Introduction

Investigations of intraspecific variation among different populations provide important clues to understand the evolutionary mechanisms of phenotypic traits of animals. Different environmental conditions in geographically remote areas can induce phenotypic variation among a species (Schlichting and Pigliucci, 1998; Foster and Endler, 1999). Elucidation of factors that have caused such intraspecific variation is one of the ultimate goals in evolutionary biology. The essential first step toward this goal is, however, to clarify and describe the pattern of intraspecific geographic variation. Widely distributed species are, therefore, ideal subjects to untangle the complex process of the phenotypic evolution.

Gloydius blomhoffii is a common, geographically widespread venomous snake in Japan, distributed in all main islands (Hokkaido, Honshu, Shikoku, and Kyushu) and their adjacent islands, covering various habitats from the subfrigid to temperate climatic zones (Gloyd and Conant, 1990; Goris and Maeda, 2004). Gloydius blomhoffii is a pit viper endemic to Japan and is a well-known snake even to local people because of the frequent occurrences of human envenomation in rural areas (Okamoto et al., 2017; Chiba et al., 2018). Accordingly, numerous studies on the toxicity of its venom and medical treatments have been reported (e.g., Hifumi et al., 2011, 2015; Taki et al., 2014; also see references in Toriba and Sawai, 1990). In contrast, basic ecological studies in the field, particularly population-level studies such as Kadowaki (1996) and Mori and Nagata (2016), are quite limited, in spite of the species' apparent familiarity to biologists and its potential importance for applied medical sciences.

During a long-term ecological study on diurnal colubrid snakes conducted in Kyoto, western Honshu area, I had a number of opportunities to encounter G. blomhoffii and to record some of its ecological traits. In this study, I present basic data on body size, monthly and daily activity patterns, thermal and reproductive characteristics, and diet of G. blomhoffii collected over 30 years. Then, I briefly compare these traits with those reported in previous studies conducted in its northernmost and southernmost geographic ranges, Hokkaido (Sasaki et al., 2008, 2012) and southern Kyushu (Yomeishu Seizo Co., Ltd., 1999), respectively, and in the eastern side of its central geographical range, Tsukuba (Kadowaki, 1996). Here, I provide basic ecological features of a population in the western side of the central range of G. blomhoffii and briefly analyze the pattern of geographic variation in the species, observed over the environmentally diverse Japanese islands.

Materials and Methods

The study site (35°18'N, 135°43'E) is located in the northeastern area of Kyoto Prefecture, Japan, close to the borders with Shiga and Fukui Prefectures. The study was conducted in the Ashiu Forest Research Station of the Field Science Education and Research Center, Kyoto University and in adjacent paddy fields owned by local residents. The elevation of the study area ranges from 355 to 688 m. Air temperature of the study site is lowest in January (the average air temperature is 0°C), and highest in August (the average air temperature is 25°C). The Yura River runs across the study site, and many small streams flow into the river. The study site consists of hilly terrain, and most areas are covered with dense forest consisting mainly of Aesculus turbinata, Pterocarya rhoifolia, Quercus cripula, Q. salicina, and coniferous plantation (mostly Cryptomeria japonica). Grasslands dominated by Miscanthus and Phragmites are scattered in open flat areas along the banks of the river. There is a small hiking trail along the river that extends more than 5 km.

The field survey was conducted mainly along the hiking trail and in and around the nearby paddy fields between April and October, from 1989 to 2020. The frequency of the survey varied year to year. I haphazardly walked along the survey area basically between 0800 h and 1800 h, and occasionally at dusk and night between 1700 h and 0300 h. Hence, research efforts were not constant nor controlled over months, hours, or areas. Therefore, I avoided analyzing data that are biased by these restrictions (see Results). I attempted to collect snakes whenever I found them on the surface. I did not collect snakes hiding themselves in shelters. Immediately after a snake was collected, I measured cloacal body temperature to the nearest 0.1°C with a thermistor (Takara, Digimulti D611 or Sato, Sk-1260). The ambient air temperature at 1 m above the ground (AT) (in shade in case of daytime) and the substrate temperature (ST) of the site, at which the snake was first observed were also measured. Sunlight exposure of the snake's body was recorded either as full sun, filtered sun, or shade. The snake was brought to the field station, and snout-vent length (SVL), tail length (TaL), and body mass (BM) were measured, and its sex was determined by examining the external shape of the tail base or by everting hemipenes. Reproductive condition of females (pregnant or non-pregnant) was determined by gentle palpation of the ventral side of the posterior part of the body. Stomach contents were detected by gentle palpation and were recovered by force-regurgitation.

I did not release the snakes because it was either not allowed by local land owners or not recommended by the Ashiu Forest Research Station, due to the possibility of snake bites to local residents or hikers by the released snakes. Thus, all collected snakes were brought back to the laboratory at Kyoto University for behavioral experiments or for preservation and deposit in the Kyoto University Museum. Reproductive condition and stomach contents of some individuals were examined by the dissection of the preserved specimens. Because of occasional logistic limitations related to the handling of venomous snakes, I was not able to record all variables described above for each snake, and thereby sample size varies depending on analyses. If females gave birth in the laboratory, I measured SVL, TaL, and BM of neonates and determined their sex, if possible.

Sexual differences in SVL were examined using Student t-test. Relative TaL and BM was compared between the sexes using ANCOVA with SVL as covariate. Because of the small range of SVL and high linear correlation coefficients of SVL with TaL and BM, I did not log-transform the data for ANCOVA. Paired t-test was used to compare BT with corresponding AT and ST. BT was compared among males, pregnant females, and non-pregnant females using ANCOVA with AT or ST as covariate. In ANCOVA only the results of elevation are shown because significant effects in slope were not detected in any comparison. Sex ratio of collected snakes was examined using the Binomial test. Differences in activity time (day or night) and sunlight condition between the sexes was examined using the Fisher exact test. All statistical analyses were conducted using JMP Pro (ver. 15.1.0), with a significance level of P=0.05. To examine a possible long-term change in body size, which could be caused by environmental changes in the study site (e.g., Kato and Okuyama, 2004; Fukushima and Tokuchi, 2008), the survey period was divided into three decades (1989–2000, 2001–2010, and 2011–2020), and SVL was compared among these periods.

Results

In total, 16 males and 53 females were collected. This sex ratio was significantly biased from 1:1 (Binomial test, P<0.0001). The earliest and latest dates I collected G. blomhoffii were 30 April and 12 October, respectively (both were females). The monthly pattern of fluctuation in the number of collected snakes was similar in males and females (Table 1). Pregnant females were collected from 6 May to 9 October. Overall, 61.0% of females were determined to be pregnant (Table 1). Eleven of 16 males were collected during the day (0800–1700 h), whereas 41 of 49 females were collected during the day (Table 2). Females tended to be collected more frequently during the day than at dusk or at night (1700–0300 h) compared to males, but the difference was not significant (Fisher exact test, P=0.279). There was also no significant difference in the activity times of pregnant and non-pregnant females (P=0.658).

Table 1.

The number of Gloydius blomhoffii collected each month. PF: pregnant female, NPF: non-pregnant female, NDF: female for which reproductive condition was not determined.

Table 2.

The number of Gloydius blomhoffii collected during each time period of the day. PF: pregnant female, NPF: non-pregnant female, NDF: female for which reproductive condition was not determined.

Females were significantly larger than males in SVL (t-test, t=2.36, df=52, P=0.0218; Table 3). SVL of pregnant females completely overlapped with that of non-pregnant females (range, 420–522 mm and 442–502 mm in pregnant and non-pregnant females, respectively; Table 3). Relative TaL was significantly larger in males than in females (ANCOVA, F=92.2, df=1, 53, P<0.0001). Relative BM was significantly larger in females (all individuals combined) than in males (ANCOVA, F=8.02, df=1, 50, P=0.0067). Relative BM of pregnant females was significantly larger than that of males (ANCOVA, F=26.2, df=1, 29, P<0.0001), but there was no significant difference in relative body mass between non-pregnant females and males (ANCOVA, F=0.005, df=1, 22, P=0.946).

Among 15 males, three individuals were observed in full sun and the remaining 12 were sighted in shade. Among pregnant females, five, three, and 14 individuals were observed in full sun, filtered sun, and shade, respectively. Of the non-pregnant females, two, three, and nine individuals were observed in full sun, filtered sun, and shade. The proportion of snakes observed in sunlight (full sun or filtered sun) was not significantly different among males, pregnant females, and non-pregnant females (20%, 36.4%, and 35.7%, respectively; Fisher exact test, P=0.553).

Mean BT of males, pregnant females, and non-pregnant females were 22.6°C, 25.8°C, and 23.8°C, respectively (Table 4). In males and pregnant females BT was significantly higher than corresponding AT and ST (paired t-test; males, AT vs. BT, t=2.6, df=10, P=0.0402, ST vs. BT, t=3.5, df=10, P=0.0072; pregnant females, AT vs. BT, t=6.1, df=19, P<0.0001, ST vs. BT, t=4.0, df=19, P=0.0008), whereas in non-pregnant females there was no significant difference either between AT and BT or between ST and BT (former, t=1.6, df=7, P=0.1640; latter, t=2.3, df=7, P=0.0527). BT was significantly correlated with both AT and ST in males, pregnant females, and non-pregnant females (AT, males, r²=0.79, F=34.54, df=1, 9, P=0.0002, pregnant females, r²=0.30, F=7.77, df=1, 18, P=0.0121, non-pregnant females, r²=0.66, F=11.68, df=1.6, P=0.0142; ST, males, r²=0.74, F=25.69, df=1, 9, P=0.0007, pregnant females, r²=0.74, F=52.21, df=1, 18, P<0.0001, non-pregnant females, r²=0.90, F=54.55, df=1.6, P=0.0003). There were significant differences in elevation of regression lines between AT and BT among these three sex/reproduction categories (ANCOVA, F=5.47, df=2, 33, P=0.0089). Paired comparisons showed significantly higher BT in pregnant females than in males and non-pregnant females (former, F=10.38, df=1, 28, P<0.0001; latter F=3.75, df=1, 25, P=0.0001) and no significant difference between males and non-pregnant females (F=0.31, df=1, 16, P=0.5849). On the other hand, there was no significant difference in elevation of regression lines between ST and BT among the three sex/reproduction categories (ANCOVA, F=1.07, df=2, 33, P=0.3559).

Table 3.

Mean±SE of body size measurements of Gloydius blohmhoffii in Kyoto. Sample size and range are shown in parentheses. SVL: snout-vent length, TaL: tail length, ToL: total length, BM: body mass. ToL is shown to facilitate comparison with previous studies.

Table 4.

Mean±SE of body temperature (BT) of Gloydius blohmhoffii and corresponding air temperature (AT) and substrate temperature (ST). Sample size and range are shown in parentheses.

Although sample sizes were small, mean SVLs were similar throughout the three decadal periods in both sexes. In males, mean SVL of the earliest, middle, and latest periods were 439.2 mm (n=6), 438.0 mm (n=3), and 452.2 mm (n=5), respectively, and in females those were 460.4 mm (n=28), 466.2 mm (n=10), and 457.8 mm (n=6), respectively.

Table 5.

Details for snakes that contained stomach contents. SVL: snout-vent length.

Stomach contents were found only in two out of 16 males and four out of 50 females (9.1% in total). Both males, which were collected in May and September each, contained rodents, and all four females contained amphibians, including two frogs and two newts (Table 5).

Ten females gave birth to neonates in the laboratory between 17 September and 20 October. Litter size varied from 2–6 (mean=4.0, including still-born neonates). Mean±SE of SVL, TaL, and BM of neonates were 183.4±10.2 (n=14), 42.1±3.1 (n=14), and 5.9±0.9 (n=14) for males, and 181.4±9.4 (n=10), 35.7±2.5 (n=10), and 5.9±1.1 (n=10) for females, respectively (sex of 16 neonates was not determined). Relative TaL was significantly larger in males than in females (ANCOVA, F=76.6, df=1, 20, P<0.0001).

Discussion

Sexual size dimorphism observed in this study agrees with the general trend reported in previous studies of G. blomhoffii conducted in various areas: females have larger SVL and larger relative BM than males, and males have longer relative TaL than females (Uchida and Imaizumi, 1939; Kadowaki, 1996; Yomeishu Seizo Co., Ltd., 1999; Sasaki et al., 2012; Mori and Nagata, 2016). A previous study conducted in Hokkaido showed, by comparison with studies in two southern areas, that adults are larger in northernmost populations than in southern Japan (Sasaki et al., 2012). A comparison of the present data with those previously reported from seven other areas partially supports this tendency: largest body size in the northernmost area and smallest body size in the southernmost area (Table 6). However, the trend does not necessarily show a clear latitudinal gradation, and especially on Honshu, the mainland of Japan, the pattern of geographic variation does not seem simple. For example, G. blomhoffii in Aomori and Kyoto, which are separated from each other by more than 800 km in straight distance, exhibit similar body sizes, whereas G. blomhoffii on Kinkasan Island have larger body size compared to those of the close mainland area, Aomori. Intraspecific variation in body size is a well-known phenomenon in snakes (e.g., Boback, 2003), and such variation would reflect differences in longevity, plastic responses in growth to local environments, or local adaptation (Hasegawa and Mori, 2008). Obviously, more intensive, comparative field studies focusing on locally characteristic environments are required to elucidate the proximate and ultimate factors underlying geographic body size variation of G. blomhoffii.

Although my survey efforts were not constant across months and previous studies employed various survey methods, the active season of G. blomhoffii in my study site generally agrees to that in other areas, i.e., activity from late April to October (Uchida and Imaizumi, 1939; Kadowaki, 1996; Okawa, 2002), although in its southernmost range the snake has been reported to be active until mid-November (Yomeishu Seizo Co., Ltd., 1999). Anecdotal descriptions often mention that G. blomhoffii is primarily nocturnal, and diurnal activity mostly reflects basking by pregnant females to raise the developmental temperature of their embryos (e.g., Toriba, 1996). However, there is no behavioral study that tracks the actual activity pattern of individual snakes. Because I conducted the survey primarily during the day, the higher proportion of females than males observed in my study site may reflect the frequent basking behavior of pregnant females. Nonetheless, my observation of males and non-pregnant females in the day throughout the study period indicates that diurnal activity is not limited to pregnant females in summer. Actually, despite the fact that previous field studies on G. blomhoffii were primarily conducted during the day (Kadowaki, 1996; Sasaki et al., 2008; Mori and Nagata, 2016), sex ratios at their study sites were not highly biased toward females. Furthermore, the low proportion of the pregnant females observed in sunlight (36.4%) implies that their diurnal activity is not restricted to the basking purpose. These lines of evidence suggest that G. blomhoffii as a whole is more active in the day than previously believed. On the other hand, pregnant females maintained BT higher than males and non-pregnant females at a given AT, although BT is highly correlated with ambient temperatures in all three groups. A similar trend was observed in G. blomhoffii in the Tsukuba population (Kadowaki, 1996), suggesting that pregnant G. blomhoffii thermoregulates more precisely than males and non-pregnant females, as in many other viviparous snakes (e.g., Peterson et al., 1993; Wittenberg, 2017). Again, individually based research, such as a radio-tracking study, will be essential to clarify the activity pattern and thermoregulatory behavior of G. blomhoffii.

Table 6.

Comparison of mean body length (SVL: snout-vent length, ToL: total length) of Gloydius blomhoffii among areas thoughout its geographical range. Hokkaido and southern Kyushu are its northernmost and southernmost geographical ranges, respectively. Areas are listed from left to right in order of their approximate geographic position from north to south. Numbers in parentheses are sample sizes.

The proportion of pregnant females between June and August, when the presence of embryos in the female's abdomen is accurately determined by palpation, was 60.3% (14/23) in my study site. Previous studies show that this proportion varies year to year from 0.27 to 0.59 (Sasaki et al., 2012), from 0.40 to 0.82 (Kadowaki, 1996), and from 0.24 to 0.53 (Yomeishu Seizo Co., Ltd., 1999) in Hokkaido, Tsukuba, and southern Kyushu, respectively. Parturition period of G. blomhoffii at my study site (17 September to 20 October) largely overlaps with that in Hokkaido (12 September to 24 October; Sasaki et al., 2012), Kansai (31 August to 14 October; Fukada, 1962), and southern Kyushu (late August to early October; Yomeishu Seizo Co., Ltd., 1999) despite the obvious climatic differences among these study areas. Body size of neonates and litter size are also similar to those reported in previous studies (see summary Table 2 in Sasaki et al., 2012) although mean litter size in the present study (4.0) seems slightly smaller (6.6, 3.6, 5.5, and 4.6 in Hokkaido, Tsukuba, Kansai, and southern Kyushu, respectively; Fukada, 1962; Kadowaki, 1996; Yomeishu Seizo Co., Ltd., 1999; Sasaki et al., 2012). In general, these reproductive traits of snakes are influenced by many intrinsic and extrinsic factors, such as size and nutrient conditions of females and environmental thermal conditions (Seigel and Ford, 1987; Ford and Seigel, 2011). Thus, geographic effects, even if present, would not be easily detected by a simple comparison among populations. In the future, multivariate analysis approaches coupled with controlled laboratory experiments would be helpful for revealing exclusively geographic effects on reproductive characteristics of G. blomhoffii.

Gloydius blomhoffii is a generalist feeder, eating on a variety of vertebrates and some invertebrates, such as centipedes (Mori and Moriguchi, 1988; Hamanaka et al., 2014). All four prey species identified in the present study have been reported previously in the diet of G. blomhoffii. Although the sample size is quite small, it should be noted that one-third of the prey items were newts, Cynops pyrrhogaster, which has a highly toxic defensive chemical, tetrodotoxin, in its body (Wakely et al., 1966; Brodie et al., 1974). Exploitation of newts by G. blomhoffii has been reported in other areas (Mori and Moriguchi, 1988; Yomeishu Seizo Co., Ltd., 1999; Hamanaka et al., 2014), implying that this snake has physiological tolerance against tetrodotoxin. The relatively low proportion of snakes that contained stomach contents (9.1%) may reflect the ambushing foraging mode employed by G. blomhoffii, because sit- and-wait predators generally show low feeding frequency compared to active foragers (Huey and Pianka, 1981). Nonetheless, this value is lower than that of G. blomhoffii on Kinkasan Island (26.5%), the only available comparable information for this snake. Investigation of the foraging strategy of G. blomhoffii is another important issue to clarify in order to understand its plasticity and possible adaptation to local food resources.

During the last several decades the environments of the Ashiu forest have been drastically altered by the extensive increase in the population density of deer (Cervus nippon) (e.g., Kato and Okuyama, 2004; Fukushima and Tokuchi, 2008), as well as the loss of local paddy fields (Mori, personal observation) and possible climate change due to the global warming. These drastic environmental changes may have affected basic ecological traits, such as growth, reproduction, and population dynamics of wild animals living in this forest (e.g., Kato and Okuyama, 2004; Sakai et al., 2012; Nakagawa, 2019). Fortunately, I did not detect any obvious change in adult body size of G. blomhoffii during this 31-year-research period. Continuous monitoring of this population as well as comparative studies in other areas are strongly desired in order to clarify temporal and spatial ecological responses of G. blomhoffii to different environments. This could reveal possible ecological versatility, plasticity, and local adaptation of G. blomhoffii, which has been evolving in various environments over Japan. Further intensive field studies are crucial to understand this culturally well-known, but ecologically poorly understood, venomous snake.