Camera traps are widely used to survey wildlife (Meek et al., Reference Meek, Ballard, Vernes and Fleming2015; Suzuki et al., Reference Suzuki, Thong, Tan and Iwata2017). However, theft and vandalism of cameras are frequent, significantly affecting studies both within (Hossain et al., Reference Hossain, Barlow, Barlow, Lynam, Chakma and Savini2016) and outside (Widodo et al., Reference Widodo, Imron, Sunarto and Giordano2022) protected areas. An international study revealed that theft and vandalism not only incur costs because of equipment loss (c. USD 1.48 million from 309 practitioners during 2010–2015) and theft prevention (c. USD 800,000 during 2010–2015) but also affect survey design (Meek et al., Reference Meek, Ballard, Sparkes, Robinson, Nesbitt and Fleming2018). However, wildlife surveys are more important than ever, particularly in human-dominated landscapes, if we are to establish human–wildlife coexistence despite increasing global urbanization rates (mean expansion rate of 9,687 km² per year for the past 30 years; Liu et al., Reference Liu, Huang, Xu, Li, Li and Ciais2020).

In particular, we need to understand how wildlife responds to movement barriers such as roads. Road underpasses and overpasses have been shown to mitigate the negative effects of roads on wildlife (Donaldson, Reference Donaldson2007; Teixeira et al., Reference Teixeira, Printes, Fagundes, Alonso and Kindel2013; Flatt et al., Reference Flatt, Basto, Pinto, Ortiz, Navarro and Reed2022). However, these structures are expensive and complex to build (Ascensão & Mira, Reference Ascensão and Mira2007). Bridges have the potential to act as multiple-use structures. They are usually constructed by transportation companies and government agencies to facilitate human mobility over waterbodies. These bridges could also serve as underpasses for wildlife, providing dispersal routes. However, the most widely studied road underpasses are drainage coverts (Taylor & Goldingay, Reference Taylor and Goldingay2010; Sparks & Gates, Reference Sparks and Gates2012; Denneboom et al., Reference Denneboom, Bar-Massada and Shwartz2021; Monge-Velázquez & Saenz, Reference Monge-Velázquez and Saenz2022). Studies that have included bridges have been concentrated in North America (Warnock-Juteau et al., Reference Warnock-Juteau, Bolduc, LoScerbo, Anderson, Daguet and Jaeger2022) and Australia (Goldingay et al., Reference Goldingay, Rohweder, Taylor and Parkyn2022).

The Osa Peninsula is home to the largest remnant tract of Pacific lowland wet forest in Mesoamerica (Holdridge, Reference Holdridge1967) and contains four protected areas (Fig. 1; Weissenhofer et al., Reference Weissenhofer, Huber, Weber and Gonzalez2001). The region is traversed by a network of unpaved roads and two paved highways: National Route 245 (7–8 m wide and stretching 77 km from Puerto Jimenez to Chacarita, overpassing 15 substantial rivers) and Inter-American Highway 2 (7–10 m wide and stretching 57 km from Sierpe to Rio Claro, overpassing 13 substantial rivers). A region-wide camera-trap study in 2018 verified the presence of wide-ranging megafauna species and documented 23 wild terrestrial mammal species (Vargas et al., Reference Vargas Soto, Beirne, Whitworth, Cruz Diaz, Flatt and Pillco-Huarcaya2022). The wildlife community is recovering throughout the region, with many species that were once restricted to Corcovado National Park now widespread across the landscape (Carrillo et al., Reference Carrillo, Wong and Cuarón2000; Vargas et al., Reference Vargas Soto, Beirne, Whitworth, Cruz Diaz, Flatt and Pillco-Huarcaya2022). This confirms the presence of a mammal community that could utilize and benefit from safe road crossings for further dispersal.

Fig. 1 Locations of camera traps installed in bridge underpasses, and of protected areas, in the Osa Peninsula, Costa Rica. (Readers of the printed journal are referred to the online article for a colour version of this figure.)

We installed single camera-trap monitoring stations (Bushnell Trophy Cam HD Aggressor, Bushnell, USA; set to record 13 s videos, with 30 s resting periods) underneath 17 bridges along two paved highways in the Osa Peninsula for 4 months (February–May 2019) during the dry season, before the heavy rains caused the rivers to rise, which can make the underpasses inaccessible to people and wildlife (Fig. 1). The rivers were 13.5–21.6 m wide and bridge heights 4.95–7.54 m. We installed locks and informative signs on all camera traps to reduce the potential for theft (Plate 1).

Plate 1 Camera trapping with theft protection. (a) A camera trap from the study with a metal case, lock and laminated sign. (b) Close-up of the sign in (a), which explains the project in Spanish and English, with visual representations and a contact number (blacked out).

Of the 17 camera traps, five obtained data, 11 were stolen and one broke, probably because of humidity or an electrical fault. The five working camera traps accumulated a total of 167 trap-nights (open habitat caused a high number of false triggers by quick-growing vegetation, and exposure to high temperatures and heavy rain resulted in short battery life). The camera traps detected two wild mammal species, eight bird species, one reptile species, three domestic species and people (Table 1). The two wild mammal species were habitat-generalist omnivores: the northern raccoon Procyon lotor pumilus and the white-nosed coati Nasua narica. The northern raccoon was detected under four of the bridges and was observed moving through these structures and foraging. The white-nosed coati was detected moving beneath just one of the bridges. Dogs were detected at three bridges, horses and cattle at one bridge and human activity (passing through and socializing) was detected at all five bridges from which we retrieved data.

Table 1 Summary of all domestic and wild animal species detected by camera traps at five bridges monitored by camera traps in Osa Peninsula, Costa Rica (Fig. 1).

We detected only two of the 23 wild terrestrial mammal species recorded in the region. This was surprising for two reasons: firstly, we conducted the study during the dry season, which is when the rivers are at their lowest, facilitating wildlife movements, and secondly, a similar study in the Guanacaste region of Costa Rica detected 14 mammal species using six drainage culverts (Monge-Velázquez & Saenz, Reference Monge-Velázquez and Saenz2022). Furthermore, the culverts in the Guanacaste region, just like the potential multiple-use structures surveyed in our study, were not built specifically for wildlife use, with no techniques being adopted to encourage wildlife movements (Monge-Velázquez & Saenz, Reference Monge-Velázquez and Saenz2022). The reasons for the low number of species detected in our study could be the surrounding land uses (cattle farming, agriculture and small towns) and the fact that the bridges are not established wildlife crossings. It is possible that species might risk crossing the roads in preference to using underpasses. Strategically placed fencing or tree planting, which are proven techniques for funnelling wildlife (Littlewood et al., Reference Littlewood, Rocha, Smith, Martin, Lockhart and Schoonover2020), could help promote wildlife use of these underpasses. However, the principal reason for the low number of species detected is the limited number of cameras and trap-nights because of the theft of 65% of the camera traps.

Theft was significant despite the installation of security cases, locks and signs. Even with these security measures, this study resulted in a financial loss of USD 2,970 (the total material cost of the study was USD 4,590, with USD 1,020 of this invested specifically in theft protection). There was also a cost in terms of the missed opportunity to contribute knowledge on wildlife movement and to use this to improve underpasses to facilitate their use by wildlife. To reduce camera-trap theft, some studies have installed camera traps at greater heights than usual, to avoid their detection by people, but this has resulted in a dramatic decrease in wildlife detections (Meek et al., Reference Meek, Ballard and Falzon2016). Studies that focus on nocturnal species have collected the deployed camera traps each day to limit their theft (Athreya et al., Reference Athreya, Odden, Linnell, Krishnaswamy and Karanth2013), but this is not a suitable or sustainable solution for large-scale studies assessing multiple species. However, a security post for camera traps, installed in a bollard-style housing to limit theft, has been developed and is proving successful (Meek et al., Reference Meek, Ballard, Abell, Perrie, Blackford, Jones and Fleming2022). Perhaps the best potential solution to combat camera-trap theft is the development of small and cryptic camera traps that can evade detection by people but still detect wildlife. Conservation organizations are making advances in camera-trap technology to develop improved camera traps (Meek & Pittet, Reference Meek and Pittet2012), but there is still work to be done to make these solutions accessible and scalable (Curnick et al., Reference Curnick, Davies, Duncan, Freeman, Jacoby and Shelley2022; Westworth et al., Reference Westworth, Chalmers, Fergus, Longmore, Piel and Wich2022). These improvements will limit resource losses and fill data gaps in wildlife surveys conducted in areas where the chance of theft of equipment is high.