Seasonal Changes in Movement Patterns and Body Exposure Frequencies of Mamushis (Gloydius blomhoffii) and Their Diurnal Activity in a Mountainous Habitat of Northeastern Kyoto, Japan
Abstract
Mamushis (Gloydius blomhoffii) are widespread and common venomous snakes in Japan. Generally, G. blomhoffii are considered a nocturnal species, but some literature suggests that these snakes may be active also during daytime. We conducted radiotelemetric surveys in a mountainous forest in Kyoto to clarify the seasonal and diurnal activities of G. blomhoffii. We defined snake body exposure (visible or hiding) as an index of snake activity. We recorded a total of 295 fixes (187 during daytime and 108 at night) for nine snakes from 2020 to 2021. Time of day did not affect snake body exposure. Environment, season, and air temperature influenced body exposure tendency. Monitored snakes were usually visible when in forest habitats, whereas they were frequently hiding under shelters in riverside habitats. Mean home range size of G. blomhoffii was 1.03 ha and was relatively small compared to other snake species. Mean daily displacement was higher in August and September and lower in other months. We conclude that G. blomhoffii are a cathemeral species, active both during daytime and at night, and vary their activity and movement depending on season, temperature, and environment. The different tendencies of body exposure between forest and riverside habitats likely reflect a strategy to avoid predators and/or overheating. Unlike previous visual-based anecdotal information, use of radiotelemetry enabled us to reveal the true activity patterns of G. blomhoffii without biases related to searching ability.
Animal movements and activity patterns are closely related to their life history. Information on movements and activity patterns is essential to understanding the ecology of animals and is also important for their conservation and management. Animal movements are usually driven by activities such as foraging, resting, and mating and vary within species depending on sex, life stage, and environment (Bowler and Benton, 2005). Animals are usually considered to have fixed daily activity patterns and are classified as diurnal, nocturnal, or crepuscular. However, some species change their activity patterns according to both intrinsic and extrinsic factors, such as environmental temperature, predator activity, interspecific competition, and food availability (Curtis and Rasmussen, 2006; Fox and Bellwood, 2011). A flexible activity pattern is called cathemerality, with many such species active both in daytime and at night (Tattersall, 2006).
Mamushis (Gloydius blomhoffii) are common venomous snakes that are distributed widely across subarctic to temperate climates in Japan. Gloydius blomhoffi inhabit various environments, including mountains, forests, wetlands, and rural areas such as paddy fields (Takeuchi, 2021). A typical ambush predator, G. blomhoffii feed on fish, amphibians, reptiles, birds, mammals, and centipedes (Mori and Moriguchi, 1988). Gloydius blomhoffi have been the focus of many researchers aiming to improve safety measures for local people against their bites. However, although numerous studies on toxicity of their venom and medical treatments have been reported (Fukuda et al., 2006; Hifumi et al., 2013), basic ecological information is quite limited (Kadowaki, 1996; Sasaki et al., 2012; Mori, 2021). For example, G. blomhoffii are generally referred to as a nocturnal species anecdotally (Takeuchi, 2021), but several exploratory studies have reported that G. blomhoffii are found to be active both during the day and at night (Kadowaki, 1996; Mori, 2021; Sawada et al., 2025). The diurnal activities of G. blomhoffii are sometimes regarded as basking behavior of gravid females, but they are sometimes reported to eat diurnal prey animals such as lizards (chiefly Plestiodon and Takydromus; Mori and Moriguchi, 1988). Therefore, G. blomhoffii may be more active diurnally than previously known, and their diurnal activities may not be restricted to simply basking in the sun. Additionally, activity patterns of G. blomhoffii may have seasonal variation. Some snakes shift diel activities seasonally according to temperatures (Landreth, 1973; Abom et al., 2012) with greater degrees of diurnal activity during cold seasons and nocturnal activity during warm seasons. Indeed, anecdotal literature reports have stated that G. blomhoffii are frequently observed to be active during the daytime in cold seasons such as early spring and late autumn (Takeuchi, 2021). An open-air experiment showed a similar trend in seasonality (Yomeishu Seizo Co., 1999). However, quantitative surveys to reveal their seasonality in the field, such as radio-tracking individuals, have not been conducted.
In this study, we focused on seasonal changes in habitat use and activity of G. blomhoffii. To clarify their seasonal and diurnal activities in nature, we conducted a census survey and a radio-telemetric survey in a mountainous habitat in northeastern Kyoto. We hypothesized that G. blomhoffii would be active both in daytime and at night and that their activity would change seasonally according to ambient temperature. Additionally, we compared their movement patterns across seasons. We also estimated their home range sizes and examined their diet, which may be related to activity patterns.
Materials and Methods
Study Site
The study site is a mountainous forest in the northeastern part of Kyoto Prefecture, Japan (Fig. 1). We conducted this research at the Ashiu Forest Research Station of the Field Science Education and Research Center, Kyoto University. The study site is an area where a long-term survey of G. blomhoffii has been conducted (Mori, 2021). Therefore, basic information was already available, which was beneficial. Additionally, the study site is a protected area where people are not allowed to collect or kill wild animals without permission. The study site is in a valley surrounded by steep mountains, and elevation ranges from 355 to 610 m. The Yura River runs along the valley, and several small streams flow into the river. Most of the area is covered by secondary and planted forests, and only small riverside areas are not covered by a forest canopy. A small hiking trail runs along the river. The air temperature (AT) at the study site is lowest in January (average 0 °C) and highest in August (average 24 °C; Nakagawa et al., 2020). At our study site, G. blomhoffii have been observed to be active between April and October (Mori, 2021).
Citation: Journal of Herpetology 59, 3; 10.1670/23-065
Census Survey
We walked the study site from 9 July to 4 November 2020 and from 30 April to 23 November 2021 primarily along the trail, river, and brooks both in the daytime and at night. When we encountered a G. blomhoffii individual, we collected it for morphometric measurements, including snout-vent length (SVL), tail length, and body mass (BM), and determined sex by examining the external shape of the tail base or by everting hemipenes. SVL and tail length were measured using a tape measure by straightening snakes. We measured BM using a digital scale (KJ-212, Tanita) to the nearest 0.1 g. Reproductive condition of females (pregnant or nonpregnant) was determined by gentle palpation of the ventral side of the posterior part of the body. Stomach contents were extracted through forced regurgitation. After measurements, we marked snakes with ventral scale clippings (Brown and Parker, 1976) and released them at the collection site.
Radiotelemetry
We walked at the study site in July 2020 and May 2021 to collect G. blomhoffii for a radiotelemetric survey. After capturing snakes, we surgically implanted radio transmitters in four adult male snakes (GB02, GB04, GB05, and GB07: mean SVL 432 mm, range 420–441 mm; BM 57.9 g, range 49.4–63.5 g) and five adult female snakes (GB01, GB03, GB06, GB08, and GB09: mean SVL 424 mm, range 418–437 mm; BM 64.5 g, range 48.4–89.3 g). Snake GB03 was pregnant in 2020. We followed surgical implantation procedures described by Weatherhead and Anderka (1984) under regulations on animal experimentation at Kyoto University (Permission number 202022 and 202118). We conducted surgeries 3–7 d after capture. After a recovery period of 4–14 d from surgery (average approximately 7 d), snakes were released at their initial site of capture. Transmitters were 3.8 g, and battery life was 5 months (Model PD-2T, Holohil Systems). Due to short battery life of transmitters, we surgically replaced them in May 2021 (GB03) and September 2021 (GB03, GB06, GB07, GB08, and GB09). Transmitters represented an average of 5.9% (range 4.3–7.9%) and 6.6% (range 6.3–7.7%) of BM in male and female snakes, respectively. Although percentage values slightly exceeded general recommendations for weight of transmitters (5%: Újvári and Korsós, 2000), we considered them acceptable for the following reasons: (1) we conducted transmitter implantation surgeries 15 times, and 13 individuals survived throughout the study, (2) G. blomhoffii have stout and short bodies compared to typical colubrid snakes, and body cavities seem to have sufficient space to store the relatively large transmitter, and (3) during our survey, we did not notice any obvious effects caused by extra weight of transmitters, such as abnormal locomotion. The two individuals that died during our telemetry survey were likely unrelated to transmitter weight: snake GB04 was found dead in the river with its head cut by a sharp tool, suggesting that it was probably killed by a human. Snake GB06 died due to a handling accident during transportation prior to release.
Snake Monitoring and Body Exposure
Snakes were monitored both during daylight hours (from 0700 h to sunset) and at night (from sunset to 0400 h), between July 2020 and December 2021 with an average monitoring period of 137.3 d (SE = 20.2, range 41–210 d). One individual (GB03) was monitored in both years, and we separated its data for each year. We located each radio-tracked individual an average of every 6.40 d (SE = 0.41, range 1–35 d), and in each survey, we located individuals once or twice each day. Exact positions of snakes were confirmed visually, if possible. If snakes were not visible, we attempted to confirm the exact site as precisely as possible. To avoid disturbing snakes, we always attempted to locate them at a secure distance (approximately 5 m).
In each observation, we recorded degree of body exposure of snakes. Activity of G. blomhoffii is difficult to estimate accurately due to the following two reasons: (1) G. blomhoffii are sitand-wait foragers and can still be active even when not moving and (2) snakes, in general, do not have eyelids, and thus, we cannot determine superficially whether they are sleeping. On the other hand, several studies (Shine et al., 2003; Sivan et al., 2013) and anecdotal literature (e.g., field guidebooks) have considered body exposure of vipers as an indicator of their activity status. Therefore, we used body exposure to estimate activity of G. blomhoffii. If a snake's entire body was covered by objects such as rocks or plants, we recorded its condition as “hiding.” Otherwise, we recorded the condition as “visible.” We also recorded other variables that could affect activity of snakes. When snakes were visible, we recorded sunlight exposure of the body as either sun or shade. We also measured AT with a thermistor (ASF-270T, As One). Body temperature (BT) of the snake was calculated by the pulse interval of the transmitter. Due to failure in calibrating pulse interval for several individuals, we only used data from five individuals (GB03, GB06, GB07, GB08, and GB09). Because habitats within our study site were clearly divided into two environmental types, we recorded habitat as either forest or riverside. In forest habitats, the sky was covered by canopy, and most areas of the ground were in shade. On the other hand, most areas of the riverside were not covered by canopy.
Daily Displacement and Home Range Size
In each observation, we recorded location of snakes. We occasionally disturbed snakes, but they did not move farther than 1 m due to disturbance and did not show other reactions that might affect results. We recorded locations of snakes by smartphone GPS; however, these locations had considerable positioning errors. Thus, when we determined positions, we also checked locations with a topographical map and aerial photographs and modified the GPS locations accordingly.
At each relocation, we measured distance moved between the last and current locations using Quantum Geographic Information System (QGIS v.3.8.2). We divided distance by number of days between relocations and calculated mean daily displacement (MDD) to estimate movement distance. We calculated MDD each month and analyzed seasonal variation in snake movement distances.
We estimated home range size of snakes. Although we initially intended to use autocorrelated kernel density estimation (AKDE), most snakes had insufficient effective sample sizes for reliable AKDE results. Therefore, instead of applying AKDE, we calculated 100% minimum convex polygons (MCP) using the adehabitatHR package (Calenge, 2006) in Program R (version 4.4.1; R Development Core Team, 2024). Use of MCPs also facilitated comparisons with home range data of other snake species that used MCP as an estimation. We plotted locations on a topographic map with 100% MCP using QGIS v.3.8.2. To evaluate movement of snakes, we calculated the dynamic Brownian Bridge Movement Model (dBBMM; Kranstauber et al., 2012) with the R package move (Kranstauber et al., 2020) to estimate occurrence distributions for individuals. We selected a window size of seven, a margin size of three, and a location error of 3 m. We used two different contour levels (95% and 99%) to estimate dBBMM occurrence distributions.
Data Analysis
We used a generalized linear mixed model (GLMM) with a binomial distribution (logit-link function) to test which variables affected body exposure of snakes. The analysis was conducted using the R package lme4. The random effect was individual ID. We set snake body exposure (hiding/visible) as the dependent variable and AT, time of day (daytime/nighttime), sex, environment (forest/riverside), sunlight, and season (summer/spring and autumn) as independent variables. We considered potential interactions between several independent variables (time of day and season/time of day and AT/time of day and environment/time of day and sex/season and sunlight/sunlight and environment) because these were expected to influence outcomes. A significance level of P < 0.05 was used for hypothesis testing. Initially, we incorporated all candidate terms using the 'glmer’ function for the GLMM analysis. Non-significant interaction terms were subsequently removed from the model to obtain the best model.
We also examined effects of several variables (AT, time of day, sex, environment, snake body exposure) on BT using a linear mixed-effects model with a Gaussian distribution (identity-link function) implemented via the ‘lmer’ function in the R package lmerTest. Random effects included individual ID. For this model, degrees of freedom and P-values for the fixed effects were calculated using Satterthwaite's approximation. As with the GLMM, we initially included potential interactions between several independent variables (AT and time of day/AT and environ-ment/AT and snake body exposure/time of day and environ-ment/time of day and snake body exposure/environment and snake body exposure) and removed non-significant interaction terms from the best model. For these two models, we assessed variance inflation factors (VIF) using the 'vif’ function of the R package car to determine the potential multicollinearity among independent variables. Some VIFs exceeded 10 in the second model, and thus we removed the interactions (AT and environment/ AT and snake body exposure) from the best model.
Results
Daily Patterns of Appearance
In the census survey, we collected 18 males (12 in daytime and 6 at night) and 22 females (16 in daytime and 6 at night). Earliest and latest dates on which we collected snakes were 1 May and 3 October, respectively.
In radiotelemetry data, number of “visible” observations was larger than that of “hiding” observations, and “visible” G. blomhoffii were observed both during day and night. In forest habitats, snakes tended to be found on ground covered with dead leaves, and when we approached, snakes usually did not escape to shelters. On the other hand, in riverside habitats, snakes tended to be found on bare or mossy ground near rocks or dead wood during daytime, whereas they were typically found in an ambush posture near water without shelter at night. When disturbed during daytime, snakes at the riverside usually escaped to nearby shelters. In both environments, snakes tended to expose their bodies both during the day and at night in summer. On the other hand, in spring and autumn, snakes tended to appear in daytime and hide at night.
The GLMM analysis showed that body exposure was significantly affected by environment (Table 1). The interaction between AT and season was also significant. The effects of time of day, sunlight, sex, and other interactions were not significant. Proportion of visible snakes was higher in summer (July to September) and lower in spring, autumn, and winter (May, June, and October to December; Fig. 2A).
Citation: Journal of Herpetology 59, 3; 10.1670/23-065
Effects of AT were different between summer and other seasons. In summer, AT did not affect body exposure. However, in spring, autumn, and winter, snakes tended to appear under higher ATs and hide under lower ATs.
BT values were significantly affected by AT and environment (Table 2). Interactions between time of day and body exposure were also significant. Slope of the regression line between AT and BT was smaller in “hiding” individuals (Fig. 3A). BT at a given AT was higher in riverside habitats than in forest, especially at higher AT (Fig. 3B). Additionally, BT at a given AT of “hiding” individuals was higher than that of “visible” individuals at night. In contrast, BT at a given AT of “hiding” individuals was lower than that of “visible” individuals in daytime.
Citation: Journal of Herpetology 59, 3; 10.1670/23-065
Seasonal Movement
We recorded a total of 295 fixes (187 in daytime and 108 at night) for the nine snakes (Fig. 4). MDD of G. blomhoffii demonstrated short daily movement distances throughout most of the year (Fig. 2B). Except for one individual (GB02), snakes tended to move longer distances in August and September, and most individuals did not move frequently in May and November. The first movement event (exceeding 5 m) in a year was recorded between 21 May and 26 May (GB03, GB07, and GB09), and the last was recorded between 10 November and 18 November (GB09). After that, all snakes hid and hibernated underground.
Citation: Journal of Herpetology 59, 3; 10.1670/23-065
Home Range Size
Mean home range of monitored snakes was estimated as 1.03 ha (SD = 0.71, range = 0.13–2.57 ha) by MCP (Table 3). Means of the 95% and 99% dBBMM confidence areas were 2.09 ha (SD = 2.00, range = 0.22–7.32 ha) and 3.33 ha (SD = 3.16, range = 0.35–11.58 ha), respectively.
Diet
Stomach contents were found in only one out of 18 males and one of 22 females (5.0% overall). The male, which was collected on 19 August 2020, contained two stomach items: a Tago's Brown Frog (Rana tagoi) and a Japanese odd-scaled snake (Achalinus spinalis). The female (GB06), which was collected on 1 May 2021, contained three rodent nestlings. Two snakes (GB06 and GB08) regurgitated Scolopendra sp. while in captivity for transmitter replacement.
Discussion
Gloydius blomhoffii have generally been considered to be nocturnal animals, and their daytime activity has been believed to represent basking behavior or to be limited to cold seasons (Takeuchi, 2021). However, our results suggest that G. blomhoffii are not exclusively nocturnal. Many individuals exposed their bodies not only to the sun but also to shade in daytime. Although proportions of visible snakes varied by season, they were frequently observed in daytime, even on hot summer days. Gloydius blomhoffii seemed to be diurnal in spring and autumn, and both diurnal and nocturnal in summer at our field site. As is known in other snake species of temperate zones (Krysko, 2002; Brito, 2003; Weatherhead et al., 2012), the seasonal shift of activity in G. blomhoffii may reflect differences in ambient temperature. As in other cathemeral species, G. blomhoffii are likely to change diurnal activity according to season to avoid exposure to cold conditions of spring and autumn nights.
Body exposure of G. blomhoffii was also affected by environment with snakes being more visible in forest habitats than in riverside habitats. We suggest that these pattern differences may be related to thermoregulation. One of the main purposes of shelter use by snakes is thermoregulation (Peterson et al., 1993; Lillywhite, 2014). In our study, BT of “hiding” snakes was relatively stable compared to that of “visible” snakes, suggesting that snakes may maintain suitable BT by hiding under shelters. Previous studies (Kadowaki, 1996; Yomeishu Seizo Co., 1999) also suggested that G. blomhoffii may hide under shelters to prevent BT from becoming too high or too low, and a similar example has been shown in another species of Gloydius (Shine et al., 2003). In forest habitats, the sky is covered by canopy, and most areas of the ground are in shade, whereas most areas of riverside habitats are open and exposed to sunshine. Predation avoidance also is likely to be a reason for shelter use. Gloydius blomhoffii are known to be preyed upon by avian and mammalian predators (Tanaka and Mori, 2000), and local humans also kill the snake frequently (Fukuyama, unpubl. report). In forest habitats, G. blomhoffii was very cryptic on leaf litter. In contrast, they were conspicuous on white stones of the riverside. Gloydius blomhoffii may hide under shelters more frequently in riverside habitats to maintain BT and avoid detection by predators and humans.
Our results suggest that G. blomhoffii are a cathemeral species, active both in daytime and at night. If so, why have they been considered nocturnal animals? The main reason could be environmental differences among research sites. Our site is mountainous and colder than those of previous research (Kadowaki, 1996; Yomeishu Seizo Co., 1999; Nakagawa et al., 2020). In lowland warm areas, snakes may be nocturnal during hot seasons to avoid overheating in daytime. Additionally, previous visual encounter surveys may have introduced a bias in snake sightings between day and night. In open habitats, G. blomhoffii may be more difficult to spot in the daytime because snakes are more likely to be sensitive to approaching humans and quickly retreat to shelters when disturbed. At night, G. blomhoffii tend to be more sessile and, therefore, are easier to detect. Our study is the first to use radio-telemetry to investigate activity and movement of G. blomhoffii. We believe that we have revealed the true activity patterns of G. blomhoffii by minimizing biases associated with visual detection.
Monthly movements of the monitored G. blomhoffii showed a single peak and were highest in August and September, a pattern that is similar to movements of other Viperidae species in temperate zones (Brito, 2003; Waldron et al., 2006). Because G. blomhoffii copulate in August and September (Isogawa et al., 1995; Sasaki et al., 2012), high activity during these months may reflect mating behavior. Another hypothesis to explain high MDD in September is migration to hibernation sites because several species of snakes are known to move long distances before hibernation (Gregory et al., 1987). In our study period, the longest moves occurred between September and October, and after that, snakes tended to remain within a small area until hibernation. Hibernation sites of tracked individuals were open and sunny rock walls or scree-covered slopes, and similar habitat was also reported to be the hibernation site of G. blomhoffii (Yomeishu Seizo Co., 1999). Generally, suitable hibernation sites for snakes are different from their suitable foraging sites. Although we only monitored four snakes throughout the year, three of these (GB03 [2021], GB07, and GB09) moved from the hibernation site in spring, stayed around other areas of riverside and forest habitats in summer, and returned to the same hibernation site in autumn. Therefore, long-distance movements of G. blomhoffii in September may reflect migration to suitable hibernation sites.
Home range size of G. blomhoffii appears to be smaller than that of many other snake species (Todd and Nowakowski, 2021). According to Todd and Nowakowski (2021), larger snakes tend to have larger home ranges, and this relationship is more evident in active foragers than in sit-and-wait foragers. The relatively small home range size of G. blomhoffii may reflect their small body size and sit-and-wait foraging behavior.
In our study, stomach contents were detected in only 5% of snakes, lower than reported in previous studies: 9.1% in another study in Ashiu (Mori, 2021) and 26.5% on Kinkasan Island (Mori and Nagata, 2016). As mentioned by Mori (2021), these low proportions may reflect the ambushing foraging mode employed by G. blomhoffii because sit-and-wait predators generally show low feeding frequency compared to active foragers (Hailey and Davies, 1986; Secor and Nagy, 1994). Nonetheless, we also recorded a diet that is assumed to be obtained by active foraging. One individual (GB06), which was collected on 1 May at a rock wall, contained three nestling rodents as stomach contents. The juvenile rodents were almost the same size and were likely to be littermates, suggesting that the snake invaded a nest and consumed the littermates as an active predator. Other recorded prey animals were mainly nocturnal species, indicating that these snakes likely foraged at night.
Our results indicate that G. blomhoffii vary their activity and behavior depending on the season, temperature, and environment. For this reason, previous visual surveys and anecdotal reports might have described only the more conspicuous aspects of their behavior. Year-round radio-tracking surveys are important for understanding the ecology of snakes without the biases associated with detection ability. However, we did not find any sexual differences, likely due to a limited number of monitored snakes. In many other viviparous snakes, pregnant females are known to maintain higher BT than males and non-pregnant females (Peterson et al., 1993). As a result, pregnant females may exhibit distinct activity patterns, such as being more active during the daytime. For a better understanding of G. blomhoffii movement ecology, further field studies with larger sample sizes and diverse habitats are needed.
Maps (A, B) and photographs (C, D) of the study site. White areas in B indicate forest habitat (D) and gray areas indicate river and riverside habitat (C). Black lines in B are the 5 m contour line.
Seasonal change of activity of Gloydius blomhoffii. (A) Proportion of visible G. blomhoffii in each month. The left bar is daytime, and the right bar is at night. Numbers within each bar represent numbers of “visible” snakes and the numbers of all snakes. Numbers above bars are number of individuals. (B) Seasonal change of MDD. Each point represents the MDD of each individual snake. Triangles are males, and circles are females. Numbers above the box are number of individuals.
(A) Thermal relations of Gloydius blomhoffii. Red circles are “visible” individuals, and blue triangles are “hiding” individuals. (B) Thermal relations of G. blomhoffii when hiding and visible. Green dots represent individuals recorded in the forest, and gray dots represent those recorded at the riverside.
Home ranges of nine radio-tracked Gloydius blomhoffii. White areas indicate forest habitat and gray areas indicate river and riverside habitat. Black lines are the 5 m contour line.
Contributor Notes
