Research Paper

INTRODUCTION

Mapping the breeding distribution of declining, habitat-specialist songbirds is essential for accurately assessing their conservation status, and discovery of previously unknown breeding sites can improve understanding of vegetation associations, population size, and extirpation risk. The “sagebrush” subspecies of the Brewer’s Sparrow (Spizella breweri breweri) is a migratory songbird that is widely considered a sagebrush-obligate because it breeds in greatest abundance in sagebrush (Artemisia spp.) shrublands across the contiguous western U.S. and southwestern Canada (Rotenberry et al. 2020). Trend estimates from the North American Breeding Bird Survey for breweri in sagebrush indicate a range-wide population decline of -1.01% per year from 1966-2015 (Sauer et al. 2017). In Colorado, breweri has been identified as a Tier 2 species of greatest conservation need in the State Wildlife Action Plan (Colorado Parks and Wildlife 2015) due to statewide population declines (-2.06%/yr) (Boyle and Reeder 2005) and historical and ongoing threats to their sagebrush breeding habitat. In contrast, the “timberline” subspecies (S. b. taverneri), or “Timberline Sparrow”, typically breeds in shrubs and conifer krummholz (i.e., stunted, wind-deformed trees) at or above tree line in mountain ranges from east-central Alaska to northwestern Montana (Doyle 1997, Griffin et al. 2003, Rotenberry et al. 2020). Breeding taverneri were first found in northern British Columbia (Swarth and Brooks 1925), but later discoveries expanded the known breeding range of the subspecies south to southeastern Alberta and northwestern Montana (McTaggart-Cowan 1946, Semenchuk 1992, Griffin et al. 2003) and northeast to east-central Alaska (Doyle 1997).

Several authors have speculated that taverneri may also breed in mountain ranges farther south in the contiguous western U.S. based on summer records of singing male Brewer’s Sparrows in shrubs or krummholz at or above tree line in California, Colorado, Idaho, Nevada, Oregon, Utah, and Wyoming (e.g., Lambeth 1998, Griffin et al. 2003, Hansley and Beauvais 2004). Colorado, in particular, has had numerous reports of Brewer’s Sparrows in willow and conifer krummholz at alpine sites (i.e., those at or above tree line) in summer, but the taxonomic affiliation of such birds in Colorado and other western states remains unknown (Righter et al. 2004, Leukering 2008, Spencer 2014).

We identified several possible explanations for the summer occurrence of singing males in alpine areas (Table 1), each with different implications for Brewer’s Sparrow breeding ecology and conservation status. First, alpine birds could be previously undocumented southern breeding populations of taverneri (Lambeth 1998, Righter et al. 2004). If so, this would extend the known breeding distribution of taverneri >1300 km south. Second, alpine birds could simply be “sagebrush” breweri breeding in an atypical habitat type (Andrews and Righter 1992, Righter et al. 2004). This subspecies occasionally nests in shrub communities other than sagebrush and at elevations up to ~3000 m (Righter et al. 2004, Walker et al. 2020, Zillig et al. 2023). Confirming breweri breeding in alpine areas would expand our understanding of the breeding biology and vegetation associations of this subspecies and potentially expand estimates of breweri population size. Under the second hypothesis, alpine populations of breweri might or might not be genetically distinct from those in sagebrush. If alpine breweri are distinct, it would suggest that sagebrush and alpine populations do not interbreed. If they are indistinguishable, it could suggest that sagebrush and alpine populations interbreed, that drift or selection has been insufficient for sagebrush and alpine populations to have diverged, or that alpine birds are itinerant breeders (i.e., breweri that first nest in sagebrush, then move upslope to renest in alpine areas; Spencer 2014). Third, alpine areas could represent a previously unknown contact zone between breweri and taverneri. The two subspecies are currently considered allopatric during the breeding season (Klicka et al. 2001, Mayr and Johnson 2001), but taverneri breeding range may extend farther south (Griffin et al. 2003). Fourth, alpine birds could be a third, previously undocumented subspecies. Existing genetic data only support the existence of two, rather than three taxa (Klicka et al. 1999), but Brewer’s Sparrows in alpine areas south of taverneri breeding range have never been sampled. Finally, Brewer’s Sparrows in alpine areas could also be non-breeding breweri or taverneri. However, an observation of recent fledglings being fed by adults at one alpine site in Colorado in early August (Lambeth 1998) makes non-breeding explanations less likely. Determining the taxonomic affiliation and breeding status of alpine birds would allow us to distinguish among most of these hypotheses.

Determining their taxonomic affiliation may also help resolve ongoing taxonomic debate. The Timberline Sparrow was originally proposed as a new species, Spizella taverneri (Swarth and Brooks 1925), but it has always been considered a subspecies (AOU 1931, Chesser et al. 2022). The two taxa are estimated to have diverged within the past 35,000-80,000 years, so they are closely related (Klicka et al. 1999). Some authors argue that taverneri should be treated as a full species based on differences in genetics, morphology, song, vegetation associations, and timing of breeding (Klicka et al. 1999, Klicka et al. 2001, Mayr and Johnson 2001). If alpine birds in Colorado are breeding taverneri, this would raise the possibility that taverneri diverged from breweri via parapatric or peripatric speciation, then colonized alpine areas farther north, rather than diverging in allopatry at the northernmost end of breweri breeding range (Klicka et al. 1999). In contrast, if alpine birds are breeding breweri, it would counter arguments that differences in breeding habitat and timing of breeding necessarily support full species status for taverneri (Klicka et al. 1999, 2001).

The objectives of this study were to determine the taxonomic affiliation, breeding status, habitat characteristics, and summer distribution of Brewer’s Sparrows in alpine areas of mountain ranges in western Colorado. To do this, we first compiled and reviewed historical summer records of Brewer’s Sparrows to identify potential alpine breeding sites. We then surveyed a broadly distributed subset of alpine sites and lower-elevation sagebrush sites nearby. At each site, we surveyed for, counted, and recorded singing males, documented habitat features, looked for evidence of breeding, and captured birds to obtain morphological data, photographs, and blood and feather samples. Finally, we compared external morphology, acoustic structure of songs, plumage and maxilla color, and mitochondrial and nuclear DNA sequences between alpine birds and sagebrush breweri and against published data for each subspecies (Klicka et al. 1999, Walker 2024) to determine the taxonomic affiliation of alpine birds.

METHODS

Historical records

We compiled potential breeding-season (April-September) observations of Brewer’s Sparrows in western Colorado from monitoring data from the Rocky Mountain Bird Observatory (1999-2005) and the Integrated Monitoring of Bird Conservation Regions program (2006-2020), a coordinated landbird monitoring effort administered by Bird Conservancy of the Rockies (Pavlacky et al. 2017), vetted eBird checklists from April-September (eBird 2020a, 2022), breeding bird survey data from the Boulder County Nature Association, field observations from U.S. Forest Service biologists, trip reports on birding list servers, and museum specimens from VertNet (www.vertnet.org). We summarized alpine records in June-July separately from those in August-September because the latter may represent post-breeding, dispersing, or migrating birds rather than breeding birds.

Site selection

We reviewed historical records in relation to elevation (above mean sea level), landcover, date, and extent of shrubs visible in imagery to identify potential sites to survey with the goal of maximizing our chances of locating breeding Brewer’s Sparrows at each site. We first reviewed records in relation to a combination of elevation from a 10-m digital elevation model and landcover from a classified 25-m resolution vegetation layer (Colorado Vegetation Classification Project) to separate sites into high-elevation alpine shrubs or krummholz, subalpine shrubs, or lower-elevation sagebrush. We then restricted records to the first two-thirds of the breeding season (~June-July in alpine areas and ~April-June in sagebrush) to minimize inclusion of non-breeding records. We then examined the resulting records in relation to 1-m resolution, natural color imagery (National Agriculture Imagery Program) to qualitatively assess the extent of shrubs suitable for nesting. We then selected a non-random subset of alpine sites to survey based on a combination of elevation (> 3250 m), the number of historical breeding-season records, the extent of shrub-dominated landcover, land ownership (public only), ease of access (< 12 km from a road), and distance from other alpine sites to ensure broad spatial coverage. Finally, we selected a non-random subset of sagebrush sites to survey based on their proximity to selected alpine sites, elevation (1700-3100 m), number of historical breeding-season records, the extent of shrub-dominated landcover, ease of access, and land ownership.

Site descriptions

Sagebrush sites were on land owned by the Bureau of Land Management, the U.S. Forest Service, or Colorado Parks and Wildlife (CPW) with elevations ranging from 1746 to 3042 m. Most sagebrush sites were dominated by mountain big sagebrush (A. tridentata vaseyana), Wyoming big sagebrush (A. t. wyomingensis), or mountain silver sagebrush (A. cana viscidula). The San Luis Lakes State Wildlife Area site was dominated by black greasewood (Sarcobates vermiculatus) and the Tarryall State Wildlife Area site by shrubby cinquefoil (Dasiphora fruticosa). The highest elevation sagebrush sites, Land’s End and Indian Point, had mountain silver sagebrush and mountain big sagebrush interspersed with shrubby cinquefoil and common juniper (Juniperis communis). Alpine sites were on U.S. Forest Service land, including several in designated Wilderness Areas, with elevations ranging from 3338 to 3764 m. Alpine sites were dominated by patches of willow (diamondleaf willow [S. planifolia], shortfruit willow [S. brachycarpa], and grayleaf willow [S. glauca]), or patches of willow mixed with sparse Engelmann spruce (Picea engelmanni) or subalpine fir (Abies lasiocarpa) krummholz or saplings. Willow patches varied in size but were typically surrounded by or interspersed with alpine tundra.

Surveys, recording, and breeding status

At each site, we searched for Brewer’s Sparrows within patches of shrubs or conifer krummholz ≤ 3 m tall (Körner 1998). We visually tracked each singing male detected for up to 20 minutes and recorded songs with a Sennheiser^® MKE600 shotgun microphone and Sound Devices^® Mix-Pre 3 II digital sound recorder. We marked use locations (i.e., locations where males were singing or captured) using handheld GPS units (Garmin® GPSMap 64). If no birds were seen or heard during the first few minutes, we used brief song playback to elicit territorial responses. We recorded observations at alpine sites using eBird checklists and breeding codes (Confirmed, Probable, Possible) and evidence of breeding using behavior codes (eBird 2020b). We solicited assistance from experienced CPW and National Park Service volunteers and birders with surveying alpine sites as part of a citizen science effort. Volunteers and birders surveyed sites using similar protocols but did not record songs or capture birds.

Capture and sampling

Following observation and recording, we set up 3-m or 6-m Avinet^® mist nets and used song playback with wireless speakers to attract and capture territorial, singing males. We banded each captured individual with a numbered, aluminum, size 0A, butt-end, U.S. Fish and Wildlife Service leg band. We confirmed birds were breeding-age (after hatch-year) by skull pneumaticization and wing and tail shape (Pyle 1997). We determined sex by the size and angle of the cloacal protuberance and extent of brood patch development. Only females are known to develop brood patches (Pyle 1997). We examined wing and tail feathers for evidence of active molt. We photographed males using a smartphone camera (Samsung Galaxy S7 or iPhone 11) in natural sunlight. Most images were stored in raw (DNG) format in each of three positions (front, back, and side) from ~25-35 cm away with a color standard card (X-Rite^® ColorChecker Classic Mini). We plucked the two outermost tail feathers from each bird and stored feathers in glassine envelopes. We collected blood samples by puncturing the brachial-ulnar vein with a sterile, 26-gauge needle and collected the resulting droplet with a ~70μL non-heparinized capillary tube. We immediately transferred blood into a 1.5 mL centrifuge tube with Longmire’s buffer. We closed and sealed each tube with Parafilm^® prior to transport and storage. We kept feather and blood samples cool and shaded in the field, refrigerated them at ~1.7° C, then shipped them to the laboratory on ice. At the laboratory, we stored blood samples in a -70° C freezer (Owen 2011) and feather samples at room temperature.

Alpine habitat characteristics

Features of breweri breeding habitat in sagebrush are well studied (Rotenberry et al. 2020, Walker et al. 2020), so we only characterized habitat features at alpine sites. At each use location, we estimated the mean height of dominant shrub and tree species using 1.5 m tall mist net poles as a reference. We quantified landcover within a 100-m circular buffer around bird use locations and extracted elevation, slope, and aspect values from a 10-m digital elevation model in ArcPro^® software, version 10.8.2 (ESRI, Redlands, CA). We first calculated the mean of each habitat variable across use locations for each individual, then used the mean for each individual to calculate summary statistics across individuals.

External morphology

We collected standard external morphometric measurements on males (Pyle 1997). We measured unflattened wing chord and tail length using an Avinet^® stainless steel 15-cm wing/tail rule. We measured culmen length, bill width, bill depth, and tarsus length using Mitutoyo^® stainless steel dial calipers. We measured culmen length from the distal edge of the nares to the tip of the bill. Bill width and bill depth were measured at the distal edge of the nares. With a sample size of males > 30, we anticipated statistical power > 0.971-0.999 to detect differences of 2.4-7.2% in morphological measurements based on means and standard deviations from breweri and taverneri measured in Montana (B. Walker, unpublished data). Only two observers collected morphological measurements, and both observers measured 35 of the same individuals. We used morphological measurements taken by the first observer (BLW) in analyses whenever they were available (n = 68). For birds only measured by the second observer (AAY; n = 14), we added the mean difference in measurements between the first and second observer to the second observer’s values to account for potential inter-observer bias. The second observer’s measurements differed from those of the first observer by an average of 0.38% (tarsus), -0.40% (tail), 0.61% (wing), 1.09% (culmen depth), 2.49% (culmen width), and -3.02% (culmen length). We measured mass using a 30-g Pesola^® scale.

We compared external morphology between sagebrush and alpine males using two-sided t-tests with a sequential, ordered α adjustment for multiple tests (Benjamini and Hochberg 1995). We also compared values for alpine birds against morphological data from breweri and taverneri from range-wide studies (Klicka et al. 1999, Rotenberry et al. 2020). Male taverneri are larger on average, with longer wing, tail, and tarsus, larger mass (Klicka et al. 1999), and they reportedly have a narrower and shallower bill (Swarth and Brooks 1925, Doyle 1997, Klicka et al. 1999). We also used principal components analysis (PCA) to reduce the suite of seven morphological variables to fewer dimensions. We plotted the first two principal components and generated minimum convex hull polygons around songs from sagebrush and alpine sites to visualize and estimate overlap in overall acoustic structure. We also quantified the proportion of males that could be correctly classified to sagebrush or alpine by external morphology using random forest analysis (Cutler et al. 2007). We ran random forest analysis using the rfPermute function in the rfPermute package (version 2.5.1; Archer 2016) in R (version 4.1.3; R Core Team 2022). We included two random predictor variables (from the seven listed above) at each node split (mTry = 2), selected bootstrap samples with replacement, generated 5000 trees to ensure stable error estimates, and used a sample size equal to the number of males captured in each site type (n = 41 in sagebrush, n = 41 in alpine). We used random forest proximity plots to visualize the extent of overlap in external morphology between sagebrush and alpine males.

Song

Male Brewer’s Sparrows sing two categories of song types, short songs and long songs, but each male typically gives only one short song type, rarely two or three (Walker 2000, Rich 2002). Males generally only give short songs when unpaired, so short songs are thought to play a key role in mate attraction (Walker 2000) and, therefore, in the potential for reproductive isolation between subspecies (Mayr and Johnson 2001). Acoustic elements of taverneri songs are reported to have lower maximum frequencies, higher minimum frequencies, and, therefore, cover a narrower range of frequencies than those of breweri (Klicka et al. 1999, Walker 2024).

We analyzed songs using Raven Pro^® 1.6.3 (K. Lisa Yang Center for Bioacoustics 2023). We reviewed all recordings and selected one high-quality example of each short song type from each male for analysis. We added one recording from a public repository of a male singing short songs in willows in the Flattops Wilderness in Colorado to our alpine sample (XenoCanto 14188). Following Walker (2024), we used the selection tool to select (a) the entire song, (b) each section of the song, (c) one representative syllable type within each section, and (d) each note type within each selected syllable type. For each song, we measured song duration, total number of syllable types, and total number of note types. We then measured the following acoustic features on each selection: (a) peak frequency (the frequency at which maximum power occurred), (b) maximum frequency, (c) 95% frequency (the frequency that divides the selection into two intervals containing the lower 95% and the upper 5% of sound energy in the selection), (d) minimum frequency, (e) 5% frequency (the frequency that divides the selection into two intervals containing the lower 5% and the upper 95% of the sound energy in the selection), (f) 90% bandwidth (the difference between the 5% and 95% frequencies), and (e) aggregate entropy (the disorder in sound energy within the selection, a measure of sound complexity). We then calculated means for each variable across sections, across syllable types, and across note types for each song.

We tested for differences in acoustic features of short songs between sagebrush and alpine males using two-sided t-tests with α = 0.05 and with a sequential, ordered α adjustment for multiple tests (Benjamini and Hochberg 1995). We also compared the acoustic features of songs from sagebrush and alpine sites against those from breweri and taverneri songs across the species’ breeding range (Walker 2024). We included ten acoustic variables that showed the largest difference between sagebrush and alpine sites in random forest analysis to identify their relative importance and significance in classification. We used the same random forest analysis parameters as those used in the analysis of morphological data (above), except that sample size for each site type in each run (n = 26) was one-half the smallest sample (alpine, n = 52) to avoid classification bias due to unequal sample sizes (Archer et al. 2017).

Most acoustic variables were moderately or strongly correlated, so we used PCA to reduce the suite of correlated acoustic variables to fewer dimensions. We included six variables identified by the random forest analysis as significant predictors in the PCA. We plotted the first two principal components and generated minimum convex hull polygons around songs from sagebrush and alpine sites to visualize and estimate overlap in overall acoustic structure.

Following Walker (2024), we also used song bandwidth and mean note bandwidth, two acoustic variables useful for separating short songs of breweri and taverneri, to estimate the probability that songs of sagebrush and alpine males were from breweri.

Plumage and maxilla color

We measured and compared the color of seven features reported to differ between breweri and taverneri between sagebrush and alpine males, including the top of the maxilla, base color of the back, breast, dorsal streaks, flanks, submoustachial stripe, and supercilium (Swarth and Brooks 1925, Pyle and Howell 1996, Doyle 1997), as well as the color of the eyestripe. Male taverneri were described as having the top of the maxilla dark brown to blackish and much darker than breweri. Male taverneri also have darker plumage overall, including grayish breast, flanks, submoustachial stripe, and supercilium, and a grayish-brown back with blackish dorsal streaks. In contrast, male breweri have a whitish breast, flanks, submoustachial stripe, and supercilium and a sandy-brown back with dark brown dorsal streaks.

We used the micaToolbox plugin (version 1.22; Troscianko and Stevens 2015) for ImageJ software (Schneider et al. 2012) to measure color in the visible spectrum (400-700 nm) from digital photographs. We converted raw images in DNG format to linear, normalized, grayscale red, green, and blue (RGB) reflectance stacks standardized against reflectance values for neutral gray 2 (3.22) and neutral gray 6.5 (38.40) standards to control for variation in ambient light. We calibrated each image using the 24-color standard card included in each photo to create a standardized, color-calibrated, multi-spectral image. We outlined a region of interest (ROI) for each feature and extracted mean linear, normalized mean RGB values across pixels within each ROI. This produced a dataset in which the color of each feature for each individual was represented as a point in three-dimensional linear, normalized RGB color space, with each point having three coordinates (RGB) scaled from 0-100. Points with coordinates closer to the origin (0, 0, 0; pure black) have lower reflectance on one or more axes.

We first checked for differences in color between sagebrush and alpine males by visually assessing whether the colors of each feature for each individual plotted in three-dimensional RGB color space clustered by site type and whether alpine males clustered closer to the origin as predicted if alpine birds are taverneri. We tested for statistical differences in the distributions of color between males at sagebrush and alpine sites using distance-based, non-parametric, permutational multivariate analysis of variance (PERMANOVA; Maia and White 2018). This procedure tests whether the centroids for points from sagebrush and alpine sites are in different locations in RGB color space, with the null hypothesis being that centroids are in the same location. Under the null, the observed distance between sagebrush and alpine centroids should be small and equivalent to distances obtained by permutation (i.e., repeated random reallocation of individuals to sagebrush or alpine). Distance-based PERMANOVA assumes that colors of each individual in the sample are independent and the dispersion of distances around the centroid within each site type are approximately equal. We ran PERMANOVA using the adonis2 function in the R package vegan (Oksanen et al. 2022) with pairwise distances between points as the dependent variable and site type as the independent variable. If alpine birds are taverneri, the distance between centroids for sagebrush and alpine males should be greater than those obtained via permutation.

Genetics

We extracted DNA from blood and residual tissue attached to feather quills and analyzed sequence data for the mtDNA cytochrome b gene following Klicka et al. (1999) (Appendix 1). Klicka et al. (1999) identified one base-pair substitution (locus 639) diagnostic for separating the subspecies. We then shipped DNA extractions on ice to the Bird Genoscape Project laboratory at Colorado State University and sequenced and analyzed the full genomes of 83 birds (Appendix 1). After identifying and removing three closely related individuals (kinship >0.0884), we calculated fixation indices for autosome (F_ST) and sex chromosome (F_ST-Z) scaffolds separately to test for genetic differentiation between sagebrush and alpine birds.

RESULTS

Historical records

We identified 186 June-July records of Brewer’s Sparrows at 59 alpine sites with shrubs at or above tree line at elevations ranging from 3334 to 4288 m, as well as 37 records at 23 subalpine sites with shrubs below tree line at elevations ranging from 2929 to 3323 m, between 1914-2022 (Fig. 1; Appendix 1: Table S1). Of the 223 total records, 90 were from Rocky Mountain Bird Observatory/Bird Conservancy of the Rockies monitoring data (1999-2017), 86 from eBird (1995-2022), 33 from the Boulder County Nature Association’s Indian Peaks Bird Count database, six from trip reports on birding list servers, four from U.S. Forest Service biologist field notes, two from VertNet specimen records, and two from The Colorado Breeding Bird Atlas (Lambeth 1998). Brewer’s Sparrows were also reported in eBird in August-September (1999-2022) at 11 of those 59 alpine sites plus 30 additional alpine sites (Appendix 1: Fig. S1, Table S2).

Surveys, capture, and banding

We conducted 57 surveys at 48 sites in May-July 2021-2023 (Fig. 2; Appendix 1: Table S3). We detected 181 males at 22 of 26 sagebrush sites surveyed and 78 males at 14 of 22 alpine sites surveyed (Appendix 1: Table S4). Alpine sites with detections ranged in elevation from 3395 to 3754 m. We captured and banded 41 males, one female, and one bird of unknown sex at 14 sagebrush sites and 41 males and one female at 12 alpine sites. No captured birds showed evidence of flight feather molt. Volunteers and birders reported 19 additional singing males, three adults of unknown sex, and four juveniles across 15 of 21 alpine sites they surveyed (including four sites also surveyed by CPW) (Appendix 1: Table S5). In combination, we detected total of 97 males, three adults of unknown sex, and four juveniles at 26 of 39 alpine sites surveyed. Combining our 2021-2023 survey results with historical records, Brewer’s Sparrows have now been documented in June-July from at least 72 alpine sites in western Colorado. That number jumps to 102 sites if August-September records are included. Volunteers and birders also counted 16 males at five of five subalpine sites they surveyed (Appendix 1: Table S5).

Breeding status and timing

We confirmed breeding at three alpine sites (Appendix 1: Tables S4, S5). We captured a female with a nearly fully developed brood patch at Rollins Pass on 23 June 2022. Birders confirmed adults with dependent fledglings at Rollins Pass on 17 July 2022. We found an active nest with three eggs ~21 cm off the ground in a conifer sapling growing within a 44-cm tall willow at Guanella Pass on 23 July 2023. We found a dependent juvenile being fed by an adult at Hoosier Pass on 27 July 2023. We considered breeding probable at 10 other alpine sites (and one subalpine site) based on the number of territorial, singing males detected, the presence of suspected breeding pairs, observations of territorial defense, or males captured with fully developed cloacal protuberances.

The absence of males at some alpine sites early in the season was informative about the timing of male arrival. Surveys at the Cumberland Pass, Scarp Ridge, Buck Mountain, and Kennebec Pass sites during 4-10 June detected only two males at the Buck Mountain site. At that time, willow leaves had just started to emerge and some willows were still buried under snow. In contrast, surveys from 15 June to 11 July detected seven males across those same four sites (Appendix 1: Tables S4, S5). Thus, males likely started arriving at alpine sites the second week of June (~June 8-10). Females likely first initiated nests during the last week of June. Based on an active nest with eggs on 23 July, nesting appears to continue through at least the end of July.

Alpine habitat characteristics

Use locations at alpine sites were typically in patches of willows, with or without conifer krummholz, growing in drier soil along the margins of larger, mesic willow patches or in isolated patches of willows surrounded by dry alpine tundra on ridges, slopes, or plateaus or in shallow basins (Fig. 3). At 172 use locations (n = 79 males) for which we recorded vegetation composition at alpine sites, 61.0% were in willow patches, 38.4% were in willow-dominated patches with sparse conifer krummholz cover, and 0.6% were in krummholz-dominated patches with sparse willows. At 136 use locations where we also collected vegetation height data, mean willow height averaged 1.3 m (range 0.9-1.8 m; n = 58 males) and mean conifer krummholz height averaged 2.6 m (range 1.3-5.0 m; n = 28 males). When conifer krummholz was present, males commonly sang from the tops of conifers above the surrounding willows.

Landcover within 100 m around use locations averaged 30.8% upland willow (range 0.0-100.0%), 3.9% conifer (range 0.0-67.3%), 56.3% alpine tundra (range 0.0-100.0%), and 9.0% bare ground/rock/snow (range 0.0-76.6%). Elevation at use locations at alpine sites ranged from 3458 to 3754 m (n = 172). Average slope at use locations had a median value of 18.5% (range 2.2-79.7%, n = 78 males). More males had use locations with a southern (136-225°), western (226-315°), or eastern (46-135°) aspect (33.3%, 28.2%, and 23.1%, respectively, of 78 males) than a northern (316-45°) aspect (15.4%).

External morphology

We collected morphological data on 82 males. We removed two outliers prior to analysis, including one sagebrush male with an abnormally long bill and an extreme value for culmen length (9.07 mm) and another sagebrush male that was abnormally heavy (14.7 g). We replaced each outlier with the mean for that variable across all other males in sagebrush in the PCA and random forest analyses.

Mean culmen length was 5.1% longer among alpine males, but there were no statistical differences in the six other metrics (Table 2). The distribution of culmen length values substantially overlapped between sagebrush and alpine males (Appendix 1, Fig. S2). Means for morphological metrics for both sagebrush and alpine males fell below (wing length) or within (tail length, tarsus length, and mass) the range of means reported for male breweri and, with the exception of tarsus, below the means reported for male taverneri by Klicka et al. (1999) and Rotenberry et al. (2020) (Table 2). Mean tarsus lengths for both sagebrush and alpine birds were similar to the mean reported for male taverneri in Rotenberry et al. (2020) but below the mean reported for male taverneri in Klicka et al. (1999).

The first two principal components explained 27.5% and 20.9%, respectively, of the variance in seven morphological variables in the PCA (Appendix 1: Tables S6, S7). Plotting the first two components indicated substantial overlap in external morphology between sagebrush and alpine males (Fig. 4A). Random forest analysis correctly classified 74.4% (95% CI: 63.6-83.4%) of 82 males to site type by external morphology, including 75.5% (95% CI: 59.7-83.4%) of sagebrush males and 73.2% (95% CI: 57.1-85.8%) of alpine males. Proximity plots also indicated substantial overlap in external morphology between sagebrush and alpine males (Fig. 4B).

Song

We analyzed 143 short song types from 134 males recorded at 19 sagebrush sites and 52 short song types from 49 males at 11 alpine sites (Appendix 1: Table S4). Five of 35 acoustic variables differed between sagebrush and alpine males (Table 3). Short songs of alpine males had narrower mean section, syllable, and note bandwidths, and higher mean note minimum and 5% frequencies than those of sagebrush males. The magnitude of differences between sagebrush and alpine males in those five metrics ranged from -11.8% to +5.3% (Table 3). Distributions of all acoustic variables from sagebrush and alpine sites showed substantial overlap (Appendix 1: Figs. S3-S6).

The first two principal components from PCA explained 55.4% and 15.2%, respectively, of the variance in ten acoustic variables included in the analysis (Appendix 1: Table S8). Bandwidth and maximum frequency variables loaded negatively on the first principal component, whereas mean minimum frequency and mean 5% frequency loaded positively on the second principal component (Appendix 1: Table S9). A plot of the first two principal components indicated that songs of alpine males were largely a subset of songs in sagebrush (Fig. 5A).

Random forest analysis indicated that three pairs of correlated variables were significant predictors of site type: mean note minimum and mean note 5% frequency, mean section bandwidth and mean section maximum frequency, and mean syllable bandwidth and mean syllable maximum frequency (Appendix 1: Fig. S7). Nonetheless, random forest analysis only correctly classified 67.2% (95% CI: 60.1-73.7%) of short songs to site type, including 65.7% (95% CI: 57.3-73.5%) of sagebrush songs and 71.2% (95% CI: 56.9-82.9%) of alpine songs. Proximity plots indicated that songs of alpine males were a subset of songs of sagebrush males (Fig. 5B).

The PCA including range-wide data indicated that sagebrush and alpine short songs from Colorado more closely matched those of range-wide breweri than taverneri (Fig. 6). Based on song bandwidth and mean note bandwidth, 98.6% of 143 songs from sagebrush sites and 96.1% of 52 songs from alpine sites had a > 0.50 probability of being from breweri.

Plumage and maxilla color

We obtained images suitable for analysis from 36 males at sagebrush sites and 38 males at alpine sites (Appendix 1: Fig. S8). The distributions of RGB reflectance values substantially overlapped between sagebrush and alpine males for all features (Appendix 1: Fig. S9). Results of PERMANOVA indicated that alpine males had lower reflectance (i.e., darker colors) than sagebrush males for four of the eight features, including back (F_1,72 = 19.81, P < 0.01), breast (F_1,72 = 8.48, P < 0.01), eyestripe (F_1,72 = 5.62, P = 0.02), and flanks (F_1,72 = 20.31, P < 0.01), but not for dorsal streaks (F_1,72 = 0.68, P = 0.42), submoustachial stripe (F_1,72 = 1.45, P= 0.23), supercilium (F_1,72 = 2.25, P = 0.13), or maxilla (F_1,72 = 3.04, P = 0.08).

Genetics

We obtained feathers from all 85 birds captured. We obtained blood from 39 of 43 birds captured at sagebrush sites and 42 of 42 birds at alpine sites. We successfully extracted DNA and analyzed mtDNA from blood for 79 males and one bird of unknown sex (later confirmed as female), and from feathers from the remaining three males and two females. Sagebrush and alpine birds had similar mtDNA haplotype distributions. At sagebrush sites, 95.3% of 43 birds had a breweri haplotype and the remaining 4.7% had a taverneri haplotype (one male at Parlin and one male at Green Mountain Reservoir). At alpine sites, 95.2% of 42 birds had a breweri haplotype and 4.8% had a taverneri haplotype (one male at Rollins Pass and one male at Hoosier Pass).

We sequenced and analyzed the full genomes of 41 birds from 11 alpine sites and 42 birds from 13 sagebrush sites. After removing data from three closely related males (two from California Park and one from Land’s End), we found 14,660,401 bi-allelic single nucleotide polymorphisms across the genomes of 80 individuals. Sagebrush and alpine birds showed no genomic differentiation in autosomes (F_ST = 0.000130, 95% CI: -0.0193-0.0485) or sex chromosomes (F_ST-Z = 0.000031, 95% CI: -0.0197-0.0481).

DISCUSSION

Three lines of evidence indicate that alpine Brewer’s Sparrows found in western Colorado in June-July are breweri breeding in atypical habitat. First, despite similarities between alpine birds and taverneri in breeding phenology and habitat features and alpine males showing minor differences in song and plumage in the expected direction for taverneri, alpine males largely overlapped with sagebrush breweri males in most morphological and acoustic features and in coloration. Second, birds at sagebrush and alpine sites both had > 95% breweri mtDNA haplotypes and showed no genomic differentiation of either autosomes or sex chromosomes. Third, including an historical record of fledglings from the Flattops Wilderness on 3 August 1988 (Lambeth 1998), breeding has now been confirmed at four alpine sites in Colorado. Overall, our results point to three possibilities: (1) alpine breweri and sagebrush breweri interbreed, (2) alpine breweri and sagebrush breweri do not interbreed but have not diverged, or (3) alpine birds are itinerant breeders. Although breweri regularly raise two or three broods in sagebrush (Mahony et al. 2002), documentation of itinerant breeding by North American songbirds is rare (Baldassarre et al. 2019).

We ruled out several other potential explanations. The lack of genomic differentiation between sagebrush breweri and alpine birds ruled out the possibility that alpine birds represent a distinct genetic cluster within breweri. Alpine areas are not a zone of introgressed breweri-taverneri hybrids or backcrosses because alpine birds were genetically indistinguishable from sagebrush breweri and fewer than 5% of alpine birds had taverneri haplotypes. There was also no geographic pattern in the occurrence of taverneri haplotypes as would be expected if Colorado contained a contact zone between subspecies. Alpine birds are not a third subspecies because we only detected breweri and taverneri mtDNA haplotypes. Although breeding has only been confirmed at four sites, as outlined previously, we suspect that all alpine birds we encountered in June-July were breeding birds. Although some birds in alpine areas conceivably could be dispersing, transient, or migrating breweri, all of the birds we captured were adults rather than juveniles and none showed evidence of prebasic flight feather molt typical of upslope post-breeding dispersers (Pyle et al. 2018).

Our results also indicate that Colorado supports a larger and more widely distributed breeding population of alpine breweri than currently recognized. First, despite limited surveys, we detected 100 adults at alpine sites statewide. This suggests Brewer’s Sparrows are not uncommon in alpine areas, at least within specific vegetation associations and elevation ranges. Second, Brewer’s Sparrows have now been documented in June-July at a total of 72 alpine sites across most major mountain ranges in the state. Farther north, taverneri often remain on alpine breeding territories through early September (Swarth 1936), so if birds reported at the 30 additional alpine sites in August-September also represent local breeders, then the breeding distribution of alpine breweri in Colorado is even broader. Third, birders unaffiliated with our citizen science effort reported Brewer’s Sparrows at additional alpine sites in Colorado in 2023 (eBird 2023). Finally, there are undoubtedly many more alpine breeding sites in Colorado that have not yet been found considering the extent of unsurveyed willow and willow-krummholz patches at or above tree line statewide. Additional surveys and habitat modeling will be required to map their breeding distribution and estimate their abundance.

It remains unclear why the Brewer’s Sparrow was not previously known as a widespread breeding species in alpine areas in Colorado. It may be because access to alpine areas is poor early in the breeding season when male song rates and detection probability are highest (Walker 2000) and birds have simply gone undetected in many alpine areas. Breeding taverneri were largely unknown prior to intensive, targeted surveys in east-central Alaska (Doyle 1997) and northwestern Montana (Griffin et al. 2003). Alternatively, Brewer’s Sparrows may have only recently started colonizing alpine areas.

Notably, Brewer’s Sparrows have also now been reported from 24 subalpine sites below tree line in June-July, including sites dominated by willows, shrubby cinquefoil, and snowberry (Symphoricarpos sp.). Confirmation of breeding at subalpine sites would further expand the known breeding distribution and vegetation associations of this subspecies. Although breweri is widely considered a sagebrush-obligate (e.g., Donnelly et al. 2017), our results support the conclusion that it is instead a shrub-obligate that nests in many different shrub species (Zillig et al. 2023).

Based on our results, we suspect that June-July records of territorial male Brewer’s Sparrows in alpine shrub or krummholz communities in western states south of taverneri breeding range (e.g., California, Idaho, Nevada, Oregon, Utah, and Wyoming) are more likely breeding breweri, but additional surveys in British Columbia, Montana, and Washington are needed to confirm where the boundary occurs between breeding breweri and taverneri. Notably, singing and displaying male Sage Thrashers (Oreoscoptes montanus), another “sagebrush-obligate” species, have also been reported in willow stands at several alpine sites in Colorado (Righter et al. 2004, eBird 2023), including during our surveys, but whether they are breeding remains unknown.

Similarities in vegetation associations and timing of breeding between alpine breweri in Colorado and taverneri farther north were striking. Both are generally restricted to shrubs or conifer krummholz within a narrow range of elevations above tree line (Doyle 1997, Griffin et al. 2003, Starzomski 2015, Stuyck et al. 2021). In Alaska, taverneri occur in large patches of 1.0-1.2 m tall alpine shrubs (diamondleaf, grayleaf, barrenground [S. niphoclada], Richardson’s [S. richardsonii], and tealeaf willow [S. pulchra]) with an understory of stunted resin birch (Betula glandulosa), blueberry (Vaccinium uliginosum), and shrubby cinquefoil (Doyle 1997, Stuyck et al. 2021). In British Columbia and Alberta, taverneri breed in resin birch, bog birch (B. pumila), willow, and subalpine fir krummholz (Swarth 1930, Nordin et al. 1988, Doyle 1997). In Montana, taverneri are primarily found in subalpine fir krummholz sometimes with an understory of shrubby cinquefoil or common juniper (Griffin et al. 2003).

Although our survey approach precluded collecting comprehensive data on the timing of breeding at any given site, accumulated observations indicate that breeding phenology at alpine sites in Colorado was 5-7 weeks later than at sagebrush sites and closely matched that of taverneri farther north. Our data suggest that alpine males start arriving the second week of June and alpine females nest from the third week of June through the last week of July. Male taverneri start arriving as early as 29 May in British Columbia and Yukon Territory and mid-June in Alaska. Female taverneri are thought to initiate nests starting the third week of June, with fledging starting the second week of July (Swarth 1930, Swarth 1936, Doyle 1997, Stuyck et al. 2021). In contrast, male breweri first start arriving in mid-elevation sagebrush sites in western Colorado in mid-April and initiate nests by mid-May (Lambeth 1998, Righter et al. 2004, Magee 2016, eBird 2022). Alpine snow melted early in Colorado in 2021 due to below-average snowpack and minimal May-June precipitation, so the timing of arrival and breeding at alpine sites in Colorado may be even later in years with normal or above-average snowpack. Determining whether alpine Brewer’s Sparrows are itinerant breeders is a priority for future research.

Differences in breeding phenology and habitat characteristics between populations are widely thought to contribute to reproductive isolation and may ultimately lead to speciation. Sagebrush and alpine breweri breed in close proximity in Colorado (< 12 km), but timing of breeding and habitat features used by alpine birds more closely match those of taverneri > 1300 km away. Our findings complicate interpretation of differences in timing of breeding and breeding habitat as supporting criteria for subspecific identification and taxonomic delineation in this species (contra Klicka et al. 1999, 2001). The proximity of sagebrush and alpine breeding sites in Colorado may facilitate interbreeding (or itinerant breeding), which in turn, would prevent reproductive isolation and subsequent genetic divergence between sagebrush and alpine populations.

Some plumage and song features of alpine birds showed tendencies toward taverneri. However, differences in plumage color between breweri and taverneri have never been quantified, so it is unclear if the differences in plumage we found are taxonomically relevant or whether alpine birds represent an intermediate phenotype between subspecies. Future studies of plumage color in the Brewer’s Sparrow should measure the full spectrum of light visible to songbirds (~300-700 nm). We also lacked species-specific data on the spectral sensitivity of photoreceptors, so were unable to convert reflectance to cone-catch images that more closely represent what birds see. For that reason, we were unable to confirm that the color differences we measured were perceptible to other males and females (Maia and White 2018). Alpine males also gave short songs with narrower mean frequency bandwidths, which is in the expected direction if alpine males were taverneri, but the absolute magnitudes of differences we observed (5.3-11.8%) were much smaller than those between subspecies (36.7-42.6%, Walker 2024). Nonetheless, the cause of narrower frequency bandwidths among alpine males would be worth investigating in light of their relevance to reproductive isolation and taxonomy. In the absence of genetic differences, environmental influences associated with high-elevation environments could produce similarities in plumage and song between alpine breweri and taverneri farther north.

Our study highlights the value of combining data from formal monitoring programs, citizen science efforts, and species-specific field research to document the distribution of bird species that breed in remote, mountainous regions. Most state and federal conservation assessments for the Brewer’s Sparrow in the western U.S. typically only consider sagebrush shrublands as breweri breeding habitat (e.g., Hansley and Beauvais 2004, Boyle and Reeder 2005). Our results will need to be incorporated into updated state and federal conservation assessments and breeding bird atlas accounts for breweri in Colorado and other western states to reflect the alpine (and possibly subalpine) breeding distribution and vegetation associations of this subspecies.

LITERATURE CITED

American Ornithologist’s Union. 1931. Check-list of North American birds. 4th edition. American Ornithologist’s Union. Washington, D. C., USA.

Andrews, R., and R. Righter. 1992. Colorado birds: a reference to their distribution and habitat. Denver Museum of Natural History Press, Denver, Colorado, USA.

Archer, E., 2016. rfPermute: estimate permutation p-values for Random Forest importance metrics. https://github.com/EricArcher/rfPermute

Archer, F. I., K. K. Martien, and B. L. Taylor. 2017. Diagnosability of mtDNA with Random Forests: using sequence data to delimit subspecies. Marine Mammal Science 33:101-131. https://doi.org/10.1111/mms.12414

Baldassarre, D. T., L. Campagna, H. A. Thomassen, J. W. Atwell, M. Chu, L. H. Crampton, R. C. Fleischer, and C. Riehl. 2019. GPS tracking and population genomics suggest itinerant breeding across drastically different habitats in the Phainopepla. Auk 136:ukz058. https://doi.org/10.1093/auk/ukz058

Benjamini, Y., and Y. Hochberg. 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society Series B 57:289–300. https://doi.org/10.1111/j.2517-6161.1995.tb02031.x

Boyle, S. A., and D. R. Reeder. 2005. Colorado sagebrush: a conservation assessment and strategy. Colorado Parks and Wildlife. Grand Junction, Colorado, USA. https://cpw.state.co.us/learn/Pages/SagebrushSpeciesConservationStrategy.aspx

Chesser, R. T., S. M. Billerman, K. J. Burns, C. Cicero, J. L. Dunn, B. E. Hernández-Baños, R. A. Jiménez, A. W. Kratter, N. A. Mason, I. J. Lovette, P. C. Rasmussen, J. V. Remsen, Jr., D. F. Stotz, and K. Winker. 2022. Checklist of North American birds. American Ornithological Society, Washington, D.C., USA. https://checklist.americanornithology.org/taxa

Colorado Parks and Wildlife. 2015. State wildlife action plan. Denver, Colorado, USA. https://cpw.state.co.us/aboutus/pages/statewildlifeactionplan.aspx

Cutler, D. R., T. C. Edwards, Jr., K. H. Beard, A. Cutler, K. T. Hess, J. Gibson, and J. J. Lawler. 2007. Random forests for classification in ecology. Ecology 88:2783-2792. https://doi.org/10.1890/07-0539.1

Donnelly, J. P., J. D. Tack, K. E. Doherty, D. E. Naugle, B. W. Allred, and V. J. Dreitz. 2017. Extending conifer removal and landscape protection strategies from sage-grouse to songbirds, a range-wide assessment. Rangeland Ecology and Management 70:95-105. https://doi.org/10.1016/j.rama.2016.10.009

Doyle, T. J. 1997. The Timberline Sparrow, Spizella (breweri) taverneri, in Alaska, with notes on breeding habitat and vocalizations. Western Birds 28:1-12.

eBird. 2020a. eBird Basic Dataset. Version: EBD_relNov-2020. Cornell Lab of Ornithology, Ithaca, New York. Nov 2020.

eBird. 2020b. Cornell Lab of Ornithology, Ithaca, New York, USA. https://support.ebird.org/en/support/solutions/articles/48000837520-ebird-breeding-and-behavior-codes

eBird. 2022. eBird Basic Dataset. Version: EBD_relAug-2022. Cornell Lab of Ornithology, Ithaca, New York, USA.

eBird. 2023. eBird Basic Dataset. Version: EBD_relNov-2023. Cornell Lab of Ornithology, Ithaca, New York, USA.

Griffin, S. C., B. L. Walker, and M. M. Hart. 2003. Using GIS to predict the distribution of timberline sparrows in northwestern Montana. Northwest Science 77:54-63.

Hansley, P. L., and G. P. Beauvais. 2004. Species assessment for Brewer’s sparrow (Spizella breweri) in Wyoming. Bureau of Land Management, Cheyenne, Wyoming, USA.

Klicka, J., R. M. Zink, J. C. Barlow, W. B. MacGillivray, and T. J. Doyle. 1999. Evidence supporting the recent origin and species status of the Timberline Sparrow (Spizella taverneri). Condor 101:577-588. https://doi.org/10.2307/1370187

Klicka, J., R. M. Zink, J. C. Barlow, W. B. MacGillivray, and T. J. Doyle. 2001. The taxonomic rank of Spizella taverneri: a response to Mayr and Johnson. Condor 103:420-422. https://doi.org/10.1093/condor/103.2.420

K. Lisa Yang Center for Conservation Bioacoustics at the Cornell Lab of Ornithology. 2023. Raven Pro: Interactive Sound Analysis Software (Version 1.6.4). The Cornell Lab of Ornithology, Ithaca, New York, USA. https://ravensoundsoftware.com/

Körner, C. 1998. A re-assessment of high elevation treeline positions and their explanation. Oecologia 115:445-459. https://doi.org/10.1007/s004420050540

Lambeth, R. 1998. Brewer’s Sparrow (Spizella breweri). Pages 456-457 in H. E. Kingery, editor. Colorado breeding bird atlas. Colorado Wildlife Heritage Foundation, Denver, Colorado, USA.

Leukering, T. 2008. Here’s looking at you...Spizella? American Birding Association. Birding 40:72-76.

Magee, P. 2016. Brewer’s Sparrow (Spizella breweri) in L. Wickersham, editor. The second Colorado breeding bird atlas. Colorado Bird Atlas Partnership, Denver, Colorado, USA.

Mahony, N. A., W. M. Vander Haegen, B. L. Walker, and P. G. Krannitz. 2002. Male incubation and multiple brooding in Brewer’s sparrows. Wilson Bulletin 113:441-444. https://doi.org/10.1676/0043-5643(2001)113[0441:MIAMBI]2.0.CO;2

Maia, R., and T. E. White. 2018. Comparing colors using visual models. Behavioral Ecology 29:649-659. https://doi.org/10.1093/beheco/ary017

Mayr, E., and N. K. Johnson. 2001. Is Spizella taverneri a species or subspecies? Condor 103:418-419. https://doi.org/10.1093/condor/103.2.418

McTaggart-Cowan, I. 1946. Notes on the distribution of Spizella breweri taverneri. Condor 48:93-94.

Nordin, K. E., W. B. MacGillivray, and G. C. Biermann. 1988. Courtship feeding in Brewer’s sparrows (Spizella breweri). Journal of Field Ornithology 59:33-36.

Oksanen J., G. Simpson, F. Blanchet, R. Kindt, P. Legendre, P. Minchin, R. O'Hara, P. Solymos, M. Stevens, E. Szoecs, H. Wagner, M. Barbour, M. Bedward, B. Bolker, D. Borcard, G. Carvalho, M. Chirico, M. De Caceres, S. Durand, H. Evangelista H, R. FitzJohn, M. Friendly, B. Furneaux, G. Hannigan, M. Hill, L. Lahti, D. McGlinn, M. Ouellette, E. Ribeiro Cunha, T. Smith, A. Stier, C. Ter Braak, and J. Weedon. 2022. vegan: community ecology package. R package version 2.6-4. https://CRAN.R-project.org/package=vegan

Owen, J. C. 2011. Collecting, processing, and storing avian blood: a review. Journal of Field Ornithology 82:339-354. https://doi.org/10.1111/j.1557-9263.2011.00338.x

Pavlacky, D. C. Jr, P. M. Lukacs, J. A. Blakesley, R. C. Skorkowsky, D. S. Klute, B. A. Hahn, V. J. Dreitz, T. L. George, and D. J. Hanni. 2017. A statistically rigorous sampling design to integrate avian monitoring and management within Bird Conservation Regions. PLoS ONE 12(10):e0185924. https://doi.org/10.1371/journal.pone.0185924

Pyle, P. 1997. Identification guide to North American birds. Part I. Slate Creek Press, Bolinas, California, USA.

Pyle, P., and S. N. G. Howell. 1996. Intraspecific variation and the identification of Spizella sparrows. Birding 28:374-387.

Pyle, P., J. F. Saracco, and D. F. DeSante. 2018. Evidence of widespread movements from breeding to molting grounds by North American landbirds. Auk 135:506-520. https://doi.org/10.1642/AUK-17-201.1

R Core Team. 2022. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/

Rich, T. D. 2002. The short song of Brewer’s Sparrow: individual and geographic variation in southern Idaho. Western North American Naturalist 62:288-299.

Righter, R., R. Levad, C. Dexter, and K. Potter. 2004. Birds of western Colorado Plateau and mesa country. Grand Valley Audubon Society, Grand Junction, Colorado, USA.

Rotenberry, J. T., M. A. Patten, and K. L. Preston. 2020. Brewer’s Sparrow (Spizella breweri), version 1.0. in A. F. Poole and F. B. Gill, editors. Birds of the world. Cornell Lab of Ornithology, Ithaca, New York, USA. https://doi.org/10.2173/bow.brespa.01

Sauer, J. R., J. E. Hines, J. E. Fallon, K. L. Pardieck, Jr., D. J. Ziolkowski, and W. A. Link. 2017. The North American breeding bird survey, results and analysis 1966-2015 (Version 2.07.17). USGS Patuxent Wildlife Research Center, Laurel, Maryland, USA. http://www.mbr-pwrc.usgs.gov/bbs/

Schneider, C. A., W. S. Rasband, and K. W. Eliceiri. 2012. NIH Image to ImageJ: 25 years of image analysis. Nature Methods 9:671-675. https://doi.org/10.1038/nmeth.2089

Semenchuk, G. P. 1992. The atlas of breeding birds of Alberta. Federation of Alberta Naturalists, Edmonton, Alberta, Canada.

Spencer, A. 2014. How to identify a Timberline Sparrow. https://earbirding.com/blog/archives/4764

Starzomski, B. 2015. Brewer’s Sparrow in P. J. A. Davidson, R. J. Cannings, A. R. Couturier, D. Lepage, and C. M. DiCorrado, editors. The atlas of the breeding birds of British Columbia, 2008-2012. Bird Studies Canada, Delta, British Columbia, Canada.

Stuyck, C. M., P. B. Leonard, G. V. Frost, M. L. S. Lake, and J. D. Walters. 2021. Update on the status and breeding phenology of the Timberline Sparrow (Spizella breweri taverneri) in Alaska. Western Birds 52:252-260. https://doi.org/10.21199/WB52.3.5

Swarth, H. S. 1930. Nesting of the Timberline Sparrow. Condor 32:255-257. https://doi.org/10.2307/1363398

Swarth, H. S. 1936. A list of the birds of the Atlin region, British Columbia. Proceedings of the California Academy of Sciences 23:35-58.

Swarth, H. S., and A. Brooks. 1925. The Timberline Sparrow: a new species from northwestern Canada. Condor 27:67-69. https://doi.org/10.2307/1363056

Troscianko J., and M. Stevens. 2015. Image calibration and analysis toolbox - a free software suite for objectively measuring reflectance, colour and pattern. Methods in Ecology and Evolution 6:1320-1331. https://doi.org/10.1111/2041-210X.12439

Walker, B. L. 2000. Structure, use, and function of song categories in Brewer’s Sparrows (Spizella breweri). Thesis. University of Montana, Missoula, Montana, USA.

Walker, B. L. 2024. Applying citizen science data to quantify differences in song between controversial avian taxa, the sagebrush and timberline subspecies of the Brewer’s Sparrow (Spizella breweri). Western North American Naturalist 83(4):10 https://scholarsarchive.byu.edu/wnan/vol83/iss4/10/

Walker, B. L., L. D. Igl, and J. A. Shaffer. 2020. The effects of management practices on grassland birds—Brewer’s Sparrow (Spizella breweri breweri). Chapter AA in D. H. Johnson, L. D. Igl, J. A. Shaffer, and J. P. DeLong, editors. The effects of management practices on grassland birds: U.S. Geological Survey Professional Paper 1842. https://doi.org/10.3133/pp1842AA

Zillig, M. W., F. A. Fogarty, and E. Fleishman. 2023. Bird associations with floristics and physiognomy differ across five biogeographic subregions of the Great Basin, USA. Ornithological Applications 125:1-15. https://doi.org/10.1093/ornithapp/duac040