The effects of urbanization on highly mobile flying vertebrates, such as bats, have been historically underestimated (Threlfall et al., 2013), in part from the assumption that they are less influenced by habitat fragmentation than are non-volant species (Gilbert, 1989). Realization of this misconception has facilitated increased research efforts on volant species (e.g., birds: Pell and Jones, 2015; bats: Lesinski, 2008; Bennett and Zurcher, 2013), ultimately showing that urbanization, and the subsequent decrease in habitat connectivity, reduction of remnant habitat and increased anthropogenic disturbance, can reduce species richness, abundance and activity in volant species including bats (see review by Russo and Ancillotto, 2014). Roads in particular, have been shown to directly influence bat mortality (Medinas et al., 2013) by posing significant barriers to movement (Abbott et al., 2012; Bennett and Zurcher, 2013) and restricting dispersal for roosting and foraging (Abbott et al., 2012). Bat density and diversity have also been shown to decrease with proximity to major roads (Berthinussen and Altringham, 2012) while the presence of vehicles themselves can restrict foraging and commuting behaviors (Kerth and Melber, 2009; Zurcher et al., 2010; Bennett et al., 2013). Direct effects such as vehicle strikes (Gaisler et al., 2009; Lesinski et al., 2011), street lighting and traffic noise (Longcore and Rich, 2004; Stone et al., 2009; Bennett and Zurcher, 2013) are also known to impact urban bat communities by altering foraging, commuting, roosting and activity times (Stone et al., 2015). For example, increased artificial light is known to dramatically reduce foraging activity and delay commuting behavior in lesser horseshoe bats (Rhinolophus hipposideros) (Stone et al., 2009), while road noise greatly impairs the passive listening abilities of greater mouse–eared bats (Myotis myotis) (Schaub et al., 2008). Vegetated fauna overpasses have the potential to mitigate these impacts for bats by reintroducing structural habitat complexity, providing safe passage across roads, as has been shown for other taxa (Bond and Jones, 2008; Mata et al., 2008; Jones and Pickvance, 2013; McGregor et al., 2015).

Fauna overpasses are gaining recognition as a key mitigation strategy for increasing or restoring landscape connectivity and permeability affected by roads (van der Ree et al., 2015). Overpasses have traditionally targeted medium to large mammals (Forman et al., 2003; Glista et al., 2009; Clevenger and Ford, 2010); however, the presence of vegetation has been found to enhance utility by non-target taxa such as birds, reptiles and bats, by establishing a continuation of the natural landscape (Bond and Jones, 2008; Pell and Jones, 2015). Fully vegetated overpasses, particularly those which closely resemble the surrounding natural habitat, have been shown to encourage diverse overpass use by non-target species (McGregor et al., 2015) including small mammals (Bond and Jones, 2008), herpetofauna (Bond and Jones, 2008; Mata et al., 2008), birds (Pell and Jones, 2015) and potentially bats (Bach and Muller-Steiss, 2005; Abbott et al., 2015). Although some studies have observed the use of underpasses by bats (Abson and Lawrence, 2003; Bach et al., 2004; Abbott et al., 2012; Berthinussen and Altringham, 2012), overpasses have only recently been suggested as a mitigation tool for potentially improving road permeability for bats (Bach and Muller-Steiss, 2005; Berthinussen and Altringham, 2012; Abbott et al., 2015). The current understanding of overpass use by bats is extremely limited (Altringham and Kerth, 2015), with only one study, to our knowledge, specifically addressing vegetated overpass use by bats (Bach and Muller-Steiss, 2005). A relatively small number of studies have documented bat interactions with over–road structures (e.g., foot bridges) or unvegetated overpasses (Abbott et al., 2012; Berthinussen and Altringham, 2012; Bhardwaj et al., 2017). These studies noted that features such as strategic location, size, connectivity of tree lines and mature vegetation encouraged use of overpasses by bats, whereas road flyovers (high level road bridges) or footbridges, did not facilitate effective road crossings for bats (Abbott et al., 2015; Altringham and Kerth, 2015).

Roadside vegetation directly affects bat behavior (Zurcher et al., 2010). For example, vegetation along road verges may facilitate commuting, while gaps in vegetation such as tree canopies can prevent the establishment of bat commuting routes (Bennett and Zurcher, 2013). Vegetation near roads can also provide protection as well as foraging habitat (Verboom and Huitema, 1997) and increased insect prey (Avila-Flores and Fenton, 2005). Considering the importance of vegetation to the movement (Russo and Ancillotto, 2014) and behavior (Bennett and Zurcher, 2013) of many bat species, it is likely that a key determinant of overpass use by bats is the presence and structure of appropriate vegetation. The Compton Road fauna overpass is a component of a diverse array of fauna crossing structures in south Brisbane, Queensland, Australia, which were constructed specifically to facilitate the movement of medium to large mammals. The overpass has since been shown to provide road permeability for a range of target and non-target taxa including mammals (Bond and Jones, 2008; Taylor and Goldingay, 2012), herpetofauna (McGregor et al., 2015) and birds (Pell and Jones, 2015). The success of this overpass for non-target species in particular, has been attributed to the established vegetation that integrates the overpass into the surrounding forest (McGregor et al., 2015; Pell and Jones, 2015), facilitating habitat continuity across Compton Road. The aim of the present study was to explore whether the Compton Road overpass, as a fully vegetated fauna crossing, facilitates road permeability for microbat species. It was expected that the road would influence bat activity and species richness, the impacts of which may be lessened where the overpass is present.

Compton Road is a major urban arterial consisting of two dual lanes, located in southern Brisbane, Queensland, Australia. The speed limit of Compton Road is 70 km/h and current traffic volume is estimated at approximately 10,000 cars daily (Brisbane City Council, 2014). The road bisects Karawatha Forest Reserve (940 ha) to the south and Kuraby Bushland (140 ha) to the north, two of Brisbane's most significant urban bushland remnants. Both reserves provide natural bushland habitat for a range of native flora and fauna, including numerous significant or threatened species (Mack, 2005; Veage and Jones, 2007). A creek line runs throughout the study area, from Kuraby, underneath Compton Road and parallel with the road within Karawatha forest. Some areas of the forest have been cleared for service roads, fire brakes and power lines (see Figure 2).

During 2004–2005, the widening of Compton Road from two to four lanes threatened to permanently isolate the remnant forests, as well as increase collisions between wildlife and vehicles (Veage and Jones, 2007). To mitigate these impacts, the road upgrade incorporated the Compton Road fauna array (27°36′53.11″S, 153°05′03.12″E) which included a vegetated overpass (Figure 1) and two underpasses. The overpass is hourglass shaped, 70 m long, 15 m wide at the midpoint and 20 m wide at the ends. Roadside exclusion fencing (2.5 m high) extends the full length of the overpass and along the length of each forest at the roadside (Bond and Jones, 2008; Jones, 2010). The natural vegetation of Karawatha and Kuraby is classified as dry eucalypt woodland with native heath understories (Veage and Jones, 2007; Jones, 2010). The vegetation structure and composition on the overpass is similar to the surrounding forests and was planted at a rate of 70 shrubs and six trees per 100 m² shortly after construction was complete (2005). This planting regime, primarily consisting of mature saplings, was maintained to facilitate the movement of macropods (Jones, 2010). Surveys undertaken in 2009 confirmed that 95% of the planted trees and shrubs had survived (Jones et al., 2011), maintaining compositional similarity with the surrounding forest.

Call analysis was completed using Kaleidoscope V3.1 and SongScope V4.1.3 (Wildlife Acoustics). Identification was based on local call libraries (south east Queensland and northern New South Wales; from Pennay et al., 2004; Churchill, 2009). Search phase calls, consisting of a series of regular sound pulses (Pennay et al., 2004), were used in this study for species identification, while feeding buzzes (a series of often erratic pulses), were used to identify active hunting behavior if occurring within a clearly identifiable call (Figure 3).

A complete call was identified by the presence of multiple (three or more), clearly identifiable pulses. Pulse characteristics such as initial sweep shape, body shape and terminal sweep shape (e.g., presence of tail) were used to distinguish between similar species (Pennay et al., 2004). Recordings were identified to species level by M.M. and K.M. independently, while calls that could not be confidently identified were analyzed by a third party for verification. All remaining unidentified calls (less than 5%) were removed from the data set. Calls from species that could not be reliably distinguished from each other, for example Scotoropens sp., were grouped together as species groups (Hourigan et al., 2010) (see Table 1).

Vocalization analyses focussed on exploring three major trends over the 7 months: microbat activity and richness between the three study areas (Karawatha, Kuraby and the overpass); whether microbat activity and richness varied at the roadside compared with the forest; and finally, whether microbat activity and richness was altered at the roadside by the presence of the overpass. SIMPER analysis (PRIMER 7, PRIMER–E Ltd.) was used to determine the similarity of species activity with distance from the roadside. SIMPER was also used to compare overpass transects to the other transects to determine the impact of vegetation on species similarity at the roadside. Species accumulation curves (SPSS 24, IBM Statistics¹; Microsoft Excel, 2016²) based on richness detected at the beginning and end of the transects were used to explore the likelihood of methodological bias, as well as variation in detectability (likelihood of detecting representative species richness), particularly between the overpass and the forest, and the roadside and the forest. The latter was also used to examine if species richness varied considerably between the roadside and the forest. QGIS 2.10.1 geoprocessing software was used to produce heat maps of total call activity using Google Earth base layers (Google Earth version 7.1.5, 2015) based on Kernel Density Estimation (using night activity as a single data point to determine mean activity over 7 months). Heat map values used call activity numbers and are shown at a radius of 15 m to reflect the maximum potential range of the EMT devices. Transect points were used to display activity within the transect in its entirety and were not treated as discrete data points for analysis, except when compared with each other, to avoid sample bias.

The cryptic nature and high mobility of the study species presented methodological challenges during monitoring. Bat call detection was the only feasible way to monitor sufficient quantities of bats; harp traps were trialed and deemed ineffective due to the lack of captures relative to what was a highly demanding procedure. However, using detectors restricted analysis and application, primarily due to the inability to identify individuals. In this respect, “activity” was explored, rather than abundance, removing any requirement to identify individuals, but was considered a reasonable basis for determining general patterns of bat richness and activity around Compton Road and the overpass. This study is limited to a single study site. Therefore, it relies on replication of transects within the area to determine patterns of activity and species richness, while replication for monitoring the overpass apex uses two monitoring points for comparison. Due to these limitations, analyses were limited to exploratory and descriptive statistical methods, with a view to helping to construct viable hypotheses that can be tested in future studies.

Nine bat species were reliably identified within the study area from 1,550 clear echolocation call recordings. Clear calls that could not reliably distinguished between species were grouped (Group 1 and Group 2; see Table 1). All nine species and both species groups were recorded in Karawatha, Kuraby and on the overpass. Call activity in the two forests were similar, with a total of 387 recorded calls in Karawatha (X¯ = 27.07 ± 23.34 SD) and 402 in Kuraby (X¯ = 28.71 ± 27.46). Call activity recorded on the overpass (apex points) was almost double (726 calls; X¯ = 51.86 ± 59.97) that of the surrounding forest (Table 1).

South east Queensland supports high microbat diversity, with at least 24 known species (Churchill, 1998), 18 of which occur around Brisbane (Churchill, 2009; Hourigan et al., 2010). Previous studies conducted in Karawatha Forest detected at least nine species (Hourigan, 2011), all of which were detected in the present study (although total species richness may be as high as 14, considering indistinguishable call groupings). The presence of microbats on the overpass was substantial, with a high level of activity compared with either forest, and a species richness equal to that found in Karawatha and Kuraby (Table 1). Comparisons of species accumulation between the three areas showed detected species richness on the overpass reaching asymptote before either of the forest areas (Figure 4). However, both forests did achieve asymptote, suggesting the use of apex point counts on the overpass did not contribute to sample bias, irrespective of using different observation methods. In comparing the overpass transects to the forest transects directly, where the overpass intersects the road had comparatively higher species richness, and high detectability (Figures 5A,B), compared to both the unvegetated roadsides and the inner forest. Although the “road” end of the overpass transects had higher species richness and greater detectability than any other roadside or forest records, neither side reached asymptote, suggesting that more species may be using the sides of the overpass, but have not yet been detected. High species detectability, equal species richness and high activity observed throughout the study suggest that diverse microbat communities were using the Compton Road overpass.

Contrary to initial predictions, call activity did not appear to vary substantially with proximity to the roadside. Detectability was similar between the cleared roadside and the forest, as was species richness, while the total activity density map shows activity hotspots that did not appear to be correlated with road presence (Figure 6a). Typical foraging behaviors were used to explore patterns of activity throughout the study area, based on differences between microbats that may not be affected by road presence, compared with those that are more likely to be susceptible. These groups form the basis for determining how foraging behaviors influence activity patterns, and although some bats such as Mormopterus sp. can alter feeding behaviors as required, these groups represent the primary feeding behavior typically displayed by each species. When the data were explored in this way, patterns of activity between bats with typically varied behaviors appear apparent (Figure 5A). Call activity along the roadside varied from activity observed in the forest, with most roadside activity dominated by open foragers (M. beccarii, C. gouldii and Group 1). Forest specialists such as Nyctophilus (Group 2) were not specifically correlated with road presence or dense forest. Instead, two hotspots of Group 2 detections were present: transect 4 in Kuraby, and on the overpass. Landscape variation is likely to have contributed to activity patterns observed during the study, a recognized aspect of bat behavior (Bennett and Zurcher, 2013), while it is possible a nearby roost may have skewed activity results. Call activity among all species appears to be positively associated with creek presence and open areas.

Current literature recognizes that roads act as barriers to bat populations (Kerth and Melber, 2009; Berthinussen and Altringham, 2012; Bennett and Zurcher, 2013) by restricting dispersal (Abbott et al., 2012) or forcing changes in behavior, such as road avoidance (Zurcher et al., 2010; Bennett and Zurcher, 2013), although the degree to which this occurs is debated. Berthinussen and Altringham (2011) observed that roads act as considerable, if not complete, barriers to microbat species, while Abbott et al. (2012) suggested that roads of similar construction to Compton Road are not absolute barriers to bat movement. How severely bat communities are affected by road barriers often depends on their morphology (Rydell, 1992; Kerth and Melber, 2009; Bhardwaj et al., 2017). Therefore, observing activity patterns between bats with different foraging behaviors may assist in determining whether, and how, Compton Road affects bat communities living in the surrounding forest. The high levels of bat activity, richness and detectability on the overpass (apex and sides) compared with the cleared roadside suggests that the overpass does provide a level of relief of road presence; however this same trend is notably present when comparing the sides of the overpass with the inner forest. One explanation for this may be the hourglass shape of the overpass creates a bottleneck, restricting bat activity to a narrow area, and therefore concentrating activity to within range of the EMTs. This particular point would benefit from a more intensive monitoring regime which may include monitoring further into each forest.

Bats are known to alter their behavior in response to the structure and availability of vegetation (Bennett and Zurcher, 2013) with suitable vegetation facilitating higher bat diversity (Abbott et al., 2012, 2015). The presence of comparatively high bat activity on the overpass, by all species and groups detected, suggests that the reason for high detectability, richness and activity on the overpass is the presence of complexity in the form of vegetation. The overpass appears to provide habitat sufficient enough to support, and indeed encourage, diverse bat species. This is evident in the comparative concentration of gleaning species activity on the overpass apex, suggesting appropriately complex and suitable habitat for forest specialists. The vegetation on the overpass may be sufficient to reduce negative aspects of the road which are known to affect bats, for example artificial light (Stone et al., 2009) and traffic noise (Schaub et al., 2008), two elements of the road that are noticeably reduced on the apex, although these were not measured in the present study. In addition, the presence of open and edge foragers suggests a planting regime which does not exclude these species, a known consequence of dense vegetation (Medinas et al., 2013). The planting regime employed on the overpass aimed to replicate the local open woodland forest, which may have encouraged overpass use by species rich microbats. To fully understand the benefit of the overpass to bats, ascertaining crossing data by tagging and tracking individual bats would be of great benefit. This would specifically determine whether individuals are using the overpass as a crossing route.

Limitations of the study area size restricted the size of the transects which, due to the potential for road effects to extend far into forest habitat, may have been influenced by the road effect zone. The limited call activity from gleaning species throughout the forest, such as long–eared Nyctophilus which are recognized as sensitive forest specialists (Lumsden et al., 2002; Threlfall et al., 2013), compared with far higher call activity from edge feeding free–tailed and sheath–tailed bats, suggests that the road effect zone may extend into the forest, beyond the length of the 75 m transects. This possible outcome is consistent with taxa-diverse studies which suggest road effect zones can extend from 100 to 1,000 m from the road (Reijnen et al., 1995; Eigenbrod et al., 2009), which includes aspects of disturbance (such as noise and light) known to affect bat activity (Schaub et al., 2008; Stone et al., 2009). The broad spread and high call activity from edge foraging species may also suggest that edge effects extend well into the forest, while the dominant presence of Molossidae species, as open-space foragers (Jung et al., 2014), along the roadside indicates substantial influence of road. These patterns suggest that Compton Road may be affecting more sensitive bat species, and is likely to present a complete barrier if this is the case. The presence of Nyctophilus/Myotis sp. activity on the overpass may provide evidence of increased habitat permeability, and most likely some level of road effect mitigation for these species. It is unlikely that Compton Road impacts open and edge foraging species to the same intensity, although it does not prevent them from being at risk of barrier effects, traffic injury or other road related impacts (Kerth and Melber, 2009).

A variety of management practices have been suggested to reduce road impacts on bat species and potentially prevent further fragmentation. Bennett and Zurcher (2013) recommend management practices that involve interlinking tree canopies across roads, as well as restoring and enhancing vegetation along roadsides. One goal of habitat restoration is to enhance landscape permeability, which is also a primary goal of overpass construction (Forman et al., 2003). Although vegetated overpasses have not traditionally been built or considered for bat communities, the overlap between what an overpass achieves, and what benefits bat communities, appears to be substantial. Bach et al. (2004) concluded that over– road crossing features are more restrictive to bat species than are under–road (typical underpasses as well as culverts) crossings, while underpasses have been suggested as a potential solution to reducing road impacts on some bat species (Abson and Lawrence, 2003; Kerth and Melber, 2009; Boonman, 2011; Abbott et al., 2012). The ecological value of the underpasses at Compton Road has not been determined; however, call activity and species richness on the overpass suggests that the overpass is a substantial asset to the facilitation of habitat continuity across Compton Road.

Current research has highlighted the necessity of maintaining remnant forests within urbanized environments for the long–term persistence of microbats (Avila-Flores and Fenton, 2005; Oprea et al., 2009; Threlfall et al., 2013). The increasing reduction and isolation of patches of remnant forest in urban areas worldwide continues to jeopardize local biodiversity, particularly for species sensitive to habitat fragmentation. Retention of tree cover and linear vegetation, in particular, are critical to the sustainable protection and long–term persistence of urban forests for use by bat communities (Threlfall et al., 2013). Additionally, the presence of overpasses and similar infrastructure that enhances and reintroduces habitat connectivity needs to be appreciated in areas that are already highly fragmented. The Compton Road fauna overpass has shown to be successful in facilitating road permeability for a wide variety of taxa (Bond and Jones, 2008; McGregor et al., 2015; Pell and Jones, 2015); this may now include microbats, with the potential for maintaining habitat connectivity and enabling crossing between Karawatha and Kuraby. Without determining whether bats are crossing the road without the aid of the overpass, it is difficult to quantify “effectiveness” (as defined in Forman et al., 2003; van der Ree et al., 2007; van der Grift and van der Ree, 2015). However, this study shows that the overpass itself is being frequented by a diverse range of microbat species. This study is the first indication that a vegetated fauna overpass can facilitate activity and provide road permeability for diverse microbats in Australia, and one of very few worldwide. It is, therefore, significant in demonstrating the potentially important role of fauna passages in facilitating the long–term persistence of microbat communities in urban landscapes.

MM devised, managed and conducted field research and wrote the majority of the final manuscript as well as facilitating the editing and submission processes. KM contributed significantly to expertise required in the field and assisted in correctly writing the technical aspects of the manuscript as well as editing. DJ supervised the field research and writing process, contributing substantially to editing, research design and execution of the project.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer CT and handling Editor declared their shared affiliation.