Material

We characterized the carotenoids in 2 Yellow-shafted Flickers that showed no indication of intermediate characteristics due to hybridization with the Red-shafted Flicker except for the presence of red feathers in the wings and tail. The first bird, collected in West Stockbridge, Berkshire County, Massachusetts, on September 9, 1995 (Massachusetts College of Liberal Arts #16 [hereafter MCLA 16]; Figure 1), was a hatch-year (HY) male still actively growing its last 3 primaries (p8–p10) and the innermost rectrix (r1). Red tones are evident on 6 primaries, p4–p9 (p10 still in pin), identical on both sides, as well as on rectrices r2–r5, with r5 still growing and r1 in pin. P6 was the reddest primary with a shaft matching Munsell chip 7.5R 6/10 (from the Munsell Books of Color, Glossy Collection, Macbeth Division, Kollmorgen Instruments Corporation, New Windsor, New York, USA), similar to Smithe's (1975) Peach Red [color #94], which is actually redder than the shafts of Red-shafted Flickers in collections at the Royal Alberta Museum (RAM; Figure 1), which vary in hue from 10R to 1.25YR (Hudon et al. 2015); p4 had red hints only on its proximal fourth, whereas p5 turned red fairly abruptly about half-way from yellow distally; p7, p8, and p9 were of relatively uniform colors but of progressively less red coloration (Figure 1). On the tail, r2 was the reddest rectrix (shaft 10R 6/10; vane 2.5YR 4/8; compared to Munsell hues of 1.25YR – 2.5YR and 10R – 2.5YR, respectively, for Red-shafted Flickers), and r3 to r5 were of relatively uniform orange color but with progressively less red, r5 being the yellowest.

FIGURE 1.

Coloration of the shafts of the outer primaries of MCLA 16 compared to those of a Red-shafted Flicker from Alberta (RAM Z66.64.13; at the top).

The second was a second-year (SY) female found dead on April 28, 2008, in Bryn Mawr, Montgomery County, Pennsylvania, and prepared as a study skin and spread wing (Academy of Natural Sciences of Drexel University #192905 [hereafter ANSP 192905]). This bird is similar to MCLA 16, except that it has fewer and slightly different primaries with red tints: p3–p7. P5 was the reddest with a shaft matching Munsell chip 2.5YR 7/8, closest to Smithe's Salmon Color [#106]. On the tail, only r2–r4 had red tints (right r2 missing); the shaft in r3, the reddest, matched Munsell chip 7.5YR 7/8, approaching Smithe's Salmon Color [#6] (vane 7.5YR 5/6); left r4 had only hints of red (largely yellow-orange, almost yellow), while right r4 was entirely Spectrum Yellow [#55], like r1 and r5.

We compared the carotenoids and spectral characteristics of feathers in these birds to those of Red-shafted Flickers and hybrid Yellow-shafted × Red-shafted flickers from Alberta held at the Royal Alberta Museum. We note that the type of color aberration scrutinized here has not yet been documented in Alberta.

Furthermore, we examined the distribution of red color on study skins of additional Yellow-shafted Flickers with aberrant red feathers from eastern Canada (Ontario and Quebec) held at the Canadian Museum of Nature and the Royal Ontario Museum, and digital images of more flickers, primarily spread wings or specimens mounted with a spread wing, held in various museums in eastern North America and the American Midwest (Appendix Table 1). For each aberrant flicker, we recorded the affected flight feathers, and assessed the fraction of each feather with reddish tones. For the study skins from eastern Canada we also determined the coloration of the shaft at the base (where it is widest) of all primaries of aberrant color, under natural daylight, by matching to the closest color chip in the Munsell Book of Colors. Color chips vary in hue increments of 1.25 for colors of high chroma like those in the flicker, spanning 80 pages of color chips. In order to estimate and test for correlations we converted the Munsell Hue notations, ranging from the most yellow (2.5Y) to most red (7.5R), to numeric values. Intervals between hue notations in the Munsell color scheme represent equal amounts of change in hue.

Spectrophotometry

We acquired reflectance spectra for the shafts of primary feathers of MCLA 16, ANSP 192905, and several Alberta specimens (RAM Z82.27.118, Z82.33.73, Z83.44.01, and Z03.01.05) at their widest at the base of the feather in situ over the range of 350 to 800 nm using an Ocean Optics USB2000 spectrophotometer (Ocean Optics, Dunedin, Florida, USA) fitted with a R400-7 Y-shaped probe (Ocean Optics) shining light from a Mikropack DH-2000-Bal deuterium tungsten halogen lamp (Ocean Optics). The probe was placed 2.2 mm directly above the shaft of interest using a RPH-1 reflection probe holder (Ocean Optics). The spectrophotometer was operated with the OOIBase32 2.0.1.4 software using the following settings: integration time: 10 msec; spectra averaged: 100; boxcar smoothing: 5. Reflectance was compared to that of 99% Spectralon white standard (Ocean Optics) “light” control, while turning the light source off served as the “dark” control.

Biochemistry

We sampled vane sections roughly 1 cm² in area from both normally and aberrantly colored feathers of the 2 aberrant flickers from the eastern USA, as well as Red-shafted Flickers from western Alberta (RAM specimens Z93.13.08 and Z96.18.06). To rule out environmental contaminants like adventitious coloration or the application of colored material, we washed the material first with a dilute aqueous solution of dishwashing detergent (Sunlight, Phoenix Canada, Toronto, Ontario, Canada) and, once dried, with petroleum spirit prior to extraction. We weighed the vane samples with a Denver Instruments Pinnacle Series PI-225D (Denver Instruments, Arvada, Colorado, USA) precision analytical balance.

We extracted feather carotenoids using either methanol overnight in the dark or warm pyridine over a hot plate following Hudon et al. (2013) in sealed 1-dram borosilicate glass vials flushed with N₂ gas. The extracted carotenoids were transferred into methyl-tert-butyl ether (MTBE) with deionized water. We replaced the deionized water twice to remove as much of the methanol (or pyridine) as possible, and transferred the MTBE epiphase with the carotenoids to a clean vial. The ether extract was evaporated over a stream of N₂ gas and the pigments redissolved in a known volume of hexanes (OmniSolv, EM Science, Gibbstown, New Jersey, USA), using anhydrous sodium sulfate powder to remove any trace of water. We extracted rhodoxanthin from the berries of local bushes of L. tatarica, and the 4-keto-carotenoids in the Red-shafted Flicker from RAM specimens Z93.13.08 and Z96.18.06.

We acquired absorption spectra of the pigment extracts using the Ocean Optics spectrophotometer fitted with a CUV-FL-DA cuvette holder (Ocean Optics). Pigment concentration was determined in mg carotenoid per gram of feather tissue using the formula (A_peak × volume of extract [mL] × 10) / ( × feather mass (g)), where A_peak is the absorption at the extract's maximum, and is the extinction coefficient of a generic carotenoid in hexane, taken to be 2,500 (Britton 1985). We initially identified the carotenoids in the red sections of flicker feathers on the basis of color, relative mobility (R_f) and comparison to known standards after analytical thin-layer chromatography (TLC) on flexible sheets of silica gel (PE SIL G; Whatman Ltd, Maistone, Kent, England) and aluminum oxide (60 neutral; EM Science, Cherry Hill, New Jersey, USA) using mixtures of hexane and acetone (2:1 and 3:1, respectively) (Hudon et al. 2013).

We quantified individual carotenoids by high-performance liquid chromatography (HPLC) using a Waters instrument (Waters, Milford, Massachusetts, USA) equipped with 2 Waters 501 pumps, a 712 WISP autoinjector, a System Interface Module, and a Lambda Max 481 UV detector. We separated the carotenoid pigments isocratically using a Phenomenex Luna normal-phase column (150 mm × 4.6 mm i.d., 3 μM; Phenomenex, Torrance, California, USA) assisted with a hexane:acetone (86:14) mobile phase flowing at 1 mL min⁻¹ (Panfili et al. 2004, Prum et al. 2012). Carotenoids were detected at 450 nm. We determined the areas under the curve of the different pigments using the Millennium 2010 Chromatography software (v2.15.01; Waters, Milford, Massachusetts, USA). On this system, rhodoxanthin from L. tatarica separated into 3 large peaks at 7.8, 9.2, and 10.7 min roughly in 1:2:1 proportions, representing common stereoisomers of the retro-carotenoid (Englert and Vecchi 1982), and a few minor peaks. By comparison the 5 principal yellow carotenoids of the Yellow-shafted Flicker, β-carotene, β-cryptoxanthin, 3′-dehydro-lutein, lutein, and zeaxanthin (Hudon et al. 2015), eluted at 2.0, 5.5, 13.7, 21.2, and 22.6 min, respectively. We confirmed the identification of 3′-dehydro-lutein through reduction of its carbonyl group (C=O) with sodium borohydride in methanol, yielding lutein (Andrewes et al. 1974). The 4 principal red 4-keto-carotenoids of the Red-shafted Flicker, canthaxanthin, adonirubin, astaxanthin, and α-doradexanthin (Hudon et al. 2015), eluted at 4.9, 7.4, 11.9, and 14.8 min, respectively.

Timing of Deposition of the Red Pigments

We can estimate when a flicker deposited red pigments in its primaries from the knowledge of the timing of flight feather replacement during the prebasic or preformative molt in the Northern Flicker. Test (1945) and Pyle (1997) provide general information on timing of molt in the Northern Flicker, but lack specifics as to individual feathers. Fortunately, the molt of individual flight feathers is routinely recorded by banders at a few migration monitoring stations in North America in the form of molt scores. This involves assigning a score from 0 (an old feather) to 5 (new feather fully grown with no trace of sheath at its base), with intermediate scores for conditions in between, to each flight feather on both wings and the tail (Ginn and Melville 1983). The scores of individual feathers in a tract (primaries, secondaries, and rectrices) are added to yield a molt score for that tract (Ginn and Melville 1983). Flickers typically replace their primaries centrifugally starting with the innermost primary (p1), staggering feather replacement so that at any time a few primaries are at different stages of growth (Test 1945).

We obtained data on primary molt scores of 134 Northern Flickers banded at Manomet in Plymouth, Massachusetts, between 1978 and 2014, including 105 birds actively molting one or more primaries on either wing. For the flickers with aberrantly colored feathers that we solicited from various museums (see above), we calculated the molt score corresponding to the score the birds would have had just before they started depositing red pigments, as if the red sections and subsequent new primaries had not grown yet. For earlier-molted primaries that show a transition from yellow to red we measured the distance from the tip of the feather to the point of transition to red (to the closest mm), as well as the total length of the primary. A molt score was calculated for all fully grown primaries with distal yellow and red basal sections using the formula—molt score = 3.5 (l_y/l_t) + 1.0—where l_y is the length of the yellow (normally colored) section of a primary with red color, and l_t is the total length of the feather.

We developed this equation by extrapolating the score various lengths of primaries would elicit in a bander observing it in situ on a molting bird as a continuous variable. An earlier-molting primary with no trace of red got a score of 5. Primaries that were lost after the bird started depositing red pigments got a score of 0, unless they were preceded by a primary with a score of 3 (2.5 as a continuous variable) or more, in which case they received a score of 1 as they probably were missing when the bird started depositing red pigments based on the Manomet banding data. We then added the scores of individual primaries to get a total primary molt score. The molt scores for the 2 wings were highly correlated in specimens that had them (r = 0.99, n = 18). When information was available for only one wing (most spread wings), we doubled the score for the one wing to get a full primary molt score.

Statistics

We conducted statistical tests using StatView for Windows 5.0 (SAS Institute 1998), except when we used a one-tailed t-test to assess the significance of the correlation between the proportion of feathers with evidence of red and the redness of the reddest feather. We used Sigma Plot 9.0 to fit linear (Newton and Rothery 2009) and nonlinear (sigmoid, logistic, and Gompertz) regressions to primary molt score data of flickers at Manomet as a function of day of year, and calculate model-averaged parameter estimates and 95% confidence intervals. In the Results section, we report the values as means ± SE, except where indicated otherwise.

Spectrophotometry

The shafts and vanes of flight feathers in the Northern Flicker owe their colors mainly to a broad peak of absorption in the blue to green part of the visible spectrum resulting in a dip in reflectance between 400 and 510 nm (Figure 2). The absorption spectra of the feathers we examined differed mainly in the position and steepness of a slope of increasing reflectance above 500 nm (Figure 2). Increased redness in the hybrids and the Red-shafted Flicker resulted mainly from a shift of the slope of increasing reflectance to longer wavelengths (Figure 2A). In contrast, in the aberrantly colored sections of both MCLA 16 and ANSP 192905 (not shown) redness resulted mainly from increased absorption centered ~570 nm resulting in a shallowing of the slope of increasing reflectance (Figure 2B).

FIGURE 2.

Reflectance spectra of shafts of primary feathers (A) of auratus and cafer (Z82.33.73 [solid line] and Z03.01.05 [dotted line], respectively) and 2 hybrids between them (Z82.27.118 and Z83.44.01), (B) 4 different primary feathers of MCLA 16 (non-solid lines), compared to auratus individual (Z82.33.73; solid line).

The carotenoid extract of the red vane sections of both MCLA 16 and ANSP 192905 (not shown) absorbed maximally at 472 nm compared to 445 nm for the yellow sections of the same individuals (Figure 3) and 468 nm for cafer. The absorption spectrum of the red sections also had a shoulder at ~520 nm, and approximated the spectrum of rhodoxanthin, for example from L. tatarica in hexanes (Figure 3).

FIGURE 3.

Absorption spectra in hexanes of carotenoids in feathers of different colors for MCLA 16: aberrant red vane of p6 (solid line), normal yellow vane of p3 (dotted line), and extract of the berries of L. tatarica (stippled line).

Biochemistry

The red carotenoids in the reddish sections of aberrant Yellow-shafted Flickers matched rhodoxanthin in number, color, and mobility on thin-layer chromatography (TLC). The 3 large peaks of rhodoxanthin in L. tatarica were observed on HPLC as well, but were present in different ratios (roughly 17:22:8) in the birds than in the berries. There was an additional peak at 7.3 min, not present in the yellow sections of the 2 aberrant flickers. We suspect that this is a metabolite of the third peak, as it was nearly as abundant as the peak at 10.7 min (third), now reduced by an equivalent amount compared to its abundance in the berries, such that the combined abundance of the peaks at 7.3 and 10.7 min added up broadly to the abundance of the first peak of rhodoxanthin.

Together the 3 main and the minor peaks of rhodoxanthin accounted for 54% and 38% of the absorption at 450 nm of extracts from the reddest sections of MCLA 16 and ANSP 192905, respectively. When including the peak at 7.3 min, this figure increased to 64% and 49% of absorption at 450 nm, respectively.

As a result of the deposition of rhodoxanthin (and the peak at 7.3 min) in the red sections of flicker feathers the fraction of absorption at 450 nm assignable to the usual, yellow carotenoids in the Yellow-shafted Flicker decreased to 38% (MCLA 16) and 51% (ANSP 192905) from 100%. This decrease was most pronounced for lutein, the main yellow pigment in the Yellow-shafted Flicker, its absorption decreasing from 54% of absorption to 6% in MCLA 16 and from 50% to 17% in ANSP 192905. The decrease was not simply the result of the deposition of an additional pigment (rhodoxanthin) in the red sections, as amounts of lutein in the feathers also decreased in absolute terms, from 170 μg g⁻¹ feather to 11 μg g⁻¹ in MCLA 16, a 12-fold decrease, and from 81 μg g⁻¹ to 20 μg g⁻¹ in ANSP 192905 compared to its abundance in the corresponding yellow sections. A decrease by a third of absolute amounts of carotenoids in the red sections compared to the yellow sections can only account for a 3-fold reduction.

Interestingly, the absorption assignable to 3′-dehydro-lutein, a normal yellow carotenoid in the Yellow-shafted Flicker, increased in both proportion and absolute terms in the red sections, from 3% to 10% of absorption (9 μg g⁻¹ to 20 μg g⁻¹) in MCLA 16 and from 3% to 11% (4.4 μg g⁻¹ to 12 μg g⁻¹) in ANSP 192905, also in spite of a decrease by a third of total carotenoids in the red sections compared to the yellow sections.

Timing of Pigment Deposition

Feathers affected. We recorded reddish tones on p5 to p7 of approximately three-quarters of the flicker specimens we examined. Flickers also had reddish tones on other primaries, though rarely on p2 and never on p1. The transition from bright yellow (Spectrum Yellow) to reddish tones usually occurred on p3 to p6, rarely on p2, p7, or p8. Redness usually appeared fairly abruptly (over a few mm), rising to a maximum in the second or third primary with an aberrant coloration, usually the first mostly red primary (Figure 4). The redness then usually progressively faded in subsequent primaries until the color returned to the typical bright yellow color (Figure 4). Occasionally, the redness rebounded but then subsided again afterward (for example CMN 85284 on Figure 4) or built slowly (CMN 85269). We observed a general trend for individuals with redder primaries, presumed to have ingested more rhodoxanthin, to show a greater number of feathers with red tones (Pearson's r = 0.59, P = 0.005, n = 18; Figure 5). This is consistent with most birds having had a single “serving” of the pigment.

FIGURE 4.

Change in Munsell Hue of primary shafts (at their bases) of flickers from southern Canada and the eastern United States with more than 3 red-tinted primaries. Asterisks identify aberrant primaries that were yellow distally.

FIGURE 5.

Relationship between the proportion of molted primaries that are of aberrant colors and the redness of the shaft of the reddest primary in flickers from southern Canada and the eastern United States (n = 18).

Molt scores. Assessed molt scores of aberrant flickers at the onset of deposition of red pigments ranged from 17.1 to 69.5, with a mean of 44.6 ± 1.8 (n = 40), corresponding broadly to the replacement of the second, eighth, and fifth primary, respectively. Molt scores averaged slightly higher in flickers from eastern Canada (x̄ = 47.1 ± 3.0, n = 18) compared to those from the northeastern United States (Massachusetts, New York, Pennsylvania, and Maryland; x̄ = 40.8 ± 3.4, n = 13) and the American Midwest (Michigan, Wisconsin, and Illinois; x̄ = 45.2 ± 2.4, n = 9), but the differences were not statistically significant (Kruskal–Wallis Test df = 2, H = 3.507, P = 0.17).

The primary molt score of Northern Flickers banded at Manomet varies as a function of day of year and can be approximated by the following linear regression: molt score = 0.77 × day of the year − 124 (r = 0.90; Figure 6). The replacement of primaries occurs slightly earlier, is faster and less variable in AHY/ASY birds (slope = 0.85, y₀ = −146; r = 0.96, n = 15) compared to HY/SY birds (slope = 0.71, y₀ = −108; r = 0.86, n = 87; Figure 6). Because it was not always possible to determine the age of the flickers we examined, we used the regression for birds of all ages to estimate when the birds would have been depositing red pigments. Nonlinear regressions produced even closer fits (r = 0.96 for a 3-parameter logistic regression, for example), in part because primary score does not increase as fast at the beginning and end of the molt when few primaries are molted (Pimm 1973), but the choice of regression did not result in date projections that differed by more than 2 or 3 days in the middle of the feather molt from those of the linear regression for all birds.

FIGURE 6.

Relationship between the molt score of Northern Flickers banded at Manomet between 1968 and 2014 and the day of year. ASY individuals are identified by hollow circles. The data points were fitted to a simple linear regression that was used subsequently to estimate when the aberrant flickers would have deposited the red pigment. Bird with finished molts (scores of 0 or 100) were excluded from the fitted model. We also show 95% confidence intervals (stippled lines).

When estimating the date of acquisition of the red pigment, variability in the timing of replacement of primary feathers in the Northern Flicker created a far greater source of uncertainty than the choice of a particular regression. Indeed, a bird caught on August 10, and expected to have an average molt score of ~48, could be expected to have molt scores between ~24 and ~71 based on the 95% confidence interval calculated for the birds at Manomet. This is a difference in molt score of 48 (it is smaller for AHY/ASY [42]; greater for HY/SY [49]), corresponding to the replacement of a little over 4 or 5 primaries. We minimize the effect of this uncertainty by focusing on the average date for a large number of birds (41) from a wide range of localities in eastern North America and the American Midwest rather than on the individual dates. This results in dates from July 1 to September 7, and an average date of August 6, ranging from August 1 for the eastern U.S. to August 9 for southern Canada.

Using the linear regression to backdate dates of acquisition of the red pigment for birds that were still molting at the time of collection and had replaced fewer than 5 primaries since growing red feathers yielded dates of July 15 (CUMV uncatalogued), July 28 (ROM 90666), August 2 (CMN 85240), August 4 (UMMZ 155570), August 7 (ROM 72487), August 9 (UMMZ 126556), and August 27 (CMN 85241).

Biochemical Considerations

The presence of a novel, yet uncharacterized carotenoid in the feathers harboring rhodoxanthin and the change in proportions of the 3 main stereoisomers of rhodoxanthin in these feathers are consistent with the exogenous pigment being metabolized by the flicker. If confirmed, this would be the first report of the modification of exogenous rhodoxanthin by a bird after ingestion. However, we note that past attempts to separate rhodoxanthin chromatographically by HPLC were done using reverse phase media that yielded complex tracings with many peaks (for example Hudon et al. 2007, 2013), lacking the simple 1:2:1 pattern of stereoisomers generated by normal phases, where changes in proportions could easily be missed.

It is apparent that rhodoxanthin, or a metabolite, interferes with either the production or the deposition of lutein in flicker flight feathers to explain its great reduction in not just relative, but also absolute, terms in the feathers with rhodoxanthin. A clue as to what is happening in these feathers comes from the accumulation in both relative and absolute terms of 3′-dehydro-lutein, a yellow carotenoid normally present in the feathers of yellow color (Hudon et al. 2015). 3′-dehydro-lutein is thought to be an intermediate in the interconversion of ϵ- and β-end-rings and the production of lutein from zeaxanthin (Matsuno et al. 1986, Katsuyama and Matsuno 1988, Nagao et al. 2015). It is plausible that rhodoxanthin, or a metabolite, interferes with the reduction of 3′-dehydro-lutein to produce lutein, resulting in an accumulation of the former and a decrease in lutein. We note that rhodoxanthin shares with 3′-dehydro-lutein a conjugated keto group at C-3′ and could potentially bind, perhaps competitively, to the enzyme that reduces 3′-dehydro-lutein, preventing its transformation into lutein.

In the future it should be possible to assess the presence of exogenous rhodoxanthin in feathers spectrophotometrically without the need to extract and isolate the pigments, from the shoulder at ~570 nm on reflectance spectra of feathers that contain it (see also Hudon et al. 2013). This is in contrast to the situation in the red head feathers of the Western Tanager (Piranga ludoviciana), a species where rhodoxanthin occurs naturally, where no such shoulder exists (Hudon 1991).

Taxonomic Considerations

Our finding that diet largely explains the presence of red flight feathers in eastern flickers also has implications for the systematics of the Northern Flicker. Indeed, it is no longer possible to invoke the presence of red feathers in eastern flickers as evidence of extensive past or present introgression of Red-shafted genes into eastern Yellow-shafted populations, as intimated by Short (1965). Though the scale of hybridization between the 2 forms east of the Rockies constrains the splitting of the forms into distinct species, our findings are consistent with the notion that the 2 forms are genetically more distinct than previously understood, and perhaps approaching species status under the new comprehensive biologic species concept (Johnson et al. 1999).

Our findings also counter the notion expounded by Short (1965) that some populations of Yellow-shafted Flickers might be actively selecting for the red color to explain the prevalence and patchy distribution of aberrantly colored birds in the east. Such selection for red color would have been surprising given that flickers do not appear to discriminate on the basis of the color differences when pairing in the hybrid zone (Moore 1987, Flockhart and Wiebe 2009). Pockets of aberrantly colored flickers can be explained by the patchy distribution of the rhodoxanthin-bearing berries in the east.