Geography of roadkills within the Tropical Andes Biodiversity Hotspot: Poorly known vertebrates are part of the toll

We explore the effect of roads in animal mortality within Biodiversity Hotspot with highest number endemic species vertebrates on Earth, Tropical Andes. Our objectives were to know which are killed this particularly biodiversity-rich area and how landscape composition configuration influences roadkills. systematically looked for roadkills along that border three protected areas Ecuadorian To evaluate our hypotheses, we used correlation, logistic regression, GIS analyses. surveyed a total 7128 km observed roadkill rate 6.24 (95% CI = 5.35–7.14) individuals per 100 km/day. Roadkills included poorly known endangered vertebrates; among them, one undescribed snake genus Atractus. Most by pastures, dominant vegetation study area. more likely occur near bridges frequent at greater distances from natural vegetation, towns, rivers. conclude pastures may be functioning as ecological traps small vertebrates. Mitigation measures could include increasing road permeability wildlife constructing culverts critical points where is high, adaptation beneath them function effectively underpasses. These should complemented fences exclude passages pastures. encourage development similar studies inform mitigation can adapted local conditions. Exploramos el efecto que tienen las carreteras en la mortalidad de animales dentro del Biodiversidad con mayor número especies endémicas Tierra, los Andes Tropicales. Nuestros objetivos fueron conocer qué son atropelladas esta área particularmente biodiversa, y cómo composición configuración paisaje influencian atropellamientos. De forma sistemática, buscamos atropellamientos vida silvestre borde tres áreas protegidas ecuatorianos. Para evaluar nuestras hipótesis, usamos métodos estadísticos tales como correlación regresión logística, análisis SIG. Muestreamos un 7218 obtuvimos una tasa individuos por km/día. Los incluyeron amenazadas, poco conocidas; entre ellas, especie serpiente no descrita género La mayoría ocurrió junto pastizales, vegetación domina ambos márgenes carretera nuestro sitio estudio. tuvieron probabilidad ocurrir puentes más frecuentes distancia natural, poblados ríos. Concluimos pastizales pueden estar funcionando trampas ecológicas para pequeños conocidos vertebrados. Las medidas mitigación podrían incluir incremento permeabilidad fauna medio construir drenajes comunican lados puntos críticos donde es alta, adaptación superficies bajo funcionen pasos fauna. Estas mediadas deberían complementadas vallas excluir vertebrados cercanas pastizales. Alentamos se realicen estudios similares poseen alta biodiversidad finalidad informar adapten condiciones locales. There over 40,000,000 paved roadway lanes world (Dulac, 2013). Many those dissect habitats they cause multiple detrimental effects. For example, catalyzers habitat loss fragmentation, facilitate introduction invasive species, interfere hydrological cycles, associated pollution water sources soil, other effects (Coffin, 2007; Forman et al. 2003). Wildlife heavily affected presence roads, hinder dispersal an important source due vehicle collisions (Ament 2008; As example global mortality, it estimated every year 194 million birds 29 mammals European (Grilo 2020), 80 United States (Erickson al., 2005), while eight two Brazilian (González-Suárez 2018). Therefore, suspected vehicles have negative consequences genetic diversity populations dissected (Jackson & Fahrig, 2011). This threat numbers; between 2015 2016, 79 percent Florida panther (Felis concolor coryi) was caused (FWC, 2016). The varies considerably across regions. Goiás-Brazil, represent 74.21 roadkilled (Silveira-Miranda 2017); Costa Rica, amphibians accounted 93.5 events (Arévalo Tanzania, 50 animals (Kioko 2015). variation responds intrinsic differences communities, configuration, characteristics (Ciocheti 2017; Clevenger 2003; Medinas 2013), will unique each site. However, relationship rates still needs explored, tropical In study, analyze relationships different attributes network dissects vertebrates, (Myers 2000). Understanding anthropogenic disturbances, such roadkills, megadiverse because large organisms, including environmental impacts. specific research follows: (1) determine frequency roadkills; (2) related (3) vehicular traffic located Ecuador 2000) adjacent areas: Antisana Ecological Reserve (120,000 ha), Sumaco-Napo-Galeras National Park (205,751 Cayambe-Coca (404,103 ha) (Figure 1). studied 33-km segments two-lane run these link town Baeza towns Papallacta, El Chaco, Cosanga Road segment Baeza-Papallacta altitude 1827–2995 m, Baeza-El Chaco 1476–1914 Baeza-Cosanga 1805–2285 m. All had lanes, speed limits varied 20 90 km/h S1). corresponds evergreen low mountain forest lower elevations (1300–2000 m) cloud higher (2000–2900 (Valencia 1999). Agricultural lands exist margins dominated pasturelands cattle ranching. Based climatic stations closer area, Papallacta (3150 Sardines (1615 m), annual rainfall ranges 1189 mm 3180 mm, average temperature fluctuates 10.17 19.5°C, respectively. weekly, periods or consecutive days March August 2014. During survey day, driving both directions 40 km/h. monitored during day 0800 h 1700 same observer (PMV) who accompanied driver. Each photographed, identified lowest possible taxonomic level, georeferenced GPS unit (Garmin GPSmap 76 Cx). Also, when identification field not (e.g., some reptiles amphibians), carcasses taken Museum Zoology Pontificia Universidad Católica (QCAZ) obtain feedback group specialists. removed after their avoid double counting days. results previous temperate regions (e.g. given vertebrates’ biological traits (Rytwinski 2012), hypothesized importance predict vary taxa. test hypothesis, developed set regression models taxon. confronted UTM coordinates equal random represented places observed. Both, five explain Euclidean to: nearest river, remnant bridge (i.e., intersections ravines rivers roads), (4) town. fifth attribute percentage cover inside buffer m distance point. Distance calculated based shapefile ecosystems (MAE, 2013); rest variables using shapefiles Instituto Geográfico Militar (IGM). covariates scaled 0 1 R package scales (Wickham Seidel, 2020). rationale hypotheses behind predictor follow: Rivers availability; if river close road, expect approach resource therefore exposed collisions. true organisms dependent capability vehicles, 2012; Seo measure fragmentation (Turner 2001). fragmented landscape, spatial requirements highly mobile vulnerable therefore, isolation larger patch) roads. Bridges crossing above underpasses (Clevenger 2003), infrastructure positively traffic, towns; find risk being (Lodé, 2000; Orlowski Nowak, 2006). (5) expected sections cross cover, has been (e.gMalo 2004; Sillero 2019). Before model implementation, evaluated multicollinearity variance inflation factor (VIF), vif car (Fox Weisberg, analyzed vertebrate classes amphibians, reptiles, birds, mammals) separately. datasets, all but opossum (Didelphis pernigra) another only opossum, most study. group, 32 regressions glm olsrr (Hebbali, 2020); null combinations predictive variables. corrected Akaike information criterion (AICc) select best fit (Burnham Anderson, 2002), dredge MuMIn (Barton, When than competitive ? AICc ?2), performed averaging model.avg estimate full averaged coefficients parameters. also cumulative Akaike's weights parameters identify group. there uncertainty choose best-fit several uninformative present, calculating weight covariate useful method (Arnold, 2010; Burnham 2002). Additionally, performance measuring under receiver operating characteristic (ROC) curve roc pROC (Robin describe amount surrounding land registered category sides forest, shrubs, crops) quantified combination category. Finally, visualize hot spots heatmaps visual representation depict high point density) kernel density estimation geographic system (Quantum v.3.12). shape, quartic interpolation search radius 500 m; chose resolution × pixel size output. combined data separately excluding opossums opossums. rates. Pearson correlation association daily frequency. purpose, datasets: dataset wildlife, sub-dataset first survey, older 24 h, excluded 48 h. age carcasses, trained ourselves spending 10 marking revisiting degree decomposition. passed 15 minutes before segment. 72 non-consecutive effort km. found 445 dead animals—domestic results—with least 44 level). majority corresponded 207 (46.31% roadkills) (Table 1); Didelphis pernigra (n 153; Table second 107 (23.94%) 16 species. Reptiles 88 (19.69%) snakes; notable Atractus S2; Torres-Carvajal pers. comm.). 43 (9.62%) 3 species; 34 belonged caecilians (Gymnophiona). Baeza–El 193); followed Baeza–Cosanga 149) Baeza–Papallacta 103). heterogeneous; approximately 23 forested remaining next human-modified types ranching, crops, urban areas. taxa, (40%, n 180) sites occupied covered (20%, 92) 2). did collinearity any VIF values 10). amphibians: (5.54) (5.71) Covariates predicted classes. bridge; model, appeared ?2, S2). negatively moderate collinearity; however, removing resulted worse performance; without rivers, ROC (AUC) 47.1%, while, bridges, AUC 75.3%. Modeling 85.29% S3). clear model; bridge, although four S2), 71.14% town, 76.32% mammals, Other ten ?2; S2); 65.5% covariates: S3), 63.4% heatmap shows road; frequently occurred pattern clearer case appear Baeza, Oritoyacu, Consanga S4). concentrated altitudes site, 1900 2300 groups, respectively Correlation analysis complete presented non-significant (r ?0.101, p 0.137, 216), result obtained trip ?0.140, 0.088, 150). less old weak significant ?0.141, 0.038, 216). It (Atractus sp.), exemplifies relevance understanding impact Hotspot. Three decades ago botanist Alwyn Gentry witnessed rapid disappearance plant deforestation Ecuador, Chocó Darién/Western (Dodson Gentry, 1991). believe deforestation, major lead extinctions. Populations capabilities might striking fossorial vagility) reptiles. Roads alarming snakes even relatively numbers enough disproportionally increase probability extinction (Row 2007). A noteworthy situation (Filius 2020; Quintero-Ángel impacted. Forty duboisi, whose distribution much eastern (Cisneros-Heredia Bustamante, 2017). Snakes primarily (Martins Oliveira, 1993) endemism. Of present 13 either deficient conservation status (Torres-Carvajal Similarly, caecilians, widely understudied endemic; 19 lack (Ron Another reported findings support need conduct Hotspots practically unknown severely networks. Among oncilla Leopardus tigrinus categorized IUCN red list (Payán felids Neotropics. geographical range thought Amazonia Andean foothills 2016); detected extensive camera trap surveys Ecuador's Amazonian correspond humid 400 m.a.s.l. (e.gEspinosa 2018; Mena Given poor knowledge available medium-sized charismatic oncilla, highlights studying biology ecology threatened understand having populations. contrast (D. pernigra), far 2 Opossums seem Neotropics (Monge-Nájera, 2018), reflects its abundance rural agriculture provide resources animals. systematic allows yearly 2279 1951–2606) wild site consider fraction die year. detectability limited factors weather, difficult see), removal predators, scavengers, fact (Santos despite underestimating communities. taxa pasturelands, landcover Central Colombia, (Quintero-Ángel 2012). type explained rodents) agricultural (Stenseth areas, crossed density. classes, positive words, further movement looking resource. finding opposite South Korea, amphibian bodies (Seo last unexpected would work passages; around Although seems counterintuitive, not. effective, fencing excludes (Cunnington 2014; Rytwinski Hence, tropics, Andes, numerous infrastructures surroundings reduce mortality. Distances That is, patches, killed, consistent prediction habitats. Animals capacity look (Rytwinski, increases, biodiversity hotspots catalyze extinction. Malo concentrate structural connectivity, use disperse observed, (Orlowski, patches small. fragments inhabited patch, reduces exposure do move often. birds; towns. dwelling habituated pressures, learning reflected bird Brown Brown, 2013; Mumme possibility already depleted, reflect numerical availability away Previous volume Lin, 2016; Tejera variable Carvalho Conard Gipson, contrary initial hypothesis surprising mixed literature regarding direction see Likely, threshold until perceived venture road. aspects length determining association. predicting showed groups management. current management tools, crossings effective wide variety animals, insects big carnivores, providing safe place maintain functional connectivity (Forman Polak expensive, existing drainage culverts, make permeable wildlife. structures barriers roads; methods 83% Dodd (2004) increased 10-fold installed Florida. diameter ?50 cm) promote (Lesbarrères Patrick Woltz 2008), placement transformed traps. An alternative complement creation modification limit deter high-risk which, ungulates (Rea, Tanner Leroux, generally private lands, requested landowners fence properties Ideally, permeability. recognize social economic reality unlikely able assume cost financial governmental non-governmental institutions. managed through secure underneath gullies amounts litter garbage, reason, simple low-cost strategy educational campaigns create public awareness policies people garbage deposits. Animal detection systems, mechanisms activate flashing lights warn drivers proven reducing signals become recommended times periods, movement) certain evidenced (Grace reflectors warning signs preventing drivers, respectively, installation reducers effectiveness transited problems traffic. notice various communities behavior (Jacobson 2016), itself construction drainages 2001, conditions (Garriga Rosa Bager, others. kind replicated develop informed tools respond particularities especially thank Fundación EcoCiencia funding research. grateful David Brito his assistance fieldwork Omar Torres-Carvajal, Fernando Ayala, Diego Paucar, Jefferson García help openly Dryad Digital Repository: https://doi.org/10.5061/dryad.xwdbrv1cp (Medrano-Vizcaíno Espinosa, 2021). Please note: publisher responsible content functionality supporting supplied authors. Any queries (other missing content) directed corresponding author article.