A Design Proposal for an Eco-Tunnel for Anurans Based on Behavioral Experiments and Species Characteristics

Abstract

Infrastructure development has caused extensive habitat fragmentation, and increased traffic raises the number of wildlife roadkills. Wildlife crossing structures (WCSs) can prevent roadkills and protect wildlife. Amphibians commonly comprise the highest proportion of roadkills. Nevertheless, adequate WCSs suitable for amphibian crossing and migration are lacking. This study aimed to suggest a design proposal for a WCS for amphibians based on a behavioral experiment. First, we evaluated the home range, protection value, and roadkill trends of Bufo gargarizans, Bombina orientalis, and Rana uenoi. We selected six factors that can affect anuran migration preferences: noise level, tunnel width, light-shielding level, construction material, tunnel substrate, and tunnel shape. Each species selected a migration path, and mobility was confirmed in different structures. The animals’ preferences were evaluated by counting the number of migrating individuals in each migration experiment. Overall, the three species showed a preference for 60 dB of noise and a tunnel width of D700 (mm). Migration frequencies were higher at 35% light-shielding level and when the substrate was soil. PVC was the preferred construction material for tunnels, and mobility was the highest in circular tunnels. We proposed a design for WCSs that incorporated the preferences of these three target species.

1. Introduction

Animal migration promotes species diversity and helps increase the population size of wild animals [1], thereby increasing their long-term survival. Migration to a new habitat helps animals avoid the dangers and disturbances associated with previous habitats, such as natural predators, disasters, and diseases [2]. However, anthropogenic construction projects and linear transport infrastructure (such as roads, motorways, and railways) have major negative impacts on natural landscapes and ecosystems [3]. Korea has approximately 110,714 km of roads. Considering the area of the country (100,210 km²), this means that on average, more than one road is encountered for every kilometer traveled. This type of infrastructure development causes habitat fragmentation, which prevents animal migration and poses a threat to wildlife.

Wildlife crossing structures (WCSs)—also known as eco-bridges, eco-links, and green bridges—have been developed as a means to alleviate habitat fragmentation and have been implemented in Asia, America, Africa, Oceania, and Europe. WCSs help to protect biodiversity and improve wildlife connectivity networks, and they also help to integrate conservation efforts with land used for other purposes [4]. In Korea, Article 2 of the Natural Environment Conservation Act defines WCSs as ecological structures that have been installed to prevent the habitats of wild animals from being isolated, damaged, or destroyed by anthropogenic structures such as roads, dams, and underwater beams [5]; that is, WCSs allow the continued movement of wild animals and plants and help maintain the continuity of the ecosystem. As of 2021, Korea had a total of 532 WCSs, including 333 viaduct-type crossing structures, 181 tunnel-type crossing structures, and 18 crossing structures for the movement of amphibians. Considering the total length of roads in Korea, only one WCS exists for every 231 km of road, which is a low ratio. According to the Korea Ministry of Environment’s guidelines, the width of the tunnel should be at least D500 on two-lane roads. For species whose utilization rate increases when sunlight enters the tunnel, tunnels should be constructed in a manner that allows sunlight to penetrate the ceiling. However, most installed ecological crossing structures are not created in a manner that aligns with their original purpose. The location and size of WCSs are determined using customary methods and field surveys without ecological considerations based on the target species. Therefore, crossing structures are largely neglected by wild animals [6,7]. In addition, these WCSs are often of suboptimal quality.

WCSs are especially emphasized for large mammals because these are typically charismatic species whose involvement in accidents can lead to the loss of human life [7]. However, there is little interest in developing WCSs for amphibians, even though amphibians comprise the highest number of roadkills overall [8]. Owing to their activity patterns, population structure, and habitat preferences, amphibians are strongly affected by traffic intensity and road density and are one of the groups most frequently involved in roadkills [9]. Specifically, amphibian roadkills account for 60–90% of all roadkill observations [10,11]. Nevertheless, only 18 WCSs aimed at amphibians have been installed in Korea, and the locations of only 14 of these have been identified to date. This implies that even the authorities are not aware of all WCSs constructed for amphibians. Fourteen WCSs for amphibians with confirmed locations have been installed on highways. This implies that they were installed at locations that are difficult for people to access. Therefore, regular monitoring and maintenance were not conducted. Even in Japan, where WCSs were introduced earlier than that in Korea, most WCSs are underpasses, box culverts, and pipe culverts targeted at large and mid-sized mammals, and there are fewer eco-bridges targeting arboreal mammals or tunnels targeting reptiles or amphibians [12]. Amphibian populations have been declining worldwide because habitat fragmentation and roads have the potential to strongly impact amphibian population dynamics. Indeed, increasing evidence suggests that a considerable amount of amphibian mortality is associated with roads [13]. Glista et al. [14] also mentioned that one potential contributor to the global amphibian decline is mortality due to roadkill. However, the measures implemented for their protection are largely inadequate.

In this study, we aimed to design a WCS that protects amphibians and promotes their survival based on the specific characteristics of a given species. We focused on the order Anura, which includes amphibians that can be legally captured and experimented upon in Korea. First, we identified the factors affecting the migration of target species, which should be considered to develop adequate WCSs. We performed behavioral experiments with each species to elucidate their migration preferences. We expected the target species to exhibit preferences for specific conditions, thereby enabling tunnel design that considered species’ characteristics. Our study provides the baseline data for the installation of migratory crossing structures in the future. Our results may also contribute to the survival of anurans, which are largely excluded from protection schemes and are an overlooked taxon compared with large mammals.

2. Materials and Methods

2.1. Species Selection

We evaluated a total of 12 species of Anura: Bufo gargarizans, Bufo stejnegeri, Bombina orientalis, Kaloula borealis, Rana huanrenensis, Rana uenoi, Rana coreana, Rana nigromaculata, Rana chosenica, Rana rugosa, Hyla japonica, and Hyla suweonensis. Finally, three species with a wide home range, high protection value, and high roadkill numbers were selected: Bufo gargarizans, Bombina orientalis, and Rana uenoi.

The home range was evaluated by referring to the study by Cho [15]. It was divided into four categories: very wide (0.4 points), wide (0.3 points), narrow (0.2 points), and very narrow (0.1 points). Species with a wider home range are likelier to suffer a decline in their numbers owing to migration and roadkills and were, therefore, assigned a score of 0.4. Species with different locations of habitat, spawning, and hibernation sites were given high scores. In contrast, species living in a single area, such as an inner forest area, a rice paddy area, and a wetland area, were given a low score.

The higher the protection value is, the higher the need is for the priority installation of the crossing structures. The IUCN (International Union for Conservation of Nature) classifies Korean amphibians into nine official categories, from extinct to data deficient. The protection value was classified into high (0.3 points), medium (0.2 points), and low (0.1 points) by referring to the criteria of the IUCN: endangered and vulnerable species were assigned a protection value of 0.3, near-threatened species and species of least concern were assigned a protection value of 0.2, and data-deficient species were assigned a protection value of 0.1.

Roadkill scores were given to each species by referring to the results of the 2006–2012 report by the Korea National Park Research Institute. Roadkill numbers were divided into high (0.3 points), medium (0.2 points), and low (0.1 points): species representing >4% of the total number of roadkills were assigned a score of 0.3, species representing <4% of the total number of roadkills were assigned a score of 0.2, and species that were very rarely recorded as roadkills or were never recorded were assigned a score of 0.1. Most roadkills occurred between March and May and intensively in April. This was consistent with the spawning season, when amphibians are generally active.

These three items were comprehensively evaluated to select the target species, and the scores were weighted at 40% (home range), 30% (protection value), and 30% (roadkill). As shown in

Table 1. Total evaluation score for selecting target species along with its score for behavioral studies.

Category/Species	Home Range (40%)		Protection Value (30%)		Road Kill (30%)		Total
Bufo gargarizans	very wide	0.4	middle	0.2	high	0.3	0.9
Bufo stejnegeri	wide	0.3	middle	0.2	high	0.3	0.8
Bombina orientalis	wide	0.3	middle	0.2	high	0.3	0.8
Kaloula borealis	wide	0.3	high	0.3	low	0.1	0.7
Rana huanrenensis	wide	0.3	middle	0.2	middle	0.2	0.7
Rana uenoi	very wide	0.4	middle	0.2	high	0.3	0.9
Rana coreana	wide	0.3	middle	0.2	middle	0.2	0.7
Rana nigromaculata	narrow	0.2	middle	0.2	middle	0.2	0.6
Rana chosenica	narrow	0.2	high	0.3	low	0.1	0.6
Rana rugosa	wide	0.3	middle	0.2	low	0.1	0.6
Hyla japonica	narrow	0.2	middle	0.2	middle	0.2	0.6
Hyla suweonensis	narrow	0.2	high	0.3	low	0.1	0.6

, the scores were totaled, and among them, species with a score of 0.8 or more were selected. The total score for Bufo stejnegeri was 0.8; however, it was excluded from the experiment because it is not widely distributed in Korea. Additionally, as it hibernates underwater, its living range is limited to water. Therefore, Bufo gargarizans (Asiatic toad), Bombina orientalis (Oriental fire-bellied toad), and Rana uenoi (Dybowski’s brown frog) were selected.

2.2. Animal Capture and Housing

We obtained the relevant permissions from the appropriate national institution to collect individuals of several species for use in academic research (Permission Number: 484000085201900004). Individuals of the target species were captured throughout the rainy season (corresponding to the spawning season), when they were most active (March–May). The target species were captured in a forest near a reservoir and from around the wetland located at the edge of the forest area, where individuals of these species are frequently observed during the rainy season.

We housed them for three days till the day of the experiment. In accordance with the mating behavior of anurans, their activity was lowered during storage at low temperatures (<10 ℃, in which mating behavior is halted to prevent gamete loss). To maintain a water temperature of less than 10 ℃, it was stored in a low-temperature tank. During transportation to the experiment site, the temperature was maintained using cooler bags and ice packs as transport boxes. After moving, the water temperature was gradually raised to increase activity. In addition, we provided a separate room for male and female species to reduce the fatality rate owing to mating. For poisonous anurans (Bufo gargarizans), the water was changed at regular intervals (twice a day) during storage. The experiments were designed based on a sample size of 50 individuals per species; however, a total of 55 individuals per species were captured and stored to account for any unpredictable loss (Figure 1). All species were released to their original location after the experiment.

2.3. Structure Design and Experimental Setting

The experiments were conducted at Namgye-ri, town (Jeollabuk-do, South Korea). The surrounding area consists of farmland adjacent to mountainous areas (total area, approximately 2300 m²; Figure 2), and the site was selected based on its high suitability as an anuran habitat. To prevent errors in the migration experiment owing to visual or odor cues, the migratory crossing structure was buried in the ground.

The environmental factors that can potentially influence anuran migration were selected based on a literature review. As a few studies were limited to only anurans, a more comprehensive range of studies was reviewed. Woltz et al. [16] analyzed the migration preferences of amphibians using the tunnel size, floor material, tunnel length, and light transmission as variables. The results revealed that the amphibians preferred wider and shorter tunnels and tunnels where the floor material was soil. Patrick et al. [17] used the size, construction material, length, and orientation of the tunnel as variables and reported that the wider and shorter the tunnel, the higher the preference. Brehm [18] suggested that migration preferences differ based on the shape of the tunnel (square or round) and that migration frequency was higher when the crossing structure was round and larger. Podloucky [19] highlighted the importance of selecting appropriate materials to improve the maintenance of tunnels. Langton [20] reported that although a high noise level can cause animals to hesitate, it ultimately does not deter migration. Karraker and Gibbs [21] also considered the tunnel material, the presence of guide fences, and light transmission to be major environmental factors influencing the migration of amphibians. Helldin and Petrovan [22] reported that the effectiveness of tunnels could depend on several factors related to their technical construction, including construction materials, tunnel substrate, and the width, shape, and length of the tunnels. Based on the literature review, we included the following variables in our experiment: noise level, tunnel width, light-shielding level, construction material, tunnel substrate, and tunnel shape. For each of these variables that could influence anuran migration, we created a structure with different characteristics totaling six different structures (

Table 2. Features of the six structures installed to evaluate anuran mobility preferences in Jeollabuk-do, South Korea.

Environmental Factor	Conditions
Noise level	Noise standards were set at 40, 50, 60, and 70 dB by installing the sound source at the destination point of the crossing structure and playing a recording of the noises generated by vehicles. The noise standard was expanded considering the 50–60 dB of background noise generated in daily life.
Tunnel width	Crossing structures of different diameters were installed: D300 (mm), D500 (mm), D700 (mm), and D1000 (mm). Four sizes were used by referring to the standards recommended by the Korean Ministry of Environment: D500 (mm) and D1000 (mm).
Light-shielding level	A U-shaped plume pipe of the same size was installed in four directions, and a light-shielding film (35%, 50%, 75%, and 90%) was installed on the pipe. Behavioral experiments were conducted in conditions of sunlight without shade.
Tunnel substrate	PE (polyethylene) pipes of the same size (D600) were installed in four directions. The floor material inside the pipe was either soil, concrete, gravel, or PVC (polyvinyl chloride).
Construction material	PVC and concrete were used based on their improved performance and to facilitate tunnel maintenance.
Tunnel shape	Circular and square crossing structures were arranged horizontally, and the same tunnel width (D600) was maintained.

and Figure 3).

To confirm the mobility of the selected target species, the experiment was repeated 3 times using 50 individuals of each species. The experiments were repeated to increase reliability. An experiment was conducted by placing 50 individuals of each species at once at the inlet, the center of the testing ground. On the first occasion, all movement routes were blocked to provide 30 min to adapt to the environment and acclimatize at the center of the testing ground. Subsequently, the partition was removed to observe the movement pattern. This allowed each species to select a migration path, and mobility was confirmed by installing different structures. When experimenting with each environmental factor related to mobility, the other factors were controlled to avoid confounding effects. The animals’ preferences were evaluated by counting the number of migrating individuals in each migration experiment (Figure 4). We only counted the number of anurans that moved from one side of the tunnel to another side, because some anurans did not move. When the experiment was repeated, the next experiment was conducted at intervals of approximately one hour after the completion previous experiment.

2.4. Data Analysis

Tests of independence and chi-square tests were performed using SPSS statistical software. The test of independence (chi-square test) was used to verify whether the results of three repeated experiments using each species showed consistent trends. Unlike many other non-parametric and some parametric statistics, the calculations needed to compute the chi-square provide considerable information about how each of the groups performed in the study. Additionally, this richness of detail allows the researcher to understand the results and thus to derive more detailed information from this statistic than from many others [23]. The data that did not show similar patterns in the three repeated experiments per species were excluded from the analysis. A goodness-of-fit test (chi-square test) was used to verify whether any species showed a biased distribution (preference) between the factors influencing migration. In addition, Fisher’s exact test was used because when performing the chi-square test, the statistic test may not follow a chi-square distribution if there is at least one value with an expected frequency of <5. For the goodness-of-fit test, we used the data that showed significance (p < 0.05) in the independence tests, indicating a biased distribution of migration frequency owing to the influence of various environmental factors.

3. Results

A total of 50 individuals of each target species were included in 3 experiments with a total of 6 environmental factors, and their preferences were evaluated based on migration frequencies (

Table 3. Number of individuals that crossed the different tunnel structures in Jeollabuk-do, South Korea.

Environmental Factor		Experimental Species			Total
		Bombina orientalis	Bufo gargarizans	Rana uenoi
Noise	40 dB	19	34	23	76 (19.79%)
	50 dB	29	36	25	90 (23.44%)
	60 dB	45	43	39	127 (33.07%)
	70 dB	37	35	19	91 (23.70%)
Tunnel width	D1000	16	72	38	126 (31.82%)
	D700	50	20	53	123 (31.06%)
	D500	23	26	34	83 (20.96%)
	D300	21	32	11	64 (16.16%)
Light transmission	90%	33	41	57	131 (31.19%)
	75%	19	38	51	108 (25.71%)
	50%	25	39	25	89 (21.19%)
	35%	43	32	17	92 (21.90%)
Tunnel substrate	Soil	35	41	65	141 (34.22%)
	Concrete	27	46	24	97 (23.54%)
	Gravel	33	36	23	92 (22.33%)
	PVC	24	29	29	82 (19.90%)
Construction material	PVC	51	97	86	234 (55.71%)
	Concrete	75	53	58	186 (44.29%)
Tunnel shape	Circle	89	61	79	229 (53.88%)
	Rectangle	52	89	55	196 (46.12%)

). Across species, the animals showed the greatest preference for a noise level of 60 dB, tunnel diameter of D1000, light shielding of 90%, soil as substrate, PVC as construction material, and circular tunnels.

The independence test for noise levels (

Table 4. Test of independence for responses to noise levels from the behavioral experiment aiming to evaluate anuran mobility preferences in Jeollabuk-do, South Korea.

Species	Noise					Chi-Square (p-Value)
	Experiment	40 dB	50 dB	60 dB	70 dB
Bombina orientalis	First	4	9	18	10	7.646 (0.265)
	Second	6	11	13	7
	Third	9	9	13	19
Bufo gargarizans	First	12	16	12	10	3.489 (0.745)
	Second	12	10	14	13
	Third	10	10	17	12
Rana uenoi	First	5	9	15	4	13.336 (0.038)
	Second	3	9	15	6
	Third	15	7	9	9

) revealed that only Rana uenoi showed the same patterns of migration behavior in response to noise in all three experiments (p = 0.038), whereas Bombina orientalis and Bufo gargarizans did not.

Bombina orientalis (p = 0.010) responded similarly to tunnel width across experiments (

Table 5. Test of independence for responses to tunnel width from the behavioral experiment aiming to evaluate anuran mobility preferences in Jeollabuk-do, South Korea.

Division	Tunnel Width					Chi-Square (p-Value)
	Experiment	D1000	D700	D500	D300
Bombina orientalis	First	2	20	5	8	Fisher (0.010)
	Second	5	8	12	2
	Third	9	22	6	11
Bufo gargarizans	First	28	4	6	12	4.464 (0.614)
	Second	23	7	10	10
	Third	21	9	10	10
Rana uenoi	First	14	19	13	4	Fisher (0.479)
	Second	14	23	9	2
	Third	10	11	12	5

), whereas Bufo gargarizans and Rana uenoi did not.

Bombina orientalis (p < 0.001) and Bufo gargarizans (p = 0.009) responded similarly to light transmission levels across experiments (

Table 6. Test of independence for responses to light transmission level from the behavioral experiment aiming to evaluate anuran mobility preferences in Jeollabuk-do, South Korea.

Species	Light Transmission					Chi-Square (p-Value)
	Experiment	90%	75%	50%	35%
Bombina orientalis	First	16	4	8	5	34.076 (<0.001)
	Second	11	2	12	8
	Third	6	11	5	28
Bufo gargarizans	First	16	15	13	6	16.928 (0.009)
	Second	17	15	6	12
	Third	8	8	20	14
Rana uenoi	First	19	18	8	5	5.653 (0.463)
	Second	18	21	5	6
	Third	20	12	12	6

), whereas Rana uenoi did not.

Bombina orientalis (p < 0.001) and Rana uenoi (p < 0.001) responded similarly to tunnel substrate across experiments (

Table 7. Test of independence for responses to tunnel substrate from the behavioral experiment aiming to evaluate anuran mobility preferences in Jeollabuk-do, South Korea.

Species		Tunnel Substrate				Chi-Square (p-Value)
	Experiment	Soil	Concrete	Gravel	Rice Straw
Bombina orientalis	First	3	9	15	6	31.089 (<0.001)
	Second	4	9	13	7
	Third	27	8	5	10
Bufo gargarizans	First	18	15	8	9	9.336 (0.155)
	Second	14	10	14	12
	Third	9	20	13	8
Rana uenoi	First	34	10	3	3	24.447 (<0.001)
	Second	19	8	8	15
	Third	12	6	12	11

), whereas Bufo gargarizans did not.

Rana uenoi (p = 0.045) responded similarly to the tunnel’s construction materials across experiments (

Table 8. Test of independence for responses to tunnel construction material.

Species	Construction Material			Chi-Square (p-Value)
	Experiment	Concrete	PE Pipe
Bombina orientalis	First	25	15	0.300 (0.860)
	Second	21	14
	Third	29	21
Bufo gargarizans	First	20	30	0.758 (0.684)
	Second	17	33
	Third	16	34
Rana uenoi	First	23	27	6.189 (0.045)
	Second	24	26
	Third	11	33

). In contrast, Bombina orientalis and Bufo gargarizans did not show similar patterns for specific materials across experiments.

Rana uenoi (p = 0.008) responded similarly to tunnel shape across experiments, whereas Bombina orientalis and Bufo gargarizans did not (

Table 9. Test of independence for responses to tunnel shape from the behavioral experiment aiming to evaluate anuran mobility preferences in Jeollabuk-do, South Korea.

Species	Tunnel Shape			Chi-Square (p-Value)
	Experiment	Circle	Square
Bombina orientalis	First	30	13	4.184 (0.123)
	Second	30	13
	Third	27	23
Bufo gargarizans	First	18	32	1.049 (0.591)
	Second	23	27
	Third	20	30
Rana uenoi	First	32	18	9.480 (0.008)
	Second	29	11
	Third	18	26

).

Table 10. Goodness-of-fit test from the behavioral experiment aiming to evaluate anuran mobility preferences in Jeollabuk-do, South Korea.

Division		Species
		Bombina orientalis	Bufo gargarizans	Rana uenoi
Noise	40 dB	19	34	23
	50 dB	29	36	25
	60 dB	45	43	39
	70 dB	37	35	19
Chi-square (p-value)		11.415 (0.009)	1.351 (0.717)	8.566 (0.035)
Width	D1000	16	72	38
	D700	50	20	53
	D500	23	26	34
	D300	21	32	11
Chi-square (p-value)		25.491 (<0.001)	44.24 (<0.001)	26.647 (<0.001)
Light transmission	90%	33	41	57
	75%	19	38	51
	50%	25	39	25
	35%	43	32	17
Chi-square (p-value)		10.8 (0.012)	1.2 (0.753)	30.373 (<0.001)
Tunnel substrate	Soil	35	41	65
	Concrete	27	46	24
	Gravel	33	36	23
	PVC	24	29	29
Chi-square (p-value)		2.647 (0.449)	4.157 (0.244)	34.064 (<0.001)
Construction material	PVC	51	97	86
	Concrete	75	53	58
Chi-square (p-value)		4.571 (0.032)	12.907 (<0.001)	5.444 (0.019)
Tunnel shape	Circle	89	61	79
	Rectangle	52	89	55
Chi-square (p-value)		9.709 (0.001)	5.226 (0.022)	4.298 (0.038)

shows the results of the goodness-of-fit test based on the results of the behavioral tests. For this analysis, we only utilized the data that showed significant results in the independence tests. In terms of noise, Rana uenoi showed a significant preference (p = 0.035) for a specific noise decibel (60 dB). Bombina orientalis (p < 0.001) showed a preference for a tunnel width of D700. The behavioral results of tests for both Bombina orientalis and Bufo gargarizans were significant in terms of light transmission. However, in the goodness-of-fit test, only Bombina orientalis showed a significant preference (p = 0.012) for a light-shielding level of 35% (the brightest environment). In the conformity test, Rana uenoi showed statistically significant preferences for soil as the tunnel substrate (p < 0.001), PVC as the construction material (p = 0.019), and a circular tunnel shape (p = 0.038). In addition, Bombina orientalis and Bufo gargarizans also showed preferences for PVC as the construction material and for a circular tunnel. Overall, the three target species responded differently to each environmental factor. Nevertheless, all six factors expected to affect anuran migration were found to have a significant impact on the migration preferences of one or more species.

4. Discussion and Conclusions

This is the first study in Korea to assess anuran responses to different WCSs and can help future conservationists implement more effective mitigation measures for this group, which has received limited attention. Establishing migratory crossing structures for these target species is of great importance, as they inhabit a wide home range, represent a high proportion of roadkills, and have high protection value [15].

We evaluated the migration preference of anurans by designing and constructing an experimental site. The type of data that we obtained through the experiment was frequency data, and it was necessary to determine whether the number of those data was statistically significant. A test of independence was conducted to find out whether the numbers obtained from the three repeated experiments showed the same distribution. In addition, a goodness-of-fit test was conducted to determine whether anurans preferred a specific environment. Both analyses were based on the chi-square test, and all the data that we wanted to analyze were obtained through the tests.

Based on our results, we proposed a design for anuran WCSs. First, the tunnel should be circular. Using PVC rather than concrete as the material and using soil as the substrate will be advantageous for anuran migration. According to Hirai’s experiment [22], pipes made of PVC material can be used as a shelter for anurans. He installed PVC pipes in the banks of rice fields. As a result, most frogs were concentrated near the pipe refuges. Few anurans were found on the banks where pipes were not installed. A tunnel width of D700 would be most appropriate. This size is not presently included in the standard recommended by the Korean Ministry of Environment; therefore, we suggest that this aspect be considered in the future. A noise level of 60 dB, which is higher than the noise level in their natural habitat, was most preferred in our experiments. A study by Nakano et al. [24] also demonstrated through experiments that the average noise level of the habitat of the anurans was 63 dB. This result is similar to that of another study conducted by Ishikuro and Iwai [25]. The authors of that study investigated the effect of noise on the choice of the best breeding site for anurans; however, their results suggested that noise did not have a significant effect. According to the Noise and Vibration Control Act of the Korean Government, noise in green areas and natural environmental protection areas should be maintained at 60 to 65 dB or less. Normal daily noise falls within this range. Our results suggest that daily noise, with the exception of loud noises, does not appear to generally affect the movement of anurans. The anurans in this study showed mobility even in a bright environment, which is contrary to conventional knowledge. Conventionally, anurans are expected to prefer dark environments, possibly because excessive light can desiccate their skin and interfere with skin respiration. However, our results suggest that if humidity is maintained above a certain level, anuran migration can be induced even in a bright environment. Therefore, crossing structures with a low light-blocking rate may be suitable at sites with sufficient humidity.

We proposed a single design for anurans because our experiment was conducted with only three species found in Korea. However, species characteristics may vary based on the country or region, and the design will have to be modified accordingly. The diverse behavioral characteristics of each species make it difficult to design a single crossing structure that incorporates every preference of each species. Therefore, different types of migratory crossing structures should be created by accounting for these factors. It is also advisable to investigate the environment of the site at which the WCS is to be installed and create a differentiated WCS for each site based on the predominant species that inhabit the region.

This study has several limitations. First, only three were included in the behavioral experiments. Behavioral characteristics and preferences can differ between species, and future studies should conduct parallel experiments on other target species. Second, environmental factors such as the season and humidity are other important factors affecting migration. Therefore, these factors should be included in future studies and weighted accordingly.