A Blind Climber: the First Evidence of Ultrasonic Echolocation in Arboreal

Integrative Zoology 2017; 12: 172–184 doi: 10.1111/1749-4877.12249

1 ORIGINAL ARTICLE 1 2 2 3 3 4 4 5 5 6 6 7 A blind climber: The first evidence of ultrasonic echolocation in 7 8 8 9 arboreal mammals 9 10 10 11 11 12 12 13 Aleksandra A. PANYUTINA,1,2 Alexander N. KUZNETSOV,2 Ilya A. VOLODIN,2,3 Alexei V. 13 14 14 4,5 2 15 ABRAMOV and Irina B. SOLDATOVA 15 16 1Severtsov Institute of Ecology and Evolution, Russian Academy of Sciences, Moscow, Russia, 2Department of Vertebrate Zoology, 16 17 17 Faculty of Biology, Lomonosov Moscow State University, Moscow, Russia, 3Scientific Research Department, Moscow Zoo, 18 18 Moscow, Russia, 4Zoological Institute, Russian Academy of Sciences, Saint Petersburg, Russia and 5Joint Vietnam–Russian Tropical 19 19 20 Research and Technological Centre, Hanoi, Vietnam 20 21 21 22 22 23 23 24 24 25 Abstract 25 26 The means of orientation is studied in the Vietnamese pygmy dormouse Typhlomys chapensis, a poorly known 26 27 enigmatic semi-fossorial semi-arboreal rodent. Data on eye structure are presented, which prove that Typhlo- 27 28 mys (translated as “the blind mouse”) is incapable of object vision: the retina is folded and retains no more than 28 29 2500 ganglion cells in the focal plane, and the optic nerve is subject to gliosis. Hence, Typhlomys has no oth- 29 30 er means for rapid long-range orientation among tree branches other than echolocation. Ultrasonic vocalization 30 31 recordings at the frequency range of 50–100 kHz support this hypothesis. The vocalizations are represented by 31 32 bouts of up to 7 more or less evenly-spaced and uniform frequency-modulated sweep-like pulses in rapid suc- 32 33 cession. Structurally, these sweeps are similar to frequency-modulated ultrasonic echolocation calls of some bat 33 34 species, but they are too faint to be revealed with a common bat detector. When recording video simultaneous- 34 35 ly with the ultrasonic audio, a significantly greater pulse rate during locomotion compared to that of resting an- 35 36 imals has been demonstrated. Our findings of locomotion-associated ultrasonic vocalization in a fast-climbing 36 37 but weakly-sighted small mammal ecotype add support to the “echolocation-first theory” of pre-flight origin of 37 38 echolocation in bats. 38 39 39 40 Key words: arboreal locomotion, reduced eyes, Rodentia, Typhlomys, ultrasonic echolocation 40 41 41 42 42 43 43 44 INTRODUCTION 44 45 45 46 Solid surfaces reflect both light and sound, which 46 47 gives animals an opportunity to perceive out-of-touch 47 48 Correspondence: Aleksandra A. Panyutina, Severtsov Institute surroundings using vision or echolocation. Use of vi- 48 49 of Ecology and Evolution, Russian Academy of Sciences, sion is much more common because life on Earth is full 49 50 Moscow, 119071, Russia. of sunlight. Only under poor light conditions are ani- 50 51 Email: [email protected] mals forced to expend their own metabolic energy to 51

1 emit sounds and listen to echoes instead of using vision. “flight-first theory,” echolocation arose in bats after -ac 1 2 To approach acuity of vision echolocation calls must be quisition of flight ability (Speakman 2001, 2008; Sim- 2 3 composed of as short as possible wavelengths to come mons et al. 2010). The “echolocation-first theory” in- 3 4 into ultrasonic range. Among mammals, ultrasonic echo- sists that echolocation was acquired earlier than flight 4 5 location is used in water by odontocetes, from dolphins by quickly moving small mammals adapted to complex 5 6 up to pygmy sperm whale (Madsen et al. 2005), and is but poorly-lit environments (Fenton et al. 1995; Jones 6 7 used on land by chiropterans (excluding most fruit bats & Teeling 2006; Teeling 2009; Teeling et al. 2012). Ac- 7 8 [Pteropodidae]) (Jones & Teeling 2006). Beyond these cording to the “tandem theory,” the flight and echoloca- 8 9 species no other mammal has yet been shown to rely on tion developed simultaneously (Schnitzler et al. 2003). 9 10 ultrasonic echolocation as the main means of orienta- The “flight-first theory” is best supported by current -ev 10 11 tion, either in water or on land. idence due to the fact that there is an absence of extant 11 12 Early reports on ultrasonic echolocation in shrews examples of non-volant mammals using ultrasonic echo- 12 13 (Gould et al. 1964; Tomasi 1979; Forsman & Malmquist location. Followers of the “echolocation-first theory” 13 14 1988) were subject to doubt (Konstantinov & Movchan must find a peculiar prototype: a mammal that has ac- 14 15 1985). Indeed, ultrasonic echolocation is likely disad- quired ultrasonic echolocation and lives in an environ- 15 16 vantageous as compared to touch in the typical environment more appropriate as a runway for flight than the 16 17 ment of shrews. Inside leaf-bedding the high-frequency depth of leaf-bedding. 17 18 calls would be reflected as noise from all the nearby ob- Zoologists have identified many mammals with re- 18 19 stacles, which are readily accessible to vibrissae. These duced eyes and many fast-climbing mammals, but no 19 20 calls would not pass beyond these obstacles to pro- mammal except the poorly studied Typhlomys (Fig. 1) 20 21 vide far-away echoes for long-range orientation. New- combines these extreme features together. Typhlomys, 21 22 er studies have failed to show ultrasonic echolocation in which means “the blind mouse,” belongs to an enigmat- 22 23 shrews (Catania et al. 2008; Siemers et al. 2009; Volo- ic rodent family, Platacanthomyidae (Jansa et al. 2009). 23 24 din et al. 2012, 2015; Zaytseva et al. 2015). It has been The genus Typhlomys consists of two species, of which 24 25 argued that lower-frequency calls (audible clicks and Typhlomys chapensis Osgood, 1932 (the Vietnamese 25 26 twitters) must be more appropriate for echolocation in pygmy dormouse) lives in northern Vietnam and Typh- 26 27 cluttered substrates, and can be used by shrews to gain lomys cinereus Milne-Edwards, 1877 lives in southern 27 28 echoes for orientation (Siemers et al. 2009; Volodin et China (Abramov et al. 2014). The pygmy dormice in- 28 29 al. 2015; Zaytseva et al. 2015). habit subtropical forests at the altitudes of 360–2000-m 29 30 30 All available speculations on echolocation regarding 31 31 small rodents come from behavioral experiments: dor- 32 32 mice, hamsters, voles and rats were vision-deprived and 33 33 forced to look for food (see review by Thomas & Jali- 34 34 li 2004). The only instance of potential ultrasonic echo- 35 35 location was seen in Norway rats [Rattus norvegicus 36 36 (Berkenhout, 1769)]. However, there was very limited 37 37 acoustic analysis provided in the study (Thomas & Jalili 38 38 2004). The study concluded that the calls were echo-lo- 39 39 cating based on weak evidence including the pulse-like 40 40 waveform of the calls, and on the fact that they were 41 41 produced by rats left alone in the dark. 42 42 43 Therefore, neither shrews nor typical rodents aid to 43 44 portray how a quadrupedal mammal can rely on ul- 44 45 trasonic echolocation. The issue is interesting not only 45 46 by itself, but also because such an animal, if discov- 46 47 ered, could serve as a model for the hypothesized ances- 47 48 tral state of bat echolocation. There are three current- 48 ly debated theories on how echolocation in bats may 49 Figure 1 Vietnamese pygmy dormouse Typhlomys chapen- 49 have evolved (Speakman 2001; Jones & Teeling 2006; 50 sis. Its reduced eyes are reflected in the generic name, which 50 51 Simmons 2008; Maltby et al. 2010). According to the means “the blind mouse.” 51

1 a.s.l. (Smith 2008; Abramov et al. 2012). The biology with R2 = 0.650. 1 2 of these rodents is poorly studied; they are reported to We calculated the regression equation for bats, using 2 3 be both terrestrial and burrowing (Smith 2008), and we data on 11 bat species from Howland et al. (2004) as: 3 4 4 trapped T. chapensis by live cage-traps not only on the Log (eye axial length in mm) = 0.9566+ 0.3255*Log 5 5 ground but on tree branches too (Abramov et al. 2014), (body mass in kg), (3) 6 which indicates they have climbing ability. Keepers at 6 with R2 = 0.745. 7 the Moscow Zoo, which obtained 2 of these animals, 7 8 confirm their outstanding agility, especially in an arbo- After measurements were taken, all 4 eyeballs with 8 9 real environment, and also report their strictly nocturnal a bit of surrounding tissues were embedded in paraffin, 9 10 activity and the absence of any vocalizations. One can cut into 4 or 5-μm thick serial sections and stained ac- 10 11 hypothesize that with reduced eyes, the arboreal mobil- cording to Mallory or with hematoxylin and eosin using 11 12 ity in the dark and silence could be guided by ultrasonic standard procedures. Sections were studied under a Lei- 12 13 echolocation. ca DM 2500 M microscope. 13 14 In this study we have conducted the following exper- Experimental design 14 15 iments to provide evidence for use of ultrasonic echo- 15 16 location in Typhlomys: (1) checked vision quality using Two adult males of T. chapensis (#1 and #2) were 16 17 histology to study the morphology of the eye structure; studied experimentally at the Scientific Department of 17 18 (2) investigated the presumed existence of ultrasonic the Moscow Zoo (Moscow, Russia). Like the speci- 18 19 vocalization with sensitive acoustic equipment and; (3) mens used in eye morphology research, they were col- 19 20 upon finding ultrasonic vocalization we used video and lected in northern Vietnam in the mountain tropical for- 20 21 audio recordings to correlate vocalization with locomo- est in 2010–2012 by live cage-traps set up either on the 21 22 tion in an environment that is difficult to navigate using branches or on the ground. In the Moscow Zoo, they 22 23 non-visual orientation. were kept for 10 months before the acoustic experi- 23 24 ments, under a natural light regime and room tempera- 24 25 ture (24–26 °C); individuals were isolated in glass-and- 25 26 MATERIALS AND METHODS wire-mesh cages 40 × 40 × 80 cm with mulch bedding 26 and various shelters and branches. The animals were fed 27 Eye morphology 27 28 with small rodent chow and water ad libitum. 28 29 Two ethanol-preserved adult specimens of T. chapen- Ultrasound-and-video recordings were carried out on 29 30 sis from the Zoological Institute of the Russian Acade- 2 days in July 2013. Two adult males (#1 and #2) were 30 31 my of Science (Saint Petersburg, Russia) were studied. tested singly in a separate room without other animals. 31 32 They were collected in northern Vietnam in the moun- Tests were performed in the experimental chamber (30 32 33 tain tropical forest in 2010–2012 by live cage-traps set × 50 × 100 cm) having the back 50 × 100-cm wall made 33 34 up either on branches or on the ground. These speci- of smooth plastic, the front 50 × 100-cm wall and one 34 35 mens are: a 16.7-g female (ZIN 101563) and a 18.3-g 30 × 50-cm wall made of glass, and the other 30 × 50- 35 36 male (ZIN 101566). In each specimen, both eyes were cm and both 30 × 100-cm wall made of wire mesh (10 36 37 excised, and the eyeball axial length was measured ex- × 10 mm). The chamber contained straight and furcat- 37 38 ternally with a vernier caliper (precision 0.1 mm) under ed dry branches 0.5–5.0-cm thick and the mulch bed- 38 39 a Stereomicroscope Carl Zeiss Stemi SV 11. ding. During test trials the cage was set either vertically 39 40 The expected eyeball axial length was calculated us- (see Supplementary_1.mov video) or horizontally (in re- 40 41 ing an equation from Howland et al. (2004) for mam- spect of direction of its 100-cm side), with the branches 41 42 mals in general: being re-positioned from time to time (when explorato- 42 43 Log (eye axial length in mm) = 0.9354+ 0.225*Log ry activity of the animal ceased) so that the test animal 43 44 (body mass in kg), (1) remained unfamiliar with the environment. This was 44 45 2 aimed at sustaining exploratory activity of the animals. 45 with coefficient of determination R = 0.835. 46 In total, 13 test trials (7 with male #1 and 6 with male 46 In addition, we calculated the regression equation 47 #2) were conducted. A test trial lasted 2 to 12 min, de- 47 48 separately for rodents, using data on 28 rodent species 48 pending on the behavior of the subject animal. All the from Howland et al. (2004) as: 49 test trials were performed between 1400 and 1800 hours 49 50 Log (eye axial length in mm) = 0.9243+ 0.2161*Log at temperatures of 24–26 °C; there were two trials with 50 51 (body mass in kg), (2) 51

1 natural light from windows and 11 trials with bright Computer, Tokyo, Japan) was used for a high-speed vid- 1 2 light of halogen lamps (1.5 kWt in total) for better qual- eo (HS-video, 299.7 fps, shutter speed 1/2000 s, frame 2 3 ity high-speed video recording. The test trials were au- size 512 × 384 pixels, no soundtrack). 3 4 dio and video recorded with an ultrasonic recorder and 4 Acoustic analysis 5 two camcorders simultaneously; the recordings were 5 6 synchronized by clicker signals. All acoustic files were inspected for the presence of 6 7 The Pettersson D 1000X recorder with a built-in mi- ultrasound by means of the Avisoft SASLab Pro soft- 7 8 crophone (Pettersson Elektronik AB, Uppsala, Swe- ware (Avisoft Bioacoustics, Berlin, Germany). Next, 8 9 den) was used for audio recording (sampling rate 768 540 pulses of good quality (325 pulses from male #1 9 10 kHz, 16 bit; frequency response from 5 to 235 kHz). and 215 pulses from male #2) were selected for acoustic 10 11 During recordings, the distance from the hand-held mi- analysis. After a high-pass filtration at 10 kHz (sampling 11 12 crophone to a test animal varied from 10 to 50 cm. The frequency 768 kHz, Hamming window, FFT 512 points, 12 13 microphone was oriented as far as possible directly to frame 50%, overlap 93.75%, frequency resolution 1500 13 14 the muzzle of tested animal because the sector for ultra- Hz, time resolution 0.04 ms), the maximum fundamen- 14 15 sound recording is rather narrow (see Supplementary_2. tal frequency (fmax) and the minimum fundamental fre- 15 16 mov video). Each test trial was recorded as a WAV-file. quency (fmin) were measured for each pulse, and the 16 17 The total duration of audio recordings was 62 min (30 pulse bandwidth as bandw = fmax–fmin (Fig. 2b) was 17 18 min from male #1 and 32 min from male #2). calculated. The pulse duration (dur), the duration to the 18 19 19 Two camcorders were used simultaneously for vid- pulse maximum amplitude in percentage of the entire 20 20 eo recording: the JVC GC-PX10 camcorder (Vic- pulse duration (disttomax) (Fig. 2b) and the peak fre- 21 21 tor Company of Japan, Yokohama, Japan) was used quency of a pulse (fpeak) (Fig. 2c) were measured by 22 22 for a high-definition video (HD-video, 50 fps, shut- applying the automated parameter measuring option of 23 23 ter speed 1/1000 s, frame size 1920 × 1080 pixels, with Avisoft. The periods between pulses (period), from the 24 24 soundtrack) and the Casio EX-F1 camcorder (Casio beginning of a pulse to that of the next pulse (Fig. 2a) 25 25 26 26 27 27 28 28 29 29 30 30 31 31 32 32 33 33 34 34 35 35 36 36 37 37 38 Figure 2 The waveforms, spectrograms 38 39 and power spectrum (c) representing 39 40 acoustic patterns and acoustic variables 40 measured from the vocal pulses of Typh- 41 41 lomys chapensis. (a) Natural sequence of 42 42 pulses, of which the first 4 comprise a typi- 43 43 cal bout; period, the period between pulses. 44 44 (b) and (c) A pulse without echo; fmax, the 45 maximum fundamental frequency; fmin, 45 46 the minimum fundamental frequency; 46 47 bandw, the pulse bandwidth; dur, the pulse 47 48 duration; disttomax, the duration to the 48 49 pulse maximum amplitude; fpeak, the peak 49 50 frequency. (d) A pulse with weak echo. (e) 50 51 A pulse with strong echo. 51

1 were also measured. Measurements were exported auto- Some ultrasonic pulses together with their echoes 1 2 matically to Microsoft Excel software (Microsoft, Red- could have been missed by the microphone due to the 2 3 mond, WA, USA) for further processing. narrow beam of ultrasound propogation and the move- 3 4 As the study sample (only 2 animals) was too small ment of the subject animal outside of the audio record- 4 5 for individual-based statistical analysis, we used the ing reach. At the same time, virtually all pulses emitted 5 6 pooled samples from both individuals according to by resting animals were collected. However, some un- 6 7 Leger and Didrichsons (1994). Analysis was carried out derestimation of the pulses emitted during locomotion is 7 8 with STATISTICA v. 8.0 software (StatSoft, Tulsa, OK, unlikely to affect the validity of the present results. 8 9 USA). Significance levels were set at 0.05, and 2-tailed 9 10 probability values were tested. Means are given as mean RESULTS 10 11 ± SD. 11 12 Eye morphology 12 13 Video analysis 13 In each of the 2 studied specimens with an average 14 To estimate the probable bearing of orientation of ex- 14 body mass of 17.5 g, the eyeball axial length was mea- 15 perimental animals on the uncovered ultrasonic puls- 15 sured as 1.4 mm. According to Equation (1), the expect- 16 es, the relationship between vocalization and locomo- 16 17 ed eye axial length for a 17.5-g mammal should be 3.5 17 tion was studied. The respective ultrasonic audio-tracks mm. According to Equation (2), the expected eye axi- 18 and video-tracks were superimposed by using Virtu- 18 19 al length for a 17.5-g rodent should also be 3.5 mm. Ac- 19 alDub (http://www.virtualdub.org/) and AviSynth (http:// 20 cording to Equation (3), the expected eye axial length 20 sourceforge.net/projects/avisynth2/) software. Ultra- 21 for a 17.5-g bat should be 2.4 mm. Thus, the eye axi- 21 sonic audio tracks and HD-video tracks were superim- 22 al length of Typhlomys is very small, not only compared 22 posed based on clicker signals. The HD-video tracks 23 to an average mammal or rodent, but even to an average 23 and HS-video tracks were superimposed based on ani- 24 bat. 24 mal postures. The total duration of the resulting HD-vid- 25 The measurement of the total eyeball inevitably in- 25 eo or HS-video clips with the ultrasonic audio track was 26 cluded some soft tissue at its posterior side. Measure- 26 27 57 min (see Online Resources Video for example). ments on histological sections (Fig. 3) give even a 27 28 The recordings were sampled by selecting the first smaller value of approximately 1 mm for the net eye- 28 29 0.5-s fragment from every 5-s interval. Each selected ball diameter. The lens front surface, which faces the 29 30 0.5-s fragment was classified into locomotion or resting pupil, is somewhat irregular (Fig. 3b1). The lens matter 30 31 state of the animal by viewing the video track in Vir- is very damaged on our histological sections, but looks 31 32 tualDub software. The audio track was analyzed using rather normally everywhere except for the pupil-facing 32 33 Syrinx software (http://www.syrinxpc.com/). The num- portion, which is filled with irregularly packed fibers 33 34 ber of ultrasonic pulses was counted and their arrange- (Fig. 3b2). 34 35 ment in bouts was registered for each 0.5-s fragment Our histological sections show very restricted, if any, 35 36 (detailed definition of a bout is presented in the Results vitreous chamber between the lens and retina. Some 36 37 section). The corresponding pulse rates (pulses per sec- parts of the retina tightly adjoin the lens, and the other 37 38 ond) and numbers of single pulses and 2-pulse, 3-pulse parts form irregular folds (Fig. 3a). These folds are defi- 38 39 etc. bouts per second were calculated. Fragments, in nitely not an artifact of hard ethanol fixation but the nat- 39 40 which the animal was beyond the video frame or in ural condition of the live animal, as is shown by the spe- 40 41 which ultrasonic pulses were overlapped with clicker cific structure of the nerve fiber layer in the deep folds 41 42 strikes or other noises, were excluded from the analysis. (Fig. 3b3). If they were an artifact the nerve fiber lay- 42 43 In total, 531 0.5-s fragments (247 fragments from male er would be folded together with the other layers of ret- 43 44 #1 and 284 fragments from male #2) were analyzed. ina to form a double lamina. Contrary to that, axons of 44 45 To compare the pulse rates in the 0.5-s fragments ganglion cells converge from the opposite walls of the 45 46 during rest and locomotion, we used the online Kolm- fold to form, inside it, a single lamina of the nerve fi- 46 47 47 ogorov–Smirnov test (http://www.physics.csbsju.edu/ ber layer; if one decided that it was an artifact and tried 48 48 stats/KS-test.html). In addition, the medians ± quartiles to spread the fold, these axons would be inevitably torn. 49 49 were estimated for the 0.5-s fragments containing pulses In the retina, ganglion cells are widely spaced from each 50 50 (pulse rate >0). other. In our series of sections, we did not find any local 51 51

1 1 2 2 3 Figure 3 Eye structure in Typhlomys chap- 3 4 ensis. (a) Section parallel to the optic axis 4 5 stained according to Mallory represents 5 6 extensive folding of retina. (b) Section 6 7 through the optic axis stained according to 7 8 Mallory at different magnifications: (b1) 8 9 general eye composition, (b2) close-up 9 10 view of the retina and (b3) close-up view 10 11 of the retinal fold. (c) Longitudinal section 11 12 of the optic nerve at its exit out of retina 12 13 stained with hematoxylin–eosin shows gli- 13 14 osis of the optic nerve. AC, anterior cham- 14 15 ber; Ch, choroid; Co, cornea; GCL, gan- 15 16 glion cell layer; I, iris; INL, inner nuclear 16 17 layer; L, lens; NFL, nerve fiber layer; ON, 17 optic nerve; ONL, outer nuclear layer (rod 18 18 nuclei); P, pigment epithelium; R, retina; S, 19 19 sclera. Scale bars 0.1 mm. 20 20 21 21 22 22 23 23 24 24 concentrations of ganglion cells. The section represent- sae, jumps down onto it (see Supplementary_3.mov vid- 25 25 ed on Fig. 3b2 allows one to count them 1 by 1. Their eo). 26 26 number in this section is as few as 70–75, including 21– 27 Acoustics 27 23 ganglion cells in the retinal fold. In accordance with 28 28 the very small amount of the ganglion cells, the nerve Almost all recordings, which we obtained in various 29 29 fiber layer is very thin too (see again the nerve fibers in experiments with Typhlomys, were surprisingly full of 30 30 the deep retinal fold represented in Fig. 3b3). Finally, soft ultrasonic vocalizations composed of rather uniform 31 31 the optic nerve of Typhlomys is characterized by an ir- units (Fig. 2). These units are pulses represented in most 32 32 regular arrangement of nuclei inside it (Fig. 3c), whose recordings by the fundamental frequency band only; the 33 33 pattern is indicative of reactive gliosis. We found no second (harmonic) frequency band was pronounced in 34 34 blood vessels supplying retina inside the optic nerve. less than 1% of pulse records. The fundamental frequen- 35 35 cy band shows a descending pattern of frequency modu- 36 Locomotor behavior 36 lation from the maximum at 127.3 ± 6.3 kHz to the min- 37 37 In captivity, Typhlomys likes to dig in mulch and, at imum at 64.1 ± 4.6 kHz, resulting in pulse bandwidth of 38 38 the same time, shows evidence of arboreal adaptation. 63.2 ± 7.3 kHz. The pulse peak frequency is 93.4 ± 7.4 39 39 Its agile movements over branches seem to indicate kHz, the pulse duration is 0.68 ± 0.15 ms, and the pulse 40 40 that its eyes function well (this cannot be true, howev- amplitude reaches a maximum in the middle of the pulse 41 41 er, with the given eye morphology). It moves with con- (51.6% ± 11.1% from the start). 42 42 fidence in an unfamiliar environment over sloping and 43 Single pulses alternate with double-pulse bouts (dy- 43 vertical branches, while also making rapid aimed leaps 44 ads), triple-pulse bouts (triads), and up to 7-pulse bouts. 44 from branch to branch over gaps several times great- 45 Bouts can be distinguished by shorter silent intervals 45 er than the tactile reach of its vibrissae (see Supplemen- 46 between the pulses: intra-bout intervals were always 46 tary_1.mov video). It can also upwardly jump 30–40 47 shorter than 23 ms, while inter-bout intervals were al- 47 cm (body length being 7–9 cm, and vibrissae not lon- 48 ways longer than 26 ms. In fact, an essential bout spec- 48 ger than 5 cm). From time to time, the animal hangs for 49 ificity exists (Fig. 2a): if there are 3 or more pulses in a 49 a while upside-down on its hind limbs and, detecting an 50 bout, they are evenly spaced, so that the intervals are al- 50 appropriate surface below, beyond the reach of vibris- 51 most equal or change gradually differing from each oth- 51

1 comotion it was more than 12.5 pulses per second. Fi- 1 2 nally, at rest, the maximum pulse rate was 38 pulses per 2 3 second, whereas during locomotion it was 64 pulses per 3 4 second. There is a significant correlation between in- 4 5 creased pulse rate and locomotion (Kolmogorov–Smirn- 5 6 ov P-value < 0.001). The Kolmogorov–Smirnov D-val- 6 7 ue reaches its maximum of 0.56 when the pulse rate is 7 8 zero. Simply speaking, even at pulse rates as low as 1, 8 9 the animal is likely to be in motion rather than at rest. 9 10 On the whole, the vocal activity during locomotion is 10 11 convincingly greater than at rest. 11 12 This does not simply mean that the more the animal 12 13 moves, the more it vocalizes: the composition of calls 13 14 changes abruptly with locomotion. We have compared 14 15 the rest and locomotion pulse distributions in the sub- 15 16 set of fragments where vocal activity was present (pulse 16 17 Figure 4 Relationship between vocalization and locomotion in rate > 0) (Fig. 4b,c). In this subset, the mean pulse rate 17 18 Typhlomys chapensis. (a) Proportion of 0.5-s fragments with- at rest was 9.1 pulses per second (Fig. 4b), of which on 18 19 out vocal pulses (white bars) and with pulses (grey or black average 5 pulses were single, 2 pulses were combined 19 20 bars) at rest and during locomotion. (b) Total pulse rates and (c) in a 2-pulse bout (dyad) and longer bouts were rare 20 21 pulse rate distributions in different bouts, from single pulse to (Fig. 4c). Contrary to that, the mean pulse rate during 21 22 a 5-pulse bout (pentad), at rest (grey bars) and during locomo- locomotion was 18.8 pulses per second (Fig. 4b), of 22 tion (black bars) in 0.5-s fragments where the vocalization was 23 which on average less than 4 pulses were single, 6 puls- 23 present (pulse rate > 0). On (b) central points show medians, 24 es were combined in three 2-pulse bouts, another 6 puls- 24 25 boxes show first and third quartiles, and whiskers show mini- 25 mum and maximum values. es were combined in two 3-pulse bouts (triads), and a 26 4-pulse bout (tetrad) was rather frequent too (Fig. 4c). 26 27 In the dataset considered, the top pulse rate (64 pulses 27 28 per second) was registered in male #1, which produced 28 29 three 4-pulse bouts and four 5-pulse bouts (pentads) 29 30 during half a second. 30 31 er only fractionally; contrary to that, the intervals before 31 32 and after the bout are several times longer, reaching the 32 DISCUSSION 33 full bout duration or exceeding it. The criterion of sub- 33 34 equal intra-bout intervals is inapplicable to dyads, but Degree of blindness 34 35 they are well distinguished from 2 single pulses in that 35 36 the total duration of the true dyad is less than that of tri- Presented data on eye morphology (Fig. 3) strong- 36 37 ad and so on (Fig. 2a). The mean period in a bout is ly suggest vision degeneration in Typhlomys, which can 37 38 13.02 ± 3.00 ms, with the minimum and maximum val- be best appreciated by comparison with subterranean 38 39 ues of 8.70 and 23.27 ms, respectively. Further analysis blind mammals such as the mole Talpa (Carmona et al. 39 40 of bouts is beyond the scope of this study. 2008; Carmona et al. 2010; Quilliam 1966) and the na- 40 41 ked mole rat Heterocephalus glaber Ruppell, 1842 (Ni- 41 Vocal activity 42 kitina et al. 2004). Contrary to mice and similar to more 42 43 In 432 of the 531 0.5-s fragments, analyzed for a re- mature naked mole rats (Nikitina et al. 2004), the lens 43 44 lationship between vocalization and locomotion, the surface in Typhlomys shows some wavy irregularities. 44 45 test animals were resting, whereas in 99 fragments The lens matter in Typhlomys looks rather similar to that 45 46 they were moving. At rest, vocal pulses were scored in in the mouse and the naked mole rat; it is definitely not 46 47 only 10.6% of fragments (n = 46), whereas during lo- as bad for light conduction as in the mole, whose lens 47 48 comotion the pulses were scored in two-thirds of frag- is entirely filled with disorganized fibers and cell nuclei 48 49 ments (n = 66) (Fig. 4a). At rest, the mean pulse rate (Carmona et al. 2008; Quilliam 1966). In Typhlomys, 49 50 was less than 1.0 pulse per second, whereas during lo- the lens fibers are disorganized at the pupil-facing side 50 51 51

1 only and no cell nuclei are visible. On the whole, the more normal, with the blood vessels in the middle and 1 2 lens of Typhlomys is diminished approximately like that nerve fibers on the periphery (Quilliam 1966). 2 3 in the mole rat but less than in the mole. On the whole, the type of eye diminishment observed 3 4 Irregular retinal folds are even more extensive in Ty- is rather similar in Typhlomys and the naked mole rat. 4 5 phlomys than in the naked mole rat (Nikitina et al., This is in contrast to the mole eye, which has a more de- 5 6 2004). Such folds protrude beyond the focal plane; they stroyed light entrance (the lens) but has better retained 6 7 are ineffective with respect to object vision and destroy those parts that are located downstream in respect of 7 8 continuity of image projection on the retina. The reduc- the signal propagation. We can conclude that Typhlo- 8 9 tion of vitreous chamber resulting in too short a distance mys, like the two subterranean mammals mentioned 9 10 between lens and retina is detrimental for focusing. above, cannot focus on an obstacle or target such as a 10 11 Notably, the mole eye appears better suited for focus- tree branch; at best, its eyes can distinguish dark from 11 12 ing than the eyes of Typhlomys and the naked mole rat light or help photoperiodicity, like in the mole (Carmo- 12 13 in that it retains big depth of the vitreous chamber and na et al. 2010). The retention of eyes for photoperiod- 13 14 does not have image-destructive retinal folds (Quilliam icity may be crucial for survival of Typhlomys as an aid 14 15 1966; Carmona et al. 2008). in avoiding vision-relying predators. Indeed, observa- 15 16 16 The acuity of the retina can be estimated by the num- tions by the Moscow zookeepers indicate that Typhlo- 17 17 ber of its ganglion layer cells. Mice possess a single lay- mys avoids coming out of the shelter when the darkness 18 18 er of tightly packed ganglion cells (Nikitina et al. 2004; is incomplete. Adhering to complete darkness, which 19 19 Carmona et al. 2010). In the Iberian mole, a nearly was noticed by keepers in the Moscow Zoo, may help it 20 20 blind rodent, this layer does not differentiate structurally in the wild to avoid any vision-relying predators. 21 21 from neuroblasts even in adults (Carmona et al. 2010). 22 Possible evolutionary scenario of Typhlomys 22 The naked mole rat (Nikitina et al. 2004) and Typhlo- 23 23 mys show an intermediate state: the ganglion cell layer Generally, reduced eyes are characteristic of small 24 24 is present, but its cells are widely spaced from each oth- subterraneous or leaf-bedding dwellers, who mainly rely 25 25 er because their number is reduced. Our data allow us to upon smelling with whisking for orientation (Deschenes 26 26 roughly estimate that the retina of Typhlomys spread on et al. 2012; Catania 2013) and also rely upon seismic 27 27 a plane must have an area of a square of 50 × 50 gangli- sensitivity (Narins et al. 1997; Kimchi et al. 2005; Ma- 28 28 on cells, if the visually useless retinal folds are exclud- son & Narins 2010). Therefore, the degenerated eyes 29 29 ed. Therefore, in total there are approximately 2500 vi- along with the Typhlomys’ digging ability suggest that 30 30 sually useful ganglion cells in the eye of Typhlomys. To this lineage may have evolved from a semi-fossorial an- 31 31 appreciate the poor image quality of such an eye, imag- imal residing in the leaf-bedding of tropical forests. 32 32 ine a digital photographic camera with a 2.5-kilopixel A peculiarity of Typhlomys is its unique juxtapo- 33 33 sensor which is pressed against the back surface of the sition of fossorial and scansorial adaptations. We hy- 34 34 lens. Note also that the common mole’s eye is not much pothesize that the eyes of Typhlomys lost visual acui- 35 35 worse in this respect, bearing about 2000 ganglion cells ty in its leaf-bedding but non-arboreal ancestor, as in the 36 36 (Quilliam 1966). mole rat. However, we can also hypothesize with confi- 37 37 dence that high-frequency hearing did not degrade as it 38 Typically, axons of the ganglion cells converge into 38 did in subterranean rodents and moles (Konstantinov & 39 the optic nerve over the concave lens-facing surface of 39 Movchan 1985; Heffner et al. 2006; Heffner & Heffner 40 the retina, lining it with a nerve fiber layer. Reduction 40 2008). Once the chance to invade the arboreal niche in 41 of the ganglion cells’ number in Typhlomys results in 41 the tropical forest appeared, the almost blind emigrant 42 a very thin nerve fiber layer. In the naked mole rat, the 42 from a leaf-bedding environment would only be able to 43 nerve fiber layer tends to disappear after birth (Nikitina 43 crawl slowly, caring not to lose tactile contact with the 44 et al. 2004). As for the optic nerve, its normal structure 44 tree surface. The re-acquisition of vision from a mole- 45 is characterized by regular arrangement of glial cells, 45 rat-like degree of blindness is likely to have been diffi- 46 characterized by longitudinal rows of their nuclei (e.g. 46 cult. If so, ultrasonic echolocation would have become 47 in a mouse; Nikitina et al. 2004). In the naked mole rat, 47 the only available replacement of vision for long-range 48 as well as Typhlomys (Fig. 3c), this regular pattern is de- 48 orientation and agile locomotion in the complex 3-D 49 stroyed, and the optic nerve corridor is plugged with 49 environment full of branches, which can suddenly be- 50 cluttering nuclei indicating reactive gliosis (Nikitina et 50 come footholds or obstacles. Indeed, the definite posi- 51 al. 2004). In contrast, the optic nerve of moles looks 51

1 tive correlation of ultrasound production with acrobatics The higher the perceived echo frequency, the closer 1 2 on branches, as well as the serial arrangement of these the use of sonar approaches vision with respect to acu- 2 3 short ultrasonic calls is definitively echolocation. ity. Therefore we can definitely conclude that the ultra- 3 4 The ultrasonic vocalization of Typhlomys is reminis- sonic activity of this unique rodent is rather advanced. 4 5 cent of bats’ echolocation signals. Both are represent- Its ancestors could hardly hear above 70–90 kHz, like 5 6 ed by pulses produced with very high peak frequency shrews (70 kHz limit), mice (80 kHz limit) and com- 6 7 (93 kHz) and combined in bouts. Structurally, the pulses mon dormice (90 kHz limit) (Konstantinov & Movchan 7 8 are most similar to frequency-modulated (FM) calls of 1985). They could have used primitive and less pre- 8 9 bats, which are relatively short sweeps with fast descent cise echolocation in the low ultrasonic range from 20 to 9 10 of frequency from the start to the end (Schnitzler Kalko 70 kHz, or even human-audible twittering as shrews in 10 11 2001; Maltby et al. 2010). In Typhlomys the frequency a leaf-bedding environment use (Siemers et al. 2009). 11 12 descends from 127 to 64 kHz during 0.68 ms. Yet, the The rise of produced and perceived frequencies must 12 13 sweeps of Typhlomys are much more faint than in bats, have accompanied the ascension from leaf-bedding onto 13 14 which is revealed by the fact that the vocalizations can- trees: fossorial-to-scansorial transition. 14 15 15 not be noticed at all with a common bat detector. The Rethinking the evolutionary scenario of bats 16 low intensity of echolocation sweeps may be explained 16 17 by their use in relatively short-range orientation only, of Ultrasonic echolocation calls have never been dis- 17 18 approximately a few decimeters in front of the animal (it covered and recorded in any climbing mammals before. 18 19 is a short range as compared to bat echolocation, but a More than that, Typhlomys is the first non-volant mam- 19 20 long range as compared to the length of vibrissae). Note mal, which relies on on-air ultrasonic echolocation for 20 21 that signals of the so called “whispering bats” hunting long-range orientation during locomotion more than on 21 22 among tree branches are poorly audible with a common any other senses. Thus, our finding allows us to recon- 22 23 bat detector as well (e.g. Griffin 2004). sider a possible evolutionary scenario of bats; name- 23 24 24 Though faint, the sweeps of Typhlomys are much ad- ly, the circumstances of echolocation origins. Based on 25 25 vanced with respect to their high frequency and, espe- molecular grounds, it is now clear that microbats, which 26 26 cially, in their frequency-modulated and bout-organized are all characterized by echolocation, are paraphylet- 27 27 structure, which was yet unknown beyond bats. Typh- ic relative to the almost exclusively non-echolocating 28 28 lomys’ sound-producing apparatus seems to have un- megabats (Tsagkogeorgasend et al. 2013; see also pure 29 29 dergone some peculiar specialization in respect of the molecular trees supplementary to O’Leary et al. [2013]). 30 30 available frequency range. It has likely entirely lost the This new fact, that non-echolocating megabats are root- 31 31 ability to produce any calls audible to the human ear. ed inside echolocating microbats, should now be tak- 32 32 This follows from the fact that during several years in en into account when answering an old question of 33 33 captivity neither keepers nor ourselves ever heard any whether or not the first ancestral bat, capable of flapping 34 34 calls of Typhlomys, which is in sharp contrast with the flight, was already able to echolocate. There two sce- 35 35 well-known ability of other small rodents, as well as narios are possible: (1) the origin of echolocation fol- 36 36 bats, for both sonic and ultrasonic vocalizations. lowed that of flight several times in parallel and the ab- 37 sence of the former in megabats is plesiomorphic, or (2) 37 The peak frequency of 93 kHz is just above the nor- 38 echolocation was already developed in the flying ances- 38 mal hearing range of non-chiropteran mammals of 39 tral bat and then secondarily lost in megabats. The Typh- 39 the size of Typhlomys, such as shrews, mice and dor- 40 lomys’ precedent seems to be good evidence in favor of 40 mice (Konstantinov & Movchan 1985; Heffner & Heff- 41 the second alternative; that is, the so-called “echoloca- 41 ner 2008). These animals can hear well only at the low- 42 tion-first theory.” This theory implies that the origin of 42 er end of Typhlomys’ calls and echoes. It is known that 43 echolocation was in small quadrupedal mammals adapt- 43 the rise of call frequencies in bats is accompanied by re- 44 ed to fast locomotion in complex but poorly-lit environ- 44 spective improvement of high-frequency hearing sensi- 45 ments (Fenton et al. 1995; Teeling et al. 2012). Simi- 45 tivity (Heffner et al. 2006), and the same must be neces- 46 lar to Typhlomys, the non-flying bat ancestor could have 46 sary to Typhlomys to make use of the energy-expensive 47 been an arboreal animal with previously reduced eyes 47 calls at the 93-kHz peak frequency. In this study, the 48 (most probably not impaired structurally as in Typhlo- 48 perception of frequencies around 90 kHz by Typhlomys 49 mys), which developed ultrasonic echolocation for com- 49 has not been shown directly: this could be the subject of 50 pensation. It must have been of great advantage when 50 51 future studies. 51

1 leaping and gliding as a means to find the best surface clusions, although rather convincing, are still prelimi- 1 2 to land upon in the dark. The initial function of echolo- nary. Additional research is required to describe in detail 2 3 cation was orientation rather than prey detection. First, the acoustic patterns of ultrasonic pulses and bouts in 3 4 this is due to the fact that for prey detection perfect Typhlomys and to compare them with the known acous- 4 5 echolocation is necessary, while even the most primitive tics of bats and with non-echolocation ultrasonic calls 5 6 usage of echoes, such as in humans (Rice et al. 1965) or of other rodents. A remaining question is the mechanism 6 7 in the piebald shrew (Volodin et al. 2012), is enough to of signal production: Is it located in the larynx? In ad- 7 8 aid orientation. The second reason is that the prey can dition, is the animal entirely incapable of communicat- 8 9 effectively be detected by echo in a homogeneous me- ing in the human-audible range? It will be of interest to 9 10 dium, such as air or water. On solid surfaces, such as investigate the degree of eye degeneration and develop- 10 11 ground or branches, even bats must detect prey by other ment of echolocation in a closely related and very sim- 11 12 means than echolocation. Namely, most of the gleaning ilar species, the Chinese pygmy dormouse, Typhlomys 12 13 bats use echolocation for orientation, but locate the prey cinereus. 13 14 on the substrate by sounds produced by prey itself (e.g. The next very interesting direction of research is as- 14 15 Arlettaz et al. 2001; Jones et al. 2003). As to the hearing sociated with the closest, and almost as rare, extant rela- 15 16 ability of ultrasonic echoes, the current understanding tive of Typhlomys: the Malabar spiny dormouse Platac- 16 17 in the field is exemplified by this statement: “the typi- anthomys. Scarce scientific data suggest their nocturnal 17 18 cal high-frequency sensitivity of small non-echolocat- activity in spite of rather small eyes and arboreal capa- 18 19 ing mammals would have been sufficient to support ini- bilities (Mudappa et al. 2001; Jayson & Jayahari 2009). 19 20 tial echolocation in the early evolution of bats” (Heffner Taken together, these traits are similar in the 2 related 20 21 et al. 2006, p. 17). genera; this suggests that Platacanthomys may too use 21 22 In the course of flight development in the first true echolocation as a means for orientation. However, the 22 23 bats, the power of ultrasonic echolocation calls must eyes in this genus, although small, are much larger than 23 24 have increased in order to satisfy the new speed of pro- in Typhlomys, judging from available photos. Therefore, 24 25 gression; flight is much faster than running or jumping there is a chance of finding a more basal stage of echo- 25 26 over branches and so sensing echoes from the more dis- location development. Perhaps Platacanthomys pro- 26 27 tant obstacles would have become crucial. The second- duces vocalizations in the low-ultrasonic or even the 27 28 ary loss of ultrasonic calls, which we propose for Ptero- human-audible range, and its calls may not be yet orga- 28 29 podidae, may have directly resulted from their transition nized in bouts. If this was true, Platacanthomys would 29 30 to the fruit diet. This is due to the associated increase of fill the “evolutionary gap” at the very beginning of the 30 31 body size with this diet and of the larynx in particular. suggested evolutionary scenario for Typhlomys, as well 31 32 This results in a corresponding decrease of fundamental as of the grand scenario for bats. 32 33 frequency of laryngeal vocalization, preventing echolo- 33 34 cating. However, some of them (e.g. the cave-living ge- ACKNOWLEDGMENTS 34 35 nus Rousettus) developed ultrasonic vocalization again 35 The field works in Vietnam were possible due to sup- 36 based on different morphological structure: by using 36 port of the Joint Vietnam–Russian Tropical Research 37 brief, broadband tongue clicks, instead of laryngeal calls 37 and Technological Centre. We are sincerely grateful to 38 (Yovel et al. 2011). Note that non-echolocating Pteropo- 38 V. V. Rozhnov for substantial organization efforts which 39 didae possess their best hearing sensitivities around 8–10 39 allowed us to undertake this study, to A. V. Shchinov, 40 and 20–25 kHz (Heffner et al. 2006), while in Rousettus 40 Nguyen Dang Hoi and Vu Van Lien for their great help 41 the second optimum of the hearing profile is shifted to 41 and scientific expertise during the field works, to the 42 45 kHz (Yovel et al. 2011). 42 43 staff of the Scientific Department of the Moscow Zoo, 43 44 and especially to O. G. Ilchenko, for the animal and 44 45 CONCLUSION room supply, to E. L. Yakhontov for assistance in ex- 45 46 The major limitations of our study were the small periments and work with literature, to B. D Vasiliev for 46 47 number of live individuals to experiment with and the his kind transport support in the crucial situation on the 47 48 poor quality of dead specimens for histology. This is rainy 15 July 2013, our first shooting day, to A. G. Bush 48 49 due to the extreme rarity of the Vietnamese pygmy dor- and E. D. Pavlov for assistance with microscopes, to O. 49 50 mouse, or “blind mouse” in nature. That is why our con- Yu. Orlov and T. V. Khokhlova for consulting regarding 50 51 eye morphology, to Al. B. Savinetsky for consulting in 51

1 statistics, to A. A. Lissovsky for help with Syrinx soft- Catania KC (2013). Stereo and serial sniffing guide 1 2 ware usage, to C. Voigt and D. N. Lapshin for valuable navigation to an odour source in a mammal. Nature 2 3 discussion on echolocation, to E. V. Volodina for partic- Communications 4, 1441. 3 4 ipation in English translation of the initial Russian text, Deschenes M, Moore J, Kleinfeld D (2012). Sniffing 4 5 and to M. Kaczmarek, D. V. Logunov and S. V. Koval- and whisking in rodents. Current Opinion in Neuro- 5 6 sky for improving the English of the final draft. biology 22, 243–50. 6 7 In all experiments we adhered to the “Guidelines for Fenton MB, Audet D, Obrist MK, Rydell J (1995). Sig- 7 8 the treatment of animals in behavioural research and nal strength, timing and self-deafening: The evolu- 8 9 teaching” (2012, Animal Behaviour, 83, 301–309). This tion of echolocation in bats. Paleobiology 21, 229– 9 10 study was approved by the Committee of Bio-ethics of 42. 10 11 11 the Lomonosov Moscow State University (research pro- Forsman KA, Malmquist MG (1988). Evidence for 12 12 tocol no. 2011-36). Video processing was performed echolocation in the common shrew Sorex araneus. 13 13 with support of the Russian Science Foundation (project Journal of Zoology 216, 655–62. 14 14-50-00029 “Scientific basis of the national biobank 14 Gould E, Negus NC, Novick A (1964). Evidence for 15 – depository of the living systems”). Acoustic analysis 15 echolocation in shrews. Journal of Experimental Zo- 16 was supported by the Russian Science Foundation (proj- 16 ology 156, 19–38. 17 ect 14-14-00237) and the Program of Basic Research 17 18 of the Presidium of the Russian Academy of Scienc- Griffin DR (2004). The past and future history of bat 18 19 es “Wildlife: Current Status and Problems of Develop- detectors. In: Brigham RM, Kalko EKV, Jones G, 19 20 ment.” Parsons S, Limpens HJGA, eds. Bat Echolocation 20 Research: Tools, Techniques and Analysis. Bat Con- 21 All the other work, including experimental setting 21 servation International, Austin, Texas, pp. 6–9. 22 and recording, eye morphology research and manuscript 22 23 preparation, was financed by the Russian Foundation Heffner HE, Heffner RS (2008). High-frequency hear- 23 24 for Basic Research (projects 14-04-01132 and 15-04- ing. In: Dallos P, Oertel D, Hoy R, eds. Handbook of 24 25 05049). the Senses: Audition. Elsevier, NY, pp. 55–60. 25 26 Heffner RS, Koay G, Heffner HE (2006). Hearing 26 27 REFERENCES in large (Eidolon helvum) and small (Cynopterus 27 28 brachyotis) non-echolocating fruit bats. Hearing Re- 28 29 Abramov AV, Aniskin VM, Rozhnov VV (2012). Karyo- search 221, 17–25. 29 30 types of two rare rodents, Hapalomys delacouri and Howland HC, Merola S, Basarab JR (2004). The allom- 30 31 Typhlomys cinereus (Mammalia, Rodentia), from etry and scaling of the size of vertebrate eyes. Vision 31 32 Vietnam. ZooKeys 164, 41–9. Research 44, 2043–65. 32 33 33 Abramov AV, Balakirev AE, Rozhnov VV (2014). An Jansa SA, Giarla TC, Lim BK (2009). The phylogenetic 34 34 enigmatic pygmy dormouse: Molecular and morpho- position of the rodent genus Typhlomys and the geo- 35 35 logical evidence for the species taxonomic status of graphic origin of Muroidea. Journal of Mammalogy 36 Typhlomys chapensis (Rodentia: Platacanthomyidae). 90, 1083–94. 36 37 Zoological Studies 53, 34. 37 38 Jayson EA, Jayahari KM (2009). Distribution of spiny 38 Arlettaz R, Jones G, Racey PA (2001). Effects of acous- tree mouse (Platacanthomys lasiurus Blyth, 1859) in 39 39 tic clutter on prey detection by bats. Nature 414, the Western Ghats of Kerala, India. Mammalia 73, 40 40 742–5. 331–7. 41 41 Carmona FD, Jiménez R, Collinson JM (2008). The mo- 42 Jones G, Teeling EC (2006). The evolution of echolo- 42 lecular basis of defective lens development in the 43 cation in bats. Trends in Ecology and Evolution 21, 43 Iberian mole. BMC Biology 6, 44. 44 149–56. 44 45 Carmona FD, Glösmann M, Ou J, Jiménez R, Col- Jones G, Webb PI, Sedgeley JA, O’Donnell CF (2003). 45 46 linson JM (2010). Retinal development and function Mysterious Mystacina: How the New Zealand short- 46 47 in a ‘blind’ mole. Proceedings of the Royal Society of tailed bat (Mystacina tuberculata) locates insect prey. 47 48 London B 277, 1513–22. Journal of Experimental Biology 206, 4209–16. 48 49 Catania KC, Hare JF, Campbell KL (2008). Water Kimchi T, Reshef M, Terkel J (2005). Evidence for the 49 50 shrews detect movement, shape, and smell to find use of reflected self-generated seismic waves for 50 51 prey underwater. PNAS 105, 571–6. 51

1 spatial orientation in a blind subterranean mammal. Schnitzler HU, Moss CF, Denzinger A (2003). From 1 2 Journal of Experimental Biology 208, 647–59. spatial orientation to food acquisition in echolocating 2 3 Konstantinov AI, Movchan VN (1985). Sounds in the bats. Trends in Ecology and Evolution 21, 386–94. 3 4 Life of Animals. Isdatelstvo Leningradskogo Univer- Siemers BM, Schauermann G, Turni H, von Merten S 4 5 siteta, Leningrad, Russia. (In Russian.) (2009). Why do shrews twitter? Communication or 5 6 6 Leger DW, Didrichsons IA (1994). An assessment of simple echo-based orientation. Biology Letters 5, 7 7 data pooling and some alternatives. Animal Behavior 593–6. 8 8 48, 823–32. Simmons NB, Seymour KL, Habersetzer J, Gunnell GF 9 9 (2008). Primitive Early Eocene bat from Wyoming 10 Madsen PT, Carder DA, Beed-holin K, Ridgway SH 10 and the evolution of flight and echolocation. Nature 11 (2005). Porpoise clicks from a sperm whale nose – 11 451, 818–21. 12 Convergent evolution of 130 kHz pulses in toothed 12 13 whale sonars? Bioacoustics 15, 195–206. Simmons NB, Seymour KL, Habersetzer J, Gunnell GF 13 14 Maltby A, Jones KE, Jones G (2010). Understanding the (2010). Inferring echolocation in ancient bats. Nature 14 15 evolutionary origin and diversification of bat echo- 466, E8–E9. 15 16 location calls. In: Brudzynski SM, ed. Handbook of Smith AT (2008). Family Platacanthomyidae. In: Smith 16 17 Mammalian Vocalization: An Integrative Approach. AT, Xie Y, eds. A Guide to the Mammals of China. 17 18 Elsevier, London, pp. 37–47. Princeton University Press, Princeton, pp. 208–9. 18 19 Mason MJ, Narins PM (2010). Seismic sensitivity Speakman JR (2001). The evolution of flight and echo- 19 20 and communication in subterranean mammals. In: location in bats: Another leap in the dark. Mammal 20 21 O’Connell-Rodwell CE, ed. The Use of Vibrations in Review 31, 111–30. 21 22 Communication: Properties, Mechanisms and Func- Speakman JR (2008). Evolutionary biology: A first for 22 23 tion across Taxa. Transworld Research Network, bats. Nature 451, 774–5. 23 24 24 Kerala, India, pp. 121–39. Teeling EC (2009). Hear, hear: The convergent evolu- 25 25 tion of echolocation in bats? Trends in Ecology and 26 Mudappa D, Kumar A, Chellam R (2001). Abundance 26 Evolution 24, 351–4. 27 and habitat selection of the Malabar spiny dormouse 27 28 in the rainforests of the southern Western Ghats, In- Teeling EC, Dool S, Springer MS (2012). Phylogenies, 28 29 dia. Current Science 80, 424–9. fossils and functional genes: The evolution of echo- 29 location in bats. In: Gunnell GF, Simmons NB, eds. 30 Narins PM, Lewis ER, Jarvis JJUM, O’Riain J (1997). 30 Evolutionary History of Bats. Fossils, Molecules and 31 The use of seismic signals by fossorial Southern Af- 31 32 Morphology. Cambridge University Press, Cam- 32 rican mammals: A neuroethological gold mine. Brain bridge, pp. 1–22. 33 Research Bulletin 44, 641–6. 33 34 Thomas JA, Jalili M (2004). Review of echolocation 34 Nikitina NV, Maughan-Brown B, O’Riain MJ, Kidson 35 in insectivores and rodents. In: Thomas JA, Moss C, 35 36 SH (2004). Postnatal development of the eye in the Vater V, eds. Echolocation in Bats and Dolphins. The 36 37 naked mole rat (Heterocephalus glaber). Anatomical University of Chicago Press, pp. 547–64. 37 38 Record Part A 277A, 317–37. Tomasi TE (1979). Echolocation by the short-tailed 38 39 O’Leary MA, Bloch JI, Flynn JJ et al (2013). The pla- shrew Blarina brevicauda. Journal of Mammalogy 39 40 cental mammal ancestor and the post-K-Pg radiation 60, 751–9. 40 41 of placentals. Science 339, 662–7. Tsagkogeorgasend G, Parker J, Stupka E, Cotton JA, 41 42 42 Quilliam AT (1966). The problem of vision in the ecolo- Rossiter SJ (2013). Phylogenomic analyses elucidate 43 43 gy of Talpa europaea. Experimental Eye Research 5, the evolutionary relationships of bats. Current Biolo- 44 44 63–78. gy 23, 2262–7. 45 Volodin IA, Zaytseva AS, Ilchenko OG, Volodina EV, 45 Rice CE, Feinstein SH, Schusterman RJ (1965). Echo de- 46 Chebotareva AL (2012). Measuring airborne compo- 46 tection ability of the blind: Size and distance factors. 47 nents of seismic body vibrations in a Middle-Asian 47 48 Journal of Experimental Psychology 70, 246–51. sand-dwelling Insectivora species, the piebald shrew 48 49 Schnitzler HU, Kalko EKV (2001). Echolocation by in- (Diplomesodon pulchellum). Journal of Experimental 49 50 sect-eating bats. BioScience 51, 557–69. Biology 215, 2849–52. 50 51 51

1 Volodin IA, Zaytseva AS, Ilchenko OG, Volodina EV Supplementary_2.mov 1 2 (2015) Small mammals ignore common rules: A 2 3 comparison of vocal repertoires and the acoustics be- Blind climbing of the Vietnamese pygmy dormouse 3 4 tween pup and adult piebald shrews Diplomesodon Typhlomys chapensis male #2. Soundtrack recorded at 4 5 pulchellum. Ethology 121, 103–15. 768 kHz sampling rate is synchronized with videotrack 5 recorded at 50 fps. Together, they are slowed down 10- 6 Yovel Y, Geva-Sagiv M, Ulanovsky N (2011). Click- 6 7 fold, so that the resulting sampling rate is 76.8 kHz, fps 7 based echolocation in bats: Not so primitive after all. is 5 and the peak frequency (fpeak) of ultrasonic pulses 8 Journal of Comparative Physiology A 197, 515–30. 8 9 appears audible at approximately 9 kHz. 9 Zaytseva AS, Volodin IA, Mason MJ et al. (2015). Vo- 10 10 cal development during postnatal growth and ear Supplementary_3.mov 11 11 morphology in a shrew that generates seismic vibra- 12 Nine seconds of blind climbing of the Vietnam- 12 tions, Diplomesodon pulchellum. Behavioural Pro- 13 ese pygmy dormouse Typhlomys chapensis male #1. 13 cesses 118, 130–41. 14 Soundtrack recorded at 768 kHz sampling rate is syn- 14 15 chronized with videotrack recorded at 299.7 fps. At the 15 16 SUPPLEMENTARY MATERIALS beginning the record is reproduced at normal speed; 16 17 then videotrack and soundtrack are slowed down 20- 17 18 Supplementary_1.mov fold, so that the resulting sampling rate is 38.4 kHz, fps 18 is 14.985 and the peak frequency (fpeak) of ultrasonic 19 Typical exploratory behavior of the Vietnamese pyg- 19 pulses appears audible at approximately 4.5 kHz. Note 20 my dormouse Typhlomys chapensis (male #1) in the ex- 20 especially, the non-stop vocalization of the animal hang- 21 perimental cage set vertically. Natural speed playback, 21 ing upside-down on its hind limbs. 22 50 fps. Note especially, 2 jumps from branch to branch. 22 23 23 24 24 25 25 26 Cite this article as: 26 27 27 Panyutina AA, Kuznetsov AN, Volodin IA, Abramov AV, Soldatova IB (2017). A blind climber: The first evidence 28 28 of ultrasonic echolocation in arboreal mammals. Integrative Zoology 12, 172–84. 29 29 30 30 31 31 32 32 33 33 34 34 35 35 36 36 37 37 38 38 39 39 40 40 41 41 42 42 43 43 44 44 45 45 46 46 47 47 48 48 49 49 50 50 51 51