Brighter is better: bill fluorescence increases social attraction in a colonial seabird and reveals a potential link with foraging

Abstract

Crested auklets (Aethia cristatella) are colonial seabirds with brilliant orange bills during the breeding season. We characterized the bill pigment with spectroscopy methods (resonance Raman, fluorescence, absorbance). We excluded carotenoids as a possible chromophore and showed that the pigment most closely resembles pterins. Like pterins the bill pigment fluoresces, and it occurred in two phenotypes that may differ geographically, perhaps due to environmental heterogeneity. The pigment is unique in the Genus Aethia, and its spectra did not match any known molecule. The UV-Vis absorbance spectrum of the bill pigment overlaps with the extracted pigment of euphausiids, a favored food of crested auklets. A color preference associated with prey may have favored characteristics of the crested auklet’s accessory bill plates. Crest size, a signal of dominance, tended to correlate positively with highest fluorescence in the single-band phenotype. Brighter bills may function in self-advertisement and verify the status signal of the crest ornament. We tested for a behavioral preference using identical decoys that differed only in bill fluorescence. Crested auklets approached models with fluorescent bills at a higher frequency. In cases where sex of crested auklets was known, males responded at a higher frequency to fluorescent bills, but females did not. In an evolutionary context, bill fluorescence could have conferred an advantage in social interactions, e.g., in dimly lit rock crevices. Bill brightness and color may communicate success in foraging and may function as an honest signal of mate quality.

INTRODUCTION

Luminescence in animal signals has a long evolutionary history, and recent discoveries suggest there is much yet to discover. Photoluminescent displays probably date back at least as early as the Mesozoic to the elaborate head ornaments of archosaurs (Woodruff et al. 2020). Bioluminescence, a luminescence generated by biochemical reactions, was recently discovered in the kitefin shark (Dalatias licha), the largest known vertebrate to express the trait (Mallefet et al. 2021). Fluorescence, a type of photoluminescence, was recently discovered in the platypus (Ornithorhynchus anatinus), making it the first observation of the phenomenon in a monotreme (Anich et al. 2020). In both bioluminescence and fluorescence, photons are released from molecules that have achieved an excited electronic state. In the case of bioluminescence, biochemical processes cause a chemical reaction. Energy from the chemical reaction is absorbed by a molecule that releases the energy as photons. In fluorescence, it is an external light source, e.g., sunlight, that causes excitation of molecules that then release photons. An analogy for bioluminescence from everyday life is the glow stick. When the glow stick is bent, chemicals that were separated in two compartments are combined. A chemical reaction ensues, and a molecule acquires an excited electronic state, causing the glow. Humans have long been fascinated with luminescence and fluorescence dating back to at least the early Greeks and Romans (Kane 2014). Examples of fluorescence are ubiquitous in everyday life, e.g., optical brighteners added to laundry detergent. These organic dyes absorb ultraviolet (UV) light and emit visible light in the blue and violet wavelengths (St. Laurent et al. 2007). This helps to compensate for loss of the blue spectrum of reflected light in faded white fabrics. Those fabrics would appear dull and discolored if not for the optical brighteners, which add to the blue spectrum of visible light (St. Laurent et al. 2007). Fluorescence went from novelty to norm in post-WWII America and elevated a cultural value of cleanliness through chemical means (Kane 2014). Fluorescence became associated with other cultural values during the next decades. During the 1960’s vivid fluorescent colors became associated with the California counterculture in part because of how they stimulate the human nervous system (Kane 2014). In recent years, ultraviolet light and visible blue light sources have been used to reveal the presence of fluorescence in several species of seabirds (Wails et al. 2017; Dunning et al. 2018; Wilkinson et al. 2019), terrestrial birds (Camacho et al. 2019) and various vertebrate species (Gruber and Sparks 2015; Kohler et al. 2019; Lamb and Davis 2020). Although these studies revealed that the animals fluoresce under an artificial light source, in most cases the studies did not provide evidence that the fluorescence functions in communication. The mere presence of fluorescence does not necessarily imply communication. Various natural objects are fluorescent or bioluminescent, including rocks and minerals (Schneider 2006), plants and invertebrates (Lagorio et al. 2015), and fungi (Oliveira et al. 2012). Even human skin contains naturally occurring fluorophores, some which mark skin aging (Na et al. 2001). Bacteria are ubiquitous and abundant and many are fluorescent (Ammor 2007). For example, Propionibacterium acnes, a commensal bacterium in human skin, produces porphyrins that fluoresce under a black light (Shu et al. 2013). Communication functions of luminescence in biological systems may be less than obvious. For example, some marine bacteria apparently “bait” zooplankton with bioluminescence, which in turn facilitates predation by the fish from which the bacteria will once again proliferate and disperse (Zarubin et al. 2012).

Fluorescence in avian plumage has been hypothesized to function in social signaling and sexual signaling. Fluorescence in avian plumage is not necessarily uncommon. Flight feathers of owls have been aged based on the amount of pink fluorescence that is evident under UV light (Weidensaul et al. 2011). Porphyrins, the source of the pink fluorescence, are rapidly degraded in the presence of sunlight (Galván et al. 2016). Most of the species that have received attention for porphyrins in plumage are nocturnal or crepuscular (Camacho et al. 2019). In red-necked nightjars (Caprimulgus ruficollis) body condition predicted the extent of UVA-induced pink fluorescence evident on the underwings for juvenile birds but not adult birds (Camacho et al. 2019). Illumination with a UV light source makes the porphyrins more apparent to human observers. It is unknown whether the pink fluorescence from porphyrins serves a communication function (Camacho et al. 2019). Certain types of fluorescence do appear to have special significance. Penguins (Spheniscidae) have unique fluorescent pigments (Jouventin et al. 2005; McGraw et al. 2007; Thomas et al. 2013). Fluorescent spots on the beaks of king penguins (Aptenodytes patagonicus) appear to function in mate choice (Ismaël et al. 2015). Bill spots and fluorescent auricular feather patches are hypothesized to signal individual quality, but apparently the potential information conveyed by these signals differs between the sexes (Viblanc et al. 2015). Two thirds of parrot species have fluorescent plumage, and there is an association between fluorescence and courtship displays (Hausmann et al. 2003). The juxtaposition of UV reflective plumage and fluorescent plumage has been hypothesized to increase the chromatic contrast in parrots’ visual systems (Hausmann et al. 2003). This may be the case for budgerigars (Melopsittacus undulatus), which have sexual dimorphism in their crown plumage. Males have brighter crowns than females in the visible wavelengths, and they are also brighter in a UVA wavelength (365 nm) that is close to the budgerigar’s peak sensitivity in UV visual perception (Pearn et al. 2003b). Fluorescent yellow plumage on the crown and cheeks of budgerigars have been hypothesized to function in sexual signaling and mate choice, but experimental studies are not in agreement (Pearn et al. 2001, 2003a, b; Arnold et al. 2002). In behavioral choice experiments, when UVA reflectance was removed by lighting and filters, UVA-induced fluorescence alone did not appear to function in mate choice (Pearn et al. 2003a). In conditions approximating normal daylight the contribution of fluorescence to the radiance of the budgerigars’ crown feathers in the yellow wavelengths was very low (Pearn et al. 2003b). UVA reflectance may be important for mate-choice in budgerigars, because it is important for perception of color hues (Pearn et al. 2001).

Crested auklets (Aethia cristatella) are colonial seabirds of Alaska and Siberia that express a constellation of ornamental traits during the breeding season including a forehead feather crest, white facial plumes, and brilliant orange accessory bill plates (Jones 1993a; Jones et al. 2000). Wails et al. (2017) found that the accessory bill plates of adults and subadults fluoresce under blue light, but other plumage did not fluoresce. Although some types of avian fluorescence such as porphyrins may not be readily noticeable to a human observer, the accessory bill plates of crested auklets have a bright appearance that is readily apparent without artificial illumination. We tested five hypotheses regarding the properties and potential functions of the crested auklet’s bill pigment, and we discuss them in the context of our anticipated outcomes. Our first objective was to identify the crested auklet’s pigment molecule, and we hypothesized (H1) that we could match the chemical characteristics of the pigment with a known compound. We used three spectroscopy techniques: resonance Raman spectroscopy (RRS), fluorescence spectroscopy (FS), and absorbance spectroscopy (AS). Similar techniques have been used to characterize orange and yellow fluorescent pigments important in arts and industry (Colombini and Kaifas 2010), as well as the unique yellow fluorescent pigments of penguins (McGraw et al. 2007; Thomas et al. 2013). We started our work with RRS, and based on insights from that work, we used FS to make comparisons of bill pigments between crested auklets and congenerics. We used AS to compare the crested auklet bill pigment to known chemical compounds. Our second objective was to test for geographic differences in the crested auklet fluorophore. We suspected that there could be qualitative or quantitative differences between breeding colonies due to environmental heterogeneity. Crested auklets migrate south from their northern Bering Sea colonies in fall and return during latter May to June. They acquire their bill pigmentation in spring. Birds from different breeding colonies may have different migrations routes and may encounter different ocean conditions and different prey. Therefore, we hypothesized (H2) that we could detect differences in the crested auklet bill pigment between colonies. Our third objective was to test for a relationship between bill fluorescence and crest size. The crested auklet’s crest ornament is an intrasexual and intersexual signal of dominance, and the bill is used in fighting (Jones and Hunter 1999; Zubakin et al. 2010). The crest ornament makes the head and bill appear larger, and the crested auklet’s species-specific citrus-like odorant causes individuals to smell stronger (Douglas et al. 2018). Octanal, the most abundant volatile chemical in the odorant, is positively correlated with crest size in males, indicating that odor is associated with dominance in this species (Douglas et al. 2018). We hypothesized (H3) that bill fluorescence would correlate with size of the crest ornament. In phenology research at the Cincinnati Zoo, HDD noted that adults acquire the bill pigment coincident with the ornamental feather crest (Fig. 1). We reasoned that expression of crest ornaments and ornamental bill plates may be governed by similar patterns of regulation. The crest ornament is a conventional signal that probably is not costly to produce (Jones and Hunter 1999). By contrast, bill fluorescence could be costly to produce in that the intensity of the signal depends on the allocation of pigment molecules to accessory bill plates. Crested auklets shed their ornamental crests and accessory bill plates late in the breeding season (Bédard and Sealy 1984; Jones 1993a). During the nonbreeding season crested auklets have a smaller stubby crest, and the bill is smaller and appears dark gray to black (Jones 1993b; HDD unpubl. data).

Fig. 1.

a Bill pigmentation and the crest feather ornament characteristic of breeding plumage are acquired during the same time in spring, April 18, 2006, live adult crested auklet. Photo by Steve Malowski, Cincinnati Zoo. b Colored dots on crested auklet rictal plate show points where measurements for fluorescent spectroscopy were taken from samples: yellow - lateral, blue – center, green – medial. Photo of live adult male by HD, Little Diomede I., AK, Aug 6, 2015

Our fourth hypothesis (H4) was that crested auklets would discriminate models of conspecifics with respect to bill fluorescence. We tested H4 with identical decoys that differed only in bill fluorescence. We predicted that crested auklets would respond preferentially to the fluorescence treatment. We compared the responses of males and females, but it was not possible to determine the sex of the responding bird in all cases. We reasoned that crested auklets perceive the bill fluorescence as increased luminance in yellow and orange wavelengths. Reflected light is a primary factor affecting color perception. Fluorescence can provide a value-added function by increasing the amount of reflected light. It is reasonable to suspect that crested auklets can detect added luminescence in the yellow-orange wavelengths. Birds have relatively fine discrimination of hues. In experimental tests accompanied with modeling birds performed similar to humans on color discrimination tests (Olsson et al. 2015). Birds have four spectral sensitivities that are relatively evenly distributed in their chromatic spacing (Marshall and Arikawa 2014). The avian visual system includes five types of visual pigments and seven types of photoreceptors (Hart 2001). Colored oil droplets in cone cells act as filters of light, narrowing the spectral absorption curves of visual pigments and resulting in efficient and powerful color discrimination (Hart 2001; Vorobyev 2003; Hart and Vorobyev 2005). The color visual systems are highly conserved in birds. The ancestral color visual system is known as violet-sensitive (VS), and it includes crested auklets and their Family Alcidae (Håstad et al. 2005). Yellow-orange wavelengths are within the sensitivities of the VS color visual system (Håstad et al. 2005).

A color preference could have favored the evolution of the crested auklet’s bill pigments. Crested auklets are specialist foragers on large zooplankton, particularly euphausiids and large copepods (Piatt et al. 1990; Jones 1993a). Their prey search image and visual system may be attuned to the color characteristics of these prey. We compared extracts of the crested auklet’s bill pigment to extracts of a favored food, euphausiids. We reasoned that if the pigments overlap in UV-Vis absorbance, then the bill pigment may be derived from prey (H5) and could have evolved based on a color preference associated with prey.

METHODS

We tested five null hypotheses regarding the crested auklet’s fluorescent bill pigment. We began work with RRS to test, H1, that the crested auklet bill pigment is not unique. Based on insights from RRS, we adopted FS to make comparisons of bill pigments among the crested auklet and its congenerics. Our second hypothesis, H2, was there would be no differences in fluorescence between breeding colonies. We took measurements with FS at three points on the anterior face of the crested auklet’s rictal plate (Fig. 1b). We validated the reproducibility of these measurements by taking three replicates of each measurement on a set of samples. We tested H3 that fluorescence does not correlate with the size of the crest ornament. We tested H4 that crested auklets do not discriminate between crested auklet models with respect to bill fluorescence. We used AS to test H5 that the crested auklet’s bill pigment and euphausiids, a favored prey, do not overlap in their UV-Vis spectrum.

We collected samples from three different field sites. The samples used for RRS were collected by HDD at Sivuqaq Mountain, St. Lawrence I. (SL), AK, on June 23, 2005 and the analysis was carried out later that summer. HDD obtained samples (n=10 crested auklets) for FS from Little Diomede I. (LD) in June 2008 and from Sivuqaq Mtn., SL (n=11 crested auklets) in latter June 2010. He also collected parakeet auklets (A. psittacula), n=3, and least auklets (A. pusilla), n=3, at SL in latter June 2010. HDD collected one whiskered auklet (A. pygmaea) from Little Tanaga Strait, Aleutian Is. on Jun 5, 2010. Almost no whiskered auklets occurred in this ocean pass in May-June 2010; in a typical year there are flocks of thousands. All specimens were frozen soon after collection and maintained at −80° C. Samples were shipped to and from University of Utah on dry ice. The crested auklet bill pigment degrades if not frozen, and the change in color and brightness is noticeable.

We attempted to characterize the physical structure of the crested auklet chromophore with resonance Raman spectroscopy (RRS). Initially we did not know the chemical composition of the bill pigment. We tested for carotenoids with the plan to quantify differences in carotenoids if they were present. RRS can be used to obtain quantitative measurements of total carotenoids in situ without altering the sample (Ermakov et al. 2004). Our results were negative for carotenoids, but we did measure a fluorescence peak at 527 nm. Three replicate readings were taken from four specimens. Samples were cut with a fine tipped scalpel from rictal plates. The experiment was performed with a laser (488 nm) directed on to the surface of a small (1 mm) piece. The custom-built Raman spectrometer consisted of a fiber-coupled light delivery and detection module, a compact air-cooled 488 nm argon laser excitation source (National Laser, Inc., Model H210), a small spectrograph (Jarrel Ash, Inc., Model MonoSpec 18) and coupled CCD camera (Santa Barbara Instrument Group, Inc., Model ST6v), and a laptop computer operating custom-tailored software for instrumentation control and data acquisition. The argon laser was routed through a 200 μm diameter multi-mode light delivery fiber to the optics module, expanded with a 25 mm focal length lens, sent through a 488 nm band-pass laser line filter and dichroic beam splitter, and focused with a 50 mm focal length lens onto the surface of the sample where it forms a ~1 mm diameter light excitation disk. The band-pass filter removed any light components other than the laser excitation light. The Raman-scattered and spectrally shifted light was collected in a 180-degree backscattering geometry with the same focusing lens. The light was expanded and reflected by the dichroic beam splitter into a separate light detection path. This path contained a holographic Rayleigh light rejection filter, and the light was focused via a 25 mm focal length fiber coupling lens into a fiber bundle that routes the collected light to the spectrograph. The light collection fiber bundle consisted of 61 multi-mode fibers each having 70 mm core diameter multi-mode fibers. The f/3.8 spectrograph employed a 1200 grooves/mm grating blazed for 500 nm. The CCD camera operated at −20 °C. The setup was designed for high light throughput at a moderate spectral resolution of ~25 cm⁻¹ that still allowed for easy separation of the expected spectrally narrow Raman lines from spectrally broad fluorescence backgrounds. The spectral response of the system was defined exclusively by the grating and CCD camera chip. We were unable to extract a Raman signature of the chromophore due to its strong intrinsic fluorescence background. Nevertheless, we identified fluorescence spectroscopy as a promising approach for characterizing the crested auklet’s fluorophore.

In analyses with FS we used an excitation wavelength of 488 nm. Our earlier work with RRS provided evidence that this wavelength would function well for our purposes. The λ_max of the fluorophore lies in the UV, but that is not necessarily the optimal excitation wavelength for quantitative spectroscopy measurements. In quantitative fluorescence spectroscopy it is necessary to establish experimental conditions appropriate for an optically thin sample. Optical density (OD) of the sample needs to be less than 1.0. Otherwise, fluorescence intensity will not reflect the actual concentrations of the fluorophore in biological samples (Guibault 1990; Gryczynski and Gryczynski 2019). Consequently, the measurements will not represent the differences in sampling volumes in samples with higher and lower concentrations. OD is proportional to both fluorophore concentration and its extinction coefficient (absorbance) for the given wavelength. In order to ensure lower OD we used a longer excitation wavelength (with low extinction coefficient). Our method was optimal for obtaining high resolution measurements from solid laminate substances. For additional technical information refer to chapter 4 “Fluorescence –Steady-State Phenomena” in Practical Fluorescence Spectroscopy (Gryczynski and Gryczynski 2019) and chapter 1 “General Aspects of Luminescence” (Guibault 1990). Our FS methods yielded highly reproducible measurements with minimal problems of light scattering and reflection. The spectroscopic instrumentation for fluorescence measurement was similar to that described above for RRS, with two exceptions. First, the fiber bundle was replaced with 200 μm multimode fiber. Second, the spectrometer was replaced with an Ocean Optics QE65000 scientific-grade spectrometer featuring spectral resolution of approximately 0.9 nm. Care was taken to ensure photo stability of the samples during the measurements. We verified that replicate measurements under the selected laser power (2 milliWatts), spot size, and exposure time, produced consistent values, and did not cause bleaching. The spectra were acquired in the range from 490 nm to 1000 nm. Spectral response from the samples exhibited either single or two-band structure. All of the spectra that we measured were easily classified as either a single or a two-band fluorescence phenotype. As a part of data reduction, we reported fluorescence intensity values in arbitrary units (a.u.) at 585 nm for the single-band spectrum and at 595 nm for the two-band spectrum. A typical example of each spectrum is shown in Fig. 2. These spectra were corrected for the spectral sensitivity of the spectrometer. The correction coefficients were obtained by comparing the measured spectrum of a Newport 6332 50W quartz tungsten halogen lamp with Planck curve corresponding to radiation of a blackbody with a temperature of 3400K. However, after the correction was applied the spectra were not substantially different from the uncorrected spectra (see SI 9, file name:ESM_9). Therefore, we deemed that the correction was not critical for purposes of our data analysis. Although Fig. 2 depicts a difference in fluorescence counts, these are individual spectra, and we do not imply that one phenotype is more fluorescent than the other.

Fig. 2.

Typical fluorescence spectra of ornamental bill plates under excitation with 488 nm: a single band phenotype and a two-band phenotype. (Note: The Fig. does not imply that one phenotype is more fluorescent)

Quantitative measurements of fluorescence were taken from the anterior surface of the rictal plate at its center and at points medial and lateral to the center. Fig. 1b shows these points of measurement on a live bird; we measured FS from samples. We took three replicates at each point of measurement on the LD samples. These replicate measurements were highly reproducible, and this validated our method. We compared measurements between colonies to test H2, and we analyzed these data with an independent samples t-test. The arithmetic mean of six readings for each individual was computed and compared between SL and LD. The Shapiro-Wilk Test was used to test for normality, and Levene’s Test was used to test for equality of variances. We tested H3 by comparing the size of crest ornaments to highest fluorescence and mean fluorescence. We reasoned that crested auklets may evaluate the brightest stimulus or overall brightness. Birds have relatively low ability to detect subtle differences in patterns and shading, referred to as contrast sensitivity (Ghim and Hodos 2006). Birds use spatial summation to improve color vision in low light levels (Olsson et al. 2017). At lower light levels the intensity thresholds needed for discrimination of color are greater. Photons falling simultaneously on larger areas of the retina must be added together to evoke the sensation of light. We chose to compare fluorescence and crest size in the single band phenotype, because it was the largest sample size. The longest crest feather was held straightened and measured with dial calipers (Central Toll Company, stainless steel 15 cm ± 0.02 mm). We selected a Spearman’s correlation test, because the data were not normally distributed. We confirmed the sex of crested auklet specimens by dissection.

Absorbance spectroscopy (AS) – We compared the UV-Vis spectrum of the crested auklet’s bill pigment to other candidate molecules and substances. Small slivers were cut from frozen rictal plates (n=12 LD, 15 SL) of crested auklet specimens used for fluorescence spectroscopy, including 22 right and left rictal plates of the same individuals. These data are available as SI 3 (ESM_3.xlsx). The samples were extracted in 200 μl of 0.1 M NaOH with a Max Q Mini 4000 at 26.0 C and 200 rpm for two hours. Optical density of the bill pigment extracts was scanned with a BioTek Epoch (200–700 nm, 1 nm resolution) and a quartz microplate (Hellma Analytics 730–009-44) with 100 μl per well, measured with a Rainin Pipet Lite (LTS 20–200 μl). A Hellma quartz microplate was selected because among the microplates tested by BioTek, it had the lowest background absorbance in the UV spectrum, as well as varying little across the visible range (Held 2009). We later used similar methods with a BioTek Cytation 1 to generate absorbance spectra for Figs in this manuscript. Those spectra were compared to pterins. We selected 6-biopterin, pterin, and pterin-6-carboxylic based on a survey of the literature and the commercial availability of these compounds. Our initial research with RRS suggested that the crested auklet fluorophore could be a pterin. Yellow pterin pigments fluoresce (Bradbury and Vehrencamp 2011, p. 129), and pterins were identified as the likely pigment in yellow fluorescent penguin feathers (McGraw et al. 2007; Thomas et al. 2013). HD obtained 6-biopterin (≥ 97%, B2517-MG), pterine (~ 95%, P1132–50MG), and pterine-6-carboxylic acid (≥ 98.0%, 82553–100MG) from Sigma-Aldrich, Inc. The solubilities of these compounds were 10 mg/ml in 1 M NaOH for 6-biopterin, 1 mg/ml of 0.05 M NaOH for pterine, and 1 mg/ml of 0.01 M NaOH for pterin-6-carboxylic acid, according to Sigma-Aldrich technical staff. Solutions were prepared at their solubilities and diluted. The 6-biopterin solution was diluted by a factor of two, and the other pterins by a factor of five, in 0.1 M NaOH. We also compared extracts of euphausiids to the crested auklet bill pigment. Frozen euphausiids (0.26 g, Thysanoessa spp.) were extracted in 400 μl 1 M NaOH, and 10.0 μl of the extract was combined with 90.0 μl deionized water, resulting in a 0.1 M NaOH solution. HD obtained these euphausiids from crested auklet regurgitations at St. Lawrence I., AK. He also obtained a sample of T. inermis from the buccal cavity of a crested auklet, recovered from commercial fishing gear in the eastern Aleutian Is. This sample was obtained in spring, preceding acquisition of the bill pigment. Ten euphausiids from this sample were preserved in 1 ml chloroform and frozen. A subsample of this extract (10 μl) was combined with 90 μl of 0.1 M NaOH. Optical density of the pterin compounds and the T. inermis were read in the quartz microplate with a BioTek Cytation 1 spectrophotometer. Optical density of the St. Lawrence euphausiids was measured with a BioTek Epoch 2. Data were processed with Gen5 version 3.09 software by BioTek.

Behavioral assay - Behavioral assays were conducted for a total of 32 hours between June 16–24 and on June 29, 2016. This was the pre-laying period. Some breeding adult crested auklets were observed in courtship and pair formation during this time. Behavioral assays were conducted during morning (0830–1305) and evening (2100–0200) social activity. We used three replicates to increase the likelihood of having crested auklets interact with models. Social behavior is variable, and the rate of response to static models can be relatively low. Six identical decoys of life-like crested auklets were fabricated (see details at end of methods). Three models were randomly assigned to have fluorescent orange bills, and three models had orange bills that lacked fluorescence (Fig. 3). The decoys’ bills had reflectance in wavelength bands that closely approximated the bill reflectance spectrum of crested auklets. Fig. 4 shows the specular reflectance spectrum of a live crested auklet captured on Little Diomede in summer 2016 compared to the bills of decoys from the experimental and control treatments. The reflectance measurements were taken with an Ocean Optics JAZ spectrometer using the same exact equipment and sampling parameters. Data for Fig. 4 are available as SI 8 (file name ESM_8.xlsx). The pairing of the individual experimental and control models was random for each trial. In each pair, the experimental and control models were placed approximately 1.5 m apart in erect posture. The models were not anchored. The base of each model was cradled on rock, and this was so the decoys would not present too much resistance against agonistic behavior. The configurations were set up so that they could easily be monitored by a single observer from a natural hide. HDD worked alone, and therefore; we could not record the data blind. The experimental setup was moved to a new location with each trial. Crested auklets have breeding-site fidelity. Breeding adults return to the same breeding sites in subsequent years (Zubakin 1990; Jones et al. 2000; Fraser et al. 2004; Major et al. 2017; Wails et al. 2021). Therefore, we think new individuals were sampled with each trial. During social activity periods crested auklets interact on the surface of the colony and descend below the surface to visit prospective nesting crevices or exchange duties at the nest (i.e., incubation). The birds in our study were not individually marked, but HDD was able to carefully monitor behavior of individual birds during trials. A bird that approached the same model twice was counted as a single approach. This occurred four times. A bird that approached both models was not included in the statistical analysis; this occurred twice. A marine heat wave affected attendance of crested auklets at LD in 2016 (Douglas et al. 2020). Densities of crested auklets were lower than expected. There were no behavioral responses in some observation periods due to very few crested auklets being present. For example, on June 17 HDD set up the models in a colony on the east side of LD where there would usually be thousands of auklets, but there were almost none. Social behavior was subdued in some activity periods. For example, on the evening of June 21, crested auklets roosted on the colony, and there were few social interactions. Birds were able to knock over the models, and this was part of the design so as not to offer too much resistance to agonistic behavior.

Fig. 3.

Crested auklet decoys used in behavioral research. The control treatment, the decoy with the nonfluorescent bill is on the left. The experimental treatment, a decoy with the fluorescent bill is on the right. Photo May 28, 2021

Fig. 4.

Specular reflectance spectra obtained from the bill surface of the experimental and control decoys and the rictal plate of a crested auklet captured at LD in summer 2016

Behavioral assays were scored based on simple behaviors that could be easily and efficiently quantified. Behavior was tabulated in consecutive 15-minute observation periods. This was optimal, because it allowed for earlier termination of observations at the end of social activity periods or adjustments to experiments as needed. We used three simple categories of behavior, because this made it easier for one observer to score three replicates. First, approaches were categorized as a direct approach within one body length of the model (e.g., Fig. 5). Second, sniffing behavior was categorized as the positioning of the bird’s nares in close proximity to the model in a manner consistent with odor assessment. Third, touching behavior was categorized as direct intentional contact with the model. The sex of responding birds was identified in 54% of approaches. Sex was determined based on bill characteristics (Jones 1993b), and the male’s larger “bull” neck (Douglas 2008). HDD noticed that some males at LD appeared to be noticeably larger, and subsistence hunters agreed. Some crested auklets are intermediate in their bill characteristics and can be difficult to distinguish as male or female, even when held in the hand (Jones 1993b). HDD classified sex based on morphology. If the sex of the bird was not determined definitively by morphology it was categorized as unidentified. HDD recorded qualitative data on specific behaviors to document how crested auklets interacted with the models. However, he decided a priori not to include a rigorous recording of frequency and intensity of social behaviors. He observed social behaviors including threat postures, trumpeting, arch display, hunch display, oblique posture with throat presentation, and billing (Jones 1993a; Zubakin et al. 2010). Billing behavior was described briefly by Jones (1993a) as nibbling of the bill by the opposite sex, in the context of courtship. Billing appears to be important early in the breeding season for reducing aggression and pair formation (Zubakin et al. 2010). The behavior can be somewhat aggressive early in the breeding season with touches to a partner’s plumage changing into pecks (Zubakin et al. 2010). In this study, the term billing describes a bird using its bill to touch, rub, and/or nibble the bill of the model. Crested auklets secrete volatile aldehydes from small wick-like feathers in the intrascapular region, and prospective mates transfer these compounds during alloanointing (Douglas 2008). In this study, the term alloanointing describes a crested auklet rubbing its body over what is the scent-secreting or scent-bearing region of a display partner (Douglas 2008). Alloanointing appears to be an intentional transfer of chemicals between conspecifics, particularly prospective mates (Douglas 2008). Some sniffing behaviors may have been similar to a behavior named “ruff-sniff”, and some touch behaviors may have been similar to a behavior named “touch” (Jones 1993a). Behaviors associated with chemical signaling include the ruff-sniff behavior (Jones 1993a; Hagelin et al. 2003) and alloanointing (Douglas 2008). Ruff-sniff was described as a crested auklet burying its bill in the nape feathers of a conspecific (Jones 1993a). The decoys had what appeared to be contour feathers as part of their molded features, but they did not have actual contour feathers. Although the decoys lacked actual plumage on the nape, some of the behaviors we described as sniffing may have been similar to ruff-sniff. Some of the behaviors described as approaches may have been similar to a behavior named “touch”. The “touch” behavior is preceded by an exaggerated posture with an extended neck (Jones 1993a). During approaches some crested auklets leaned in and extended neck, head, and bill towards the decoy, without proceeding to behavior categories 2 and 3, sniffing and touching.

Fig. 5.

Crested auklet beside a decoy model at LD in summer 2016

The three replicates were collapsed into single columns of experimental and control treatments for data analysis. As Fig. 4 shows, the fluorescent bills of decoys had higher reflectance than the bills of the control decoys within the color spectrum of crested auklet bills. This was expected because fluorescence increases the apparent intensity of colors (reflected light) within specific wavelengths (Springsteen 1999). We predicted that crested auklets would respond more strongly to fluorescent bills. We chose to use the Wilcoxon Matched Pairs Signed-Ranks Test, after evaluating normality with the D’Agostino-Pearson omnibus test (D’agostino et al. 1990; Yap and Sim 2011), because the difference scores between variables were not normally distributed. Parametric tests such as the paired samples t-test assume that difference scores between variables are normally distributed (Zar 1999). Statistical analyses were performed using IBM SPSS v27 and MS Excel 2016 with the Real Statistics data analysis tool.

HDD fabricated decoys from a silicone mold. He made the mold with materials from Fiberlay Inc. HDD took the mold impression from a wood carving of a crested auklet. D. Wedl created the carving for the mold. He used crested auklet specimens as a reference. The decoys were completed more than a year prior to the behavioral study; this was ample time for off gassing of the decoys. All decoys were finished in the same manner, with round wiggle eyes (5 mm Creativity Street 409848) and acrylic paints to match plumage (M. Graham & Co. Mars Black Series 9 52–115, Payne Gray Series 122–128, Titanium White Series 7 52–180). The bills of three control models were painted with an acrylic orange paint (Delta Ceramacoat Tangerine 02043). The bills of three experimental models were painted with an acrylic fluorescent orange paint (Palmer Prism Tempera 55–4408). Actual crest feather ornaments of the same size were affixed to the foreheads of the models with a glue. These crest feather ornaments were obtained from birds taken by subsistence hunters at LD. The decoys did not differ with respect to odor, and odors were not added during the experiments.

RESULTS

Resonance Raman spectroscopy – The results were negative for carotenoids, but we observed strong fluorescence at 527 nm, similar to Yellow Fluorescent Protein (YFP-10C). Our fluorescence measurements were reproducible, and the more intense the color of the bill pigment the more reliable the fluorescent signal. The standard error of measurements, expressed as a percentage of the mean value, ranged from 0.6 to 16 percent. There was a statistically significant difference between individuals as determined by one-way ANOVA (F (3,8) =10.57, p=0.004). A Tukey’s HSD post-hoc test revealed that the intensity of fluorescence for one individual was higher (p=0.005 to 0.01, Cohen’s d=3.56 to 3.99). Our results showed that the chromophore is not a carotenoid. Under the same experimental conditions, a carotenoid would have emitted a characteristic Raman spectrum as described for zeaxanthin and lutein (Ermakov et al. 2004) and for astaxanthin and canthaxanthin (Ermakov et al. 2006). Our RRS data are available as supplementary information (SI) 1 (file name ESM_1.xlsx).

Fluorescence spectroscopy -- No other species in the Genus Aethia manifested bill fluorescence in response to the experimental conditions (488 nm excitation wavelength). Crested auklets exhibited two distinct phenotypes with respect to fluorescence, a single-band spectral signature and a two-band spectral signature. These were distinct phenotypes. Every spectrum that we measured was easily classified. Typical examples of the fluorescence spectra for these two phenotypes are shown in Fig. 2 (note that this Fig. does not imply that one phenotype is more fluorescent). Maximum fluorescence was at approximately 585 nm for the single-band phenotype and 595 nm for the two-band phenotype. All crested auklets from SL were of the single-band phenotype, but both phenotypes were present at LD. The spectral differences were not attributable to sex. Eight males and three females comprised samples from SL; six males and four females comprised samples from LD. At LD, two females and one male had the single-band signature, and the others had the two-band signature. Fluorescence did not differ quantitatively between colonies (t_{2-tailed (19)} =0.68, p=0.5); assumptions of normality and homogeneity of variances were met. Fluorescence did not differ quantitatively between sexes (t_{2-tailed (19)} = 0.85, p = 0.41). Our FS data are available as SI 2 (file name ESM_2.xlsx).

Although crest size correlated with fluorescence (Fig. 6), it was not significant. Length of the longest crest feather tended to correlate positively with the highest fluorescence readings (r_{s 2-tailed} = 0.48, p=0.08, n=14) for the single-band phenotype. A positive correlation between crest size and mean fluorescence was not significant (p=0.12). The mean values for crest size did not differ significantly between colonies (p= 0.9). Our data for the crest size and sex of specimens are available as SI 2 (file name ESM_2.xlsx).

Fig. 6.

Correlation of highest fluorescence reading versus length of longest crest feather (mm) in crested auklets with the single-band phenotype

Absorbance spectroscopy - The UV-Vis spectrum for the crested auklet’s bill pigment was similar to some pterin molecules (Fig. 7a, b). Measurements of λ_max for the crested auklet bill pigment were 243–250 nm. The λ_max for the pterins were 253 nm, 6-biopterin; 249 nm, pterine; and 263 nm, pterin-6-carboxylic acid. These pterin molecules had a smaller band of absorption at 221 nm for pterine and 217–218 nm for the other pterins. The crested auklet bill pigment absorbs in these wavelengths, and it also has a smaller sideband at 290–292 that overlaps with pterin-6-carboxylic acid. The pterins absorbed strongly in wavelengths around 350–360, but the crested auklet fluorophore did not absorb in these wavelengths. Data for Figs 7a, b are available in SI 4 and 5, respectively (file names ESM_4.xlsx and ESM_5.xlsx).

Fig. 7.

UV-Vis’s absorbance of crested auklet bill pigment and pterin molecules. a) Absorbance of crested auklet bill extracts from LD and SL. b) Absorbance of 6-biopterin, pterin-6-carboxylic acid, and pterine compared to a crested auklet bill extract from SL

Behavioral assay - Crested auklets responded at a higher frequency to models with fluorescent bills (63 vs. 37 percent). A Wilcoxon Matched Pairs Signed-Ranks Test indicated that more crested auklets approached models with fluorescent bills than the control models, z=2.60, p=0.009, n=128. Fig 8 shows how the frequency of approaches differed cumulatively between treatments across the study. The sex of the responding bird was identified by morphology in 53 percent of cases, and this total included more males than females (44 vs. 28). Males tended to approach fluorescent models more frequently than controls (68 vs. 32 percent), z=1.75, p=0.08 (Wilcoxon Matched Pairs Signed-Ranks Test). Females approached the treatments at the same rate. Crested auklets sniffed more fluorescent models than controls, and this difference approached significance z=1.85, p=0.06, n=128 (Wilcoxon Matched Pairs Signed-Ranks Test). Birds sniffed the nape and hindneck, where the species’ citrus-like odorant is usually evident (Jones and Hunter 1999; Douglas et al. 2001). They also sniffed the back and the location where the uropygial gland is located on a live bird. More crested auklets touched fluorescent models than controls (61 vs. 39 percent), but this difference was not significant (p=0.14). Touch involved billing and alloanointing. In billing behavior, the bird used its bill to touch, rub, or nibble the bill of the model. In alloanointing birds rubbed their bodies over what would be scent-secreting and scent-bearing regions in conspecifics (Douglas 2008). Billing behaviors can become aggressive when the display partners are not well acquainted (Zubakin et al. 2010). In two instances males billed with and then began biting control models. Pecking of the crest ornament, a common agonistic behavior (Jones 1993a) was directed once at a fluorescent decoy, and one fluorescent decoy was knocked over when it was pecked. Crested auklets adopted postures and behaviors in response to decoys, including trumpeting, arch display, hunch, touch, threat postures, and oblique posture with throat presentation (Jones 1993a; Zubakin et al. 2010). Some sniffing behaviors may have been similar to a behavior named “ruff-sniff”, and some touch behaviors appeared to be similar to a behavior named “touch” (Jones 1993a; Zubakin et al. 2010). Also, some crested auklets that approached decoys leaned in and extended neck head and bill towards the decoy, a posture that generally precedes a behavior named “touch.” Crested auklets vocalized at the models, but vocalizations, other than trumpeting, could not be identified based on posture. Behavioral data are available in SI 6 (file name ESM_6.xlsx).

Fig. 8.

Cumulative approaches by wild crested auklets to decoys with the fluorescent bill (experimental) and control treatments across 15 min. observation periods

Absorption spectra of the bill pigment and extracts of euphausiids (Thysanoessa spp.) overlapped (Figs 7a, 9). The maximum absorption for the euphausiid extracts occurred at 222–223 nm, comparable to the smaller absorption band in pterine (Fig. 7b). The other pterins and the crested auklet fluorophore overlapped with this band. The SL euphausiids had a sideband at 271 nm, and this overlapped with the sideband in the crested auklet bill pigment (Figs 7a, 9). Extracts of the crested auklet bill pigment and the euphausiids were similar in color. Data for Fig. 9 are available as SI 7 (file name ESM_7.xlsx).

Fig. 9.

UV-Vis absorbance spectrum of extracts from euphausiids (Thysanoessa spp.) and a crested auklet bill from SL

DISCUSSION

We characterized properties and functions of a unique fluorescent pigment from the marine environment. This pigment occurs in the accessory bill plates of the crested auklet, a colonial seabird of Alaska and Siberia. We showed that the crested auklet’s fluorophore is unique in its genus. The bill pigments of parakeet, least, and whiskered auklets did not emit fluorescence under FS. Accordingly, we did not reject H1, that the crested auklet bill pigment is not unique. We excluded carotenoids as possible chromophores in the bill pigment, because we did not observe the characteristic spectrums under RRS. The Raman spectra of carotenoids produce prominent Stokes lines at 1008-, 1159-, and 1524 cm⁻¹ (e.g., Fig. 3 in Ermakov et al. 2004). If carotenoids had been present, this spectral Raman signature should have been observable even on the top of a strong intrinsic fluorescence background. Hence, our RRS data were not consistent with carotenoids. IE observed that the more intense the color of the bill pigment the more reliable the fluorescent signal. We interpreted this as evidence that bill fluorescence is directly related to the yellow-orange pigment. A fluorescence emission spectra obtained with RRS matched YFP-10C (527 nm) (McAnaney et al. 2005), but absorbance data did not match literature values for YFP-10C. Our estimate of position for the fluorescent peak may have been imprecise, due to limitations of RRS. The crested auklet fluorophore is similar to some pterin compounds. Pterins may have two peak emission bands (McGraw et al. 2007) like the two-band phenotype of the crested auklet fluorophore (Fig. 2). Extracts of the bill pigment have a yellow-orange color and this is also characteristic of some pterins. We found that three pterin compounds (6-biopterin, pterine, pterin-6-carboxylic acid) had a λ_max similar to the crested auklet fluorophore (Fig. 7a, b). The UV-Vis absorbance spectrums that we reported for 6-biopterin and pterine (Fig. 7b) were similar to data reported in the literature (Nisshanthini et al. 2015, Fig. 5). The three pterins absorbed strongly around 350–360 nm (Fig. 7b), but the crested auklet fluorophore did not absorb in these wavelengths (Fig. 7a). Some examples of fluorescence are the result of structural properties or the arrangement of molecules, but this does not appear to be the case for the crested auklet fluorophore. Our data indicate the crested auklet’s bill is fluorescent due to a pigment molecule.

We found two phenotypes of the crested auklet’s fluorophore (Fig. 2), and there were geographic differences. All of our samples from SL were of the single band phenotype, while our samples from LD were both single-band and two-band phenotypes. Differences in fluorescence phenotypes could not be explained based on sex, because males and females were of either phenotype. Based on this evidence, we rejected H2 of no differences in fluorescence between breeding colonies. The mean values of fluorescence did not differ quantitatively between LD and SL, or between males and females. Intensity of fluorescence did not differ systematically between the two fluorescence phenotypes, i.e., individuals with one of the fluorescence phenotypes did not consistently have brighter bills than the other phenotype. A previous study did not find geographic variation in genetic markers, vocalizations, or morphometrics in crested auklets (Pshenichnikova et al. 2015). Therefore, environmental heterogeneity may be a more parsimonious explanation for our observations of apparent geographic differences in fluorescence phenotypes. Samples for FS were collected from LD in 2008 and from SL in 2010. Both years were affected by ENSO (El Niño-Southern Oscillation) patterns but in different ways. A negative phase of the Pacific Decadal Oscillation (PDO) began in September 2007, coincidental to the start of the 2007–2008 La Niña of ENSO in August 2007 (Xue and Reynolds 2009). During La Niña and negative PDO sea surface temperatures are cooler, and the negative PDO that persisted through 2008 was the lowest since 1971 (Xue and Reynolds 2009). In 2010, the warm phase of ENSO led off the year, followed by a neutral period in late spring and a switch to the cool phase of ENSO in summer (Xue et al. 2011). Oceanographic conditions set the stage for the composition of plankton communities. Euphausiids, a favored food of crested auklets, have species-specific habitat requirements, and their oceanic distributions and abundances can be affected by climate (Cimino et al. 2020). Crested auklets from LD could have encountered different prey species in their spring foraging areas, if they followed different migratory routes. Differences in the fluorescent bill pigments might be attributable to specific differences in the zooplankton the birds consumed.

The crested auklet fluorophore is probably important in communication. Crested auklet bills with higher fluorescence are brighter in the yellow-orange visible spectrum. Yellow-orange wavelengths are within the sensitivities of the VS color visual system (Håstad et al. 2005), the visual system of crested auklets. We noted that the more intense the color of the bill pigment the more reliable the fluorescent signal in RRS. In terms of color discrimination, birds have been shown to have capabilities comparable to humans (Olsson et al. 2015). Birds have four visual spectral sensitivities relatively evenly spaced; colored oil droplets within cone cells filter and more narrowly define spectral absorption curves, providing excellent color discrimination (Hart 2001; Vorobyev 2003; Marshall and Arikawa 2014). Fluorescence can augment a signal, making the signaler appear brighter and perhaps stronger to conspecifics. The size of the crest ornament tended to correlate positively with highest fluorescence in the single-band phenotype (Fig. 6). However, the correlation was not significant, and therefore, we did not reject H3, which hypothesized no correlation between fluorescence and crest ornament. Bill fluorescence may help to validate a symbolic message conveyed by crest size. The crest ornament is probably relatively cheap to produce, and this status signal may be vulnerable to “cheaters” (Jones and Hunter 1999). By contrast, allocation of fluorescent pigments to the bill plates could be more costly, and signaling at higher fluorescence would require committing greater amounts of the fluorophore. This is because differences in the amount of fluorescence are attributable to differences in the concentration of fluorescent molecules present (Baker et al. 1985). Brighter bills may advertise success in foraging and greater capacity to allocate endogenous resources for display. In our behavioral assay, crested auklets were more attracted to models with fluorescent bills (Fig. 8). Reflectance from the fluorescent treatment was higher than the control treatment in the same wavelengths of visible light, and this was in approximately the same wavelengths as crested auklet bills (Fig. 4). In total, crested auklets approached models with fluorescent bills at a higher frequency than controls. Therefore, we rejected H4, that crested auklets do not discriminate bill fluorescence. We concluded that bill fluorescence is attractive to crested auklets. Crested auklets interacted with decoys as though they were conspecifics. They adopted behavioral postures that are typical of social behavior. More crested auklets sniffed the fluorescent model than controls, and the difference approached significance. Chemical signaling and alloanointing, the transfer of chemicals between conspecifics, is an aspect of crested auklet social behavior. Although more crested auklets touched the fluorescent decoys, this difference was not significant. There were differences in behavioral responses to decoys based on sex. More males approached decoys than females, and more males approached the fluorescent models than the controls. This was not true for females; females approached the treatments in equal numbers. This could be explained by previously observed differences in social behavior between sexes as well as our methods. Agonistic behavior is more frequent among male crested auklets (Jones and Hunter 1999). Interactions between males are more likely to progress to physical aggression (Jones and Hunter 1999). Female-female interactions are mediated by signaling with the crest ornament and may not progress to overt physical combat as frequently as in males (Jones and Hunter 1999). Static models may not elicit as strong a response from female crested auklets. Jones and Hunter (1999) suggested that females may need a more direct stimulus to provoke aggression. In our study we scored approaches within one body length of the models. Males may have been more motivated to closely approach the fluorescent treatment, particularly if fluorescence is perceived as a status signal.

The crested auklet’s bill pigment could be patterned on color preferences related to their prey. The crested auklet’s bill pigment and the bird’s euphausiid prey have similar chemical properties, suggesting the bill pigment could be acquired from prey. The UV-Vis spectrum of the crested auklet bill pigment overlapped and partially enveloped the absorbance band of its euphausiid prey (Fig. 9). Therefore, we rejected H5 of no overlap between the bill pigment and pigments of prey. Prey from foraging areas used by crested auklets during spring as well as prey from foraging areas near the breeding colonies may contribute to the bill pigment. We included samples of both in our study. We included T. inermis from the eastern Aleutians, where crested auklets occur in large aggregations during winter and spring. Crested auklets begin acquiring bill pigmentation in spring while foraging in these areas. We also included Thysanoessa spp. from SL, collected during the breeding season. Euphausiids are a favored prey of crested auklets (Bédard 1969; Piatt et al. 1990; Jones 1993a). The crested auklet’s fluorophore may have been sequestered or metabolized from its zooplankton prey. Pterins occur in decapod crustaceans (Palmer et al. 2018), and the crested auklet’s fluorescent pigment has chemical properties similar to pterins. Furthermore, nearly all species of euphausiid crustaceans have bioluminescent photophores (Clarke 1963; Herring and Locket 1978). Those photophores could be a source of molecules that contribute to luminescence and possibly bill fluorescence. Although pterin pigments account for coloration of the iris in some birds (Oliphant 1987), the genes involved are not expressed in the integument of birds or mammals (Andrade et al. 2021). The fluorescent pigment must be supplied via the circulating blood plasma to the cells that secrete the bill pigment, and we suspect it is sequestered or metabolized from prey.

Fluorescence appears to accentuate the accessory bill plates. This could augment the function of the bill in social signaling. Birds as a group have relatively low contrast sensitivity (Ghim and Hodos 2006). Low contrast sensitivity means that objects of similar luminance values are more difficult to distinguish. The plumage of crested auklets is dark sooty gray overall, not unlike the relatively dark rock substrates of their colonies. Meteorological conditions at many breeding colonies include frequent fog and mist, thick enough to obscure visibility. Fluorescence could help to increase the chromatic contrast of the bill against dark backgrounds and in conditions of diminished visibility. The crested auklet’s bill is prominently displayed during the trumpeting display (Jones 1993a; Zubakin et al. 2010; Klenova et al. 2011a). This is a self-advertisement behavior performed by males on the surface of the colony during activity periods. Acoustic properties of the call are associated with body size and may signal individuality of the caller (Klenova et al. 2011a, b). Bill fluorescence may augment the self-advertisement function of this display. A bright bill may help a trumpeting male, in combination with the acoustic cue, to precisely mark its location on the colony surface. Crested auklets also give trumpeting calls under the colony surface. At some colonies with increased threat of avian predation, social activity is audible for long periods under the surface of the colony, and crested auklets spend less time on the surface (Douglas et al. 2018). A bill with enhanced luminescence could help mark a bird’s location visually in the reduced light of rock crevices. Wails et al. (2017) also suggested that fluorescence may advertise the crested auklet’s bill to prospective mates and conspecifics. Bill fluorescence may be important in the marine environment. HDD observed crested auklets flocks plunge diving into the ocean in synchronicity after evading the aerial pursuits of peregrine falcons (Falco peregrinus). It was a behavior that occurred nearly every day during the breeding season at Yukon Harbor, Big Koniuji I., AK. HDD noted the waters of the harbor were relatively transparent. The fluorescent bills could help with coordination of underwater movements by providing location and orientation cues for conspecifics. This might be important for evading predation threats from beneath the surface, e.g., river otters frequently predated crested auklets in Yukon Harbor. Although red photons of light and much of the UV radiation are attenuated close to the surface of the ocean; the UV-A band and blue-green light penetrate to greater depths than previously thought (Lee et al. 2013). Reef fish (Meadows et al. 2014) and mantis shrimp (Mazel et al. 2004) have fluorescent signals that depend upon excitation by longer wavelengths in the blue-green visible light spectrum, and our data suggest that blue light could also cause crested auklet bill plates to fluoresce in nature. If the bills fluoresce at depth it might help to mediate interactions with prey and improve foraging success. However, bright bills could also increase the risk of predation as some predatory fish consume crested auklets (Ulman et al. 2015).

Status signals are thought to communicate an individual’s ability to win contests over resources, but the underlying mechanisms are not well understood (Westneat and Diep 2013). The crest ornaments of crested auklets signal status, but are not directly employed as weapons in physical contest (Jones and Hunter 1999; Jones et al. 2000). Rictal plates, like crest ornaments, vary among individuals (Jones et al. 2000). The rictal plates of subadult crested auklets are smaller and often not as bright in color compared to adults (Jones 1993a; HDD pers. obs.). We suspect that the concentration of fluorescent pigments may be less in subadults. As we showed in this study, the highest fluorescence of rictal plates tended to correlate positively with the longest crest feather in breeding adults. HDD has observed that the rictal plates flare out from the corners of the mouth during agonistic displays. This may cause the bill to appear larger or wider at the base of the mandibles, where increased size may be perceived as increased bite force. Interestingly, nonbreeding adult crested auklets have larger rictal plates than breeding adult crested auklets (Jones et al. 2000). A larger rictal plate has a larger surface area to absorb and reflect light. Larger size and increased brightness command attention. Research in psychology has shown that increasing an object’s size biases the competition for attention in preference to the larger object, and likewise increasing an object’s brightness biases the competition for attention in preference to the brighter object (Proulx and Egeth 2008). Larger rictal plates present larger luminance patches, and this could be important in low light conditions such as rock crevices where nonbreeders are prospecting for future nest sites. Size of a rictal plate may be like a conventional signal. It may provide an advantage to a nonbreeder in social interactions, but perhaps at some risk of inviting social challenges. Among crested auklet males, allostatic load appears to be distributed disproportionately on lower status males (Douglas et al. 2009). Crest size is negatively correlated with baseline corticosterone in breeding adult males, and this may be related in part to vigorous challenges over status (Douglas et al. 2009). The same pattern is not evident in females (Douglas et al. 2009). The ways in which negotiations over status are incorporated into ornament phenotypes via social feedback and physiological regulation may differ between the sexes. Female crested auklets have greater variability in their crest ornaments (Jones et al. 2000). Females tend to signal with their crest ornaments in female-female interactions, and are less likely than males to escalate to physical combat (Jones and Hunter 1999). Female breeding adults also have more variability in rictal plate height (a measure of plate size) than do male breeding adults (Jones et al. 2000). Like the variability in crest size, the variability in the size of the breeding adult females’ rictal plates may relate to social signaling. Females contend for resources that improve fecundity, and evolution of female ornamentation may be more comprehensible in a model that incorporates sexual selection as a subset of social selection (Tobias et al. 2012). In terms of evolution of the trait, bill fluorescence could have conferred an advantage in social competition, and this could have helped to increase the frequency of the trait in populations. In animal societies, individuals may learn not only from direct conflict, but also from the conflicts they observe among conspecifics. Prominent ornaments such as large crest ornaments, large rictal plates, and fluorescent bill pigments could help individuals stand out in social groups. While size of the rictal plate may be a conventional signal like the size of feathers in the crest ornament, we think that bill fluorescence could be a condition-based signal. A condition-based signal such as bill fluorescence could increase the reliability of a signaling system by verifying that a signaler merits the status signal that it bears. Bill brightness and color may signal an individual’s ability to present pigments captured from prey, an indication of foraging ability. Euphausiids, the likely source of the crested auklet’s fluorescent pigments, are fast swimmers (Miyashita et al. 1996; De Robertis et al. 2003; Klevjer and Kaartvedt 2011) and can be difficult to capture, even for scientists with plankton nets. Because the quality of the prey delivered to chicks is crucial for chick survival (Gall et al. 2006; Sheffield Guy et al. 2009; Bond et al. 2012), it could be adaptive for crested auklets of both sexes to evaluate prospective mates based on the color and brightness of accessory bill plates.

Significance statement.

Preferences embedded in sensory systems may influence evolution of ornamental traits. Overlap in chemical properties of zooplankton and brilliant orange bills, suggest that crested auklets prefer a bill pigment patterned on a food preference. The pigment is unique in its genus, but occurs in two fluorescent phenotypes that may vary geographically, possibly due to environmental heterogeneity. The pigment resembles a class of compounds known as pterins but could be novel. Rictal plates are bill plates that flare out from the corners of the mouth, and they tend to be brighter in crested auklets with larger feather crests. Birds with larger crests are more attractive and dominant. Brighter bills may advertise foraging success and verify status. Bill fluorescence is attractive to crested auklets. Crested auklets approached decoys with fluorescent bills at a higher frequency.