyn-Emel” and “Balkhash” (Table 1). All the 3 habi- 12 13 et al. 2003; Herron et al. 2004; Thorington & Hoffmann tats represented a dry grazing steppe. We examined 38 13 14 2005; Helgen et al. 2009). living subjects: 30 live-trapped and 8 free-ranging; 10 14 were from the Tara, 10 from the Altyn-Emel and 18 15 In rodents, vocalizations are not learned but inherit- 15 from the Balkhash population. Of the 38 living subjects, 16 ed genetically (Kikusui et al. 2011). Ground squirrels 16 22 provided matched samples of the alarm calls and 17 have species-specific alarm calls (Nikol’skii 1979, 1984; 17 DNA, whereas 16 individuals from the Balkhash pop- 18 Schneiderová & Policht 2011; Matrosova et al. 2012). 18 ulation provided unmatched samples of alarm calls and 19 Bioacoustic analysis has proved to be a useful taxonom- 19 DNA (8 individuals provided only alarm calls where- 20 ic tool for identifying ground squirrel species (Titov et 20 as the other 8 individuals provided only DNA samples). 21 al. 2005; Nikol’skii et al. 2007; Schneiderová & Policht 21 In addition, we sampled DNA from 8 museum speci- 22 2012a,b), subspecies (Matrosova et al. 2016) and inter- 22 mens: “brevicauda” (2), “intermedius” (4), “carruther- 23 species hybrids (Nikol’skii et al. 1984; Formozov & Ni- 23 si” (1) and “iliensis” (1) forms obtained from their type 24 kol’skii 1986; Nikol’ski & Starikov 1997; Titov et al. 24 locality or adjacent areas (Table 1, Fig. 1). We also in- 25 2005). 25 26 cluded in the analyses 2 sequences of the form “iliensis” 26 Within species, alarm calls of ground squirrels are 27 from Genbank (AF157856 and AF157857). Nucleotide 27 similar in the acoustics across ages and sexes (Matroso- 28 sequences of the cyt b gene of S. relictus (AF157876), S. 28 va et al. 2007, 2011; Swan & Hare 2008; Volodina et 29 pallidicauda (AF157866, AF157869) and S. erythroge- 29 al. 2010), including the red-cheeked ground squirrels S. 30 nys (AF157875) were used as outgroups (Table 1, Fig. 30 erythrogenys (Zhilin 2002). Alarm calls of red-cheeked 31 1). The latter sequence was obtained near the terra typi- 31 ground squirrels are either single notes (hereinafter “sin- 32 ca (Vasilyeva 1968; Thorington & Hoffmann 2005) and 32 gle-note alarm calls”) or represent clusters of several 33 is, therefore, considered a nominative subspecies S. e. 33 similar notes (“multi-note alarm calls”). In both the sin- 34 erythrogenys. 34 gle-note and multi-note alarm calls, the maximum fun- 35 35 damental frequency of the notes ranges from 2.91 to Alarm call, DNA and GPS data collection 36 36 5.52 kHz, whereas note duration ranges from 22 to 179 37 For acoustic recordings (48 kHz, 16 bit), we used a 37 ms (Nikol’skii 1979; Nikol’skii & Starikov 1997; Zhil- 38 solid state recorder Marantz PMD-660 (D&M Profes- 38 in 2002). Both the single-note and multi-note alarm call 39 sional, Kanagawa, Japan) with an AKG-C1000S car- 39 types have similar frequency modulation patterns of the 40 dioid electret condenser microphone (AKG-Acoustics 40 notes. 41 Gmbh, Vienna, Austria). Twenty-two individuals (10 in- 41 42 In this study, we used representative genetic and dividuals from the Tara population, 10 individuals from 42 43 acoustic samples to obtain new data that enables test- the Altyn-Emel population and 2 individuals from the 43 44 ing of a potentially separated taxonomical position of Balkhash population) were live-captured and recorded 44 45 ground squirrels from Southeast Kazakhstan (on the ter- for calls emitted toward a researcher when sitting singly 45 46 ritory of the Dzungarian Alatau mountains), taking into in wire-mesh live-traps (10 × 10 × 35 cm). The alarm 46 47 account the existing natural eco-geographical barri- calls were either produced spontaneously or in response 47 48 ers (Fig. 1). The aims of this study were to compare in- to additional stimulation (walking the researcher near 48 49 trapopulation and interpopulation variation of the alarm the traps or in response to movements of a hand-held 49 50 call structure and to provide, by using 2 mtDNA mark- hat). The distance from the microphone to a caller was 50 51 ers (control region and cytochrome b gene), insights approximately 1 m. After recording, the animals were 51

1 Table 1 Geographic origin of all Spermophilus specimens analyzed 1 2 2 Population/ Location Region Date of collection/Collector(s) 3 3 Museum specimen 4 4 Tara, 44.9166N Alma-Ata region, Koksus district, village Jarlyosek (former 10-24.06.2015/ 5 5 78.0833E village Oktyabr). 6 n = 10 Ivanova A.D., Alexandrov D.Yu., 6 Sibiryakova O.V. 7 Altyn-Emel, 44.1923N Alma-Ata region, Kerbulak district, National Park Altyn- 7 8 n = 10 78.5833E Emel, the pass «Stariy». 8 9 Balkhash, 46.2530N Alma-Ata region, Sarkand district, village Lepsy. 9 10 n = 18 78.9200E 10 11 126* (1) 46.1951N Alma-Ata region, Alakol district, town Usharal. 30.05.1953/ 11 12 80.9165E Strautman E.I. 12 13 13 142* (1) 46.4277N Alma-Ata region, Alakol district, lake Baybol. 20.06.1948/ 14 14 80.9363E Sludsky A.A. 15 15 16 59552* (2) 45.3807N Alma-Ata region, Sarkand district, aul (village) Ekiasha 07.06.2007/ 16 80.1397E (former village Pokatilovka). 17 Lopatina N.V. 17 18 59553* (2) 46.7275N Alma-Ata region, Sarkand district, lake Balkhash, estuary 08.06.2007/ 18 19 79.2150E of Ayaguz River. Lopatina N.V. 19 20 S-64538* (3) 45.5835N Dzungarian gates, Alma-Ata region, Alakol district, lake 28.05.1959/ 20 21 82.0498E Zhalanashkol. Kuzmina M.A. 21 22 S-151799* (3) 47.6121N Zaysan depression, East-Kazakhstan region, Zaysan 11.06.1954/ 22 23 23 85.1692E district, village Karatal. Ivanov O.A. 24 24 S-131114* (3) 47.9751N Zaysan depression, East-Kazakhstan region, Kurchum 06.05.1967/ 25 25 85.2406E district, aul (village) Boran (former Buran). Prokopov K.P. 26 26 27 S-143254* (3) 48.3547N Kazakh Upland, East-Kazakhstan region, Ayagoz district, 22.06.1987/ 27 80.4635E station Enkerey. 28 Shenbrot G.I. 28 29 Museum specimens are labeled with asterisks (*): (1) the Zoological Museum of Institute of Zoology of the Republic of Kazakh- 29 30 stan; (2) the Zoological Museum of Institute of Systematics and Ecology of Animals, Siberian Branch of Russian Academy of Sci- 30 31 ences; and (3) the Zoological Museum of Moscow University. n, sample size 31 32 32 33 33 34 34 35 35 36 sampled for the genomic samples (claws of fingers from as alarm calls of red-cheeked ground squirrels do not 36 37 a hind leg). The place of the cut was treated with anti- differ between ages and sexes (Zhilin 2002). 37 38 septic solution of hydrogen peroxide; the presence of the To reduce the likelihood of collecting individuals 38 39 cut was checked to prevent a repeated inclusion of ani- from the same family, we distributed the traps as far 39 40 mals in the analyses. After sampling, the animals were as possible based on the size of the colony. The larg- 40 41 released back at the place of capture. In addition, in the est distances between traps, up to several hundred me- 41 42 Balkhash population, unmatched call and genomic sam- ters, were in the Balkhash population, which covered 42 43 ples were collected, as 8 live-trapped individuals pro- the long coastline of the lake. The smallest distances be- 43 44 vided only the genomic samples and 8 free-ranging in- tween traps, up to a few dozens meters, were in the Al- 44 45 dividuals provided only the alarm calls. Inclusion in the tyn-Emel population, represented by a small isolat- 45 46 same analyses of calls from live-trapped and free-ranged valley. In mid-June, the body weight of the juveniles 46 47 ing animals should not affect the results because pattern was already close to those of adults and the natal disper- 47 48 of calling toward humans and structure of alarm calls sion had occurred, so we did not observe any mothers 48 49 are similar in ground squirrels in live-traps and in natu- associated with the young, which minimized the possi- 49 50 ral conditions (Nikol’skii 1979; Matrosova et al. 2010). bility of individuals belonging to the same families. The 50 51 Age and sex of the animals were not taken into account, trapping points from which the alarm calls and/or gene 51

1 samples were labeled by GPS coordinates using Garmin ed (HKY+G) (–lnL = 2592.53, BIC = 5831.58, AICc = 1 2 GPSmap 60CS (Garmin, Olathe, KS, USA) at a point 5389.03). The minimum evolution (ME) and the neigh- 2 3 close to the point of recording/sampling. Geographical bor-joining (NJ) methods were used to check the stabil- 3 4 distances between populations were calculated as lin- ity of the branch topology. Node support values in phy- 4 5 ear distances (in km) based on GPS coordinates using logenetic trees were estimated according to bootstrap 5 6 the Google Earth Internet resource (https://earth.google. support (1000 replicates). The S. relictus, S. pallidicau- 6 7 com/web). da and S. erythrogenys were selected as outgroups for 7 8 creating the phylogenetic tree. Haplotype diversity (h) 8 Molecular analyses 9 and nucleotide diversity (π) within each population were 9 10 We used a full-size mitochondrial control region and calculated in Arlequin v. 3.5 (Excoffier & Lischer 2010) 10 11 cytochrome b gene (cyt b) as genomic markers. The and DnaSP v. 5.10.01 (Librado et al. 2009). Haplotype 11 12 control region is highly variable, so it was previously networks were constructed using the median joining 12 13 used for both estimating the within-species geographical method in the PopART software (Leigh & Bryant 2015). 13 14 14 differences (Ochoa et al. 2012; Matrosova et al. 2016; Acoustic analyses 15 Ivanova et al. 2017) and for identification of species 15 16 (Gündüz et al. 2007). The сyt b is the most commonly All calls were analyzed spectrographically using 16 17 used marker for comparison of taxa, and is widely ap- Avisoft SASLab Pro software (Avisoft Bioacoustics, 17 18 plied in studies of different species of ground squirrels Berlin, Germany). Before analysis, the calls were high- 18 19 (e.g. Harrison et al. 2003; Herron et al. 2004; Helgen et pass filtered at 1 kHz to reduce the low-frequency 19 20 al. 2009). background noise, as preliminary visual analysis of the 20 21 Whole genomic DNA was extracted from tissue using spectrograms showed that call fundamental frequen- 21 22 the phenol–chloroform method (Sambrook et al. 1989). cy always exceeded 1 kHz. Spectrograms were created 22 23 The complete control region flanked by fragments of the with Hamming window, FFT 1024 points, frame 50% 23 24 tRNA–Pro and the tRNA–Phe genes was amplified us- and overlap 93.75%. For the acoustic analysis, we ran- 24 25 ing the primers MDL1 and H00651 following Ermakov domly selected up to 10 alarm calls of good quality 25 26 et al. (2002). Polymerase chain reaction (PCR) com- per individual, 10 individuals per population, with 282 26 27 prised 35 cycles of 1 min at 94 °C, 1 min at 62 °C and 1 alarm calls in total, because 6 individuals provided few- 27 28 min at 72 °C. er than 10 (from 5 to 8) measurable alarm calls. 28 29 29 The cyt b gene was amplified using 2 primer pairs: Because the alarm call of red-cheeked ground squir- 30 30 L14723 (L7)/GlCbR5a and 397-Sp (L397)/GlCben- rels may contain a different number of notes, we mea- 31 31 dR (Ducroz et al. 2001; Faerman et al. 2017). The PCR sured the acoustic parameters of each note sequential- 32 32 comprised 35 cycles of 1 min at 94 °C, 1 min at 56 °C, ly and uniformly. In total, we analyzed 344 notes (245 33 33 and 1 or 2 min at 72 °C for 2 pairs of primers, respec- notes from the 245 single-note alarm calls and 99 notes 34 34 tively. from the 37 multi-note alarm calls). Alarm calls of 35 12 individuals who provided both the single-note and 35 36 Sequencing was performed in an ABI 3730 automat- 36 ed genetic analyzer using a BigDye Terminator v3.1 Cy- multi-note alarm calls were used for comparison of the 37 acoustic structure between the notes of the single-note 37 38 cle Sequencing Kit (Applied Biosystems, Foster City, 38 CA, USA) using PCR primers. The nucleotide sequenc- alarm calls and the first notes of the multi-note alarm 39 calls. For this analysis, we averaged within individu- 39 40 es were aligned both with BioEdit (Hall 1999) soft- 40 ware and manually. We used the MEGA v. 7.0. software als all available alarm calls of each alarm call type (sin- 41 gle-note or multi-note). 41 42 (Kumar et al. 2016) for data processing. Nucleotide in- 42 In the alarm calls, the fundamental frequency band 43 dels were treated as missing data. DNA raw and net di- 43 had the highest energy relative to the harmonics (Fig. 44 vergences were estimated under the Kimura 2-parame- 44 2a). As alarm calls of red-cheeked ground squirrels rep- 45 ter (K2P) model using MEGA v. 7.0. SE were estimated 45 resent tonal notes deeply modulated in frequency (Zhil- 46 from 1000 bootstrap replicates. For constructing the 46 in 2002), the maximum fundamental frequency (f0 max) 47 phylogenetic tree, the maximum likelihood (ML) meth- 47 was clearly visible on the spectrogram (Fig. 2a). We 48 od was used. The most appropriate DNA substitution 48 measured the f0 max, the start fundamental frequency (f0 49 model for the dataset was established using jModelT- 49 st) and the end fundamental frequency (f0 end) of a note 50 est 2.1.10 (Posada 2008). The ML tree was created with 50 manually from the screen with the reticule cursor. We 51 the Hasegawa–Kishino–Yano model, gamma distribut- 51

1 measured duration of each note within alarm calls man- sented. All measurements were exported automatically 1 2 ually with the standard marker cursor. From the values to Microsoft Excel (Microsoft, Redmond, WA, USA). 2 3 of f0 st and f0 end, we automatically selected the mini- 3 Statistical analysis 4 mum fundamental frequency (f0 min). The depth of fre- 4 5 quency modulation (df) was calculated as the difference Statistical analyses were made with STATISTICA 5 6 between the f0 max and the f0 min. In addition, the be- v. 6.0 (StatSoft, USA); all means are given as mean ± 6 7 tween-note interval (Fig. 2a) was calculated where pre- SD. Significance levels were set at 0.05 and 2-tailed 7 8 probability values are reported. We used repeated mea- 8 9 sures analysis of variance (RM ANOVA) to compare 9 10 the acoustics between the averaged per individual notes 10 a 11 of the single-note alarm calls and the first notes of the 11 12 multi-note alarm calls. To estimate the acoustic differ- 12 13 ences between populations, we performed a standard 13 14 procedure of discriminant function analysis (DFA) us- 14 15 ing the 3 least correlated acoustic variables that mainly 15 16 contributed to discrimination (duration, f0 min, f0 max). 16 17 We used ANOVA with a Tukey honestly significant dif- 17 18 ference (HSD) test with the averaged per individual val- 18 19 ues of the acoustic parameters of the single-note alarm 19 20 b calls and/or the first note of the multi-note alarm calls to 20 21 assess whether the acoustic variables differed between 21 22 populations. 22 23 23 24 RESULTS 24 25 25 26 General acoustic structure of alarm calls 26 27 27 28 The subject red-cheeked ground squirrels mostly pro- 28 29 duced the single-note alarm calls (87% of the total num- 29 30 Time (s) ber). Much more rarely (13%) they produced the multi- 30 31 note alarm calls, consisting of 2–6 notes separated with 31 Figure 2 (a) Measured alarm call variables: f0 max – the 32 internote intervals ranging from 33 to 312 ms (on av- 32 maximum fundamental frequency; f0 min – the minimum 33 erage, 132 ± 74 ms). To account for the potential dif- 33 fundamental frequency; duration – duration of a note; interval 34 ferences between the first and other alarm calls with- 34 – between-note interval. (b) Examples of three single-note in clusters (Randall & Rogovin 2002), we compared 35 alarm calls of three red-cheeked ground squirrels from study 35 the acoustic structure between notes of the single-note 36 populations (Tara, Altyn-Emel and Balkhash). 36 37 37 38 38 39 39 40 Table 2 Values (mean ± SD) of variables of the notes differing in position within the alarm call (single, first, other position) and RM 40 41 ANOVA results for their comparison 41 42 42 43 Acoustic variable Single-note Multi-note alarm calls, ANOVA 43 44 alarm calls, n = 99 44 45 n = 245 First notes, n = 37 Other notes, n = 62 Single notes vs First notes First notes vs Other notes 45

46 duration (ms) 168 ± 25 153 ± 17 139 ± 17 F1,11 = 0.11, P = 0.75 F1,11 = 29.37, P < 0.001 46 47 47 f0 min (kHz) 4.48 ± 0.46 4.66 ± 0.47 4.26 ± 0.41 F1,11 = 0.52, P = 0.49 F1,11 = 8.15, P = 0.02 48 48 f0 max (kHz) 8.45 ± 0.72 8.56 ± 0.63 8.07 ± 0.68 F = 0.004, P = 0.95 F = 28.02, P < 0.001 49 1,11 1,11 49 50 df (kHz) 3.98 ± 0.60 3.90 ± 0.61 3.81 ± 0.52 F1,11 = 0.21, P = 0.66 F1,11 = 2.16, P = 0.17 50 51 n, number of averaged per individual notes (one note per alarm call). 51

1 alarm calls, the first notes of the multi-note alarm calls 1 2 and other notes of the multi-note alarm calls. In the 2 3 multi-note alarm calls, duration, f0 max and f0 min were 3 4 significantly higher in the first notes compared to other 4 5 notes (2nd, 3rd, etc.), whereas df did not differ signifi- 5 6 cantly between the first and other notes (RM ANOVA, 6 Table 2). At the same time, the first notes of the multi- 7 note alarm calls and the notes of the single-note alarm 7 8 calls did not differ in any acoustic variable (RM ANO- 8 9 VA, Table 2). Thus, for each individual, we used for fur- 9 10 ther analysis only the averaged per individual first notes 10 11 of the alarm calls regardless of the presence or absence 11 12 of other notes. 12 13 The duration of the first alarm call note (n = 282) 13 14 ranged from 125 to 237 ms (on average, 167 ± 25 ms); 14 15 f0 max ranged from 7.19 to 9.90 kHz (on average, 8.46 15 16 ± 0.70 kHz), f0 min ranged from 3.63 to 5.33 kHz (on 16 17 average, 4.49 ± 0.42 kHz) and df ranged from 3.01 to 17 18 5.05 kHz (on average, 3.96 ± 0.54 kHz). 18 19 Interpopulation acoustic differences 19 20 Alarm call variables were very similar between the 20 21 3 study populations (Fig. 2b). Minor, although signifi- 21 22 cant, differences were only revealed in duration (Table 22 23 3). DFA, based on the averaged per individual values 23 24 of 3 acoustic variables of the first notes of alarm calls, 24 25 showed 56.7% correct assignment of the alarm calls to 25 population. 26 26 27 Genetic comparison with closely related taxa 27 28 We found 15 unique cyt b haplotypes: 7 haplotypes 28 29 among the examined 8 museum specimens obtained near 29 30 the type localities of different closely related forms and Figure 3 Suggested phylogenetic relationships of red- 30 31 8 haplotypes among the 30 examined live-trapped indi- cheecked ground squirrels and the closely related taxa. (a) 31 32 viduals (Table S1). The cyt b sequence data displayed a ML phylogenetic tree (HKY+G model) based on analy- 32 ses of 15 unique cyt b haplotypes found among study speci- 33 clear phylogeographic structure (Fig. 3a). On the phylo- 33 genetic tree, ground squirrels from Southeast Kazakh- mens from Southeast Kazakhstan. Numbers near the branches 34 34 stan display 3 clades with very high (98–100%) boot- denote percentage bootstrap resampling support from 1000 35 35 strap support irrespective of the applied algorithm for replications (ML/ME/NJ). Bootstrap support is only shown 36 36 creating the tree (NJ, ML or ME). The first clade com- for the values exceeding 70%. n – number of sequences repre- 37 prises ground squirrels from the Balkhash, Tara and Al- senting each haplotype. (b) Suggested distribution areas for the 37 38 tyn-Emel (form “intermedius” MH518104-MH518107); given forms. Figures near a sample name designate clustering 38 39 among them, the Altyn-Emel ground squirrels are most of the sample with a particular taxon accordingly to the cyt b 39 40 separated. The second clade comprises specimens from polymorphism (triangles – “brevicauda”, squares – “iliensis”, 40 41 the Zaysan depression (form “brevicauda” MH518109- circles – “intermedius”). Stars designate type locality 41 MH518110) and from surroundings of the Dzungarian 42 (“brevicauda”, “carruthersi”, “iliensis”, “intermedius”). For 42 Gate (form “carruthersi” MH518108). The third clade 43 details, see Table S1. 43 comprises specimens from the left bank of the Ily River 44 (form “iliensis” MH518111) and from the southeast part 44 45 of the Kazakh Upland (Fig 3b; see also Fig 1). 45 46 Raw genetic distances between the 3 clades inhab- 46 47 47 iting Southeast Kazakhstan (“brevicauda,” “iliensis” ic distances between clades “brevicauda,” “iliensis” and 48 48 and “intermedius”) varied from 2.9% to 3.9% (Table 4). “intermedius” varied from 2.3% to 2.9% (Table 4) and 49 49 The raw distances between these clades and S. eryth- from 6.1% to 6.5% between these clades and S. eryth- 50 50 rogenys varied from 6.2% to 7.0%. Similarly, net genet- rogenys. 51 51

1 Table 3 Values (mean ± SD) of the alarm call first notes and ANOVA results for their comparison between 3 populations 1 2 2 Acoustic variable Population ANOVA results for 3 3 4 Tara, n = 10 Altyn-Emel, n = 10 Balkhash, n = 10 3 populations 4 5 a a b 5 duration (ms) 158 ± 15 158 ± 16 185 ± 32 F2,27 = 4.57, P < 0.05 6 6 f0 min (kHz) 4.48 ± 0.42 4.39 ± 0.35 4.62 ± 0.50 F = 0.73, P = 0.49 7 2,27 7 8 f0 max (kHz) 8.36 ± 0.80 8.56 ± 0.67 8.45 ± 0.67 F2,27 = 0.21, P = 0.81 8

9 df (kHz) 3.87 ± 0.58 4.18 ± 0.60 3.84 ± 0.39 F2,27 = 1.21, P = 0.31 9 10 10 The same superscripts indicate which populations did not differ significantly (P > 0.05, Tukey HSD post-hoc test). n, number of av- 11 eraged alarm calls (1 per individual). 11 12 12 13 13 14 14 15 Interpopulation genetic differences full-size mtDNA control region (1005-1006 bp, Gen- 15 16 Overall genetic diversity Bank Acc. No.: MH518112-MH518141) and full cyt b 16 17 (1140 bp, MH518074-MC518103) in 30 red-cheeked 17 18 We analyzed in detail the nucleotide sequences of the ground squirrels from Tara, Altyn-Emel and Balkhash 18 19 populations (10 individuals per population). For the 19 20 control region, 33 sites (approximately 3% of the full 20 21 fragment length) were variable, of which 21 were parsi- 21 22 Table 4 Raw (below the diagonal) and net (above the diago- mony informative. Similarly, for the cyt b, 29 sites (ap- 22 23 nal) K2P-distances between the 3 main genetic (cyt b) linages proximately 2.5% of the full fragment length) were vari- 23 24 of ground squirrels inhabiting the Southeast Kazakhstan, with able, of which 21 were parsimony informative. 24 25 associated standard errors (SE) based on 1000 bootstrap repli- We summarize the genetic characteristics in Table 5. 25 26 cates in parentheses Both mtDNA markers yielded similar results. Nucleo- 26 27 intermedius brevicauda iliensis tide diversity (π) was very low; in the pooled sample set 27 28 N = 34 N = 3 N = 3 of the 30 individuals it was 0.0097 ± 0.0051 for the con- 28 29 intermedius — 0.023(0.004) 0.029(0.006) trol region and 0.0088 ± 0.0046 for cyt b. Haplotype di- 29 30 30 brevicauda 0.029(0.005) — 0.025(0.005) versity (h) was substantially higher; in the pooled sam- 31 ple set of the 30 individuals it was 0.8575 ± 0.0423 for 31 iliensis 0.039(0.006) 0.032(0.005) — 32 control region and 0.8184 ± 0.0457 for cyt b. 32 33 N, number of sequences. 33 34 34 35 35 36 36 Table 5 Genetic characteristics of 3 study populations of red-cheeked ground squirrels 37 37 38 Population C-region Cytochrome b 38

39 Nhapl π (SD) h (SD) Nhapl π (SD) h (SD) 39 40 Tara, 3† 0.0046 0.6444 3† 0.0012 (0.0009) 0.6444 (0.1012) 40 41 n = 10 (0.0028) (0.1012) 41 42 Altyn-Emel 3 0.0006 0.3778 (0.1813) 2 0.0007 (0.0006) 0.2000 (0.1541) 42 43 n = 10 (0.0006) 43 44 Balkhash 8† 0.0087 0.9333 (0.0773) 6† 0.0069 (0.0040) 0.8444 (0.1029) 44 45 n = 10 (0.0049) 45 46 Total 13 0.0097 0.8575 (0.0423) 10 0.0088 (0.0046) 0.8184 (0.0457) 46 47 n = 30 (0.0051) 47 48 Mean nucleotide 33.5% (T), 30.5% (A), 34.1% (T), 28.3% (A), 48 49 composition 23.7% (C), 12.3% (G) 24.8% (C), 12.7% (G) 49 50 †One haplotype was shared between Tara and Balkhash populations. Designations: h, haplotype diversity; n, the number of assayed 50

51 animals in population; Nhapl, number of found haplotypes; π, nucleotide diversity (averaged over loci); SD, standard deviation. 51

1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10 10 11 11 12 12 13 13 14 14 15 15 16 16 17 17 18 Figure 4 Unrooted haplotype network of absolute distances between mitochondrial DNA haplotypes of (a) control region and (b) 18 19 cytochrome b gene. Each circle represents a unique haplotype; its size proportional to the haplotype frequency. Small black circles 19 20 represent hypothetical haplotypes. Colors represent the three study populations. The hatches on the lines connecting haplotypes 20 21 represent nucleotide substitutions. For details, see Table S1. 21 22 22 23 23 24 24 25 Interpopulation genetic diversity 0.014 ± 0.003 (Altyn-Emel–Balkash) and 0.005 ± 0.001 25 26 (Tara–Balkhash) for the cyt b. 26 Among the 30 individuals examined, 13 control re- 27 27 gion haplotypes and 10 cyt b haplotypes were identi- 28 28 fied (Table 4 and Table S1). Both markers demonstrat- DISCUSSION 29 29 ed 1 shared haplotype between individuals from Tara 30 30 and Balkhash populations. Most unique haplotypes Alarm call differences 31 31 were found in the Balkhash population. In contrast, in- 32 Based on the acoustics of alarm calls, 3 study pop- 32 dividuals from the Altyn-Emel population were practi- 33 ulations were very similar, in agreement with findings 33 cally identical genetically; only 2 individuals had single 34 for Spermophilus suslicus (Matrosova et al. 2016), indi- 34 substitutions in their haplotypes. Both the control re- 35 cating that the bioacoustical changes of alarm calls are 35 gion and the cyt b sequence data showed a weak phylo- 36 more conservative compared to the genetic changes in 36 geographic structure, without any subdivision into phy- 37 speciation of ground squirrels. At the same time, our re- 37 logroups (Fig. 4). Clusters of close haplotypes do not 38 sults confirmed the early evidence of noticeable differ- 38 correspond to certain geographical areas, with the ex- 39 ences in the acoustics of the alarm calls between the 39 ception of the Altyn-Emel population, which is the most 40 red-cheeked ground squirrels of Southeast Kazakhstan 40 remote and isolated within the Dzungarian Alatau south- 41 and the red-cheeked ground squirrels of the Southern 41 west ridges. Within-group pairwise distances (K2P) 42 Trans-Ural (Kurgan region) of Russia (Nikol’skii 1979; 42 were 0.0046 ± 0.0016 (Tara), 0.0006 ± 0.0004 (Al- 43 Nikol’skii & Starikov 1997; Zhilin 2002; Nikol’skii & 43 tyn-Emel) and 0.008 ± 0.0020 (Balkhash) for the con- 44 Rumyantsev 2004). 44 trol region and 0.0012 ± 0.0007 (Tara), 0.0007 ± 0.0004 45 In the 3 studied “intermedius” populations from 45 (Altyn-Emel) and 0.007 ± 0.0016 (Balkhash) for the cyt 46 Southeast Kazakhstan, the alarm call note duration was 46 b. 47 167 ± 25 ms compared to much shorter (119.8 ± 73.98 47 48 Between-group distances (K2P) comprised 0.014 ± ms) duration reported for the “iliensis” red-cheeked 48 49 0.003 (Altyn-Emel–Tara), 0.012 ± 0.003 (Altyn-Emel– ground squirrels from Russia (Zhilin 2002). In addi- 49 50 Balkash) and 0.008 ± 0.002 (Tara–Balkhash) for the tion, an average maximum fundamental frequency was 50 51 control region and 0.016 ± 0.003 (Altyn-Emel–Tara), substantially higher in “intermedius” populations from 51

1 Southeast Kazakhstan (8.46 ± 0.75 kHz) than in “ilien- wise comparisons (Kapustina et al. 2015). 1 2 sis” red-cheeked ground squirrels from Russia (5.62 ± Net genetic distances between the S. erythrogenys 2 3 0.06 kHz) (Zhilin 2002). Further differences were ob- from one side and the “intermedius,” “brevicauda” or 3 4 served in ratios of single-note/multi-note alarm calls: “iliensis” lineages from another side were highly sig- 4 5 whereas the ground squirrels from Southeast Kazakh- nificant and amounted of 6.1–6.5%, which indicates 5 6 stan mostly produced the single-note alarm calls (this paraphilia of these taxa. Recently described as separat- 6 7 study), the ground squirrels from Russia mostly pro- ed species, the Taurus ground squirrel S. taurensis dif- 7 8 duced the multi-note alarm calls (Zhilin 2002; Nikol’skii fers from the most closely related S. citellus by cyt b 8 9 & Rumyantsev 2004). The degree of acoustic differ- genetic distance of 5.0%, whereas the difference be- 9 ences between the “intermedius” and “iliensis” ground 10 tween S. taurensis and S. xanthoprymnus comprised 10 squirrels was comparable with those between the Tau- 11 9.6%, and the difference between S. citellus and S. 11 rus (Spermophilus taurensis), Anatolian (Spermophilus 12 xanthoprymnus comprised 9.8% (Gündüz et al. 2007). 12 xanthoprymnus) and European ground squirrels (Sper- Our analyses of specimens from Southeast Kazakh- 13 mophilus citellus) (Gündüz et al. 2007). The Taurus and 13 stan suggest the presence of 3 major mtDNA lineages, 14 Anatolian ground squirrels were recently separated to 14 15 different species based on their alarm call traits (Schnei- displaying mtDNA distances of potential species-rank 15 16 derová & Policht 2011), as they differed from each other levels, “intermedius,” “brevicauda” and “iliensis,” 16 17 by the alarm call structure as strongly as from the close- and point to the potential role of geographical barriers 17 18 ly related species, the European ground squirrel (Schnei- in a substantial genetic differentiation of red-cheeked 18 19 derová & Policht 2012). Our results provide similar ar- ground squirrels in Southeast Kazakhstan. 19 20 guments in favor of species autonomy of potential S. e. We conclude that the Spermophilus erythrogenys sen- 20 21 intermedius from Southeast Kazakhstan and potential S. su lato complex needs further taxonomical revision. The 21 22 e. iliensis from Russia. Information about the acoustics 3 mtDNA lineages revealed in this study point, respec- 22 23 of the third line of red-cheeked ground squirrels (“brevi- tively, to the 3 putative species, which should be vali- 23 24 cauda”) is scarce; the frequency pattern of the alarm call dated with inclusion of data on the alarm call acoustics 24 25 is probably different (Nikol’skii & Rumyantsev 2004). from the “brevicauda” and “iliensis” lineages, with anal- 25 26 Further research is necessary to investigate the acoustic yses of nuclear markers of specimens representative for 26 27 differences between the potential species S. e. interme- all the 3 mtDNA lineages, “intermedius,” “brevicauda” 27 28 dius and S. brevicauda etc. and “iliensis,” and with a rigorous morphological ex- 28 29 Mitochondrial DNA differences amination. The inclusion of nuclear DNA markers, and 29 30 acoustic and morphological data for validating potential 30 31 The within-species level of control region variation, species of ground squirrels is necessary because the re- 31 32 studied in the 30 living specimens in this study, showed cently speciated Eurasian Spermophilus species may be 32 33 that 3% positions were variable. This degree of vari- genetically distinct but not morphologically well-differ- 33 ation was comparable with that reported for the Per- entiated, as was revealed in the case of the geographi- 34 34 ote ground squirrel Xerospermophilus perotensis (2%, cally widespread Anatolian ground squirrel, S. xantho- 35 35 Ochoa et al. 2012) but was substantially lower than that prymnus (Gündüz et al. 2007). 36 36 reported for the S. suslicus (8%, Matrosova et al. 2016), 37 “Intermedius” lineage 37 which diverged to 2 different geographically separated 38 38 chromosome races with p-distance of 4.4–4.6% between The 3 study populations (Tara, Altyn-Emel and 39 them (Brandler et al. 2015; Matrosova et al. 2016). The Balkhash), comprising a common mtDNA clade, are all 39 40 within-species level of cyt b variation of 2.5% of vari- isolated by the Balkhash-Alakol depression from the 40 41 able positions was substantially lower than in the Eu- north and east, by the Ili River and by the Dzungarian 41 42 ropean ground squirrel S. citellus (6%, Kryštufek et al. Alatau Mountains from the west and south (Fig. 1). Fol- 42 43 2009) but higher than in the Alashan ground squirrel S. lowing Belyaev (1955), this clade can be considered as 43 44 alashanicus (0.11%, Kapustina et al. 2015). a potential new species S. intermedius Brandt, 1843, de- 44 45 scribed earlier from this territory (southeast bank of the 45 Interspecies genetic distances based on cyt b sequenc- Balkhash Lake, village Lepsy) (Belyaev 1955). 46 es for “intermedius,” “brevicauda” and “iliensis” lineag- 46 47 es were 2.7–3.6% between these 3 forms. This degree of “Brevicauda” lineage 47 48 48 variation is comparable with the distances between such Ground squirrels inhabiting territories from the Zay- 49 49 well-separated species as Spermophilus fulvus, Spermo- san Lake and the Mongolian Altai Mountains to the 50 50 philus rallii and S. pallidicauda, of 2.9–3.2% in all pair- Ebinur Lake and the Borohoro Mountains (from the 51 51

1 north to the south) (Fig. 1) are isolated from the “inter- treatment of animals in behavioral research and teaching 1 2 medius” lineage by the Balkhash-Alakol depression and (Anim Behav, 2006, 71: 245–253). Animal disturbance 2 3 probably also by the Tarbagatai, Saur, Urkashar and Bir- and invasive procedures were kept at a minimum, in ac- 3 4 liktau Mountains. Grounds squirrels of this territory can cordance with recommendation of the Advisory Board 4 5 be considered as a potential new species S. brevicauda on Bioethics of the Engelhardt Institute of Molecular Bi- 5 6 Brandt, 1843 by a principle of priority, so S. carruther- ology. The conservation status of the ground squirrels in 6 7 si (Thomas, 1912) would be a synonym (Smith et al. the South Kazakhstan is “Least Concern,” so there is no 7 8 2008). need for special conservation measures across their dis- 8 9 “Iliensis” lineage tribution area. This study was supported by the Russian 9 Foundation for Basic Research (grant 18-04-00400). 10 Ground squirrels inhabiting a territory to the west 10 11 from the Ily River barrier (including the right bank of 11 12 the Ily River, the Chu-Ily Mountains, the western and REFERENCES 12 13 northern bank of the Balkhash Lake (Fig. 1), probably Afanas’ev AV, Bazhanov VS, Korelov MN, Sludskij 13 14 belong to the same “iliensis” mtDNA lineage and can be AA, Strautman EI (1953). Mammals of Kazakhstan. 14 15 considered as a potential new species S. iliensis (Belyaev, Academy of Sciences of the Kazakh SSR, Alma-Ata. 15 16 1945), described from the right bank of the Ily River or S. 16 Belyaev AM (1955). Ground squirrels of Kazakhstan. selevini (Argyropulo, 1941) described from the South- 17 Proceedings of the Republic Plant Protection Station 17 east Kazakh Uplands. 18 2, 1–102. 18 19 For further research, we suggest new molecular anal- 19 Brandler OV, Biryuk IYu, Ermakov OA et al. (2015). yses of the Spermophilus erythrogenys sensu lato com- 20 Interspecies and intraspecific molecular genetic vari- 20 plex with more specimens included in analyses to iden- 21 ability and differentiation in speckled ground squir- 21 tify potential “cryptic” species. For example, the studies 22 rels Spermophilus suslicus and S. odessanus (Ro- 22 by Harrison et al. (2003) and Herron et al. (2004) pre- 23 dentia, Sciuridae, Marmotini). The Journal of V.N. 23 sented genealogies on the basis of the cyt b polymor- 24 Karazin Kharkiv National University. Series: Biology 24 phism that demonstrated that Eurasian Spermophilus 25 24, 58–67. 25 26 form a monophyletic group within Marmotini. Howev- 26 er, they studied too few specimens per species to confi- Brandt JF (1843). Observations sur les différentes es- 27 27 dently validate new cryptic species. pèces de sousliks de Russie, suivies de remarques 28 sur l’arrangement et la distribution géographique du 28 29 genre Spermophilus, ansé que sur la classification de 29 30 ACKNOWLEDGMENTS la familie des Ecureuils (Sciurina) en général. Bulle- 30 31 We are grateful to the staff of the DNA sequencing tin Scientifique l’Académie Impériale des Sciences de 31 32 center “Genome” of the Engelhardt Institute of Molec- Saint-Pétersbourg 2, 357–82. 32 33 ular Biology, Russian Academy of Sciences and to the Ducroz JF, Volobouev V, Granjon L (2001). An assess- 33 34 Laboratory of genetic engineering of the National medi- ment of the systematics of Arvicanthine rodents us- 34 35 cal research center for Hematology, Ministry of Health- ing mitochondrial DNA sequences: Evolutionary and 35 36 care of the Russian Federation for their help in conduct- biogeographical implications. Journal of Mammalian 36 37 ing the genetic analyses. We are grateful to the staff of Evolution 8, 173–206. 37 38 the Altyn-Emel National Park for their assistance in the Ermakov OA, Surin VL, Titov SV, Tagiev AF, Luk’ya- 38 39 fieldwork. We are grateful to the staff of the Zoologi- nenko AV, Formozov NA (2002). A molecular genet- 39 40 cal Museum of Institute of Zoology of the Republic of ic study of hybridization in four species of ground 40 41 Kazakhstan and to the staff Zoological Museum of In- squirrels (Spermophilus: Rodentia, Sciuridae). Rus- 41 42 stitute of Systematics and Ecology of Animals, Siberi- sian Journal of Genetics 38, 796–809. 42 43 an Branch of Russian Academy of Sciences and the Zo- Ermakov OA, Simonov E, Surin VL et al. (2015). Impli- 43 44 ological Museum of Moscow University for providing cations of hybridization. NUMTs, and overlooked di- 44 45 the genetic samples. We thank Vladimir Lebedev, Ev- versity for DNA barcoding of Eurasian ground squir- 45 geniy Simonov, Vadim Surin, Eugenia Solovjeva and 46 rels. PLoS ONE 10, e0117201. 46 Semen Bodrov for their methodical advices. We are sin- 47 Excoffier L, Lischer HEL (2010). Arlequin suite ver 3.5: 47 cerely grateful to the 3 anonymous reviewers whose A new series of programs to perform population ge- 48 comments were very helpful and inspiring. This study 48 netics analyses under Linux and Windows. Molecu- 49 has been conducted in accordance with ethical and an- 49 lar Ecology Resources 10, 564–67. 50 imal welfare standards and with the guidelines for the 50 51 Faerman M, Bar-Gal GK, Boaretto E et al. (2017). DNA 51

1 analysis of a 30,000-year-old Urocitellus glacial- Kryštufek B, Bryja J, Buzan EV (2009). Mitochondri- 1 2 is from northeastern Siberia reveals phylogenetic re- al phylogeography of the European ground squirrel, 2 3 lationships between ancient and present-day arctic Spermophilus citellus, yields evidence on refugia for 3 4 ground squirrels. Scientific Reports 7, 42639. steppic taxa in the southern Balkans. Heredity 103, 4 5 Formozov NA, Nikol’skii AA (1986). The sound sig- 129–35. 5 6 nal of the hybrid between the major and the pyg- Kryštufek B, Vohralík V (2013). Taxonomic revision of 6 7 my ground squirrels (Citellus major x C. pygmaeus). the Palaearctic rodents (Rodentia). Sciuridae: Xerinae 7 8 Bulletin of Moscow University, ser. 16. Biology 4, 3–8. 1 (Eutamias and Spermophilus). Lynx 43, 17–111. 8 9 Gromov IG, Bibikov DI, Kalabuchov NI, Meier MN Kumar S, Stecher G, Tamura K (2016). MEGA7: Mo- 9 10 (1965). Ground Squirrels (Marmotinae). Fauna lecular Evolutionary Genetics Analysis version 7.0. 10 11 USSR, Mammals, Vol. 3, Issue. 2. Nauka, Moscow. Molecular Biology and Evolution 33, 1870–4. 11 12 Gündüz I, Jaarola M, Tez C, Yeniyurt C, Polly PD, Sear- Leigh JW, Bryant D (2015). PopART: Full-feature soft- 12 13 le JB (2007). Multigenic and morphometric differ- ware for haplotype network construction. Methods of 13 14 entiation of ground squirrels (Spermophilus, Sciuri- Ecology and Evolution 6, 1110–6. 14 15 da, Rodentia) in Turkey, with a description of a new Librado P, Rozas J (2009). DnaSP v5: A software for 15 16 species. Molecular Phylogenetics and Evolution 43, comprehensive analysis of DNA polymorphism data. 16 17 916–35. Bioinformatics 25, 1451–2. 17 18 Hall TA (1999). BioEdit: A user friendly biological se- Matrosova VA, Volodin IA, Volodina EV, Babitsky AF 18 19 quence alignment editor and analysis program for (2007). Pups crying bass: Vocal adaptation for avoid- 19 20 Windows 95/98/NT. Nucleic Acids Symposium Series ance of age-dependent predation risk in ground squir- 20 21 41, 95–8. rels? Behavioral Ecology and Sociobiology 62, 181– 21 22 Harrison RG, Bogdanowicz SM, Hoffmann RS, Yensen 91. 22 23 E, Sherman PW (2003). Phylogeny end evolutionary Matrosova VA, Volodin IA, Volodina EV, Vasilieva 23 24 history of Ground Squirrels (Rodentia: Marmotinae). NA (2010). Stability of acoustic individuality in the 24 25 Journal of Mammalian Evolution 10, 249–76. alarm calls of wild yellow ground squirrels Spermo- 25 26 Helgen KM, Cole FR, Helgen LE, Wilson DE (2009). philus fulvus and contrasting calls from trapped and 26 27 Generic revision in the holarctic ground squirrel ge- free-ranging callers. Naturwissenschaften 97, 707– 27 28 nus Spermophilus. Journal of Mammalogy 90, 270– 15. 28 29 305. Matrosova VA, Blumstein DT, Volodin IA, Volodina EV 29 30 Herron MD, Castoe TA, Parkinson CL (2004). Sciu- (2011). The potential to encode sex, age and individ- 30 31 rid phylogeny and the paraphyly of Holarctic ground ual identity in the alarm calls of three species of Mar- 31 32 squirrels (Spermophilus). Molecular Phylogenetics motinae. Naturwissenschaften 98, 181–92. 32 33 and Evolution 31, 1015–30. Matrosova1 VA, Schneiderová I, Volodin IA, Volodi- 33 34 Hoffmann RS, Smith AT (2008). Subfamily Sciurinae. na EV (2012). Species-specific and shared features in 34 35 In: Smith AT, Xie Y, eds. A Guide to the Mammals vocal repertoires of three Eurasian ground squirrels 35 36 of China. Princeton University Press, Princeton, pp. (genus Spermophilus). Acta Theriologica 57, 65–78. 36 37 174–82. Matrosova VA, Rusin MY, Volodina EV et al. (2016). 37 38 Ivanova AD, Surin VL, Ermakov OA, Formozov NA Genetic and alarm call diversity across scattered pop- 38 39 (2017). The hybridization zone of red-cheeked (Sper- ulations of speckled ground squirrels (Spermophilus 39 40 mophilus erythrogenys) and little (Spermophilus pyg- suslicus). Mammalian Biology 81, 255–65. 40 41 maeus) ground squirrels (Rodentia, Sciuridae). Rus- Nikol’skii AA (1979). Species specificity of alarm call 41 42 sian Journal of Zoology 96, 449–57. in sousliks (Citellus, Sciuridae) of Eurasia. Zoolog- 42 43 Kapustina SYu, Brandler OV, Adiya Y (2015). Phylog- icheskii Zhurnal 58, 1183–94. 43 44 eny of genus Spermophilus and position of Alashan Nikol’skii AA, Rumyantsev V.Yu. (2004). Variabili- 44 45 ground squirrel (Spermophilus alashanicus, Buchner, ty of alarm call in ground squirrels of major group 45 46 1888) on phylogenetic tree of Paleartic short-tailed (Rodеntia, Sciuridae, Spermоphilus) as a model of 46 47 ground squirrels. Molecular Biology 49, 391–6. geographic species formation. Zoologicheskii Zhur- 47 48 Kikusui T, Nakanishi K, Nakagawa R, Nagasawa M, nal 83, 1008–17. 48 49 Mogi K, Okanoya K (2011). Cross fostering experi- Nikol’skii AA, Starikov VP (1997). The variabili- 49 50 ments suggest that mice songs are innate. PLoS ONE ty of the audible signal of a warning russet ground 50 6, e17721. 51 squirrel (Spermophilus major) and red-cheeked (S. 51

1 еrуthrogеnуs) ground squirrels in the contact zone do not affect responses to Richardson’s ground squir- 1 2 in the territory of the Kurgan region, Zoologicheskii rel alarm calls. Journal of Mammalogy 89, 889–94. 2 3 Zhurnal 76, 845–57. Thomas O (1912). On mammals from Central Asia, col- 3 4 Nikol’skii AA, Denisov VP, Stoiko TG, Formosov NA lected by Mr. Douglas Carruthers. Annals and Maga- 4 5 (1984). The alarm call in F1 hybrids Citellus pyg- zine of Natural History, Series 8 9, 391–408. 5 6 maeus X C. suslicus (Sciuridae, Rodentia). Zoolog- Thorington RW, Hoffmann RS (2005). Family Sciuri- 6 7 icheskii Zhurnal 63, 1216–25. dae. In: Wilson DE, Reeder DM, eds. Mammal Spe- 7 8 Nikol’skii AA, Ermakov OA, Titov SV (2007). Geo- cies of the World: A Taxonomic and Geographic Re- 8 9 graphical variability of the little ground squirrel ference, 3rd edn. Johns Hopkins University Press, 9 10 (Spermophilus pygmaeus): A bioacoustical analysis. Baltimore, pp. 754–818. 10 11 Zoologicheskii Zhurnal 86, 1379–88. Titov SV, Ermakov OA, Surin VL et al. (2005). Molec- 11 12 Ochoa A, Gasca J, Ceballos GJ, Eguiarte LE (2012). ular genetic and bioacoustic diagnostics russet (Sper- 12 13 Spatiotemporal population genetics of the endan- mophilus major Pallas, 1778) and yellow (S. fulvus 13 14 gered Perote ground squirrel (Xerospermophilus per- Lichtenstein, 1823) ground squirrels from mixed col- 14 15 otensis) in a fragmented landscape. Journal of Mam- ony. Bulletin of Moscow Society of Naturalists 110, 15 16 malogy 93, 1061–74. 72–7. 16 17 Ognev SI (1947). Mammals of the U.S.S.R. and adja- Vasilyeva MV (1968). Taxonomical interactions of two 17 18 cent countries. Vol. 5 Rodents. Izdatelstvo Akademii closely related species of ground squirrels, Citellus 18 19 Nauk SSSR, Moscow. major Pall., 1778 and Citellus erythrogenys Br., 1843 19 20 Posada D (2008). jModelTest: Phylogenetic Model Av- (Mammalia, Glires). Studies on the fauna of the So- 20 21 eraging. Molecular Biology and Evolution 25, 1253– viet Union (Mammals). Proceedings of the Zoologi- 21 22 6. cal Museum of Moscow State University 10, 94–108. 22 23 Randall JA, Rogovin KA (2002). Variation in and mean- Volodina EV, Matrosova VA, Volodin IA (2010). An un- 23 24 ing of alarm calls in a social desert rodent Rhombo- usual effect of maturation on the alarm call funda- 24 25 mys opimus. Ethology 108, 513–27. mental frequency in two species of ground squirrels. 25 26 Sambrook J, Fritch E, Maniatis T (1989). Molecular Bioacoustics 20, 87–98. 26 27 Cloning: A Laboratory Manual. Cold Spring Harbor Zachos FE (2018). Mammals and meaningful taxonom- 27 28 Laboratory Press, New York. ic units: the debate about species concepts and con- 28 29 Schneiderová I, Policht R (2012a). Acoustic analysis of servation. Mammal Review 48, 153–9. 29 30 the alarm call of the Anatolian ground squirrel Sper- Zhilin ME (2002). Ground squirrels of Southern Transu- 30 31 mophilus xanthoprymnus: A description and compar- ralia (distribution, vocalization, interpopulation pecu- 31 32 ison with alarm calls of the Taurus S. taurensis and liarities (PhD thesis). Kurgan State University, Kur- 32 33 European S. citellus ground squirrels. Naturwissen- gan. 33 34 schaften 99, 55–64. 34 35 Schneiderová I, Policht R (2012b). Acoustic analysis of SUPPLEMENTARY MATERIALS 35 36 alarm calls of the European ground squirrel (Sper- 36 37 mophilus citellus) and the Taurus ground squirrel (S. Additional supporting information may be found in 37 38 taurensis) (Mammalia: Sciuridae). Zoologischer An- the online version of this article at the publisher’s web- 38 39 zeiger 251, 139–46. site. 39 40 Sludsky AA, Varshavsky SN, Ismagilov MI, Kapitonov Table S1 List of samples. In the names of the sam- 40 41 VI, Shubin IG (1969). Mammals of Kazakhstan. Ro- ples, “A” stands for the origin of the Altyn-Emel pop- 41 42 dents (Marmots and Ground squirrels). Nauka of the ulation, “T” for the Tara population and “B” for the 42 43 Kazakh SSR, Alma-Ata. Balkhash population 43 44 Swan DC, Hare JF (2008). Signaler and receiver ages 44 45 45 46 46 47 Cite this article as: 47 48 48 Matrosova VA, Ivanova AD, Volodina EV et al. (2019). Phylogenetic relationship and variation of alarm call traits 49 49 of populations of red-cheeked ground squirrels (Spermophilus erythrogenys sensu lato) suggest taxonomic de- 50 50 lineation. Integrative Zoology 14, 341–53. 51 51