Urbanization and elevated cholesterol in American Crows

Abstract

Although urban areas can be sources of abundant food for wildlife, anthropogenic foods may be lower in quality than natural food sources, with possible consequences for birds. We examined how urbanization and anthropogenic food were linked to cholesterol levels, condition, and survival of American Crows (Corvus brachyrhynchos). We collected cholesterol and landscape data from 140 crow nestlings along an urban-to-rural gradient in Davis, California, USA. We also ran a supplementation experiment with high-cholesterol fast food (McDonald’s cheeseburgers) on 86 nestlings in a rural population in Clinton, New York, USA. Plasma cholesterol increased with percentage of impervious surface along the urban-to-rural gradient. Cholesterol levels were sensitive to anthropogenic foods: crows supplemented with fast food cheeseburgers had higher cholesterol levels than unsupplemented crows. Elevated cholesterol levels had no detectable effects on survival and were associated with higher indices of body condition, although urbanization itself was linked to lower survival. Elevated cholesterol levels could indicate access to high-calorie, high-fat anthropogenic foods, which might, in some contexts, improve body condition, potentially offsetting other negative effects of urbanization. Observations over a longer time scale, assessing additional indices of health and fitness, are needed to evaluate long-term costs or benefits of elevated cholesterol for urban crows.

Resumen

Aunque las áreas urbanas pueden ser una fuente abundante de alimentos para la vida silvestre, la comida antropogénica puede ser de menor calidad que las fuentes naturales de alimento, con posibles consecuencias para las aves. Examinamos cómo la urbanización y la comida antropogénica estuvieron vinculados con los niveles de colesterol, la condición y la supervivencia de Corvus brachyrhynchos. Colectamos datos de colesterol y de paisaje de 140 polluelos a lo largo de un gradiente urbano-rural en Davis, California, EEUU. También realizamos un experimento de suplementación con comida rápida con alto contenido de colesterol (hamburguesas con queso de McDonald’s) en 86 polluelos en una población rural en Clinton, Nueva York, EEUU. El colesterol del plasma aumentó con el porcentaje de superficie impermeable a lo largo del gradiente urbano-rural. Los niveles de colesterol fueron sensibles a los alimentos antropogénicos: los individuos suplementados con hamburguesas con queso presentaron niveles de colesterol más altos que los individuos no suplementados. Los niveles elevados de colesterol no tuvieron efectos detectables en la supervivencia y estuvieron asociados con índices más altos de condición corporal, aunque la urbanización en sí misma se vinculó con una supervivencia más baja. Los niveles elevados de colesterol podrían indicar un acceso a comida antropogénica con alto contenido calórico y de grasas, lo que podría, en algunos contextos, mejorar la condición corporal, compensando potencialmente otros efectos negativos de la urbanización. Se necesitan observaciones sobre escalas temporales más largas, evaluando índices adicionales de salud y de adecuación biológica, para evaluar los costos o beneficios a largo plazo del colesterol elevado para los cuervos urbanos.

INTRODUCTION

The ability to exploit anthropogenic food resources—provided intentionally and unintentionally—is a characteristic common to many species of wildlife in urban settings (McKinney 2002). Urban humans produce approximately twice as much waste as their rural counterparts (Hoornweg and Bhada-Tata 2012), and ~21% of municipal solid waste is food (U.S. Environmental Protection Agency 2014). Some of this food waste may become available to urban wildlife that scavenge from litter, compost piles, and trash. In some cases, anthropogenic food sources might be nutritionally deficient, particularly during early developmental stages (Isaksson and Andersson 2007, Chamberlain et al. 2009, Heiss et al. 2009). For these species, cities might act as ecological traps, in which the abundant yet (potentially) low-quality food resources attract animals but are unable to sustain a population (Battin 2004). In other cases, however, anthropogenic subsidies may improve the condition of wild animals (Cypher and Frost 1999), increase reproductive output (Schoech et al. 2008), and buffer adults in harsh winter months (Chamberlain et al. 2009). The relative costs and benefits of anthropogenic food subsidies on wildlife health are still poorly understood, and are considered both a major conservation concern (Wilcoxen et al. 2015, Murray et al. 2016) and a global research opportunity (Jones and Reynolds 2008).

Human diet is changing in ways that may affect the health of urban scavengers. Meat consumption, for example, is positively correlated with degree of urbanization, and the availability of fast food has increased in the human diet on a global scale (Drewnowski and Popkin 1997, Jahren and Kraft 2008, Bonhommeau et al. 2013). This alteration is reflected in the diet of urban wildlife. Urban ants, for example, have isotopic signatures associated with processed foods, and the magnitude of these signatures increases with human density (Penick et al. 2015). Similarly, several studies have documented an increase in cholesterol levels of wild animals in association with human activity. House Sparrows (Passer domesticus) and endangered San Joaquin kit foxes (Vulpes macrotis mutica) from urban populations have higher cholesterol levels than conspecifics from nonurban populations (Gavett and Wakeley 1986, Cypher and Frost 1999), as do endangered male Northern Bahamian rock iguanas (Cyclura cyclura; Knapp et al. 2013) and green turtles (Chelonia mydas; Monzon-Arguello et al. 2018) in touristed areas. An adequate supply of cholesterol and its precursors is critical to physiological function, playing essential roles in cellular membrane permeability, bile production, synthesis of steroid hormones (e.g., cortisol, progesterone, testosterone), and calcium metabolism (Tabas 2002). In humans and laboratory animals, however, excessive cholesterol is linked to coronary and cardiovascular disease (Yuan et al. 1997, Stamler et al. 2000), although it has also been linked to declines in diseases associated with nutrient deficiency (Drewnowski and Popkin 1997). The consequences of elevated cholesterol—positive or negative—on condition and survival in wild animal populations have not been evaluated.

The objectives of this study were (1) to determine if plasma cholesterol level increased with urbanization in American Crows (Corvus brachyrhynchos); (2) to assess the whether plasma cholesterol level predicted body condition or survival, either positively or negatively; and (3) to experimentally test the likelihood that elevation in cholesterol levels was driven by access to anthropogenic food. We examined the links between plasma cholesterol and urbanization of American Crows along an urban-to-rural gradient in Davis, California, USA, using percentage of impervious surface surrounding each nest site as an index of urbanization (Townsend and Barker 2014). We also examined the sensitivity of nestling cholesterol levels to anthropogenic food in a supplementation experiment with high-cholesterol fast food in a rural crow population near Clinton, New York, USA. We used a mass by size residual as an index of body condition (Townsend et al. 2010), and examined survival in a mark–recapture framework based on 2–3 yr of weekly monitoring data in the California population. We hypothesized that cholesterol level would be positively correlated with urbanization and body condition because plasma cholesterol may indicate access to reliable, high-calorie human food sources. However, we also hypothesized that nestling cholesterol levels would be negatively correlated with survival, if these foods are lower in quality (i.e. lacking other key micronutrients) than natural foods (Heiss et al. 2009).

METHODS

Study Sites

We sampled crows in Davis, California, from 2012 to 2015 and Clinton, New York, from 2016 to 2018. The California population spanned the urban campus of the University of California at Davis to the surrounding agricultural areas (Figure 1A). The New York population encompassed the relatively rural Hamilton College campus and adjacent residential areas (Figure 1B). These crows lived in family groups, including a pair of socially monogamous breeders and up to 5 auxiliaries of either sex (Townsend et al. 2018b), many of which helped to provision offspring.

FIGURE 1.

Study sites in (A) Davis, California, and (B) Clinton, New York. Histograms indicate the distribution of the amount of impervious surface within 10 ha buffers around nests.

To quantify degree of urbanization, we created 10 ha buffers surrounding each nest site (Figure 1), and estimated the percentage of land cover types within each buffer using the 2011 National Land Cover Database (Homer et al. 2015). Following Townsend and Barker (2014), we used the percentage of impervious surface as our index of urbanization. Percentage of impervious surface within these buffers strongly correlated with percentage of medium-intensity and high-intensity development, another index of urbanization (r^{2 =}0.85, P < 0.0001). Mean percent coverages of impervious surface in territories were mean ± SE = 34.5 ± 2.4% (range: 1.2–67.3%; Figure 1A) and 4.2 ± 0.8% (range: 0–15.6%; Figure 1B) in the California and New York sites, respectively. Spatial analyses, including the sp and raster packages, were run in R 3.0.1 (R Core Team 2016). Means ± SE are given throughout.

Cholesterol, Landscape, Condition, and Survival

In California we collected data from 140 nestlings from 66 crow nests (2012: 60 nestlings, 28 nests; 2013: 80 nestlings, 38 nests). In New York we collected data from 86 nestlings from 29 nests (2016: 34 nestlings, 13 nests; 2017: 36 nestlings, 10 nests; 2018: 16 nestlings, 6 nests). Following Townsend et al. (2018a, 2018b, 2018c), nestlings were banded 22–35 days after hatching with both a numbered USGS band and a unique color band. Nestling age (±3 days) was estimated from approximate hatch date. We calculated an index of body condition for each nestling as the residual from a regression of mass against size + (size*size), defining nestling size as the first principal component on covariances of exposed culmen, skull, bill width and depth, and tarsus, following Townsend et al. (2010). The first principal component explained 97.5% of the variation in these measurements. Blood samples (~300 µL) were drawn at time of banding from the brachial or jugular vein using 27-gauge ½-inch needles and 1 mL syringes and stored in heparinized tubes. We extracted DNA from blood samples using DNeasy tissue kits (Qiagen, Valencia, California, USA) and sexed nestlings with the P2/P8 sexing test (Griffiths et al. 1998). Plasma cholesterol levels were evaluated using a Roche Cobas c501 analyzer (Roche Diagnostics, Indianapolis, Indiana, USA) at the Veterinary Medical Teaching Hospital Clinical Diagnostic Laboratories at the University of California, Davis (California samples), or the Cornell Veterinary Diagnostics Laboratory (New York samples).

Nestlings were returned to their nests immediately after sampling. Banded birds were monitored for survival 3–7 days per week until May 2015 in the California population (up to 24 or 36 mo for birds hatched in 2012 and 2013, respectively) along established census routes (Hinton et al. 2015, Taff and Townsend 2017; Figure 1A). The effects of urbanization on cholesterol level were examined in a linear mixed effects model (lme in R package NLME; Pinheiro et al. 2019) with percent coverage of impervious surfaces, nestling age (days), and sex as fixed effects. The effects of cholesterol on body condition were examined in a linear mixed effects model with cholesterol, impervious surface, age, and sex as fixed effects. The effect of cholesterol on fledging success (0/1) was examined in a generalized linear mixed model (binomial distribution; glmmPQL in R package MASS; Venables and Ripley 2002) with cholesterol, impervious surface, age, and sex as fixed effects. Family group was included as a random effect in all models to account for nonindependence among nestmates. Age and sex had nonsignificant effects and were removed from final models.

Supplementation Experiment

To evaluate the effect of anthropogenic foods on plasma cholesterol, we randomly selected nests from 10 family groups in the Clinton, New York, population (8 in 2016 and 2 in 2017) for supplementation. Each group was provided 3 cheeseburgers from a McDonald’s restaurant (42 mg cholesterol per burger; 126 mg per baiting session; U.S. Department of Agriculture 2018), placed within 10 m of the nest tree, 5–6 days per week. Groups were supplemented for 1–6 weeks prior to banding and cholesterol sampling, depending on the date at which the nest was first discovered.

The effects of supplementation on nestling cholesterol were examined in linear mixed effects models with cholesterol as the response, “supplementation” (yes/no), number of supplementation days, or number of burgers provided as fixed effects, and family group as a random effect. In the final model, we used “supplementation” as the predictor because it explained more variation in the data than number of supplementation days or number of burgers. We also specified nestling age, group size, and brood size as covariates, but they had nonsignificant effects and were removed from the final model. Variation in mean cholesterol level across populations with urbanization and supplementation was further evaluated in a linear mixed effects model with urbanization as a categorical fixed effect (rural = 0–16% impervious surface, corresponding to the range of impervious surfaces in the Clinton, New York, territories; intermediate = 16–40% impervious surface; and urban = 40–67%, the upper boundary corresponding to the maximum percentage of impervious surfaces in the Davis, California, territories), and family group as a random effect. Population had a nonsignificant effect and was removed from the final model.

Mark–Recapture Analysis

Apparent survival (φ) and recapture (p) parameters were estimated in MARK 8.1. Model fit was verified by dividing the deviance estimate by the mean of simulated deviances from a parametric goodness-of-fit test. C-hat was adjusted to 1.35. Specifying [φ(t)p(t)] as the underlying model, we generated models to evaluate the effects of cholesterol (“chol”) and urbanization (“urb”; percent coverage of impervious surfaces) on φ, starting with [φ(t+chol+urb)p(t)] as the global model. The model set was balanced with respect to factors of interest [e.g., φ (chol), φ (urb)]. Models with lower Akaike information criterion (AIC) scores, corrected for sample size (AIC_c), were considered more parsimonious (Akaike 1974, Lebreton et al. 1992).

RESULTS

Urban-to-Rural Gradient

Plasma cholesterol increased with urbanization (Figure 2A). Along the urban-to-rural gradient in the Davis, California, population, nestling cholesterol levels increased with percentage of impervious surface within a 10 ha area surrounding the nest site (β [10% increase in impervious surface] ± SE = 6.0 ± 2.2 mg dL⁻¹ cholesterol; t₈₃ = 2.8; P = 0.006; lme with cholesterol as the response, impervious surface as a fixed effect, and family group as a random effect). There was no effect of either percentage of impervious surface or cholesterol on nestling body condition or fledging success (P > 0.05; Tables 1, 2).

FIGURE 2.

Associations between impervious surface in Davis, California, and (A) plasma cholesterol level or (B) survival. Impervious surface is shown categorically here for illustration but treated as a continuous variable in the mark–recapture analysis.

TABLE 1.

Fixed effects from a linear mixed model predicting the body condition of American Crow nestlings in Davis, California.^a

Response variable	Explanatory variable	ß ± SE	t80	P
Condition	Cholesterol	0.01 ± 0.08	0.1	0.91
	Impervious surface (10 ha)	−0.17 ± 0.20	−0.8	0.42
	Age (days)	−0.22 ± 0.56	−0.4	0.70
	Sex (M)	2.16 ± 5.0	0.4	0.67

^a Family specified as random effect. N = 139 nestlings; 55 family groups.

TABLE 2.

Fixed effects from a generalized linear mixed model examining factors affecting the fledging success (0/1; binomial distribution) of American Crow nestlings in Davis, California.^a

Response variable	Explanatory variable	ß ± SE	t80	P
Fledging success	Cholesterol	−0.01 ± 0.01	−1.6	0.12
	Impervious surface (10 ha)	−0.02 ± 0.02	−0.7	0.46
	Age (days)	0.04 ± 0.05	0.9	0.40
	Sex (M)	−0.22 ± 0.43	−0.5	0.61

^a Family specified as random effect. N = 139 nestlings; 55 family groups.

Urbanization was negatively associated with long-term survival in the California population (Figure 2B). Apparent survival in the first 24–36 mo of life declined with increasing impervious surface: mark–recapture models that included impervious surface as a predictor had the most support (Table 3; cumulative model support: 0.81). In contrast, cumulative model support for an effect of cholesterol was relatively weak (0.33); it did not substantially improve model fit (∆AIC_c < 2) over a model with impervious surface alone.

TABLE 3.

Candidate set of approximating models generated to fit American Crow mark–recapture data along an urban-to-rural gradient in Davis, California (w_i = AIC_c weight; k = number of parameters; φ = survival; imp = percentage of impervious surface; chol = plasma cholesterol level; p = recapture; t = time).

Candidate models	AICc	∆AICc	wi	k	Deviance
φ (imp + t) p(t)	0.49	0	0.49	25	651.0
φ (chol + imp + t) p(t)	0.32	0.86	0.32	26	649.7
φ (chol + t) p(t)	0.12	2.78	0.12	25	653.8
φ (t) p(t)	0.08	3.71	0.08	24	656.9
φ (chol + imp * t) p(t)	0	50.65	0	42	663.0

Supplementation Experiment

Adult crows from all supplemented groups immediately descended to and removed food from under their nest trees after the researchers moved away from bait. Opportunistic observations indicated that some adults brought pieces of burger immediately and directly to the nestlings, whereas others cached or ate them. We did not quantify the amount of burger actually fed to the nestlings in each group. Nevertheless, supplementation increased nestling cholesterol levels. Mean nestling cholesterol levels were 142.1 ± 2.1 mg dL⁻¹ plasma in unsupplemented groups (n = 64 nestlings from 19 nests) and 149 ± 5.5 mg dL⁻¹ plasma in supplemented groups (n = 22 supplemented nestlings from 10 nests). Nestling cholesterol levels in supplemented groups were, on average, higher (β [supplemented] ± SE = 11.3 ± 5.6 mg dL⁻¹ plasma; t₆₆ = 2.0; P = 0.04; unequal variances specified between treatments) and more variable (O’Brien test for unequal variances; P = 0.005) than levels in unsupplemented groups. Cholesterol level was associated with higher indices of nestling body condition (β [cholesterol] ± SE = 0.4 ± 0.2; t₆₆ = 2.3; P = 0.03; lme with condition as the response, cholesterol as a fixed effect, and family group as a random effect); supplementation and impervious surface were not significant predictors of condition (P > 0.05) and were removed from final models.

Population Comparison

Mean cholesterol levels in the rural territories were similar across populations (i.e. territories with 0–16% impervious surface; 142.1 ± 2.1 mg dL⁻¹ plasma and 144.8 ± 5.3 mg dL⁻¹ plasma in the [unsupplemented] Clinton and Davis rural territories, respectively). In contrast, mean nestling cholesterol levels in the most urban crow territories in the Davis, California, population (i.e. comprising 40–67% impervious surface) were significantly higher than rural territories (171.5 ± 4.4 mg dL⁻¹ plasma in the urban territories; lme with urbanization as a categorical fixed effect and family group as a random effect). Nestlings in both the supplemented rural territories in Clinton, New York, and the territories with an intermediate level of development (i.e. comprising 16–40% impervious surface) in the Davis, California, population had intermediate cholesterol levels (Figure 3).

FIGURE 3.

Cholesterol levels of crow nestlings with population (Clinton, New York, or Davis, California), urbanization (categorized here as rural, intermediate, or urban), and supplementation. Means and standard errors shown. Levels that are not connected by the same letter were significantly different from one another in a linear mixed model with urbanization as a categorical fixed effect and family group as a random effect (Tukey’s HSD, α < 0.05).

DISCUSSION

Anthropogenic food can have positive and negative effects on wildlife. It can buffer animals during food shortages, reduce foraging time, advance breeding, and improve condition (Kaneko and Maruyama 2005, Jones and Reynolds 2008, Shanahan et al. 2014, Wilcoxen et al. 2015). Studies in another cooperative corvid, the Florida Scrub-Jay (Aphelocoma coerulescens), for example, suggest that anthropogenic food supplementation leads to earlier breeding, increased reproductive output, and decreased annual variation in reproductive success, and can (in some cases) increase total body lipids (Schoech and Bowman 2003, Schoech et al. 2008, Schoech 2009). Other studies, however, indicate that supplementary food can decrease reproductive output and increase pathogen prevalence and stress (Harrison et al. 2010, Wilcoxen et al. 2015, Murray et al. 2016, 2018,). Anthropogenic food can alter dietary quality (Murray et al. 2018) in ways that affect blood chemistry (e.g., plasma protein, calcium, and cholesterol; Schoech and Bowman 2003, Ishigame et al. 2006, Heiss et al. 2009), although the extent to which these changes appear beneficial or detrimental vary among studies.

In this study, we found that plasma cholesterol levels in crow nestlings increased with degree of urbanization, and experimental supplementation indicated that mean cholesterol level increased with access to anthropogenic food. Plasma cholesterol levels among the supplemented nestlings were also more variable than among control nestlings (Figure 3), which might have been due to variation in the extent to which adults fed the burgers to the nestlings or ate the burgers themselves. However, we could not test this hypothesis because we did not quantify the amount of burger eaten by the adults vs. fed to nestlings in each group.

Cholesterol level did not have detectable survival consequences for these birds, and elevated cholesterol levels were associated with higher indices of nestling body condition in the New York (although not in the California) population. This result is consistent with work showing that urban San Joaquin kit foxes had higher cholesterol levels and fewer signs of nutritional stress than their exurban counterparts (Cypher and Frost 1999). Indices of body condition in birds can be difficult to interpret because what constitutes “good condition” can vary with developmental stage (Schoech 2009). However, previous work has indicated that nestling crows in “good condition” along the index used in this study (mass:size residuals) do appear to have long-term survival advantages: crow nestlings with higher condition indices in the first month after hatching had a lower disease probability in the first 3 yr of life than nestlings with lower condition indices (Townsend et al. 2010). Elevated cholesterol levels could be linked to better body condition among crow nestlings if they indicate access to reliable, high-calorie human food sources, which might, in some contexts, offset other negative effects of urbanization (Suri et al. 2017).

Adequate cholesterol is essential to physiological function, but excessive cholesterol is linked to a suite of diseases, including atherosclerotic vascular disease and liver dysfunction (Tabas 2002). Cholesterol levels that would be “excessive” in crows, however, are unknown. The range of values that we observed were similar to those reported for American Crows (n = 67) by the Zoological Information Management System (ZIMS; http://zims.Species360.org): ZIMS: mean = 166.02 mg dL⁻¹, reference interval: 88.8–243.2 mg dL⁻¹; Clinton, New York: mean ± SE = 143.9 ± 2.1 dL⁻¹, range: 108–211 mg dL⁻¹; Davis, California: mean = 160.7 ± 3.1 dL⁻¹, range: 84–266 mg dL⁻¹. Such comparisons should be made with caution, however, as the ZIMS data were collected from putatively healthy adult zoo crows that were unlikely to be consuming a natural diet, whereas our data were collected from wild-caught nestling crows, some of which harbored one or more infections (Wheeler et al. 2014, Taff et al. 2016, Townsend et al. 2018a). A laboratory study of Japanese Quail (Coturnix japonica) indicated that birds fed a high-cholesterol diet had cholesterol levels of 1,423.0–1,678.3 mg dL⁻¹ and were significantly more likely to develop aortic plaques than the control quail (control quail cholesterol levels: 206.1–299.7 mg dL⁻¹; Yuan et al. 1997). These “high cholesterol” levels are approximately an order of magnitude larger than the values observed in our crows, and do not clarify the threshold level at which cholesterol could become harmful in crows or other wild birds.

Although cholesterol levels did not have detectable effects on crow survival, urbanization itself did: urbanization was associated with lower apparent survival over the first 3 yr of life along the urban-to-rural gradient in the California population. Likewise, a suburban–rural comparison of American Crows in another population (Ithaca, New York) reported that nestlings were smaller and number of fledglings per successful nest lower in a suburban site relative to a rural site, although post-fledging survival was higher in the suburban site (McGowan 2001). A complex array of factors is likely to play a role in these survival differences between urban and rural crows. Predator and disease pressures, for example, can vary with urbanization, although the direction of these differences can vary geographically as well as with the predator or pathogen in question (McGowan 2001, Delgado and French 2012). Urbanization is also likely to reduce hunting pressure and have variable effects on the frequency of other anthropogenic sources of mortality (e.g., car collisions, entanglement in plastic, electrocution; Townsend et al. 2009, Townsend and Barker 2014). Diet could also play a role if urban food sources are deficient in key micronutrients, which may be particularly important in early development (Chamberlain et al. 2009, Heiss et al. 2009, Meillere et al. 2017, Murray et al. 2018). For example, American Crow nestlings in the Ithaca, New York, population were smaller and had lower blood protein and calcium levels in suburban areas than in rural areas, which the authors attributed to a nutrient-limited (but not a calorie-limited) diet on suburban territories (Heiss et al. 2009). Food quality on territory might be less of an issue in the nonbreeding season, when crows can foray off-territory at communal foraging sites (Taff et al. 2016; A. K. Townsend personal observation). The costs or benefits of abundant, reliable anthropogenic food on crow survival is therefore likely to vary with developmental stage and the availability and relative quality of natural foods.

We found no evidence that elevated cholesterol levels were costly for crows in our populations; indeed, elevated cholesterol was associated with better body condition in the New York (although not the California) population. We note, however, that we only monitored nestlings for 2–3 yr after hatching. Elevated cholesterol might have costs that we did not detect, as the negative effects of excessive cholesterol levels can take years to manifest (Tabas 2002). Moreover, some authors have proposed that anthropogenic food could increase the prevalence of degenerative disease in wildlife populations (Giraudeau et al. 2018). Although intriguing, we found no evidence consistent with this idea: we have seen no signs of degenerative diseases, such as cancer and atherosclerosis, in the course of extensive histopathology work in 2 crow populations (Davis, California, and Ithaca, New York; Miller et al. 2010, Wheeler et al. 2014). Observations over a longer time scale that assess additional indices of health and fitness are needed to fully evaluate long-term costs or benefits of elevated cholesterol on urban crows.

ACKNOWLEDGMENTS

Thanks to C. Taff and S. Wheeler for valuable field assistance and to C. Downs for access to the ZIMS cholesterol data for American Crows. We have no conflicts of interest for this manuscript.

Ethics statement: This project was carried out under protocols approved by the Institutional Animal Care and Use Committees of UC Davis and Hamilton College.

Author contributions: A.K.T. wrote the manuscript. C.M.B. carried out the spatial analysis. A.K.T. and H.S. collected and analyzed the data.

Data accessibility: Urbanization, cholesterol, and fitness data: Dryad Digital Repository (doi:10.5061/dryad.t7r7899).