Introduction

Various types of roads have negative impacts on faunal biodiversity. Direct impacts are due to killing of individuals, or habitat destruction (Muñoz et al. 2015). Indirectly, roads may present dispersal barriers to fauna, restrict their gene flow, or reduce their biodiversity levels through increased dust loads, altering vegetation characteristics across the landscape, and increased chemical and noise pollution by vehicles (Jackson and Fahrig 2011; Phillips et al. 2020). Roads cover relatively small proportions of total landscapes, yet their direct and indirect impacts often extend far beyond their verges (Swart et al. 2019). Overall, less mobile taxa are negatively affected by roads, while more mobile ones are more resilient (Perumal et al. 2021).

In plantation forestry landscapes, road networks provide access to various plantation compartments, and enable transport of harvested timber (Smith et al. 2007). Roads surround plantation compartments, and often extend beyond the planted area, dissecting surrounding natural vegetation patches. Primary roads are frequently used during daylight hours and are regularly maintained through scraping and compaction. Secondary roads, often only present as tracks, are used periodically to access remote localities within timber plantation mosaics and are rarely maintained (Edwards et al. 2017). Due to differences in their physical characteristics and maintenance regime, these two types of roads can vary in their impacts on biodiversity (Pickering and Norman 2017).

Most road ecology studies have focused on birds (Morelli et al. 2014) and mammals (Rytwinski and Fahrig 2015), with only a few studies providing information on invertebrate taxa (reviewed by Muñoz et al. (2015). As with most other taxa, effects of roads are more pronounced for sedentary insects compared to mobile taxa (de la Puente et al. 2008). Most invertebrate studies have focused on iconic insect groups such as butterflies (e.g., Skórka et al. 2013), bees and wasps (e.g., Fitch and Vaidya 2021), and dragonflies (e.g., Rao and Girish 2007), indicating mixed results among these aerial insect groups. Among the ground-dwelling insects, ants (e.g., Tshiguvho et al. 1999), and rare and threatened ground beetles have featured in research elucidating the ecological impacts of roads, especially in forested areas (e.g., Koivula and Vermeulen 2005). Currently, not much is known about the impacts of roads on insects in other habitat types.

Dung beetles are an important terrestrial group, contributing to overall habitat function. Dung beetles are highly abundant in regions where mammals roam the landscape, and require certain substrate (e.g., soil particle size and leaf litter), and vegetation characteristics (e.g., open vs. shaded habitats) to complete their life cycles (Nependa et al. 2021). Adults spend much of their lives on the ground or in the soil, and scour the landscape for resources, especially food and microhabitats for egg laying (Davis et al. 2008). Roads can affect dung beetle assemblages in various ways, with wingless and large tunneling species being particularly vulnerable (Hayward et al. 2010; Edwards et al. 2017).

While some studies have highlighted the impacts of roads on forest-dwelling dung beetles (e.g., Yamada et al. 2014; Edwards et al. 2017), the impact of roads on dung beetles in grasslands have rarely been studied. Grasslands often support high densities of dung beetles, due to overall high productivity and the high diversity of mega-herbivores they support, especially in Africa (Davis and Scholtz 2020; Pryke et al. 2022). However, grassland integrity is threatened by human activities such as livestock farming, agriculture, and plantation forestry (Neke and du Plessis 2004). These activities decrease landscape-scale habitat heterogeneity, and lead to long-term habitat loss and fragmentation.

Our overall aim was to assess the effects of unpaved roads on dung beetles in a plantation forestry area where grassland conservation corridors make up an ecological network among cultivated compartments. Focusing specifically on roads dissecting grassland patches, our objectives were to: (1) determine whether dung beetles are affected by maintained roads vs. unmaintained roads, (2) determine whether species richness, abundance, and species assemblage composition change with increasing distance from unpaved roads, and (3) determine whether dung beetle diversity patterns are influenced by other environmental factors impacted by roads. We hypothesise that frequently maintained unpaved roads had a greater impact on dung beetles compared to unmaintained roads and expected that species richness and abundances were low near roads, while dung beetle assemblages varied from road verges to areas farther away. We also hypothesise that roads had an indirect impact on dung beetle diversity patterns, including alteration of vegetation and substrate characteristics. Support for these hypotheses would indicate that careful management of the existing road network is required to alleviate impacts on dung beetles and other ground-dwelling insects.

Materials and methods

Study area and sampling design

The study was conducted in the northeastern part of the Maputaland-Pondoland-Albany (MPA) biodiversity hotspot, in the northern coastal area of the KwaZulu-Natal Province of South Africa (Myers et al. 2000). Natural vegetation in the area is dominated by Maputaland Wooded Grassland and Maputaland Coastal Belt vegetation types (Mucina et al. 2006). The overall landscape is characterized by maintained (primary) unpaved roads and unmaintained (secondary) tracks connecting Eucalyptus spp. plantation compartments on two neighbouring plantation forestry estates (managed by SiyaQhubeka Forestry) on the eastern portion of the study area, and iSimangaliso Wetland Park, a World Heritage Site on the western portion of the study area (Fig. 1). The boundary between the forestry estates and the protected area is unfenced, and large mammals (e.g., African elephant, African buffalo, giraffe, and wildebeest) freely roam the landscape (Pryke et al. 2022).

Fig. 1
figure 1

Eight primary unpaved roads (red) and eight secondary unpaved roads (green) were selected across the study area. Black indicates the primary road network, while grey indicates the secondary road network. Yellow indicates Eucalyptus spp. plantation compartments

We selected sixteen road sites dissecting natural grassland, distributed across the plantation forestry estates and the protected area. Eight sites were regularly maintained unpaved roads, and eight were unmanaged tracks. In the case of study sites on the plantation forestry estates, suitable study sites were those where the natural grassland corridors among forestry compartments were at least 300 m wide.

Field sampling

Field data were collected between 7 and 27 February 2022. To limit attraction to traps and obtain a true representation of the dung beetles occupying habitat at increasing distances from roads, we used baitless pitfalls. At each study site, we placed a row of 5 open pitfall traps on both sides of roads (ten traps total for each study site). Traps were placed at the road edge (0 m), and at 4 m, 16 m, 64 and 128 m away from the edge of the road. Each pitfall trap was filled with 75 ml ethylene glycol and 50 ml water as a killing and preserving agent. Pitfall traps were left in the field for 15 full days. At 5-day intervals, trap catches were removed, and killing and preserving agents were replaced. In total, 480 samples were collected across 16 study sites. All collected dung beetles were counted and identified to species level using Davis et al. (2020). Dung beetles were classified into their respective broad functional guilds (tunnellers vs. rollers) and activity period category (diurnal vs. nocturnal). The replicated data collected over the sampling period was pooled for each distance class per study site, totaling 80 samples used for analysis.

For all study sites, we conducted a vegetation survey using 10 quadrats of 4 m2 each, to estimate average vegetation height, % total vegetation cover, % grass cover, % herbaceous cover, and plant species richness. We also recorded soil substrate (clay vs. fine sand vs. coarse sand), compaction within quadrats, and elevation (m a.s.l.) at the site level. Each sampling quadrat coincided with the placement of pitfall traps. We then extracted the orientation of each road (N-S vs. E-W) using a 10 m resolution Sentinel2 spatial dataset, captured on the 9th of February 2022 (Copernicus Sentinel2 Data 2022).

Data analysis

We calculated the estimated number of species (Jackknife2, Chao2 and ICE species estimators) and constructed a species accumulation curve for all collected species, using the vegan and fossil packages for R (Vavrek 2011; Oksanen et al. 2020; R Core Team 2020). Observed species richness neared species estimations, and the species accumulation curve neared an asymptote, indicating that sampling effort was sufficient (Online Resource 1). Using the ‘Cor()’ function in R, we then tested for covariation among all predictor variables. To reduce multicollinearity, variables with correlation coefficients larger than 0.65 and smaller than − 0.65 were excluded from final analysis. These were, % total vegetation cover, road substrate, road edge substrate, and road compaction. Final models were built with sampling distance away from roads, vegetation height, % grass cover, % herbaceous cover, elevation, road type, edge compaction and road orientation as predictor variables.

Species richness and abundance were interrogated for normality, and tested for spatial autocorrelation through a Mantel test using ade4 for R (Dray and Dufour 2007).

To select the best fitting error distributions, all models were tested for overdispersion. Generalized linear mixed modelling, using site location as a random spatial variable, was used for overall dung beetle species richness (Poisson), roller species richness (negative binomial), roller abundance (negative binomial), and nocturnal species richness (negative binomial). A generalized linear modelling approach was used for overall dung beetle abundance (Poisson), tunneller species richness (negative binomial), tunneller abundance (negative binomial), diurnal species richness (negative binomial), diurnal abundance (negative binomial), and nocturnal species abundance (negative binomial). Modelling was performed using lme4 for R (Bates et al. 2015), while using road type and trap distance away from roads as explanatory variables, respectively.

We then tested for effects of other environmental variables on species richness and abundance. The set of environmental variables were tested for variance inflation, using the car package for R (Fox and Weisberg 2019) and only those variables with variance inflation factors (VIF) of < 2 were included to reduce the statistical effects of strong correlations among variables. To select the best model for each modelling scenario, we performed forward selection of explanatory variables for each response, using the glmulti package for R (Calcagno and Mazancourt 2010). Final models were built including only the selected important variables.

To determine variation in overall dung beetle assemblage composition, we tested for homogeneity of dispersion among sampling distances away from roads, implemented through the ‘betadisper’ function in vegan. Dispersion among distances were homogenous. We then performed a pairwise permutational analysis of variance (PERMANOVA) to determine differences in dung beetle assemblage structuring among the various distances away from roads, implemented through the ‘adonis2’ function in vegan. To test the effects of environmental factors on dung beetle assemblage structure, we performed a model-based analysis of multivariate abundance data in mvabund for R (Wang et al. 2022). Only variables with a VIF < 2 were considered for the multivariate component of the study. The analyses were repeated for tunnellers, rollers, diurnal species and nocturnal species individually.

Results

Dung beetle species richness and abundance

Across the 16 sampling sites, we recorded 26 dung beetle species, represented by 2414 individuals and eight dung beetle tribes (Online Resource 2). Onthophagus leroyi was the most abundant species with 424 individuals caught. Onthophagus aeriginosus, O. viridulus, and Catharsius heros had overall the lowest abundance, each with one individual trapped.

Dung beetle species richness was higher for primary unpaved roads compared to secondary tracks, but road type had no effect on dung beetle abundance (Table 1; Fig. 2A and B). Distance away from roads had no significant influence on overall species richness nor abundance (Fig. 2C and D). High overall species richness and abundance were associated with overall lower % grass cover (Fig. 2E and F). Taller vegetation led to lower species richness (Fig. 2G) but had no significant effect on abundance (Fig. 2H).

Fig. 2
figure 2

Effects of road type A and B, distance away from roads C and D, and vegetation characteristics E-H, on dung beetle species richness and abundance. Red dotted lines indicate direction of correlations. Bold horizontal lines indicate medians. For pairwise comparisons, lowercase letters indicate statistical significance. Significance levels: *: p < 0.05; **: p < 0.01; *** : p < 0.001

Investigating effects of environmental variables on the two broad functional guilds (tunnellers vs. rollers) and those species with different activity periods (diurnal vs. nocturnal), species richness of tunnellers (Fig. 3A), diurnal species (Fig. 3B) and nocturnal species (Fig. 3C) decreased with an increase in vegetation height. Although % grass cover was selected as import to rollers, the effects were not significant (Fig. 3D). Tunneller, roller, and diurnal species abundances decreased with an increase in % grass cover (Fig. 3E-G). Nocturnal species abundance was lower with relatively taller vegetation (Fig. 3H), while their abundance was also higher at 64 m away from roads compared to 0 m (z = 2.81; p = 0.04) and 4 m (z = 2.73; p = 0.49) and 16 m (z = 3.11; p = 0.02), and higher at 128 m compared to 16 m (z = 2.86; p = 0.03; Fig. 3I).

Fig. 3
figure 3

Effects of vegetation and distance away from roads on dung beetle species richness A-D, and abundance E-I, groupings. Red dotted lines indicate direction of correlations. Bold horizontal lines indicate medians. For pairwise comparisons, lowercase letters indicate statistical significance. Significance levels: *: p < 0.05; **: p < 0.01; *** : p < 0.001

Dung beetle assemblage composition

There was no significant assemblage variation among trap distances away from roads, explaining only 5% of total variation (Fig. 4). Among the other variables, only vegetation height had a significant effect on overall assemblage structure (Table 1). Vegetation height was also the only significant driver of assemblage structure for the roller, tunneller, and diurnal subsets of dung beetle species. In the case of the nocturnal dung beetle species subset, assemblage structure was significantly influenced by road orientation.

Fig. 4
figure 4

Constrained ordination results indicating dung beetle assemblage composition similarity among trapping distances away from roads. Colored circles represent model standard errors

Table 1 Test statistics of road type, sampling distance away from roads, and important environmental variables relative to dung beetle species richness and abundance

Discussion

We investigated the effects of unpaved roads on dung beetles in a plantation forestry-natural grassland mosaic in the MPA biodiversity hotspot, South Africa. We focused specifically on the roads dissecting remnant grassland patches and/or grasslands corridors among plantation compartments. Overall dung beetle species richness was higher for maintained unpaved roads compared to unmaintained tracks. Only nocturnal dung beetle abundance was impacted by distance from roads, while nocturnal species assemblage composition was influenced by road orientation. These findings lend only partial support to our first and second hypotheses. However, other environmental variables related to vegetation, but not substrate characteristics, were important, supporting our third hypothesis. These findings suggest that the indirect impacts of roads may outweigh their direct impacts on dung beetles in this grassland area.

In forested regions, road construction has been linked to declines in dung beetle species richness and abundance, regardless of road width, substrate, and usage frequency (Hosaka et al. 2014; Yamada et al. 2014). Main effects originate from roads clearing patches of forest and introducing landscape contrast, increasing sunlight penetration, and in turn leading to higher surface temperatures and increased desiccation rates (Peyras et al. 2013).

In the open grassland setting investigated here, unpaved roads had little impact on dung beetles. Although road construction and usage alter overall landscape characteristics, the contrast introduced in this grassland setting was lower compared to forested settings (Yamada et al. 2014). The insect species occupying this region are likely pre-adapted to survive abrupt changes to their habitats, as the region is prone to occasional fire events, while climate and habitat characteristics can vary between years. The habitat openness introduced by unpaved roads is likely not enough to hinder occupancy of dung beetles in this grassland region, as was the case elsewhere (Koivula 2005). These findings were perhaps related to dung beetles being able to detect food resources more efficiently in open areas (Melis et al. 2010), while roads can also provide attractive microhabitats (Itzhak 2008).

The weak relationships with distance away from roads likely result from these roads having an overall low traffic volume compared to the national road network (Melis et al. 2010). Another reason was that animals such as elephant, buffalo, wildebeest, and a range of antelope species freely roam the study area, even crossing into plantation compartments. Occupancy by these large mammals is an important driver of local dung beetle diversity patterns (Pryke et al. 2022). Although some mammal species prefer areas farther away from roads where vegetation is taller and denser (Naidenko et al. 2021), many mammal species also use roads to traverse the region (Hill et al. 2021). As a result, there is a relatively even distribution of animal dung across the landscape, meaning that food resources for dung beetles are scattered across the landscape.

Most grassland dung beetle species are highly mobile and generalist scavengers, enabling them to be more resilient to the direct impacts of roads. However, roads influence vegetation height, composition and percentage cover, especially when road verges are frequently managed (Jakobsson et al. 2018). These vegetation characteristics are important drivers of dung beetle diversity patterns in forested regions (Yamada et al. 2014). Complementary to previous findings, our results indicated that dung beetle species richness, abundance, and assemblage composition were driven by vegetation characteristics. Vegetation was overall short and sparse close to road verges, while taller farther away. Most of the dung beetles recorded here were sun-loving, with some species tolerating partially shaded habitats (Davis et al. 2020). This meant that exposed areas in the landscape associated with roads and their verges, were favored by most species recorded here. Linear landscape features such as road verges, with short and sparse vegetation, may also act as dispersal corridors for ground-dwelling beetles (Koivula and Vermeulen 2005), and enable dung beetles to process and move dung more easily (Stanbrook et al. 2021).

These findings do not necessarily mean that densely vegetated areas away from roads are unimportant, as they are attractive habitats to shade-tolerant species and offer dung beetles shelter against unfavourable weather conditions. An interesting result of this study was that nocturnal dung beetle abundances were higher farther away from roads, where vegetation was tall and dense. It is possible that nocturnal beetles select for areas with dense vegetation to obscure themselves from nocturnal insectivores (Young 2015). Overall, microhabitat structure (related to vegetation cover and composition) is important for dung beetle occupancy, as was previously suggested for the region (van Schalkwyk et al. 2017) and elsewhere (Davis et al. 2020). Our results showed that the variation in microhabitat structure led to variation in assemblages across this grassy landscape. Substrate characteristics, which are also influenced by road networks, were unimportant to the dung beetles investigated here, suggesting that road surfacing and degrees of compaction have little impact on grassland dung beetles.

Interestingly, nocturnal dung beetle assemblages varied between north-south orientated roads relative to those that were orientated from east to west. Although it has been suggested that dung beetles ignore landscape features during navigation (Dacke et al. 2013), the interaction between road orientation and astral cues has not been studied in depth, warranting further investigation to make definite conclusions.

Limitations

Dung beetle diversity is strongly related to type and spatial distribution of dung (Davis et al. 2008). In the area, dung beetles have a high preference for elephant, buffalo and zebra dung (Pryke et al. 2022), all of which freely roamed the landscape. We acknowledge that using baited pitfall traps instead of baitless pitfalls could have potentially produced higher yields of trapped individuals. However, our aim was to assess dung beetles traversing the grassland landscape relative to the existing road network, and using baited pitfalls would have led to interference among traps (Silva and Hernández 2015), so further obscuring the effects of roads. We also acknowledge that climatic conditions overall influence dung beetle diversity and their activity patterns (Davis 2002; Correa et al. 2018). Yet, data collection was performed over a relatively short time frame, during which climatic fluctuations were minor. Long-term monitoring may reveal that road effects vary between consecutive years, and in response to moisture gradients, dung resources, and microhabitat availability.

While the results presented here indicate how secondary attributes of roads influence grassland-dwelling dung beetle diversity patterns, it provides little information on the direct effects of roads (Hayward et al. 2010). Dung beetle individuals going near roads are likely killed by passing vehicles, yet it is challenging to quantify the magnitude of direct killing, especially since dung beetle population sizes are not known for the area. However, direct killing may lead to long term reduction in abundances, especially for weak fliers.

Conservation implications

The habitat openness introduced by roads in this grassland setting had little effect on dung beetles, suggesting that roads do not alter opportunities for dung beetles to traverse the landscape. However, road construction and maintenance change vegetation structure, which was an important factor for the set of dung beetle species investigated here, and interestingly, nocturnal species in particular had a higher preference for areas away from roads where vegetation was overall tall and dense.

These findings suggest that road management should consider the wider landscape-level influence of roads, rather than focusing on road characteristics per se. Maintaining habitat heterogeneity of the landscape relative to roads is essential to conserve the full set of dung beetles in the region. With road verges being attractive habitats for the dung beetle species of the region, maintaining a combination of open and densely vegetated areas away from roads would reduce interaction with vehicles, so limiting direct killing of both open habitat and shade tolerant dung beetles. As mammals roam the area regardless of road presence, it is imminent for dung beetles to occur on or near roads. Road users should keep on road tracks to limit impact on adjacent favorable habitats, but exercise caution when traveling on roads to avoid direct killing of dung beetles. While larger and more mobile dung species may be able to move away as vehicles approach, road users should avoid trampling dung pats on the road surface to reduce killing pressure on less mobile species. To limit future impacts on dung beetles in the region, extending the road network should be considered with caution as this could lead to increased dung beetle mortality rates.