Crypsis in a polymorphic mammal along an urbanization gradient

Crypsis in a polymorphic mammal along an urbanization gradient | 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 Crypsis in a polymorphic mammal along an urbanization gradient Jessica Proctor, Alessandra Bryan, Bradley J. Cosentino, James P. Gibbs This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5462361/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 21 Mar, 2025 Read the published version in Urban Ecosystems → Version 1 posted 11 You are reading this latest preprint version Abstract Urbanization transforms landscapes and alters visual environments for polymorphic species that rely on cryptic coloration for survival, potentially generating urban-rural clines in pigmentation. Such clines are evident in eastern gray squirrel ( Sciurus carolinensis ) populations, for which a melanic morph is currently more prevalent in cities but was historically more prevalent in rural woodlands prior to urbanization. We compared the degree of crypsis between the two primary color morphs of gray squirrels – gray and melanic – among the suite of habitats that predominate along an urbanization gradient to test whether an altered visual environment may contribute to the maintenance of an urban-rural cline in morph prevalence. Crypsis was quantified using an online game with human observers and image pixel classification to measure detectability of taxidermic mounts of each morph against their backgrounds in replicate sites within each habitat. The melanic morph was more conspicuous than the gray morph in all habitat types and across seasons, as evidenced by greater detection probabilities by human observers and lower background matching. Coat color in gray squirrels likely mediates visual detection by predators, potentially resulting in selection against the more conspicuous, melanic morph in rural woodlands. Conversely, selection via road mortality may favor the melanic morph in urban areas if vehicular collisions with melanics are more easily avoided due to their visual conspicuity. We conclude that differential crypsis between color morphs across the habitat continuum of urban-rural gradients may play an important role in maintaining urban-rural clines in coat color. animal color camouflage citizen science city squirrel evolution Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Urban areas are the most rapidly expanding of any ecosystem on Earth (United Nations, 2018 ). Urbanization often reduces biodiversity through habitat transformations that include increased impervious surfaces, ambient temperature, artificial light, and noise (Grimm et al. 2008 , Rivkin et al. 2019 ). Yet urban areas can also represent new habitat for some species, driving novel evolutionary trajectories that can affect trait variation (Hahs et al., 2023 ), as evidenced by documentation in many species of morphological variation between urban and rural populations (Wandeler et al. 2003 , Weller and Ganzhorn 2004 , Vakhlamova et al. 2014 , Diamond et al. 2017 ). The mechanistic processes causing trait differentiation between urban and rural areas are often unknown. Non-adaptive evolution, driven by genetic drift or gene flow may play a role (Miles et al. 2021 ), whereas fitness differences between urban populations and surrounding, rural populations might also lead to adaptive evolution. Color pattern affects individual fitness in many species by mediating the degree to which individuals are concealed or revealed to predators in different environments (Hultgren and Stachowicz 2008 , Cook et al. 2012 , Zimova et al. 2014 , Troscianko et al. 2016 ). Crypsis is a potential component involved in selection that might be mediated by urbanization via altered visual environments (reviewed by Leveau 2021 ). Replacement of vegetation with impervious cover dramatically reshapes the landscape in cities versus rural areas, which may amplify or weaken selective pressures on conspicuity in species that rely on it for survival. For example, urban grasshoppers show biased movement to pavement that best matches their coloration to avoid predation (Edelaar et al. 2019 ). Conversely, bold coloration in cities may be advantageous for some species, where greater visibility to humans on roads could reduce the likelihood of being struck by motorists (Kreling 2023 ). Here we explore drivers of morphological evolution in eastern gray squirrels ( Sciurus carolinensis ), an arboreal rodent common in treed habitats in urban and rural areas within its native range of eastern North America. The species exhibits two distinct coat color morphs (Fig. 1 ): gray and melanic (black) morphs associated with a 24-bp deletion allele on the melanocortin-1 receptor gene (MC1R; McRobie et al. 2009 ). Urban-rural clines in melanism have been documented in northern parts of the gray squirrel range where melanism is regionally common (Cosentino and Gibbs, 2022 ; Cosentino et al. 2023 ), but the selective factors generating these clines are not known. We examined relative crypsis of gray squirrel morphs in relation to two axes of environmental change associated with urbanization that may most affect survival of these tree-dwelling animals: change in structure and composition of the forest habitats in which squirrels live, and visibility of each morph on road surfaces. Predation is the primary cause of mortality in tree squirrels in rural woodlands (Havera and Nixon 1980 ; Bowers and Breland 1996 ), whereas road-mortality is the primary mortality source in urban areas (McCleery et al. 2008 ). We imaged squirrels of each color morph in representative habitats across the urbanization gradient and compared crypsis between morphs in two ways: 1) employing human observers in an online game to measure morph-specific detection probabilities and 2) using image pixel-classification to measure morph-specific background matching. Methods Study area We examined crypsis of gray squirrels in Syracuse, New York (43.04068° N, − 76.14373° W), a mid-sized city with approximately 150,000 residents (U.S. Census Bureau 2023). The city hosts a well-documented cline in coat color of gray squirrels, with prevalence of melanic forms decreasing from ~ 50% in the city center to < 10% in nearby (< 20 km away) rural forests surrounding the city (Cosentino et al. 2023 ). Experimental design and imaging We used image analysis techniques common in animal coloration studies (Karpestam et al. 2013 , Stevens et al. 2015 , Troscianko et al. 2018 , Edelaar et al. 2019 , Barnett et al. 2020 ) to compare background matching between melanic and gray morphs across the urbanization gradient. To do so we staged taxidermy models of each color morph at replicate locations (hereafter “scenes”) across five habitat types (Table 1 , Supplemental Table S1 ): old growth forest, secondary growth forests, residential areas, urban parks, and urban developed (i.e., core city). This suite of habitats captured both a spatial sequence of land use change encountered along contemporary urban-rural land use gradients (secondary forest to urban core), as well as a historical sequence of land use change that has occurred within the gray squirrel’s native range over the last 300 years (old growth forest to second growth forest). We also staged taxidermy models on road surfaces, which represent a major source of mortality to squirrels from traffic collisions (McCleery et al. 2008 ) and are increasing in extent over both time and space in relation to urbanization (Medley et al. 1995 , Meijer et al. 2018 ). A used 12 taxidermy models for imaging (six of each color morph). Models selected for imaging in each scene were standardized for size between morphs. In each habitat scene, squirrel models were staged in three postures: running on the ground, running on a branch or elevated structure, and climbing (Fig. 2 A). For each posture location, images were taken from two viewpoints to mimic visual perspectives of ground (e.g., humans and canids) and aerial predators (e.g., raptors) (Fig. 2 B-C). Ground-level images were taken at 1.6 m above ground, and aerial images were taken at 6 m above ground. A single squirrel model was included in each image, and each color morph was imaged in identical locations (Table 1 ). For the road environment, images were taken at 1.6 m above ground centered in a road lane. Only squirrels in a running posture were used on roads, oriented perpendicular to the road and randomly placed facing left or right (Fig. 2 D). Table 1 Sampling frame used for examining crypsis of gray and melanic color morphs of the eastern gray squirrel along an urbanization gradient. Images were taken of both morphs in up to three postures (climbing, running on branch, running on ground) and two viewpoints (ground or aerial) during leaf-on (fall) and leaf-off (spring) seasons. In total, 12 images per non-road scene and two images per road scene across two seasons generated 1,760 unique images composed of 880 melanic-gray pairs. Habitat Scenes Postures Viewpoints Total images Old growth forest 20 Climb, branch, ground Aerial, ground 480 Secondary forest 20 Climb, branch, ground Aerial, ground 480 Urban park 10 Climb, branch, ground Aerial, ground 240 Urban yard 10 Climb, branch, ground Aerial, ground 240 Urban developed 10 Climb, branch, ground Aerial, ground 240 Road surface 20 Ground Ground 80 Imaging was undertaken with an iPhone XR camera (Apple Inc., Cupertino, California, United States) using the ProCamera application (Cocologics GmbH, Manheim, Germany) controlled by a remote shutter trigger. Standard ISO was set at 100, and focus, white balance, and shutter speed set to automatic. Each melanic-gray pair was photographed at each posture location within a scene within two minutes of one another to control for lighting conditions. Model position within each image was randomized (left, center, right), as was focal distance to each model (5–20 m in non-road habitats, 10–40 m on roads). Imaging was replicated at each scene during leaf-off and leaf-on seasons using postures and locations identical between seasons (Table 1 ). Quantifying background matching as measured by human observers We used an internet-based gamification approach (Karpestam et al. 2013 , Troscianko et al. 2016 , Troscianko et al. 2018 , Barnett et al. 2020 ) to engage human observers to measure morph-specific detection in each habitat: “Squirrel Spotter” ( https://bcosentino.shinyapps.io/squirrelspotter/ ). The game presented observers with a sequence of 36 images of squirrel models from three randomly chosen habitat types. Before playing, participants were asked if they had played the game before (yes or no). To play the game, each observer was given 15 seconds to detect (by clicking) the squirrel model in each image. For each of the three habitat types, six scenes were randomly chosen, and observers were presented in random order with paired images of both gray and melanic color morphs in identical posture, viewpoint, and season (i.e., 3 habitats x 6 scenes x 2 morphs = 36 images). Images were presented to observers in a blocked fashion by habitat type, such that each sequence of three images included one scene of each habitat type. Images with paired gray-melanic images were never shown sequentially. The first six images were considered a “learning phase," allowing users to habituate to the game (Troscianko et al. 2018 ), and the remaining 30 images (5 gray-melanic pairs from five scenes each of the three randomly selected habitats) considered as “testing phase" and included in the analysis. We used survival analysis to examine how detection of squirrel models was related to color morph and habitat with Cox proportional hazards models fit with the coxme package (Therneau 2024 ) in R 4.2.3 (R Core Team 2023 ). For non-road habitats, we modeled the hazard ratio (i.e., risk of being detected in an image) as a function of color morph, habitat, the morph x habitat interaction, distance of squirrel from the camera, and image order (i.e., order in which an image was presented during the game) with fixed effects. A separate model was fit for the road habitat, which included fixed effects of morph, distance, and image order. Scene and user were included as random intercept terms in both models. We only used the initial game play of users who played multiple times. Images for which squirrel models were not detected within 15 seconds after presentation were right censored. Exploratory analyses revealed a strong carry-over effect such that users tended to find the second of each pair of squirrel morphs faster than the first squirrel encountered within scenes; to address this bias, we limited our analyses to detection data from only the first morph presented in each scene to a user (i.e., a two-sample design with n = 15 scenes for each game play). Tests of significance of fixed effects were made with Wald Chi-square tests using the car package (Fox and Weisberg 2019 ). Separate models were fit for each combination of levels of squirrel posture, viewpoint, and season, resulting in 66 hypothesis tests. We applied a Bonferroni correction to maintain a familywise Type I error rate of 0.05. To do so, we quantified adjusted P -values by multiplying raw P -values by the number of tests (Jafari and Ansari-Pour 2019 ). Quantifying background matching with machine learning We also measured background matching via pixel classification (e.g., Stevens et al. 2015 , Barnett et al. 2020 , Nokelainen et al. 2021 ). We used the Waikato Environment for Knowledge Analysis (Weka Machine Learning) toolkit (Arganda-Carreras et al. 2017 ) as implemented in Fiji ImageJ (Schindelin et al. 2012 ) to create unique FastRandomForest classifiers for each melanic-gray image pair (880 pairs from 1,760 total images, Table 1 ). For each pair, we trained classifiers on an equal proportion of pixels on the dorsal surface of each color morph (Fig. 3 ) based on Red, Green, and Blue (RGB) intensity scores, using the mean and variance of RGB intensity scores in a 1-, 2-, 4-, and 8-pixel radius from each target pixel. The classification model was then used to predict all remaining non-squirrel pixels in the image (i.e., the background of the squirrel) as melanic or gray (Fig. 3 ), and we quantified the proportion of background pixels that matched each morph. Deviations in the proportion of pixels classified as melanic from 0.5 would indicate a better match to melanic (> 0.5) or gray (< 0.5). Because the image background for each melanic-gray pair was virtually identical (see Fig. 3 ), we only applied the classification model to the image with the melanic morph to quantify background matching. Moreover, because crypsis can be a scale-dependent phenomenon (more complex backgrounds can reduce detection; Merilaita, 2003 ), we analyzed background matching for each image at two scales: every pixel in the image (“wide scale”) and within a rectangular region of interest surrounding the squirrel (“local scale”). The immediate background was defined as a region of interest with dimensions double the length and width of the squirrel in each given image (Fig. 3 ) We used generalized linear mixed models to examine variation in the proportion of background pixels classified as melanic. For images in non-road habitats, we specified fixed effects of habitat, posture, perspective, and season. Two-way interaction terms were included between habitat and each of the other fixed effects to test whether variation in background matching among habitats depended on posture, perspective, and season. Random intercept terms were included for scene. We specified a beta distribution for the response variable because it was a proportion bounded between 0 and 1. The beta distribution is not inclusive of the values 0 and 1, so we added a trace value of 0.0001 to five instances of zero values in our dataset (Damgaard and Irvine 2019 ). Separate models were fit for wide and local background scales. For images on roads, we fit separate models with a fixed effect of season, as there was no variation in posture or viewpoint in the road scenes. All models were fit with the glmmTMB package (Brooks et al. 2017 ) in R 4.2.3, with Wald Chi-square tests to examine the significance of each fixed effect. Fitting four models (full and immediate backgrounds for road and non-road habitats) resulted in 16 hypothesis tests, to which we applied a Bonferroni correction to P- values to maintain a familywise Type I error rate of 0.05. Results A total of 1,765 unique participants generated 24,050 observations that were included in our detection analysis from the online game. Detection probability was greater for the melanic morph than the gray morph in virtually all cases (Fig. 4 , Supplemental Table S2). The difference in detection probability between morphs varied among habitats in some postures (morph X habitat interaction effects, Supplemental Table S2). For example, the difference in detection probability between color morphs was weaker in urban than forest habitats when placed in the branch posture (Fig. 4 ). For squirrel models in running position on the ground, detection probability was generally greater in urban than forested habitats for both morphs (Fig. 4 , Supplemental Table S2). Squirrel models were more detectable from the ground than aerial viewpoint in forests, whereas the effect of viewpoint was less consistent and pronounced in urban habitats (Fig. 4 ). Squirrels tended to be more readily detected in ground and climbing positions than in the branch position, particularly in urban habitats (Fig. 4 ). Squirrel detection was similar between seasons, although detection differences between morphs were greatest during the leaf-off season in forested habitats (Fig. 4 ). Scene order had a positive effect on detection in the road habitat (i.e., greater detection probability for images presented later in the game), whereas there was no effect of scene order on detection in non-road habitats (Supplemental Table S2). Squirrel model detection was negatively related to distance from the camera when squirrels were imaged from the ground perspective in climbing and branch postures in non-road habitats (Supplemental Table S2). There were also negative associations between squirrel model detection and distance in the leaf-off season when imaged on roads, as well as the non-road habitats when imaged from the aerial viewpoint in branch posture (Supplemental Table S2). Pixel matching analyses indicated that the gray morph always matched more closely the full and immediate background than did the melanic morph (i.e., < 50% background classified as melanic; Fig. 5 ). The gray morph’s matching advantage at the full background scale tended to be weakest in urban habitats, particularly urban yards, urban developed, and roads (Fig. 5 ). There was little variation in the degree of immediate background matching among habitats (Supplemental Table S3), except for the gray morph’s matching advantage being weakest in urban yards and developed areas when in the branch posture (Fig. 5 ). The gray morph showed greater background matching in the leaf-off season, especially at the full background in urban yard, urban developed areas, and old growth forests (Fig. 5 ). Patterns of background matching were generally consistent between imaging viewpoints (Fig. 5 , Supplemental Table S3). Discussion The melanic morph of the eastern gray squirrel was more conspicuous than the gray morph among all habitats assessed across the urban-rural gradient, a conclusion supported by two measurements of crypsis. First, melanic squirrels had greater probabilities of detection and were detected faster by human observers compared to the gray morph. Second, the melanic morph was a weaker match to its background than the gray morph when using quantitative image analysis. These results highlight a clear effect of color polymorphism on visibility of squirrels across the urbanization gradient, the consequences of which may play an important role in contributing to the maintenance of urban-rural clines in pigmentation. Our results clearly show that the melanic morph was more detectable than the gray morph in forested habitats, including the secondary forests that dominate rural areas and the urban greenspaces where squirrels commonly occur in the city. Visual conspicuity is an important driver of individual predation risk. Previous experimental studies of small mammal species have found melanic morphs are attacked by predators more than other morphs when on mismatched backgrounds (e.g., Hoekstra et al. 2004 , Vignieri et al. 2010 , Linnen et al. 2013 ). Given that predation is a significant source of mortality for tree squirrels in rural woodlands (Bowers and Breland 1996 , McCleery et al. 2008 ), predators may play an important role in causing selection against the melanic morph where secondary forests dominate the landscape. The difference in camouflage between color morphs may also help explain historical trends in the prevalence of melanism in eastern gray squirrels in rural forests. The melanic morph was the prevailing color morph in forests of the northeastern U.S. prior to European settlement (Schorger 1949 , Robertson 1973 ). However, as forests were cleared for agriculture, gray squirrels were viewed as agricultural pests, leading to bounties and a period of intensive hunting (Benson 2013 ). The greater visual conspicuity of the melanic morph to hunters may have contributed to their decline. Curiously, we did not find that the melanic morph was more cryptic in old growth forests where it was once common. Old growth forests have structural features that should provide greater concealment to melanic individuals, including vertical stratification of vegetation that creates deep patches of shade and greater prevalence of coniferous species with dark bark and foliage (Franklin and Van Pelt 2004 , Bauhus et al. 2009 ). Whether our study provides a valid representation of old growth forest conditions is not clear, given that old growth forests remaining today are rare (< 1% of forest cover; Foster et al. 2010 ) and differ in structure and composition from historical old growth stands. Increased deer browsing has caused significant change in the structure of contemporary old growth forests (White 2012 ), and tree species composition has changed dramatically, such as the disappearance of the American chestnut ( Castanea dentata ) (Elliott and Swank 2008 ). Notably, the melanic morph was far more conspicuous on roads than the gray morph. This may create a visual advantage for the melanic morph by providing motorists more time to react and avoid striking them. A previous study with citizen science data found that the melanic morph is underrepresented among roadkill proportional to its frequency in live squirrel populations and suggested that conspicuousness on pavement may give melanic morphs an advantage over gray morphs (Gibbs et al. 2019 ). Roadways can cause intense selection pressures on trait variation in wildlife (Brady and Richardson 2017 ), and high density of roadways associated with urban areas may introduce a novel pressure on visual conspicuity. Roadways are a leading cause of mortality in urban squirrel populations (McCleery 2008), however, when possible, most motorists attempt to avoid striking animals on roads (Beckmann and Shine 2012 ). We found a modest effect of viewpoint on the detection of squirrel models, particularly in secondary and old growth forests. Images of squirrel models from an aerial perspective yielded slower detection times than those from the ground view in forests. This could be the result of human observers being inexperienced at searching for squirrels from an aerial angle. We also found that distance sometimes plays a significant role in detection times, which may also explain the viewpoint effect. Aerial perspectives created greater distances between the camera and squirrel models than ground perspectives, an effect on detection that may be compounded in the more visually complex forested habitats than urban greenspaces. Overall, detection probability was similar between seasons, with a weak seasonal effect detectable in secondary forests (e.g., greater difference in detection between morphs in leaf-off season), where the primarily deciduous tree canopies create stronger visual variation between seasons. It is important to note that while our detection methods are an accurate model for detection by human hunters and motorists, they assume human detection is also representative of detection by avian predators. Previous studies have shown human detection experiments can reliably predict natural detection, including that of birds (Karpestam et al. 2013 ). However, using humans as proxies for detection still has shortcomings because doing so does not account for the cognitive processing associated with color perception (Cuthill et al. 2017 ). The quantitative background matching analysis removes some of this human bias to support our results, but future research should focus on using visual models specific to different types of predators and quantifying actual predation rates between morphs. This study builds upon previous work on gray squirrel color morph frequencies and urban coloration in general. Previous work has shown that the gray morph has a greater survival rate in rural woodlands than the melanic morph, but urban survival rates are comparable between morphs (Cosentino et al. 2023 ). Our study showed the melanic morph is more visible than the gray morph in all habitats. This conspicuousness may contribute to selection against the melanic morph in rural forests, likely from predators and human hunters. In contrast, the greater visual conspicuity of the melanic morph on roads may be to its benefit, particularly in cities where road densities and traffic volume are greatest. It is notable that squirrels in general tended to be more detectable in urban parks and yards than in forests when running on the ground (Fig. 4 ), likely due to the predominance of open turf grass in parks and yards. The melanic morph may be at particularly high predation risk in these open urban habitats, and the conflicting selection pressures of roads and predation in cities may offset, explaining the comparable survival rates between color morphs in the city. Additional studies that quantify morph-specific predation and road-mortality rates are needed to provide insight into the spatial variation of these selection pressures along the urbanization gradient. Overall, this study highlights how visual landscape changes across urban-rural gradients may cause shifts in evolutionary pressures and novel interactions among those pressures, leading to opportunities for evolution of novel animal coloration in cities. Declarations Acknowledgments We thank Cayuga Nature Center, Cornell Botanic Gardens, New York State Department of Environmental Conservation, New York State Office of Parks, Recreation and Historic Preservation, and private landowners for access to sites. We are also grateful for field assistance from Sam Denenberg, Gaby Devo, Aidan Dougherty, Rae Dunstan, Andrew Ferguson, and Paul Raucci. Funding This research was funded by the U.S. National Science Foundation (DEB-2018140, DEB-2018249). Data availability Data and code used for analyses in this manuscript are available at https://github.com/bcosentino/urban-squirrel-crypsis/. References Arganda-Carreras I, Kaynig V, Rueden C, Eliceiri KW, Schindelin J, et al (2017) Trainable Weka Segmentation: A machine learning tool for microscopy pixel classification. Bioinformatics 33(15):2424–2426 Barnett JB, Michalis C, Anderson HM, McEwen BL, Yeager J, et al (2020) Imperfect transparency and camouflage in glass frogs. Proc Natl Acad Sci USA 117(23):12885–12890 Bauhus J, Puettmann K, Messier C (2009) Silviculture for old-growth attributes. For Ecol Manage 258(4):525–537 Beckmann C, Shine R (2012) Do drivers intentionally target wildlife on roads? Austral Ecol 37(5):629–632 Benson E (2013) The urbanization of the eastern gray squirrel in the United States. J Am Hist 100:691–710 Bowers MA, Breland B (1996) Foraging of gray squirrels on an urban-rural gradient: use of the GUD to assess anthropogenic impact. Ecol Appl 6(4):1135–1142 Brady S, Richardson J (2017) Road ecology: shifting gears toward evolutionary perspectives. Front Ecol Environ 15(2):91–98 Brooks ME, Kristensen K, van Benthem KJ, Magnusson A, Berg CW, et al (2017) glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. R J 9(2):378–400 Cuthill IC, Allen WL, Arbuckle K, Caspers B, Chaplin G, et al (2017) The biology of color. Science 357 Cook LM, Grant BS, Saccheri IJ, Mallet J (2012) Selective bird predation on the peppered moth: the last experiment of Michael Majerus. Biol Lett 8(4):609–612 Cosentino BJ, Gibbs JP (2022) Parallel evolution of urban–rural clines in melanism in a widespread mammal. Sci Rep 12:1752 Cosentino BJ, Vanek JP, Gibbs JP (2023) Rural selection drives the evolution of an urban-rural cline in coat color in gray squirrels. Ecol Evol 13:e10544 Damgaard CF, Irvine KM (2019) Using the beta distribution to analyze plant cover data. J Ecol 107:2747–2759 Diamond SE, Chick L, Perez A, Strickler SA, Martin RA (2017) Rapid evolution of ant thermal tolerance across an urban-rural temperature cline. Biol J Linn Soc 121(2):248–257 Edelaar P, Baños-Villalba A, Quevedo DP, Escudero G, Bolnick DI, Jordán-Andrade A (2019) Biased movement drives local cryptic coloration on distinct urban pavements. Proc R Soc B 286(1912):20191343 Elliott KJ, Swank WT (2008) Long-term changes in forest composition and diversity following early logging (1919–1923) and the decline of American chestnut ( Castanea dentata ). Plant Ecol 197(2):155–172 Foster DR, Donahue BM, Kittredge DB, Lambert KF, Hunter ML, et al (2010) Wildlands and Woodlands: A Vision for the New England Landscape. Harvard University Press Fox J, Weisberg S (2019) An R Companion to Applied Regression Third Edition. Sage Publications Franklin JF, Van Pelt R (2004) Spatial aspects of structural complexity in old-growth forests. J For 102(3):22–28 Gibbs JP, Buff MF, Cosentino BJ (2019) The biological system—urban wildlife, adaptation, and evolution: urbanization as a driver of contemporary evolution in gray squirrels (Sciurus carolinensis). In: Hall M, Balogh S (eds) Understanding Urban Ecology. Springer, Cham, pp 269–286 Grimm NB, Faeth SH, Golubiewski NE, Redman CL, Wu J, et al (2008) Global change and the ecology of cities. Science 319(5864):756–760 Hahs AK, Fournier B, Aronson MFJ, Nilon CH, Herrera-Montes A, et al (2023) Urbanisation generates multiple trait syndromes for terrestrial animal taxa worldwide. Nat Commun 14:4751 Havera SP, Nixon CM (1980) Winter feeding of fox and gray squirrel populations. J Wildl Manage 44(1):41–55 Hoekstra HE, Drumm KE, Nachman MW (2004) Ecological genetics of adaptive color polymorphism in pocket mice: geographic variation in selected and neutral genes. Evolution 58:1329–1341 Hultgren KM, Stachowicz JJ (2008) Alternative camouflage strategies mediate predation risk among closely related co-occurring kelp crabs. Oecologia 155(3):519–528 Jafari M, Ansari-Pour N (2019) Why, when, and how to adjust your P values? Cell J 20(4):604–607 Karpestam E, Merilaita S, Forsman A (2013) Detection experiments with humans implicate visual predation as a driver of colour polymorphism dynamics in pygmy grasshoppers. BMC Ecol 13(1):17 Kreling SES (2023) So overt it’s covert: Wildlife coloration in the city. BioScience 73(5):333–346 Leveau L (2021) United colours of the city: A review about urbanisation impact on animal colours. Austral Ecol 46(4):670–679 Linnen CR, Poh YP, Peterson BK, Barrett RDH, Larson JG, et al (2013) Adaptive evolution of multiple traits through multiple mutations at a single gene. Science 339:1312–1316 McCleery RA, Lopez RR, Silvy NJ, Gallant DL (2008) Fox squirrel survival in urban and rural environments. J Wildl Manage 72(1):133–137 McRobie H, Thomas A, Kelly J (2009) The genetic basis of melanism in the gray squirrel ( Sciurus carolinensis ). J Hered 100(6):709–714 Medley KE, McDonnell MJ, Pickett STA (1995) Forest-landscape structure along an urban-to-rural gradient. Prof Geogr 47:159–168 Meijer JR, Huijbregts MA, Schotten KCGJ, Schipper AM (2018) Global patterns of current and future road infrastructure. Environ Res Lett 13:064006 Merilaita S (2003) Visual background complexity facilitates the evolution of camouflage. Evolution 57(6):1248–1254 Miles LS, Carlen EJ, Winchell KM, Johnson MTJ (2021) Urban evolution comes into its own: Emerging themes and future directions of a burgeoning field. Evol Appl 14(1):3–11 Nokelainen O, Scott-Samuel NE, Nie Y, Wei F, Caro T (2021) The giant panda is cryptic. Sci Rep 11:21287 R Core Team (2023) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria Rivkin LR, Santangelo JS, Alberti M, Aronson MFJ, de Keyzer CW, et al (2019) A roadmap for urban evolutionary ecology. Evol Appl 12(3):384–398 Robertson GI (1973) Distribution of color morphs of Sciurus carolinensis in eastern North America. MS Thesis, University of Western Ontario Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, et al (2012) Fiji: an open-source platform for biological-image analysis. Nat Methods 9(7):676–682 Schorger AW (1949) Squirrels in early Wisconsin. Trans Wis Acad Sci Arts Lett 39:195–247 Stevens M, Broderick AC, Godley BJ, Lown AE, Troscianko J, et al (2015) Phenotype–environment matching in sand fleas. Biol Lett 11(8):20150494 Therneau TM (2024) coxme: Mixed effects cox models. R package version 22-20. https://CRAN.R-project.org/package=coxme Troscianko J, Wilson-Aggarwal J, Stevens M, Spottiswoode CN (2016) Camouflage predicts survival in ground-nesting birds. Sci Rep 6:19966 Troscianko J, Skelhorn J, Stevens M (2018) Camouflage strategies interfere differently with observer search images. Proc R Soc B 285(1886):20181386 United Nations (2018) World Urbanization Prospects: The 2018 Revision. https://population.un.org/wup/ US Census Bureau (2023) US Census Bureau QuickFacts: Syracuse City, New York. https://www.census.gov/quickfacts/fact/table/syracusecitynewyork/PST045223 Vakhlamova T, Rusterholz HP, Kanibolotskaya Y, Baur B (2014) Changes in plant diversity along an urban–rural gradient in an expanding city in Kazakhstan, Western Siberia. Landsc Urban Plan 132:111–120 Vignieri SN, Larson JG, Hoekstra HE (2010) The selective advantage of crypsis in mice. Evolution 64(7):2153–2158 Wandeler P, Funk SM, Largiadèr CR, Gloor S, Breitenmoser U (2003) The city-fox phenomenon: Genetic consequences of a recent colonization of urban habitat. Mol Ecol 12(3):647–656 Weller B, Ganzhorn JU (2004) Carabid beetle community composition, body size, and fluctuating asymmetry along an urban-rural gradient. Basic Appl Ecol 5(2):193–201 White MA (2012) Long-term effects of deer browsing: Composition, structure and productivity in a northeastern Minnesota old-growth forest. For Ecol Manage 269:222–228 Zimova M, Mills LS, Lukacs PM, Mitchell S (2014) Snowshoe hares display limited phenotypic plasticity to mismatch in seasonal camouflage. Proc R Soc B 281(1782):20140029 Additional Declarations No competing interests reported. Supplementary Files supplement20241115.pdf Cite Share Download PDF Status: Published Journal Publication published 21 Mar, 2025 Read the published version in Urban Ecosystems → Version 1 posted Editorial decision: Revision requested 24 Jan, 2025 Reviews received at journal 24 Jan, 2025 Reviews received at journal 14 Jan, 2025 Reviews received at journal 04 Jan, 2025 Reviewers agreed at journal 03 Jan, 2025 Reviewers agreed at journal 03 Jan, 2025 Reviewers agreed at journal 03 Jan, 2025 Reviewers invited by journal 03 Jan, 2025 Editor assigned by journal 18 Nov, 2024 Submission checks completed at journal 18 Nov, 2024 First submitted to journal 15 Nov, 2024 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-5462361","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":382627450,"identity":"7819ba8a-6e31-463a-8e58-8ba667f2198f","order_by":0,"name":"Jessica Proctor","email":"","orcid":"","institution":"State University of New York College of Environmental Science and Forestry","correspondingAuthor":false,"prefix":"","firstName":"Jessica","middleName":"","lastName":"Proctor","suffix":""},{"id":382627451,"identity":"72e3d0dc-92f1-4994-b252-2b2cecb63d02","order_by":1,"name":"Alessandra Bryan","email":"","orcid":"","institution":"Hobart and William Smith Colleges","correspondingAuthor":false,"prefix":"","firstName":"Alessandra","middleName":"","lastName":"Bryan","suffix":""},{"id":382627452,"identity":"5942facd-da2c-498e-a075-4d926a956cad","order_by":2,"name":"Bradley J. Cosentino","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAvUlEQVRIiWNgGAWjYPACCQYG9h4wi7GBOB0JQC08Z0jTArIoh0gt/O3tDx8X/rDI45d8e0zyB4ON7IYDBLRInDljbDwjQaJYcnZemjQPQ5oxQS0GEjls0jwJEokbbueY3WZgOJxIhJb057/BWm6eMbv5g+E/MVoSzJjBWm7wmN3gYThAWAvIL9I8aRKJM3tyzH/zGCQbzySkBRRin3ls6hL72c8YG/6osJPtI6QF3Z2kKR8Fo2AUjIJRgAMAAJE+P1RsuNJSAAAAAElFTkSuQmCC","orcid":"","institution":"Hobart and William Smith Colleges","correspondingAuthor":true,"prefix":"","firstName":"Bradley","middleName":"J.","lastName":"Cosentino","suffix":""},{"id":382627453,"identity":"5153f7b9-00b5-4f57-a18a-2a79fd0c7ace","order_by":3,"name":"James P. Gibbs","email":"","orcid":"","institution":"State University of New York College of Environmental Science and Forestry","correspondingAuthor":false,"prefix":"","firstName":"James","middleName":"P.","lastName":"Gibbs","suffix":""}],"badges":[],"createdAt":"2024-11-15 18:08:22","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5462361/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5462361/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s11252-025-01708-4","type":"published","date":"2025-03-21T15:57:15+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":71277825,"identity":"367b325f-6652-4328-8692-e471370219d4","added_by":"auto","created_at":"2024-12-12 22:09:32","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1040133,"visible":true,"origin":"","legend":"\u003cp\u003eGray (left) and melanic (right) color morphs of the eastern gray squirrel (\u003cem\u003eSciurus carolinensis\u003c/em\u003e). Image source: Elizabeth Hunter.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-5462361/v1/e0c2738120108fea80171ca2.png"},{"id":71277826,"identity":"375eb850-d308-4751-8721-c0835c30fc24","added_by":"auto","created_at":"2024-12-12 22:09:32","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":947044,"visible":true,"origin":"","legend":"\u003cp\u003eExample\u003cstrong\u003e \u003c/strong\u003eplacement of taxidermied mounts of eastern gray squirrels within habitat scenes.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-5462361/v1/434c2bc7a18e60febf2e0412.png"},{"id":71277824,"identity":"5bda8838-ecb3-4a43-8605-97837ac24a9d","added_by":"auto","created_at":"2024-12-12 22:09:32","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":633840,"visible":true,"origin":"","legend":"\u003cp\u003eExample set of images used for background matching analysis at wide (A) and local (B) scales. Panels include image of gray morph (left), melanic morph (middle), and the classified output (right). Classifiers were trained to pixels on the torso of each squirrel model to create a binary classification of the entire scene indicating a better match to the gray or melanic morph. Pixels associated with the model were excluded from the analysis.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-5462361/v1/1d69f48888a8995304964653.png"},{"id":71277963,"identity":"a6a31bf5-d284-4679-b864-9ce80d51f0f0","added_by":"auto","created_at":"2024-12-12 22:17:32","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":137973,"visible":true,"origin":"","legend":"\u003cp\u003ePredicted detection probability for melanic and gray color morphs of eastern gray squirrels during leaf-on (A) and leaf-off (B) seasons and from ground (solid line) and aerial (dotted line) perspectives and among habitats (columns) and squirrel model position (rows).\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-5462361/v1/a09cf18b14f2cff61d8e0b06.png"},{"id":71277830,"identity":"d680fc38-f552-4808-868a-078567f4f923","added_by":"auto","created_at":"2024-12-12 22:09:32","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":110769,"visible":true,"origin":"","legend":"\u003cp\u003ePredicted proportion of background pixels classified as matching the melanic morph of eastern gray squirrel during\u003cem\u003e \u003c/em\u003eleaf-on (A) and leaf-off (B) seasons\u003cem\u003e \u003c/em\u003eand\u003cem\u003e \u003c/em\u003efrom ground\u003cem\u003e \u003c/em\u003eand aerial perspectives\u003cem\u003e \u003c/em\u003eamong habitats\u003cem\u003e \u003c/em\u003e(columns) and squirrel model\u003cem\u003e \u003c/em\u003eposition (rows). Error bars represent 95% confidence intervals.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-5462361/v1/bc191953872530a1af6f5db2.png"},{"id":79121178,"identity":"7fc510b6-b1b3-49c1-ac4b-31f9ec99c1bc","added_by":"auto","created_at":"2025-03-24 16:11:40","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4049813,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5462361/v1/364e8832-6af9-4906-af44-200a49033bf3.pdf"},{"id":71277828,"identity":"b4a6b1e0-499c-437f-80f2-06d40721daf7","added_by":"auto","created_at":"2024-12-12 22:09:32","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":201922,"visible":true,"origin":"","legend":"","description":"","filename":"supplement20241115.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5462361/v1/98e6505397b3625f373d7ef0.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Crypsis in a polymorphic mammal along an urbanization gradient","fulltext":[{"header":"Introduction","content":"\u003cp\u003eUrban areas are the most rapidly expanding of any ecosystem on Earth (United Nations, \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Urbanization often reduces biodiversity through habitat transformations that include increased impervious surfaces, ambient temperature, artificial light, and noise (Grimm et al. \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2008\u003c/span\u003e, Rivkin et al. \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Yet urban areas can also represent new habitat for some species, driving novel evolutionary trajectories that can affect trait variation (Hahs et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), as evidenced by documentation in many species of morphological variation between urban and rural populations (Wandeler et al. \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2003\u003c/span\u003e, Weller and Ganzhorn \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2004\u003c/span\u003e, Vakhlamova et al. \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2014\u003c/span\u003e, Diamond et al. \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2017\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe mechanistic processes causing trait differentiation between urban and rural areas are often unknown. Non-adaptive evolution, driven by genetic drift or gene flow may play a role (Miles et al. \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), whereas fitness differences between urban populations and surrounding, rural populations might also lead to adaptive evolution. Color pattern affects individual fitness in many species by mediating the degree to which individuals are concealed or revealed to predators in different environments (Hultgren and Stachowicz \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2008\u003c/span\u003e, Cook et al. \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2012\u003c/span\u003e, Zimova et al. \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2014\u003c/span\u003e, Troscianko et al. \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Crypsis is a potential component involved in selection that might be mediated by urbanization via altered visual environments (reviewed by Leveau \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Replacement of vegetation with impervious cover dramatically reshapes the landscape in cities versus rural areas, which may amplify or weaken selective pressures on conspicuity in species that rely on it for survival. For example, urban grasshoppers show biased movement to pavement that best matches their coloration to avoid predation (Edelaar et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Conversely, bold coloration in cities may be advantageous for some species, where greater visibility to humans on roads could reduce the likelihood of being struck by motorists (Kreling \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eHere we explore drivers of morphological evolution in eastern gray squirrels (\u003cem\u003eSciurus carolinensis\u003c/em\u003e), an arboreal rodent common in treed habitats in urban and rural areas within its native range of eastern North America. The species exhibits two distinct coat color morphs (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e): gray and melanic (black) morphs associated with a 24-bp deletion allele on the melanocortin-1 receptor gene (MC1R; McRobie et al. \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). Urban-rural clines in melanism have been documented in northern parts of the gray squirrel range where melanism is regionally common (Cosentino and Gibbs, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Cosentino et al. \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), but the selective factors generating these clines are not known.\u003c/p\u003e \u003cp\u003eWe examined relative crypsis of gray squirrel morphs in relation to two axes of environmental change associated with urbanization that may most affect survival of these tree-dwelling animals: change in structure and composition of the forest habitats in which squirrels live, and visibility of each morph on road surfaces. Predation is the primary cause of mortality in tree squirrels in rural woodlands (Havera and Nixon \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e1980\u003c/span\u003e; Bowers and Breland \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e1996\u003c/span\u003e), whereas road-mortality is the primary mortality source in urban areas (McCleery et al. \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). We imaged squirrels of each color morph in representative habitats across the urbanization gradient and compared crypsis between morphs in two ways: 1) employing human observers in an online game to measure morph-specific detection probabilities and 2) using image pixel-classification to measure morph-specific background matching.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy area\u003c/h2\u003e \u003cp\u003eWe examined crypsis of gray squirrels in Syracuse, New York (43.04068\u0026deg; N, \u0026minus;\u0026thinsp;76.14373\u0026deg; W), a mid-sized city with approximately 150,000 residents (U.S. Census Bureau 2023). The city hosts a well-documented cline in coat color of gray squirrels, with prevalence of melanic forms decreasing from ~\u0026thinsp;50% in the city center to \u0026lt;\u0026thinsp;10% in nearby (\u0026lt;\u0026thinsp;20 km away) rural forests surrounding the city (Cosentino et al. \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eExperimental design and imaging\u003c/h3\u003e\n\u003cp\u003eWe used image analysis techniques common in animal coloration studies (Karpestam et al. \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2013\u003c/span\u003e, Stevens et al. \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2015\u003c/span\u003e, Troscianko et al. \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2018\u003c/span\u003e, Edelaar et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2019\u003c/span\u003e, Barnett et al. \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) to compare background matching between melanic and gray morphs across the urbanization gradient. To do so we staged taxidermy models of each color morph at replicate locations (hereafter \u0026ldquo;scenes\u0026rdquo;) across five habitat types (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, Supplemental Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e): old growth forest, secondary growth forests, residential areas, urban parks, and urban developed (i.e., core city). This suite of habitats captured both a spatial sequence of land use change encountered along contemporary urban-rural land use gradients (secondary forest to urban core), as well as a historical sequence of land use change that has occurred within the gray squirrel\u0026rsquo;s native range over the last 300 years (old growth forest to second growth forest). We also staged taxidermy models on road surfaces, which represent a major source of mortality to squirrels from traffic collisions (McCleery et al. \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2008\u003c/span\u003e) and are increasing in extent over both time and space in relation to urbanization (Medley et al. \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e1995\u003c/span\u003e, Meijer et al. \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2018\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eA used 12 taxidermy models for imaging (six of each color morph). Models selected for imaging in each scene were standardized for size between morphs. In each habitat scene, squirrel models were staged in three postures: running on the ground, running on a branch or elevated structure, and climbing (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). For each posture location, images were taken from two viewpoints to mimic visual perspectives of ground (e.g., humans and canids) and aerial predators (e.g., raptors) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB-C). Ground-level images were taken at 1.6 m above ground, and aerial images were taken at 6 m above ground. A single squirrel model was included in each image, and each color morph was imaged in identical locations (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). For the road environment, images were taken at 1.6 m above ground centered in a road lane. Only squirrels in a running posture were used on roads, oriented perpendicular to the road and randomly placed facing left or right (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD).\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\u003eSampling frame used for examining crypsis of gray and melanic color morphs of the eastern gray squirrel along an urbanization gradient. Images were taken of both morphs in up to three postures (climbing, running on branch, running on ground) and two viewpoints (ground or aerial) during leaf-on (fall) and leaf-off (spring) seasons. In total, 12 images per non-road scene and two images per road scene across two seasons generated 1,760 unique images composed of 880 melanic-gray pairs.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\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=\"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=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHabitat\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eScenes\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePostures\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eViewpoints\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eTotal images\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOld growth forest\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eClimb, branch, ground\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAerial, ground\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e480\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSecondary forest\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eClimb, branch, ground\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAerial, ground\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e480\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUrban park\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eClimb, branch, ground\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAerial, ground\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e240\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUrban yard\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eClimb, branch, ground\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAerial, ground\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e240\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUrban developed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eClimb, branch, ground\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAerial, ground\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e240\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRoad surface\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGround\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGround\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e80\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\u003eImaging was undertaken with an iPhone XR camera (Apple Inc., Cupertino, California, United States) using the ProCamera application (Cocologics GmbH, Manheim, Germany) controlled by a remote shutter trigger. Standard ISO was set at 100, and focus, white balance, and shutter speed set to automatic. Each melanic-gray pair was photographed at each posture location within a scene within two minutes of one another to control for lighting conditions. Model position within each image was randomized (left, center, right), as was focal distance to each model (5\u0026ndash;20 m in non-road habitats, 10\u0026ndash;40 m on roads). Imaging was replicated at each scene during leaf-off and leaf-on seasons using postures and locations identical between seasons (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eQuantifying background matching as measured by human observers\u003c/h3\u003e\n\u003cp\u003eWe used an internet-based gamification approach (Karpestam et al. \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2013\u003c/span\u003e, Troscianko et al. \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2016\u003c/span\u003e, Troscianko et al. \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2018\u003c/span\u003e, Barnett et al. \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) to engage human observers to measure morph-specific detection in each habitat: \u0026ldquo;Squirrel Spotter\u0026rdquo; (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://bcosentino.shinyapps.io/squirrelspotter/\u003c/span\u003e\u003cspan address=\"https://bcosentino.shinyapps.io/squirrelspotter/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). The game presented observers with a sequence of 36 images of squirrel models from three randomly chosen habitat types. Before playing, participants were asked if they had played the game before (yes or no). To play the game, each observer was given 15 seconds to detect (by clicking) the squirrel model in each image. For each of the three habitat types, six scenes were randomly chosen, and observers were presented in random order with paired images of both gray and melanic color morphs in identical posture, viewpoint, and season (i.e., 3 habitats x 6 scenes x 2 morphs\u0026thinsp;=\u0026thinsp;36 images). Images were presented to observers in a blocked fashion by habitat type, such that each sequence of three images included one scene of each habitat type. Images with paired gray-melanic images were never shown sequentially. The first six images were considered a \u0026ldquo;learning phase,\" allowing users to habituate to the game (Troscianko et al. \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), and the remaining 30 images (5 gray-melanic pairs from five scenes each of the three randomly selected habitats) considered as \u0026ldquo;testing phase\" and included in the analysis.\u003c/p\u003e \u003cp\u003eWe used survival analysis to examine how detection of squirrel models was related to color morph and habitat with Cox proportional hazards models fit with the \u003cem\u003ecoxme\u003c/em\u003e package (Therneau \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) in R 4.2.3 (R Core Team \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). For non-road habitats, we modeled the hazard ratio (i.e., risk of being detected in an image) as a function of color morph, habitat, the morph x habitat interaction, distance of squirrel from the camera, and image order (i.e., order in which an image was presented during the game) with fixed effects. A separate model was fit for the road habitat, which included fixed effects of morph, distance, and image order. Scene and user were included as random intercept terms in both models. We only used the initial game play of users who played multiple times. Images for which squirrel models were not detected within 15 seconds after presentation were right censored. Exploratory analyses revealed a strong carry-over effect such that users tended to find the second of each pair of squirrel morphs faster than the first squirrel encountered within scenes; to address this bias, we limited our analyses to detection data from only the first morph presented in each scene to a user (i.e., a two-sample design with \u003cem\u003en\u003c/em\u003e\u0026thinsp;=\u0026thinsp;15 scenes for each game play).\u003c/p\u003e \u003cp\u003eTests of significance of fixed effects were made with Wald Chi-square tests using the \u003cem\u003ecar\u003c/em\u003e package (Fox and Weisberg \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Separate models were fit for each combination of levels of squirrel posture, viewpoint, and season, resulting in 66 hypothesis tests. We applied a Bonferroni correction to maintain a familywise Type I error rate of 0.05. To do so, we quantified adjusted \u003cem\u003eP\u003c/em\u003e-values by multiplying raw \u003cem\u003eP\u003c/em\u003e-values by the number of tests (Jafari and Ansari-Pour \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e\n\u003ch3\u003eQuantifying background matching with machine learning\u003c/h3\u003e\n\u003cp\u003eWe also measured background matching via pixel classification (e.g., Stevens et al. \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2015\u003c/span\u003e, Barnett et al. \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2020\u003c/span\u003e, Nokelainen et al. \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). We used the Waikato Environment for Knowledge Analysis (Weka Machine Learning) toolkit (Arganda-Carreras et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2017\u003c/span\u003e) as implemented in Fiji ImageJ (Schindelin et al. \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2012\u003c/span\u003e) to create unique FastRandomForest classifiers for each melanic-gray image pair (880 pairs from 1,760 total images, Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). For each pair, we trained classifiers on an equal proportion of pixels on the dorsal surface of each color morph (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) based on Red, Green, and Blue (RGB) intensity scores, using the mean and variance of RGB intensity scores in a 1-, 2-, 4-, and 8-pixel radius from each target pixel. The classification model was then used to predict all remaining non-squirrel pixels in the image (i.e., the background of the squirrel) as melanic or gray (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e), and we quantified the proportion of background pixels that matched each morph. Deviations in the proportion of pixels classified as melanic from 0.5 would indicate a better match to melanic (\u0026gt;\u0026thinsp;0.5) or gray (\u0026lt;\u0026thinsp;0.5). Because the image background for each melanic-gray pair was virtually identical (see Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e), we only applied the classification model to the image with the melanic morph to quantify background matching. Moreover, because crypsis can be a scale-dependent phenomenon (more complex backgrounds can reduce detection; Merilaita, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2003\u003c/span\u003e), we analyzed background matching for each image at two scales: every pixel in the image (\u0026ldquo;wide scale\u0026rdquo;) and within a rectangular region of interest surrounding the squirrel (\u0026ldquo;local scale\u0026rdquo;). The immediate background was defined as a region of interest with dimensions double the length and width of the squirrel in each given image (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e)\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eWe used generalized linear mixed models to examine variation in the proportion of background pixels classified as melanic. For images in non-road habitats, we specified fixed effects of habitat, posture, perspective, and season. Two-way interaction terms were included between habitat and each of the other fixed effects to test whether variation in background matching among habitats depended on posture, perspective, and season. Random intercept terms were included for scene. We specified a beta distribution for the response variable because it was a proportion bounded between 0 and 1. The beta distribution is not inclusive of the values 0 and 1, so we added a trace value of 0.0001 to five instances of zero values in our dataset (Damgaard and Irvine \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Separate models were fit for wide and local background scales. For images on roads, we fit separate models with a fixed effect of season, as there was no variation in posture or viewpoint in the road scenes. All models were fit with the \u003cem\u003eglmmTMB\u003c/em\u003e package (Brooks et al. \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2017\u003c/span\u003e) in R 4.2.3, with Wald Chi-square tests to examine the significance of each fixed effect. Fitting four models (full and immediate backgrounds for road and non-road habitats) resulted in 16 hypothesis tests, to which we applied a Bonferroni correction to \u003cem\u003eP-\u003c/em\u003evalues to maintain a familywise Type I error rate of 0.05.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eA total of 1,765 unique participants generated 24,050 observations that were included in our detection analysis from the online game. Detection probability was greater for the melanic morph than the gray morph in virtually all cases (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, Supplemental Table S2). The difference in detection probability between morphs varied among habitats in some postures (morph X habitat interaction effects, Supplemental Table S2). For example, the difference in detection probability between color morphs was weaker in urban than forest habitats when placed in the branch posture (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). For squirrel models in running position on the ground, detection probability was generally greater in urban than forested habitats for both morphs (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, Supplemental Table S2). Squirrel models were more detectable from the ground than aerial viewpoint in forests, whereas the effect of viewpoint was less consistent and pronounced in urban habitats (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). Squirrels tended to be more readily detected in ground and climbing positions than in the branch position, particularly in urban habitats (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). Squirrel detection was similar between seasons, although detection differences between morphs were greatest during the leaf-off season in forested habitats (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eScene order had a positive effect on detection in the road habitat (i.e., greater detection probability for images presented later in the game), whereas there was no effect of scene order on detection in non-road habitats (Supplemental Table S2). Squirrel model detection was negatively related to distance from the camera when squirrels were imaged from the ground perspective in climbing and branch postures in non-road habitats (Supplemental Table S2). There were also negative associations between squirrel model detection and distance in the leaf-off season when imaged on roads, as well as the non-road habitats when imaged from the aerial viewpoint in branch posture (Supplemental Table S2).\u003c/p\u003e \u003cp\u003ePixel matching analyses indicated that the gray morph always matched more closely the full and immediate background than did the melanic morph (i.e., \u0026lt;\u0026thinsp;50% background classified as melanic; Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). The gray morph\u0026rsquo;s matching advantage at the full background scale tended to be weakest in urban habitats, particularly urban yards, urban developed, and roads (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). There was little variation in the degree of immediate background matching among habitats (Supplemental Table S3), except for the gray morph\u0026rsquo;s matching advantage being weakest in urban yards and developed areas when in the branch posture (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). The gray morph showed greater background matching in the leaf-off season, especially at the full background in urban yard, urban developed areas, and old growth forests (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). Patterns of background matching were generally consistent between imaging viewpoints (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e, Supplemental Table S3).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe melanic morph of the eastern gray squirrel was more conspicuous than the gray morph among all habitats assessed across the urban-rural gradient, a conclusion supported by two measurements of crypsis. First, melanic squirrels had greater probabilities of detection and were detected faster by human observers compared to the gray morph. Second, the melanic morph was a weaker match to its background than the gray morph when using quantitative image analysis. These results highlight a clear effect of color polymorphism on visibility of squirrels across the urbanization gradient, the consequences of which may play an important role in contributing to the maintenance of urban-rural clines in pigmentation.\u003c/p\u003e \u003cp\u003eOur results clearly show that the melanic morph was more detectable than the gray morph in forested habitats, including the secondary forests that dominate rural areas and the urban greenspaces where squirrels commonly occur in the city. Visual conspicuity is an important driver of individual predation risk. Previous experimental studies of small mammal species have found melanic morphs are attacked by predators more than other morphs when on mismatched backgrounds (e.g., Hoekstra et al. \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2004\u003c/span\u003e, Vignieri et al. \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2010\u003c/span\u003e, Linnen et al. \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Given that predation is a significant source of mortality for tree squirrels in rural woodlands (Bowers and Breland \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e1996\u003c/span\u003e, McCleery et al. \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2008\u003c/span\u003e), predators may play an important role in causing selection against the melanic morph where secondary forests dominate the landscape.\u003c/p\u003e \u003cp\u003eThe difference in camouflage between color morphs may also help explain historical trends in the prevalence of melanism in eastern gray squirrels in rural forests. The melanic morph was the prevailing color morph in forests of the northeastern U.S. prior to European settlement (Schorger \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e1949\u003c/span\u003e, Robertson \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e1973\u003c/span\u003e). However, as forests were cleared for agriculture, gray squirrels were viewed as agricultural pests, leading to bounties and a period of intensive hunting (Benson \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). The greater visual conspicuity of the melanic morph to hunters may have contributed to their decline. Curiously, we did not find that the melanic morph was more cryptic in old growth forests where it was once common. Old growth forests have structural features that should provide greater concealment to melanic individuals, including vertical stratification of vegetation that creates deep patches of shade and greater prevalence of coniferous species with dark bark and foliage (Franklin and Van Pelt \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2004\u003c/span\u003e, Bauhus et al. \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). Whether our study provides a valid representation of old growth forest conditions is not clear, given that old growth forests remaining today are rare (\u0026lt;\u0026thinsp;1% of forest cover; Foster et al. \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2010\u003c/span\u003e) and differ in structure and composition from historical old growth stands. Increased deer browsing has caused significant change in the structure of contemporary old growth forests (White \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2012\u003c/span\u003e), and tree species composition has changed dramatically, such as the disappearance of the American chestnut (\u003cem\u003eCastanea dentata\u003c/em\u003e) (Elliott and Swank \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2008\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eNotably, the melanic morph was far more conspicuous on roads than the gray morph. This may create a visual advantage for the melanic morph by providing motorists more time to react and avoid striking them. A previous study with citizen science data found that the melanic morph is underrepresented among roadkill proportional to its frequency in live squirrel populations and suggested that conspicuousness on pavement may give melanic morphs an advantage over gray morphs (Gibbs et al. \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Roadways can cause intense selection pressures on trait variation in wildlife (Brady and Richardson \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2017\u003c/span\u003e), and high density of roadways associated with urban areas may introduce a novel pressure on visual conspicuity. Roadways are a leading cause of mortality in urban squirrel populations (McCleery 2008), however, when possible, most motorists attempt to avoid striking animals on roads (Beckmann and Shine \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2012\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eWe found a modest effect of viewpoint on the detection of squirrel models, particularly in secondary and old growth forests. Images of squirrel models from an aerial perspective yielded slower detection times than those from the ground view in forests. This could be the result of human observers being inexperienced at searching for squirrels from an aerial angle. We also found that distance sometimes plays a significant role in detection times, which may also explain the viewpoint effect. Aerial perspectives created greater distances between the camera and squirrel models than ground perspectives, an effect on detection that may be compounded in the more visually complex forested habitats than urban greenspaces. Overall, detection probability was similar between seasons, with a weak seasonal effect detectable in secondary forests (e.g., greater difference in detection between morphs in leaf-off season), where the primarily deciduous tree canopies create stronger visual variation between seasons.\u003c/p\u003e \u003cp\u003eIt is important to note that while our detection methods are an accurate model for detection by human hunters and motorists, they assume human detection is also representative of detection by avian predators. Previous studies have shown human detection experiments can reliably predict natural detection, including that of birds (Karpestam et al. \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). However, using humans as proxies for detection still has shortcomings because doing so does not account for the cognitive processing associated with color perception (Cuthill et al. \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). The quantitative background matching analysis removes some of this human bias to support our results, but future research should focus on using visual models specific to different types of predators and quantifying actual predation rates between morphs.\u003c/p\u003e \u003cp\u003eThis study builds upon previous work on gray squirrel color morph frequencies and urban coloration in general. Previous work has shown that the gray morph has a greater survival rate in rural woodlands than the melanic morph, but urban survival rates are comparable between morphs (Cosentino et al. \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Our study showed the melanic morph is more visible than the gray morph in all habitats. This conspicuousness may contribute to selection against the melanic morph in rural forests, likely from predators and human hunters. In contrast, the greater visual conspicuity of the melanic morph on roads may be to its benefit, particularly in cities where road densities and traffic volume are greatest. It is notable that squirrels in general tended to be more detectable in urban parks and yards than in forests when running on the ground (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e), likely due to the predominance of open turf grass in parks and yards. The melanic morph may be at particularly high predation risk in these open urban habitats, and the conflicting selection pressures of roads and predation in cities may offset, explaining the comparable survival rates between color morphs in the city. Additional studies that quantify morph-specific predation and road-mortality rates are needed to provide insight into the spatial variation of these selection pressures along the urbanization gradient. Overall, this study highlights how visual landscape changes across urban-rural gradients may cause shifts in evolutionary pressures and novel interactions among those pressures, leading to opportunities for evolution of novel animal coloration in cities.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgments\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cbr\u003e\u003c/strong\u003eWe thank Cayuga Nature Center, Cornell Botanic Gardens, New York State Department of Environmental Conservation, New York State Office of Parks, Recreation and Historic Preservation, and private landowners for access to sites. We are also grateful for field assistance from Sam Denenberg, Gaby Devo, Aidan Dougherty, Rae Dunstan, Andrew Ferguson, and Paul Raucci.\u003c/p\u003e\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research was funded by the U.S. National Science Foundation (DEB-2018140, DEB-2018249). \u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData and code used for analyses in this manuscript are available at https://github.com/bcosentino/urban-squirrel-crypsis/.\u003c/p\u003e"},{"header":"References","content":"\u003cp\u003eArganda-Carreras I, Kaynig V, Rueden C, Eliceiri KW, Schindelin J, et al (2017) Trainable Weka Segmentation: A machine learning tool for microscopy pixel classification. Bioinformatics 33(15):2424\u0026ndash;2426\u003c/p\u003e\n\u003cp\u003eBarnett JB, Michalis C, Anderson HM, McEwen BL, Yeager J, et al (2020) Imperfect transparency and camouflage in glass frogs. Proc Natl Acad Sci USA 117(23):12885\u0026ndash;12890\u003c/p\u003e\n\u003cp\u003eBauhus J, Puettmann K, Messier C (2009) Silviculture for old-growth attributes. For Ecol Manage 258(4):525\u0026ndash;537\u003c/p\u003e\n\u003cp\u003eBeckmann C, Shine R (2012) Do drivers intentionally target wildlife on roads? Austral Ecol 37(5):629\u0026ndash;632\u003c/p\u003e\n\u003cp\u003eBenson E (2013) The urbanization of the eastern gray squirrel in the United States. J Am Hist 100:691\u0026ndash;710\u003c/p\u003e\n\u003cp\u003eBowers MA, Breland B (1996) Foraging of gray squirrels on an urban-rural gradient: use of the GUD to assess anthropogenic impact. Ecol Appl 6(4):1135\u0026ndash;1142\u003c/p\u003e\n\u003cp\u003eBrady S, Richardson J (2017) Road ecology: shifting gears toward evolutionary perspectives. Front Ecol Environ 15(2):91\u0026ndash;98\u003c/p\u003e\n\u003cp\u003eBrooks ME, Kristensen K, van Benthem KJ, Magnusson A, Berg CW, et al (2017) glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. R J 9(2):378\u0026ndash;400\u003c/p\u003e\n\u003cp\u003eCuthill IC, Allen WL, Arbuckle K, Caspers B, Chaplin G, et al (2017) The biology of color. Science 357\u003c/p\u003e\n\u003cp\u003eCook LM, Grant BS, Saccheri IJ, Mallet J (2012) Selective bird predation on the peppered moth: the last experiment of Michael Majerus. Biol Lett 8(4):609\u0026ndash;612\u003c/p\u003e\n\u003cp\u003eCosentino BJ, Gibbs JP (2022) Parallel evolution of urban\u0026ndash;rural clines in melanism in a widespread mammal. Sci Rep 12:1752\u003c/p\u003e\n\u003cp\u003eCosentino BJ, Vanek JP, Gibbs JP (2023) Rural selection drives the evolution of an urban-rural cline in coat color in gray squirrels. Ecol Evol 13:e10544\u003c/p\u003e\n\u003cp\u003eDamgaard CF, Irvine KM (2019) Using the beta distribution to analyze plant cover data. J Ecol 107:2747\u0026ndash;2759\u003c/p\u003e\n\u003cp\u003eDiamond SE, Chick L, Perez A, Strickler SA, Martin RA (2017) Rapid evolution of ant thermal tolerance across an urban-rural temperature cline. Biol J Linn Soc 121(2):248\u0026ndash;257\u003c/p\u003e\n\u003cp\u003eEdelaar P, Ba\u0026ntilde;os-Villalba A, Quevedo DP, Escudero G, Bolnick DI, Jord\u0026aacute;n-Andrade A (2019) Biased movement drives local cryptic coloration on distinct urban pavements. Proc R Soc B 286(1912):20191343\u003c/p\u003e\n\u003cp\u003eElliott KJ, Swank WT (2008) Long-term changes in forest composition and diversity following early logging (1919\u0026ndash;1923) and the decline of American chestnut (\u003cem\u003eCastanea dentata\u003c/em\u003e). Plant Ecol 197(2):155\u0026ndash;172\u003c/p\u003e\n\u003cp\u003eFoster DR, Donahue BM, Kittredge DB, Lambert KF, Hunter ML, et al (2010) Wildlands and Woodlands: A Vision for the New England Landscape. Harvard University Press\u003c/p\u003e\n\u003cp\u003eFox J, Weisberg S (2019) An R Companion to Applied Regression Third Edition. Sage Publications\u003c/p\u003e\n\u003cp\u003eFranklin JF, Van Pelt R (2004) Spatial aspects of structural complexity in old-growth forests. J For 102(3):22\u0026ndash;28\u003c/p\u003e\n\u003cp\u003eGibbs JP, Buff MF, Cosentino BJ (2019) The biological system\u0026mdash;urban wildlife, adaptation, and evolution: urbanization as a driver of contemporary evolution in gray squirrels (Sciurus carolinensis). In: Hall M, Balogh S (eds) Understanding Urban Ecology. Springer, Cham, pp 269\u0026ndash;286\u003c/p\u003e\n\u003cp\u003eGrimm NB, Faeth SH, Golubiewski NE, Redman CL, Wu J, et al (2008) Global change and the ecology of cities. Science 319(5864):756\u0026ndash;760\u003c/p\u003e\n\u003cp\u003eHahs AK, Fournier B, Aronson MFJ, Nilon CH, Herrera-Montes A, et al (2023) Urbanisation generates multiple trait syndromes for terrestrial animal taxa worldwide. Nat Commun 14:4751\u003c/p\u003e\n\u003cp\u003eHavera SP, Nixon CM (1980) Winter feeding of fox and gray squirrel populations. J Wildl Manage 44(1):41\u0026ndash;55\u003c/p\u003e\n\u003cp\u003eHoekstra HE, Drumm KE, Nachman MW (2004) Ecological genetics of adaptive color polymorphism in pocket mice: geographic variation in selected and neutral genes. Evolution 58:1329\u0026ndash;1341\u003c/p\u003e\n\u003cp\u003eHultgren KM, Stachowicz JJ (2008) Alternative camouflage strategies mediate predation risk among closely related co-occurring kelp crabs. Oecologia 155(3):519\u0026ndash;528\u003c/p\u003e\n\u003cp\u003eJafari M, Ansari-Pour N (2019) Why, when, and how to adjust your P values? Cell J 20(4):604\u0026ndash;607\u003c/p\u003e\n\u003cp\u003eKarpestam E, Merilaita S, Forsman A (2013) Detection experiments with humans implicate visual predation as a driver of colour polymorphism dynamics in pygmy grasshoppers. BMC Ecol 13(1):17\u003c/p\u003e\n\u003cp\u003eKreling SES (2023) So overt it\u0026rsquo;s covert: Wildlife coloration in the city. BioScience 73(5):333\u0026ndash;346\u003c/p\u003e\n\u003cp\u003eLeveau L (2021) United colours of the city: A review about urbanisation impact on animal colours. Austral Ecol 46(4):670\u0026ndash;679\u003c/p\u003e\n\u003cp\u003eLinnen CR, Poh YP, Peterson BK, Barrett RDH, Larson JG, et al (2013) Adaptive evolution of multiple traits through multiple mutations at a single gene. Science 339:1312\u0026ndash;1316\u003c/p\u003e\n\u003cp\u003eMcCleery RA, Lopez RR, Silvy NJ, Gallant DL (2008) Fox squirrel survival in urban and rural environments. J Wildl Manage 72(1):133\u0026ndash;137\u003c/p\u003e\n\u003cp\u003eMcRobie H, Thomas A, Kelly J (2009) The genetic basis of melanism in the gray squirrel (\u003cem\u003eSciurus carolinensis\u003c/em\u003e). J Hered 100(6):709\u0026ndash;714\u003c/p\u003e\n\u003cp\u003eMedley KE, McDonnell MJ, Pickett STA (1995) Forest-landscape structure along an urban-to-rural gradient. Prof Geogr 47:159\u0026ndash;168\u003c/p\u003e\n\u003cp\u003eMeijer JR, Huijbregts MA, Schotten KCGJ, Schipper AM (2018) Global patterns of current and future road infrastructure. Environ Res Lett 13:064006\u003c/p\u003e\n\u003cp\u003eMerilaita S (2003) Visual background complexity facilitates the evolution of camouflage. Evolution 57(6):1248\u0026ndash;1254\u003c/p\u003e\n\u003cp\u003eMiles LS, Carlen EJ, Winchell KM, Johnson MTJ (2021) Urban evolution comes into its own: Emerging themes and future directions of a burgeoning field. Evol Appl 14(1):3\u0026ndash;11\u003c/p\u003e\n\u003cp\u003eNokelainen O, Scott-Samuel NE, Nie Y, Wei F, Caro T (2021) The giant panda is cryptic. Sci Rep 11:21287\u003c/p\u003e\n\u003cp\u003eR Core Team (2023) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria\u003c/p\u003e\n\u003cp\u003eRivkin LR, Santangelo JS, Alberti M, Aronson MFJ, de Keyzer CW, et al (2019) A roadmap for urban evolutionary ecology. Evol Appl 12(3):384\u0026ndash;398\u003c/p\u003e\n\u003cp\u003eRobertson GI (1973) Distribution of color morphs of \u003cem\u003eSciurus carolinensis\u003c/em\u003e in eastern North America. MS Thesis, University of Western Ontario\u003c/p\u003e\n\u003cp\u003eSchindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, et al (2012) Fiji: an open-source platform for biological-image analysis. Nat Methods 9(7):676\u0026ndash;682\u003c/p\u003e\n\u003cp\u003eSchorger AW (1949) Squirrels in early Wisconsin. Trans Wis Acad Sci Arts Lett 39:195\u0026ndash;247\u003c/p\u003e\n\u003cp\u003eStevens M, Broderick AC, Godley BJ, Lown AE, Troscianko J, et al (2015) Phenotype\u0026ndash;environment matching in sand fleas. Biol Lett 11(8):20150494\u003c/p\u003e\n\u003cp\u003eTherneau TM (2024) coxme: Mixed effects cox models. R package version 22-20. https://CRAN.R-project.org/package=coxme\u003c/p\u003e\n\u003cp\u003eTroscianko J, Wilson-Aggarwal J, Stevens M, Spottiswoode CN (2016) Camouflage predicts survival in ground-nesting birds. Sci Rep 6:19966\u003c/p\u003e\n\u003cp\u003eTroscianko J, Skelhorn J, Stevens M (2018) Camouflage strategies interfere differently with observer search images. Proc R Soc B 285(1886):20181386\u003c/p\u003e\n\u003cp\u003eUnited Nations (2018) World Urbanization Prospects: The 2018 Revision. https://population.un.org/wup/\u003c/p\u003e\n\u003cp\u003eUS Census Bureau (2023) US Census Bureau QuickFacts: Syracuse City, New York. https://www.census.gov/quickfacts/fact/table/syracusecitynewyork/PST045223\u003c/p\u003e\n\u003cp\u003eVakhlamova T, Rusterholz HP, Kanibolotskaya Y, Baur B (2014) Changes in plant diversity along an urban\u0026ndash;rural gradient in an expanding city in Kazakhstan, Western Siberia. Landsc Urban Plan 132:111\u0026ndash;120\u003c/p\u003e\n\u003cp\u003eVignieri SN, Larson JG, Hoekstra HE (2010) The selective advantage of crypsis in mice. Evolution 64(7):2153\u0026ndash;2158\u003c/p\u003e\n\u003cp\u003eWandeler P, Funk SM, Largiad\u0026egrave;r CR, Gloor S, Breitenmoser U (2003) The city-fox phenomenon: Genetic consequences of a recent colonization of urban habitat. Mol Ecol 12(3):647\u0026ndash;656\u003c/p\u003e\n\u003cp\u003eWeller B, Ganzhorn JU (2004) Carabid beetle community composition, body size, and fluctuating asymmetry along an urban-rural gradient. Basic Appl Ecol 5(2):193\u0026ndash;201\u003c/p\u003e\n\u003cp\u003eWhite MA (2012) Long-term effects of deer browsing: Composition, structure and productivity in a northeastern Minnesota old-growth forest. For Ecol Manage 269:222\u0026ndash;228\u003c/p\u003e\n\u003cp\u003eZimova M, Mills LS, Lukacs PM, Mitchell S (2014) Snowshoe hares display limited phenotypic plasticity to mismatch in seasonal camouflage. Proc R Soc B 281(1782):20140029\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"urban-ecosystems","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"ueco","sideBox":"Learn more about [Urban Ecosystems](https://www.springer.com/journal/11252)","snPcode":"11252","submissionUrl":"https://submission.nature.com/new-submission/11252/3","title":"Urban Ecosystems","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"animal color, camouflage, citizen science, city, squirrel, evolution","lastPublishedDoi":"10.21203/rs.3.rs-5462361/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5462361/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eUrbanization transforms landscapes and alters visual environments for polymorphic species that rely on cryptic coloration for survival, potentially generating urban-rural clines in pigmentation. Such clines are evident in eastern gray squirrel (\u003cem\u003eSciurus carolinensis\u003c/em\u003e) populations, for which a melanic morph is currently more prevalent in cities but was historically more prevalent in rural woodlands prior to urbanization. We compared the degree of crypsis between the two primary color morphs of gray squirrels \u0026ndash; gray and melanic \u0026ndash; among the suite of habitats that predominate along an urbanization gradient to test whether an altered visual environment may contribute to the maintenance of an urban-rural cline in morph prevalence. Crypsis was quantified using an online game with human observers and image pixel classification to measure detectability of taxidermic mounts of each morph against their backgrounds in replicate sites within each habitat. The melanic morph was more conspicuous than the gray morph in all habitat types and across seasons, as evidenced by greater detection probabilities by human observers and lower background matching. Coat color in gray squirrels likely mediates visual detection by predators, potentially resulting in selection against the more conspicuous, melanic morph in rural woodlands. Conversely, selection via road mortality may favor the melanic morph in urban areas if vehicular collisions with melanics are more easily avoided due to their visual conspicuity. We conclude that differential crypsis between color morphs across the habitat continuum of urban-rural gradients may play an important role in maintaining urban-rural clines in coat color.\u003c/p\u003e","manuscriptTitle":"Crypsis in a polymorphic mammal along an urbanization gradient","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-12-12 22:09:27","doi":"10.21203/rs.3.rs-5462361/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-01-25T03:56:06+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-01-24T05:29:23+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-01-14T21:41:47+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-01-04T08:26:42+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"316969052537223493982306740473475392756","date":"2025-01-03T23:10:25+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"251276415271548223152462180780157485380","date":"2025-01-03T21:01:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"125182548362247884271096643590796099965","date":"2025-01-03T18:31:58+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-01-03T16:42:16+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-11-19T03:36:20+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-11-18T08:18:01+00:00","index":"","fulltext":""},{"type":"submitted","content":"Urban Ecosystems","date":"2024-11-15T18:01:46+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"urban-ecosystems","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"ueco","sideBox":"Learn more about [Urban Ecosystems](https://www.springer.com/journal/11252)","snPcode":"11252","submissionUrl":"https://submission.nature.com/new-submission/11252/3","title":"Urban Ecosystems","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"c7b6e31a-8009-46db-b498-4998a6ed6a8b","owner":[],"postedDate":"December 12th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-03-24T16:10:01+00:00","versionOfRecord":{"articleIdentity":"rs-5462361","link":"https://doi.org/10.1007/s11252-025-01708-4","journal":{"identity":"urban-ecosystems","isVorOnly":false,"title":"Urban Ecosystems"},"publishedOn":"2025-03-21 15:57:15","publishedOnDateReadable":"March 21st, 2025"},"versionCreatedAt":"2024-12-12 22:09:27","video":"","vorDoi":"10.1007/s11252-025-01708-4","vorDoiUrl":"https://doi.org/10.1007/s11252-025-01708-4","workflowStages":[]},"version":"v1","identity":"rs-5462361","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5462361","identity":"rs-5462361","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}