Skip to main content
NCBI home page
Scientific Reports logoLink to Scientific Reports
. 2025 Jul 13;15:25333. doi: 10.1038/s41598-025-10402-6

Impact of linear infrastructure and landscape characteristics on wildlife roadkill in the Nelliyampathy Hills, Western Ghats, India

S Sushanth 1,2,3,5, G Praphul 5, S R Ganesh 4, Sanjay Molur 1, Honnavalli N Kumara 5, Mewa Singh 2,✉
PMCID: PMC12256599  PMID: 40653534

Abstract

The expansion of linear infrastructures, such as roads, which are integral to human society, poses a serious threat to wildlife, leading to road accidents, which have become major causes of terrestrial and avian wildlife mortality worldwide. This study, conducted from June 2023 to May 2024 in the Nelliyampathy Hills, Western Ghats, Kerala, India aimed to assess terrestrial and avian vertebrate roadkill, and the environmental parameters influencing it. In 22 roadkill surveys, 330 roadkills were recorded, representing 72 species, that included 228 individuals of reptiles (43 species) (66.09%), 70 amphibian individuals (11 species) (20.29%), 23 mammal individuals (10 species) (9.57%), and 9 bird individuals (8 species) (4.06%). The annual roadkill estimate was 5,490 along a 50 km transect, with an overall roadkill rate of 0.3 roadkills/km/year. Environmental factors such as plantations, road pavement, water sources, terrain, and undergrowth were found to significantly influence roadkill occurrences. However, there was negligible spatial and seasonal variation in roadkill hotspots. This communication presents advisory measures to reduce wildlife mortality from vehicle collisions to support ecosystem health and minimize such wildlife mortality.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-10402-6.

Keywords: Wildlife mortality, Roadkill, Environmental parameters, Terrestrial and avian vertebrates

Subject terms: Ecology, Zoology

Introduction

Urbanization is one of the most rapidly growing processes around the world, leading to a high demand for linear infrastructure, resulting in the expansion of road networks and connectivity to accommodate growing populations and economic activities1,2. Globally, there are at least 36 million km of roads3, and according to the Indian Ministry of Road Transport and Highways (2021)4, India boasts of the second-largest road network, stretching over 6.2 million km, and also indicating a substantial increase in expansion in the coming years. From 1951 to 2015, the total length of roads in India has grown more than 11-fold, expanding from 399,000 km to 4.67 million km5. By 2050, it is anticipated that at least 25 million km of new roads will be constructed worldwide, representing a 60% increase in road lengths compared to 20106.

With the rapid development of this scale, mortality from road accidents is among the leading direct human causes of death for animals worldwide7–16. Roads are known to negatively impact animals in multidimensional ways. They are the most obvious and leading cause of human-induced mortality for many vertebrates, second only to legal harvesting, as they significantly contribute to deaths through animal-vehicle collision (AVC hereafter)17,18. Roads pose a threat to both endangered and common species primarily due to the increased mortality risk from AVCs19. The other negative effects of roads on wildlife include habitat destruction and fragmentation, distribution and movement patterns, breeding and population density, and heterozygosity and genetic polymorphism20–26. Such effects have likely intensified over the past 50 years due to the expansion of road networks and increased vehicular traffic27. Such an enormous increase in road network will push animals into proximity with vehicular traffic (especially, the ones that crawl or move across roads) by fragmenting and altering the habitat, often leading to mortality from AVCs28,29. With an anticipated 60% increase in road lengths by 2050 compared to 2010, we can expect a corresponding increase in AVCs (Mammal-vehicle collisions have already risen significantly since the 1970s)6,18.

It is, therefore, essential to evaluate the scale of wildlife deaths resulting from road traffic. It is estimated that in the United States alone, over a million vertebrates are killed each day and 340 million birds annually30, 194 million birds and 29 million mammals are killed annually in Europe31, and 12.4 million birds and 5.1 million mammals are killed in Latin America32. An annual estimate of 7 million birds in Bulgaria, 5 million frogs and reptiles in Australia, 13.8 million birds in Canada, and 159,000 mammals along with 653,000 birds are killed annually in the Netherland9,33. Roadkill in India have also been reported by various investigators15,34–39.

These issues not only pose a threat to wildlife populations but also have ecological and economic impacts, such as disrupted predator–prey relationships, fluctuated population dynamics, and increased vehicle maintenance costs9,40. Environmental parameters surrounding roadways are known to be strongly associated with AVCs such as the presence of nearby water bodies, shrublands, forests, and grasslands along with road-specific characteristics41–46. Understanding the patterns and causes of roadkill can provide crucial insights into mitigating their effects, such as improving road design, enhancing wildlife corridors, and implementing better driver awareness campaigns47,48. Given these concerns, studying roadkill should be prioritized to develop effective mitigation strategies for reducing animal fatalities, protecting biodiversity, and making our transportation systems more harmonious with the natural world.

Western Ghats, which is a biological global hotspot housing over 30% of wildlife in India, has undergone extensive ecological transformations due to anthropogenic impacts, particularly over the last century by losing over 35% of forest cover49–52. Apart from a few short-term studies on wildlife roadkill in Kerala, India53–56, no major studies have been conducted, highlighting that there is a lacuna in understanding how roads and highways are affecting wildlife in this biological global hotspot. Thus, the present study was conducted to assess the extent of roadkill occurrences with a focus on quantifying the frequency and distribution of incidents across different taxa in the Nelliyampathy Hills, which lie in the southern Western Ghats. Additionally, this study seeks to assess the influence of surrounding environmental factors—such as habitat type, road characteristics, and surroundings—on the prevalence of roadkill, and to determine how they disproportionately affect particular taxa, thereby providing a clear understanding of taxa-specific vulnerabilities and informing targeted conservation strategies.

Results

In our study area (Fig. 1), 330 roadkill individuals were recorded belonging to four vertebrate classes (Amphibia, Reptilia, Aves, and Mammalia) with 72 species. The count included 228 reptile individuals (with 8 unidentified snakes), 70 individuals of amphibians (with 14 unidentified individuals), mammals with 23 individuals (with 5 unidentified carcasses), and birds with 9 individuals (with 1 unidentified individual). In total, 72 roadkill species (identified) were recorded including 43 species of reptiles, 11 species of amphibians, 8 species of birds, and 10 species of mammals (Table 1; Figs. 2, 3a,b).

Fig. 1.

Fig. 1

Open in a new tab

Study area map created using QGIS version 3.26.2 (https://qgis.org) along with roadkill survey track (Pothundi dam to Kaikatty junction 20 km, Kaikatty junction to Seethargundu 9 km, Kaikatty junction to Victoria bridge junction 9 km, Victoria bridge junction to Karadi bridge 5 km, Victoria bridge junction to KFDC Pakuthipalam 7 km).

Table 1.

Species list and total number of roadkills recorded during the study period (Jun 2023–May 2024).

Common name Scientific name Roadkill individuals (n) Endemic IUCN WPA
I WG
Amphibians
 Bicoloured frog Clinotarsus curtipes 22 + + LC
 Bombay caecilian Uraeotyphlus bombayensis 3 + + LC
 Bull frog Hoplobatrachus tigerinus 3 – – LC II
 Common Indian toad Duttaphrynus melanostictus 13 – – LC
 Don’s bronze frog Indosylvirana doni 2 + + NT
 Fungoid frog Hydrophylax malabaricus 2 + – LC
 Malabar gliding frog Rhacophorus malabaricus 2 + + LC
 Sharp-tailed caecilian Uraeotyphlus cf. oxyurus 3 UK UK DD
 Spinular night frog Nyctibatrachus acanthodermis 1 + + EN
 Sreeni’s golden-backed frog Indosylvirana sreeni 2 + – LC
 Tricolored caecilian Ichthyophis tricolor 2 + + LC
 Golden backed frog Indosylvirana sp. 2 – – – –
 Leaping frog Indirana sp. 1 – – – –
 Unidentified frog – 12 – – – –
 Total amphibians 70
Reptiles
 Anamalai pit viper Trimeresurus anamallensis 5 + + LC II
 Anamalai vine Snake Ahaetulla isabellina 12 – – – II
 Annandale’s grass skink Eutropis dawsoni 1 – – –
 Ashok’s bronzeback tree snake Dendrelaphis ashoki 1 + + LC II
 Beaked worm Snake Grypotyphlops acutus 2 + – LC II
 Beddome’s coral snake Calliophis beddomei 1 + – DD II
 Beddome’s keelback Sahyadriophis beddomei 14 – – – II
 Beddome’s shieldtail Uropeltis beddomii 7 + + DD II
 Bibron’s coral snake Calliophis bibroni 1 + + LC II
 Brahminy blind snake Indotyphlops braminus 1 – – LC II
 Brown vine snake Ahaetulla sahyadrensis 5 – – – II
 Checkered keelback Fowlea cf. piscator 7 – – LC I
 Cochin shieldtail Uropeltis nitida 2 + + DD II
 Collared cat snake Boiga nuchalis 3 + – LC II
 Common krait Bungarus caeruleus 1 – – LC II
 Common sand boa Eryx conicus 5 – – NT II
 Common vine snake Ahaetulla oxyrhyncha 2 – – – II
 Common wolf snake Lycodon aulicus 3 – – LC II
 Elliot’s forest lizard Monilesaurus ellioti 2 + + LC
 Ferguson’s shieldtail Rhinophis fergusonianus 7 + + DD II
 Green forest lizard Calotes calotes 5 – – LC
 Gunther’s grass skink Eutropis brevis 1 – – –
 Hill keelback Amphiesma monticola 4 + + LC II
 Hump-nosed pit viper Hypnale hypnale 8 – – LC II
 Indian Flying lizard Draco dussumieri 1 + – LC II
 Indian garden lizard Calotes versicolor 24 – – LC
 Indian black turtle Melanochelys trijuga 1 – – LC II
 Keeled skink Eutropis carinata 32 – – LC
 Kerala shieldtail Uropeltis cf. ceylanica 7 UK UK DD II
 Large eyed bronzeback Dendrelaphis grandoculis 3 + + LC II
 Large scaled forest lizard Calotes grandisquamis 1 + + LC
 Montane trinket Coelognathus helena monticollaris 4 – – – II
 Nelliyampathy shieldtail Uropeltis cf. ocellata 15 + – LC II
 Nilgiri forest lizard Calotes nemoricola 6 + + LC
 Pied bellied shieldtail Melanophidium punctatum 2 + + VU II
 Rat snake Ptyas mucosa 1 – – LC I
 Rock agama Psammophilus dorsalis 1 + – LC
 Striped coral Calliophis nigrescens 4 + + LC II
 Thackrey’s cat snake Boiga thackerayi 1 – – – II
 Travancore wolf snake Lycodon travancoricus 13 + – LC II
 Variegated kukri Oligodon taeniolatus fasciatus 2 – – – II
 Western ghats bronzeback Dendrelaphis chairecacos 1 + + DD II
 Western kukri snake Oligodon affinis 1 + + LC II
 Unidentified shieldtail Uropeltis sp. 2 – – – II
 Unidentified snake – 6 – – – –
 Total reptiles 228
Birds
 Greater coucal Centropus sinensis 1 – – LC II
 House crow Corvus splendens 1 – – LC
 Indian nightjar Caprimulgus asiaticus 1 – – LC II
 Jerdon’s nightjar Caprimulgus atripennis 1 – – LC II
 Red whiskered bulbul Pycnonotus jocosus 1 – – LC II
 Southern hill myna Gracula indica 1 – – LC I
 Streak-throated woodpecker Picus xanthopygaeus 1 – – LC II
 White-bellied drongo Dicrurus caerulescens 1 – – LC II
 Unidentified bird – 1 – – – –
 Total birds 9
Mammals
 Asiatic long-tailed climbing mouse Vandeleuria oleracea 1 – – LC
 Blanford’s rat Madromys blanfordi 3 – – LC
 Bonnet macaque Macaca radiata 1 + – VU I
 Feral cat Felis catus 1 – – –
 Flat-haired mouse Mus platythrix 2 + – LC
 Greater bandicoot rat Bandicota indica 1 – – LC
 Indian black rat Rattus rattus 2 – – LC
 Jungle striped squirrel Funambulus tristriatus 4 + + LC II
 Leopard Panthera pardus 1 – – VU I
 Lesser bandicoot rat Bandicota bengalensis 1 – – LC
 Mouse sp. Mus sp. 2 – – – –
 Shrew sp. Suncus sp. 1 – – – –
 Unidentified mammal – 3 – – – –
 Total mammals 23
 Total roadkills 330
Open in a new tab

LC, Least Concerned; NT, Near Threatened; VU, Vulnerable; EN, Endangered; DD, Data Deficient; UK, Unknown.

Fig. 2.

Fig. 2

Open in a new tab

Total monthly roadkill counts for all the recorded taxa along with the average monthly rainfall.

Fig. 3.

Fig. 3

Open in a new tab

(a) Total number of roadkill, and (b) Mosaic plot of the proportion of roadkill by animal class and season.

The highest number of overall roadkills was recorded in the months of January and September with 57 individuals each (including all taxa), followed by May with 51 individuals, and the least was in the month of August with no roadkills (Fig. 2).

Of the 72 recorded road-killed species, 1 amphibian species is listed as Endangered, 2 species as Near-threatened (1 amphibian and 1 reptile), 3 species as Vulnerable (1 reptile and 2 mammals), 7 species as Data deficient (1 amphibian and 6 reptile), and 49 species as Least concerned (8 amphibians, 26 reptiles, 8 birds and 7 mammals)57. In terms of endemicity, 33 species are endemic to India (8 amphibians, 22 reptiles, and 3 mammals) while 22 species are endemic to only the Western Ghats (6 amphibians, 15 reptiles, and 1 mammal)57. According to Wild Life (Protection) Amendment Act (2022)58 of India, 5 species belong to Schedule-I (endangered species) (2 reptiles, 1 bird, and 2 mammals) and 41 species belong to Schedule-II (given high protection and trade prohibited) (1 amphibian, 33 reptiles, 6 birds, and 1 mammal).

During the Monsoon, the roadkill number was 12 individuals of amphibians, 49 of reptiles, 6 of birds, and 7 of mammals. In the post-monsoon, the roadkills included 36 individuals of amphibians, 104 of reptiles, 1 bird, and 11 of mammals. During the Dry season, the amphibian roadkill number was 22, the reptile was 75, the bird was 7 and the mammals was 15 (Fig. 3).

For the Overall Roadkills model, the random forest algorithm predicted Coffee Plantation, Muddy Terrain, Water source Dam, High Undergrowth, and Low Canopy cover as the top predictors. The Coffee plantation was the major predictor (%IncMSE = 14.91) and (IncNodePurity = 139.86) (Fig. 4), Muddy Terrain and Water source Dam were ranked second and third making moderate contributions to both metrics, High Undergrowth and Low Canopy cover, though demonstrated lower importance, still played a meaningful role in model performance (Table 2). The Coffee Plantation (β = − 1.02, p = 0.00), Water source Dam (β = − 0.63, p = 0.01) and Low Canopy cover (β = − 0.36, p = 0.07) exhibited negative association, with the first two only being significant. Muddy Terrain (β = 0.32, p = 0.11) and High Undergrowth (β = 0.27, p = 0.18) exhibited a positive but nonsignificant association (Table 3).

Fig. 4.

Fig. 4

Open in a new tab

Impact of different environmental factors affecting the roadkill (a) overall roadkills, (b) amphibian roadkills, (c) reptile roadkills, (d) bird and mammal roadkills (Top 5 predictors of Overall: coffee plantation, muddy surrounding terrain, nearby by dam water source, high undergrowth and low canopy cover; Amphibians: coffee and paddy plantations, nearby by dam water source, roadtype tar, boulder & concrete mix and muddy surrounding terrain; Reptiles: coffee plantation, nearby by dam & pond water source, winding and linear road pattern; BM: muddy surrounding terrain, roadtype tar, boulder & concrete mix, explicitly boulder roadtype, coffee plantation and low undergrowth).

Table 2.

Top predictors ranked by their importance scores identified by the Random Forest algorithm for roadkill.

Species Variables %IncMSE IncNodePurity
Overall Estate Coffee 14.91 139.86
Terrain Muddy 7.94 107.36
Water Source Dam 7.88 92.77
Undergrowth High 5.63 49.11
Canopy Cover Low 5.33 38.44
Amphibian Estate Coffee 10.43 8.99
Estate paddy 9.65 4.46
Water Source Dam 6.21 5.61
Roadtype TBC (mixture of Tar, Boulder and Concrete) 3.72 0.67
Terrain Muddy 2.64 7.14
Reptile Estate Coffee 13.78 74.08
Water Source Dam 6.31 35.03
Water Source Pond 6.17 3.24
Road pattern winding 6.10 55.45
Road Pattern Linear 5.83 42.63
Birds and Mammal Terrain Muddy 8.62 5.46
Terrain RGM(Mixture of Rocky, Grass and Muddy) 7.57 4.96
Roadtype Boulder 5.97 9.49
Estate Coffee 5.81 2.94
Undergrowth Low 5.41 2.67
Open in a new tab

Table 3.

Summary of models obtained from the GLM explaining environmental and surrounding factors influencing roadkills in Nelliyampathy Hills, Kerala.

Model Variable Estimate SE Z value Pr (>|z|) Significance AIC
Overall roadkill’s
 Amphibians + Reptiles + BM ~ E.coffee + Ter.Muddy + WS.dam + Undergrowth.H + CC.L (Intercept) 1.72 0.21 8.25 0.00 *** 444.15
E.coffee − 1.02 0.23 − 4.44 0.00 ***
Ter.Muddy 0.32 0.20 1.60 0.11
WS.dam − 0.63 0.23 − 2.74 0.01 **
Undergrowth.H 0.27 0.20 1.33 0.18
CC.L − 0.36 0.20 − 1.83 0.07
Amphibian roadkill’s
 Amphibians ~ E.coffee + E.paddy + WS.dam + Ter.Muddy (Intercept) − 0.18 0.30 − 0.60 0.55 217.32
E.coffee − 0.69 0.37 − 1.86 0.06
E.paddy 2.37 0.92 2.57 0.01 *
WS.dam − 1.27 0.48 − 2.62 0.01 **
Ter.Muddy 0.58 0.35 1.64 0.10
Reptile roadkill’s
 Reptiles ~ E.coffee + WS.dam + Roadpat.W (Intercept) 1.22 0.19 6.59 0.00 *** 367.77
E.coffee − 1.18 0.25 − 4.80 0.00 ***
WS.dam − 0.66 0.24 − 2.81 0.00 **
Roadpat.W 0.38 0.20 1.86 0.06
Birds and Mammal roadkill’s
 BM ~ Roadtype.Bould + E.coffee + Undergrowth.L (Intercept) − 0.35 0.36 − 0.98 0.33 202.94
Roadtype.Bould 1.66 0.91 1.84 0.07
E.coffee − 0.18 0.42 − 0.43 0.67
Undergrowth.L − 0.28 0.41 − 0.70 0.48
Open in a new tab

The table includes the model intercept and coefficients for predictor variables, with their associated standard errors (SE), Z values, p values (Pr (>|z|)), and levels of statistical significance (Significance codes: ***< 0.001, **< 0.01, *< 0.05, < 0.1), based on the specified model formula.

For the Amphibian Roadkills model, the random forest algorithm predicted Coffee Plantation, Paddy Plantation, Water source Dam, Road type TBC (mixture of Tar, Boulder, and Concrete), and Muddy Terrain as the top predictors. While, the Coffee Plantation (%IncMSE = 10.43; IncNodePurity = 8.98) (Fig. 4) and Paddy Plantation (%IncMSE = 9.64; IncNodePurity = 4.46) were the most important predictors, the Water source Dam ranked third making moderate contributions to both metrics, Road type TBC and Muddy Terrain demonstrated lower importance but still played a meaningful role in model performance (Table 2). Coffee Plantation (β = − 0.69, p = 0.06) and Water source Dam (β = − 1.27, p = 0.01) exhibited negative associations with the first being nonsignificant and the second being significant. Plantation Paddy (β = 2.37, p = 0.01) and Muddy Terrain (β = 0.58, p = 0.10) exhibited a positive association, with Plantation Paddy being significant (Table 3).

For the Reptile Roadkills model, the random forest algorithm predicted Coffee Plantation, Water source Dam, Water source Pond, Road Pattern Winding, and Road Pattern Linear as the top predictors. Coffee Plantation (%IncMSE = 13.78; IncNodePurity = 74.08) (Fig. 4) was predicted as the most important predictor. Water source Dam, Water source Pond, and Road Pattern winding made moderate contributions to both metrics, Road Pattern Linear demonstrated lower importance but still played a meaningful role in model performance (Table 2). While Coffee Plantation (β = − 1.18, p = 0.00) and Water source Dam (β = − 0.66, p = 0.00) exhibited negative association being significant, Road Pattern winding (β = 0.38, p = 0.06) exhibited a non-significant positive association (Table 3).

In the Birds and Mammal Roadkills model, the random forest algorithm predicted Terrain Muddy, Terrain RGM (Mixture of Rocky, Grass, and Muddy), Roadtype Boulder, Coffee Plantation, and Low Undergrowth as the top predictors. Terrain Muddy (%IncMSE = 8.62; IncNodePurity = 5.46) (Fig. 4) was predicted as the most important predictor. While, Terrian RGM (Mixture of Rocky, Grass, and Muddy) made moderate contributions to both metrics, Roadtype Boulder, Plantation Coffee, and Low Undergrowth demonstrated lower importance, though still playing a meaningful role in model performance (Table 2). While, Roadtype Boulder (β = 1.66, p = 0.07) exhibited a non-significant positive association, Coffee Plantation (β = − 0.18, p = 0.67) and Low Undergrowth (β = − 0.28, p = 0.48) exhibited a nonsignificant negative association (Table 3).

The spatial representation of roadkills provides a comprehensive overview of the frequency and distribution patterns and locations for each of the four taxa, (Fig. 5). The heatmaps of different taxa in the same season show significant differences in distribution, frequency, and locations of roadkills, but heatmaps of same taxa across different seasons shows least to minimal variation. Birds and mammals were two of the least affected taxa, hence, we had minimal data on those taxa which makes it difficult to draw conclusive patterns on its roadkill seasonal variation and spatial distribution. Furthermore, the combined seasonal heatmap of all taxa facilitates a broader analysis of roadkill hotspots, helping to identify areas of consistent wildlife vulnerability across multiple seasons. Anyhow, in the case of all taxa combined, there was no significant variation in distribution and location (Figs. 5, 6). Images of all roadkills are shown in Fig. 7.

Fig. 5.

Fig. 5

Open in a new tab

Spatial representation of all the roadkill taxa wise (a) Amphibians, (b) Birds, (c) Mammals, (d) reptiles.

Fig. 6.

Fig. 6

Open in a new tab

Heatmaps of the density of roadkill events for each class and season with seasonal overall.

Fig. 7.

Fig. 7

Open in a new tab

Roadkill images of some individuals recorded in Nelliyampathy during the study period (a) Indian black turtle, (b) Indian flying lizard, (c) Common sand boa, (d) Malabar gliding frog, (e) Spinular night frog, (f) Tricoloured caecilian, (g) Jungle striped squirrel, (h) Bonnet macaque, (i) Asiatic long-tailed climbing mouse.

Discussion

In the current study, a total of 330 roadkill individuals belonging to four vertebrate classes (Amphibia, Reptilia, Aves, and Mammalia) were recorded (Fig. 7), comprising 72 species. Reptiles constituted the highest number of roadkills, followed by amphibians, mammals, and birds, the latter being the least affected. Of the total species identified, reptiles included 43 species, amphibians 11, birds 8, and mammals 10.

The results highlight that reptiles are the most affected taxon, potentially due to several factors. Reptiles’ high exposure to roads, their tendency to bask on roads for thermoregulation59, their mode of locomotion28, and their feeding behavior60,61 make them particularly vulnerable. Additionally, snakes and other reptiles often evoke fear in people, leading to intentional killings on roads62. The proximity of roads to habitats offering resources such as food and basking spots further elevates risk63,64. This highlights the dual challenge of direct vehicular threats and human-induced risks.

Amphibians exhibited relatively lower roadkill numbers, which may be influenced by survey methodologies. As the surveys were conducted bi-monthly during the day, carcasses of nocturnal amphibians may not have been observed65. Amphibians are primarily active during the night, especially in conditions of high rainfall and humidity66. This nocturnal activity could lead to increased predation or scavenging of carcasses before surveys67. Furthermore, amphibians’ low detectability on roads during daytime surveys68,69 underscores the need for optimized temporal sampling. Future studies should incorporate nocturnal surveys to better capture the impact of roadkill on amphibians.

Mammals were relatively less affected, likely due to their higher detectability owing to their larger sizes, which may enable drivers to avoid them70. However, small mammals, which are more susceptible to roadkills, face risks similar to reptiles due to their locomotion patterns, small size, and low visibility29,71. Moreover, the high density of small mammals near roadside verges makes them more vulnerable72,73. This dynamic highlights the potential for roadside habitats to function as ecological traps for smaller mammals.

Birds were the least affected class, likely due to their ability to fly, which reduces their exposure to vehicular collisions74. The auditory and visual sensitivity of birds to approaching vehicles further enhances their likelihood of escaping collisions74,75. However, ground-dwelling bird species, which are more vulnerable to roadkill, require further investigation to determine the specific environmental factors contributing to their mortality.

Environmental variables significantly influenced roadkill frequency across all classes. Factors such as coffee plantations (even though natural intact rainforest had the highest adjacent land cover), mixed terrain (rocky, grassy, and muddy), water sources, and low canopy coverage played critical roles (Tables 2, 3). These variables are known to attract wildlife, either due to resource availability or habitat suitability39,76. Reptiles were notably affected in areas near water bodies and straight road stretches (Tables 2, 3) where vehicles tend to accelerate, reducing reaction times for both drivers and animals77. Amphibians were significantly impacted in coffee and paddy plantation areas, while muddy terrains attracted herpetofauna due to insect presence63.

The comprehensive range of environmental parameters collected and their correlation with roadkills constitute a central novelty of the study. The findings underscore the interaction between habitat features and animal behavior, particularly the role of roads as ecological traps. Roadsides often provide favourable microhabitats, attracting animals and increasing the likelihood of vehicle collisions60–64. Habitat-specific factors such as resource availability, road structure, and vegetation types must be considered when designing mitigation strategies.

Birds and mammals showed relatively low roadkill numbers, making it difficult to draw conclusive patterns. However, habitat features such as mixed terrains and undergrowth appeared to play a role in roadkill incidents of these taxa. Future research should aim to identify critical hotspots for these groups to address their specific needs.

Seasonal analysis indicated a relatively consistent roadkill rate throughout the year, with reptiles consistently representing over 60% of roadkill across all seasons. Amphibians showed moderate seasonal variation, with higher roadkills during the post-monsoon period. Interestingly, contrary to our hypothesis, roadkill numbers were lower during peak monsoon, potentially due to the low detectability of carcasses, getting washed away by rain and scavenging65–67. This indicates the importance of integrating climatic factors into roadkill assessments.

Among reptiles, snakes were the most frequently road-killed group, with endemic species such as uropeltids showing particularly high mortality during the pre-monsoon period. Amphibians were predominantly represented by frogs and toads, with the Bicoloured frog (Clinotarsus curtipes) being the most affected species. Mammalian roadkills were dominated by rodents, while in birds, aerial species were more affected than ground-dwelling ones, contrary to expectations. These patterns suggest that species-specific traits, such as behavior and ecology, critically influence roadkill vulnerability.

During our study, the average roadkill per survey per day is estimated at 15 (Fig. 8). If we project it for a year, the estimate amounts to 5,490 roadkill in the selected 50 km road segment. The numbers further indicate 0.3 roadkills/kilometer/day. This value represents a general roadkill estimate rather than a species/taxa-specific estimate. Furthermore, no correction factor has been applied due to the unavailability of carcass persistence data. Globally, quantifying the extent of roadkills occurring per kilometer per year would provide a clear picture of the intensity at which roadkills are happening. Out of several studies across the world that have reported all terrestrial vertebrate roadkill, analysis of raw data indicates that Australia has the highest extent of roadkill with 15.64 roadkill/kilometer/year (hereafter rk/km/yr), followed by India with 15.45 rk/km/yr and least extent is occurring in Canada with 0.003 rk/km/yr (Supplementary table 1). In Nelliyampathy Hills, the roadkill extent is 0.3 rk/km/yr, which is not extremely high, but it is significant enough to impact negatively the wildlife population dynamics of the area. We should also not forget that the true extent of roadkill mortality is likely significantly higher than reported, as many roadkills go undiscovered or unreported, leading to a severe underestimation of the actual magnitude of mortality occurring78. Vertebrate wise rk/km/year values in Nelliyampathy, amphibian’s rk/km/year is 0.064, reptiles are 0.21, birds are 0.0082 and mammals are 0.021. While this extrapolation in itself may not serve as a definitive indicator of ecological impact, it nonetheless offers a useful basis for comparative analysis across different geographical regions. Such comparisons can reveal patterns or discrepancies that warrant further ecological interpretations, thereby enriching the broader understanding of AVC dynamics.

Fig. 8.

Fig. 8

Open in a new tab

The mean number of roadkill of different animals in the year 2023–2024 in Nelliyampathy Hills, Kerala.

Roadkill mortality is strongly influenced by local abundance, and its effects are particularly pronounced in less abundant populations65. While reptiles and amphibians showed high roadkill numbers, the impact on less abundant taxa, such as certain mammals and birds, can still be significant. Mitigation strategies should prioritize local species and regions most vulnerable to roadkill.

Effective measures for accident-prone taxa include constructing wildlife corridors, such as overpasses and underpasses, implementing speed limits in high-risk areas, and enhancing public awareness of wildlife conservation. Additionally, road design improvements, such as incorporating reflective surfaces, speed bumps, and barriers, can reduce wildlife-vehicle collisions. In the Nelliyampathy hills, mitigating measures of the night traffic ban have already been undertaken by not allowing vehicle movements from 9 pm to 6 am, to protect nocturnal wildlife, and reduce roadkill incidents. These interventions align with global recommendations to mitigate the ecological impact of infrastructure development.

While the findings of this study are based on a rigorous and systematic methodology, it is acknowledged that the exclusion of night-time surveys may have limited the detection of nocturnal fatalities or species active during low-light conditions. Additionally, although the current results provide a reliable assessment of the mortality rates, the absence of carcass detection and removal trials, which is generally recommended for quantifying potential biases associated with scavenger activity and observer detection efficiency could not be fully quantified due to logistics reasons. Incorporating these elements in future studies would build on the strong foundation laid here, further enhancing the accuracy and comprehensiveness of the impact assessment. Nevertheless, the data reported in this article are sufficient to inform the authorities of effective conservation management plans.

  1. Survey enhancements: Future studies should incorporate nocturnal surveys to better assess the impact on nocturnal species, particularly amphibians. Comprehensive, year-round monitoring can provide deeper insights into seasonal and taxonomic variations.

  2. Habitat-specific strategies: Targeted mitigation measures such as wildlife crossings, speed limits, and fencing should be implemented in high-risk areas. Special attention should be given to ecological traps created by roadsides.

  3. Policy and public awareness: Promoting environmental awareness through education campaigns and incorporating eco-friendly designs in infrastructure projects can aid in wildlife conservation. Collaborative efforts between policymakers, conservationists, and local communities are critical.

  4. Long-term monitoring and research: Continuous monitoring of roadkill incidents is essential to track changes in population dynamics and evaluate the effectiveness of mitigation measures. Research should also explore the indirect impacts of roadkill, such as population fragmentation and genetic isolation.

This study underscores the urgent need to address roadkill mortality to conserve biodiversity in the Nelliyampathy Hills and similar regions. The findings provide valuable insights into the interplay between infrastructure development and wildlife mortality, emphasizing the critical need for sustainable development practices. Roads, while essential for human progress, must be designed and managed to coexist with natural ecosystems. Integrating conservation science into infrastructure planning can mitigate wildlife losses and foster resilience in vulnerable populations.

Sustainable infrastructure development must prioritize coexistence with natural ecosystems to mitigate the impact on wildlife populations. By implementing targeted conservation measures, fostering public awareness, and advancing scientific research, we can work towards a balanced approach to development that safeguards biodiversity for future generations.

Methods

Study area

The Nelliyampathy hills (10.30–10.41N and 76.58–76.75E) are located in the Palakkad district, Kerala, India (Fig. 1). Situated at the southern edge of the Palghat gap in the Western Ghats, these hills encompass three forest ranges: Nelliyampathy range, Alathur range, and Kollangod range, covering an area about 285 Sq. Km79. The hills contain vegetation of evergreen, semi-evergreen, and moist deciduous forests with scattered tea, coffee, and cardamom plantations80. The altitude ranges between 100 m in plains and 1700 msl in the shola grasslands81. This area encounters the southwest monsoon, with pre-monsoon rains in April and May, followed by the onset of the monsoon by late May or early June, lasting through October. The temperature ranges from 21 °C to 42 °C in plains and 8 °C to 32 °C in the hills, and the mean annual rainfall is 3644 mm with maximum rain in July and minimum in January82. The study area is adjacent to the Parambikulam Tiger Reserve.

Data was collected in Nemmara-Nelliyampathy Main Road (From Pothundi dam till KFDC Pakuthipalam via Seethargundu and Karadi hanging bridge). The survey was carried out in a stretch which is about 50 km (one way). The adjacent land cover comprised 44% natural intact rainforest, 37.6% coffee plantation, 12.9% tea plantation, 2.1% rubber and paddy plantations, and 1.1% orange plantation.

Methodology

We carried out the roadkill surveys from June 2023 to May 2024. The road was systematically searched during the day using a motor vehicle at a steady speed of about 10 km/h. We surveyed the road stretch twice in a month. In total, there were 22 systematic roadkill surveys which were carried out only on Mondays. Counted and recorded dead animals were removed from the road to avoid repetition. Each encountered carcass was identified to the species level (whenever possible), otherwise to the genus level or the family level. We used standard texts for identification83–85. All the documented roadkills were grouped by their class as amphibians, reptiles, birds, and mammals. Each carcass encountered was photographed for identification. We have not collected or preserved any physical samples of any of the roadkill animals during our survey as we did not have permits for collections.

Along with the AVC mortality data, 22 different environmental parameters were recorded at every 500 m distance along the stretch of the survey road in order to cover a large area for sampling. These environmental parameters were collected as a single set of measurements during the study period and were not repeatedly collected at each carcass detection, thereby representing static rather than dynamic environmental conditions. The environmental parameters collected through visual estimation were—the type of vegetation (natural forest or plantation), plantation type (paddy, coffee, rubber, tea, orange), terrain (rocky, grassy, muddy), road type (tar, concrete, boulder), road structure (flat or dome), road pattern (linear or winding), electric line (present or absent), nearest water source type (dam, pond, river, falls), nearest human settlement type (houses, colony, village, town), trench (present or absent), garbage (low, medium, high), sidewalk (present or absent). Through quadrat sampling, we collected vegetation density (low, medium, high), leaf litter (low, medium, high), and undergrowth (low, medium, high). Road width and sidewalk widths were measured using a measuring tape. GPS (Garmin MONTANA 650) was used to record the elevation, nearest water source distance, and nearest human settlement distance. The surrounding slope was measured using a clinometer and a spherical densiometer was used to measure canopy coverage. For parameters classified (low, medium, and high), the range from the minimum to the maximum value was divided into three equal intervals, which were subsequently designated as low, medium, and high.

Ethics declaration

The study was completely based on the observation of dead animals on the road without collecting them, and thus did not involve any handling of animals. The study was carried out with the approval of the research committee of the University of Mysore and approval of the methods by the Forest Department, Kerala.

Data analysis

Our analysis was conducted in R (version 4.3.1)86 using RStudio87 version 2023.06.0 + 421. Random Forest was employed in this study as an initial modeling technique to identify the most influential predictors among a potentially large set of variables88, by using the “RANDOM FOREST” package89 in R. The Random Forest models were trained using the “RANDOM FOREST” package with 500 trees (mtry = 13, seed = 123) to evaluate variable importance based on both %IncMSE (Percentage Increase in Mean Squared Error) and IncNodePurity metrics. No formal hyperparameter tuning was performed, and default parameters were used aside from the number of trees. However, model tuning was explored in the initial stages when the dataset included 22 main response variables, each further subdivided by ecological strata, e.g., canopy cover (high, medium, low), resulting in a larger set of response variables (n = 52). These divisions were treated as separate variables to refine model accuracy while preserving ecological relevance, particularly concerning species-specific habitat preferences. %IncMSE (Percentage Increase in Mean Squared Error), measured the impact of each variable on prediction accuracy, and IncNodePurity showed the overall decrease in node impurity that each variable contributed. Based on their significance, the top five predictors were identified. A total of ten models were developed: one global Random Forest, three individual Random Forests (one per response: amphibians, reptiles, birds, and mammals), and three Poisson and three negative binomial GLMs (one per response, using 5 predictors selected via Random Forest importance). In Random Forest models, the %IncMSE metric properly measures the predictors’ contribution to model performance, making it adequate for evaluating variable relevance90–92. To visualize the findings, variance importance plots were generated. Further, the identified top five predictors were used to run the Generalized Linear Model (GLM), and Poisson GLMs were fitted using the base glm() function in R. Before performing each GLM, we verified multicollinearity among predictor variables, retaining only those variables with acceptable Variance Inflation Factor (VIF) values using the “CAR” package and ecological significance, ensuring that the models were both statistically sound and ecologically significant. Overdispersion was tested using the “DHARMa” package, whenever overdispersion was detected, we used Negative Binomial models instead of Poisson models using glm.nb() function from the “MASS” package77, as Negative Binomial models include an additional parameter to handle excess variance, improving model accuracy and reliability93.

The study area map, roadkill maps, and heat maps were created using QGIS94 by overlaying all the different seasonal data of all other taxa. The estimate of roadkill per kilometer per year was determined by total roadkills recorded in one year / total road length surveyed in one year (km).

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1 (533.2KB, pdf)

Acknowledgements

We thank Kerala Forest Department for providing permission (KFDHQ-4025/2021-CWW/WL10) to conduct research and support in the field. This research is supported by Science and Engineering Research Board, Government of India grant under Distinguished Fellowship to Mewa Singh. Mewa Singh also thanks Indian National Science Academy for the award of Distinguished Professorship during which this article was prepared. We are also thankful to the Kerala Forest Developmental Corporation (KFDC) for providing the accommodation facility and for their support. Thanks are due to Mr. Gaurav Singh for helping in some identification and to Mr. MK Darshan, Mr. Venu Gopal Reddy, Mr. RP Prajwal, Ms. NA. Swaathi for help in some field visits.

Author contributions

MS and HNK conceived the study. SS collected the field data. SS and GP primarily analysed the data. SS, GP, GSR, SM, HNK and MS contributed to article writing and editing. MS procured the research funds.

Data availability

All raw data are available with the corresponding author and can be made accessible with a reasonable request. It is not publically available otherwise.

Declarations

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  • 1.Seto, K. C., Güneralp, B. & Hutyra, L. R. Global forecasts of urban expansion to 2030 and direct impacts on biodiversity and carbon pools. PNAS109, 16083–16088 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Piketty, T. Capital in the twenty-first century: A multidimensional approach to the history of capital and social classes. Br. J. Sociol.65, 736–747 (2014). [DOI] [PubMed] [Google Scholar]
  • 3.CIA The world factbook. Central Intelligence Agency. https://www.cia.gov/library/publicatio%20ns/the-world-factbook/rankorder/2085r%20ank.html. Accessed on 25 August 2024 (2020).
  • 4.Ministry of Road Transport & Highways. https://morth.nic.in/. Accessed on 26 September 2024 (2021).
  • 5.ROADS-Statistical Year Book, 2018. Ministry of Statistics & Programme Implementation, 2024. https://www.mospi.gov.in/statistical-year-book-india/2018/190. Accessed on 15 December 2024.
  • 6.Laurance, W. F. et al. A global strategy for road building. Nature513, 229–232. 10.1038/nature13717 (2014). [DOI] [PubMed] [Google Scholar]
  • 7.Drews, C. Road kills of animals by public traffic in Mikumi National Park, Tanzania, with notes on baboon mortality. Afr. J. Ecol.33, 89–100. 10.1111/j.1365-2028.1995.tb00785.x (1995). [Google Scholar]
  • 8.Fedigan, L. M. & Zohar, S. Sex differences in mortality of Japanese macaques: Twenty-one years of data from the Arashiyama West population. Am. J. Phys. Anthropol.102, 161–175. 10.1002/(SICI)10968644(199702)102:2%3c161::AID-AJPA2%3e3.0.CO;2-1 (1997). [DOI] [PubMed] [Google Scholar]
  • 9.Forman, R. T. & Alexander, L. E. Roads and their major ecological effects. Annu. Rev. Ecol. Syst.29, 207–231. 10.1146/annurev.ecolsys.29.1.207 (1998). [Google Scholar]
  • 10.Printes, R. C. The Lami biological reserve, Rio Grande do Sul, Brazil and the danger of power lines to howlers in urban reserves. Neotrop. Primates7, 135–136 (1999). [Google Scholar]
  • 11.Chhangani, A. K. Killing of Hanuman Langur in road accidents in Knumbhalgarh Wildlife Sanctuary, Rajasthan, India. Primate Rep.69, 49–57 (2004). [Google Scholar]
  • 12.Lokschin, L. X., Rodrigo, C. P., Cabral, J. N. H. & Buss, G. Power lines and howler monkey conservation in Porto Alegre, Rio Grande do Sul, Brazil. Neotrop. Primates14, 76–80. 10.1896/044.014.0206 (2007). [Google Scholar]
  • 13.Parker, L., Nijman, V. & Nekaris, K. A. I. When there is no forest left: Fragmentation, local extinction, and small population sizes in the Sri Lankan western purple-faced langur. Endanger. Species Res.5, 29–36. 10.3354/esr00107 (2008). [Google Scholar]
  • 14.Caceres, N. C. et al. Mammal occurrence and roadkill in two adjacent ecoregions (Atlantic Forest and Cerrado) in south-western Brazil. Zoologia27, 709–717. 10.1590/S1984-46702010000500007 (2010). [Google Scholar]
  • 15.Pragatheesh, A. Effect of human feeding on the road mortality of Rhesus Macaques on National Highway-7 routed along Pench Tiger Reserve, Madhya Pradesh, India. JoTT3, 1656–1662. 10.11609/JoTT.o2669.1656-62 (2011). [Google Scholar]
  • 16.Umapathy, G., Hussain, S. & Shivaji, S. Impact of habitat fragmentation on the demography of lion-tailed macaque (Macaca silenus) populations in the rainforests of Anamalai Hills, Western Ghats India. Int. J. Primatol.32, 889–900. 10.1007/s10764-011-9508-9 (2011). [Google Scholar]
  • 17.Trumbulak, S. C. & Frissell, C. A. Review of ecological effects of roads on terrestrial and aquatic communities. Conserv. Biol.14, 18–30 (2000). [Google Scholar]
  • 18.Hill, J. E., DeVault, T. L. & Belant, J. L. Cause-specific mortality of the world’s terrestrial vertebrates. Glob. Ecol. Biogeogr.28, 680–689. 10.1111/geb.12881 (2019). [Google Scholar]
  • 19.Rytwinski, T. & Fahrig, L. The impacts of roads and traffic on terrestrial animal populations. In Handbook of Road Ecology (eds van der Ree, R. et al.) 237–246 (Springer, 2015). [Google Scholar]
  • 20.Reh, W. & Seitz, A. The influence of land use on the genetic structure of populations of the common frog Rana temporaria. Biol. Conserv.54, 239–249 (1990). [Google Scholar]
  • 21.Burnett, S. E. Effects of a rainforest road on movements of small mammals: Mechanisms and implications. Wildl. Res.19, 95–104 (1992). [Google Scholar]
  • 22.Reijnen, R., Foppen, R., Braak, C. T. & Thissen, J. The effects of car traffic on breeding bird populations in woodland. III. Reduction of density in relation to the proximity of main roads. J. Appl. Ecol.32, 187–202 (1995). [Google Scholar]
  • 23.Newmark, W. D., Boshe, J. I., Sariko, H. I. & Makumbule, G. K. Effects of a highway on large mammals in Mikumi National Park, Tanzania. Afr. J. Ecol.34, 15–31 (1996). [Google Scholar]
  • 24.Richardson, J. H., Shore, R. F., Treweek, J. R. & Larkin, S. B. C. Are major roads a barrier to small mammals?. J. Zool.243, 840–846 (1997). [Google Scholar]
  • 25.Desai, A. A. & Baskaran, N. Ecology of Malabar Giant Squirrel (Ratufa indica) in Mudumalai Wildlife Sanctuary, South India. Technical Report Bombay Natural History Society, Bombay (1998).
  • 26.Carr, L. W. & Fahrig, L. Effect of road traffic on two amphibian species of differing vagility. Conserv. Biol.15, 1071–1078 (2001). [Google Scholar]
  • 27.Schwartz, A. L., Williams, H. F., Chadwick, E., Thomas, R. J. & Perkins, S. E. Roadkill scavenging behaviour in an urban environment. J. Urban Ecol.4, juy006. 10.1093/jue/juy006 (2018). [Google Scholar]
  • 28.Andrews, K. M. & Gibbons, J. W. How do highways influence snake movement? Behavioral responses to roads and vehicles. Copeia4, 772–782 (2005). [Google Scholar]
  • 29.Ibisch, P. L. et al. A global map of roadless areas and their conservation status. Science354, 1423–1427 (2016). [DOI] [PubMed] [Google Scholar]
  • 30.Loss, S. R., Will, T. & Marra, P. P. Estimation of bird-vehicle collision mortality on US roads. J. Wildl. Manag.78, 763–771. 10.1002/jwmg.721 (2014). [Google Scholar]
  • 31.Grilo, C., Koroleva, E., Andrášik, R., Bíl, M. & González-Suárez, M. Roadkill risk and population vulnerability in European birds and mammals. Front. Ecol. Environ.18(6), 323–328 (2020). [Google Scholar]
  • 32.Medrano-Vizcaíno, P. et al. Roadkill patterns in Latin American birds and mammals. Glob. Ecol. Biogeogr.31(9), 1756–1783 (2022). [Google Scholar]
  • 33.Bishop, C. A. & Brogan, J. M. Estimates of Avian Mortality Attributed to Vehicle Collisions in Canada Estimation de la mortalité aviaire attribuable aux collisions automobiles au Canada. Avian Conserv. Ecol.8, 2. 10.5751/ACE-00604-080202 (2013). [Google Scholar]
  • 34.Kumara, H. N., Sharma, A. K., Kumar, A. & Singh, M. Roadkills of wild fauna in Indira Gandhi wildlife sanctuary, Western Ghats, India: Implications for management. Biosphere Conserv.3, 41–47 (2000). [Google Scholar]
  • 35.Vijayakumar, S. P., Vasudevan, K. & Ishwar, N. M. Herpetofaunal mortality on roads in the Anamalai Hills, southern Western Ghats. Hamadryad26, 265–272 (2001). [Google Scholar]
  • 36.Sunder, K. S. G. Mortality of herpetofauna, Birds and Mammals due to vehicular traffic in Etawah District Uttra Pradesh, India. JBNHS103, 392–398 (2004). [Google Scholar]
  • 37.Seshadri, K. S. & Ganesh, T. Faunal mortality on roads due to religious tourism across time and space in protected areas: A case study from south India. For. Ecol. Manag.262, 1713–1721 (2011). [Google Scholar]
  • 38.Gubbi, S., Poornesha, H. C. & Madhusudan, M. D. Impact of vehicular traffic on the use of highway edges by large mammals in a South Indian wildlife reserve. Curr. Sci.102, 1047–1051 (2012). [Google Scholar]
  • 39.Jeganathan, P., Mudappa, D., Kumar, M. A. & Raman, T. R. S. Seasonal variation in wildlife roadkills in plantations and tropical rainforest in the Anamalai Hills, Western Ghats, India. Curr. Sci.114, 619–626 (2018). [Google Scholar]
  • 40.Ascensão, F. et al. Preventing wildlife roadkill can offset mitigation investments in short-medium term. Biol. Conserv.253, 108902 (2021). [Google Scholar]
  • 41.Medrano-Vizcaíno, P., Grilo, C., Brito-Zapata, D. & González-Suárez, M. Landscape and road features linked to wildlife mortality in the Amazon. Biodivers. Conserv.32(13), 4337–4352 (2023). [Google Scholar]
  • 42.Coelho, I. P. et al. Anuran road-kills neighboring a peri-urban reserve in the Atlantic Forest, Brazil. J. Environ. Manag.112, 17–26 (2012). [DOI] [PubMed] [Google Scholar]
  • 43.de Carvalho, N. C., Bordignon, M. O. & Shapiro, J. T. Fast and furious: A look at the death of animals on the highway MS-080, Southwestern Brazil. Iheringia Série Zool104, 43–49. 10.1590/1678-4766201410414349 (2014). [Google Scholar]
  • 44.Freitas, D. et al. How landscape features influence road-kill of three species of mammals in the Brazilian savanna?. Oecologia Aust.18, 35–45. 10.4257/oeco.2014.18.05.02 (2015). [Google Scholar]
  • 45.Braz, V. D. S. & Rodrigues, F. Wild vertebrate roadkill in the Chapada dos Veadeiros National Park, Central Brazil. Biota Neotrop.16, 1–11 (2016). [Google Scholar]
  • 46.Mulero-Pázmány, M. et al. Road orientation affects the impact of roads on wildlife. Wildl. Res.10.1071/WR21149 (2022). [Google Scholar]
  • 47.Langen, T. A., Gunson, K. E., Jackson, S. D., Smith, D. J. & Ruediger, W. Planning and Designing Mitigation of Road Effects on Small Animals. Roads and Ecological Infrastructure. Concepts and Applications for Small Animals 146–177 (Johns Hopkins University Press, 2015). [Google Scholar]
  • 48.Van Der Ree, R., Smith, D. J. & Grilo, C. Handbook of Road Ecology (Wiley, 2015). [Google Scholar]
  • 49.Myers, N., Mittermeier, R. A., Mittermeier, C. G., Da Fonseca, G. A. & Kent, J. Biodiversity hotspots for conservation priorities. Nature403(6772), 853–858. 10.1038/35002501 (2000). [DOI] [PubMed] [Google Scholar]
  • 50.Nampoothiri, K. M., Ramkumar, B. & Pandey, A. Western Ghats of India: Rich source of microbial biodiversity. J. Sci. Ind. Res.72, 617–623 (2013). [Google Scholar]
  • 51.Raman, T. S. & Sukumar, R. Responses of tropical rainforest birds to abandoned plantations, edges and logged forest in the Western Ghats, India. Anim. Conserv.5(3), 201–216. 10.1017/S1367943002002251 (2002). [Google Scholar]
  • 52.Reddy, C. S., Jha, C. S. & Dadhwal, V. K. Assessment and monitoring of long-term forest cover changes (1920–2013) in Western Ghats biodiversity hotspot. J. Earth Syst. Sci.125, 103–114. 10.1007/s12040-015-0645-y (2016). [Google Scholar]
  • 53.Baskaran, N. & Boominathan, D. Road kill of animals by highway traffic in the tropical forests of Mudumalai Tiger Reserve, southern India. JoTT2, 753–759 (2010). [Google Scholar]
  • 54.Roshnath, R. & Cyriac, P. V. Way back home-butterfly roadkills. Zoo’s Print28, 18–20 (2013). [Google Scholar]
  • 55.Bansal, U. D. I. T. A. A study of reptile road mortalities on an inter-state highway in the Western Ghats, India and suggestion of suitable mitigation measures. C&FH4, 15–30 (2020). [Google Scholar]
  • 56.Karthik, L. P. P. Effect of vehicular traffic on wild animals in Anaikatty Hills, Southern Western Ghats, India. Indian J. Ecol.48, 108–111 (2021). [Google Scholar]
  • 57.IUCN. The IUCN Red List of Threatened Species. Version 2024-1. https://www.iucnredlist.org. Accessed on 12th September 2024 (2024).
  • 58.Wildlife (Protection) Amendment Act, 2022. 2, No. 18 Act of Parliament, India (2022).
  • 59.Silva, M. O., Oliveira, I. D., Cardoso, M. W. & Graf, V. Road kills impact over the herpetofauna of Atlantic Forest (PR-340, Antonina, Paraná). Acta Biol. Paran.36, 103–112 (2007). [Google Scholar]
  • 60.Brehme, C. S., Tracey, J. A., McClenaghan, L. R. & Fisher, R. N. Permeability of roads to movement of scrubland lizards and small mammals. Conserv. Biol.27, 710–720. 10.1111/cobi.12081 (2013). [DOI] [PubMed] [Google Scholar]
  • 61.Meek, R. Patterns of reptile road-kills in the Vendée region of western France. J. Herpetol.19, 135–142 (2009). [Google Scholar]
  • 62.Secco, H., Ratton, P., Castro, E., Da Lucas, P. S. & Bager, A. Intentional snake road-kill: a case study using fake snakes on a Brazilian road. Trop. Conserv. Sci.7, 561–571. 10.1177/194008291400700313 (2014). [Google Scholar]
  • 63.Campioni, L., Marengo, I., Román, J. & D’Amico, M. Mud-puddling on roadsides: a potential ecological trap for butterflies. J. Insect Conserv.26, 131–134 (2022). [Google Scholar]
  • 64.Li, P. et al. The relative importance of green infrastructure as refuge habitat for pollinators increases with local land-use intensity. J. Appl. Ecol.57, 1494–1503 (2020). [Google Scholar]
  • 65.Arca-Rubio, J., Moreno-Rueda, G. & Ortega, Z. The distribution of vertebrate roadkill varies by season, surrounding environment, and animal class. Eur. J. Wildl. Res69, 42 (2023). [Google Scholar]
  • 66.Gibbs, J. P. & Shriver, W. G. Can road mortality limit populations of pool-breeding amphibians?. Wetl. Ecol. Manag.13, 281–289 (2005). [Google Scholar]
  • 67.Mazerolle, M. J., Huot, M. & Gravel, M. Behavior of amphibians on the road in response to car traffic. Herpetologica61, 380–388. 10.1655/04-79.1 (2005). [Google Scholar]
  • 68.Zhang, W. et al. Daytime driving decreases amphibian roadkill. PeerJ6, e5385 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 69.MacKenzie, D. I. et al. Occupancy Estimation and Modeling: Inferring Patterns and Dynamics of Species Occurrence (Elsevier, 2017). [Google Scholar]
  • 70.Lima, S. L., Blackwell, B. F., DeVault, T. L. & Fernández-Juricic, E. Animal reactions to oncoming vehicles: A conceptual review. Biol. Rev. Camb. Philos. Soc.90, 60–76 (2015). [DOI] [PubMed] [Google Scholar]
  • 71.Slater, F. M. An assessment of wildlife road casulties–the potential discrepancy between numbers counted and numbers killed. Web Ecol.3, 33–42 (2002). [Google Scholar]
  • 72.Sabino-Marques, H. & Mira, A. Living on the verge: are roads a more suitable refuge for small mammals than streams in Mediterranean pastureland?. Ecol. Res.26, 277–287 (2011). [Google Scholar]
  • 73.Ascensao, F., Clevenger, A. P., Grilo, C., Filipe, J. & Santos-Reis, M. Highway verges as habitat providers for small mammals in agrosilvopastoral environments. Biodivers. Conserv.21, 3681–3697 (2012). [Google Scholar]
  • 74.Husby, M. Factors affecting road mortality in birds. Ornis Fenn.93, 212–224 (2016). [Google Scholar]
  • 75.Husby, A. & Husby, M. Interspecific analysis of vehicle avoidance behavior in birds. Behav. Ecol.25, 504–508 (2014). [Google Scholar]
  • 76.Pinto, T., Santos, S. M., Mira, A. & Sillero, N. Importance of water availability for amphibian roadkill in a mediterranean landscape. Biodivers. Conserv.32, 2513–2537 (2023). [Google Scholar]
  • 77.van Langevelde, F. & Jaarsma, C. F. Using traffic flow theory to model traffic mortality in mammals. Landsc. Ecol.19, 895–907. 10.1007/s10980-005-0464-7 (2005). [Google Scholar]
  • 78.Teixeira, F. Z., Coelho, A. V. P., Esperandio, I. B. & Kindel, A. Vertebrate Road mortality estimates: effects of sampling methods and carcass removal. Biol. Conserv.157, 317–323 (2013). [Google Scholar]
  • 79.Vijayakumar, S., Yabesh, J. M., Prabhu, S., Manikandan, R. & Muralidharan, B. Quantitative ethnomedicinal study of plants used in the Nelliyampathy hills of Kerala, India. J. Ethnopharmacol.161, 238–254 (2015). [DOI] [PubMed] [Google Scholar]
  • 80.Ramachandran, K. K. & Suganthasakthivel, R. Ecology and behaviour of the arboreal mammals of the Nelliyampathy forests (No. 382). KFRI Research, Report (2010).
  • 81.Sushanth, S. et al. Mass mortality of the endangered Purple Frog, Nasikabatrachus sahyadrensis (Biju and Bossuyt, 2003), in the Nelliyampathy Hills, Southern Western Ghats, India. Amphibian Reptile Conserv.19, 48–58 (2025). [Google Scholar]
  • 82.Varghese, S. Working Plan of Nenmara Forest Division 2015–2025. Kerala Forest Department, Thiruvananthapuram (2015).
  • 83.Menon, V. & Daniel, J. C. Field guide to Indian mammals. Dorling Kindersley, India in association with Penguin Book, India (2003).
  • 84.Grimmett, R., Inskipp, C. & Inskipp, T. Birds of Indian Subcontinent (Oxford University Press, 1998). [Google Scholar]
  • 85.Daniel, J. C. The Book of Indian Reptiles and Amphibians, 238. Oxford University Press, BNHS (2002).
  • 86.R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.Rproject.org (2023).
  • 87.RStudio Team. RStudio: Integrated Development for R. RStudio, PBC. http://www.rstudio.com/ (2023).
  • 88.Venables, W. N. & Ripley, B. D. Modern applied statistics with S. In Statistics and Computing, 4th edn. (Springer, 2002).
  • 89.Breiman, L. Random forest. Mach. Learn.45, 5–32 (2001). [Google Scholar]
  • 90.Wright, M. N. & Ziegler, A. ranger: A fast implementation of random forests for high dimensional data in C++ and R. J. Stat. Softw.77, 1–17 (2015). [Google Scholar]
  • 91.Gregorutti, B., Michel, B. & Saint-Pierre, P. Correlation and variable importance in random forests. Stat. Comput.27, 659–678 (2017). [Google Scholar]
  • 92.Nicodemus, K. K., Malley, J. D., Strobl, C. & Ziegler, A. The behaviour of random forest permutation-based variable importance measures under predictor correlation. BMC Bioinform.11, 1–13 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 93.Ver Hoef, J. M. & Boveng, P. L. Quasi-Poisson vs. negative binomial regression: How should we model overdispersed count data?. Ecology88, 2766–2772 (2007). [DOI] [PubMed] [Google Scholar]
  • 94.QGIS.org. QGIS Geographic Information System. Version 3.26.2. Open Source Geospatial Foundation Project. http://qgis.org (2024).

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1 (533.2KB, pdf)

Data Availability Statement

All raw data are available with the corresponding author and can be made accessible with a reasonable request. It is not publically available otherwise.

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group

RESOURCES