As of 2021, there are over 6.7 million km of highways in the United States, with over 2 million km in urban areas and over 4.7 million km in rural areas (Federal Highway Administration, 2021). Although roads cover only about 1% of the total land area in the United States, their impact is estimated to affect more than 19% of the country’s total land (Forman, 2000). Moreover, roads pose a serious risk of wildlife-vehicle collisions (WVCs), especially when large mammals gain access to the roadway. Every year, over two million WVCs occur in the United States, involving large mammals, resulting in 200 human deaths and 29,000 injuries (Huijser et al., 2009; Allen et al., 2013). WVCs is estimated at over $8 billion US annually, including vehicle repair costs, carcass removal, human injuries, and accident investigations (Huijser et al., 2009). Efforts to minimize these accidents have led to the development of several mitigation measures, including overpass and underpass wildlife crossing structures (WCSs; Lesbarrères and Fahrig, 2012; Soanes et al., 2024), exclusionary or guidance fencing (Roy et al., 2024; Maharjan et al., 2025), and cattle guards (Belant et al., 1998; Allen et al., 2013; Roy et al., 2024).

The impact of roads is further exacerbated when they intersect protected habitats of endangered wildlife species. State Highway (SH) 100 in south Texas is one of such highways that passes through the Laguna Atascosa National Wildlife Refuge (LANWR), a habitat for endangered ocelots (Leopardus pardalis, Jackson et al., 2005). Between 2010 and 2024, three ocelot road mortalities were recorded on SH 100 due to vehicle collisions (Environmental Affairs Division, 2015). To prevent future ocelot road mortality, the Texas Department of Transportation (TxDOT) constructed five wildlife crossing structures (underpasses), 16 modified cattle guards (also referred to as wildlife guards; Allen et al., 2013; Roy et al., 2024), 10 wildlife exits (Sheikh et al., 2023), and continuous fencing over a distance of 11.9 km stretching between Laguna Vista and Los Fresnos (Sheikh et al., 2023).

Overpasses and underpasses allow animals to cross over (i.e., above) safely or under (i.e., below) roadways and structures have been constructed for a wide range of species, including the endangered Florida panther (Puma concolor) in Florida (Foster and Humphrey, 1995), red crabs (Gecarcoidea natalis) on Christmas Island (Muller and Misso, 2015), bobcats (Lynx rufus) in Montana (Huijser et al., 2011), and lemurs (Lemuroidea) in the Toamasina province of Madagascar (Mass et al., 2011). With the increase in roads, the demand for mitigation measures is likely to become more prevalent. It may require integrating of these measures for project approvals when roads traverse biologically diverse or protected habitats (Meijer et al., 2018; Maharjan et al., 2024). Additionally, fencing is an effective mitigation measure capable of excluding animals from entering the roadway along the right-of-way (ROW) and guiding them to WCSs. When combined, fencing and WCS have been shown to reduce WVC with large mammals by over 80% in central Arizona (Dodd et al., 2007) and US Highway 93 in Montana (Allen et al., 2013). In urban or developed areas, fence gaps are sometimes necessary to allow vehicles and people to access land. Without WCS along the fenced roadways, animals tend to get funneled towards those gaps, which may allow them to access the roadway (Allen et al., 2013). While vehicle gates can be an alternative solution to prevent animals from accessing the road at fence gaps, it is not always feasible to use gates, especially in areas with high traffic volumes or at road intersections (U.S. Department of Transportation, Federal Highway Administration, 2008). In this scenario, modified cattle guards used by ranchers to handle cattle have proven effective in discouraging animals from using fence gaps to access the ROW (Peterson et al., 2003).

Ranchers have utilized standard cattle guards combined with fencing for decades to prevent cattle from accessing unwanted areas; however, traditional designs are not always safe and effective for heavy emergency vehicles or pedestrians (Allen et al., 2013) and do not account for the jumping or leaping capabilities of wild animals. Therefore, researchers and the state Departments of Transportation have developed alternative designs for wildlife, termed wildlife guards (Allen et al., 2013; Roy et al., 2024). Sebesta et al. (2003) found that the effectiveness of wildlife guards depends on their position and placement. Guards placed over a pit or any raised height using concrete structures can effectively deter wildlife by discouraging them from placing their legs between the grates (Allen, 2011; Sebesta et al., 2003). Peterson et al. (2003) compared three different wildlife guard grate patterns (10.1× 12.7 cm with diagonal cross opening, 10.1× 7.6 cm without diagonal design, and 7.6 × 10.1 cm without diagonal opening) on Florida Key deer (Odocoileus virginianus clavium) and found varying repel rates. Allen et al. (2013) indicated that changes in the dimension and design of wildlife guards can create an effective barrier for larger mammalian deer species. While ranchers and farmers have long relied on cattle guards, the widespread adoption of wildlife guards in road ecology remains elusive, rendering research in this area profoundly impactful. We focused on evaluating the effectiveness of two wildlife guard designs (bridge grate and pipe) in repelling meso-carnivores and ungulates found in south Texas. Our evaluation addressed three key research questions: (1) Do wildlife guards effectively deter animals from accessing the right of way (ROW)? We assessed this by determining the percentage of wildlife breaches relative to the number of wildlife interactions or movements near the wildlife guards; (2) Which wildlife guard design is more effective in repelling animals? We compared the repel percentage across two designs; and (3) Which wildlife categories are predominantly deterred by wildlife guards? To ascertain this, we compared the repel percentage among various target species (meso-carnivores: bobcat – Lynx rufus, coyote - Canis latrans, striped skunk – Mephitis mephitis, northern raccoon – Procyon lotor, and ungulates: white-tailed deer – Odocoileus virginianus, javelina – Dicotyles tajacu, wild pig – Sus scrofa, nilgai – Boselaphus tragocamelus) across the two wildlife guard designs. Although the primary purpose of the guard installation was to prevent ocelots from using the fence gaps, no ocelots were recorded during our study period or at our study location. Therefore, this species was not included in the species categories for analysis.

Our results illustrate that wildlife guards are effective in preventing wild ungulates ( x¯ ± SE = 82.15 ± 0.20%) from entering the ROW compared to meso-carnivores ( x¯ ± SE = 16.64 ± 0.09%). We also show that PWG was more effective at repelling ungulates (86.79%) and meso-carnivores (18.83%) than BGWG (ungulates 81.51% and meso-carnivores 15.81%). Overall, we documented a total of 10,445 interactions (breached and repelled) across 14 wildlife guards (six PWG and eight BGWGs) for eight wildlife species (four meso-carnivores and four ungulates). The total number of interactions were higher in BGWGs (breached = 774 per BGWG sites, repelled = 183 per BGWG sites) compared to PWGs (breached = 372 per PWG sites, repelled = 94 per PWG sites). Among meso-carnivores, the repelled percentage was less than 36% for each species within this subset for both WG designs, while more than 50% of wildlife interactions for ungulate species were repelled by both designs (Table 1).

During all post-construction monitoring (2020-2024), the mean repelled percentage of target species between the two different WG designs was statistically significant (P = 0.049; Table 2). The mean repelled percentage was significantly higher in PWGs (20.12 ± 0.17) compared to BGWGs (19.11 ± 0.10; Figure 2A). Similarly, species categories revealed that the mean repelled percentage were significantly different (P<0.001) with a higher repelled percentage of ungulates (82.15 ± 0.20) compared to meso-carnivores (16.64 ± 0.09; Figure 2A). Although the mean repelled percentage for ungulates were much higher in both BGWGs (81.51 ± 0.22) and PWGs (86.79 ± 0.50) compared to meso-carnivores (BGWGs: 15.81 ± 0.11 and PWGs: 18.83 ± 0.17), our statistical analysis showed insignificant interactions between WG type and species category.

Additionally, the repel interaction between species category and study year was found to be statistically significant with a repel percentage for ungulate (ranging from 68.85 ± 0.72 to 91.73 ± 0.25) being significantly higher than meso-carnivores (14.50 ± 0.18 to 19.58 ± 0.26) across all five-study years (Figure 2B). No multicollinearity was observed due to the categorical nature of all variables, and no overdispersion was detected. The repel rate varied among ungulates and meso-carnivores across different study periods (P<0.001; Table 2), with higher percentage of repelled individuals noted after 2020 (Figure 2B).

Comparative studies between the two wildlife guard designs are limited in the published literature, with most focusing only on BGWG (Allen et al., 2013; Peterson et al., 2003) or PWG designs (Belant et al., 1998; Reed et al., 1974). Our findings highlight the importance of considering the specific wildlife guard designs when evaluating their effectiveness. Additionally, our study examined the repel interactions of wildlife guard designs based on species category (ungulates and meso-carnivores). These interactions were found to be statistically insignificant, except for the interaction between study years and species category (Table 2). One of the reasons for the global model being insignificant can be attributed to our limited sample size (n= 14). It might have limited the statistical power and increased the risk of failure for detecting an actual effect in our datasets. In addition, our study did not use a paired or a blocked design; however, the locations were selected to be similar in terms of road characteristics, surrounding land use, and fencing. This allowed for a practical comparison of WG performance while minimizing major environmental variability. Although statistically significant, many of our differences appear practically insignificant. These types of patterns are most common in ecological datasets, where the effects of interactions might not meet the statistically significant threshold due to variability in the nature or limited sample size (Wei et al., 2012).

The post-hoc test done for two-way interaction between study years and species category showed the significant difference between meso-carnivores and ungulates across all five-study years, although there was not much significant difference within species category (Figure 2B). Moreover, the repel percentage for each species category increased over time, which might be due to increased human activity near the WG sites, leading to greater avoidance on those sites (Webb et al., 2011). Although our GLM model didn’t provide a statistically significant result between the interactions of WG designs with species category, a higher repel percentage was observed for ungulate in both types of WG designs compared to meso-carnivores (Figure 2A). A similar observation was made regarding the interaction between gate design and species category, although we did not analyze the interactions specifically for WGs with gates in our study. In those case, the repel percentage for both ungulates and meso-carnivores were higher at WG sites with gates compared to those without gates.

The variation in the mean repelled percentage between the two different designs can be attributed to the complexity of their structural design patterns suggesting that the design of WGs may have a significant impact on their effectiveness in repelling target species. The PWG design, featuring a wider opening for foot placement, may have presented challenges for ungulates to walk on the horizontally laid cylindrical steel pipes compared to the bridge grate design, which has grating material positioned diagonally at a 45° angle. Belant et al. (1998) reported results similar to our PWG, where double cattle guards with round bars were more than 85% effective in repelling large ungulates like white-tailed deer. Allen et al. (2013) found the effectiveness of BGWG to be only 60% for white-tailed deer, while our study found it to exceed 85%. Furthermore, Allen et al. (2013) indicated that the design was less effective for carnivores like coyotes (33-55%), whereas our study observed a lower repel rate (coyote-13.03%). Peterson et al. (2003) also tested the effectiveness of a bridge grate design for large ungulates, such as Florida key deer, with findings comparable to ours. However, a study by Reed et al. (1974), which involved flat mill steel rails (1.3 cm wide) installed parallel with 10.2 cm spacing, found the design ineffective for mule deer (Odocoileus hemionus), as more than 85% began crossing it after being released, fearing they might be captured again. This finding highlights the importance of design specificity in WGs and suggests that species-specific behavioral response must also be considered to ensure its effectiveness.

The effectiveness of the wildlife guard depends on the morphological and motivational characteristics of the targeted wildlife species. Our research showed that the effectiveness of both wildlife guard designs was lower for meso-carnivores (with paws) than ungulates (with hooves), as soft-footed animals are more agile and can navigate through the guards more easily. Although the PWG design was relatively more effective than the BGWG for meso-carnivores, species such as bobcats and coyotes exhibited hesitancy before crossing, ultimately using the guards to access the roadway. This result suggests the need for additional design features, such as frightening tools like artificial lights (Belant et al., 1998; Koehler et al., 1990), sound (Honda, 2019), and odor (Bíl et al., 2018) to deter wildlife from using the structures. Wildlife guards primarily designed for ungulates should consider the size and agility of the ungulates before installation. For instance, a study conducted by Reed et al. (1974) on Interstate 70 in Colorado found that elk (Cervus canadensis) and mule deer accessed the road by crossing a 3.7 m long and 10.2 cm spaced pipe-style wildlife guard by walking, without a recorded instance of them jumping across the guard. In contrast, a field study by Sebesta et al. (2003) along US Highway 1 in Florida found white-tailed deer crossing over a 3.6 m long PWG solely by jumping through it. Our wildlife guard designs were 6 m long, significantly longer than those studied by Reed et al. (1974) and Sebesta et al. (2003), due to concerns that ocelots would be able to jump over them and the necessity to exclude nilgai, a large-bodied ungulate, from the road. We found that hooved animals such as nilgai and javelina were only two ungulate species repelled 100% of the time by PWGs, likely due to the large spacing between the pipes (7.6 cm), while the other two ungulate species (white-tailed deer and wild pig) were not completely repelled. In contrast, BGWGs were less effective, with only a 55% repel for nilgai and a 70% for javelina, while the design repelled 100% of wild pig and about 86% of white-tailed deer. The narrow spacing between the mesh of the bridge grate facilitated these species in walking over the BGWG. However, there was about 14-46% difference in the repellent percentage between nilgai, javelina, white-tailed deer, and wild pigs within BGWG and 15 - 50% in PWG. When designing wildlife guards, we recommend considering the size and agility of the target animals, and the hoof size of ungulates in the area to select an appropriate surface design (e.g., bridge grate, pipe, or other material). Additionally, the length and width of the wildlife guard should also be evaluated based on the size and agility of the species to be excluded. The difference in effectiveness between two WG designs may be attributed to several factors, including overall dimensions of the guard (length and width), species-specific body size (i.e. ungulates vs meso-carnivores), and the structural configuration, particularly the spacing between mesh elements in BGWGs and between pipes in PWGs. These factors likely influence an animal’s perception of barrier risk and crossing difficulty, thus affecting their response to the guard.

The low percentage of repel for meso-carnivores using either PWG or BGWG on SH100 suggests that implementing just wildlife guards may not be cost-effective for meso-carnivore targeted road mitigation. During our study, two wildlife guards had vehicle gates added due to high numbers of bobcat crossings, raising concerns about the potential for ocelots to do the same. They were manual chain-link gates made of Geo-mesh polypropylene (GEO 55), installed to fill fence gaps along the exclusion fence. The gate measured approximately 1.7 m in height with a mesh-size of about 5.08 cm, and required a manual operation by push or pull method to open. Although data from these two sites were not included in our analysis, we observed a substantial reduction in bobcat and ungulate crossings following the installation of vehicle gates. Among the two wildlife guards with gates, one opened to private land and had an irregular closing schedule due to high traffic flow, while the other opened to a federal wildlife reserve and was kept closed at all times. Based on our observations within two wildlife guard sites, mesocarnivores, and larger ungulates could breach the WG when the gate was open but faced a significant barrier when it was closed. This suggest that WGs with closed gates could serve as effective deterrents to prevent endangered ocelots from accessing the road through fence gaps. We observed that gates were mostly effective for bobcats and other larger ungulates but not for coyotes and virginia opossums (Didelphis virginiana), as these species were very persistent in crossing the gate and utilized the gap beneath it to access the road.

Other forms of modification that may be implemented include electrification. This approach has been explored in several states, including Utah (Cramer and Flower, 2017), Montana (Huijser and Getty, 2023), and Ohio (Seamans and Helon, 2008). They have installed and tested the effectiveness of electrified wildlife guards to repel animals. The electrified guards work by delivering a non-lethal shock to animals that come into contact with it. Despite the installation and maintenance costs, electrified wildlife guards may be effective for other non-hoofed animals. Although there is not much information on the effectiveness of electrified defenses for meso-carnivores compared to ungulates, Siepel et al. (2013) found that black bears were highly deterred from accessing the road by electric mats. Safety considerations are also important, as electrified guards may pose a risk to pedestrians and cyclists. However, some designs produce electric pulses that are not harmful to humans wearing shoes, and option like using a push-button to temporarily stop the electric pulses can allow safe passage for pedestrians or cyclists (Gagnon et al., 2020).

We demonstrate that the effectiveness of wildlife guard is affected by its design types with PWG being more effective in repelling ungulates in comparison to BGWG. When designing wildlife guards, we recommend considering the size and agility of the target animals for exclusion, as well as the hoof size of the ungulates in the area; these characteristics can assist in determining an appropriate surface (e.g., bridge grate, pipe, or other design materials) and size for wildlife guard designs. Although wildlife guards in our study were primarily installed to reduce ocelot road mortality by preventing them from accessing the ROW, these structures have been found to be less effective for meso-carnivores. However, combined with vehicle gates the effectiveness of wildlife guard designs could be increased for both meso-carnivores and ungulate species. Further monitoring of additional modifications using diverse deterrent tools such as sound, lasers, and odor may provide valuable insight into improving the effectiveness of wildlife guards at repelling meso-carnivores.