Gloger's Rule, Feather-Degrading Bacteria, and Color Variation Among Song Sparrows

Abstract

Feathers tend to be darkly colored in habitats where relative humidity is high and pale where it is low. We suggest that this correlation, known as Gloger's rule, results, in part, from selection for dark feathers that are more resistant than light feathers to bacterial degradation, which is a severe problem in humid habitats where bacteria thrive and a lesser problem in arid habitats. In May and June 2000–2002 we sampled feather-degrading bacteria (Bacillus licheniformis) from the plumage of Song Sparrows (Melospiza melodia) in southeastern Arizona and northwestern Washington. Under standardized laboratory conditions, feather-degrading bacteria from the plumage of sparrows in the humid Northwest degraded feathers more rapidly and more completely than feather-degrading bacteria from sparrows of the arid Southwest. The differences in feather-degrading bacteria and in relative humidity produce a difference in the intensity of selection, which in turn produces the difference in color described in Gloger's rule.

La Regla de Gloger, Bacterias Degradantes de Plumas y Variación de Color en Melospiza melodia

Resumen. Las plumas tienden a ser de tonos obscuros en hábitats donde la humedad relativa es alta y más pálidas en hábitats donde la humedad relativa es baja. Esta correlación, conocida como la regla de Gloger, se aplica a muchas especies de aves a través del mundo. Sugerimos que la regla de Gloger es, en parte, un producto evolutivo de la selección por plumas obscuras, que son más resistentes a la degradación bacteriana que las plumas claras. La degradación bacteriana es un problema severo en hábitats húmedos donde prosperan las bacterias y un problema menor en hábitats áridos. En mayo y junio de 2000 a 2002 tomamos muestras de bacterias degradantes de plumas (Bacillus licheniformis) del Melospiza melodia fallax, que tiene plumaje pálido y reside en la parte sureste del estado de Arizona, y comparamos la incidencia y actividad de estas bacterias con las de aquellas encontradas en el plumaje obscuro de M. m. morphna, que reside de los bosques húmedos del noroeste del estado de Washington. Sin embargo, bajo condiciones estandarizadas de laboratorio, las bacterias obtenidas de M. m. morphna, degradaron las plumas más rápida y completamente que las bacterias de M. m. fallax. Las diferencias sugieren que las plumas obscuras de M. m. morphna del noroeste húmedo están sujetas a selección más intensa para resistir la degradación bacteriana que las plumas claras del gorrión del suroeste árido. La diferencia en humedad relativa produce una diferencia en la intensidad de selección, que a su vez produce la diferencia en color descrita en la regla de Gloger.

Feather degradation by Bacillus licheniformis (Williams et al. 1990), its occurrence on the feathers of wild birds (Burtt and Ichida 1999, Muza et al. 2000), and the increased resistance of melanic feathers to bacterial degradation (Goldstein et al. 2004), raise the possibility that plumage color may be an evolutionary response to the presence of feather-degrading bacteria. In 1833 Gloger noted that birds in climates with high relative humidity were darker than conspecifics in climates with low relative humidity. Zink and Remsen (1986) found that among birds of the United States and Canada individuals in 94% of the species were darker in areas of high relative humidity than in areas of low relative humidity. Because bacteria thrive in humid habitats and because dark, melanic feathers resist bacterial degradation better than light feathers that lack melanin, we hypothesized that the geographic correlation of dark coloration with high relative humidity may be a response to selection for feathers that resist bacterial degradation.

Our test species was the Song Sparrow (Melospiza melodia), which is widely distributed and abundant (Arcese et al. 2002), easily captured, and known to harbor feather-degrading bacteria in its plumage (Burtt and Ichida 1999). We sampled individuals from a pale subspecies of the arid southwestern United States and from a dark subspecies of the humid northwestern coast. If feather-degrading bacilli have been a factor in selection for the observed color difference, then B. licheniformis should occur on a higher proportion of dark, northwestern Song Sparrows than on their pale southwestern relatives.

Methods

Bacterial Sampling of Song Sparrows

We quantified color variation and sampled plumage microorganisms in two populations of Song Sparrows, pale individuals of the subspecies M. m. fallax of southeastern Arizona and dark individuals of the subspecies M. m. morphna of western Washington (Arcese et al. 2002). In Arizona we captured sparrows in mist nets set along Sonoita Creek (31°31′N, 110°46′W) near Patagonia and along the San Pedro River (31°33′N, 110°08′W) near Sierra Vista. In Washington we captured sparrows in 3–5-year-old clearcuts near Forks (48°N, 124°W) and in gardens, fields and woodland edges in Clinton (47°59′N, 122°23′W).

We disinfected our hands with quaternary disinfectant and allowed them to air dry before removing a bird from the net. Each bird's back was rubbed with a sterile, Dacron-tipped applicator (Puritan, St. Louis, Missouri) wetted with sterile saline (0.85% NaCl). After the sample was collected the applicator was replaced in its sterile envelope. The tail and venter were sampled with the same procedure. The plumage of the back, tail, and venter was then pressed lightly on three separate trypticase soy agar (TSA; Acumedia, Troy, Michigan) plates. TSA is a rich medium on which many different microorganisms grow readily. Exposed applicators, in their sterile envelopes, and exposed plates wrapped in clean plastic sandwich bags were placed in a Styrofoam cooler with ice packs, where they remained until we returned to the laboratory 4–6 hr later. After the microbial samples were collected the birds were banded with a U.S. Fish and Wildlife Service band and released. Individuals caught more than once on the same day were sampled on the initial capture only.

In the laboratory we removed the applicators from their envelopes, placed each in an individually labeled, sterile tube of modified (pH 7.5, 7.5% NaCl) nutrient broth (BBL, Cockeysville, Maryland), and incubated the tubes for 7 days at 50°C. B. licheniformis forms spores under adverse conditions, such as occurred on the Dacron-tipped applicators, but emerges as a vegetative cell when conditions improve, such as occurred in the broth (Black 1996).

The slightly basic, moderately saline broth and high temperature favor the growth of B. licheniformis and inhibit the growth of most other microorganisms (Burtt and Ichida 1999). If the broth remained clear, the sample lacked B. licheniformis and was discarded. If the broth became cloudy, the bacilli were cultured by streaking a loopful of the medium across a sterile TSA plate and incubating at 36°C. After 48 hr we checked the plate to be sure that it contained only B. licheniformis, which forms wrinkled, cone-shaped colonies. If the culture was pure, we removed a representative colony and transferred it to a TSA slant. The slant was incubated for 48 hr at 36°C, then was removed and stored at 4°C until further testing. The combination of selective media, high incubation temperature, and visual identification of isolated colonies grown from two sequential transfers of bacteria from the media has been shown to successfully isolate B. licheniformis (Burtt and Ichida 1999).

Field samples on TSA plates were incubated at 36°C. After 48 hr all plates were removed from the incubators and B. licheniformis colonies were identified and counted. We used a sterile loop to transfer colonies of B. licheniformis to individually labeled, sterile tubes of modified nutrient broth and processed them as described above for the applicators. The procedure screened out strains of B. licheniformis that were neither thermophilic nor tolerant of moderately saline and mildly alkaline conditions. Such strains do not degrade feathers (Burtt and Ichida 1999).

All isolates were labeled to indicate not only the bird and area of the bird from which they were collected, but also the applicator or plate on which they were initially cultured.

Measuring Intraspecific Color Variation

Before release the colors of each Song Sparrow were matched by eye to color samples in Kornerup and Wanscher (1989). Matching was done under clear sky at least 1 hr after dawn and no later than 1 hr before sunset. All color matching was done by EHB, who matched the dark stripes and background color of the greater wing coverts near the “wrist” of the wing, which could be held open against color swatches in the handbook. Then the bird was held on its back and the brown stripes of the venter and their background color were matched to the color swatches. Each sparrow was released after its colors were matched. Fifty-three individuals of M. m. fallax and 98 individuals of M. m. morphna were measured.

Luminance is a measure of the amount of light that is reflected back to the observer from the colored patch. White reflects the most light and black the least. We assigned luminance scores to markings by recording the column of the matching color swatch in Kornerup and Wanscher (1989; white = 1, dark gray = 6). Stripes that were blacker than dark gray were assigned a score of 7. Thus the luminance of the stripe or its background varied between 1 and 7.

Soil Samples

We collected soil samples at each site. We used a sterile tablespoon to dig a sample of soil and place it in a sterile plastic bag, which was then sealed. The bags were kept on ice until returned to the laboratory where a sterile, Dacron-tipped applicator was rolled in the soil, the excess particles shaken off, and the applicator placed in an individually labeled, sterile tube of modified (pH 7.5, 7.5% NaCl) nutrient broth. We incubated tubes for 7 days at 50°C. If present, B. licheniformis was identified and isolated as described above.

Measuring Bacterial Activity

Plumage may be darker in response to the probability of colonization by feather-degrading bacilli. Alternatively the plumage of Song Sparrows may be darker in the humid Northwest in response to bacteria that degrade feathers more rapidly or more completely than bacilli that occur in the arid Southwest. To test this possibility we inoculated test tubes of feather medium (Williams et al. 1990) with each of the B. licheniformis isolates from Arizona and Washington sparrows and assessed feather condition (Table 1) daily for 10 days.

Table 1.

Qualitative scale used to estimate the condition of chicken feathers exposed to feather-degrading Bacillus licheniformis

We used secondary remiges from single comb white leghorn chickens (Gallus domesticus) to test bacterial isolates for feather-degrading activity. We removed and discarded the distal 1 cm from the feather; the next 4 cm were removed, cut in half across the rachis, and the two halves placed in a test tube. We added 10 mL of feather medium (Williams et al. 1990) to each tube. All tubes were sterilized at 121°C and 7.7 kg pressure for 15 min. We removed the bacterial isolates from storage at 4°C and inoculated fresh TSA cultures with the isolate to be tested. After 24 hr a loopful of bacteria was removed and suspended in sterile saline. The turbidity of the saline-bacterial suspension was adjusted to 0.5 MacFarland standard, which corresponds to ∼150 000 cells mL⁻¹. Two drops of this suspension (∼0.1 mL) were placed in a test tube of medium containing the two 2-cm pieces of feather. A replicate was prepared from the same suspension. Tubes were placed in a rack in a shaker incubator at 50°C and rotated at 125 rpm.

We checked all tubes daily for degradation of the feather pieces. Racks were removed from the incubator one at a time. Each test tube was removed from the rack and held against a vortexer until a whirlpool extended from the surface of the solution to the bottom of the test tube. This ensured that all materials were thoroughly mixed and that each sample achieved the same level of mixing every day. The contents were evaluated visually using the criteria in Table 1.

All feathers were condition 6 when placed in the medium, but after sterilization in the autoclave, many of the feathers were rated condition 5. Occasionally a feather was rated condition 4 after sterilization and before inoculation with B. licheniformis, but none were rated lower than 4. Therefore we considered the feather to be degraded when both pieces in the test tube reached condition 3 or lower (Table 1).

Statistical Analyses

Most of our data were not normally distributed; therefore most of our analyses were nonparametric (Minitab 2000). We used the Mann-Whitney U-test and its normal approximation (Zar 1999) to compare the luminance values of the stripe and background colors of the back and venter of Song Sparrows in southern Arizona to those of Song Sparrows in northwestern Washington. We used the Mann-Whitney U-test again to compare the condition of feathers (Table 1) exposed to different isolates of B. licheniformis. We used chi-square with a continuity correction (Zar 1999) to test for independence of the proportion of Song Sparrows with B. licheniformis in their plumage in Arizona and Washington. We also used the chi-square, again with the continuity correction, to compare the proportion of feathers stripped of barbs by bacilli isolated from Song Sparrows in Arizona and Washington. We used the t-test to compare the number of days required to degrade feathers by B. licheniformis isolated from Song Sparrows in Arizona and Washington. Luminance and feather condition values are given as modes. Days to degrade are given as means ± SE.

Results

Intraspecific Color Variation

The luminance values of the stripes and background colors of Song Sparrows clustered tightly about modal values characteristic of that marking for that subspecies. The venter of M. m. fallax was white (luminance 1) in all 53 sparrows measured, whereas the venter of M. m. morphna was significantly grayer (luminance 2, range 1–5; U = 2406, P < 0.001; Fig. 1). The breast stripes were dark in M. m. fallax (luminance 5, range 4–6), but significantly darker in M. m. morphna (luminance 6, range 5–6; U = 2634, P < 0.001; Fig. 1). The dorsal background color as measured on the greater wing coverts was brown (luminance 5, range 2–5) in M. m. fallax, but significantly darker (luminance 6, range 5–6; U = 1578, P < 0.001; Fig. 1) in M. m. morphna. Finally, the dorsal stripes of M. m. fallax had a modal luminance of 6 (range 5–6), but those of M. m. morphna were still darker, almost black (luminance 7, range 6–7; U = 1762, P < 0.001; Fig. 1).

Figure 1.

Comparison of the modal luminance of the dorsal and ventral stripes and backgrounds of Song Sparrows in arid Arizona (Melospiza melodia fallax; n = 53) and humid Washington (M. m. morphna; n = 98). Luminance ranges from 1 (white) to 7 (almost black). For each marking the difference between the subspecies was significant at P < 0.001, with fallax always lighter than morphna

Soil Samples

Soils at all of our sites contained feather-degrading B. licheniformis. The bacilli were active, vegetative cells under humid to damp conditions (Ichida et al. 2001), and formed inactive endospores under arid conditions (Black 1996).

Occurrence of Feather-Degrading Bacilli in Song Sparrows

Feather-degrading bacilli occurred on 33 of 142 (23%) Song Sparrows (M. m. morphna) from the humid Northwest and 14 of 77 (18%) from the arid Southwest (M. m. fallax), a nonsignificant difference (χ²₁ = 0.6, P = 0.48). Despite the small difference, individuals of M. m. morphna were more likely to have feather-degrading bacilli in their plumage than individuals of M. m. fallax in all 3 years studied. The difference may be small, but our limited data suggest that it may be consistent.

Bacterial Activity

Among white chicken feathers that degraded, those exposed to bacilli from the plumage of Washington Song Sparrows degraded more rapidly (5.1 ± 0.4 days) than those exposed to bacilli from the plumage of Arizona Song Sparrows (6.8 ± 0.6 days; t₃₁ = 2.5, P = 0.02). At the end of the 10-day experiment, the modal condition of feathers degraded by bacilli from dark sparrows was 1 (range 0–3), whereas the modal condition of feathers degraded by bacilli from pale sparrows was 2 (range 0–3). The difference was significant (U = 742, P = 0.02). Significantly more bacilli from dark sparrows (χ²₁ = 5.2, P = 0.03; Fig. 2) reduced white chicken feathers to condition 1 or 2 (Table 1) than bacilli isolated from pale sparrows.

Figure 2.

Proportion of bacilli isolated from the plumage of Song Sparrows in arid Arizona (M. m. fallax; n = 18) and humid Washington (M. m. morphna; n = 47) that stripped all barbs from the rachis of chicken feathers within 10 days of inoculation

Discussion

We confirmed quantitatively that all markings of Song Sparrows from the humid coastal forests of northwestern Washington are darker than the comparable markings of Song Sparrows from the arid Southwest, a pattern described previously by Zink and Remsen (1986), Aldrich and James (1991), James (1991), and Arcese et al. (2002). We found feather-degrading B. licheniformis in soils and on Song Sparrows of both regions, but feather-degrading bacilli isolated from the plumage of dark Song Sparrows degraded feathers more quickly and more completely than feather-degrading bacilli from the plumage of pale Song Sparrows. This result suggests that feathers of Song Sparrows in the humid Northwest are subjected to strong selection pressure to develop resistance to bacterial feather degradation.

The darker color indicates more melanin in the feathers of morphna, which inhabits the humid forests of the Northwest coast, than in the feathers of fallax, which inhabits the arid Southwest. The presence of melanin increases the abrasion resistance and thickness of feather keratin (Burtt 1979, 1986, Voitkevich 1966). The inclusion of granular melanin also increases the hardness of keratin (Bonser 1995, Bonser and Witter 1993), which increases its resistance to abrasion, independent of thickness. Recent evidence (Goldstein et al. 2004) indicates that melanin also increases the resistance of feather keratin to bacterial degradation. Regardless of whether the increased resistance is due to thicker or harder keratin, we would expect Song Sparrows in a climate with high relative humidity to evolve melanic plumage as a defense against bacterial degradation of their feathers. Thus selection would favor dark plumage in environments with high relative humidity, which is the pattern observed by Gloger (1833). That this is a dynamic, evolutionary relationship was shown by Johnston and Selander (1971), who documented the darkening of House Sparrows (Passer domesticus) as they spread into regions of the United States and Canada with high relative humidity during the last 150 years.

The plumage of birds hosts an entire ecosystem of microorganisms, among which are other feather-degrading species, for example the bacterium Streptomyces pactum (Bockle et al. 1995) and the fungi Chrysosporium spp. (Pugh and Evans 1970) and Fusarium sporotrichioides (H. M. Costello, pers. comm.). The combined action of several feather-degrading species may have a greater effect on feathers than we have demonstrated by focusing on B. licheniformis. However, the combined action may be subtle due to competition among microorganisms that are able to chemically inhibit or kill each other (Burtt 1999). Additionally, the bird may be able to limit feather degradion directly by exposing feathers to ultraviolet irradiance or heat (Moyer and Wagenbach 1995) during sunning, or indirectly by preening, dust bathing, or other maintenance behavior that could encourage the growth of microbes that excrete antibacterial chemicals (e.g., Pencillium spp.).

Gloger's rule describes color variation in 94% of the North American species that encounter substantial climatological differences in relative humidity (Zink and Remsen 1986). This is a far more robust relationship than that described by Bergmann's or Allen's rules (Zink and Remsen 1986). We suggest that the strength of the relationship may be the product of several important selection pressures (Table 2), all of which favor dark color in a humid environment. None of the pressures listed in Table 2 are mutually exclusive, and resistance to bacterial degradation is simply another pressure that selects for dark feathers. We believe the strength of the relationship between relative humidity and color rests on the number of selection pressures that act in concert to favor the evolution of dark plumage in habitats with high relative humidity.

Table 2.

Selection pressures that may contribute to the evolution of dark coloration among species of birds living in climates with high relative humidity

Acknowledgments

We thank Steve and Ruth Russell for invaluable advice on where to find Song Sparrows and Linda Kennedy and Bill Brannan at the Audubon Research Ranch. We thank the Nature Conservancy for allowing us to work on their Sonoita Creek preserve. We also thank Matt Brown who shared his friendship and vast knowledge of southeastern Arizona birds and Jack Whetstone and the Bureau of Land Management for permission to work in the San Pedro Riparian Corridor. In Washington we are indebted to Dan Varland and Rayonier Timber Co. for permission to net sparrows on their land, and to the Olympic Natural Resource Center for excellent housing and laboratory facilities. On Whidbey Island we are indebted to Sievert and Brigette Rohwer and Mary and Tom Fisher for the use of their properties, which abound in Song Sparrows. We are especially grateful to Umut Aypar, Pam Y. Burtt, Benjamin T. Lawrence, Jason C. McCarthy, Trent B. Marburger, Nga P. Nguyen, Nadinath N. M. Nillegoda, Michael L. Parrish, and James V. Whitaker for help with fieldwork and laboratory experiments, Scott R. Linder for advice on statistics, and José Manuel Galindo who prepared the Spanish abstract. We thank Michael A. Patten and an anonymous referee for their helpful comments on the manuscript. Our work was supported by the National Science Foundation (BIR-998805).

Author notes