Ecological and Climatic Drivers of Wildlife Road Mortality in Kenya’s Athi-Kapiti Plains

Ecological and Climatic Drivers of Wildlife Road Mortality in Kenya’s Athi-Kapiti Plains | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Ecological and Climatic Drivers of Wildlife Road Mortality in Kenya’s Athi-Kapiti Plains Aggrey W. Chemwa, Flora N. Namu, Odd T. Jacobson, Genevieve E. Finerty, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6944631/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 10 You are reading this latest preprint version Abstract Roads and habitat fragmentation pose significant conservation challenges for many animals, yet the environmental factors that influence roadkill risk remain poorly understood. In Kenya, the Nairobi–Mombasa Highway cuts through the ranges of numerous wildlife species within one of the country’s most biodiverse regions, the Athi-Kapiti Plains. To investigate the environmental factors influencing roadkill risk in this region, we combined remote sensing data on long-term drought (Standardized Precipitation Evapotranspiration Index, SPEI) and vegetation greenness (Normalized Difference Vegetation Index, NDVI) with 6,240 km of road transect surveys conducted in 2023 and historical roadkill records from 2020–2022. We recorded 218 wildlife roadkills involving 11 species, predominantly plains zebras ( Equus quagga , 69%), spotted hyenas ( Crocuta crocuta ), and Maasai giraffes ( Giraffa camelopardalis tippelskirchi). Prolonged drought, reduced vegetation productivity, and wildlife use intensity emerged as key predictors of roadkill risk along the Nairobi–Mombasa Highway across the Kenya’s Athi-Kapiti Plains. These factors interact to increase animal movement across roads, especially during dry periods when animals seek water and forage. We also found roadkills to be clustered near artificial water points, suggesting these sites may function as ecological traps by concentrating wildlife in high-risk zones. Our findings suggest that drought-driven declines in food availability may force animals to travel farther and cross roads in search of resources, increasing roadkill risk. This interaction between drought and road mortality is particularly concerning, as both represent significant threats that may intensify under climate change. Our results highlight how climatic stress, habitat degradation, and wildlife use intensity jointly shape roadkill patterns. Bayesian modelling drought (SPEI) Mombasa Highway NDVI Spatial-temporal drivers Wildlife-vehicle collisions Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 1. Introduction Linear infrastructure such as roads, railways, and pipelines is essential for socioeconomic development, yet it often incurs significant ecological costs, including habitat fragmentation, loss of ecological connectivity, and direct wildlife mortality (Laurance et al. 2014 ). Among these impacts, direct wildlife mortality from vehicle collisions is a frequently overlooked but substantial driver of population decline, particularly for wide-ranging and migratory species (Seiler and Helldin 2006 ; Loss et al. 2014 ). In some regions, roadkill exceeds natural predation or poaching as the leading cause of mortality for certain species (Forman et al. 2003 ). Roads not only fragment habitats physically but also facilitate human encroachment, increase human-wildlife conflict, and create barriers to movement that threaten long-term population viability. These effects are especially concerning in biodiversity-rich areas, where expanding infrastructure intersects with critical wildlife habitats and migratory corridors (Forman et al. 2003 ). In Kenya, road networks have expanded by 50% over the past four decades, increasingly ncroaching on ecologically sensitive regions such as the Athi-Kapiti Wildlife Conservancies along the Mombasa highway. The Athi-Kapiti landscape is recognized as a biodiversity hotspot, supporting endangered species such as the Maasai giraffe ( Giraffa tippelskirchi ) and migratory herds of wildebeest ( Connochaetes taurinus ) and elands ( Taurotragus oryx ) (Damania et al. 2019 ). This vital corridor of private and communal conservancies connects Nairobi National Park to the Amboseli, Tsavo, and Serengeti ecosystems. The ecological integrity of the Athi-Kapiti landscape is increasingly compromised by the Nairobi-Mombasa Highway. This high-traffic highway has fragmented habitats and elevated collision risks for wildlife, underscoring the urgent need to reconcile infrastructure development with biodiversity conservation. The Athi-Kapiti landscape supports diverse ecological communities, including large herds of herbivores such as zebras, wildebeest and eland, which rely on seasonal migrations to access forage and water. The movements of these species are closely synchronized with rainfall patterns, leading to cyclical shifts in road-crossing behaviour (Birkett et al. 2012 ). Wildebeest, for example, traverse vast distances during dry seasons in search of diminishing resources (Martin et al. 2015 ), while elands, Africa’s largest antelope, are often funnelled into hazardous roadside zones by habitat fragmentation and edge effects (Lee and Bond 2018 ). The Nairobi–Mombasa Highway exacerbates these risks, functioning as both a physical barrier and a demographic filter (Clevenger and Wierzchowski 2006 ; Jacobson et al. 2016 ). Scavengers such as spotted hyenas ( Crocuta crocuta ) and critically endangered white-backed vultures ( Gyps africanus ) are also vulnerable to road mortality, as their reliance on roadkill carcasses increases their exposure to vehicle collisions (Ogada et al. 2012 ; Naciri et al. 2023 ). Such losses not only threaten population viability but also disrupt key ecosystem functions, including seed dispersal, nutrient cycling, and predator–prey dynamics (Beasley et al. 2019 ) Roadkill risk is therefore shaped by a combination of species traits, behavioural patterns, and environmental conditions. For herbivores such as wildebeest and zebra, risk is elevated during resource-scarce dry seasons when road-crossings increase in search of forage and water (Bolger et al. 2008 ; Martin et al. 2015 ), while browsers such as giraffe and eland face risks year-round due to fragmented roadside vegetation (Kioko et al. 2008 ; Lee and Bond 2018 ). Vegetation structure can impact roadkill risk at multiple scales. At a landscape scale, it can influence large-scale movement patterns, increasing likelihood or road crossing (Blackburn et al. 2022 ) and at more local scale it can influence risk of collision by obscuring visibility (Canal et al. 2019 ). Remote sensing indices like the Normalized Difference Vegetation Index (NDVI), a proxy for vegetation productivity, have proven useful in capturing these seasonal habitat dynamics, with low NDVI values associated with periods of heightened roadkill risk in other systems (Ascensão et al. 2019 ). Rainfall patterns and proximity to water sources similarly influence crossing behavior, as animals traverse roads to reach ephemeral waterholes during dry periods (Valeix et al. 2011 ; Sánchez-Montoya et al. 2023 ). Here, we present the first study in the Athi-Kapiti region to integrate remote sensing indices; Normalized Difference Vegetation Index (NDVI), Standardized Precipitation Evapotranspiration Index (SPEI) with systematic field-based roadkill surveys. We investigated the spatio-temporal drivers of wildlife roadkills along a 60-kilometre segment of the Nairobi–Mombasa Highway, which traverses the Athi-Kapiti Conservancies. Specifically, we aimed to:(1) identify the species most vulnerable to vehicle collisions, (2) map seasonal and spatial variation in roadkills across three sections with differential levels of collision risk, and (3) quantify the effects of vegetation structure, drought severity, and wildlife use intensity on collision rates. Our findings aim to inform specific mitigation actions such as the construction of wildlife overpasses, speed bumps, implementation of seasonal speed restrictions, and habitat restoration initiatives, to support Kenya’s infrastructural development while safeguarding wild animals and their habitats. 2. Methods 2.1 Study area Our study was conducted in the Athi-Kapiti Plains Conservancies, located at latitude − 1.467663° and longitude 37.101091° (Fig. 1 ). The area encompasses the Swara Plains Conservancy, the International Livestock Research Institute (Kapiti Estate), Maanzoni Conservancy, Mwambi Hill Conservancy, Malinda Conservancy, and adjacent community lands. Theconservancies, constituting the Athi Kapiti Plains, are situated south of Nairobi city and Nairobi National Park, extending east into Machakos County and south toward Amboseli National Park via the gently descending Emarti Valley (Said et al. 2016 ). The Athi-Kapiti Plains are ecologically significant, supporting a diverse assemblage of wildlife and livestock within a semi-arid savannah landscape. The region receives an average annual rainfall of 550 mm and sits at an elevation of 1,650 to 1,900 meters above sea level. The soils are predominantly black cotton in the plains and red cotton in the ridges, supporting a variety of savannah grasses and tree species. Common grasses include Themeda triandra (Thatching grass), Cenchrus stramineus (Sandhill millet), Cynodon dactylon (Bermuda grass), and Cyperus rotundus (Purple nutsedge). Prominent tree species include Balanites aegyptiaca , Acacia xanthophloea, Acacia drepanolobium , Grewia kakothamnos , and Commiphora kataf (Kapiti Brochure, 2020). The Athi-Kapiti Plains supports a rich diversity of wildlife, including the Maasai giraffe ( Giraffa tippelskirchi ), common zebra ( Equus quagga ), common eland ( Taurotragus oryx ), wildebeest ( Connochaetes taurinus ), impala ( Aepyceros melampus ), and kongoni ( Alcelaphus buselaphus cokii ). Predators such as cheetah (Acinonyx jubatus), lion ( Panthera leo ), leopard ( Panthera pardus ), and spotted hyena ( Crocuta crocuta ) are also present. It serves as a critical wildlife dispersal and migration corridor, linking Nairobi National Park to larger ecosystems such as Amboseli, Tsavo, Magadi and Natron, underscoring their ecological importance (Ogutu et al. 2014 ; Sircely et al. 2020 ). The Nairobi-Mombasa Highway, a major transportation corridor connecting Nairobi to the port city of Mombasa, bisects the Athi-Kapiti Plains. The highway is a two-lane road that transitions into a single lane at Makutano Junction. It experiences heavy traffic, with an estimated daily flow of over 31,000 vehicles according to the Kenya Roads Board. Speed limits are set at 80 km/h for commercial and passenger vehicles and 110 km/h for private vehicles, as per Kenya’s Traffic Act (CAP 403). The high traffic volume and speeds creates significant risks for wildlife, particularly where the highway intersects known movement corridors. 2.2 Roadkill Data Collection We conducted roadkill surveys between January and December 2023, covering both wet and dry seasons (Duethmann et al. 2020 ). Two road transect surveys were conducted each week along a 60 km stretch. Surveys were carried out by a single observer driving at a low speed of 30–40 km/hr) to maximize detection probabilities. The protocol established by (Santos et al. 2011 ), was followed, with surveys taking place early in the morning at sunrise to optimize roadkill detection before scavengers removed the carcasses or decay set in (Hels and Buchwald 2001 ). Upon detecting a carcass, the location was recorded by taking coordinate, using a handheld Global Positioning System (GPS). Information on species name, sex, age group, date, and time of the carcass was documented, and the carcass removed from the road to prevent double counting. Additional secondary roadkill data from 2020 to 2022 were provided by the Kenya Wildlife Service (Machakos Station) and included information on the date, GPS coordinates, species, age, and sex of the animals, which was merged with data collected during the study. 2.3 Defining Roadkill Hotspots and Sampling Zones To better understand spatial variation in roadkill risk, we performed a kernel density analysis to identify high-incidence sections along the 60km segment of the Mombasa Highway where roadkills occurred (Lala et al. 2021 ; Galinskaitė et al. 2022 ). We imported the collected roadkill data (coordinates, species, sex, age, date, time) into ArcGIS Pro and used the Kernel Density tool to visualize roadkill intensity along the highway. This process was used to define three sections with varying levels of roadkill density, in order of highest to lowest: Swara-Maanzoni Interface (A), Kapiti-Mwambi Transition (B) and Community Edge Zone (C), hereafter referred to as A, B, and C, respectively (Fig. 3 ) (Lala et al. 2021 ). 2.4 Sampling Site-level Landscape and Habitat Characteristics We calculated five ecological characteristics for each site ID using data collected from field surveys: (1) wildlife use intensity, (2) woody vegetation density, (3) herbaceous vegetation cover, (4) visibility, and (5) artificial water source density. The first four characteristics were sampled using two parallel transects (3 km long and 2 km wide) established on either side of the highway, resulting in six transects across the three sections. Sampling points were systematically placed at 200-meter intervals along each transect, yielding 30 points per section and 90 points in total. At each of the 90 points, we sampled woody vegetation density, herbaceous vegetation cover, wildlife use intensity and visibility. Artificial water source density was assessed separately by recording the coordinates of artificial water points using a handheld GPS. To explore potential spatial clustering, these coordinates were used to generate a heatmap of water point density in ArcGIS Pro, which was then visually overlaid with roadkill hotspot locations. Woody vegetation density was measured using a Point-Cantered Quarter (PCQ) method. The area around each sampling point was divided into four 90° quadrants, and the distance to the nearest tree was measured in each quadrant following (Silva et al. 2017 ), where the distance from the quadrat centre to the nearest tree with a height of 1.3 m and above (≥ 1.3m) was measured. For each tree, diameter at Breast Height (DBH) was recorded, and species were identified. Measurements were taken using a tape measure and range finder (Kiani et al. 2013 ; Wainscott 2015 ). Additionally, all the trees within 20m radius of the quadrats were counted. Woody vegetation density (D) was calculated assuming a random spatial distribution using the following formula: D = 1/rˉ2⋅π. Where rˉ is the mean distance to the nearest tree (in meters), and π ≈ 3.1416. Density was expressed as individuals per square meter (ind/m²) Herbaceous vegetation cover At each of the 90 sampling points, eight 1 × 1 m quadrats were randomly placed, yielding a total of 720 quadrats. Within each quadrat, herbaceous species and their heights were recorded. Herbaceous vegetation cover at each sampling point was estimated by averaging the cover values from the eight quadrats. Visibility was measured by calculating the mean distance to the nearest line-of-sight obstruction. At each sampling point, distances were recorded in 24 directions from the centre using a rangefinder (Riginos 2015 ). The 24 measurements were then averaged to obtain a visibility value for each point. Wildlife use intensity : Wildlife presence at each of the three sections (A, B and C) was assessed by counting dung pellet groups within 50 m × 4 m belt transects placed at 200-meter intervals along each 3 km transect. This yielded 15 sampling points per transect, resulting in 30 points per section and 90 points across all sections. Dung counts were used instead of direct observations to minimize disturbance (Takenoshita and Yamagiwa 2008 ) and to detect species that are present but not easily observable (Eggert et al. 2003 ). Wildlife use intensity was calculated as dung density using the formula: Dung Density = Number of Dung Groups/Area of the Transect, where: Area of Transect = Length × Width = 50m × 4m = 200 m². We used total dung per plot as a relative assay of individual herbivore species use of the different plots. Thus, our approach averts issues related to inferring population densities from dung counts, such as differential decomposition rates across seasons and habitats (Nchanji and Plumptre 2001 ; Hema et al. 2017 ). There is ample evidence from our study system (Augustine 2003 ; Young et al. 2005 ; Riginos et al.) and elsewhere (Cromsigt et al. 2009 ; Hema et al. 2017 ) that dung counts are robust for comparing relative habitat use by large mammals within a species. 2.5 Remote Sensing Data and Spatial Data Processing Environmental covariates were derived for each site using a combination of remote sensing data from Google Earth Engine (GEE) as well as R (v4.5.0) packages: terra, and sf (Haviv et al. 2025 ). Drought conditions were assessed using the Standardized Precipitation Evapotranspiration Index (SPEI), derived from ERA5 climate reanalysis datasets (Keune et al. 2025 ). SPEI as an integrative drought index that is calculated by subtracting potential evapotranspiration from precipitation providing a more comprehensive measure of climatic water availability than precipitation alone (Nosrati 2014 ). Positive SPEI values indicate wetter than average conditions and negative values indicate drier than average conditions. SPEI values were calculated at 1, 3, 6 and 12-month accumulation intervals to capture both short- and long-term water stress (Keune et al. 2025 ). We extracted SPEI values from ERA5-derived NetCDF layers and calculated the monthly mean SPEI for each section using a 2km buffer on either side of the highway (consistent with the methodology for field data collection). Vegetation productivity was evaluated using the Normalized Difference Vegetation Index (NDVI), derived from Sentinel-2 surface reflectance imagery between 2020 and 2023. The analysis focused on a 2 km-wide corridor along both sides of the Mombasa Highway. Monthly NDVI composites were generated in Google Earth Engine (GEE) by spatially filtering imagery using the digitized extent polygon and temporally selecting specific monthly intervals. Scenes with more than 15% cloud cover were excluded using the CLOUDY_PIXEL_PERCENTAGE metadata. NDVI was computed for each image using the standard formula: NDVI = (B8 – B4) / (B8 + B4), based on the Near-Infrared (Band 8) and Red (Band 4) bands. Monthly mean NDVI rasters were then generated using a mean reducer, exported as GeoTIFFs, and processed in R. We then calculated the monthly mean NDVI within the polygon for each zone (2km buffer on each side of the highway). 3. Statistical Analysis To estimate the relationship between road kills and our environmental variables, we used a set of Bayesian hurdle Poisson regression models using the brms package in R (Bürkner 2021 ). Hurdle models accommodate excess zeros (as is common in event data such as roadkills) by modeling both the occurrence (zero vs non-zero) and the count of roadkills where they occur. The response variable, roadkill counts (n_roadkills), was regressed against two climatic stress variables SPEI and NDVI in separate models. For SPEI, we ran four separate models using accumulation periods of 1, 3, 6, and 12 months to evaluate differences in the effects of short-term water stress versus long-term drought. All models included an indicator variable for site ID (i.e., A, B, C) to account for spatial variation of road kills. This allowed us to indirectly assess the influence of ecological factors such as wildlife use intensity, visibility, and vegetation cover (herbaceous and woody) by relating the effects of site ID to measurements from transect surveys post hoc. The models were specified as: SPEI model : log(n_roadkills) ~ spei+ site_ID NDVI model : log(n_roadkills) ~ ndvi_std + site_ID, with a binomial hurdle component (hu) estimating zero-inflation. Weakly informative priors were applied: Normal (0, 1) for fixed effects (climatic variables, density terms), Normal (0, 1.2) for intercepts, and Beta (1, 1) for the hurdle probability. Four Markov chains were run (2,000 iterations, 1,000 warmup) using cmdstanr (v0.7.1), with convergence confirmed by Rhat 1,000. Model fit was evaluated via posterior predictive checks. Analyses were conducted in R (v4.3.1) with tidybayes and ggplot2 for visualization. We used the Kruskal–Wallis test to compare wildlife use intensity, herbaceous vegetation cover, woody vegetation density, and visibility across the six sampling points selected along the highway based on roadkill intensity. Dunn’s test was applied for post hoc pairwise comparisons to identify statistically significant differences among groups. 4. Results 4.1 Species-specific and temporal trends in roadkill incidences along the Mombasa Highway in the Athi-Kapiti Region A total of 11 animal species were involved in roadkills. Zebras had the highest incidents of roadkills, accounting for 69% of the total followed by the spotted hyena at 11%, and the Maasai giraffe at 8% (Table 1 ). The total roadkills recorded between the year 2020 and 2023 were 218. Year 2020 had the lowest kills with 6 carcasses while 2023 had the highest with 84 carcasses (Fig. 2 ). Table 1 Animal species killed during the study and their percentage along Mombasa Highway in Athi- Kapiti Plains Conservancies Number Species Name Total roadkill Percentage Roadkill 1 Cheetah 2 0.92 2 Duiker 1 0.46 3 Eland 7 3.21 4 Giraffe 17 7.8 5 Hyena 23 10.55 6 Impala 2 0.92 7 Kongoni 1 0.46 8 Ostrich 1 0.46 9 Serval Cat 1 0.46 10 Wildebeest 13 5.96 11 Zebra 150 68.81 Total 218 100 4.2 Roadkill risk highly impacted by drought conditions and vegetation greenness We found that long-term drought severity and reduced vegetation productivity led to more predicted road kills. Specifically, we observed a considerable negative relationship between road kill frequency and both SPEI (β SPEI_6 = -0.61[-0.79,-0.42]) and NDVI (β NDVI = -0.92[-1.21,-0.64]) (Fig. 3 and Fig. S1 ). The strength of the negative drought effect (SPEI) increased considerably from the 1-month to the 3-month timescale, and again from 3 to 6 months, but no clear difference was observed between the 6-month and 12-month timescales (Fig. 4 ). 4.3 Ecological characteristics across sites: Swara-Maanzoni Interface (A), Kapiti-Mwambi Transition (B) and Community Edge Zone (C). Wildlife use intensity varied significantly across the three sections (χ² = 41.6, df = 2, p < 0.001), with section A having the highest mean wildlife use intensity (0.0765 ± 0.0493), followed by B (0.0410 ± 0.0558) and C (0.0090 ± 0.0120). Pairwise comparisons showed that wildlife use intensity was significantly higher in A compared to both B (p < 0.001) and C (p < 0.001), while B was also significantly higher than C (p = 0.0206) (Fig. 5 )). Herbaceous vegetation cover also differed significantly among the three sections (χ² = 20.9, df = 2, p < 0.001), with section B exhibiting the highest mean herbaceous vegetation cover (81.4 ± 14.6%), followed by C (64.6 ± 17.4%) and A (63.4 ± 13.4%). Herbaceous vegetation cover was significantly higher in B compared to both A (p < 0.001) and C (p < 0.001), while A and C did not differ significantly (p = 1.000) (Fig. 5 ). Visibility was significantly different across the three sections (χ² = 54.0, df = 2, p < 0.001), with C having the highest visibility (437.0 ± 262.0), followed by A (69.3 ± 64.3) and B (54.0 ± 34.1). Visibility in C was significantly higher than both A (p < 0.001) and B (p < 0.001), while A and B did not differ (p = 1.000) (Fig. 5 ). Woody vegetation density varied significantly across the three sections (χ² = 20.9, df = 2, p = 0.0000284), with section B exhibiting the highest mean woody vegetation density (0.0360 ± 0.0364), followed by A (0.0296 ± 0.0392), and C having the lowest (0.0634 ± 0.139). Post-hoc comparisons indicated that C had significantly lower woody vegetation density than both A (p = 0.00171) and B (p = 0.0000446), while there was no significant difference between A and B (p = 1.000) (Fig. 5 )). 4.4 Roadkill risk impacted by spatial variation in wildlife use intensity and ecological characteristics across sites Roadkill frequency varied significantly across sites. Compared to site A, site B had a markedly lower rate of road kills (β B : -1.18 [-1.87 to -0.61]), and site C had the lowest roadkill frequency (β C : -1.98 [-3.20 to -0.99]). These estimates indicate that, conditional on NDVI and SPEI, road kills at site B and C were approximately 69% and 86% lower, respectively, than at site A. Transect measurements revealed that site A (the highest roadkill risk) was associated with the highest wildlife use intensity, while site C (the lowest roadkill risk) had the greatest visibility and lowest woody vegetation cover. Site B (moderate roadkill risk) was characterized by the highest herbaceous vegetation cover (Fig. 5 and Fig. 6 ). 5. Discussion This study highlights how prolonged drought, vegetation degradation, and wildlife use intensity interact to influence roadkill patterns in the Athi-Kapiti Plains. The increasingly strong negative association between roadkills and SPEI with longer accumulation periods suggests that sustained drought (see Fig. S2 ), rather than short dry spells, prompts extensive wildlife movement that increases road exposure and ultimately leads to higher roadkill rates. During exceedingly dry periods, access to water and forage becomes limited and unevenly distributed, causing animals to expand their ranges and encounter roads more frequently (Mulero-Pázmány et al. 2022 ). The concentration of roadkills near permanent water sources further emphasizes roads as lethal barriers to critical resources during droughts, exacerbating mortality in fragmented landscapes (Clevenger and Wierzchowski 2006 ; Okita-Ouma et al. 2016 ). At the same time, low NDVI values indicated reduced vegetation greenness and declining forage availability, especially among herbaceous species (Egeru et al. 2015 ). These conditions fragmented suitable feeding areas and extended the distances animals needed to travel, increasing the likelihood of road encounters leading heightened roadkill rates (Ascensão et al. 2019 ). Together, drought and vegetation loss may have created a reinforcing effect, where animals displaced by poor habitat quality were funnelled into higher-risk areas near roads. Wildlife use intensity was also a key driver. We observed a strong positive relationship between wildlife abundance and roadkill rates, with no evidence of a threshold beyond which risk plateaus. This suggests that even moderate increases in local intensity can significantly raise the likelihood of collisions, particularly when combined with drought and food scarcity. As competition intensifies in densely populated areas, animals may travel more widely, further increasing their road exposure. Species-specific traits shaped susceptibility to collisions. Giraffes, although less frequently involved in roadkills, are especially vulnerable due to their tall stature, slow movement, and limited agility (More et al. 2013 ). Observations during the study showed giraffes foraging on acacia trees near the road, likely because preferred woody browse was less available farther away, especially during dry months (Bond et al. 2016 ). This behavior increased their proximity to traffic and risk of collisions. Zebras, which accounted for the majority of roadkill incidents, were especially affected by the loss of herbaceous vegetation and water scarcity during the dry season. During this period, they are forced to traverse roads more frequently in search of remnant forage and water, a behavior intensified by habitat fragmentation (Boyers and Parrini 2024 ). Their gregarious nature and tendency to move in large herds further increase the likelihood of road crossings, raising the risk of vehicle collisions (Kioko et al. 2015 ). According to the 2021 Kenya Wildlife Census conducted during the study period, zebras were the most abundant species in the Athi-Kapiti Plains, with a population of 11,355, significantly higher than any other species. Hence there high population density lead to high roadkill mortality. Spotted hyenas were likely drawn to carcasses along roads, as their scavenging behavior leads them to revisit sites of previous mortality. This creates a feedback loop where roadkills attract scavengers, which then become roadkill themselves (Naciri et al. 2023 ). As primarily nocturnal animals, hyenas are especially vulnerable during periods of low visibility. These results underscore the importance of implementing carcass removal protocols to reduce attractant cues and prevent cascading mortality (Hager et al. 2012 ). Roadkill rates varied across the three study sections. Section A had the highest number of incidents, likely due to high wildlife use intensity, limited forage, and the presence of artificial watering points that drew wildlife closer to the road, particularly during dry seasons. Its open landscape also made it attractive to herbivores seeking visibility to detect predators. Section B experienced moderate roadkill levels, despite having relatively high vegetation cover. This may reflect its moderate wildlife use intensity, and although higher herbaceous cover may reduce road crossings during wet periods, it can attract grazers as vegetation senesces, creating delayed dry-season hotspots (Munyati 1997 ). Additionally, both Section A and Section B act as part of a larger conservancy corridor linking Swara Plains, Maanzoni, Mwambi Hill, and Kapiti Plains. These areas maintain intact wildlife populations and minimal anthropogenic disturbance, which likely channels animal movement between protected habitats, creating predictable crossing zones and elevating roadkill risk (HP-Ruffino et al. 2016 ; Lala et al. 2021 ). In contrast, Section C recorded the lowest roadkill frequency. This site had sparse vegetation and higher levels of human activity, which may have discouraged wildlife presence due to habitat degradation and negative human–wildlife interactions (Murray and St. Clair 2015 ). These conditions likely reduced both the intensity of animals and their need to cross roads in the area. These spatial patterns suggest that conditions even a few kilometres from roads can shape animal movement and influence collision risk. Using a Bayesian modeling framework enabled us to identify both the probability of roadkill occurrence and the ecological factors that heighten roadkill risk. This approach provided robust inference while explicitly accounting for uncertainty in parameter estimates, which is critical when modeling complex ecological systems. The models identified key ecological drivers specifically, prolonged drought conditions and reduced vegetation productivity that influence roadkill risk which can inform targeted conservation strategies. Management efforts should prioritize high-risk zones like Section A. Recommended actions include constructing wildlife crossings, restoring vegetation, redistributing water points more evenly across the ecosystem, reducing speed limits during dry seasons, and removing carcasses promptly. Involving local communities in land-use planning and ecological monitoring will be essential to ensure sustainable and effective mitigation. We demonstrate that drought, habitat quality, wildlife use intensity, and animal behavior jointly influence roadkill risk in savannah ecosystems. In the Athi-Kapiti region, these ecological drivers are further intensified by anthropogenic pressures, reflecting the complex realities wildlife faces in increasingly fragmented landscapes. Our integrated approach combining field-based data with climate and vegetation modeling provides a transferable framework for other arid regions facing similar conditions. As climate change intensifies the frequency and severity of drought, wildlife movements in search of water and forage are likely to rise, leading to greater risk of collisions. These findings underscore the need to incorporate both biodiversity conservation and climate resilience into regional infrastructure planning. Declarations Statements and Declarations Funding This research was funded by Save Giraffes Now as part of the author's professional role within the organization. No grant number was issued Competing Interests The authors have no competing interests to declare. Author Contributions Aggrey W. Chemwa: Conceptualization (equal), Data curation (lead), Formal analysis (lead), Methodology (equal), Writing – original draft (lead), Writing – review and editing (equal); Flora N. Namu: Conceptualization (equal), Methodology (equal), Writing – review and editing (equal); Odd T. Jacobson: Conceptualization (equal), Data curation (equal), Formal analysis (equal), Methodology (equal), Writing – review and editing (equal); Genevieve E. Finerty: Conceptualization (equal), Formal analysis (equal), Methodology (equal), Writing – review and editing (equal); Duncan M. Kimuyu: Conceptualization (equal), Formal analysis (equal), Methodology (equal), Writing – review and editing (equal). Acknowledgements We extend our heartfelt gratitude to all those who contributed to this research. We are especially thankful to Save Giraffes Now, and to its CEO Susan Myers and her husband Richard Myers, for their generous financial support, which made this study possible. We sincerely appreciate the Athi-Kapiti Conservancies for granting us access to conduct fieldwork on their land. Special thanks to the late Steve Tankard (KWS Honorary Warden) and the Kenya Wildlife Service (KWS) for their invaluable assistance in providing secondary roadkill data. We are also grateful to Vincent Kipkazi for his dedicated support during fieldwork. Additional thanks go to Fiona Sandeman, Kayla Walisch, Philip Winter, Ilona Gluecks, and Nobert Muriungi for their valuable contributions. Your support was instrumental to the success of this research. Data Availability The datasets generated and analysed during the current study are publicly available in the Figshare repository at https://doi.org/10.6084/m9.figshare.29373971. Ethical Compliance This study did not involve direct handling of animals. References Ascensão F, Yogui D, Alves M et al (2019) Predicting spatiotemporal patterns of road mortality for medium-large mammals. J Environ Manage 248:109320. https://doi.org/10.1016/j.jenvman.2019.109320 Augustine DJ (2003) Long-term, livestock‐mediated redistribution of nitrogen and phosphorus in an East African savanna. J Appl Ecol 40:137–149. https://doi.org/10.1046/j.1365-2664.2003.00778.x Beasley JC, Olson ZH, Selva N, DeVault TL (2019) Ecological functions of vertebrate scavenging. Carrion Ecol Manag 125–157. https://doi.org/10.1007/978-3-030-16501-7_6 Birkett PJ, Vanak AT, Muggeo VM et al (2012) Animal perception of seasonal thresholds: changes in elephant movement in relation to rainfall patterns. PLoS ONE 7:e38363. https://doi.org/10.1371/journal.pone.0038363 Blackburn A, Veals AM, Tewes ME et al (2022) If you build it, will they come? A comparative landscape analysis of ocelot roadkill locations and crossing structures. PLoS ONE 17:e0267630. https://doi.org/10.1371/journal.pone.0267630 Bolger DT, Newmark WD, Morrison TA, Doak DF (2008) The need for integrative approaches to understand and conserve migratory ungulates. Ecol Lett 11:63–77. https://doi.org/10.1111/j.1461-0248.2007.01109.x Bond ML, Strauss MK, Lee DE (2016) Soil correlates and mortality from giraffe skin disease in Tanzania. J Wildl Dis 52:953–958. https://doi.org/10.7589/2016-02-047 Boyers M, Parrini F (2024) Resource selection at fine scale: what drives the decision of a generalist herbivore? J Mammal 105:1336–1344. https://doi.org/10.1093/jmammal/gyae091 Bürkner P-C (2021) Bayesian item response modeling in R with brms and Stan. J Stat Softw 100:1–54. https://doi.org/10.18637/jss.v100.i05 Canal D, Camacho C, Martín B et al (2019) Fine-scale determinants of vertebrate roadkills across a biodiversity hotspot in Southern Spain. Biodivers Conserv 28:3239–3256. https://doi.org/10.1007/s10531-019-01817-5 Clevenger AP, Wierzchowski J (2006) Maintaining and restoring connectivity in landscapes fragmented by roads. Conserv Biol Ser-Camb- 14:502. https://doi.org/10.1017/cbo9780511754821.023 Cromsigt JP, van Rensburg SJ, Etienne RS, Olff H (2009) Monitoring large herbivore diversity at different scales: comparing direct and indirect methods. Biodivers Conserv 18:1219–1231. https://doi.org/10.1007/s10531-008-9506-1 Damania R, Desbureaux S, Scandizzo PL et al (2019) When Good Conservation Becomes Good Economics: Kenya’s Vanishing Herds. World Bank. https://doi.org/10.1596/33083 Duethmann D, Blöschl G, Parajka J (2020) Why does a conceptual hydrological model fail to correctly predict discharge changes in response to climate change? Hydrol Earth Syst Sci 24:3493–3511. https://doi.org/10.5194/hess-24-3493-2020 Egeru A, Wasonga O, Mburu J et al (2015) Drivers of forage availability: An integration of remote sensing and traditional ecological knowledge in Karamoja sub-region, Uganda. Pastoralism 5:1–18. https://doi.org/10.1186/s13570-015-0037-6 Eggert LS, Eggert JA, Woodruff DS (2003) Estimating population sizes for elusive animals: the forest elephants of Kakum National Park, Ghana. Mol Ecol 12:1389–1402. https://doi.org/10.1046/j.1365-294x.2003.01822.x Forman RT, Sperling D, Bissonette JA et al (2003) Road ecology. Sci Solut 482 Galinskaitė L, Ulevičius A, Valskys V et al (2022) The influence of landscape structure on wildlife–vehicle collisions: Geostatistical analysis on hot spot and habitat proximity relations. ISPRS Int J Geo-Inf 11:63. https://doi.org/10.3390/ijgi11010063 Hager SB, Cosentino BJ, McKay KJ (2012) Scavenging affects persistence of avian carcasses resulting from window collisions in an urban landscape. J Field Ornithol 83:203–211. https://doi.org/10.1111/j.1557-9263.2012.00370.x Haviv D, Remšík J, Gatie M et al (2025) The covariance environment defines cellular niches for spatial inference. Nat Biotechnol 43:269–280. https://doi.org/10.1038/s41587-024-02193-4 Hels T, Buchwald E (2001) The effect of road kills on amphibian populations. Biol Conserv 99:331–340 Hema EM, Di Vittorio M, Barnes RF et al (2017) Detection of interannual population trends in seven herbivores from a West African savannah: a comparison between dung counts and direct counts of individuals. Afr J Ecol 55:609–617. https://doi.org/10.1111/aje.12397 HP-Ruffino P, Rodríguez-C KG, Ramazzotto LA, Freitas PD (2016) Roadkill hotspots in a protected area of Cerrado in Brazil: planning actions to conservation. Rev MVZ Córdoba 21:5441–5448. https://doi.org/10.21897/rmvz.609 Jacobson SL, Bliss-Ketchum LL, de Rivera CE, Smith WP (2016) A behavior‐based framework for assessing barrier effects to wildlife from vehicle traffic volume. Ecosphere 7:e01345. https://doi.org/10.1002/ecs2.1345 Keune J, Di Giuseppe F, Barnard C et al (2025) ERA5–Drought: Global drought indices based on ECMWF reanalysis. Sci Data 12:616 Kiani B, Fallah A, Tabari M et al (2013) A comparison of distance sampling methods in Saxaul (Halloxylon ammodendron CA Mey Bunge) Shrub-Lands. Pol J Ecol 61:207–219 Kioko J, Kiffner C, Jenkins N, Collinson WJ (2015) Wildlife roadkill patterns on a major highway in northern Tanzania. Afr Zool 50:17–22 Kioko J, Muruthi P, Omondi P, Chiyo PI (2008) The performance of electric fences as elephant barriers in Amboseli, Kenya. South Afr J Wildl Res-24-Mon. Delayed Open Access 38:52–58. https://doi.org/10.3957/0379-4369-38.1.52 Lala F, Chiyo PI, Kanga E et al (2021) Wildlife roadkill in the Tsavo Ecosystem, Kenya: identifying hotspots, potential drivers, and affected species. Heliyon 7. https://doi.org/10.1016/j.heliyon.2021.e06364 Laurance WF, Clements GR, Sloan S et al (2014) A global strategy for road building. Nature 513:229–232 Lee DE, Bond ML (2018) Quantifying the ecological success of a community-based wildlife conservation area in Tanzania. J Mammal 99:459–464. https://doi.org/10.1093/jmammal/gyy014 Loss SR, Will T, Marra PP (2014) Estimation of bird-vehicle collision mortality on US roads. J Wildl Manag 78:763–771. https://doi.org/10.1002/jwmg.721 Martin J, Benhamou S, Yoganand K, Owen-Smith N (2015) Coping with spatial heterogeneity and temporal variability in resources and risks: adaptive movement behaviour by a large grazing herbivore. PLoS ONE 10:e0118461. https://doi.org/10.1371/journal.pone.0118461 More HL, O’Connor SM, Brøndum E et al (2013) Sensorimotor responsiveness and resolution in the giraffe. J Exp Biol 216:1003–1011. https://doi.org/10.1242/jeb.067231 Mulero-Pázmány M, Rollán L, D’Amico M, González-Suárez M (2022) Road orientation affects the impact of roads on wildlife. Wildl Res 50:39–46. https://doi.org/10.1071/wr21149 Munyati C (1997) Wetland change assessment on the Kafue Flats. a remote sensing approach, Zambia Murray MH, St. Clair CC (2015) Individual flexibility in nocturnal activity reduces risk of road mortality for an urban carnivore. Behav Ecol 26:1520–1527. https://doi.org/10.1093/beheco/arv102 Naciri M, Planillo A, Gicquel M et al (2023) Three decades of wildlife-vehicle collisions in a protected area: Main roads and long-distance commuting trips to migratory prey increase spotted hyena roadkills in the Serengeti. Biol Conserv 279:109950. https://doi.org/10.1016/j.biocon.2023.109950 Nchanji AC, Plumptre AJ (2001) Seasonality in elephant dung decay and implications for censusing and population monitoring in south-western Cameroon. Afr J Ecol 39:24–32. https://doi.org/10.1046/j.1365-2028.2001.00265.x Nosrati K (2014) Assessment of Standardized Precipitation Evapotranspiration Index (SPEI) for Drought Identification in Different Climates of IranK. Environ Sci 12 Ogada DL, Keesing F, Virani MZ (2012) Dropping dead: causes and consequences of vulture population declines worldwide. Ann N Y Acad Sci 1249:57–71. https://doi.org/10.1111/j.1749-6632.2011.06293.x Ogutu JO, Piepho H-P, Said MY, Kifugo SC (2014) Herbivore dynamics and range contraction in Kajiado County Kenya: climate and land use changes, population pressures, governance, policy and human-wildlife conflicts. Open Ecol J 7. https://doi.org/10.2174/1874213001407010009 Okita-Ouma B, Lala F, Moller R et al (2016) Preliminary indications of the effect of infrastructure development on ecosystem connectivity in Tsavo National Parks, Kenya. Pachyderm 57:109–111. https://doi.org/10.69649/pachyderm.v57i.396 Riginos C (2015) Climate and the landscape of fear in an A frican savanna. J Anim Ecol 84:124–133 Riginos C, Veblen KE, Young TP The KLEE experiment Said MY, Ogutu JO, Kifugo SC et al (2016) Effects of extreme land fragmentation on wildlife and livestock population abundance and distribution. J Nat Conserv 34:151–164. https://doi.org/10.1016/j.jnc.2016.10.005 Sánchez-Montoya MM, Datry T, Ruhi A et al (2023) Intermittent rivers and ephemeral streams are pivotal corridors for aquatic and terrestrial animals. Bioscience 73:291–301. https://doi.org/10.1093/biosci/biad004 Santos SM, Carvalho F, Mira A (2011) How long do the dead survive on the road? Carcass persistence probability and implications for road-kill monitoring surveys. PLoS ONE 6:e25383. https://doi.org/10.1371/journal.pone.0025383 Seiler A, Helldin JO (2006) Mortality in wildlife due to transportation. The ecology of transportation: Managing mobility for the environment. Springer, pp 165–189. https://doi.org/10.1007/1-4020-4504-2_8 Silva LB, Alves M, Elias RB, Silva L (2017) Comparison of-square, point centered quarter, and-tree sampling methods in Pittosporum undulatum invaded woodlands. Int J For Res 2017: https://doi.org/10.1155/2017/2818132 Sircely JA, Gluecks IV, Kimengich N (2020) ILRI Kapiti Plains Research Station: Grazing plan for early 2020 Takenoshita Y, Yamagiwa J (2008) Estimating gorilla abundance by dung count in the northern part of Moukalaba-Doudou National Park, Gabon. Afr Study Monogr Suppl Issue 39:41–54 Valeix M, Fritz H, Sabatier R et al (2011) Elephant-induced structural changes in the vegetation and habitat selection by large herbivores in an African savanna. Biol Conserv 144:902–912. https://doi.org/10.1016/j.biocon.2010.10.029 Wainscott BC (2015) Surveying Forest Diversity and Health Using the Point-Centered Quarter Method. Proc Assoc Biol Lab Educ 36:18 Young TP, Palmer TM, Gadd ME (2005) Competition and compensation among cattle, zebras, and elephants in a semi-arid savanna in Laikipia, Kenya. Biol Conserv 122:351–359. https://doi.org/10.1016/j.biocon.2004.08.007 Additional Declarations No competing interests reported. Supplementary Files Fig.S1.jpg Fig.S2.jpg Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 01 Mar, 2026 Reviews received at journal 16 Feb, 2026 Reviewers agreed at journal 31 Jan, 2026 Reviewers agreed at journal 27 Jan, 2026 Reviews received at journal 17 Sep, 2025 Reviewers agreed at journal 29 Aug, 2025 Reviewers invited by journal 29 Aug, 2025 Editor assigned by journal 05 Jul, 2025 Submission checks completed at journal 23 Jun, 2025 First submitted to journal 21 Jun, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6944631","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":507342451,"identity":"c45b903b-5fc1-4d75-adbd-1a0c6af79495","order_by":0,"name":"Aggrey W. Chemwa","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA4klEQVRIiWNgGAWjYHAD5gNAQkKGFC1sCSAtPKRo4TEAkwTV8TcAVf74VZvPL5Hz+dWNGgseBvbDRzfg0yJxgMeAsbfvuOXMGbnbrHOOAR3Gk5Z2A58WAwbeDQy8PccMDG7kbjPOYQNqkeAxI6iF8S9Qi/2NnGfGOf+I1MLM86PGwEAih/lxbhsRWiQO8384LNtwwEDizDMz5tw+CR42Qn7hb29LfPjmT50Bf3vy48853+rk+NkPH8OrhYGZgeEAY9thBgaBBDYJkAAbXuVw8KcOaN8B5g/EqR4Fo2AUjIKRBgAWEkWEOPoghQAAAABJRU5ErkJggg==","orcid":"","institution":"Karatina University","correspondingAuthor":true,"prefix":"","firstName":"Aggrey","middleName":"W.","lastName":"Chemwa","suffix":""},{"id":507342452,"identity":"14cd0455-5e01-4afc-9286-cd6e8af9fe1f","order_by":1,"name":"Flora N. Namu","email":"","orcid":"","institution":"Karatina University","correspondingAuthor":false,"prefix":"","firstName":"Flora","middleName":"N.","lastName":"Namu","suffix":""},{"id":507342453,"identity":"e863b96c-db0a-43cd-8bb1-048b5d50dfea","order_by":2,"name":"Odd T. Jacobson","email":"","orcid":"","institution":"Max Planck Institute for Animal Behaviour","correspondingAuthor":false,"prefix":"","firstName":"Odd","middleName":"T.","lastName":"Jacobson","suffix":""},{"id":507342454,"identity":"32bd389f-b342-4837-a077-eb9755874565","order_by":3,"name":"Genevieve E. Finerty","email":"","orcid":"","institution":"Max Planck Institute for Animal Behaviour","correspondingAuthor":false,"prefix":"","firstName":"Genevieve","middleName":"E.","lastName":"Finerty","suffix":""},{"id":507342455,"identity":"c7fc7ba9-53d9-4d74-8769-877511fa1ee6","order_by":4,"name":"Duncan M. Kimuyu","email":"","orcid":"","institution":"Karatina University","correspondingAuthor":false,"prefix":"","firstName":"Duncan","middleName":"M.","lastName":"Kimuyu","suffix":""}],"badges":[],"createdAt":"2025-06-21 11:23:19","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6944631/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6944631/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":90710232,"identity":"342835a2-c14c-4623-9bac-46f137df7a39","added_by":"auto","created_at":"2025-09-06 06:27:35","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":3961203,"visible":true,"origin":"","legend":"\u003cp\u003eMap of the study area showing the Mombasa Highway intersecting Athi-Kapiti Plains conservation area\u003c/p\u003e","description":"","filename":"Fig.1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6944631/v1/acffd2c81de88fa0b5e75930.jpg"},{"id":90709914,"identity":"08b720dd-2040-44e5-9bac-f1a6dc13e7c2","added_by":"auto","created_at":"2025-09-06 06:19:35","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1087692,"visible":true,"origin":"","legend":"\u003cp\u003eAnnual wildlife roadkill totals recorded along the Mombasa Highway in the Athi-Kapiti Plains from 2020 to 2023 showing the highest incidents in 2022 and 2023 and the lowest in 2020\u003c/p\u003e","description":"","filename":"Fig.2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6944631/v1/0f6cc484ac378a9b07ead767.jpg"},{"id":90709911,"identity":"0cfe4ec6-e078-4958-88c1-6c9a71105eec","added_by":"auto","created_at":"2025-09-06 06:19:35","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":31037,"visible":true,"origin":"","legend":"\u003cp\u003ePredicted number of roadkills as a function of environmental variables and location. (A) Relationship between the 6-month accumulation Standardized Precipitation-Evapotranspiration Index (SPEI) and predicted roadkill frequency. (B) Relationship between Normalized Difference Vegetation Index (NDVI) and predicted roadkill frequency. (C) Comparison of predicted roadkill numbers across three different plots (A, B, and C), with plot A exhibiting significantly higher predicted roadkill counts compared to plots B and C. Error bars represent 95% credible intervals. Shaded areas in panels A and B represent the 95% credible intervals around the predicted means from the fitted models.\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6944631/v1/cb90c6c20263b34a3608db27.jpg"},{"id":90709920,"identity":"b7efa05a-1bb9-4454-8f6c-3e75981734f2","added_by":"auto","created_at":"2025-09-06 06:19:35","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":127067,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of posterior estimates for the effect of Standardized Precipitation Evapotranspiration Index (SPEI) accumulation periods (1, 3, 6, and 12 months) on wildlife-vehicle collision rates. Negative values indicate higher roadkill frequency under drier conditions.\u003c/p\u003e","description":"","filename":"Fig.4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6944631/v1/1ac4a1a75c5c6c2eabdf89c4.jpg"},{"id":90709930,"identity":"7bac9256-938b-4dd0-8186-afd5df4cd651","added_by":"auto","created_at":"2025-09-06 06:19:36","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":230884,"visible":true,"origin":"","legend":"\u003cp\u003eWildlife use intensity, wooden vegetation density, herbaceous vegetation cover and visibility at Swara-Maanzoni Interface (A), Kapiti-Mwambi Transition (B) and Community Edge Zone (C), along Mombasa road in Athi Kapiti Plains. The x-axis labels display the site ID and relative risk of roadkill (Low, Moderate, and High) according to model predictions.\u003c/p\u003e","description":"","filename":"Fig.5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6944631/v1/a945c20eb78540ba21313b78.jpg"},{"id":90710237,"identity":"3c2f43c3-2bfc-4db4-8767-d3bd403184ad","added_by":"auto","created_at":"2025-09-06 06:27:36","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":3461054,"visible":true,"origin":"","legend":"\u003cp\u003eSpatial distribution of wildlife roadkill intensity at Swara-Maanzoni Interface (A), Kapiti-Mwambi Transition (B) and Community Edge Zone (C) along Mombasa Highway in Athi-Kapiti plains.\u003c/p\u003e","description":"","filename":"Fig.6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6944631/v1/a2b68af4d6a95bca8f37bee5.jpg"},{"id":90710235,"identity":"094298b7-deef-4f9a-90ce-9cb7ae3a03de","added_by":"auto","created_at":"2025-09-06 06:27:36","extension":"jpg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":3796788,"visible":true,"origin":"","legend":"\u003cp\u003eSpatial distribution of roadkill hotspots in relation to watering points along the Mombasa Highway within the Athi-Kapiti Plains Conservancies.\u003c/p\u003e","description":"","filename":"Fig.7.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6944631/v1/c4c873c66d92a5921fc31544.jpg"},{"id":90710673,"identity":"154ba478-80cb-4996-8ebb-1d6aed3e0123","added_by":"auto","created_at":"2025-09-06 06:43:39","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":13545262,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6944631/v1/ec68e9a7-588f-45c7-9030-2dcfc5401706.pdf"},{"id":90709912,"identity":"5fe225e9-6ab9-407f-a2b2-569569121474","added_by":"auto","created_at":"2025-09-06 06:19:35","extension":"jpg","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":59035,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.S1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6944631/v1/a0a8ff7ed9d9967aeaed3826.jpg"},{"id":90709934,"identity":"c93a446b-7ab5-4a80-aae6-88bdb62e60c7","added_by":"auto","created_at":"2025-09-06 06:19:36","extension":"jpg","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":2525277,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.S2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6944631/v1/fc0d21f6e30e609b977a6769.jpg"}],"financialInterests":"No competing interests reported.","formattedTitle":"Ecological and Climatic Drivers of Wildlife Road Mortality in Kenya’s Athi-Kapiti Plains","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eLinear infrastructure such as roads, railways, and pipelines is essential for socioeconomic development, yet it often incurs significant ecological costs, including habitat fragmentation, loss of ecological connectivity, and direct wildlife mortality (Laurance et al. \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). Among these impacts, direct wildlife mortality from vehicle collisions is a frequently overlooked but substantial driver of population decline, particularly for wide-ranging and migratory species (Seiler and Helldin \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2006\u003c/span\u003e; Loss et al. \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). In some regions, roadkill exceeds natural predation or poaching as the leading cause of mortality for certain species (Forman et al. \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2003\u003c/span\u003e). Roads not only fragment habitats physically but also facilitate human encroachment, increase human-wildlife conflict, and create barriers to movement that threaten long-term population viability. These effects are especially concerning in biodiversity-rich areas, where expanding infrastructure intersects with critical wildlife habitats and migratory corridors (Forman et al. \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2003\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eIn Kenya, road networks have expanded by 50% over the past four decades, increasingly ncroaching on ecologically sensitive regions such as the Athi-Kapiti Wildlife Conservancies along the Mombasa highway. The Athi-Kapiti landscape is recognized as a biodiversity hotspot, supporting endangered species such as the Maasai giraffe (\u003cem\u003eGiraffa tippelskirchi\u003c/em\u003e) and migratory herds of wildebeest (\u003cem\u003eConnochaetes taurinus\u003c/em\u003e) and elands (\u003cem\u003eTaurotragus oryx\u003c/em\u003e) (Damania et al. \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). This vital corridor of private and communal conservancies connects Nairobi National Park to the Amboseli, Tsavo, and Serengeti ecosystems. The ecological integrity of the Athi-Kapiti landscape is increasingly compromised by the Nairobi-Mombasa Highway. This high-traffic highway has fragmented habitats and elevated collision risks for wildlife, underscoring the urgent need to reconcile infrastructure development with biodiversity conservation.\u003c/p\u003e\u003cp\u003eThe Athi-Kapiti landscape supports diverse ecological communities, including large herds of herbivores such as zebras, wildebeest and eland, which rely on seasonal migrations to access forage and water. The movements of these species are closely synchronized with rainfall patterns, leading to cyclical shifts in road-crossing behaviour (Birkett et al. \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). Wildebeest, for example, traverse vast distances during dry seasons in search of diminishing resources (Martin et al. \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2015\u003c/span\u003e), while elands, Africa\u0026rsquo;s largest antelope, are often funnelled into hazardous roadside zones by habitat fragmentation and edge effects (Lee and Bond \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). The Nairobi\u0026ndash;Mombasa Highway exacerbates these risks, functioning as both a physical barrier and a demographic filter (Clevenger and Wierzchowski \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2006\u003c/span\u003e; Jacobson et al. \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Scavengers such as spotted hyenas (\u003cem\u003eCrocuta crocuta\u003c/em\u003e) and critically endangered white-backed vultures (\u003cem\u003eGyps africanus\u003c/em\u003e) are also vulnerable to road mortality, as their reliance on roadkill carcasses increases their exposure to vehicle collisions (Ogada et al. \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Naciri et al. \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Such losses not only threaten population viability but also disrupt key ecosystem functions, including seed dispersal, nutrient cycling, and predator\u0026ndash;prey dynamics (Beasley et al. \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2019\u003c/span\u003e)\u003c/p\u003e\u003cp\u003eRoadkill risk is therefore shaped by a combination of species traits, behavioural patterns, and environmental conditions. For herbivores such as wildebeest and zebra, risk is elevated during resource-scarce dry seasons when road-crossings increase in search of forage and water (Bolger et al. \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2008\u003c/span\u003e; Martin et al. \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2015\u003c/span\u003e), while browsers such as giraffe and eland face risks year-round due to fragmented roadside vegetation (Kioko et al. \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2008\u003c/span\u003e; Lee and Bond \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Vegetation structure can impact roadkill risk at multiple scales. At a landscape scale, it can influence large-scale movement patterns, increasing likelihood or road crossing (Blackburn et al. \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) and at more local scale it can influence risk of collision by obscuring visibility (Canal et al. \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Remote sensing indices like the Normalized Difference Vegetation Index (NDVI), a proxy for vegetation productivity, have proven useful in capturing these seasonal habitat dynamics, with low NDVI values associated with periods of heightened roadkill risk in other systems (Ascens\u0026atilde;o et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Rainfall patterns and proximity to water sources similarly influence crossing behavior, as animals traverse roads to reach ephemeral waterholes during dry periods (Valeix et al. \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; S\u0026aacute;nchez-Montoya et al. \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eHere, we present the first study in the Athi-Kapiti region to integrate remote sensing indices; Normalized Difference Vegetation Index (NDVI), Standardized Precipitation Evapotranspiration Index (SPEI) with systematic field-based roadkill surveys. We investigated the spatio-temporal drivers of wildlife roadkills along a 60-kilometre segment of the Nairobi\u0026ndash;Mombasa Highway, which traverses the Athi-Kapiti Conservancies. Specifically, we aimed to:(1) identify the species most vulnerable to vehicle collisions, (2) map seasonal and spatial variation in roadkills across three sections with differential levels of collision risk, and (3) quantify the effects of vegetation structure, drought severity, and wildlife use intensity on collision rates. Our findings aim to inform specific mitigation actions such as the construction of wildlife overpasses, speed bumps, implementation of seasonal speed restrictions, and habitat restoration initiatives, to support Kenya\u0026rsquo;s infrastructural development while safeguarding wild animals and their habitats.\u003c/p\u003e"},{"header":"2. Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1 Study area\u003c/h2\u003e\u003cp\u003eOur study was conducted in the Athi-Kapiti Plains Conservancies, located at latitude \u0026minus;\u0026thinsp;1.467663\u0026deg; and longitude 37.101091\u0026deg; (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The area encompasses the Swara Plains Conservancy, the International Livestock Research Institute (Kapiti Estate), Maanzoni Conservancy, Mwambi Hill Conservancy, Malinda Conservancy, and adjacent community lands. Theconservancies, constituting the Athi Kapiti Plains, are situated south of Nairobi city and Nairobi National Park, extending east into Machakos County and south toward Amboseli National Park via the gently descending Emarti Valley (Said et al. \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2016\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe Athi-Kapiti Plains are ecologically significant, supporting a diverse assemblage of wildlife and livestock within a semi-arid savannah landscape. The region receives an average annual rainfall of 550 mm and sits at an elevation of 1,650 to 1,900 meters above sea level. The soils are predominantly black cotton in the plains and red cotton in the ridges, supporting a variety of savannah grasses and tree species. Common grasses include \u003cem\u003eThemeda triandra\u003c/em\u003e (Thatching grass), \u003cem\u003eCenchrus stramineus\u003c/em\u003e (Sandhill millet), \u003cem\u003eCynodon dactylon\u003c/em\u003e (Bermuda grass), and \u003cem\u003eCyperus rotundus\u003c/em\u003e (Purple nutsedge). Prominent tree species include \u003cem\u003eBalanites aegyptiaca\u003c/em\u003e, \u003cem\u003eAcacia xanthophloea, Acacia drepanolobium\u003c/em\u003e, \u003cem\u003eGrewia kakothamnos\u003c/em\u003e, and \u003cem\u003eCommiphora kataf\u003c/em\u003e (Kapiti Brochure, 2020). The Athi-Kapiti Plains supports a rich diversity of wildlife, including the Maasai giraffe (\u003cem\u003eGiraffa tippelskirchi\u003c/em\u003e), common zebra (\u003cem\u003eEquus quagga\u003c/em\u003e), common eland (\u003cem\u003eTaurotragus oryx\u003c/em\u003e), wildebeest (\u003cem\u003eConnochaetes taurinus\u003c/em\u003e), impala (\u003cem\u003eAepyceros melampus\u003c/em\u003e), and kongoni (\u003cem\u003eAlcelaphus buselaphus cokii\u003c/em\u003e). Predators such as cheetah (Acinonyx jubatus), lion (\u003cem\u003ePanthera leo\u003c/em\u003e), leopard (\u003cem\u003ePanthera pardus\u003c/em\u003e), and spotted hyena (\u003cem\u003eCrocuta crocuta\u003c/em\u003e) are also present. It serves as a critical wildlife dispersal and migration corridor, linking Nairobi National Park to larger ecosystems such as Amboseli, Tsavo, Magadi and Natron, underscoring their ecological importance (Ogutu et al. \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Sircely et al. \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThe Nairobi-Mombasa Highway, a major transportation corridor connecting Nairobi to the port city of Mombasa, bisects the Athi-Kapiti Plains. The highway is a two-lane road that transitions into a single lane at Makutano Junction. It experiences heavy traffic, with an estimated daily flow of over 31,000 vehicles according to the Kenya Roads Board. Speed limits are set at 80 km/h for commercial and passenger vehicles and 110 km/h for private vehicles, as per Kenya\u0026rsquo;s Traffic Act (CAP 403). The high traffic volume and speeds creates significant risks for wildlife, particularly where the highway intersects known movement corridors.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2 Roadkill Data Collection\u003c/h2\u003e\u003cp\u003eWe conducted roadkill surveys between January and December 2023, covering both wet and dry seasons (Duethmann et al. \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Two road transect surveys were conducted each week along a 60 km stretch. Surveys were carried out by a single observer driving at a low speed of 30\u0026ndash;40 km/hr) to maximize detection probabilities. The protocol established by (Santos et al. \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2011\u003c/span\u003e), was followed, with surveys taking place early in the morning at sunrise to optimize roadkill detection before scavengers removed the carcasses or decay set in (Hels and Buchwald \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2001\u003c/span\u003e). Upon detecting a carcass, the location was recorded by taking coordinate, using a handheld Global Positioning System (GPS). Information on species name, sex, age group, date, and time of the carcass was documented, and the carcass removed from the road to prevent double counting. Additional secondary roadkill data from 2020 to 2022 were provided by the Kenya Wildlife Service (Machakos Station) and included information on the date, GPS coordinates, species, age, and sex of the animals, which was merged with data collected during the study.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e2.3 Defining Roadkill Hotspots and Sampling Zones\u003c/h2\u003e\u003cp\u003eTo better understand spatial variation in roadkill risk, we performed a kernel density analysis to identify high-incidence sections along the 60km segment of the Mombasa Highway where roadkills occurred (Lala et al. \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Galinskaitė et al. \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). We imported the collected roadkill data (coordinates, species, sex, age, date, time) into ArcGIS Pro and used the Kernel Density tool to visualize roadkill intensity along the highway. This process was used to define three sections with varying levels of roadkill density, in order of highest to lowest: Swara-Maanzoni Interface (A), Kapiti-Mwambi Transition (B) and Community Edge Zone (C), hereafter referred to as A, B, and C, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) (Lala et al. \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.4 Sampling Site-level Landscape and Habitat Characteristics\u003c/h2\u003e\u003cp\u003eWe calculated five ecological characteristics for each site ID using data collected from field surveys: (1) wildlife use intensity, (2) woody vegetation density, (3) herbaceous vegetation cover, (4) visibility, and (5) artificial water source density. The first four characteristics were sampled using two parallel transects (3 km long and 2 km wide) established on either side of the highway, resulting in six transects across the three sections. Sampling points were systematically placed at 200-meter intervals along each transect, yielding 30 points per section and 90 points in total. At each of the 90 points, we sampled woody vegetation density, herbaceous vegetation cover, wildlife use intensity and visibility. Artificial water source density was assessed separately by recording the coordinates of artificial water points using a handheld GPS. To explore potential spatial clustering, these coordinates were used to generate a heatmap of water point density in ArcGIS Pro, which was then visually overlaid with roadkill hotspot locations.\u003c/p\u003e\u003cp\u003e\u003cem\u003eWoody vegetation density\u003c/em\u003e was measured using a Point-Cantered Quarter (PCQ) method. The area around each sampling point was divided into four 90\u0026deg; quadrants, and the distance to the nearest tree was measured in each quadrant following (Silva et al. \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2017\u003c/span\u003e), where the distance from the quadrat centre to the nearest tree with a height of 1.3 m and above (\u0026ge;\u0026thinsp;1.3m) was measured. For each tree, diameter at Breast Height (DBH) was recorded, and species were identified. Measurements were taken using a tape measure and range finder (Kiani et al. \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Wainscott \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Additionally, all the trees within 20m radius of the quadrats were counted. Woody vegetation density (D) was calculated assuming a random spatial distribution using the following formula: D\u0026thinsp;=\u0026thinsp;1/rˉ2\u0026sdot;π. Where rˉ is the mean distance to the nearest tree (in meters), and π\u0026thinsp;\u0026asymp;\u0026thinsp;3.1416. Density was expressed as individuals per square meter (ind/m\u0026sup2;)\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eHerbaceous vegetation cover\u003c/strong\u003e\u003cp\u003eAt each of the 90 sampling points, eight 1 \u0026times; 1 m quadrats were randomly placed, yielding a total of 720 quadrats. Within each quadrat, herbaceous species and their heights were recorded. Herbaceous vegetation cover at each sampling point was estimated by averaging the cover values from the eight quadrats.\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cem\u003eVisibility\u003c/em\u003e was measured by calculating the mean distance to the nearest line-of-sight obstruction. At each sampling point, distances were recorded in 24 directions from the centre using a rangefinder (Riginos \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). The 24 measurements were then averaged to obtain a visibility value for each point.\u003c/p\u003e\u003cp\u003e\u003cem\u003eWildlife use intensity\u003c/em\u003e: Wildlife presence at each of the three sections (A, B and C) was assessed by counting dung pellet groups within 50 m \u0026times; 4 m belt transects placed at 200-meter intervals along each 3 km transect. This yielded 15 sampling points per transect, resulting in 30 points per section and 90 points across all sections. Dung counts were used instead of direct observations to minimize disturbance (Takenoshita and Yamagiwa \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2008\u003c/span\u003e) and to detect species that are present but not easily observable (Eggert et al. \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2003\u003c/span\u003e). Wildlife use intensity was calculated as dung density using the formula: Dung Density\u0026thinsp;=\u0026thinsp;Number of Dung Groups/Area of the Transect, where: Area of Transect\u0026thinsp;=\u0026thinsp;Length \u0026times; Width\u0026thinsp;=\u0026thinsp;50m \u0026times; 4m\u0026thinsp;=\u0026thinsp;200 m\u0026sup2;. We used total dung per plot as a relative assay of individual herbivore species use of the different plots. Thus, our approach averts issues related to inferring population densities from dung counts, such as differential decomposition rates across seasons and habitats (Nchanji and Plumptre \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2001\u003c/span\u003e; Hema et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). There is ample evidence from our study system (Augustine \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2003\u003c/span\u003e; Young et al. \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e2005\u003c/span\u003e; Riginos et al.) and elsewhere (Cromsigt et al. \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2009\u003c/span\u003e; Hema et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2017\u003c/span\u003e) that dung counts are robust for comparing relative habitat use by large mammals within a species.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e2.5 Remote Sensing Data and Spatial Data Processing\u003c/h2\u003e\u003cp\u003eEnvironmental covariates were derived for each site using a combination of remote sensing data from Google Earth Engine (GEE) as well as R (v4.5.0) packages: terra, and sf (Haviv et al. \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Drought conditions were assessed using the Standardized Precipitation Evapotranspiration Index (SPEI), derived from ERA5 climate reanalysis datasets (Keune et al. \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). SPEI as an integrative drought index that is calculated by subtracting potential evapotranspiration from precipitation providing a more comprehensive measure of climatic water availability than precipitation alone (Nosrati \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). Positive SPEI values indicate wetter than average conditions and negative values indicate drier than average conditions. SPEI values were calculated at 1, 3, 6 and 12-month accumulation intervals to capture both short- and long-term water stress (Keune et al. \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). We extracted SPEI values from ERA5-derived NetCDF layers and calculated the monthly mean SPEI for each section using a 2km buffer on either side of the highway (consistent with the methodology for field data collection).\u003c/p\u003e\u003cp\u003eVegetation productivity was evaluated using the Normalized Difference Vegetation Index (NDVI), derived from Sentinel-2 surface reflectance imagery between 2020 and 2023. The analysis focused on a 2 km-wide corridor along both sides of the Mombasa Highway. Monthly NDVI composites were generated in Google Earth Engine (GEE) by spatially filtering imagery using the digitized extent polygon and temporally selecting specific monthly intervals. Scenes with more than 15% cloud cover were excluded using the CLOUDY_PIXEL_PERCENTAGE metadata. NDVI was computed for each image using the standard formula: NDVI = (B8 \u0026ndash; B4) / (B8\u0026thinsp;+\u0026thinsp;B4), based on the Near-Infrared (Band 8) and Red (Band 4) bands. Monthly mean NDVI rasters were then generated using a mean reducer, exported as GeoTIFFs, and processed in R. We then calculated the monthly mean NDVI within the polygon for each zone (2km buffer on each side of the highway).\u003c/p\u003e\u003c/div\u003e"},{"header":"3. Statistical Analysis","content":"\u003cp\u003eTo estimate the relationship between road kills and our environmental variables, we used a set of Bayesian hurdle Poisson regression models using the brms package in R (B\u0026uuml;rkner \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Hurdle models accommodate excess zeros (as is common in event data such as roadkills) by modeling both the occurrence (zero vs non-zero) and the count of roadkills where they occur. The response variable, roadkill counts (n_roadkills), was regressed against two climatic stress variables SPEI and NDVI in separate models. For SPEI, we ran four separate models using accumulation periods of 1, 3, 6, and 12 months to evaluate differences in the effects of short-term water stress versus long-term drought. All models included an indicator variable for site ID (i.e., A, B, C) to account for spatial variation of road kills. This allowed us to indirectly assess the influence of ecological factors such as wildlife use intensity, visibility, and vegetation cover (herbaceous and woody) by relating the effects of site ID to measurements from transect surveys post hoc. The models were specified as:\u003c/p\u003e\u003col start=\"1\" type=\"1\"\u003e\n \u003cli\u003e\u003cem\u003eSPEI model\u003c/em\u003e:\u0026nbsp;log(n_roadkills) ~ spei+ site_ID\u003c/li\u003e\n \u003cli\u003e\u003cem\u003eNDVI model\u003c/em\u003e:\u0026nbsp;log(n_roadkills) ~ ndvi_std + site_ID,\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003ewith a binomial hurdle component (hu) estimating zero-inflation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWeakly informative priors were applied: Normal (0, 1) for fixed effects (climatic variables, density terms), Normal (0, 1.2) for intercepts, and Beta (1, 1) for the hurdle probability. Four Markov chains were run (2,000 iterations, 1,000 warmup) using cmdstanr\u0026nbsp;(v0.7.1), with convergence confirmed by Rhat \u0026lt;1.01 and effective sample sizes \u0026gt;1,000. Model fit was evaluated via posterior predictive checks. Analyses were conducted in R (v4.3.1) with\u0026nbsp;tidybayes\u0026nbsp;and\u0026nbsp;ggplot2\u0026nbsp;for visualization.\u003c/p\u003e\n\u003cp\u003eWe used the Kruskal\u0026ndash;Wallis test to compare wildlife use intensity, herbaceous vegetation cover, woody vegetation density, and visibility across the six sampling points selected along the highway based on roadkill intensity. Dunn\u0026rsquo;s test was applied for post hoc pairwise comparisons to identify statistically significant differences among groups.\u003c/p\u003e"},{"header":"4. Results","content":"\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e4.1 Species-specific and temporal trends in roadkill incidences along the Mombasa Highway in the Athi-Kapiti Region\u003c/h2\u003e\u003cp\u003eA total of 11 animal species were involved in roadkills. Zebras had the highest incidents of roadkills, accounting for 69% of the total followed by the spotted hyena at 11%, and the Maasai giraffe at 8% (Table \u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The total roadkills recorded between the year 2020 and 2023 were 218. Year 2020 had the lowest kills with 6 carcasses while 2023 had the highest with 84 carcasses (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eAnimal species killed during the study and their percentage along Mombasa Highway in Athi- Kapiti Plains Conservancies\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNumber\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSpecies Name\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eTotal roadkill\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003ePercentage Roadkill\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eCheetah\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.92\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eDuiker\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.46\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eEland\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3.21\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGiraffe\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e17\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e7.8\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eHyena\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e23\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e10.55\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eImpala\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.92\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eKongoni\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.46\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eOstrich\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.46\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e9\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eServal Cat\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.46\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e10\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eWildebeest\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e13\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e5.96\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e11\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eZebra\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e150\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e68.81\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTotal\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e218\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e100\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003e4.2 Roadkill risk highly impacted by drought conditions and vegetation greenness\u003c/h2\u003e\u003cp\u003eWe found that long-term drought severity and reduced vegetation productivity led to more predicted road kills. Specifically, we observed a considerable negative relationship between road kill frequency and both SPEI (β\u003csub\u003eSPEI_6\u003c/sub\u003e = -0.61[-0.79,-0.42]) and NDVI (β\u003csub\u003eNDVI\u003c/sub\u003e = -0.92[-1.21,-0.64]) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e and Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). The strength of the negative drought effect (SPEI) increased considerably from the 1-month to the 3-month timescale, and again from 3 to 6 months, but no clear difference was observed between the 6-month and 12-month timescales (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003e4.3 Ecological characteristics across sites: Swara-Maanzoni Interface (A), Kapiti-Mwambi Transition (B) and Community Edge Zone (C).\u003c/h2\u003e\u003cp\u003eWildlife use intensity varied significantly across the three sections (χ\u0026sup2; = 41.6, df\u0026thinsp;=\u0026thinsp;2, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), with section A having the highest mean wildlife use intensity (0.0765\u0026thinsp;\u0026plusmn;\u0026thinsp;0.0493), followed by B (0.0410\u0026thinsp;\u0026plusmn;\u0026thinsp;0.0558) and C (0.0090\u0026thinsp;\u0026plusmn;\u0026thinsp;0.0120). Pairwise comparisons showed that wildlife use intensity was significantly higher in A compared to both B (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and C (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), while B was also significantly higher than C (p\u0026thinsp;=\u0026thinsp;0.0206) (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e5\u003c/span\u003e)).\u003c/p\u003e\u003cp\u003eHerbaceous vegetation cover also differed significantly among the three sections (χ\u0026sup2; = 20.9, df\u0026thinsp;=\u0026thinsp;2, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), with section B exhibiting the highest mean herbaceous vegetation cover (81.4\u0026thinsp;\u0026plusmn;\u0026thinsp;14.6%), followed by C (64.6\u0026thinsp;\u0026plusmn;\u0026thinsp;17.4%) and A (63.4\u0026thinsp;\u0026plusmn;\u0026thinsp;13.4%). Herbaceous vegetation cover was significantly higher in B compared to both A (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and C (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), while A and C did not differ significantly (p\u0026thinsp;=\u0026thinsp;1.000) (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e5\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eVisibility was significantly different across the three sections (χ\u0026sup2; = 54.0, df\u0026thinsp;=\u0026thinsp;2, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), with C having the highest visibility (437.0\u0026thinsp;\u0026plusmn;\u0026thinsp;262.0), followed by A (69.3\u0026thinsp;\u0026plusmn;\u0026thinsp;64.3) and B (54.0\u0026thinsp;\u0026plusmn;\u0026thinsp;34.1). Visibility in C was significantly higher than both A (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and B (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), while A and B did not differ (p\u0026thinsp;=\u0026thinsp;1.000) (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e5\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eWoody vegetation density varied significantly across the three sections (χ\u0026sup2; = 20.9, df\u0026thinsp;=\u0026thinsp;2, p\u0026thinsp;=\u0026thinsp;0.0000284), with section B exhibiting the highest mean woody vegetation density (0.0360\u0026thinsp;\u0026plusmn;\u0026thinsp;0.0364), followed by A (0.0296\u0026thinsp;\u0026plusmn;\u0026thinsp;0.0392), and C having the lowest (0.0634\u0026thinsp;\u0026plusmn;\u0026thinsp;0.139). Post-hoc comparisons indicated that C had significantly lower woody vegetation density than both A (p\u0026thinsp;=\u0026thinsp;0.00171) and B (p\u0026thinsp;=\u0026thinsp;0.0000446), while there was no significant difference between A and B (p\u0026thinsp;=\u0026thinsp;1.000) (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e5\u003c/span\u003e)).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003e4.4 Roadkill risk impacted by spatial variation in wildlife use intensity and ecological characteristics across sites\u003c/h2\u003e\u003cp\u003eRoadkill frequency varied significantly across sites. Compared to site A, site B had a markedly lower rate of road kills (β\u003csub\u003eB\u003c/sub\u003e: -1.18 [-1.87 to -0.61]), and site C had the lowest roadkill frequency (β\u003csub\u003eC\u003c/sub\u003e: -1.98 [-3.20 to -0.99]). These estimates indicate that, conditional on NDVI and SPEI, road kills at site B and C were approximately 69% and 86% lower, respectively, than at site A.\u003c/p\u003e\u003cp\u003eTransect measurements revealed that site A (the highest roadkill risk) was associated with the highest wildlife use intensity, while site C (the lowest roadkill risk) had the greatest visibility and lowest woody vegetation cover. Site B (moderate roadkill risk) was characterized by the highest herbaceous vegetation cover (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e5\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e6\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e"},{"header":"5. Discussion","content":"\u003cp\u003eThis study highlights how prolonged drought, vegetation degradation, and wildlife use intensity interact to influence roadkill patterns in the Athi-Kapiti Plains. The increasingly strong negative association between roadkills and SPEI with longer accumulation periods suggests that sustained drought (see Fig. \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e), rather than short dry spells, prompts extensive wildlife movement that increases road exposure and ultimately leads to higher roadkill rates. During exceedingly dry periods, access to water and forage becomes limited and unevenly distributed, causing animals to expand their ranges and encounter roads more frequently (Mulero-P\u0026aacute;zm\u0026aacute;ny et al. \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). The concentration of roadkills near permanent water sources further emphasizes roads as lethal barriers to critical resources during droughts, exacerbating mortality in fragmented landscapes (Clevenger and Wierzchowski \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2006\u003c/span\u003e; Okita-Ouma et al. \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2016\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eAt the same time, low NDVI values indicated reduced vegetation greenness and declining forage availability, especially among herbaceous species (Egeru et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). These conditions fragmented suitable feeding areas and extended the distances animals needed to travel, increasing the likelihood of road encounters leading heightened roadkill rates (Ascens\u0026atilde;o et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Together, drought and vegetation loss may have created a reinforcing effect, where animals displaced by poor habitat quality were funnelled into higher-risk areas near roads.\u003c/p\u003e\u003cp\u003eWildlife use intensity was also a key driver. We observed a strong positive relationship between wildlife abundance and roadkill rates, with no evidence of a threshold beyond which risk plateaus. This suggests that even moderate increases in local intensity can significantly raise the likelihood of collisions, particularly when combined with drought and food scarcity. As competition intensifies in densely populated areas, animals may travel more widely, further increasing their road exposure.\u003c/p\u003e\u003cp\u003eSpecies-specific traits shaped susceptibility to collisions. Giraffes, although less frequently involved in roadkills, are especially vulnerable due to their tall stature, slow movement, and limited agility (More et al. \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Observations during the study showed giraffes foraging on acacia trees near the road, likely because preferred woody browse was less available farther away, especially during dry months (Bond et al. \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). This behavior increased their proximity to traffic and risk of collisions.\u003c/p\u003e\u003cp\u003eZebras, which accounted for the majority of roadkill incidents, were especially affected by the loss of herbaceous vegetation and water scarcity during the dry season. During this period, they are forced to traverse roads more frequently in search of remnant forage and water, a behavior intensified by habitat fragmentation (Boyers and Parrini \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Their gregarious nature and tendency to move in large herds further increase the likelihood of road crossings, raising the risk of vehicle collisions (Kioko et al. \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). According to the 2021 Kenya Wildlife Census conducted during the study period, zebras were the most abundant species in the Athi-Kapiti Plains, with a population of 11,355, significantly higher than any other species. Hence there high population density lead to high roadkill mortality.\u003c/p\u003e\u003cp\u003eSpotted hyenas were likely drawn to carcasses along roads, as their scavenging behavior leads them to revisit sites of previous mortality. This creates a feedback loop where roadkills attract scavengers, which then become roadkill themselves (Naciri et al. \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). As primarily nocturnal animals, hyenas are especially vulnerable during periods of low visibility. These results underscore the importance of implementing carcass removal protocols to reduce attractant cues and prevent cascading mortality (Hager et al. \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2012\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eRoadkill rates varied across the three study sections. Section A had the highest number of incidents, likely due to high wildlife use intensity, limited forage, and the presence of artificial watering points that drew wildlife closer to the road, particularly during dry seasons. Its open landscape also made it attractive to herbivores seeking visibility to detect predators.\u003c/p\u003e\u003cp\u003eSection B experienced moderate roadkill levels, despite having relatively high vegetation cover. This may reflect its moderate wildlife use intensity, and although higher herbaceous cover may reduce road crossings during wet periods, it can attract grazers as vegetation senesces, creating delayed dry-season hotspots (Munyati \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e1997\u003c/span\u003e). Additionally, both Section A and Section B act as part of a larger conservancy corridor linking Swara Plains, Maanzoni, Mwambi Hill, and Kapiti Plains. These areas maintain intact wildlife populations and minimal anthropogenic disturbance, which likely channels animal movement between protected habitats, creating predictable crossing zones and elevating roadkill risk (HP-Ruffino et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Lala et al. \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eIn contrast, Section C recorded the lowest roadkill frequency. This site had sparse vegetation and higher levels of human activity, which may have discouraged wildlife presence due to habitat degradation and negative human\u0026ndash;wildlife interactions (Murray and St. Clair \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). These conditions likely reduced both the intensity of animals and their need to cross roads in the area. These spatial patterns suggest that conditions even a few kilometres from roads can shape animal movement and influence collision risk.\u003c/p\u003e\u003cp\u003eUsing a Bayesian modeling framework enabled us to identify both the probability of roadkill occurrence and the ecological factors that heighten roadkill risk. This approach provided robust inference while explicitly accounting for uncertainty in parameter estimates, which is critical when modeling complex ecological systems. The models identified key ecological drivers specifically, prolonged drought conditions and reduced vegetation productivity that influence roadkill risk which can inform targeted conservation strategies.\u003c/p\u003e\u003cp\u003eManagement efforts should prioritize high-risk zones like Section A. Recommended actions include constructing wildlife crossings, restoring vegetation, redistributing water points more evenly across the ecosystem, reducing speed limits during dry seasons, and removing carcasses promptly. Involving local communities in land-use planning and ecological monitoring will be essential to ensure sustainable and effective mitigation.\u003c/p\u003e\u003cp\u003eWe demonstrate that drought, habitat quality, wildlife use intensity, and animal behavior jointly influence roadkill risk in savannah ecosystems. In the Athi-Kapiti region, these ecological drivers are further intensified by anthropogenic pressures, reflecting the complex realities wildlife faces in increasingly fragmented landscapes. Our integrated approach combining field-based data with climate and vegetation modeling provides a transferable framework for other arid regions facing similar conditions. As climate change intensifies the frequency and severity of drought, wildlife movements in search of water and forage are likely to rise, leading to greater risk of collisions. These findings underscore the need to incorporate both biodiversity conservation and climate resilience into regional infrastructure planning.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eStatements and Declarations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research was funded by Save Giraffes Now as part of the author's professional role within the organization. No grant number was issued\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors have no competing interests to declare.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAggrey W. Chemwa: Conceptualization (equal), Data curation (lead), Formal analysis (lead), Methodology (equal), Writing – original draft (lead), Writing – review and editing (equal); Flora N. Namu: Conceptualization (equal), Methodology (equal), Writing – review and editing (equal); Odd T. Jacobson: Conceptualization (equal), Data curation (equal), Formal analysis (equal), Methodology (equal), Writing – review and editing (equal); Genevieve E. Finerty: Conceptualization (equal), Formal analysis (equal), Methodology (equal), Writing – review and editing (equal); Duncan M. Kimuyu: Conceptualization (equal), Formal analysis (equal), Methodology (equal), Writing – review and editing (equal).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe extend our heartfelt gratitude to all those who contributed to this research. We are especially thankful to Save Giraffes Now, and to its CEO Susan Myers and her husband Richard Myers, for their generous financial support, which made this study possible. We sincerely appreciate the Athi-Kapiti Conservancies for granting us access to conduct fieldwork on their land. Special thanks to the late Steve Tankard (KWS Honorary Warden) and the Kenya Wildlife Service (KWS) for their invaluable assistance in providing secondary roadkill data. We are also grateful to Vincent Kipkazi for his dedicated support during fieldwork. Additional thanks go to Fiona Sandeman, Kayla Walisch, Philip Winter, Ilona Gluecks, and Nobert Muriungi for their valuable contributions. Your support was instrumental to the success of this research.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets generated and analysed during the current study are publicly available in the Figshare repository at https://doi.org/10.6084/m9.figshare.29373971.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthical Compliance\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study did not involve direct handling of animals.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAscens\u0026atilde;o F, Yogui D, Alves M et al (2019) Predicting spatiotemporal patterns of road mortality for medium-large mammals. J Environ Manage 248:109320. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jenvman.2019.109320\u003c/span\u003e\u003cspan address=\"10.1016/j.jenvman.2019.109320\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAugustine DJ (2003) Long-term, livestock‐mediated redistribution of nitrogen and phosphorus in an East African savanna. J Appl Ecol 40:137\u0026ndash;149. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1046/j.1365-2664.2003.00778.x\u003c/span\u003e\u003cspan address=\"10.1046/j.1365-2664.2003.00778.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBeasley JC, Olson ZH, Selva N, DeVault TL (2019) Ecological functions of vertebrate scavenging. Carrion Ecol Manag 125\u0026ndash;157. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-030-16501-7_6\u003c/span\u003e\u003cspan address=\"10.1007/978-3-030-16501-7_6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBirkett PJ, Vanak AT, Muggeo VM et al (2012) Animal perception of seasonal thresholds: changes in elephant movement in relation to rainfall patterns. PLoS ONE 7:e38363. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pone.0038363\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0038363\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBlackburn A, Veals AM, Tewes ME et al (2022) If you build it, will they come? A comparative landscape analysis of ocelot roadkill locations and crossing structures. PLoS ONE 17:e0267630. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pone.0267630\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0267630\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBolger DT, Newmark WD, Morrison TA, Doak DF (2008) The need for integrative approaches to understand and conserve migratory ungulates. Ecol Lett 11:63\u0026ndash;77. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/j.1461-0248.2007.01109.x\u003c/span\u003e\u003cspan address=\"10.1111/j.1461-0248.2007.01109.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBond ML, Strauss MK, Lee DE (2016) Soil correlates and mortality from giraffe skin disease in Tanzania. J Wildl Dis 52:953\u0026ndash;958. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.7589/2016-02-047\u003c/span\u003e\u003cspan address=\"10.7589/2016-02-047\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBoyers M, Parrini F (2024) Resource selection at fine scale: what drives the decision of a generalist herbivore? J Mammal 105:1336\u0026ndash;1344. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/jmammal/gyae091\u003c/span\u003e\u003cspan address=\"10.1093/jmammal/gyae091\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eB\u0026uuml;rkner P-C (2021) Bayesian item response modeling in R with brms and Stan. J Stat Softw 100:1\u0026ndash;54. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.18637/jss.v100.i05\u003c/span\u003e\u003cspan address=\"10.18637/jss.v100.i05\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCanal D, Camacho C, Mart\u0026iacute;n B et al (2019) Fine-scale determinants of vertebrate roadkills across a biodiversity hotspot in Southern Spain. Biodivers Conserv 28:3239\u0026ndash;3256. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10531-019-01817-5\u003c/span\u003e\u003cspan address=\"10.1007/s10531-019-01817-5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eClevenger AP, Wierzchowski J (2006) Maintaining and restoring connectivity in landscapes fragmented by roads. Conserv Biol Ser-Camb- 14:502. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1017/cbo9780511754821.023\u003c/span\u003e\u003cspan address=\"10.1017/cbo9780511754821.023\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCromsigt JP, van Rensburg SJ, Etienne RS, Olff H (2009) Monitoring large herbivore diversity at different scales: comparing direct and indirect methods. Biodivers Conserv 18:1219\u0026ndash;1231. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10531-008-9506-1\u003c/span\u003e\u003cspan address=\"10.1007/s10531-008-9506-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDamania R, Desbureaux S, Scandizzo PL et al (2019) When Good Conservation Becomes Good Economics: Kenya\u0026rsquo;s Vanishing Herds. World Bank. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1596/33083\u003c/span\u003e\u003cspan address=\"10.1596/33083\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDuethmann D, Bl\u0026ouml;schl G, Parajka J (2020) Why does a conceptual hydrological model fail to correctly predict discharge changes in response to climate change? Hydrol Earth Syst Sci 24:3493\u0026ndash;3511. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5194/hess-24-3493-2020\u003c/span\u003e\u003cspan address=\"10.5194/hess-24-3493-2020\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEgeru A, Wasonga O, Mburu J et al (2015) Drivers of forage availability: An integration of remote sensing and traditional ecological knowledge in Karamoja sub-region, Uganda. Pastoralism 5:1\u0026ndash;18. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s13570-015-0037-6\u003c/span\u003e\u003cspan address=\"10.1186/s13570-015-0037-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEggert LS, Eggert JA, Woodruff DS (2003) Estimating population sizes for elusive animals: the forest elephants of Kakum National Park, Ghana. Mol Ecol 12:1389\u0026ndash;1402. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1046/j.1365-294x.2003.01822.x\u003c/span\u003e\u003cspan address=\"10.1046/j.1365-294x.2003.01822.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eForman RT, Sperling D, Bissonette JA et al (2003) Road ecology. Sci Solut 482\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGalinskaitė L, Ulevičius A, Valskys V et al (2022) The influence of landscape structure on wildlife\u0026ndash;vehicle collisions: Geostatistical analysis on hot spot and habitat proximity relations. ISPRS Int J Geo-Inf 11:63. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/ijgi11010063\u003c/span\u003e\u003cspan address=\"10.3390/ijgi11010063\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHager SB, Cosentino BJ, McKay KJ (2012) Scavenging affects persistence of avian carcasses resulting from window collisions in an urban landscape. J Field Ornithol 83:203\u0026ndash;211. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/j.1557-9263.2012.00370.x\u003c/span\u003e\u003cspan address=\"10.1111/j.1557-9263.2012.00370.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHaviv D, Remš\u0026iacute;k J, Gatie M et al (2025) The covariance environment defines cellular niches for spatial inference. Nat Biotechnol 43:269\u0026ndash;280. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41587-024-02193-4\u003c/span\u003e\u003cspan address=\"10.1038/s41587-024-02193-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHels T, Buchwald E (2001) The effect of road kills on amphibian populations. Biol Conserv 99:331\u0026ndash;340\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHema EM, Di Vittorio M, Barnes RF et al (2017) Detection of interannual population trends in seven herbivores from a West African savannah: a comparison between dung counts and direct counts of individuals. Afr J Ecol 55:609\u0026ndash;617. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/aje.12397\u003c/span\u003e\u003cspan address=\"10.1111/aje.12397\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHP-Ruffino P, Rodr\u0026iacute;guez-C KG, Ramazzotto LA, Freitas PD (2016) Roadkill hotspots in a protected area of Cerrado in Brazil: planning actions to conservation. Rev MVZ C\u0026oacute;rdoba 21:5441\u0026ndash;5448. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.21897/rmvz.609\u003c/span\u003e\u003cspan address=\"10.21897/rmvz.609\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJacobson SL, Bliss-Ketchum LL, de Rivera CE, Smith WP (2016) A behavior‐based framework for assessing barrier effects to wildlife from vehicle traffic volume. Ecosphere 7:e01345. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/ecs2.1345\u003c/span\u003e\u003cspan address=\"10.1002/ecs2.1345\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKeune J, Di Giuseppe F, Barnard C et al (2025) ERA5\u0026ndash;Drought: Global drought indices based on ECMWF reanalysis. Sci Data 12:616\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKiani B, Fallah A, Tabari M et al (2013) A comparison of distance sampling methods in Saxaul (Halloxylon ammodendron CA Mey Bunge) Shrub-Lands. Pol J Ecol 61:207\u0026ndash;219\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKioko J, Kiffner C, Jenkins N, Collinson WJ (2015) Wildlife roadkill patterns on a major highway in northern Tanzania. Afr Zool 50:17\u0026ndash;22\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKioko J, Muruthi P, Omondi P, Chiyo PI (2008) The performance of electric fences as elephant barriers in Amboseli, Kenya. South Afr J Wildl Res-24-Mon. Delayed Open Access 38:52\u0026ndash;58. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3957/0379-4369-38.1.52\u003c/span\u003e\u003cspan address=\"10.3957/0379-4369-38.1.52\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLala F, Chiyo PI, Kanga E et al (2021) Wildlife roadkill in the Tsavo Ecosystem, Kenya: identifying hotspots, potential drivers, and affected species. Heliyon 7. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.heliyon.2021.e06364\u003c/span\u003e\u003cspan address=\"10.1016/j.heliyon.2021.e06364\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLaurance WF, Clements GR, Sloan S et al (2014) A global strategy for road building. Nature 513:229\u0026ndash;232\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLee DE, Bond ML (2018) Quantifying the ecological success of a community-based wildlife conservation area in Tanzania. J Mammal 99:459\u0026ndash;464. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/jmammal/gyy014\u003c/span\u003e\u003cspan address=\"10.1093/jmammal/gyy014\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLoss SR, Will T, Marra PP (2014) Estimation of bird-vehicle collision mortality on US roads. J Wildl Manag 78:763\u0026ndash;771. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/jwmg.721\u003c/span\u003e\u003cspan address=\"10.1002/jwmg.721\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMartin J, Benhamou S, Yoganand K, Owen-Smith N (2015) Coping with spatial heterogeneity and temporal variability in resources and risks: adaptive movement behaviour by a large grazing herbivore. PLoS ONE 10:e0118461. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pone.0118461\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0118461\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMore HL, O\u0026rsquo;Connor SM, Br\u0026oslash;ndum E et al (2013) Sensorimotor responsiveness and resolution in the giraffe. J Exp Biol 216:1003\u0026ndash;1011. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1242/jeb.067231\u003c/span\u003e\u003cspan address=\"10.1242/jeb.067231\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMulero-P\u0026aacute;zm\u0026aacute;ny M, Roll\u0026aacute;n L, D\u0026rsquo;Amico M, Gonz\u0026aacute;lez-Su\u0026aacute;rez M (2022) Road orientation affects the impact of roads on wildlife. Wildl Res 50:39\u0026ndash;46. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1071/wr21149\u003c/span\u003e\u003cspan address=\"10.1071/wr21149\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMunyati C (1997) Wetland change assessment on the Kafue Flats. a remote sensing approach, Zambia\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMurray MH, St. Clair CC (2015) Individual flexibility in nocturnal activity reduces risk of road mortality for an urban carnivore. Behav Ecol 26:1520\u0026ndash;1527. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/beheco/arv102\u003c/span\u003e\u003cspan address=\"10.1093/beheco/arv102\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNaciri M, Planillo A, Gicquel M et al (2023) Three decades of wildlife-vehicle collisions in a protected area: Main roads and long-distance commuting trips to migratory prey increase spotted hyena roadkills in the Serengeti. Biol Conserv 279:109950. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.biocon.2023.109950\u003c/span\u003e\u003cspan address=\"10.1016/j.biocon.2023.109950\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNchanji AC, Plumptre AJ (2001) Seasonality in elephant dung decay and implications for censusing and population monitoring in south-western Cameroon. Afr J Ecol 39:24\u0026ndash;32. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1046/j.1365-2028.2001.00265.x\u003c/span\u003e\u003cspan address=\"10.1046/j.1365-2028.2001.00265.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNosrati K (2014) Assessment of Standardized Precipitation Evapotranspiration Index (SPEI) for Drought Identification in Different Climates of IranK. Environ Sci 12\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eOgada DL, Keesing F, Virani MZ (2012) Dropping dead: causes and consequences of vulture population declines worldwide. Ann N Y Acad Sci 1249:57\u0026ndash;71. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/j.1749-6632.2011.06293.x\u003c/span\u003e\u003cspan address=\"10.1111/j.1749-6632.2011.06293.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eOgutu JO, Piepho H-P, Said MY, Kifugo SC (2014) Herbivore dynamics and range contraction in Kajiado County Kenya: climate and land use changes, population pressures, governance, policy and human-wildlife conflicts. Open Ecol J 7. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2174/1874213001407010009\u003c/span\u003e\u003cspan address=\"10.2174/1874213001407010009\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eOkita-Ouma B, Lala F, Moller R et al (2016) Preliminary indications of the effect of infrastructure development on ecosystem connectivity in Tsavo National Parks, Kenya. Pachyderm 57:109\u0026ndash;111. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.69649/pachyderm.v57i.396\u003c/span\u003e\u003cspan address=\"10.69649/pachyderm.v57i.396\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRiginos C (2015) Climate and the landscape of fear in an A frican savanna. J Anim Ecol 84:124\u0026ndash;133\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRiginos C, Veblen KE, Young TP The KLEE experiment\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSaid MY, Ogutu JO, Kifugo SC et al (2016) Effects of extreme land fragmentation on wildlife and livestock population abundance and distribution. J Nat Conserv 34:151\u0026ndash;164. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jnc.2016.10.005\u003c/span\u003e\u003cspan address=\"10.1016/j.jnc.2016.10.005\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eS\u0026aacute;nchez-Montoya MM, Datry T, Ruhi A et al (2023) Intermittent rivers and ephemeral streams are pivotal corridors for aquatic and terrestrial animals. Bioscience 73:291\u0026ndash;301. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/biosci/biad004\u003c/span\u003e\u003cspan address=\"10.1093/biosci/biad004\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSantos SM, Carvalho F, Mira A (2011) How long do the dead survive on the road? Carcass persistence probability and implications for road-kill monitoring surveys. PLoS ONE 6:e25383. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pone.0025383\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0025383\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSeiler A, Helldin JO (2006) Mortality in wildlife due to transportation. The ecology of transportation: Managing mobility for the environment. Springer, pp 165\u0026ndash;189. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/1-4020-4504-2_8\u003c/span\u003e\u003cspan address=\"10.1007/1-4020-4504-2_8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSilva LB, Alves M, Elias RB, Silva L (2017) Comparison of-square, point centered quarter, and-tree sampling methods in Pittosporum undulatum invaded woodlands. Int J For Res 2017: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1155/2017/2818132\u003c/span\u003e\u003cspan address=\"10.1155/2017/2818132\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSircely JA, Gluecks IV, Kimengich N (2020) ILRI Kapiti Plains Research Station: Grazing plan for early 2020\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTakenoshita Y, Yamagiwa J (2008) Estimating gorilla abundance by dung count in the northern part of Moukalaba-Doudou National Park, Gabon. Afr Study Monogr Suppl Issue 39:41\u0026ndash;54\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eValeix M, Fritz H, Sabatier R et al (2011) Elephant-induced structural changes in the vegetation and habitat selection by large herbivores in an African savanna. Biol Conserv 144:902\u0026ndash;912. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.biocon.2010.10.029\u003c/span\u003e\u003cspan address=\"10.1016/j.biocon.2010.10.029\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWainscott BC (2015) Surveying Forest Diversity and Health Using the Point-Centered Quarter Method. Proc Assoc Biol Lab Educ 36:18\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYoung TP, Palmer TM, Gadd ME (2005) Competition and compensation among cattle, zebras, and elephants in a semi-arid savanna in Laikipia, Kenya. Biol Conserv 122:351\u0026ndash;359. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.biocon.2004.08.007\u003c/span\u003e\u003cspan address=\"10.1016/j.biocon.2004.08.007\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"biodiversity-and-conservation","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bioc","sideBox":"Learn more about [Biodiversity and Conservation](https://www.springer.com/journal/10531)","snPcode":"10531","submissionUrl":"https://submission.nature.com/new-submission/10531/3","title":"Biodiversity and Conservation","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Bayesian modelling, drought (SPEI), Mombasa Highway, NDVI, Spatial-temporal drivers, Wildlife-vehicle collisions","lastPublishedDoi":"10.21203/rs.3.rs-6944631/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6944631/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eRoads and habitat fragmentation pose significant conservation challenges for many animals, yet the environmental factors that influence roadkill risk remain poorly understood. In Kenya, the Nairobi\u0026ndash;Mombasa Highway cuts through the ranges of numerous wildlife species within one of the country\u0026rsquo;s most biodiverse regions, the Athi-Kapiti Plains. To investigate the environmental factors influencing roadkill risk in this region, we combined remote sensing data on long-term drought (Standardized Precipitation Evapotranspiration Index, SPEI) and vegetation greenness (Normalized Difference Vegetation Index, NDVI) with 6,240 km of road transect surveys conducted in 2023 and historical roadkill records from 2020\u0026ndash;2022. We recorded 218 wildlife roadkills involving 11 species, predominantly plains zebras (\u003cem\u003eEquus quagga\u003c/em\u003e, 69%), spotted hyenas (\u003cem\u003eCrocuta crocuta\u003c/em\u003e), and Maasai giraffes (\u003cem\u003eGiraffa camelopardalis\u003c/em\u003e tippelskirchi). Prolonged drought, reduced vegetation productivity, and wildlife use intensity emerged as key predictors of roadkill risk along the Nairobi\u0026ndash;Mombasa Highway across the Kenya\u0026rsquo;s Athi-Kapiti Plains. These factors interact to increase animal movement across roads, especially during dry periods when animals seek water and forage. We also found roadkills to be clustered near artificial water points, suggesting these sites may function as ecological traps by concentrating wildlife in high-risk zones. Our findings suggest that drought-driven declines in food availability may force animals to travel farther and cross roads in search of resources, increasing roadkill risk. This interaction between drought and road mortality is particularly concerning, as both represent significant threats that may intensify under climate change. Our results highlight how climatic stress, habitat degradation, and wildlife use intensity jointly shape roadkill patterns.\u003c/p\u003e","manuscriptTitle":"Ecological and Climatic Drivers of Wildlife Road Mortality in Kenya’s Athi-Kapiti Plains","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-06 06:19:30","doi":"10.21203/rs.3.rs-6944631/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-03-02T03:07:49+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-02-17T00:05:20+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"286199873214754044854878331049421613200","date":"2026-02-01T03:17:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"255288854144702818163506766911679985499","date":"2026-01-27T17:45:03+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-17T07:35:00+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"233565014010148536201282422331493853289","date":"2025-08-29T08:48:44+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-08-29T04:33:51+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-07-05T14:16:43+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-06-23T07:56:44+00:00","index":"","fulltext":""},{"type":"submitted","content":"Biodiversity and Conservation","date":"2025-06-21T11:19:20+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"biodiversity-and-conservation","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bioc","sideBox":"Learn more about [Biodiversity and Conservation](https://www.springer.com/journal/10531)","snPcode":"10531","submissionUrl":"https://submission.nature.com/new-submission/10531/3","title":"Biodiversity and Conservation","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"9714002a-c844-45c7-b99b-adbabfc5e5aa","owner":[],"postedDate":"September 6th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-04-21T12:25:00+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-06 06:19:30","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6944631","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6944631","identity":"rs-6944631","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}