Urban environments are often associated with resource and environmental differences, providing potential novel selection pressures compared to adjacent unmodified landscapes. While these characteristics (e.g., heat islands, reduced vegetation) can contribute to differences in certain behaviors, morphology, or physiological traits, there is mixed evidence on how and to what extent populations are responding. In this study, we compared the feather morphology of Dark-eyed Junco (Junco hyemalis) populations established across an urbanization gradient. We examined whether differential temperature regimes, related to urbanization, correspond with significant variations in the proportion of down. We sampled ventral and dorsal feathers from 256 individuals throughout central and southern California at varying degrees of urbanization. Dorsal feathers had a higher proportion of down compared to ventral feathers, but did not differ between populations. Urbanization did not significantly correlate with feather morphology. Ventral feathers had a greater proportion of down as the range of temperature increased, but this correlation was marginal. Our results show that despite urbanization altering fine-scale habitat conditions, these did not correspond with rapid feather morphological variations. Whether this is the case for other feather types or across species is still unknown, but it would provide insight into the complex effects of urbanization on wildlife biology.

We captured 256 Dark-eyed Juncos across central and southern California during the 2022 (n = 47) and 2023 (n = 209) breeding season (February–August) between 07:00 and 12:00. We mist netted at six locations on average per day, and at each location, the net was open for 30 min during which playbacks were used to attract juncos in the area. Playbacks consisted of singing juncos from Los Angeles. We color-banded each junco with a unique three-color band and one USGS aluminum band combination and recorded various metrics as part of a long-term study. Metrics included bill dimensions (width, depth, and length), body weight, wing chord length, tail length, and tarsus length. We quantified sex differences based on the presence of brood patch or cloacal protuberance, as well as plumage characteristics. The juncos were aged as after second year (ASY), second year (SY), after hatch year (AHY), or hatch year (HY) based on known plumage and molt limit. Data from HYs and recaptured juncos were excluded from this study.

We collected 2–6 contour feathers from male and female adult juncos in definitive basic plumage. To sample feathers, banders removed 1–3 feathers from the anterior region of the mantle (dorsal) and 1–3 feathers from the anterior region of the breast (ventral) of each bird. Feathers were stored dry in envelopes until processing. Our sites consist of the University of California, Los Angeles (UCLA, n = 121), the University of California, San Diego (UCSD, n = 22), the University of California, Santa Barbara (UCSB, n = 18), Occidental College (Occ, n = 22), San Francisco State University (SFSU, n = 6), Santa Monica Parks (SM Parks, n = 9), Los Angeles Parks (LA Parks, n = 10), as well as nearby mountain regions in the Santa Monica Mountains (SMM, n = 9), and the Angeles National Forest (ANF, n = 35; Fig. 1). The minimum distance between our sampling sites was 4700 m (between UCLA and SM Parks). This distance is further than the expected maximum distance of 174 m that a junco will fly during the breeding season, reducing the possibility of the same individual moving between sites. Coordinates of each banding site were recorded and used to obtain corresponding environmental and climatic conditions.

All methods were carried out by the relevant ARRIVE methods required for observational animal research and by institutional guidelines and regulations. Animal handling in this study adhered to protocols approved by the Institutional Animal Care and Use Committee (IACUC) of UCLA (ARC-2018-007-AM 004). Banding efforts were conducted in compliance with the Ethics and Responsibilities of Bird Banders published by the US Geological Survey Federal Bird Banding Laboratory (Permit #23809) and as outlined by the State of California Department of Fish and Wildlife Scientific Collecting Permit—Specific Use (S-191300002-20288-001-02) for taking/possession of wildlife for scientific purposes.

We calculated the proportion of the feather surface area that consists of plumulaceous barbs (henceforth “proportion of down”) and then averaged these values for each bird. To determine the proportion of down, we scanned each feather using a CanoScan LiDE 400 Color Image Scanner at a resolution of 600 dots per inch (dpi). When scanning the images, we placed the feathers on a transparency sheet labeled with a 10 mm scale and weighed each feather down with a 0.13–0.19 mm-thick coverslip. A colored sheet of cardstock laid on top maximized the contrast between the desired region and the background.

To measure the relative proportion of down, we developed color thresholds using a hue-saturation-brightness color space in ImageJ for the ventral and dorsal feathers in order to increase repeatability and reduce potential observer bias. For dorsal feathers, a yellow (#ebd798) background and a white (#e8e7e5) background provided the best contrast to measure the pennaceous region and total surface area, respectively (Fig. 2a, b). To analyze ventral feathers, we measured the proportion of down and total surface area against a red (#c71934) background (Fig. 2c). We then selected and measured the area above the set color threshold. To calculate the area of down in dorsal feathers, we subtracted the area of pennaceous barbs from the total area. We then calculated the proportion of down for dorsal and ventral feathers by dividing the area of down by the total surface area.

Using a subset of feathers, our values correlated with previously published metrics used to measure the proportion of down and pennaceous feathers based on a Kendall’s rank correlation tau, where the color difference measures of surface area were strongly correlated with the length of down/pennaceous regions along the shaft of the feathers (τ = 0.388, p < 0.001, n = 100)[23343545]. We selected our metric to reduce potential observer bias when quantifying feather barbs. We assessed repeatability of our measurements with feathers from museum specimens to reduce stress on living birds. Repeatability of our measurements was included to account for sampling bias. Museum specimens provided by the Occidental College Moore Laboratory of Zoology were all collected within five years of our study from Los Angeles (Specimen numbers: MLZ: Bird:70290; MLZ: Bird:70291; MLZ: Bird:70222; MLZ: Bird:70246; MLZ: Bird:70221; MLZ: Bird:70292). We randomly sampled a different number of feathers from each specimen (n = 5, 6, 8, 9, 12, 15) and calculated the proportion of down using the methods previously described. We then ran a linear model for the variance of the proportion of down of dorsal feathers against sample size, defined as the number of feathers sampled from a given specimen. We repeated this model for ventral feathers.

Temperature data from the PRISM Climate Group was available using the prism 0.2.0 package in R[48]. Our climatic variables include mean temperature (Tmean: μ = 17.3 ºC, SD = 1.1, range = 13.4 to 18.9, CV = 6.28%) and the temperature range, based on the difference between the maximum and minimum temperatures (ΔTemp: μ = 9.9 ºC, SD = 1.7, range = 6.1 to 13.6, CV = 16.90%; Fig. 1). For each variable, we extracted averages over 30 years (1991–2020) at a resolution of 800 m using the terra 1.7, sp 2.0, and sf 1.0 packages. Selected values capture the broadest conditions but may not accurately represent the strongest selective pressures. Temperature extremes, for instance, could be more impactful, but were not available for our study. Measurements of urbanization are based on the Built-Up index (BU) calculated using raster images from Collection 2 Level 2 Landsat 8 data available from the U.S. Geological Survey Earth Resources Observation and Science (EROS) Center Science Processing Architecture (ESPA). These raster images are at a 30 m resolution within one month of the sampling date, where cloud cover was less than 8%. We used QGIS[52] to calculate the BU index.

At each sampling coordinate, we then averaged the BU index within a 50 m radius (μ = −0.236, SD = 0.095, range = −0.531 to −0.041, CV = 40.33%), which represents the average territory size of Dark-eyed Juncos recorded by previous literature as well as our observations. This method of averaging the BU index is most accurately able to capture the degree of urbanization of our study sites. Detailed information on the BU, Tmean, and ΔTemp for different sites is provided (Supplemental Table S1).