Carnivore-mediated seed crossings are more likely on roadkill hotspots

Carnivore-mediated seed crossings are more likely on roadkill hotspots | 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 Carnivore-mediated seed crossings are more likely on roadkill hotspots João Craveiro, Miguel Bugalho, Pedro Vaz This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9314010/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 6 You are reading this latest preprint version Abstract Roadkill hotspots concentrate animal movement and mortality, but whether they also interfere with animal-mediated seed dispersal—a conservation-relevant process underpinning plant regeneration and landscape connectivity—remains unknown. In Mediterranean oak woodlands of southern Portugal, we used seed mimics to test how road type (paved vs. unpaved) and road–forest context (edge vs. non-edge) shape carnivore-mediated seed crossings, and whether crossings coincide with roadkill hotspots detected by kernel density analysis. Bayesian logistic models indicated that seed-crossing probability was about twice as high on unpaved as on paved roads (predicted means 0.28 vs. 0.14), with weaker evidence for a negative edge effect. Crossing probability also tended to increase with carnivore abundance and distance to streams, and decrease with rodent density, albeit with some uncertainty. Crucially, paved sections intersecting roadkill hotspots showed nearly threefold higher predicted crossing probabilities than non-hotspot sections (0.51 vs. 0.18). These results yield clear conservation implications: prioritize paved segments that overlap carnivore roadkill hotspots. Implement precision measures—short-segment speed management, directional fencing that funnels animals to existing culverts/underpasses, and verge management that removes carcasses and limits small-mammal/scavenging attractants—to cut mortality while maintaining seed-mediated connectivity. A light, repeatable monitor–act–reassess workflow (hotspot mapping + seed-mimic trials + camera checks) enables rapid deployment and evaluation, and is transferable to other human-dominated forest mosaics. Hotspot-guided mitigation offers a pragmatic path to reconcile mobility infrastructure with living connectivity. Seed dispersal Road ecology Forest fragmentation Carnivores Hotspots Landscape connectivity Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 INTRODUCTION Carnivores are important drivers of ecosystem function, namely via biotic interactions such as seed dispersal through ingestion (Roemer et al. 2009 ; Herrera et al. 2015 ). Their capacity to disperse seeds stems from their large home ranges, broad movement patterns, and extended gut-retention times enabling them to transport seeds across long distances (Salgueiro et al. 2019 ). Yet the very traits that facilitate their broad dispersal simultaneously heighten their vulnerability to anthropogenic impacts (Craveiro et al. 2019 ; Azedo et al. 2022 )—most notably roads, whose effects on wildlife are widely recognized (Grilo et al. 2015 ; Quintana et al. 2022 ). Roads act as barriers to movement, alter animal behavior, and cause direct mortality (Forman and Alexander 1998 ; Quiles and Barrientos 2024 ). Yet how different roads shape carnivore-mediated seed crossings and whether high seed-crossing sections coincide with segments of elevated animal road mortality remains untested. Road segments with disproportionately high mortality (roadkill hotspots) often coincide with high-quality habitat (Clevenger et al. 2003 ; Grilo et al. 2011 ; Frangini et al. 2022 ). Beyond mortality, these segments may also undermine mutualistic interactions, including carnivore-mediated seed dispersal (Quiles and Barrientos 2024 ; Craveiro et al. 2025 ). Because these hotspots often overlap areas of regular animal use, they may also represent key seed-crossing points, affecting ecological connectivity (Cook and Blumstein 2013 ; Fabrizio et al. 2019 ). Thus, while hotspots are typically framed in terms of mortality risk, they may simultaneously function as risky corridors for seed flow—a dual role that remains unstudied and can have important implications for the conservation of both plants and animals. Seed dispersal underpins plant recruitment and population dynamics (Nathan and Muller-Landau 2000 ; Wang and Smith 2002 ; Chen et al. 2019 ; Vaz et al. 2024 ). Carnivores that consume fleshy fruits transport viable seeds across landscape patches (Fedriani et al. 2023 ). In roaded environments, the interference of roads in seed dispersal is context dependent: both the road-forest context (edge vs. non-edge) and road type (paved vs. unpaved) shape movement and thus seed dispersal (Craveiro et al. 2025 ). Some carnivores avoid roads that delineate forest edges, reducing dispersal there, whereas others exploit edge habitats and move seeds into forest interiors (Tigas et al. 2002 ; Andersen et al. 2017 ; Craveiro et al. 2025 ). Unpaved roads—typically with lower traffic—are likely to show more carnivore-mediated seed crossings, whereas verges of paved roads can attract foraging via carcasses and prey, reinforcing road use (Ruiz-Capillas et al. 2013 ; Mata et al. 2017 ; Galantinho et al. 2020 ) and thereby enhancing dispersal in some contexts while elevating mortality. Road mortality is the most visible consequence of road presence, often apparent as carcasses along roads (Santos et al. 2011 ). It imposes substantial population losses: vehicles kill hundreds of millions of vertebrates annually in the United States (Shilling et al. 2021 ) and ~ 29 million mammals per year in Europe (Grilo et al. 2020 ). Carnivores are disproportionately affected—for example, 35% of annual Florida panther deaths are due to road collisions (Taylor et al. 2002 ); in Sweden, traffic kills 22000–33000 badgers and 6500–12500 red foxes each year (Seiler et al. 2004 ); and in southern Portugal, 47 medium-sized carnivores are recorded as roadkill per 100 km per year (Grilo et al. 2009 ). As mortality commonly clusters in hotspots, an open question is whether those segments also represent areas of elevated carnivore-mediated seed crossings. We conducted a field experiment in Mediterranean oak woodlands of southern Portugal to determine how road type (paved vs. unpaved) and road–forest context (edge vs. non-edge) shape the probability of carnivore-mediated seed crossings, and—on paved roads—to test whether road segments overlapping roadkill hotspots show higher seed-crossing probability. Specifically, we tested two hypotheses. H1: seed-crossing probability is higher on unpaved roads and in non-edge road-forest contexts, where traffic and edge-related openness are expected to impose a weaker movement barrier. H2: seed-crossing probability is higher on paved segments that overlap roadkill hotspots, because hotspots often coincide with areas of concentrated carnivore activity and movement. METHODS Study area and sampling design We conducted the study on roads in the Alentejo region, southern Portugal (Fig. 1 ). The local climate is Mediterranean, with cold-wet winters and hot-dry summers (mean annual precipitation 620 mm; mean annual temperature 18°C). The landscape consists of undulating terrain (200–400 m a.s.l.) dominated by montado (also denominated as dehesa in Spain) and pastures. Montado is characterized by open-canopy cork ( Quercus suber ) and holm ( Q. rotundifolia ) oak stands with an understory of shrubs (e.g., Cistus , Ulex , Erica , Lavandula spp.; Bugalho et al. 2009 , 2011 ; Castro and Freitas, 2009 ; Azedo et al. 2022 ) and grassland areas. In Mediterranean systems such as evergreen oak woodlands of the Iberian Peninsula, where oak decline and limited natural regeneration have been documented (Bugalho et al. 2011 ; Vaz et al. 2019 ; Wadud et al. 2024 ), carnivore-mediated seed dispersal is particularly relevant (Herrera et al. 2015 ; Craveiro et al. 2025 ). We selected 12 road sections (1.2 km each; 3.1–32.8 km apart; mean = 8.7 km) encompassing two road-forest contexts—six bordering oak forests (edge) and six traversing oak forests (non-edge)—and two road types (six paved, six unpaved) (Fig. 2 ). Edge roads were characterized by oak woodland on one side and open pasture or cropland on the other. Paved roads (~ 8 m wide; range: 6–11 m) were National Roads with asphalt pavement and a traffic volume of 95 vehicles/hour (range: 19–234) corresponding to 71 vehicles h⁻¹ (95% CI 16–125) on edge roads and 99 vehicles h⁻¹ (95% CI 193–392) on non-edge roads (see www.infraestruturasdeportugal.pt ). Unpaved roads (~ 4 m wide; range: 3–7 m) were public dirt roads with low traffic, mainly for agricultural access. To account for spatial variation, we established three replicates per factor (2 road-forest contexts × 2 road types × 3 replicates = 12 road sections; Fig. 1 a). Several carnivore species in the study area contribute to fleshy-fruit dispersal, including red foxes ( Vulpes vulpes ), European badgers ( Meles meles ), and stone martens ( Martes foina ), all of them moving seeds of brambles ( Rubus ulmifolius ), common hawthorns ( Crataegus monogyna ), and Phoenician juniper ( Juniperus phoenicea ) (Rosalino et al. 2010 ; Peredo et al. 2013 ; Farris et al. 2017 ). Other seed-dispersers include Egyptian mongoose ( Herpestes ichneumon ), Common genet ( Genetta genetta ), and Eurasian otter ( Lutra lutra ) (Craveiro et al. 2019 ). Seed-crossing sampling and covariates To assess the influence of road-forest context, road type and covariates on seed crossing probability, we supplied dried common figs ( Ficus carica ), which have been proven suitable in feeding experiments (González-Varo et al. 2013 ; Herrera et al. 2015 ; Craveiro et al. 2025 ). Figs were offered once (no replenishment) across the 12 road sections from October 4, 2022 (fall) to July 18, 2023 (summer), ensuring proportional representation of each road-forest context and road type to avoid seasonal biases (Craveiro et al. 2025 ). Within each road section, we installed six supply stations on each side of the road, approximately 200 m apart; each station was positioned 5 m beyond the verge, totaling 144 stations. At each station, we placed six aluminum trays (14.5 × 12.0 × 4.0 cm) spaced 10 m apart, each containing six dried figs. Every fig embedded 10 plastic beads uniquely color-coded for each station (seed mimics, 2.6 mm in diameter) (Herrera et al. 2015 ; Salgueiro et al. 2019 ). To document carnivore consumption of figs, we placed two Bushnell CORE S-4K model 119949C cameras at two randomly selected stations. Eight days after deployment, we systematically searched for carnivore scats containing seed mimics within a 1200 m radius buffer around the supply stations, corresponding to the average home range radius of the carnivore community (Dekker et al. 2001 ; Planillo et al. 2018 ; Craveiro et al. 2019 , 2025 ); this timing followed pilot trials indicating that recoveries within buffers stabilized by ~ 8 days. The search focused on carnivore-utilized paths (e.g., trails, firebreaks, animal tracks; Herrera et al. 2015 ), covering on average 88% of the paths per buffer (range: 81%–98%). To reduce bias in detectability, all paths were traveled by the same observer walking at a constant mean speed of 2 km h⁻¹. Scat locations were georeferenced using a Garmin Dakota 10 GPS (mean error ± 3 m). Seed mimics from each scat were matched to their station of origin via the color code, allowing us to determine whether a ‘seed’ crossing occurred (scat on the opposite side of the road from its supply station; no scats were found on the road) and to calculate the perpendicular distance to the road verge. Although scat detection is imperfect, our standardized, path-based protocol and identical effort across sections reduce heterogeneity in detectability and thus minimize bias in among-section comparisons of carnivore-mediated seed-crossing probability. Tree density (oak trees/ha) was estimated within each 1200-m buffer around the supply stations, as this may influence carnivore habitat use. Using GIS, we generated five random points per road side and created 50-m radius buffers around—twice the standard plot radius employed in Portugal’s National Forest Inventory, which improves tree detection on satellite imagery—, totaling 10 buffers per site. Oak density was calculated by dividing the satellite-derived count of oaks in the buffers by their combined area, obviating exhaustive counts across the 1200-m buffer. To account for variations in carnivore activity within the buffer, we calculated a Kilometer Abundance Index (KAI) as scats km⁻¹ along walked paths (Vincent et al. 1991 ), as a proxy for carnivore abundance. Additionally, we quantified landscape characteristics within the 1200-m buffer using QGIS (v3.20.3-Odense), considering total stream length, distance to nearest stream, the number and total area of water patches—variables selected a priori for their ecological relevance to carnivore-mediated seed dispersal and because they could be quantified consistently for all sections. To assess rodent effects on carnivore-mediated seed crossings, we conducted a 4-night live-trapping survey at 12 rodent sampling points per 1.2-km road section. Six points were placed along the road (three on each side, 3–10 m beyond the outer edge of the verge), and six points were placed in the forest interior on one side of the road, arranged along a linear interior transect to minimize road influence. All sampling points were spaced > 200 m apart. At each point, we deployed four Sherman traps along a short transect perpendicular to the road (or to the interior transect) at 0 m, 10–15 m, 20–30 m, and 40–50 m from the point center, placing traps in favorable microhabitats to maximize captures (under Cistus or Ulex shrubs or in tall grass; Vaz et al. 2024 ). Traps were baited with peanut butter on toast (4 × 4 cm), provided with cotton wool for insulation, checked at sunrise each day, and left in the same locations throughout the session. Captured rodents were marked with a small lateral fur clip (Hoffmann et al. 2010 ; Jumeau et al. 2017 ) and released at the capture site. Trap effort was 192 trap-nights per road section (12 points × 4 traps × 4 nights; total effort 2304 trap-nights). Procedures complied with Directive 2010/63/EU on animal use in research. Rodent density by road section was estimated with Rcapture (v1.4-4; Baillargeon and Rivest 2009 ) under a closed-population assumption (no mortality or movement during the 4-night session), using the bias-corrected closed.bc estimator for capture–mark–recapture data. Roadkill surveys and hotspot detection Between August 11, 2022, and August 10, 2023, we conducted weekly roadkill surveys (every Thursday) of carnivore species along four road sections (8.5–17 km in length; mean 12.8 km; Fig. 1 b) (Grilo et al. 2009 ; Santos et al. 2015 ). To evaluate the spatial relationship between roadkill hotspots and seed-crossing sections, each survey section included the previously defined 1.2-km paved segments, yielding 51 km of roads surveyed weekly (total survey effort = 2652 km; 51 km × 52 visits). A single observer conducted surveys at sunrise, driving at ~ 50 km·h⁻¹ (following standard roadkill survey protocols; e.g., Santos et al. 2018 ) and stopping to georeference each carnivore roadkill with a Garmin Dakota 10 GPS (mean positional error ± 3 m). Carcasses were removed after recording to avoid double counting. We detected roadkill hotspots using kernel density estimation (KDE) (Bíl et al. 2013 ; Medinas et al. 2021 ) with a fixed 500-m bandwidth, previously recommended for identifying roadkill clusters and planning mitigation at the road scale (Grilo et al. 2009 ; Carvalho and Mira 2011 ; Morelle et al. 2013 ; Santos et al. 2016 ). Statistical significance was evaluated against complete spatial randomness via Monte Carlo simulations (n = 999; α = 0.05; Bíl et al. 2013 ). Given the small number of hotspots and their occurrence on separate road sections, network-wide autocorrelation tests (e.g., Moran’s I) were not informative (Carrijo and da Silva 2017 ). For subsequent analyses, we treated hotspots as operationally independent, because they occurred on different roads and were > 500 m apart (i.e., farther than the KDE bandwidth). Hotspots were identified with the sparr package in R (Davies et al. 2017 ). We overlaid the KDE significance map on the six 1.2-km paved sections, classifying each section as hotspot-present or hotspot-absent (3 present, 3 absent). This binary factor was then used as a predictor in the model testing hotspot-presence effect on seed-crossing probability (see § 2.4). Analyses of seed-crossing probabilities To address both questions (effects of road attributes and of hotspot presence on seed-crossing probability), we modelled whether a seed crossed the road (1/0) with a Bernoulli model and logit link (Bayesian logistic regressions fit in brms (Stan; Bürkner 2017 ). For the analysis of road attributes vs. seed-crossing probability, road type (paved vs. unpaved) and road–forest context (edge vs. non-edge) were specified a priori given our primary hypotheses, and their interaction was evaluated. We also considered a section-level random intercept (1|site; “site” denotes each 1.2-km road section) to account for within-section clustering. Additional covariates were distance to road, carnivore abundance (KAI), total stream length, distance to nearest stream, rodent abundance, number of water patches, area of water patches, and tree density (Table 1 ). Numeric predictors were centred and scaled. For a parsimonious model, we compared the interaction model with an additive specification (no interaction) and then tested removing each covariate one at a time (road type and road–forest context were retained a priori throughout). We interpreted leave-one-out differences in expected log predictive density (ΔELPD) whose absolute value exceeded 2×SE as meaningful (e.g., Vaz et al. 2021 ). This supported a final additive model (road type + road-forest context, no interaction) that omitted total stream length, number of water patches, and area of water patches (Table 3 ). Variance inflation factors (VIF; adjusted GVIF for categorical predictors) were < 2.36 overall and < 1.70 in the final model, indicating low multicollinearity (Lüdecke et al. 2021 ). For the analysis of hotspot presence vs. seed-crossing probability, carnivore roadkill hotspot presence (yes/no; § 2.3) was the sole fixed predictor, and we considered a section-level random intercept (1|site) as above. Table 1 Names, definitions, and range of variables considered in the analysis of carnivore-mediated seed crossings. Variable Description Range Road-forest context Position of the road relative to the forest: "edge" (oak forest on one side, open area on the other) or "non-edge" (road traversing oak forest) Road type Road surface type: "paved" (asphalt) or "unpaved" (dirt road) Distance to road Shortest distance (m) from the seed’s final location within a carnivore scat to the road. Carnivore abundance Kilometric abundance of carnivore scats 0.69–3.01 Rodent density Estimated (individuals/ha) within each site 1.91–7.14 Length of streams Total length of watercourses (m) within a 1200-m buffer. 4103–9597 Distance to stream Shortest distance (m) from the seed’s final location within a carnivore scat to a stream. 45.91–1209.65 Number of water patches Number of water patches. 12–32 Area of water patches Total area (ha) occupied by water patches within. 1.68–20.74 Tree density Number of oaks per hectare estimated from ten 50-m radius buffers (five per road side). 17.60–55.40 Table 3 Summary of the fixed effects from the Bayesian mixed-effects logistic model evaluating the effects of road-forest context (non-edge vs. edge) and road-type (unpaved vs. paved) on carnivore-mediated seed crossing probability. Covariates include distance to road, carnivore abundance (carnivore kilometric abundance index; scats km⁻¹), distance to stream, and rodent density). CI = 95% credible interval; β <>0 = posterior probability that the effect is greater (or smaller) than zero, depending on the sign of the estimate. The potential scale reduction factor (Rhat) was 1.00 for all parameters. Parameter β Error 2.5% CI 97.5% CI β <>0 intercept -1.81 0.78 -3.47 -0.41 0.99 road-forest context (edge) -0.41 1.10 -2.64 1.72 0.65 road type (unpaved) 0.89 0.87 -0.75 2.67 0.85 distance to road 1.09 0.41 0.37 1.95 1.00 carnivore abundance 0.48 0.55 -0.53 1.65 0.81 distance to stream 0.36 0.47 -0.53 1.31 0.78 rodent density -0.41 0.51 -1.43 0.58 0.79 For both models, to aid convergence and limit overfitting, we assigned weakly informative priors to all effects (e.g., Gelman 2025 ). For the road-attributes model we used Normal(0, 10) priors for all β coefficients except road–forest context, for which we used Normal(0, 13); for the hotspot model we used Normal(0, 13) for the hotspot presence coefficient. We ran MCMC with 4 chains × 4000 iterations (1000 warm-up), yielding 12000 post-warm-up draws. We checked convergence using trace plots and R̂ ≤ 1.01, with adequate bulk/tail effective sample sizes; posterior predictive checks indicated good fit. Analyses were performed in R v4.5.0 (R Core Team 2025 ). For Bayesian statistical inference, we reported model coefficients ( β ) parameters of the models, their credible intervals, and tested statistical hypotheses (e.g., McElreath 2018 ). We report directional posterior probabilities [Pr( β > 0) or Pr( β < 0)] as an evidence measure. We describe values ≥ 0.95 as strong, 0.80–0.95 as moderate, and < 0.80 as weak evidence for an effect, while avoiding dichotomous ‘significance’ thresholds (Hobbs and Hooten 2015 ; Kruschke 2015 ; McElreath 2018 ). RESULTS Across approximately 350 km of systematic scat searches along the 12 sections, we recorded 612 carnivore scats; 41 contained seed mimics, yielding 59 dispersal events—some scats contained seed mimics from multiple stations—, 27% of which were seed road crossings (Table 2 ). Crossings were more frequent on unpaved vs. paved roads (11 vs. 5; 2.2×) and in non-edge vs. edge road–forest contexts (11 vs. 5; 2.2×). Dispersal events were attributed to stone martens (36%), European badgers (34%), and red foxes (30%). Among crossings, red foxes mediated 44%, followed by stone martens (31%) and European badgers (25%). Camera traps logged 359 animal detections (44% carnivores) and 153 fig removals, ~ 70% by carnivores (chiefly red foxes, stone martens, and Egyptian mongooses). Table 2 Counts of carnivore-mediated seed dispersal events (“dispersals”) and of events that crossed the road (“crossings”), by road–forest context (edge vs. non-edge) and road type (paved vs. unpaved), and—for paved sections only—by roadkill hotspot presence (absent vs. present). Totals refer to the 12 sections (six edge, six non-edge; six paved, six unpaved). For hotspot presence on paved roads, three sections were classified present and three absent. “Crossing” = scat with seed mimics deposited on the side of the road opposite its supply station. Road-forest context Road type Paved Unpaved Total edge dispersals 4 13 17 crossings 2 3 5 non-edge dispersals 13 29 42 crossings 3 8 11 Total dispersals 17 42 59 crossings 5 11 16 Paved sections: roadkill hotspot presence absent dispersals 11 crossings 2 present dispersals 6 crossings 3 Table 4 Fixed part of the Bayesian logistic model predicting the effect of hotspot presence on the probability of seed crossing mediated by carnivores. CI = 95% credible interval; β <>0 = posterior probability that the effect is greater (or smaller) than zero, depending on the sign of the estimate. The potential scale reduction factor (Rhat) was 1.00 for all parameters. Parameter β Error 2.5% CI 97.5% CI β <>0 intercept -1.59 0.82 -3.36 -0.16 0.99 carnivore roadkill hotspot (yes) 1.62 1.20 -0.64 4.06 0.92 Among our paved 1.2-km sections, the proportion of dispersal events that crossed the road was 2.7× higher in sections with roadkill hotspot presence (50%; see § 3.2 for hotspot detection details) than in non-hotspot sections (18%) (Table 2 ). During weekly surveys we recorded 33 carnivore roadkills—65 per 100 km yr⁻¹ (Table A.1, Appendix A)—of which 88% were adults. Fifteen occurred within KDE-defined hotspots intersecting the three paved 1.2-km sections where seed crossings were monitored. Within these hotspots, badgers were most affected (6, 40%), followed by genets (3, 20%) and, with two deaths each (13%), red foxes, Egyptian mongooses, and stone martens. No carnivore roadkills were detected on the three non-hotspot paved seed-crossing sections. Outside the six paved 1.2-km seed-crossing sections—that is, elsewhere along the weekly roadkill transects—red foxes were most frequently killed (9 of 18; 50%). During the live-trapping survey, we captured 57 unique rodents, 28 of which were recaptured at least once. Captures comprised Wood mouse, Apodemus sylvaticus (80%), and Algerian mouse, Mus spretus (20%), across the 144 sampling points. Closed-population capture–recapture estimates (Rcapture) yielded a mean density of 3.72 rodents ha⁻¹ (range 1.91–7.14 ha⁻¹) across the 12 sections; we used these estimates for among-section comparisons of relative rodent availability. Carnivore-mediated seed crossing probability Accounting for nesting with a section-level random intercept, our Bayesian mixed-effects logistic model (Table 3 ; Fig. 3 ) indicated that road type influences the probability of carnivore-mediated seed crossings. Consistent with our hypothesis, directional Bayesian tests for the road-type contrast supported higher crossing probabilities on unpaved than on paved sections, with an estimated 85% posterior probability for this direction of effect. Predicted mean crossing probabilities were 0.28 (unpaved) vs. 0.14 (paved), doubling on unpaved sections. In contrast, evidence for a road–forest context effect was limited, with an estimated 65% posterior probability that crossings were higher in non-edge than edge contexts. Among covariates, distance to the road showed the clearest signal, with essentially all posterior mass above zero. Illustratively, model predictions indicate that a seed deposit located 100 m from the road had 33% higher probability of being a crossing than one only 10 m from the road (Fig. 3 c). Crossing probability also tended to increase with carnivore abundance (KAI) and with greater distance to the nearest stream, and to decrease with rodent density (directional posterior probabilities of 0.81, 0.78, and 0.79, respectively). For illustration, model-based predictions show a 133% relative increase in crossing probability when KAI rises from 1 to 2 scats km⁻¹, a 43% increase when stream distance rises from 250 to 500 m, and a 27% decrease when rodent density increases from 2 to 4 individuals ha⁻¹ (Fig. 3 d–f). Carnivore roadkill hotspots versus seed-crossing sections We detected seven carnivore roadkill hotspots along the network of surveyed roads (Fig. 4 ). Of the seven hotspots, three intersected three of the six paved 1.2-km seed-crossing sections (one per section), and four occurred elsewhere on the network. Accordingly, the six seed-crossing sections were classified as hotspot-present (n = 3) or hotspot-absent (n = 3). Our Bayesian logistic regression on hotspot presence supported higher crossing probabilities on hotspot-present vs. hotspot-absent sections, with an estimated 92% directional posterior probability for this effect. Model predictions indicate a mean crossing probability of 0.51 on hotspot-present sections versus 0.18 on hotspot-absent sections—an absolute increase of 0.33 (+ 183% relative). DISCUSSION Road effects on landscapes and wildlife are well documented (Forman and Alexander 1998 ; Quiles and Barrientos 2024 ), but their influence on seed dispersal remains underexplored. Beyond viewing roads solely as barriers (Suárez-Esteban et al. 2013a ; Ferreira et al. 2022 ), we show that carnivore-mediated seed crossings vary with road attributes—especially surface type—and provide the first evidence that sections with elevated seed-crossing probability coincide with carnivore roadkill hotspots. These findings clarify how road networks modulate plant–animal interactions, with implications for connectivity, restoration, and road planning in forested systems. Raw seed-crossing proportions were higher in non-edge than edge sections, yet Bayesian inference offered only weak evidence for a road–forest context effect, suggesting any context signal is modest relative to road type. Still, habitat structure guides carnivore movement and crossing locations (Fabrizio et al. 2019 ; Valerio et al. 2019 ; Frangini et al. 2022 ): carnivores can be more active where continuous forest provides cover and resources on both sides (Tigas et al. 2002 ; Craveiro et al. 2025 ) and avoid open, altered habitats (Grünewald et al. 2010 ; Costa et al. 2022 ). For instance, red foxes—30% of our dispersal events—though generalists, often use forest more than open habitats (Cagnacci et al. 2004 ; Van Etten et al. 2007 ), consistent with the forest–open contrast at our edge sites. Pasture and cropland bordering edge roads likely diminish carnivore activity and thus reduce crossings, although generalists may also exploit edges (Andersen et al. 2017 ). Road surface played the larger role. Despite roads being less of a barrier to carnivores than to many vertebrates (Craveiro et al. 2025 ), paved roads had fewer crossings than unpaved roads—plausibly due to heavier traffic and avoidance—whereas lower disturbance on unpaved roads may facilitate movements that promote seed dispersal (Suárez-Esteban et al. 2013b ; Bueno et al. 2021 ). Beyond road type, additional factors also shaped seed crossings. The positive distance-to-road effect is consistent with long gut-retention coupled with cumulative movement: seeds ingested near roads are more likely to be carried across as animals continue travelling (Grilo et al. 2015 ). Crossing probability also increased with carnivore activity (KAI), matching an encounter-rate expectation—more carnivores yield more dispersal events and thus more opportunities to cross—and plausibly reflecting social and spatial processes (territorial patrols, avoidance of dominants, niche partitioning) that extend movements (Doncaster and Macdonald 1991 ). Resource concentration along verges (e.g., carcasses, small mammals) can further intensify itinerancy and encounters within the carnivore guild (Galantinho et al. 2020 ; Craveiro et al. 2025 ). Although KAI is a scat-based activity index rather than a direct abundance measure and is sensitive to detectability, it nonetheless supports a plausible behavioural pathway from higher carnivore activity to higher seed-crossing probability. Distance to streams and rodent density were additional but weaker effects (positive and negative, respectively). Streams can channel mesocarnivore movement by providing guidance, cover and resources (Grilo et al. 2008 , 2016 ; Delgado et al. 2018 ; Craveiro et al. 2019 ; Azedo et al. 2022 ), keeping animals on one roadside and lowering crossing odds; as stream distance increases, movements are less constrained by riparian corridors and more likely to cut across roads, raising seed-crossing probability. This mechanism likely depends on riparian condition—degraded strips can disrupt movement and weaken seed dispersal along streams (Grilo et al. 2011 ). In turn, higher rodent densities near verges may reduce the incentive to risk crossing, effectively anchoring foraging to a single roadside (Ruiz-Capillas et al. 2013 ; Galantinho et al. 2020 ). This is consistent with carnivores co-occurring in prey-rich habitats and ranging more widely where prey is scarce (Šálek et al. 2014 ), and with trophic partitioning whereby stone martens shift diet under low small-mammal abundance to avoid competition with genets (Barrientos and Virgós 2006 ). Together these processes can depress cross-road movement when prey is abundant near verges—and promote it when prey is patchy—with cascading consequences for plant regeneration via predator–prey–seed links (Roemer et al. 2009 ). Hotspots concentrated risk where seed crossings were frequent. Notably, two of the three hotspot-intersecting seed-crossing sections were non-edge (with forest on both sides of the road), whereas two of the three non-hotspot sections were edge. This pattern is consistent with continuous forest promoting both movement and mortality, pointing to paved, non-edge segments as priority targets for mitigation. Resource availability and connectivity likely draw carnivores to such places (Fabrizio et al. 2019 ; Valerio et al. 2019 ; Frangini et al. 2022 ), potentially making them focal points for plant–animal mutualisms. Consistently, European polecat roadkills were more frequent where natural habitats bordered roads in Spain (Carmona et al. 2024 ), and in southern Portugal—partly overlapping our area—favourable habitat predicted carnivore roadkills with a reported rate of 47 roadkills per 100 km yr⁻¹ (Grilo et al. 2009 ), 28% below our estimate. Mitigation at hotspots is crucial to reduce mortality while retaining seed-dispersal functions. Notwithstanding, forest edges are also important, as these transition zones may facilitate seed movement into altered habitats, potentially contributing to habitat restoration. Importantly, dispersal may continue after a disperser’s death: carcasses subsidize scavengers (Mata et al. 2017 ), which can ingest and move seeds from roadkilled animals (Hämäläinen et al. 2017 ). However, this post-mortem dispersal is complementary and cannot fully replace primary dispersal by the live animal; roadkill also removes the disperser from the system, with a time-extended cost in foregone future dispersal events. Although we sampled across all four seasons to minimize seasonal bias, roadkill risk also varies with sex, age, and species (Grilo et al. 2009 ; Ascensão et al. 2014 ; Carvalho et al. 2018 ). For instance, on a motorway in southern Portugal, 69% of roadkilled genets were subadults, with mortality peaking during the dispersal season (Carvalho et al. 2018 ), whereas in our national-road data most carcasses were adults. Such age- and season-specific patterns warrant attention because they can disproportionately affect recruitment and the ecological roles of dispersers, with consequences for long-term resilience. Conservation implications Our results point to a clear priority: treat first the paved road segments that overlap carnivore roadkill hotspots. These are the places where risk and ecological function co-occur—seed crossings are most likely and mortality is concentrated. Targeting these segments yields dual benefits: fewer deaths and maintenance of carnivore-mediated seed flow, a key pathway of functional connectivity in many managed landscapes (Genes and Dirzo 2022 ). Mitigation at these sites should be hotspot-guided, not diffuse. In practice, combine: (i) speed management tailored to short sections (e.g., dynamic signage, targeted enforcement; Jägerbrand et al. 2018 ); (ii) directional fencing that funnels animals to existing structures—culverts retrofitted with dry ledges/benches (Craveiro et al. 2019 ) and, where warranted, overpasses (Azedo et al. 2022 ); and (iii) verge management to reduce attractants (prompt carcass removal; vegetation and waste practices that limit small-mammal peaks) (Planillo et al. 2018 ) while retaining essential cover. Where culverts already exist, modest retrofits plus smart fencing are cost-effective and rapidly deployable (Craveiro et al. 2019 ). Landscape context matters. Riparian strips can channel movement; where streams meet roads, pairing hydraulic works with fauna passages improves crossing safety without sacrificing guidance/cover (Grilo et al. 2008 ). Where rodents are abundant near verges, foraging may be fixed to one roadside, reducing crossings but potentially raising exposure to vehicles—another reason to align vegetation and waste management with mitigation goals. A light-weight monitor–act–reassess workflow—kernel-density roadkill mapping + seed-mimic trials + cameras—offers a replicable protocol to detect hotspots, implement measures, and evaluate effectiveness (ideally in BACI designs). Update hotspot maps regularly and after interventions. The hotspot-first approach is transferable to other roaded mosaics where mesocarnivores disperse seeds. While evidence was strongest for road surface and more limited for road–forest context and other covariates, the expected gains and co-benefits justify pilots on paved, non-edge segments with confirmed hotspots, coupled with adaptive monitoring. This aligns road-safety and conservation mandates and offers a pragmatic route to reconcile mobility infrastructure with living connectivity. Declarations Author Contribution J.C.: Data curation, Formal analysis, Investigation, Visualization, Writing – original draft. M.N.B.: Supervision, Writing – review & editing. P.G.V.: Conceptualization, Supervision, Resources, Formal analysis, Visualization, Writing – review & editing. Acknowledgement JC was supported by the Portuguese Science and Technology Foundation (FCT) through an individual grant (2021.05551.BD) and institutional funding to CEABN-InBIO (UIDB/50027/2025) and InBIO (DOI 10.54499/LA/P/0048/2020). JC and PGV were funded by FCT through CE3C (UID/00329/2025) and CHANGE (DOI 10.54499/LA/P/0121/2020). PGV received funding from the AdaptForGrazing project (PRR-C05-i03-I-000035-LA4.3/4.4/4.6/4.7), supported by the EU (RRF) via IFAP (PRR). We are grateful to Francesco Valerio and Denis Medinas for their valuable support. Data Availability The dataset analysed during the current study is available from the corresponding authors on reasonable request. References Andersen GE, Johnson CN, Barmuta LA, Jones ME (2017) Use of anthropogenic linear features by two medium-sized carnivores in reserved and agricultural landscapes. Sci. Rep 7 :11624. https://doi.org/10.1038/s41598-017-11454-z Ascensão F, Grilo C, LaPoint S, Tracey J, Clevenger AP, Santos-Reis M (2014) Inter-individual variability of stone marten behavioral responses to a highway. PLoS One 9:e103544. https://doi.org/10.1371/journal.pone.0103544 Azedo R, Ilhéu A, Santos S, Vaz PG (2022) Carnivores cross irrigation canals more often through overpasses than through culverts. Basic Appl. Ecol. 61:53–67. https://doi.org/10.1016/j.baae.2022.03.004 Baillargeon S, Rivest L.P. (2009) Rcapture: loglinear models for capture–recapture experiments. R package version 1.4-4. https://CRAN.R-project.org/package=Rcapture Barrientos R, Virgós E. (2006) Reduction of potential food interference in two sympatric carnivores by sequential use of shared resources. Acta Oecol. 30:107–116. https://doi.org/10.1016/j.actao.2006.02.006 Bíl M, Andrášik R, Janoška Z (2013) Identification of hazardous road locations of traffic accidents by means of kernel density estimation and cluster significance evaluation. Accid. Anal. Prev. 55:265–273. https://doi.org/10.1016/j.aap.2013.03.003 Bueno RD, García D, Galetti M, La Mantia T. (2021) Trophic and spatial complementarity on seed dispersal services by birds, wild mammals, and cattle in a Mediterranean woodland pasture. Glob. Ecol. Conserv. 31:e01880. https://doi.org/10.1016/j.gecco.2021.e01880 Bugalho MN, Plieninger T, Aronson J, Ellatifi M, Crespo DG (2009) Open woodlands: a diversity of uses (and overuses). In: Aronson J, Pereira JS and Pausas J (Eds.) Cork Ak Woodlands on the Edge: Ecology, Biogeography, and Restoration of an Ancient Mediterranean Ecosystem. Island Press , Washington DC, 33–45. Bugalho MN, Caldeira MC, Pereira JS, Aronson J, Pausas JG (2011) Mediterranean cork oak savannas require human use to sustain biodiversity and ecosystem services. Front. Ecol. Environ. 9:278–286. https://doi.org/10.1890/100084 Bürkner PC (2017) brms: An R package for Bayesian multilevel models using Stan. J. Stat. Softw. 80. https://doi.org/10.18637/jss.v080.i01 Cagnacci F, Meriggi A, Lovari S (2004) Habitat selection by the red fox Vulpes vulpes (L. 1758) in an Alpine area. Ethol. Ecol. Evol. 16:103–116. https://doi.org/10.1080/08927014.2004.9522640 Carmona G, Virgós E, Burgos T, Barrientos R (2024) Factors determining roadkills in a mammal carnivore are road-type specific. Mamm. Biol. 104:175–183. https://doi.org/10.1007/s42991-024-00400-y Carrijo TB, da Silva AR (2017) Modified Moran's I for small samples. Geogr. Anal. 49:451–467. https://doi.org/10.1111/gean.12130 Carvalho F, Lourenço A, Carvalho R, Alves PC, Mira A, Beja P (2018) The effects of a motorway on movement behaviour and gene flow in a forest carnivore: Joint evidence from road mortality, radio tracking and genetics. Landsc. Urban Plann. 178:217–227. https://doi.org/10.1016/j.landurbplan.2018.06.007 Carvalho F, Mira A (2011) Comparing annual vertebrate road kills over two time periods, 9 years apart: a case study in Mediterranean farmland. Eur. J. Wildl. Res. 57:157–174. https://doi.org/10.1007/s10344-010-0410-0 Castro H, Freitas H (2009) Aboveground biomass and productivity in the Montado: from herbaceous to shrub dominated communities. J. Arid Environ. 73:506–511. https://doi.org/10.1016/j.jaridenv.2008.12.009 Chen W, Zhong J, Carson WP, Tang Z, Xie Z, Sun S, Zhou Y (2019) Proximity to roads disrupts rodents' contributions to seed dispersal services and subsequent recruitment dynamics. J. Ecol. 107:2623–2634. https://doi.org/10.1111/1365-2745.13221 Clevenger AP, Chruszcz B, Gunson KE (2003) Spatial patterns and factors influencing small vertebrate fauna road-kill aggregations. Biol. Conserv. 109:15–26. https://doi.org/10.1016/S0006-3207(02)00127-1 Cook TC, Blumstein DT (2013) The omnivore’s dilemma: Diet explains variation in vulnerability to vehicle collision mortality. Biol. Conserv. 167:310–315. https://doi.org/10.1016/j.biocon.2013.08.016 Costa IMDC, Ferreira MS, Mourão CLB, Bueno C (2022) Spatial patterns of carnivore roadkill in a high-traffic-volume highway in the endangered Brazilian Atlantic Forest. Mamm. Biol. 102:477–487. https://doi.org/10.1007/s42991-022-00247-1 Craveiro J, Bernardino J, Mira A, Vaz PG (2019) Impact of culvert flooding on carnivore crossings. J. Environ. Manage. 231:878–885. https://doi.org/10.1016/j.jenvman.2018.10.108 Craveiro J, Bugalho MN, Vaz PG (2025) Roads shape seed dispersal by rodents and carnivores at forest edges. J. Appl. Ecol. 00:1–15. https://doi.org/10.1111/1365-2664.70174 Davies T, Marshall J, Hazelton M (2017) Tutorial on Kernel Estimation of Continuous Spatial and Spatiotemporal Relative Risk with Accompanying Instruction in R, pp. 1–30. R package version 2.2–13. https://doi.org/10.48550/arXiv.1707.06888 Dekker JA, Stein A, Heitkönig IMA (2001) A spatial analysis of a population of red fox (Vulpes vulpes) in the Dutch coastal dune area. J. Zool. 255:505–510. https://doi.org/10.1017/S0952836901001583 Delgado JD, Morelli F, Arroyo NL, Durán J, Rodríguez A, Rosal A, Palenzuela MDV, Rodríguez JDGP (2018) Is vertebrate mortality correlated to potential permeability by underpasses along low-traffic roads? J. Environ. Manage. 221:53–62. https://doi.org/10.1016/j.jenvman.2018.05.043 Doncaster CP, Macdonald DW (1991) Drifting territoriality in the red fox Vulpes vulpes. J. Anim. Ecol. 60:423–439. https://doi.org/10.2307/5288 Fabrizio M, Di Febbraro M, D’Amico M, Frate L, Roscioni F, Loy A (2019) Habitat suitability vs landscape connectivity determining roadkill risk at a regional scale: a case study on European badger (Meles meles). Eur. J. Wildl. Res. 65:1–10. https://doi.org/10.1007/s10344-018-1241-7 Farris E, Canopoli L, Cucca E, Landi S, Maccioni A, Filigheddu R (2017) Foxes provide a direct dispersal service to Phoenician junipers in Mediterranean coastal environments: ecological and evolutionary implications. Plant Ecol. Evol. 150:117–128. https://doi.org/10.5091/plecevo.2017.1277. Fedriani JM, Wiegand T, Garrote PJ, Leiva MJ, Ayllón D (2023) Seed dispersal effectiveness in fragmented and defaunated landscapes. Ecosphere 14:e4658. https://doi.org/10.1002/ecs2.4658 Ferreira EM, Fernandes N, Craveiro J, Valerio F, Costa P, Carrapato C, Mira A, Medinas D, Silva JP, Santos SM (2022) Assessing behaviour states of a forest carnivore in a road-dominated landscape using Hidden Markov Models. Nat. Conserv. 47:155–175. https://doi.org/10.3897/natureconservation.47.72781 Forman RT, Alexander LE (1998) Roads and their major ecological effects. Annu. Rev. Ecol. Evol. Syst. 29:207–231. https://doi.org/10.1146/annurev.ecolsys.29.1.207 Frangini L, Sterrer U, Franchini M, Pesaro S, Rüdisser J, Filacorda S (2022) Stay home, stay safe? High habitat suitability and environmental connectivity increases road mortality in a colonizing mesocarnivore. Landsc. Ecol. 37:2343–2361. https://doi.org/10.1007/s10980-022-01491-z Galantinho A, Herrera JM, Eufrázio S, Silva C, Carvalho F, Alpizar-Jara R, Mira A. (2020) Road verges provide connectivity for small mammals: A case study with wood mice (Apodemus sylvaticus) in an agro-silvo pastoral system. J. Environ. Manage. 258:110033. https://doi.org/10.1016/j.jenvman.2019.110033 Gelman A (2025) Prior choice recommendations. Retrieved from https://github.com/stan-dev/stan/wiki/prior-choice-recommendations Genes L, Dirzo R (2022) Restoration of plant-animal interactions in terrestrial ecosystems. Biol. Conserv. 265:109393. https://doi.org/10.1016/j.biocon.2021.109393 González‐Varo JP, López‐Bao JV, Guitián J (2013) Functional diversity among seed dispersal kernels generated by carnivorous mammals. J. Anim. Ecol. 82:562–571. https://doi.org/10.1111/1365-2656.12024 Grilo C, Ascensão F, Santos-Reis M, Bissonette JA (2011) Do well-connected landscapes promote road-related mortality? Eur. J. Wildl. Res.57:707–716. https://doi.org/10.1007/s10344-010-0478-6 Grilo C, Bissonette JA, Santos-Reis M (2008) Response of carnivores to existing highway culverts and underpasses: implications for road planning and mitigation. Biodivers. Conserv. 17:1685–1699. https://doi.org/10.1007/s10531-008-9374-8 Grilo C, Bissonette JA, Santos-Reis M (2009) Spatial–temporal patterns in Mediterranean carnivore road casualties: consequences for mitigation. Biodivers. Conserv. 142:301–313. https://doi.org/10.1016/j.biocon.2008.10.026 Grilo F, Ferreira E, Alcobia S, Simões L, Santos-Reis M (2016) Do fine-scale factors shape the use of riparian galleries by carnivores in a Mediterranean agro-forested environment? Int. J. Environ. Agric. Res. 2:2454–1850. Grilo C, Koroleva E, Andrášik R, Bíl M, González‐Suárez M (2020) Roadkill risk and population vulnerability in European birds and mammals. Front. Ecol. Environ. 18:323–328. https://doi.org/10.1002/fee.2216 Grilo C, Smith DJ, Klar N (2015) Carnivores: struggling for survival in roaded landscapes. In: van der Ree R, Smith DJ, Grilo C (Eds.), Handbook of Road Ecology, first ed. John Wiley & Sons, Ltd. Grünewald C, Breitbach N, Böhning-Gaese K (2010) Tree visitation and seed dispersal of wild cherries by terrestrial mammals along a human land-use gradient. Basic Appl. Ecol. 11:532–541. https://doi.org/10.1016/j.baae.2010.07.007 Hämäläinen A, Broadley K, Droghini A, Haines JA, Lamb CT, Boutin S, Gilbert S (2017) The ecological significance of secondary seed dispersal by carnivores. Ecosphere 8:e01685. https://doi.org/10.1002/ecs2.1685 Herrera JM, Teixeira IS, Rodríguez-Pérez J, Mira A (2015) Landscape structure shapes carnivore-mediated seed dispersal kernels. Landsc. Ecol. 31:731–743. https://doi.org/10.1007/s10980-015-0283-4 Hobbs NT, Hooten MB (2015) Bayesian Models: A Statistical Primer for Ecologists. Princeton University Press, Princeton, NJ. Hoffmann A, Decher J, Rovero F, Schaer J, Voigt C, Wibbelt G (2010) Field methods and techniques for monitoring mammals. Manual on field recording techniques and protocols for all taxa biodiversity inventories 8:482–529. Jägerbrand AK, Antonson H, Ahlström C (2018) Speed reduction effects over distance of animal-vehicle collision countermeasures – a driving simulator study. Eur. Transp. Res. Rev. 10:40. https://doi.org/10.1186/s12544-018-0314-8 Jumeau J, Boucharel P, Handrich Y, Burel F. (2017) Road-related landscape elements as a habitat: A main asset for small mammals in an intensive farming landscape. Basic Appl. Ecol. 25:15–27. https://doi.org/10.1016/j.baae.2017.09.013 Kruschke JK (2015) Doing Bayesian Data Analysis: A Tutorial with R, JAGS, and Stan (2nd ed.). Academic Press (Elsevier), Amsterdam. Lüdecke D, Ben-Shachar MS, Patil I, Waggoner P, Makowski D (2021) performance: An R Package for Assessment, Comparison and Testing of Statistical Models. J. Open Source Softw. 6:3139. https://doi.org/10.21105/joss.03139 Mata C, Ruiz-Capillas P, Malo JE (2017) Small-scale alterations in carnivore activity patterns close to motorways. Eur. J. Wildl. Res. 63:1–12. https://doi.org/10.1007/s10344-017-1118-1 McElreath R (2018) Statistical rethinking: A Bayesian course with examples in R and Stan. Chapman and Hall, CRC texts in statistical science. https://doi.org/10.1201/9780429029608 Medinas D, Marques JT, Costa P, Santos S, Rebelo H, Barbosa AM, Mira A (2021) Spatiotemporal persistence of bat roadkill hotspots in response to dynamics of habitat suitability and activity patterns. J. Environ. Manage. 277:111412. https://doi.org/10.1016/j.jenvman.2020.111412 Morelle K, Lehaire F, Lejeune P (2013) Spatio-temporal patterns of wildlife-vehicle collisions in a region with a high-density road network. Nat. Conserv. 5:53–73. 10.3897/natureconservation.5.4634 Nathan R, Muller-Landau HC (2000) Spatial patterns of seed dispersal, their determinants and consequences for recruitment. Trends Ecol. Evol. 15:278–285. https://doi.org/10.1016/S0169-5347(00)01874-7 Peredo A, Martínez D, Rodríguez-Pérez J, García D (2013) Mammalian seed dispersal in Cantabrian woodland pastures: network structure and response to forest loss. Basic Appl. Ecol. 14:378–386. https://doi.org/10.1016/j.baae.2013.05.003 Planillo A, Mata C, Manica A, Malo JE (2018) Carnivore abundance near motorways related to prey and roadkills. J. Wildl. Manag. 82:319–327. https://doi.org/10.1002/jwmg.21384 Quiles P, Barrientos R (2024) Interspecific interactions disrupted by roads. Biol. Rev. 99:1121–1139. https://doi.org/10.1111/brv.13061 Quintana I, Cifuentes EF, Dunnink JA, Ariza M, Martínez-Medina D, Fantacini FM, Shrestha BR, Richard FJ (2022) Severe conservation risks of roads on apex predators. Sci. Rep. 12:2902. https://doi.org/10.1038/s41598-022-05294-9 R Core Team (2025) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/ Roemer GW, Gompper ME, Van Valkenburgh B (2009) The ecological role of the mammalian mesocarnivore. BioScience 59:165–173. https://doi.org/10.1525/bio.2009.59.2.9 Rosalino LM, Rosa S, Santos-Reis M (2010) The role of carnivores as Mediterranean seed dispersers. Ann. Zool. Fennici 47:195–205. https://doi.org/10.5735/086.047.0304 Ruiz-Capillas P, Mata C, Malo JE (2013) Road verges are refuges for small mammal populations in extensively managed Mediterranean landscapes. Biol. Conserv. 158:223–229. https://doi.org/10.1016/j.biocon.2012.09.025 Šálek M, Červinka J, Padyšáková E, Kreisinger J (2014) Does spatial co-occurrence of carnivores in a Central European agricultural landscape follow the null model?. Eur J Wildl Res 60:99–107. https://doi.org/10.1007/s10344-013-0755-2 Salgueiro V, Silva C, Eufrázio S, Salgueiro PA, Vaz PG (2019) Endozoochory of a dry‐fruited tree aids quarry passive restoration and seed soaking further increases seedling emergence. Restor. Ecol. 28:668–678. https://doi.org/10.1111/rec.13018 Santos SM, Marques JT, Lourenço A, Medinas D, Barbosa AM, Beja P, Mira A (2015) Sampling effects on the identification of roadkill hotspots: implications for survey design. J. Environ. Manage. 162:87–95. https://doi.org/10.1016/j.jenvman.2015.07.037 Santos RAL, Santos SM, Santos-Reis M, Figueiredo AP, Bager A, Aguiar LM, Ascensao F (2016) Carcass persistence and detectability: reducing the uncertainty surrounding wildlife-vehicle collision surveys. PloSone 11:e0165608. https://doi.org/10.1371/journal.pone.0165608 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. PLosone 6:e25383. https://doi.org/10.1371/journal.pone.0025383 Santos RA, Mota-Ferreira M, Aguiar LM, Ascensão F (2018) Predicting wildlife road-crossing probability from roadkill data using occupancy-detection models. Sci. Total Environ. 642:629–637. https://doi.org/10.1016/j.scitotenv.2018.06.107 Seiler A, Helldin JO, Seiler C (2004) Road mortality in Swedish mammals: results of a drivers' questionnaire. Wildlife Biol. 10:225–233. https://doi.org/10.2981/wlb.2004.028 Shilling F, Nguyen T, Saleh M, Kyaw MK, Tapia K, Trujillo G, Bejarano M, Waetjen D, Peterson J, Kalisz G, Sejour R, Croston S, Ham E (2021) A reprieve from US wildlife mortality on roads during the COVID-19 pandemic. Biol. Conserv. 256:109013. https://doi.org/10.1016/j.biocon.2021.109013 Suárez-Esteban A, Delibes M, Fedriani JM, Dickman C (2013a) Barriers or corridors? The overlooked role of unpaved roads in endozoochorous seed dispersal. J. Appl. Ecol. 50:767–774. https://doi.org/10.1111/1365-2664.12080 Suárez-Esteban A, Delibes M, Fedriani JM (2013b) Unpaved road verges as hotspots of fleshy-fruited shrub recruitment and establishment. Biol. Conserv. 167:50–56. https://doi.org/10.1016/j.biocon.2013.07.022 Taylor SK, Buergelt CD, Roelke-Parker ME, Homer BL, Rotstein, DS (2002) Causes of mortality of free-ranging Florida panthers. J. Wildl. Dis. 38:107–114. https://doi.org/10.7589/0090-3558-38.1.107 Tigas LA, Van Vuren DH, Sauvajot RM (2002) Behavioral responses of bobcats and coyotes to habitat fragmentation and corridors in an urban environment. Biol. Conserv. 108:299–306. https://doi.org/10.1016/S0006-3207(02)00120-9 Valerio F, Carvalho F, Barbosa AM, Mira A, Santos SM (2019) Accounting for connectivity uncertainties in predicting roadkills: a comparative approach between path selection functions and habitat suitability models. Environ. Manage. 64:329–343. https://doi.org/10.1007/s00267-019-01191-6 Van Etten KW, Wilson KR, Crabtree RL (2007) Habitat use of red foxes in Yellowstone National Park based on snow tracking and telemetry. J. Mammal. 88:1498–1507. https://doi.org/10.1644/07-mamm-a-076.1 Vaz PG, Bugalho MN, Fedriani JM, Branco M, Lecomte X, Nogueira C, Caldeira MC (2019) Unravelling associations between tree-seedling performance, herbivory, competition, and facilitation in high nature value farmlands. J. Environ. Manage. 232:1066–1074. https://doi.org/10.1016/j.jenvman.2018.11.082 Vaz PG, Bugalho MN, Fedriani JM (2024) Grazing hinders seed dispersal during crop failure in a declining oak woodland. Sci. Total Environ. 907:167835. https://doi.org/10.1016/j.scitotenv.2023.167835 Vaz PG, Merten EC, Robinson CT, Pinto P (2021) Severely burned wood from wildfires has low functional potential in streams. J. Appl. Ecol. 58:1346–1356 . https://doi.org/10.1111/1365-2664.13872 Vincent JP, Gaillard JM, Bideau E. (1991) Kilometric index as biological indicator for monitoring forest roe dear populations. Acta Theriol. 36:315–328. Wadud AI, Bugalho MN, Vaz PG (2024) Meta-analysis of livestock effects on tree regeneration in oak agroforestry systems. J. Environ. Manage. 364:121498. https://doi.org/10.1016/j.jenvman.2024.121498 Wang BC, Smith TB (2002) Closing the seed dispersal loop. Trends Ecol. Evol. 17:379–385. https://doi.org/10.1016/S0169-5347(02)02541-7 Additional Declarations No competing interests reported. Supplementary Files SupportingInformation.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 04 May, 2026 Reviewers agreed at journal 03 May, 2026 Reviewers invited by journal 28 Apr, 2026 Editor assigned by journal 23 Apr, 2026 Submission checks completed at journal 06 Apr, 2026 First submitted to journal 03 Apr, 2026 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-9314010","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":634350233,"identity":"964d0a76-ce26-4dd1-9562-cfb55b208adf","order_by":0,"name":"João Craveiro","email":"","orcid":"","institution":"CE3C-Centre for Ecology, Evolution and Environmental Changes \u0026 CHANGE, Global Change and Sustainability Institute, Faculdade de Ciências, Universidade de Lisboa","correspondingAuthor":false,"prefix":"","firstName":"João","middleName":"","lastName":"Craveiro","suffix":""},{"id":634350234,"identity":"6bb2a4f9-a25d-4f77-a05e-beb695952715","order_by":1,"name":"Miguel Bugalho","email":"","orcid":"","institution":"Centre for Applied Ecology “Prof. Baeta Neves” (CEABN-InBIO), School of Agriculture, University of Lisbon","correspondingAuthor":false,"prefix":"","firstName":"Miguel","middleName":"","lastName":"Bugalho","suffix":""},{"id":634350235,"identity":"62e0074a-a413-4e61-8d60-7fd792c65467","order_by":2,"name":"Pedro Vaz","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAArElEQVRIiWNgGAWjYFCCA0BcIQckmBtI0XLGGEgwArUkEKuLsY0ULeaMZww//pxnIM8gkdjAXPiDCC2WDWeMJSS3GRg2gLTMIMYWgwPHEiQMt/1JANvCQ6SW5B+JcwxI0nL4mMTBBlK0WDYcPmbZcMzAsI3nYcPhGWlEaDGXONh880eNgTw/e/LBxwU2xDhM4gCEwQbEh4nQANTC34DgMBOlZRSMglEwCkYcAABOtTUEbSEh0wAAAABJRU5ErkJggg==","orcid":"","institution":"CE3C-Centre for Ecology, Evolution and Environmental Changes \u0026 CHANGE, Global Change and Sustainability Institute, Faculdade de Ciências, Universidade de Lisboa","correspondingAuthor":true,"prefix":"","firstName":"Pedro","middleName":"","lastName":"Vaz","suffix":""}],"badges":[],"createdAt":"2026-04-03 14:40:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9314010/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9314010/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108608352,"identity":"b1c48c4e-70c5-41f2-821d-630ec39f972c","added_by":"auto","created_at":"2026-05-06 12:41:23","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":142468,"visible":true,"origin":"","legend":"\u003cp\u003eStudy area and sampling layout in the Alentejo, southern Portugal. (a) Regional context and locations of the twelve 1.2-km road sections where carnivore-mediated seed crossings were monitored. Symbols encode road–forest context (circles = edge; triangles = non-edge) and road type (filled = paved; open = unpaved). Red lines indicate the National-Road transects driven weekly for roadkill surveys, which encompassed the six paved 1.2-km sections. (b) Detail of the six paved 1.2-km sections (light-green lines) and carnivore roadkill hotspots (grey patches; KDE, 500-m bandwidth), showing that three sections intersected one hotspot each and three intersected none.\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9314010/v1/50d37e314fe613e8f5ca474d.jpeg"},{"id":108608520,"identity":"0daf29dd-50d0-4b8b-80d1-3edd0f3f5eda","added_by":"auto","created_at":"2026-05-06 12:41:40","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":372492,"visible":true,"origin":"","legend":"\u003cp\u003eSchematic representation of the road-forest contexts and road types. (a, b) Edge sites with different road types, where the forest borders a tree-free open area, typically used for pasture or cropland. (c, d) Non-edge sites with different road types traversing a continuous forest.\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9314010/v1/8714d1600b11091e45333d58.jpeg"},{"id":108608473,"identity":"203376df-6422-4859-8713-ab540c74fc79","added_by":"auto","created_at":"2026-05-06 12:41:32","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":130590,"visible":true,"origin":"","legend":"\u003cp\u003eMean fitted seed crossing probability (±95 % credible intervals) by road type (a), road-forest context (b) and covariates (c–f), as predicted by a Bayesian mixed-effects logistic model. See Table 1 for variable descriptions.\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9314010/v1/d7366b2a2db70aadf9a38f8a.jpeg"},{"id":108608393,"identity":"7ebfab8a-e871-434f-8976-a12f0e4f7e7e","added_by":"auto","created_at":"2026-05-06 12:41:27","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":55632,"visible":true,"origin":"","legend":"\u003cp\u003eDetection of carnivore roadkill hotspots. (a) Kernel density estimation of roadkill intensity along the weekly-surveyed road network (warmer colours = higher density). (b) Statistically significant hotspots (95% Monte Carlo test against spatial randomness) shown as grey patches. Coordinates: EPSG:32629 (WGS 84 / UTM zone 29N).\u003c/p\u003e","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9314010/v1/c22bf3d43c39fbb6131845f5.jpeg"},{"id":108608353,"identity":"9f2dac66-3c6d-4bd7-9303-a4c1f905c6ba","added_by":"auto","created_at":"2026-05-06 12:41:23","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":10321,"visible":true,"origin":"","legend":"\u003cp\u003eMean fitted seed crossing probability (±95 % credible intervals) by carnivore roadkill hotspots, as predicted by a Bayesian logistic regression.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-9314010/v1/9c6dcc4ac8d855e79585cecf.png"},{"id":108608742,"identity":"92411ca0-0309-40ea-90d3-a26c5536a5ea","added_by":"auto","created_at":"2026-05-06 12:42:18","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1142971,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9314010/v1/81731105-b09f-4e39-90a7-c21c5a4eca53.pdf"},{"id":108608471,"identity":"4bb402a2-b937-4566-8bab-cca4fdc303eb","added_by":"auto","created_at":"2026-05-06 12:41:31","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":16265,"visible":true,"origin":"","legend":"","description":"","filename":"SupportingInformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-9314010/v1/e11e9cd39c8368b1b03d4f1f.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Carnivore-mediated seed crossings are more likely on roadkill hotspots","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eCarnivores are important drivers of ecosystem function, namely via biotic interactions such as seed dispersal through ingestion (Roemer et al. \u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e2009\u003c/span\u003e; Herrera et al. \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Their capacity to disperse seeds stems from their large home ranges, broad movement patterns, and extended gut-retention times enabling them to transport seeds across long distances (Salgueiro et al. \u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Yet the very traits that facilitate their broad dispersal simultaneously heighten their vulnerability to anthropogenic impacts (Craveiro et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Azedo et al. \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2022\u003c/span\u003e)\u0026mdash;most notably roads, whose effects on wildlife are widely recognized (Grilo et al. \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Quintana et al. \u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Roads act as barriers to movement, alter animal behavior, and cause direct mortality (Forman and Alexander \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e1998\u003c/span\u003e; Quiles and Barrientos \u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Yet how different roads shape carnivore-mediated seed crossings and whether high seed-crossing sections coincide with segments of elevated animal road mortality remains untested.\u003c/p\u003e \u003cp\u003eRoad segments with disproportionately high mortality (roadkill hotspots) often coincide with high-quality habitat (Clevenger et al. \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2003\u003c/span\u003e; Grilo et al. \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; Frangini et al. \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Beyond mortality, these segments may also undermine mutualistic interactions, including carnivore-mediated seed dispersal (Quiles and Barrientos \u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Craveiro et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Because these hotspots often overlap areas of regular animal use, they may also represent key seed-crossing points, affecting ecological connectivity (Cook and Blumstein \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Fabrizio et al. \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Thus, while hotspots are typically framed in terms of mortality risk, they may simultaneously function as risky corridors for seed flow\u0026mdash;a dual role that remains unstudied and can have important implications for the conservation of both plants and animals.\u003c/p\u003e \u003cp\u003eSeed dispersal underpins plant recruitment and population dynamics (Nathan and Muller-Landau \u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e2000\u003c/span\u003e; Wang and Smith \u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e2002\u003c/span\u003e; Chen et al. \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Vaz et al. \u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Carnivores that consume fleshy fruits transport viable seeds across landscape patches (Fedriani et al. \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). In roaded environments, the interference of roads in seed dispersal is context dependent: both the road-forest context (edge vs. non-edge) and road type (paved vs. unpaved) shape movement and thus seed dispersal (Craveiro et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Some carnivores avoid roads that delineate forest edges, reducing dispersal there, whereas others exploit edge habitats and move seeds into forest interiors (Tigas et al. \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2002\u003c/span\u003e; Andersen et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Craveiro et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Unpaved roads\u0026mdash;typically with lower traffic\u0026mdash;are likely to show more carnivore-mediated seed crossings, whereas verges of paved roads can attract foraging via carcasses and prey, reinforcing road use (Ruiz-Capillas et al. \u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Mata et al. \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Galantinho et al. \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) and thereby enhancing dispersal in some contexts while elevating mortality.\u003c/p\u003e \u003cp\u003eRoad mortality is the most visible consequence of road presence, often apparent as carcasses along roads (Santos et al. \u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). It imposes substantial population losses: vehicles kill hundreds of millions of vertebrates annually in the United States (Shilling et al. \u003cspan citationid=\"CR73\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) and ~\u0026thinsp;29\u0026nbsp;million mammals per year in Europe (Grilo et al. \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Carnivores are disproportionately affected\u0026mdash;for example, 35% of annual Florida panther deaths are due to road collisions (Taylor et al. \u003cspan citationid=\"CR76\" class=\"CitationRef\"\u003e2002\u003c/span\u003e); in Sweden, traffic kills 22000\u0026ndash;33000 badgers and 6500\u0026ndash;12500 red foxes each year (Seiler et al. \u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e2004\u003c/span\u003e); and in southern Portugal, 47 medium-sized carnivores are recorded as roadkill per 100 km per year (Grilo et al. \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). As mortality commonly clusters in hotspots, an open question is whether those segments also represent areas of elevated carnivore-mediated seed crossings.\u003c/p\u003e \u003cp\u003eWe conducted a field experiment in Mediterranean oak woodlands of southern Portugal to determine how road type (paved vs. unpaved) and road\u0026ndash;forest context (edge vs. non-edge) shape the probability of carnivore-mediated seed crossings, and\u0026mdash;on paved roads\u0026mdash;to test whether road segments overlapping roadkill hotspots show higher seed-crossing probability. Specifically, we tested two hypotheses. H1: seed-crossing probability is higher on unpaved roads and in non-edge road-forest contexts, where traffic and edge-related openness are expected to impose a weaker movement barrier. H2: seed-crossing probability is higher on paved segments that overlap roadkill hotspots, because hotspots often coincide with areas of concentrated carnivore activity and movement.\u003c/p\u003e"},{"header":"METHODS","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy area and sampling design\u003c/h2\u003e \u003cp\u003eWe conducted the study on roads in the Alentejo region, southern Portugal (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The local climate is Mediterranean, with cold-wet winters and hot-dry summers (mean annual precipitation 620 mm; mean annual temperature 18\u0026deg;C). The landscape consists of undulating terrain (200\u0026ndash;400 m a.s.l.) dominated by \u003cem\u003emontado\u003c/em\u003e (also denominated as \u003cem\u003edehesa\u003c/em\u003e in Spain) and pastures. \u003cem\u003eMontado\u003c/em\u003e is characterized by open-canopy cork (\u003cem\u003eQuercus suber\u003c/em\u003e) and holm (\u003cem\u003eQ. rotundifolia\u003c/em\u003e) oak stands with an understory of shrubs (e.g., \u003cem\u003eCistus\u003c/em\u003e, \u003cem\u003eUlex\u003c/em\u003e, \u003cem\u003eErica\u003c/em\u003e, \u003cem\u003eLavandula\u003c/em\u003e spp.; Bugalho et al. \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2009\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; Castro and Freitas, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2009\u003c/span\u003e; Azedo et al. \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) and grassland areas. In Mediterranean systems such as evergreen oak woodlands of the Iberian Peninsula, where oak decline and limited natural regeneration have been documented (Bugalho et al. \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; Vaz et al. \u003cspan citationid=\"CR80\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Wadud et al. \u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), carnivore-mediated seed dispersal is particularly relevant (Herrera et al. \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Craveiro et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eWe selected 12 road sections (1.2 km each; 3.1\u0026ndash;32.8 km apart; mean\u0026thinsp;=\u0026thinsp;8.7 km) encompassing two road-forest contexts\u0026mdash;six bordering oak forests (edge) and six traversing oak forests (non-edge)\u0026mdash;and two road types (six paved, six unpaved) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Edge roads were characterized by oak woodland on one side and open pasture or cropland on the other. Paved roads (~\u0026thinsp;8 m wide; range: 6\u0026ndash;11 m) were National Roads with asphalt pavement and a traffic volume of 95 vehicles/hour (range: 19\u0026ndash;234) corresponding to 71 vehicles h⁻\u0026sup1; (95% CI 16\u0026ndash;125) on edge roads and 99 vehicles h⁻\u0026sup1; (95% CI 193\u0026ndash;392) on non-edge roads (see \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e\u003ca href=\"https://ror.org/00snfqn58\" target=\"_blank\"\u003ewww.infraestruturasdeportugal.pt\u003c/a\u003e\u003c/span\u003e\u003cspan address=\"http://www.infraestruturasdeportugal.pt\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Unpaved roads (~\u0026thinsp;4 m wide; range: 3\u0026ndash;7 m) were public dirt roads with low traffic, mainly for agricultural access. To account for spatial variation, we established three replicates per factor (2 road-forest contexts \u0026times; 2 road types \u0026times; 3 replicates\u0026thinsp;=\u0026thinsp;12 road sections; Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eSeveral carnivore species in the study area contribute to fleshy-fruit dispersal, including red foxes (\u003cem\u003eVulpes vulpes\u003c/em\u003e), European badgers (\u003cem\u003eMeles meles\u003c/em\u003e), and stone martens (\u003cem\u003eMartes foina\u003c/em\u003e), all of them moving seeds of brambles (\u003cem\u003eRubus ulmifolius\u003c/em\u003e), common hawthorns (\u003cem\u003eCrataegus monogyna\u003c/em\u003e), and Phoenician juniper (\u003cem\u003eJuniperus phoenicea\u003c/em\u003e) (Rosalino et al. \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; Peredo et al. \u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Farris et al. \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). Other seed-dispersers include Egyptian mongoose (\u003cem\u003eHerpestes ichneumon\u003c/em\u003e), Common genet (\u003cem\u003eGenetta genetta\u003c/em\u003e), and Eurasian otter (\u003cem\u003eLutra lutra\u003c/em\u003e) (Craveiro et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eSeed-crossing sampling and covariates\u003c/h3\u003e\n\u003cp\u003eTo assess the influence of road-forest context, road type and covariates on seed crossing probability, we supplied dried common figs (\u003cem\u003eFicus carica\u003c/em\u003e), which have been proven suitable in feeding experiments (Gonz\u0026aacute;lez-Varo et al. \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Herrera et al. \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Craveiro et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Figs were offered once (no replenishment) across the 12 road sections from October 4, 2022 (fall) to July 18, 2023 (summer), ensuring proportional representation of each road-forest context and road type to avoid seasonal biases (Craveiro et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eWithin each road section, we installed six supply stations on each side of the road, approximately 200 m apart; each station was positioned 5 m beyond the verge, totaling 144 stations. At each station, we placed six aluminum trays (14.5 \u0026times; 12.0 \u0026times; 4.0 cm) spaced 10 m apart, each containing six dried figs. Every fig embedded 10 plastic beads uniquely color-coded for each station (seed mimics, 2.6 mm in diameter) (Herrera et al. \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Salgueiro et al. \u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). To document carnivore consumption of figs, we placed two Bushnell CORE S-4K model 119949C cameras at two randomly selected stations.\u003c/p\u003e \u003cp\u003eEight days after deployment, we systematically searched for carnivore scats containing seed mimics within a 1200 m radius buffer around the supply stations, corresponding to the average home range radius of the carnivore community (Dekker et al. \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2001\u003c/span\u003e; Planillo et al. \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Craveiro et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2019\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e); this timing followed pilot trials indicating that recoveries within buffers stabilized by ~\u0026thinsp;8 days. The search focused on carnivore-utilized paths (e.g., trails, firebreaks, animal tracks; Herrera et al. \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2015\u003c/span\u003e), covering on average 88% of the paths per buffer (range: 81%\u0026ndash;98%). To reduce bias in detectability, all paths were traveled by the same observer walking at a constant mean speed of 2 km h⁻\u0026sup1;. Scat locations were georeferenced using a Garmin Dakota 10 GPS (mean error\u0026thinsp;\u0026plusmn;\u0026thinsp;3 m). Seed mimics from each scat were matched to their station of origin via the color code, allowing us to determine whether a \u0026lsquo;seed\u0026rsquo; crossing occurred (scat on the opposite side of the road from its supply station; no scats were found on the road) and to calculate the perpendicular distance to the road verge. Although scat detection is imperfect, our standardized, path-based protocol and identical effort across sections reduce heterogeneity in detectability and thus minimize bias in among-section comparisons of carnivore-mediated seed-crossing probability.\u003c/p\u003e \u003cp\u003eTree density (oak trees/ha) was estimated within each 1200-m buffer around the supply stations, as this may influence carnivore habitat use. Using GIS, we generated five random points per road side and created 50-m radius buffers around\u0026mdash;twice the standard plot radius employed in Portugal\u0026rsquo;s National Forest Inventory, which improves tree detection on satellite imagery\u0026mdash;, totaling 10 buffers per site. Oak density was calculated by dividing the satellite-derived count of oaks in the buffers by their combined area, obviating exhaustive counts across the 1200-m buffer.\u003c/p\u003e \u003cp\u003eTo account for variations in carnivore activity within the buffer, we calculated a Kilometer Abundance Index (KAI) as scats km⁻\u0026sup1; along walked paths (Vincent et al. \u003cspan citationid=\"CR83\" class=\"CitationRef\"\u003e1991\u003c/span\u003e), as a proxy for carnivore abundance. Additionally, we quantified landscape characteristics within the 1200-m buffer using QGIS (v3.20.3-Odense), considering total stream length, distance to nearest stream, the number and total area of water patches\u0026mdash;variables selected a priori for their ecological relevance to carnivore-mediated seed dispersal and because they could be quantified consistently for all sections.\u003c/p\u003e \u003cp\u003eTo assess rodent effects on carnivore-mediated seed crossings, we conducted a 4-night live-trapping survey at 12 rodent sampling points per 1.2-km road section. Six points were placed along the road (three on each side, 3\u0026ndash;10 m beyond the outer edge of the verge), and six points were placed in the forest interior on one side of the road, arranged along a linear interior transect to minimize road influence. All sampling points were spaced\u0026thinsp;\u0026gt;\u0026thinsp;200 m apart. At each point, we deployed four Sherman traps along a short transect perpendicular to the road (or to the interior transect) at 0 m, 10\u0026ndash;15 m, 20\u0026ndash;30 m, and 40\u0026ndash;50 m from the point center, placing traps in favorable microhabitats to maximize captures (under \u003cem\u003eCistus\u003c/em\u003e or \u003cem\u003eUlex\u003c/em\u003e shrubs or in tall grass; Vaz et al. \u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Traps were baited with peanut butter on toast (4 \u0026times; 4 cm), provided with cotton wool for insulation, checked at sunrise each day, and left in the same locations throughout the session. Captured rodents were marked with a small lateral fur clip (Hoffmann et al. \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; Jumeau et al. \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2017\u003c/span\u003e) and released at the capture site. Trap effort was 192 trap-nights per road section (12 points \u0026times; 4 traps \u0026times; 4 nights; total effort 2304 trap-nights). Procedures complied with Directive 2010/63/EU on animal use in research.\u003c/p\u003e \u003cp\u003eRodent density by road section was estimated with Rcapture (v1.4-4; Baillargeon and Rivest \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2009\u003c/span\u003e) under a closed-population assumption (no mortality or movement during the 4-night session), using the bias-corrected closed.bc estimator for capture\u0026ndash;mark\u0026ndash;recapture data.\u003c/p\u003e\n\u003ch3\u003eRoadkill surveys and hotspot detection\u003c/h3\u003e\n\u003cp\u003eBetween August 11, 2022, and August 10, 2023, we conducted weekly roadkill surveys (every Thursday) of carnivore species along four road sections (8.5\u0026ndash;17 km in length; mean 12.8 km; Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb) (Grilo et al. \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2009\u003c/span\u003e; Santos et al. \u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). To evaluate the spatial relationship between roadkill hotspots and seed-crossing sections, each survey section included the previously defined 1.2-km paved segments, yielding 51 km of roads surveyed weekly (total survey effort\u0026thinsp;=\u0026thinsp;2652 km; 51 km \u0026times; 52 visits). A single observer conducted surveys at sunrise, driving at ~\u0026thinsp;50 km\u0026middot;h⁻\u0026sup1; (following standard roadkill survey protocols; e.g., Santos et al. \u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e2018\u003c/span\u003e) and stopping to georeference each carnivore roadkill with a Garmin Dakota 10 GPS (mean positional error\u0026thinsp;\u0026plusmn;\u0026thinsp;3 m). Carcasses were removed after recording to avoid double counting.\u003c/p\u003e \u003cp\u003eWe detected roadkill hotspots using kernel density estimation (KDE) (B\u0026iacute;l et al. \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Medinas et al. \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) with a fixed 500-m bandwidth, previously recommended for identifying roadkill clusters and planning mitigation at the road scale (Grilo et al. \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2009\u003c/span\u003e; Carvalho and Mira \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; Morelle et al. \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Santos et al. \u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Statistical significance was evaluated against complete spatial randomness via Monte Carlo simulations (n\u0026thinsp;=\u0026thinsp;999; α\u0026thinsp;=\u0026thinsp;0.05; B\u0026iacute;l et al. \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Given the small number of hotspots and their occurrence on separate road sections, network-wide autocorrelation tests (e.g., Moran\u0026rsquo;s I) were not informative (Carrijo and da Silva \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). For subsequent analyses, we treated hotspots as operationally independent, because they occurred on different roads and were \u0026gt;\u0026thinsp;500 m apart (i.e., farther than the KDE bandwidth). Hotspots were identified with the sparr package in R (Davies et al. \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). We overlaid the KDE significance map on the six 1.2-km paved sections, classifying each section as hotspot-present or hotspot-absent (3 present, 3 absent). This binary factor was then used as a predictor in the model testing hotspot-presence effect on seed-crossing probability (see \u0026sect;\u0026nbsp;2.4).\u003c/p\u003e\n\u003ch3\u003eAnalyses of seed-crossing probabilities\u003c/h3\u003e\n\u003cp\u003eTo address both questions (effects of road attributes and of hotspot presence on seed-crossing probability), we modelled whether a seed crossed the road (1/0) with a Bernoulli model and logit link (Bayesian logistic regressions fit in brms (Stan; B\u0026uuml;rkner \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2017\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFor the analysis of road attributes vs. seed-crossing probability, road type (paved vs. unpaved) and road\u0026ndash;forest context (edge vs. non-edge) were specified a priori given our primary hypotheses, and their interaction was evaluated. We also considered a section-level random intercept (1|site; \u0026ldquo;site\u0026rdquo; denotes each 1.2-km road section) to account for within-section clustering. Additional covariates were distance to road, carnivore abundance (KAI), total stream length, distance to nearest stream, rodent abundance, number of water patches, area of water patches, and tree density (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Numeric predictors were centred and scaled. For a parsimonious model, we compared the interaction model with an additive specification (no interaction) and then tested removing each covariate one at a time (road type and road\u0026ndash;forest context were retained a priori throughout). We interpreted leave-one-out differences in expected log predictive density (ΔELPD) whose absolute value exceeded 2\u0026times;SE as meaningful (e.g., Vaz et al. \u003cspan citationid=\"CR82\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). This supported a final additive model (road type\u0026thinsp;+\u0026thinsp;road-forest context, no interaction) that omitted total stream length, number of water patches, and area of water patches (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Variance inflation factors (VIF; adjusted GVIF for categorical predictors) were \u0026lt;\u0026thinsp;2.36 overall and \u0026lt;\u0026thinsp;1.70 in the final model, indicating low multicollinearity (L\u0026uuml;decke et al. \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). For the analysis of hotspot presence vs. seed-crossing probability, carnivore roadkill hotspot presence (yes/no; \u0026sect;\u0026nbsp;2.3) was the sole fixed predictor, and we considered a section-level random intercept (1|site) as above.\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\u003eNames, definitions, and range of variables considered in the analysis of carnivore-mediated seed crossings.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\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=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDescription\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eRange\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRoad-forest context\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePosition of the road relative to the forest: \"edge\" (oak forest on one side, open area on the other) or \"non-edge\" (road traversing oak forest)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRoad type\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRoad surface type: \"paved\" (asphalt) or \"unpaved\" (dirt road)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDistance to road\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eShortest distance (m) from the seed\u0026rsquo;s final location within a carnivore scat to the road.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCarnivore abundance\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eKilometric abundance of carnivore scats\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.69\u0026ndash;3.01\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRodent density\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eEstimated (individuals/ha) within each site\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.91\u0026ndash;7.14\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLength of streams\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTotal length of watercourses (m) within a 1200-m buffer.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4103\u0026ndash;9597\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDistance to stream\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eShortest distance (m) from the seed\u0026rsquo;s final location within a carnivore scat to a stream.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e45.91\u0026ndash;1209.65\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNumber of water patches\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNumber of water patches.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e12\u0026ndash;32\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eArea of water patches\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTotal area (ha) occupied by water patches within.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.68\u0026ndash;20.74\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTree density\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNumber of oaks per hectare estimated from ten 50-m radius buffers (five per road side).\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e17.60\u0026ndash;55.40\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 \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSummary of the fixed effects from the Bayesian mixed-effects logistic model evaluating the effects of road-forest context (non-edge vs. edge) and road-type (unpaved vs. paved) on carnivore-mediated seed crossing probability. Covariates include distance to road, carnivore abundance (carnivore kilometric abundance index; scats km⁻\u0026sup1;), distance to stream, and rodent density). CI\u0026thinsp;=\u0026thinsp;95% credible interval; \u003cem\u003eβ\u003c/em\u003e\u0026lt;\u0026gt;0\u0026thinsp;=\u0026thinsp;posterior probability that the effect is greater (or smaller) than zero, depending on the sign of the estimate. The potential scale reduction factor (Rhat) was 1.00 for all parameters.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" 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=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eParameter\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cem\u003eβ\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eError\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.5% CI\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e97.5% CI\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cem\u003eβ\u003c/em\u003e\u0026lt;\u0026gt;0\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eintercept\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-1.81\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-3.47\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e-0.41\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.99\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eroad-forest context (edge)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.41\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-2.64\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.72\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.65\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eroad type (unpaved)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.87\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-0.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e2.67\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.85\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003edistance to road\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.09\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.41\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.37\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.95\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.00\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ecarnivore abundance\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.48\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.55\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-0.53\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.81\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003edistance to stream\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.36\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.47\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-0.53\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.31\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003erodent density\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.41\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.51\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-1.43\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.58\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.79\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\u003eFor both models, to aid convergence and limit overfitting, we assigned weakly informative priors to all effects (e.g., Gelman \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). For the road-attributes model we used Normal(0, 10) priors for all \u003cem\u003eβ\u003c/em\u003e coefficients except road\u0026ndash;forest context, for which we used Normal(0, 13); for the hotspot model we used Normal(0, 13) for the hotspot presence coefficient. We ran MCMC with 4 chains \u0026times; 4000 iterations (1000 warm-up), yielding 12000 post-warm-up draws. We checked convergence using trace plots and R̂ \u0026le; 1.01, with adequate bulk/tail effective sample sizes; posterior predictive checks indicated good fit. Analyses were performed in R v4.5.0 (R Core Team \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). For Bayesian statistical inference, we reported model coefficients (\u003cem\u003eβ\u003c/em\u003e) parameters of the models, their credible intervals, and tested statistical hypotheses (e.g., McElreath \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). We report directional posterior probabilities [Pr(\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0) or Pr(\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0)] as an evidence measure. We describe values\u0026thinsp;\u0026ge;\u0026thinsp;0.95 as strong, 0.80\u0026ndash;0.95 as moderate, and \u0026lt;\u0026thinsp;0.80 as weak evidence for an effect, while avoiding dichotomous \u0026lsquo;significance\u0026rsquo; thresholds (Hobbs and Hooten \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Kruschke \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; McElreath \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2018\u003c/span\u003e).\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cp\u003eAcross approximately 350 km of systematic scat searches along the 12 sections, we recorded 612 carnivore scats; 41 contained seed mimics, yielding 59 dispersal events\u0026mdash;some scats contained seed mimics from multiple stations\u0026mdash;, 27% of which were seed road crossings (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Crossings were more frequent on unpaved vs. paved roads (11 vs. 5; 2.2\u0026times;) and in non-edge vs. edge road\u0026ndash;forest contexts (11 vs. 5; 2.2\u0026times;). Dispersal events were attributed to stone martens (36%), European badgers (34%), and red foxes (30%). Among crossings, red foxes mediated 44%, followed by stone martens (31%) and European badgers (25%). Camera traps logged 359 animal detections (44% carnivores) and 153 fig removals, ~\u0026thinsp;70% by carnivores (chiefly red foxes, stone martens, and Egyptian mongooses).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eCounts of carnivore-mediated seed dispersal events (\u0026ldquo;dispersals\u0026rdquo;) and of events that crossed the road (\u0026ldquo;crossings\u0026rdquo;), by road\u0026ndash;forest context (edge vs. non-edge) and road type (paved vs. unpaved), and\u0026mdash;for paved sections only\u0026mdash;by roadkill hotspot presence (absent vs. present). Totals refer to the 12 sections (six edge, six non-edge; six paved, six unpaved). For hotspot presence on paved roads, three sections were classified present and three absent. \u0026ldquo;Crossing\u0026rdquo; = scat with seed mimics deposited on the side of the road opposite its supply station.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\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=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eRoad-forest context\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eRoad type\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePaved\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eUnpaved\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cem\u003eTotal\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eedge\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003edispersals\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e17\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\u003ecrossings\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003enon-edge\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003edispersals\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e42\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\u003ecrossings\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eTotal\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003edispersals\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e42\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e59\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\u003ecrossings\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"6\" nameend=\"c6\" namest=\"c1\"\u003e \u003cp\u003ePaved sections: roadkill hotspot presence\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eabsent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003edispersals\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\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\u003ecrossings\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003epresent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003edispersals\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\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\u003ecrossings\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eFixed part of the Bayesian logistic model predicting the effect of hotspot presence on the probability of seed crossing mediated by carnivores. CI\u0026thinsp;=\u0026thinsp;95% credible interval; \u003cem\u003eβ\u003c/em\u003e\u0026lt;\u0026gt;0\u0026thinsp;=\u0026thinsp;posterior probability that the effect is greater (or smaller) than zero, depending on the sign of the estimate. The potential scale reduction factor (Rhat) was 1.00 for all parameters.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" 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=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eParameter\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cem\u003eβ\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eError\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.5% CI\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e97.5% CI\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cem\u003eβ\u003c/em\u003e\u0026lt;\u0026gt;0\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eintercept\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-1.59\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-3.36\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e-0.16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.99\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ecarnivore roadkill hotspot (yes)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.62\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-0.64\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4.06\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.92\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\u003eAmong our paved 1.2-km sections, the proportion of dispersal events that crossed the road was 2.7\u0026times; higher in sections with roadkill hotspot presence (50%; see \u0026sect;\u0026nbsp;3.2 for hotspot detection details) than in non-hotspot sections (18%) (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e2\u003c/span\u003e). During weekly surveys we recorded 33 carnivore roadkills\u0026mdash;65 per 100 km yr⁻\u0026sup1; (Table A.1, Appendix A)\u0026mdash;of which 88% were adults. Fifteen occurred within KDE-defined hotspots intersecting the three paved 1.2-km sections where seed crossings were monitored. Within these hotspots, badgers were most affected (6, 40%), followed by genets (3, 20%) and, with two deaths each (13%), red foxes, Egyptian mongooses, and stone martens. No carnivore roadkills were detected on the three non-hotspot paved seed-crossing sections. Outside the six paved 1.2-km seed-crossing sections\u0026mdash;that is, elsewhere along the weekly roadkill transects\u0026mdash;red foxes were most frequently killed (9 of 18; 50%).\u003c/p\u003e \u003cp\u003eDuring the live-trapping survey, we captured 57 unique rodents, 28 of which were recaptured at least once. Captures comprised Wood mouse, \u003cem\u003eApodemus sylvaticus\u003c/em\u003e (80%), and Algerian mouse, \u003cem\u003eMus spretus\u003c/em\u003e (20%), across the 144 sampling points. Closed-population capture\u0026ndash;recapture estimates (Rcapture) yielded a mean density of 3.72 rodents ha⁻\u0026sup1; (range 1.91\u0026ndash;7.14 ha⁻\u0026sup1;) across the 12 sections; we used these estimates for among-section comparisons of relative rodent availability.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eCarnivore-mediated seed crossing probability\u003c/h2\u003e \u003cp\u003eAccounting for nesting with a section-level random intercept, our Bayesian mixed-effects logistic model (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e3\u003c/span\u003e; Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) indicated that road type influences the probability of carnivore-mediated seed crossings. Consistent with our hypothesis, directional Bayesian tests for the road-type contrast supported higher crossing probabilities on unpaved than on paved sections, with an estimated 85% posterior probability for this direction of effect. Predicted mean crossing probabilities were 0.28 (unpaved) vs. 0.14 (paved), doubling on unpaved sections. In contrast, evidence for a road\u0026ndash;forest context effect was limited, with an estimated 65% posterior probability that crossings were higher in non-edge than edge contexts.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eAmong covariates, distance to the road showed the clearest signal, with essentially all posterior mass above zero. Illustratively, model predictions indicate that a seed deposit located 100 m from the road had 33% higher probability of being a crossing than one only 10 m from the road (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec). Crossing probability also tended to increase with carnivore abundance (KAI) and with greater distance to the nearest stream, and to decrease with rodent density (directional posterior probabilities of 0.81, 0.78, and 0.79, respectively). For illustration, model-based predictions show a 133% relative increase in crossing probability when KAI rises from 1 to 2 scats km⁻\u0026sup1;, a 43% increase when stream distance rises from 250 to 500 m, and a 27% decrease when rodent density increases from 2 to 4 individuals ha⁻\u0026sup1; (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ed\u0026ndash;f).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eCarnivore roadkill hotspots versus seed-crossing sections\u003c/h3\u003e\n\u003cp\u003eWe detected seven carnivore roadkill hotspots along the network of surveyed roads (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). Of the seven hotspots, three intersected three of the six paved 1.2-km seed-crossing sections (one per section), and four occurred elsewhere on the network. Accordingly, the six seed-crossing sections were classified as hotspot-present (n\u0026thinsp;=\u0026thinsp;3) or hotspot-absent (n\u0026thinsp;=\u0026thinsp;3).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eOur Bayesian logistic regression on hotspot presence supported higher crossing probabilities on hotspot-present vs. hotspot-absent sections, with an estimated 92% directional posterior probability for this effect. Model predictions indicate a mean crossing probability of 0.51 on hotspot-present sections versus 0.18 on hotspot-absent sections\u0026mdash;an absolute increase of 0.33 (+\u0026thinsp;183% relative).\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eRoad effects on landscapes and wildlife are well documented (Forman and Alexander \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e1998\u003c/span\u003e; Quiles and Barrientos \u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), but their influence on seed dispersal remains underexplored. Beyond viewing roads solely as barriers (Su\u0026aacute;rez-Esteban et al. \u003cspan citationid=\"CR74\" class=\"CitationRef\"\u003e2013a\u003c/span\u003e; Ferreira et al. \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), we show that carnivore-mediated seed crossings vary with road attributes\u0026mdash;especially surface type\u0026mdash;and provide the first evidence that sections with elevated seed-crossing probability coincide with carnivore roadkill hotspots. These findings clarify how road networks modulate plant\u0026ndash;animal interactions, with implications for connectivity, restoration, and road planning in forested systems.\u003c/p\u003e \u003cp\u003eRaw seed-crossing proportions were higher in non-edge than edge sections, yet Bayesian inference offered only weak evidence for a road\u0026ndash;forest context effect, suggesting any context signal is modest relative to road type. Still, habitat structure guides carnivore movement and crossing locations (Fabrizio et al. \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Valerio et al. \u003cspan citationid=\"CR78\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Frangini et al. \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2022\u003c/span\u003e): carnivores can be more active where continuous forest provides cover and resources on both sides (Tigas et al. \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2002\u003c/span\u003e; Craveiro et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) and avoid open, altered habitats (Gr\u0026uuml;newald et al. \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; Costa et al. \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). For instance, red foxes\u0026mdash;30% of our dispersal events\u0026mdash;though generalists, often use forest more than open habitats (Cagnacci et al. \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2004\u003c/span\u003e; Van Etten et al. \u003cspan citationid=\"CR79\" class=\"CitationRef\"\u003e2007\u003c/span\u003e), consistent with the forest\u0026ndash;open contrast at our edge sites. Pasture and cropland bordering edge roads likely diminish carnivore activity and thus reduce crossings, although generalists may also exploit edges (Andersen et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2017\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eRoad surface played the larger role. Despite roads being less of a barrier to carnivores than to many vertebrates (Craveiro et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e), paved roads had fewer crossings than unpaved roads\u0026mdash;plausibly due to heavier traffic and avoidance\u0026mdash;whereas lower disturbance on unpaved roads may facilitate movements that promote seed dispersal (Su\u0026aacute;rez-Esteban et al. \u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e2013b\u003c/span\u003e; Bueno et al. \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eBeyond road type, additional factors also shaped seed crossings. The positive distance-to-road effect is consistent with long gut-retention coupled with cumulative movement: seeds ingested near roads are more likely to be carried across as animals continue travelling (Grilo et al. \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Crossing probability also increased with carnivore activity (KAI), matching an encounter-rate expectation\u0026mdash;more carnivores yield more dispersal events and thus more opportunities to cross\u0026mdash;and plausibly reflecting social and spatial processes (territorial patrols, avoidance of dominants, niche partitioning) that extend movements (Doncaster and Macdonald \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e1991\u003c/span\u003e). Resource concentration along verges (e.g., carcasses, small mammals) can further intensify itinerancy and encounters within the carnivore guild (Galantinho et al. \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Craveiro et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Although KAI is a scat-based activity index rather than a direct abundance measure and is sensitive to detectability, it nonetheless supports a plausible behavioural pathway from higher carnivore activity to higher seed-crossing probability.\u003c/p\u003e \u003cp\u003eDistance to streams and rodent density were additional but weaker effects (positive and negative, respectively). Streams can channel mesocarnivore movement by providing guidance, cover and resources (Grilo et al. \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2008\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Delgado et al. \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Craveiro et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Azedo et al. \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), keeping animals on one roadside and lowering crossing odds; as stream distance increases, movements are less constrained by riparian corridors and more likely to cut across roads, raising seed-crossing probability. This mechanism likely depends on riparian condition\u0026mdash;degraded strips can disrupt movement and weaken seed dispersal along streams (Grilo et al. \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). In turn, higher rodent densities near verges may reduce the incentive to risk crossing, effectively anchoring foraging to a single roadside (Ruiz-Capillas et al. \u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Galantinho et al. \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). This is consistent with carnivores co-occurring in prey-rich habitats and ranging more widely where prey is scarce (Š\u0026aacute;lek et al. \u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e2014\u003c/span\u003e), and with trophic partitioning whereby stone martens shift diet under low small-mammal abundance to avoid competition with genets (Barrientos and Virg\u0026oacute;s \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2006\u003c/span\u003e). Together these processes can depress cross-road movement when prey is abundant near verges\u0026mdash;and promote it when prey is patchy\u0026mdash;with cascading consequences for plant regeneration via predator\u0026ndash;prey\u0026ndash;seed links (Roemer et al. \u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e2009\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eHotspots concentrated risk where seed crossings were frequent. Notably, two of the three hotspot-intersecting seed-crossing sections were non-edge (with forest on both sides of the road), whereas two of the three non-hotspot sections were edge. This pattern is consistent with continuous forest promoting both movement and mortality, pointing to paved, non-edge segments as priority targets for mitigation. Resource availability and connectivity likely draw carnivores to such places (Fabrizio et al. \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Valerio et al. \u003cspan citationid=\"CR78\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Frangini et al. \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), potentially making them focal points for plant\u0026ndash;animal mutualisms. Consistently, European polecat roadkills were more frequent where natural habitats bordered roads in Spain (Carmona et al. \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), and in southern Portugal\u0026mdash;partly overlapping our area\u0026mdash;favourable habitat predicted carnivore roadkills with a reported rate of 47 roadkills per 100 km yr⁻\u0026sup1; (Grilo et al. \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2009\u003c/span\u003e), 28% below our estimate. Mitigation at hotspots is crucial to reduce mortality while retaining seed-dispersal functions. Notwithstanding, forest edges are also important, as these transition zones may facilitate seed movement into altered habitats, potentially contributing to habitat restoration.\u003c/p\u003e \u003cp\u003eImportantly, dispersal may continue after a disperser\u0026rsquo;s death: carcasses subsidize scavengers (Mata et al. \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2017\u003c/span\u003e), which can ingest and move seeds from roadkilled animals (H\u0026auml;m\u0026auml;l\u0026auml;inen et al. \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). However, this post-mortem dispersal is complementary and cannot fully replace primary dispersal by the live animal; roadkill also removes the disperser from the system, with a time-extended cost in foregone future dispersal events.\u003c/p\u003e \u003cp\u003eAlthough we sampled across all four seasons to minimize seasonal bias, roadkill risk also varies with sex, age, and species (Grilo et al. \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2009\u003c/span\u003e; Ascens\u0026atilde;o et al. \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Carvalho et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). For instance, on a motorway in southern Portugal, 69% of roadkilled genets were subadults, with mortality peaking during the dispersal season (Carvalho et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), whereas in our national-road data most carcasses were adults. Such age- and season-specific patterns warrant attention because they can disproportionately affect recruitment and the ecological roles of dispersers, with consequences for long-term resilience.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eConservation implications\u003c/h2\u003e \u003cp\u003eOur results point to a clear priority: treat first the paved road segments that overlap carnivore roadkill hotspots. These are the places where risk and ecological function co-occur\u0026mdash;seed crossings are most likely and mortality is concentrated. Targeting these segments yields dual benefits: fewer deaths and maintenance of carnivore-mediated seed flow, a key pathway of functional connectivity in many managed landscapes (Genes and Dirzo \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eMitigation at these sites should be hotspot-guided, not diffuse. In practice, combine: (i) speed management tailored to short sections (e.g., dynamic signage, targeted enforcement; J\u0026auml;gerbrand et al. \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2018\u003c/span\u003e); (ii) directional fencing that funnels animals to existing structures\u0026mdash;culverts retrofitted with dry ledges/benches (Craveiro et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) and, where warranted, overpasses (Azedo et al. \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2022\u003c/span\u003e); and (iii) verge management to reduce attractants (prompt carcass removal; vegetation and waste practices that limit small-mammal peaks) (Planillo et al. \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2018\u003c/span\u003e) while retaining essential cover. Where culverts already exist, modest retrofits plus smart fencing are cost-effective and rapidly deployable (Craveiro et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eLandscape context matters. Riparian strips can channel movement; where streams meet roads, pairing hydraulic works with fauna passages improves crossing safety without sacrificing guidance/cover (Grilo et al. \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). Where rodents are abundant near verges, foraging may be fixed to one roadside, reducing crossings but potentially raising exposure to vehicles\u0026mdash;another reason to align vegetation and waste management with mitigation goals. A light-weight monitor\u0026ndash;act\u0026ndash;reassess workflow\u0026mdash;kernel-density roadkill mapping\u0026thinsp;+\u0026thinsp;seed-mimic trials\u0026thinsp;+\u0026thinsp;cameras\u0026mdash;offers a replicable protocol to detect hotspots, implement measures, and evaluate effectiveness (ideally in BACI designs). Update hotspot maps regularly and after interventions.\u003c/p\u003e \u003cp\u003eThe hotspot-first approach is transferable to other roaded mosaics where mesocarnivores disperse seeds. While evidence was strongest for road surface and more limited for road\u0026ndash;forest context and other covariates, the expected gains and co-benefits justify pilots on paved, non-edge segments with confirmed hotspots, coupled with adaptive monitoring. This aligns road-safety and conservation mandates and offers a pragmatic route to reconcile mobility infrastructure with living connectivity.\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eJ.C.: Data curation, Formal analysis, Investigation, Visualization, Writing \u0026ndash; original draft. M.N.B.: Supervision, Writing \u0026ndash; review \u0026amp; editing. P.G.V.: Conceptualization, Supervision, Resources, Formal analysis, Visualization, Writing \u0026ndash; review \u0026amp; editing.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eJC was supported by the Portuguese Science and Technology Foundation (FCT) through an individual grant (2021.05551.BD) and institutional funding to CEABN-InBIO (UIDB/50027/2025) and InBIO (DOI 10.54499/LA/P/0048/2020). JC and PGV were funded by FCT through CE3C (UID/00329/2025) and CHANGE (DOI 10.54499/LA/P/0121/2020). PGV received funding from the AdaptForGrazing project (PRR-C05-i03-I-000035-LA4.3/4.4/4.6/4.7), supported by the EU (RRF) via IFAP (PRR). We are grateful to Francesco Valerio and Denis Medinas for their valuable support.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe dataset analysed during the current study is available from the corresponding authors on reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eAndersen GE, Johnson CN, Barmuta LA, Jones ME (2017) Use of anthropogenic linear features by two medium-sized carnivores in reserved and agricultural landscapes. Sci. Rep\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003cstrong\u003e7\u003c/strong\u003e:11624. https://doi.org/10.1038/s41598-017-11454-z\u003c/li\u003e\n \u003cli\u003eAscens\u0026atilde;o F, Grilo C, LaPoint S, Tracey J, Clevenger AP, Santos-Reis M (2014) Inter-individual variability of stone marten behavioral responses to a highway. PLoS One 9:e103544. https://doi.org/10.1371/journal.pone.0103544\u003c/li\u003e\n \u003cli\u003eAzedo R, Ilh\u0026eacute;u A, Santos S, Vaz PG (2022) Carnivores cross irrigation canals more often through overpasses than through culverts. Basic Appl. Ecol. 61:53\u0026ndash;67. https://doi.org/10.1016/j.baae.2022.03.004\u003c/li\u003e\n \u003cli\u003eBaillargeon S, Rivest L.P. (2009) Rcapture: loglinear models for capture\u0026ndash;recapture experiments. R package version 1.4-4. https://CRAN.R-project.org/package=Rcapture\u003c/li\u003e\n \u003cli\u003eBarrientos R, Virg\u0026oacute;s E. (2006) Reduction of potential food interference in two sympatric carnivores by sequential use of shared resources. Acta Oecol. 30:107\u0026ndash;116. https://doi.org/10.1016/j.actao.2006.02.006\u003c/li\u003e\n \u003cli\u003eB\u0026iacute;l M, Andr\u0026aacute;\u0026scaron;ik R, Jano\u0026scaron;ka Z (2013) Identification of hazardous road locations of traffic accidents by means of kernel density estimation and cluster significance evaluation. Accid. Anal. Prev. 55:265\u0026ndash;273. https://doi.org/10.1016/j.aap.2013.03.003\u003c/li\u003e\n \u003cli\u003eBueno RD, Garc\u0026iacute;a D, Galetti M, La Mantia T. (2021) Trophic and spatial complementarity on seed dispersal services by birds, wild mammals, and cattle in a Mediterranean woodland pasture. Glob. Ecol. Conserv. 31:e01880. https://doi.org/10.1016/j.gecco.2021.e01880\u003c/li\u003e\n \u003cli\u003eBugalho MN, Plieninger T, Aronson J, Ellatifi M, Crespo DG (2009) Open woodlands: a diversity of uses (and overuses). In: Aronson J, Pereira JS and Pausas J (Eds.) Cork Ak Woodlands on the Edge: Ecology, Biogeography, and Restoration of an Ancient Mediterranean Ecosystem. \u003cem\u003eIsland Press\u003c/em\u003e, Washington DC, 33\u0026ndash;45.\u003c/li\u003e\n \u003cli\u003eBugalho MN, Caldeira MC, Pereira JS, Aronson J, Pausas JG (2011) Mediterranean cork oak savannas require human use to sustain biodiversity and ecosystem services. Front. Ecol. Environ. 9:278\u0026ndash;286. https://doi.org/10.1890/100084\u003c/li\u003e\n \u003cli\u003eB\u0026uuml;rkner PC (2017) brms: An R package for Bayesian multilevel models using Stan. J. Stat. Softw. 80. https://doi.org/10.18637/jss.v080.i01\u003c/li\u003e\n \u003cli\u003eCagnacci F, Meriggi A, Lovari S (2004) Habitat selection by the red fox \u003cem\u003eVulpes vulpes\u0026nbsp;\u003c/em\u003e(L. 1758) in an Alpine area. Ethol. Ecol. Evol. 16:103\u0026ndash;116. https://doi.org/10.1080/08927014.2004.9522640\u003c/li\u003e\n \u003cli\u003eCarmona G, Virg\u0026oacute;s E, Burgos T, Barrientos R (2024) Factors determining roadkills in a mammal carnivore are road-type specific. Mamm. Biol. 104:175\u0026ndash;183. https://doi.org/10.1007/s42991-024-00400-y\u003c/li\u003e\n \u003cli\u003eCarrijo TB, da Silva AR (2017) Modified Moran\u0026apos;s I for small samples. Geogr. Anal. 49:451\u0026ndash;467. https://doi.org/10.1111/gean.12130\u003c/li\u003e\n \u003cli\u003eCarvalho F, Louren\u0026ccedil;o A, Carvalho R, Alves PC, Mira A, Beja P (2018) The effects of a motorway on movement behaviour and gene flow in a forest carnivore: Joint evidence from road mortality, radio tracking and genetics. Landsc. Urban Plann. 178:217\u0026ndash;227. https://doi.org/10.1016/j.landurbplan.2018.06.007\u003c/li\u003e\n \u003cli\u003eCarvalho F, Mira A (2011) Comparing annual vertebrate road kills over two time periods, 9 years apart: a case study in Mediterranean farmland. Eur. J. Wildl. Res. 57:157\u0026ndash;174. https://doi.org/10.1007/s10344-010-0410-0\u003c/li\u003e\n \u003cli\u003eCastro H, Freitas H (2009) Aboveground biomass and productivity in the Montado: from herbaceous to shrub dominated communities. J. Arid Environ. 73:506\u0026ndash;511. https://doi.org/10.1016/j.jaridenv.2008.12.009\u003c/li\u003e\n \u003cli\u003eChen W, Zhong J, Carson WP, Tang Z, Xie Z, Sun S, Zhou Y (2019) Proximity to roads disrupts rodents\u0026apos; contributions to seed dispersal services and subsequent recruitment dynamics. J. Ecol. 107:2623\u0026ndash;2634. https://doi.org/10.1111/1365-2745.13221\u003c/li\u003e\n \u003cli\u003eClevenger AP, Chruszcz B, Gunson KE (2003) Spatial patterns and factors influencing small vertebrate fauna road-kill aggregations. Biol. Conserv. 109:15\u0026ndash;26. https://doi.org/10.1016/S0006-3207(02)00127-1\u003c/li\u003e\n \u003cli\u003eCook TC, Blumstein DT (2013) The omnivore\u0026rsquo;s dilemma: Diet explains variation in vulnerability to vehicle collision mortality. Biol. Conserv. 167:310\u0026ndash;315. https://doi.org/10.1016/j.biocon.2013.08.016\u003c/li\u003e\n \u003cli\u003eCosta IMDC, Ferreira MS, Mour\u0026atilde;o CLB, Bueno C (2022) Spatial patterns of carnivore roadkill in a high-traffic-volume highway in the endangered Brazilian Atlantic Forest. Mamm. Biol. 102:477\u0026ndash;487. https://doi.org/10.1007/s42991-022-00247-1\u003c/li\u003e\n \u003cli\u003eCraveiro J, Bernardino J, Mira A, Vaz PG (2019) Impact of culvert flooding on carnivore crossings. J. Environ. Manage. 231:878\u0026ndash;885. https://doi.org/10.1016/j.jenvman.2018.10.108\u003c/li\u003e\n \u003cli\u003eCraveiro J, Bugalho MN, Vaz PG (2025) Roads shape seed dispersal by rodents and carnivores at forest edges. J. Appl. Ecol. 00:1\u0026ndash;15. https://doi.org/10.1111/1365-2664.70174\u003c/li\u003e\n \u003cli\u003eDavies T, Marshall J, Hazelton M (2017) Tutorial on Kernel Estimation of Continuous Spatial and Spatiotemporal Relative Risk with Accompanying Instruction in R, pp. 1\u0026ndash;30. R package version 2.2\u0026ndash;13. https://doi.org/10.48550/arXiv.1707.06888\u003c/li\u003e\n \u003cli\u003eDekker JA, Stein A, Heitk\u0026ouml;nig IMA (2001) A spatial analysis of a population of red fox (Vulpes vulpes) in the Dutch coastal dune area. J. Zool. 255:505\u0026ndash;510. https://doi.org/10.1017/S0952836901001583\u003c/li\u003e\n \u003cli\u003eDelgado JD, Morelli F, Arroyo NL, Dur\u0026aacute;n J, Rodr\u0026iacute;guez A, Rosal A, Palenzuela MDV, Rodr\u0026iacute;guez JDGP (2018) Is vertebrate mortality correlated to potential permeability by underpasses along low-traffic roads? J. Environ. Manage. 221:53\u0026ndash;62. https://doi.org/10.1016/j.jenvman.2018.05.043\u003c/li\u003e\n \u003cli\u003eDoncaster CP, Macdonald DW (1991) Drifting territoriality in the red fox Vulpes vulpes. J. Anim. Ecol. 60:423\u0026ndash;439. https://doi.org/10.2307/5288\u003c/li\u003e\n \u003cli\u003eFabrizio M, Di Febbraro M, D\u0026rsquo;Amico M, Frate L, Roscioni F, Loy A (2019) Habitat suitability vs landscape connectivity determining roadkill risk at a regional scale: a case study on European badger (Meles meles). Eur. J. Wildl. Res. 65:1\u0026ndash;10. https://doi.org/10.1007/s10344-018-1241-7\u003c/li\u003e\n \u003cli\u003eFarris E, Canopoli L, Cucca E, Landi S, Maccioni A, Filigheddu R (2017) Foxes provide a direct dispersal service to Phoenician junipers in Mediterranean coastal environments: ecological and evolutionary implications. Plant Ecol. Evol. 150:117\u0026ndash;128. https://doi.org/10.5091/plecevo.2017.1277.\u003c/li\u003e\n \u003cli\u003eFedriani JM, Wiegand T, Garrote PJ, Leiva MJ, Ayll\u0026oacute;n D (2023) Seed dispersal effectiveness in fragmented and defaunated landscapes. Ecosphere 14:e4658. https://doi.org/10.1002/ecs2.4658\u003c/li\u003e\n \u003cli\u003eFerreira EM, Fernandes N, Craveiro J, Valerio F, Costa P, Carrapato C, Mira A, Medinas D, Silva JP, Santos SM (2022) Assessing behaviour states of a forest carnivore in a road-dominated landscape using Hidden Markov Models.\u003cem\u003e\u0026nbsp;\u003c/em\u003eNat. Conserv. 47:155\u0026ndash;175. https://doi.org/10.3897/natureconservation.47.72781\u003c/li\u003e\n \u003cli\u003eForman RT, Alexander LE (1998) Roads and their major ecological effects. Annu. Rev. Ecol. Evol. Syst. 29:207\u0026ndash;231. https://doi.org/10.1146/annurev.ecolsys.29.1.207\u003c/li\u003e\n \u003cli\u003eFrangini L, Sterrer U, Franchini M, Pesaro S, R\u0026uuml;disser J, Filacorda S (2022) Stay home, stay safe? High habitat suitability and environmental connectivity increases road mortality in a colonizing mesocarnivore. Landsc. Ecol. 37:2343\u0026ndash;2361. https://doi.org/10.1007/s10980-022-01491-z\u003c/li\u003e\n \u003cli\u003eGalantinho A, Herrera JM, Eufr\u0026aacute;zio S, Silva C, Carvalho F, Alpizar-Jara R, Mira A. (2020) Road verges provide connectivity for small mammals: A case study with wood mice (Apodemus sylvaticus) in an agro-silvo pastoral system. J. Environ. Manage. 258:110033. https://doi.org/10.1016/j.jenvman.2019.110033\u003c/li\u003e\n \u003cli\u003eGelman A (2025) Prior choice recommendations. Retrieved from https://github.com/stan-dev/stan/wiki/prior-choice-recommendations\u003c/li\u003e\n \u003cli\u003eGenes L, Dirzo R (2022) Restoration of plant-animal interactions in terrestrial ecosystems. Biol. Conserv. 265:109393. https://doi.org/10.1016/j.biocon.2021.109393\u003c/li\u003e\n \u003cli\u003eGonz\u0026aacute;lez‐Varo JP, L\u0026oacute;pez‐Bao JV, Guiti\u0026aacute;n J (2013) Functional diversity among seed dispersal kernels generated by carnivorous mammals. J. Anim. Ecol. 82:562\u0026ndash;571. https://doi.org/10.1111/1365-2656.12024\u003c/li\u003e\n \u003cli\u003eGrilo C, Ascens\u0026atilde;o F, Santos-Reis M, Bissonette JA (2011) Do well-connected landscapes promote road-related mortality? Eur. J. Wildl. Res.57:707\u0026ndash;716. https://doi.org/10.1007/s10344-010-0478-6\u003c/li\u003e\n \u003cli\u003eGrilo C, Bissonette JA, Santos-Reis M (2008) Response of carnivores to existing highway culverts and underpasses: implications for road planning and mitigation. Biodivers. Conserv. 17:1685\u0026ndash;1699. https://doi.org/10.1007/s10531-008-9374-8\u003c/li\u003e\n \u003cli\u003eGrilo C, Bissonette JA, Santos-Reis M (2009) Spatial\u0026ndash;temporal patterns in Mediterranean carnivore road casualties: consequences for mitigation. Biodivers. Conserv. 142:301\u0026ndash;313. https://doi.org/10.1016/j.biocon.2008.10.026\u003c/li\u003e\n \u003cli\u003eGrilo F, Ferreira E, Alcobia S, Sim\u0026otilde;es L, Santos-Reis M (2016) Do fine-scale factors shape the use of riparian galleries by carnivores in a Mediterranean agro-forested environment? Int. J. Environ. Agric. Res. 2:2454\u0026ndash;1850.\u003c/li\u003e\n \u003cli\u003eGrilo C, Koroleva E, Andr\u0026aacute;\u0026scaron;ik R, B\u0026iacute;l M, Gonz\u0026aacute;lez‐Su\u0026aacute;rez M (2020) Roadkill risk and population vulnerability in European birds and mammals. Front. Ecol. Environ. 18:323\u0026ndash;328. https://doi.org/10.1002/fee.2216\u003c/li\u003e\n \u003cli\u003eGrilo C, Smith DJ, Klar N (2015) Carnivores: struggling for survival in roaded landscapes. In: van der Ree R, Smith DJ, Grilo C (Eds.), Handbook of Road Ecology, first ed. John Wiley \u0026amp; Sons, Ltd.\u003c/li\u003e\n \u003cli\u003eGr\u0026uuml;newald C, Breitbach N, B\u0026ouml;hning-Gaese K (2010) Tree visitation and seed dispersal of wild cherries by terrestrial mammals along a human land-use gradient. Basic Appl. Ecol. 11:532\u0026ndash;541. https://doi.org/10.1016/j.baae.2010.07.007\u003c/li\u003e\n \u003cli\u003eH\u0026auml;m\u0026auml;l\u0026auml;inen A, Broadley K, Droghini A, Haines JA, Lamb CT, Boutin S, Gilbert S (2017) The ecological significance of secondary seed dispersal by carnivores. Ecosphere 8:e01685. https://doi.org/10.1002/ecs2.1685\u003c/li\u003e\n \u003cli\u003eHerrera JM, Teixeira IS, Rodr\u0026iacute;guez-P\u0026eacute;rez J, Mira A (2015) Landscape structure shapes carnivore-mediated seed dispersal kernels. Landsc. Ecol. 31:731\u0026ndash;743. https://doi.org/10.1007/s10980-015-0283-4\u003c/li\u003e\n \u003cli\u003eHobbs NT, Hooten MB (2015) Bayesian Models: A Statistical Primer for Ecologists. Princeton University Press, Princeton, NJ.\u003c/li\u003e\n \u003cli\u003eHoffmann A, Decher J, Rovero F, Schaer J, Voigt C, Wibbelt G (2010) Field methods and techniques for monitoring mammals. Manual on field recording techniques and protocols for all taxa biodiversity inventories 8:482\u0026ndash;529.\u003c/li\u003e\n \u003cli\u003eJ\u0026auml;gerbrand AK, Antonson H, Ahlstr\u0026ouml;m C (2018) Speed reduction effects over distance of animal-vehicle collision countermeasures \u0026ndash; a driving simulator study. Eur. Transp. Res. Rev. 10:40. https://doi.org/10.1186/s12544-018-0314-8\u003c/li\u003e\n \u003cli\u003eJumeau J, Boucharel P, Handrich Y, Burel F. (2017) Road-related landscape elements as a habitat: A main asset for small mammals in an intensive farming landscape. Basic Appl. Ecol. 25:15\u0026ndash;27. https://doi.org/10.1016/j.baae.2017.09.013\u003c/li\u003e\n \u003cli\u003eKruschke JK (2015) Doing Bayesian Data Analysis: A Tutorial with R, JAGS, and Stan (2nd ed.). Academic Press (Elsevier), Amsterdam.\u003c/li\u003e\n \u003cli\u003eL\u0026uuml;decke D, Ben-Shachar MS, Patil I, Waggoner P, Makowski D (2021) performance: An R Package for Assessment, Comparison and Testing of Statistical Models. J. Open Source Softw. 6:3139. https://doi.org/10.21105/joss.03139\u003c/li\u003e\n \u003cli\u003eMata C, Ruiz-Capillas P, Malo JE (2017) Small-scale alterations in carnivore activity patterns close to motorways. Eur. J. Wildl. Res. 63:1\u0026ndash;12. https://doi.org/10.1007/s10344-017-1118-1\u003c/li\u003e\n \u003cli\u003eMcElreath R (2018) Statistical rethinking: A Bayesian course with examples in R and Stan. Chapman and Hall, CRC texts in statistical science. https://doi.org/10.1201/9780429029608\u003c/li\u003e\n \u003cli\u003eMedinas D, Marques JT, Costa P, Santos S, Rebelo H, Barbosa AM, Mira A (2021) Spatiotemporal persistence of bat roadkill hotspots in response to dynamics of habitat suitability and activity patterns. J. Environ. Manage. 277:111412. https://doi.org/10.1016/j.jenvman.2020.111412\u003c/li\u003e\n \u003cli\u003eMorelle K, Lehaire F, Lejeune P (2013) Spatio-temporal patterns of wildlife-vehicle collisions in a region with a high-density road network. Nat. Conserv. 5:53\u0026ndash;73. 10.3897/natureconservation.5.4634\u003c/li\u003e\n \u003cli\u003eNathan R, Muller-Landau HC (2000) Spatial patterns of seed dispersal, their determinants and consequences for recruitment. Trends Ecol. Evol. 15:278\u0026ndash;285. https://doi.org/10.1016/S0169-5347(00)01874-7\u003c/li\u003e\n \u003cli\u003ePeredo A, Mart\u0026iacute;nez D, Rodr\u0026iacute;guez-P\u0026eacute;rez J, Garc\u0026iacute;a D (2013) Mammalian seed dispersal in Cantabrian woodland pastures: network structure and response to forest loss. Basic Appl. Ecol. 14:378\u0026ndash;386. https://doi.org/10.1016/j.baae.2013.05.003\u003c/li\u003e\n \u003cli\u003ePlanillo A, Mata C, Manica A, Malo JE (2018) Carnivore abundance near motorways related to prey and roadkills. J. Wildl. Manag. 82:319\u0026ndash;327. https://doi.org/10.1002/jwmg.21384\u003c/li\u003e\n \u003cli\u003eQuiles P, Barrientos R (2024) Interspecific interactions disrupted by roads. Biol. Rev. 99:1121\u0026ndash;1139. https://doi.org/10.1111/brv.13061\u003c/li\u003e\n \u003cli\u003eQuintana I, Cifuentes EF, Dunnink JA, Ariza M, Mart\u0026iacute;nez-Medina D, Fantacini FM, Shrestha BR, Richard FJ (2022) Severe conservation risks of roads on apex predators. Sci. Rep. 12:2902. https://doi.org/10.1038/s41598-022-05294-9\u003c/li\u003e\n \u003cli\u003eR Core Team (2025) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/\u003c/li\u003e\n \u003cli\u003eRoemer GW, Gompper ME, Van Valkenburgh B (2009) The ecological role of the mammalian mesocarnivore. BioScience 59:165\u0026ndash;173. https://doi.org/10.1525/bio.2009.59.2.9\u003c/li\u003e\n \u003cli\u003eRosalino LM, Rosa S, Santos-Reis M (2010) The role of carnivores as Mediterranean seed dispersers. Ann. Zool. Fennici 47:195\u0026ndash;205. https://doi.org/10.5735/086.047.0304\u003c/li\u003e\n \u003cli\u003eRuiz-Capillas P, Mata C, Malo JE (2013) Road verges are refuges for small mammal populations in extensively managed Mediterranean landscapes. Biol. Conserv. 158:223\u0026ndash;229. https://doi.org/10.1016/j.biocon.2012.09.025\u003c/li\u003e\n \u003cli\u003e\u0026Scaron;\u0026aacute;lek M, Červinka J, Pady\u0026scaron;\u0026aacute;kov\u0026aacute; E, Kreisinger J (2014) Does spatial co-occurrence of carnivores in a Central European agricultural landscape follow the null model?. Eur J Wildl Res 60:99\u0026ndash;107. https://doi.org/10.1007/s10344-013-0755-2\u003c/li\u003e\n \u003cli\u003eSalgueiro V, Silva C, Eufr\u0026aacute;zio S, Salgueiro PA, Vaz PG (2019) Endozoochory of a dry‐fruited tree aids quarry passive restoration and seed soaking further increases seedling emergence. Restor. Ecol. 28:668\u0026ndash;678. https://doi.org/10.1111/rec.13018\u003c/li\u003e\n \u003cli\u003eSantos SM, Marques JT, Louren\u0026ccedil;o A, Medinas D, Barbosa AM, Beja P, Mira A (2015) Sampling effects on the identification of roadkill hotspots: implications for survey design. J. Environ. Manage. 162:87\u0026ndash;95. https://doi.org/10.1016/j.jenvman.2015.07.037\u003c/li\u003e\n \u003cli\u003eSantos RAL, Santos SM, Santos-Reis M, Figueiredo AP, Bager A, Aguiar LM, Ascensao F (2016) Carcass persistence and detectability: reducing the uncertainty surrounding wildlife-vehicle collision surveys. PloSone 11:e0165608. https://doi.org/10.1371/journal.pone.0165608\u003c/li\u003e\n \u003cli\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. PLosone 6:e25383. https://doi.org/10.1371/journal.pone.0025383\u003c/li\u003e\n \u003cli\u003eSantos RA, Mota-Ferreira M, Aguiar LM, Ascens\u0026atilde;o F (2018) Predicting wildlife road-crossing probability from roadkill data using occupancy-detection models. Sci. Total Environ. 642:629\u0026ndash;637. https://doi.org/10.1016/j.scitotenv.2018.06.107\u003c/li\u003e\n \u003cli\u003eSeiler A, Helldin JO, Seiler C (2004) Road mortality in Swedish mammals: results of a drivers\u0026apos; questionnaire. Wildlife Biol. 10:225\u0026ndash;233. https://doi.org/10.2981/wlb.2004.028\u003c/li\u003e\n \u003cli\u003eShilling F, Nguyen T, Saleh M, Kyaw MK, Tapia K, Trujillo G, Bejarano M, Waetjen D, Peterson J, Kalisz G, Sejour R, Croston S, Ham E (2021) A reprieve from US wildlife mortality on roads during the COVID-19 pandemic. Biol. Conserv. 256:109013. https://doi.org/10.1016/j.biocon.2021.109013\u003c/li\u003e\n \u003cli\u003eSu\u0026aacute;rez-Esteban A, Delibes M, Fedriani JM, Dickman C (2013a) Barriers or corridors? The overlooked role of unpaved roads in endozoochorous seed dispersal. J. Appl. Ecol. 50:767\u0026ndash;774. https://doi.org/10.1111/1365-2664.12080\u003c/li\u003e\n \u003cli\u003eSu\u0026aacute;rez-Esteban A, Delibes M, Fedriani JM (2013b) Unpaved road verges as hotspots of fleshy-fruited shrub recruitment and establishment. Biol. Conserv. 167:50\u0026ndash;56. https://doi.org/10.1016/j.biocon.2013.07.022\u003c/li\u003e\n \u003cli\u003eTaylor SK, Buergelt CD, Roelke-Parker ME, Homer BL, Rotstein, DS (2002) Causes of mortality of free-ranging Florida panthers. J. Wildl. Dis. 38:107\u0026ndash;114. https://doi.org/10.7589/0090-3558-38.1.107\u003c/li\u003e\n \u003cli\u003eTigas LA, Van Vuren DH, Sauvajot RM (2002) Behavioral responses of bobcats and coyotes to habitat fragmentation and corridors in an urban environment. Biol. Conserv. 108:299\u0026ndash;306. https://doi.org/10.1016/S0006-3207(02)00120-9\u003c/li\u003e\n \u003cli\u003eValerio F, Carvalho F, Barbosa AM, Mira A, Santos SM (2019) Accounting for connectivity uncertainties in predicting roadkills: a comparative approach between path selection functions and habitat suitability models. Environ. Manage. 64:329\u0026ndash;343. https://doi.org/10.1007/s00267-019-01191-6\u003c/li\u003e\n \u003cli\u003eVan Etten KW, Wilson KR, Crabtree RL (2007) Habitat use of red foxes in Yellowstone National Park based on snow tracking and telemetry. J. Mammal. 88:1498\u0026ndash;1507. https://doi.org/10.1644/07-mamm-a-076.1\u003c/li\u003e\n \u003cli\u003eVaz PG, Bugalho MN, Fedriani JM, Branco M, Lecomte X, Nogueira C, Caldeira MC (2019) Unravelling associations between tree-seedling performance, herbivory, competition, and facilitation in high nature value farmlands. J. Environ. Manage. 232:1066\u0026ndash;1074. https://doi.org/10.1016/j.jenvman.2018.11.082\u003c/li\u003e\n \u003cli\u003eVaz PG, Bugalho MN, Fedriani JM (2024) Grazing hinders seed dispersal during crop failure in a declining oak woodland. Sci. Total Environ. 907:167835. https://doi.org/10.1016/j.scitotenv.2023.167835\u003c/li\u003e\n \u003cli\u003eVaz PG, Merten EC, Robinson CT, Pinto P (2021) Severely burned wood from wildfires has low functional potential in streams. J. Appl. Ecol. 58:1346\u0026ndash;1356\u003cstrong\u003e.\u0026nbsp;\u003c/strong\u003ehttps://doi.org/10.1111/1365-2664.13872\u003c/li\u003e\n \u003cli\u003eVincent JP, Gaillard JM, Bideau E. (1991) Kilometric index as biological indicator for monitoring forest roe dear populations. Acta Theriol. 36:315\u0026ndash;328.\u003c/li\u003e\n \u003cli\u003eWadud AI, Bugalho MN, Vaz PG (2024) Meta-analysis of livestock effects on tree regeneration in oak agroforestry systems. J. Environ. Manage.\u003cem\u003e\u0026nbsp;\u003c/em\u003e364:121498. https://doi.org/10.1016/j.jenvman.2024.121498\u003c/li\u003e\n \u003cli\u003eWang BC, Smith TB (2002) Closing the seed dispersal loop. Trends Ecol. Evol. 17:379\u0026ndash;385. https://doi.org/10.1016/S0169-5347(02)02541-7\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"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":"Seed dispersal, Road ecology, Forest fragmentation, Carnivores, Hotspots, Landscape connectivity","lastPublishedDoi":"10.21203/rs.3.rs-9314010/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9314010/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eRoadkill hotspots concentrate animal movement and mortality, but whether they also interfere with animal-mediated seed dispersal\u0026mdash;a conservation-relevant process underpinning plant regeneration and landscape connectivity\u0026mdash;remains unknown. In Mediterranean oak woodlands of southern Portugal, we used seed mimics to test how road type (paved vs. unpaved) and road\u0026ndash;forest context (edge vs. non-edge) shape carnivore-mediated seed crossings, and whether crossings coincide with roadkill hotspots detected by kernel density analysis. Bayesian logistic models indicated that seed-crossing probability was about twice as high on unpaved as on paved roads (predicted means 0.28 vs. 0.14), with weaker evidence for a negative edge effect. Crossing probability also tended to increase with carnivore abundance and distance to streams, and decrease with rodent density, albeit with some uncertainty. Crucially, paved sections intersecting roadkill hotspots showed nearly threefold higher predicted crossing probabilities than non-hotspot sections (0.51 vs. 0.18). These results yield clear conservation implications: prioritize paved segments that overlap carnivore roadkill hotspots. Implement precision measures\u0026mdash;short-segment speed management, directional fencing that funnels animals to existing culverts/underpasses, and verge management that removes carcasses and limits small-mammal/scavenging attractants\u0026mdash;to cut mortality while maintaining seed-mediated connectivity. A light, repeatable monitor\u0026ndash;act\u0026ndash;reassess workflow (hotspot mapping\u0026thinsp;+\u0026thinsp;seed-mimic trials\u0026thinsp;+\u0026thinsp;camera checks) enables rapid deployment and evaluation, and is transferable to other human-dominated forest mosaics. Hotspot-guided mitigation offers a pragmatic path to reconcile mobility infrastructure with living connectivity.\u003c/p\u003e","manuscriptTitle":"Carnivore-mediated seed crossings are more likely on roadkill hotspots","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-06 12:39:02","doi":"10.21203/rs.3.rs-9314010/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"291443371594870583048203197863671947160","date":"2026-05-04T18:53:22+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"280991178029135238878244522723254737587","date":"2026-05-03T17:04:34+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-28T06:10:48+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-24T02:37:26+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-06T11:23:08+00:00","index":"","fulltext":""},{"type":"submitted","content":"Biodiversity and Conservation","date":"2026-04-03T14:31:17+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":"00597e29-c58a-426f-8cda-435d8530e8da","owner":[],"postedDate":"May 6th, 2026","published":true,"recentEditorialEvents":[{"type":"reviewerAgreed","content":"291443371594870583048203197863671947160","date":"2026-05-04T18:53:22+00:00","index":16,"fulltext":""},{"type":"reviewerAgreed","content":"280991178029135238878244522723254737587","date":"2026-05-03T17:04:34+00:00","index":15,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-06T12:39:02+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-06 12:39:02","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9314010","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9314010","identity":"rs-9314010","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}