Rutting Vocalizations of Formosan Sika Deer Cervus nippon taiouanus Acoustic Structure, Seasonal and Diurnal Variations, and Individuality

Rutting Vocalizations of Formosan Sika Deer Cervus nippon taiouanus Acoustic Structure, Seasonal and Diurnal Variations, and Individuality Author(s): Shih-Ching Yen, Bao-Sen Shieh, Yi-Ting Wang and Ying Wang Source: Zoological Science, 30(12):1025-1031. 2013. Published By: Zoological Society of Japan DOI: http://dx.doi.org/10.2108/zsj.30.1025 URL: http://www.bioone.org/doi/full/10.2108/zsj.30.1025 BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological, and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and books published by nonprofit societies, associations, museums, institutions, and presses. Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance of BioOne s Terms of Use, available at www.bioone.org/page/terms_of_use. Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial inquiries or rights and permissions requests should be directed to the individual publisher as copyright holder. BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, research libraries, and research funders in the common goal of maximizing access to critical research.

ZOOLOGICAL SCIENCE 30: 1025 1031 (2013) 2013 Zoological Society of Japan Rutting Vocalizations of Formosan Sika Deer Cervus nippon taiouanus Acoustic Structure, Seasonal and Diurnal Variations, and Individuality Shih-Ching Yen 1, Bao-Sen Shieh 2, Yi-Ting Wang 3, and Ying Wang 1 * 1 Department of Life Science, National Taiwan Normal University. 116 No. 88, Sec. 4, Tinchow Rd., Wenshan District, Taipei, Taiwan 2 Department of Biomedical Science and Environmental Biology, Kaohsiung Medical University. 807 No. 100, Shih-Chuan 1st Rd., Kaohsiung, Taiwan 3 Institute of Statistics, National Tsing Hua University. 300 No. 101, Sec. 2, Kuang-Fu Rd., Hsinchu, Taiwan In sika deer Cervus nippon, rutting vocalizations play an important role in breeding behavior. This study investigated two types of rutting vocalizations, the moan and the howl, of the Formosan sika deer C. n. taiouanus, including the acoustic characteristics of the vocalizations, the diurnal and seasonal variations of vocal activity, and individual acoustic variation and identification. The results showed that the sound levels were approximately 81 88 db(a) for the moan and 92 96 db(a) for the howl, at a distance of 7 m from the sources. From October 2006 to January 2007, eight days of continuous observations were conducted to record the type and amount of vocalizations. Both moan and howl began to occur in the middle of October and reached peaks in the middle and end of November. Thereafter, few vocalizations were recorded until mid-january 2007. Moreover, we found that 74.5% of the first portion of moan, 65.3% of the second portion of moan, and 64.2% of howl could be identified on an individual basis by using discriminant analysis with 200 iterations of cross-validation test. These results revealed that the sounds differed among individuals, and also that they could be correctly identified. Our findings add to the scientific knowledge of sika deer behavior and provide the basis for a novel method of monitoring sika deer populations. Key words: acoustic characteristics, Cervus nippon, individuality, rut, vocal behavior * Corresponding author. Tel. : +886-2-29301987; Fax : +886-2-29346443; E-mail: t43002@ntnu.edu.tw doi:10.2108/zsj.30.1025 INTRODUCTION Rutting vocalizations of deer males play an important role in their breeding behavior. Previous studies have shown that rutting vocalizations have various functions, such as signaling social dominance (Vannoni and McElligott, 2008), mediating intraspecific competition (Clutton-Brock and Albon, 1979; Reby et al., 2005), intersexual attraction (Charlton et al., 2007; McElligott et al., 1999; Reby et al., 2010), localization (Miura, 1984), and advancing the estrus of the female (McComb, 1987). As a first step toward understanding the functions of such vocalizations, the acoustic structures of the vocalizations should be investigated. Furthermore, animals can identify other individuals through individual acoustic characteristics. For example, the groans of fallow deer Dama dama are individually distinctive, and potentially allow individual recognition for intra-sexual competition (Reby et al., 1998b; Vannoni and McElligott, 2007). Female red deer Cervus elaphus can use sound to recognize a familiar stag and thus ensure the quality of their mate (Reby et al., 2001). Therefore, the ability to distinguish other individuals could be advantageous to both the sound producer and the receiver. In addition to these studies on acoustic structure and function, research on animal vocalizations has been used to develop tools to monitor the populations of large mammals, such as whales (Moore et al., 1999) and African elephants Loxodonta africana (Payne et al., 2003). Similar studies have been conducted for Cervidae. For example, Long et al. (1998) used rutting vocalizations for rapid identification of hybrids of red deer and sika deer Cervus nippon. This method facilitates the management of the hybrids, and provides information that may be useful for the protection of vulnerable native red deer in the wild. Pereladova et al. (1998) monitored the male rutting vocalizations to estimate the minimum population size of red deer, and Reby et al. (1998a) provoked the vocal response of roe deer Capreolus capreolus for use in the capture-mark-recapture method. The Formosan sika deer C. n. taiouanus is an endemic subspecies of Taiwan. It has been extinct in the wild since 1969 (McCullough, 1974). The administration of Kenting National Park in Taiwan launched a sika deer restoration program using the captive Formosan sika deer of Taipei Zoo in 1984. In recent years, wild Formosan sika deer populations have been established in the Hengchun Peninsula of southern Taiwan (Yen et al., 2012). Some captive populations have been established by deer farmers, who raise the

1026 S.-C. Yen et al. deer for their velvet. Studies on captive Formosan sika deer have suggested that its mating system was polygamy, but the female may also mate with other stags when the harem holder was absent (Wang and Chan, 1987). The Formosan sika deer mates during November and March, and most fawns are born during July and August (Wang and Chan, 1987; Yang and Ma, 1994). Sika deer vocalizations have been studied in some studies. For example, Yang et al. (2012) investigated the alarm calls of Sichuan sika deer C. n. sichuanicus. And Minami and Kawamichi (1992) categorized the vocalizations of the Japanese sika deer C. n. centralis into 13 types. The vocalizations were used for territory advertisement, alarms, maternal contact, sexual contact, attacks, and appeasement. Three of the 13 types are related to male-female intersexual behavior, and another four types are produced by males for intrasexual competition. However, only two of these sexual behavior-related call types are loud; the others are weak and can only be heard a few meters away. The two loud sexual vocalizations, the moan and the howl, are frequent and easy to observe during the rutting season. The moan is usually emitted by a harem holder and is associated with several sexual behaviors. The howl enables other individuals to locate the caller easily (Miura, 1984) and is related to territory advertisment (Minami and Kawamichi, 1992). Formosan sika deer and Japanese sika deer are geographically isolated and are identified as two separate subspecies. The vocalizations of the two subspecies are expected to differ (Davidson, 1995). Wang and Chan (1987) defined seven types of vocalizations in Formosan sika deer and provided some descriptions. However, the knowledge of the vocalizations of Formosan sika deer is limited. In this study, we present the first description of the acoustic structure of the Formosan sika deer s rutting vocalizations. The aim of this study is to contribute additional knowledge of the rutting vocalizations of Formosan sika deer, including acoustic characteristics and the seasonal and diurnal variations in vocal activity. Information on the acoustic structure will be used to test for vocal individuality. Finally, we will discuss the possibility of using these vocalizations to monitor Formosan sika deer populations that have been reintroduced into the wild and that require continuing monitoring. pens. Sound recording and analysis We recorded rutting vocalizations at the three study sites using a Sharp DR-7 MD player equipped with a Sony ECM-MS907 directional microphone or a Telinga EM23 omnidirectional microphone. We also measured the sound level of rutting vocalizations with a Rion NA-14 sound level meter on the deer farms. To measure the sound level, we hid ourselves at a distance of approximately 7 m from the calling animal to avoid disturbing the deer. Two types of rutting vocalizations, moan and howl, were investigated in the study. The moan usually start with a short and highpitched heeu sound (part I), followed by a long and low-pitched m-war sound (part II) (Fig. 1). We analyzed these two parts separately. The howl comprises a series of repeated syllables (Fig. 2). We treated each syllable as a sample for further analysis. We analyzed 93 moans and 55 howls of good quality recordings. The recordings from deer farms were recognized for their sound producer, and the ones in the wild could be recognized only in some cases that we were very close to the deer. The recordings were transferred to.wav files via Adobe Audition v.2.0 (Adobe Systems Inc., San Jose, California). Spectrograms were generated with a sampling rate of 22,050 Hz and at 16 bits amplitude resolution. Measurements were automatically made on the spectrograms using Avisoft-SASLab Pro v.4.40 (Avisoft Bioacoustics, Berlin, Germany). The fundamental frequency is the lowest frequency of a periodic waveform, which is determined primarily by the length and mass of vocal folds in animal vocalizations. The peak frequency is Fig. 1. The spectrogram of the moan vocalization of a Formosan sika deer (Cervus nippon taiouanus) stag. The moan consists of two parts. MATERIALS AND METHODS Study sites Our study included three sites and involved wild and captive deer. 1. The Sheding sika deer restoration area is part of Kenting National Park (21 57 N, 120 49 E), located in southern Taiwan. A wild sika deer population including more than 150 individuals inhabits the restoration area and its vicinity (Chen et al., 2007). 2. The Taipei Zoo is located in Taipei City (24 59 N, 121 34 E). Four stags and over 10 hinds and calves occupy a 0.4 ha area of the zoo. 3. Deer farms on which Formosan sika deer are raised are located in Taitung County (23 07 N, 121 21 E) in southeastern Taiwan. Some of these farms have fewer than 10 deer, whereas others have over 30 deer. The Formosan sika deer in the deer farms are usually housed in small pens. The stags live singly in the smaller pens, and the hinds and calves are housed together in the larger Fig. 2. The spectrogram of the howl vocalization of a Formosan sika deer (Cervus nippon taiouanus) stag. The howl is composed of several syllables.

Rutting Vocalizations of Sika Deer 1027 Table 1. Acoustic characteristics of the rutting vocalizations of male Formosan sika deer (Cervus nippon taiouanus). Sound Peak frequency (Hz) Fundamental frequency (Hz) Duration (sec.) level [db(a)] Start center end mean minimum maximum start center end mean minimum maximum Range 0.26 1.89 2464 ± 1178 2376 ± 540 1628 ± 1045 2420 ± 761 614 ± 494 2473 ± 607 1860 ± 1133 2009 ± 867 1010 ± 837 1976 ± 843 77 ± 26 1488 ± 854 Moan part I mean ± S.D. 0.74 ± 0.30 200 4300 900 3200 300 3500 900 3600 40 1630 1500 4320 400 3100 500 3100 300 2200 600 3100 20 120 310 2820 81 88 Range 0.56 3.18 987 ± 586 592 ± 449 573 ± 444 705 ± 410 469 ± 458 2716 ± 1551 479 ± 159 392 ± 111 337 ± 157 655 ± 464 101 ± 68 604 ± 299 Moan part II mean ± S.D. 1.47 ± 0.62 200 2000 200 1600 150 1900 200 1600 40 1890 710 6240 200 800 200 700 100 800 200 1700 20 310 160 1570 Howl Range 0.59 1.13 733 ± 557 1607 ± 668 615 ± 391 1124 ± 845 328 ± 336 1964 ± 361 598 ± 589 1471 ± 724 441 ± 305 1069 ± 834 59 ± 38 1574 ± 751 92 96 mean ± S.D. 0.83 ± 0.18 100 1500 450 2700 100 900 400 2400 20 1460 1310 2950 200 500 450 2700 200 800 350 2450 20 220 380 2950 the frequency with the highest amplitude. Values of fundamental frequency, peak frequency were measured at three locations of the sound spectrograms, which were the start point, the center point, and the end point, and on mean spectrum of entire sound (mean fundamental frequency, mean peak frequency). In addition, duration, minimum and maximum of peak frequency and fundamental frequency were also measured for entire sound (minimum peak frequency, maximum peak frequency, minimum fundamental frequency, maximum fundamental frequency). In total, 13 sound characteristics were measured for each sound recording. The values were averaged first for each identified individual and then for further analysis. Moreover, the number of syllables in each howl was counted. All data are shown as means ± SD. Seasonal and diurnal variations in vocal activity To better understand the seasonal and diurnal patterns of the vocal activity of the sika deer, we collected field observations to record the occurrence of moan and howl vocalizations from an observation point in the Sheding sika deer restoration area. The observations were conducted one day per week for 24 h during the sika deer breeding season. Rutting vocalizations we heard from the wild sika deer population were counted without identifying individuals. We collected twelve 24-h samples of observations from 18 October 2006 to 19 January 2007. To avoid the disturbances caused by wind and rain, four observations from rainy and windy days were excluded from the analysis. The remaining eight samples included a total of 2136 moans and 189 howls. Individual identification The moan part I, moan part II, and howl were used to identify individuals. For moan part I and part II, a sample consisted of one vocalization. For howls, a sample consisted of a single syllable. We used 112 recordings of identified callers, including individuals in the deer farms and from wild population, for further analysis. The 112 samples included 35 samples of moan part I from four individuals, 35 samples of moan part II from four individuals, and 42 samples of howl from seven individuals. We performed discriminant analysis to identify individuals by using all the 13 acoustic characteristics measured in this study. We calculated the percentage of sound samples assigned to the appropriate individual (correct classification) for each kind of rutting vocalization. Furthermore, we performed 200 cross-validation tests to evaluate the model s ability to classify additional vocalizations. For this purpose, we randomly selected three-fourth of the data for each rutting vocalization as training data to build the discriminant analysis models. The models were tested using the remaining one-fourth of the data. This procedure was repeated 200 times for each rutting vocalization, with different random samples. We calculated the mean percentage of correct classification over 200 tests. The percentages of correct classification were compared to the chance percentage expected to four (25%), four (25%), and seven (14.3%) individuals for moan part I, moan part II, and howl, respectively. All statistical analyses were performed using R 2.12.2 (R Foundation, Vienna, Australia). RESULTS Acoustic characteristics Part I and part II of the moan vocalization differed in several respects. The fundamental frequency of moan part I usually increased from low to high and then decreased (Fig. 1). The mean fundamental frequency and mean peak frequency of moan part I were 1976 ± 843 and 2420 ± 761, respectively. Moan part II was a low-pitched, prolonged sound (Fig. 1). Our observations indicated that moan part I sometimes occurred by itself, but moan part II always occurred following moan part I. The mean fundamental frequency and mean peak frequency of moan part II were 655 ± 464 and 705 ± 410, respectively. The sound level of moan was 81 88 db(a) at a distance of 7 m from the source (Table 1). The howl usually comprised several repeated syllables (Fig. 2). According to the field observations at the Sheding restoration area, the average number of syllables in a howl was 5.4 (n = 189), ranging from 1 to 11, with a mode of 7. The spectrograms showed that the syllables within a howl were very similar by visual inspection. The mean fundamental frequency and mean peak frequency of howl were 1069 ± 834 and 1124 ± 845, respectively. The sound level of the howl was 92 96 db(a) at a distance of 7 m from the source (Table 1). Seasonal and diurnal variation in vocal activity The first rutting vocalization observed in 2006 was on 9 October at the Sheding restoration area. The second vocalization occurred on 15 October; the rutting vocalizations became increasingly frequent after that date. The howl occurred most at the beginning and the end of the breeding season, whereas the moan was most frequent during the middle of the breeding season (Fig. 3). Two peaks of moan activity occurred at dawn (5:00 AM through 7:00 AM) and at dusk (5:00 PM through 6:00 PM). Two additional smaller peaks occurred at noon (11:00 AM) and at midnight (12:00 PM) (Fig. 4). The peak of howl was at dawn (6:00 AM). Another 2 peaks of howl were observed at 8:00 PM and 1:00 AM (Fig. 4). Moan and howl first occurred in mid-october and increased rapidly in occurrence, the highest peak in vocal activity occurred in mid- and late November. The amounts of both vocalizations then decreased markedly. Fewer vocalizations were observed in December 2006 than in November 2006. Only a very few rutting vocalizations were observed on 3 January 2007. We heard occasional rutting vocalizations during early and mid-january 2007. No rutting vocalizations were observed on 19 January 2007 (Fig. 5).

1028 S.-C. Yen et al. Fig. 3. Changes in the proportions of moan and howl by Formosan sika deer (Cervus nippon taiouanus) stags from October 2006 to January 2007. Individual identification The discriminant analyses yielded high rates of correct individual classification. Among the 35 moan part I samples, 100% were correctly classified. Among the 35 moan part II samples, 88.6% were correctly classified; among the 42 howl samples, 97.6% were correctly classified. In the 200 validation tests, the mean percentage of correct classification was 74.5%, 65.3%, and 64.2% for moan part I, moan part II, and howl, respectively. These percentages of correct classification are much higher than the chance percentages which are 25%, 25%, and 14.3% for moan part I, moan part II, and howl, respectively. Three, two, and four significant discriminant functions (eigenvalue > 1) were generated for moan part I, moan part II, and howl, respectively (Table 2). The examination of matrix structure (Table 2) suggested that the main contributors to vocal individuality were: duration, minimum peak frequency, and end peak frequency for moan part I; center fundamental frequency, maximum peak frequency, and mean fundamental frequency for moan part II; duration, start fundamental frequency, mean fundamental frequency, and mean peak frequency for howl. Fig. 4. Diurnal changes in the number of moans and howls made by Formosan sika deer (Cervus nippon taiouanus) stags. Eight observations were made from October 2006 to January 2007. Each point represents the hourly number of moans observed, averaged over the eight observations. Fig. 5. Changes in the number of moans and howls made by Formosan sika deer (Cervus nippon taiouanus) stags. Eight observations were made from October 2006 to January 2007. Each point indicates the number of moans occurring during a given observation. Table 2. Structure matrix from discriminant analysis using the duration, fundamental frequency variables, and peak frequency variables. The functions with eigenvalue > 1 are represented. The coefficients represent the contribution of each characteristic to the individual discrimination. Moan part I Moan part II Howl Sound characteristics Function 1 Function 2 Function 3 Function 1 Function 2 Function 1 Function 2 Function 3 Function 4 Duration 0.84 0.35 0.18 0.94 0.6 0.94 0.53 1.02 0.13 Start fundamental frequency 0.5 0.59 0.06 0.05 0.66 0.89 0.74 0.07 0.15 Center fundamental frequency 0.59 1.19 0.42 3.18 0.88 0.07 0.82 0.39 0.03 End fundamental frequency 0.28 0.15 1.03 1.29 1.8 0.2 0.5 0.17 0.19 Mean fundamental frequency 0.67 0.32 0.11 1.21 2.1 0.72 0.97 1.22 0.04 Min. fundamental frequency 0.7 0.12 0.03 0.88 0.03 0.12 0.2 0.08 0.01 Max. fundamental frequency 0.35 0.66 0.19 1.8 0.48 0.7 0.71 0.27 0.02 Start peak frequency 0.41 0.25 0.16 0.63 1.62 0.56 0.49 0.22 0.56 Center peak frequency 0.27 0.09 0.29 1.47 0.49 0.04 0.61 0.88 0.08 End peak frequency 0.36 1.47 0.39 0.01 0.51 0.79 0.08 0.43 0.81 Mean peak frequency 0.66 0.78 0.34 0.47 0.66 0.17 2.23 0.78 0.04 Min. peak frequency 0.73 0.24 0.16 1.32 0.93 0.52 0.64 0.7 1.45 Max. peak frequency 0.41 0.04 0.09 2.41 0.8 0.38 0.04 0.52 0.41 Eigenvalue 14.76 7.95 1.31 66.91 1.45 15.62 12.83 3.49 1.33 % variance explained 61.43 33.11 5.46 97.34 2.1 45.8 37.62 10.24 3.91

Rutting Vocalizations of Sika Deer 1029 DISCUSSION Acoustic characteristics Mitchell and Robinson (1995) and Miura (1984) suggested that the functions of moan and howl are to advertise the presence of the vocalizing individual and to identify the territory of the individual to potential competitors. The moan and howl of the Formosan sika deer have many harmonics and a wide band of frequency range. Wide band sounds are more easily localized than narrow band or pure tone sounds (Reby et al., 1999). The sound structures of moan and howl support the hypotheses proposed by Mitchell and Robinson (1995) and Miura (1984). In the past, sika deer have been moved and hybridized throughout Southeast Asia (McCullough et al., 2008). In Taiwan, many deer farms have been found to import alien deer species (red deer, elk Cervus canadensis, Ceylon sambar deer Rusa unicolor unicolor) and crossbreed them with Formosan sika deer to increase the velvet production about 20 30 years ago (Lo, 1999). The origin of some captive sika deer in Taiwan remains doubtful. In addition, Formosan sika deer had been exported to Japan and China (S. K. Yang personal communication). A technique that can quickly identify sika deer subspecies would benefit the management of sika deer. Although other factors such as habitat characteristics can affect the vocal behavior of animals, this behavior of deer is considered to be largely genetically influenced and can be used to reconstruct their phylogeny (Cap et al., 2008). Among subspecies, e.g., male red deer, the rutting vocalizations showed a great variation in fundamental frequency (Kidjo et al., 2008). Furthermore, Davidson (1995) has proposed that the rutting vocalizations of sika deer could be used to distinguish between subspecies. However, because comparative studies of the vocalizations of several different subspecies are scarce, Davidson s suggestion could not be tested satisfactorily. Here, we have reported the acoustic structure of a subspecies of sika deer in Taiwan. In comparing the sound characteristics of the rutting vocalizations between Formosan and Japanese sika deer (Minami and Kawamichi, 1992; Miura, 1984), we found that the fundamental frequency of the moan were similar in the two subspecies. However, the moan of the Japanese sika deer had a longer duration of about 3 4 sec. In addition, the howl of the Japanese sika deer comprised only 2 4 syllables and the duration of each syllable was approximately 3 sec (Minami and Kawamichi, 1992; Miura, 1984). The Formosan sika deer s moan had the duration of about 2 sec, and its howl had an average syllable number of 5.4 and the syllable duration of 0.84 sec. The moans and howls of the two subspecies exhibited some clear differences. These differences suggest that the hypothesis of Davidson (1995) may be valid. Seasonal and diurnal variation in vocal activity The results of the current study showed that the diurnal pattern of the moan had two peaks at dawn and dusk and two smaller peaks at noon and midnight. The peaks of the howl were observed at dawn and midnight. Vocal rut displays could be advantageous at these periods because it is difficult to visually estimate the body condition of the caller at twilight and midnight. Furthermore, the Formosan sika deer are more active at dawn and dusk (Chen, 2002), these observations suggested that the sika deer s vocal behavior is related to its overall activity pattern. The seasonal variation in the Formosan sika deer s vocal behavior is similar to that of the red deer. In both species, the vocalizations increase gradually to their peak of occurrence and then decrease dramatically (Pépin et al., 2001). However, the duration of the Formosan sika deer s breeding season is approximately three months, which is much longer than the 1-month breeding season of the red deer. The reason for this difference might be that the Formosan sika deer live in a tropical habitat, where the seasonal changes are not very obvious. Loe et al. (2005) showed that animals at higher latitudes have a higher degree of breeding synchrony than animals at lower latitudes. However, Japanese sika deer live in the temperate zone and have a 3-month-long breeding season. Therefore, we do not consider that the difference in the seasonal patterns of vocal behavior between sika and red deer results entirely from a difference in climate. According to the studies of Miura (1984) and Minami and Kawamichi (1992), howl could advertise the occupation of an area to potential opponents, and moan usually directed at other individuals. The advertising functions of howl and moan might be helpful to explain the proportional pattern observed in Fig. 3. At the beginning of rutting season, howl occurred when males established their harem or territories. After establishing their territories, the males used moan toward potential mates or competitors who approached their territories. At the late rutting season, dominant males got tired and new challengers came (Hirotani, 1994), the proportion of howl increased again. In the Sheding restoration area, most fawns are born in July and August (M. S. Pan, personal communication). On the deer farms, the fawns are born from July to November and mostly in July (Yang and Ma, 1994). The gestation period of female Formosan sika deer is 233.2 days (Shih et al., 1985). Therefore, most fertilizations may occur in November and December, which are the peak periods of rutting vocalizations. In addition, the operators of the deer farms observed that most mating and rutting vocalizations occurred in November and that vocal behavior decreased rapidly after copulation (Yen, 2008). We therefore believe that the rutting vocalization can signal the beginning of mating season. In comparing with other rutting behaviors, such as tree rubbing, clashing, and wallowing, the vocal behavior is more easily being observed in the field. The distribution of the reintroduced Formosan sika deer is still expanding (Yen et al., 2012). It is essential to monitor the process of dispersion and expansion. However, traditional survey methods, such as line transects, are difficult to apply in Kenting National Park owing to the dense vegetation and hot weather. Accordingly, we suggest developing a new method to monitor the Formosan sika deer distribution. In this study, we showed that the sound level of the Formosan sika deer s rutting vocalizations was high. Furthermore, our study also indicated the peak period and time of the rutting vocalizations. To define the distribution of the sika deer, an observer can monitor an area of interest from an appropriate location during appropriate period of time to determine the occurrence of rutting vocalizations. If rutting vocalizations

1030 S.-C. Yen et al. are recorded, they indicate that Formosan sika deer are present in the area. This method has the advantage of using less labor and time, and it is easy to apply in the environment of Kenting National Park. In the survey of Yen et al. (2012), this method was applied to monitor the sika deer population at the park. Rutting vocalizations were observed in several sites out of the known deer distribution range, indicating the expansion of deer distribution. Individual identification Vocal communication is very important for forest animals (Smith and Yu, 1992; Rossi t al., 2002; Fukui et al., 2004). Characteristics that distinguish the vocalizations of individual males can help the males recognize each other, and can therefore reduce direct conflicts (Stoddard, 1996). Such vocalizations can also help females to find and mate with familiar males (Zimmerman and Lerch, 1993). Reby et al. (1998b) suggested that female fallow deer may prefer to mate with familiar males that they have heard, and that the individually distinct vocalizations of the males might affect their reproductive success. Our study showed that the rutting vocalizations of male sika deer have individually distinct characteristics. We suggested that it likely reflects individual differences in morphology of the larynx and vocal tract (Taylor and Reby, 2010). The individually distinct rutting vocalizations may convey information used in intersexual choice or intrasexual competition. Although the high sound level of moan and howl is favorable for making sound recordings, these vocalizations are still difficult to record in the field because of the high level of background noise and the difficulty of getting close to the animals. The quality of many of the recordings that we made was unsuitable for individual identification. If the sound recording technique can be improved in the future, the higher-quality sound recording files that are obtained will represent a useful tool for population monitoring. For example, we should be able to use individual identifications to estimate the minimum size of the male population. Moreover, the vocally identified males can be treated as marked, and a mark-capture-recapture method using large-scale audio recordings with successive samplings can be designed for population size estimation. ACKNOWLEDGMENTS We thank C. H. Yang for his help with fieldwork; L. L. Lee, L. T. Chao, and S. C. Chen for their useful comments on this study. We are also grateful to the Sheding sika deer restoration area, the Taipei Zoo, Kenting Resort, and the deer farms in Taitung County for their permission to conduct this research at their facilities and for their help with this study. REFERENCES Cap H, Deleporte P, Joachim J, Reby D (2008) Male vocal behavior and phylogeny in deer. Cladistics 24: 917 931 Charlton BD, Reby D, McComb K (2007) Female red deer prefer the roars of larger males. Biol Lett 3: 382 385 Chen SC (2002) The activity pattern and habitat use of Formosan sika deer (Cervus nippon taiouanus) at Kenting National Park. Thesis, National Taiwan Normal University, Taipei, Taiwan, p 130 Chen SC, Wang Y, Yen SC (2007) Population distribution of Formosan sika deer (Cervus nippon taiouanus) in the greater Kenting National Park Area. J Nat Park 17: 43 70 Clutton-Brock TH, Albon SD (1979) The roaring of red deer and the evolution of honest advertising. Behaviour 69: 145 170 Davidson MM (1995) Variation of rutting cries in sika deer. In: SIKA, Cervus nippon Termminck, 1838 Ed by E Eick, R Konig, JA Willet, International sika society, Möhnesee, Germany, Chapter 5.3, pp 25 29 Fukui D, Agetsuma N, Hill DA (2004) Acoustic identification of eight species of bat (Mammalia: Chiroptera) inhabiting forests of Southern Hokkaido, Japan: potential for conservation monitoring. Zool Sci 21: 947 955 Hirotani A (1994) Dominance rank, copulatory behaviour and estimated reproductive success in male reindeer. Anim Behav 48:929 936 Kidjo N, Cargnelutti B, Charlton BD, Wilson C, Reby D (2008) Vocal behaviour in the endangered corsican deer: description and phylogenetic implications. Bioacoustic 18: 159 181 Lo SJ (1999) The future of sika deer genetic studies. Proceeding of Formosan Sika Deer Restoration Coference: 109 110 Loe LE et al. (2005) Climate predictability and breeding phenology in red deer: timing and synchrony of rutting and calving in Norway and France. J Anim Ecol 74: 579 588 Long AM, Moore NP, Hayden TJ (1998) Vocalizations in red deer (Cervus elaphus), sika deer (Cervus nippon), and red X sika hybrids. J Zool 244: 123 134 McComb K (1987) Roaring by red deer stags advances the date of oestrus in hinds. Nature 330: 684 649 McCullough DR (1974) Status of larger mammals in Taiwan. Tourism Bureau, Taiwan McCullough DR, Takatsuki S, Kaji K (2008) Sika Deer: Biology and Management of Native and Introduced Populations. Springer, Heidelberg, Germany McElligott AG, O Neill KP, Hayden TJ (1999) Cumulative long-term investment in vocalization and mating success of fallow bucks, Dama dama. Anim Behav 57: 1159 1167 Minami M, Kawamichi T (1992) Vocal repertoires and classification of the sika deer Cervus nippon. J Mammal Soc Japan 17: 71 94 Mitchell WR, Robinson J (1995) Some notes on vocal activity in Bowland. In: SIKA, Cervus nippon Termminck, 1838 Ed by E Eick, R Konig, JA Willet, International sika society, Möhnesee, Germany, Chapter 5.1, pp 1 5 Miura S (1984) Social behavior and territoriality in male sika deer (Cervus nippon Temminck 1838) during the rut. Z Tierpsychol 64: 33 73 Moore SE et al. (1999) Acoustic and visual detection of large whales in the eastern north Pacific Ocean. NOAA Technical Memorandum NMFS-AFSC-107, US Dept of Commerce Payne KB, Thompson M, Kramer L (2003) Elephant calling patterns as indicators of group size and composition: the basis for acoustic monitoring system. Afr J Ecol 41: 99 107 Pépin D, Cargnelutti B, Gonzalez G, Joachim J, Reby D (2001) Diurnal and seasonal variations of roaring activity of farmed red deer stags. Appl Anim Behav Sci 74: 233 239 Pereladova OB, Sempere AJ, Agrizkov E (1998) Monitoring of rare deer populations by bioacoustic methods. Proc Int Deer Biol Cong 4: 62 65 Reby D, Joachim J, Lauga J, Lek S, Aulagnier S (1998a) Individuality in the groans of fallow deer (Dama dama) bucks. J Zool 245: 79 84 Reby D, Hewison AJM, Cargnelutti B, Angibault JM, Vincent JP (1998b) Use of vocalizations to estimate population size of roe deer. J Wildlife Manage 62: 1342 1348 Reby D, Cargnelutti B, Hewison AJ (1999) Contexts and possible functions of barking in roe deer. Anim Behav 57: 1121 1128 Reby D, Hewison M, Izquierdo M, Pépin D (2001) Red deer (Cervus elaphus) hinds discriminate between the roars of their current

Rutting Vocalizations of Sika Deer 1031 harem-holder stag and those of neighbouring stags. Ethology 107: 951 959 Reby D, McComb K, Cargnelutti B, Darwin C, Fitch WT, Clutton- Brock T (2005) Red deer stags use formants as assessment cues during intrasexual agonistic interactions. Proc R Soc B 272: 941 947 Reby D, Charlton BD, Locatelli Y, McComb K (2010) Oestrous red deer hinds prefer male roars with higher fundamental frequencies. Proc R Soc B 277: 2747 2753 Rossi I, Mauri L, Laficara S, Apollonio M (2002) Barking in roe deer (Capreolus Capreolus) seasonal trends and possible functions. Hystrix 13: 13 18 Shih CH, Yang SK, Sung SM, Hwang KH (1985) The study on the status of Formosan sika deer farming. In The study on the restoration of Formosan sika deer Ed by Y Wang, Kenting National Park, Pintung, Taiwan, pp 248 275 Smith JI, Yu HT (1992) The association between vocal characteristics and habitat type in Taiwanese passerines. Zool Sci 9: 659 664 Stoddard P (1996) Vocal recognition of neighbors by territorial passerines. In Ecology and Evolution of Acoustic Communication in Birds Ed by DE Kroodsma, EH Miller, Cornell University Press, Ithaca, New York Taylor AM, Reby D (2010) The contribution of source-filter theory to mammal vocal communication research. J Zool 280: 221 236 Vannoni E, McElligott AG (2007) Individual acoustic variation in fallow deer (Dama dama) common and harsh groans: a sourcefilter theory perspective. Ethology 113: 223 234 Vannoni E, McElligott AG (2008) Low frequency groans indicate larger and more dominant fallow deer (Dama dama) males. PLoS ONE 3:e3113 Wang Y, Chan SC (1987) The study on the restoration of Formosan sika deer. Kenting National Park, Pintung, Taiwan Yang SK, Ma CS (1994) The reproductive performance and antler cycle of Formosan sika deer (Mei-Hau-Lok; Cervus taiouanus) in Taiwan. J Chinese Soc Anim Sci 23: 43 55 Yang C, Xiao Z, Guo Y, Xiong Y, Zhang X, Yue B (2012) Alarm signals of the Sichuan sika deer Cervus nippon sichuanicus. Zool Sci 29: 423 427 Yen SC (2008) The study on the vocalizations and related behavior of Formosan sika deer (Cervus nippon taiouanus) during rut. Thesis, National Taiwan Normal University, Taipei, Taiwan, p 40 Yen SC et al. (2012) Population status of Formosan sika deer (Cervus nippon taiouanus) in Kenting National Park and nearby area. J Natl Park 22: 27 40 Zimmerman E, Lerch C (1993) The complex acoustic design of an advertisement call in male mouse lemurs (Microcebus murinus, Prosimii, Primates) and sources of its variation. Ethology 93: 211 224 (Received March 4, 2013 / Accepted August 15, 2013)