• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

A Phylogenetic Perspective on the Evolution of Chromatic Ultraviolet Plumage Coloration in Grackles and Allies (Icteridae)

Posted on: Wednesday, 15 February 2006, 06:00 CST

By Eaton, Muir D

ABSTRACT.-

Plumage traits have been studied intensely for more than a century, especially bright and exaggerated plumage. A large body of evidence across a range of avian taxa supports sexual selection as a major evolutionary force acting on plumage colors. The discovery of ultraviolet (UV) coloration in avian plumage resulted in the extension of sexual selection hypotheses to explain the evolution of potential UV plumage traits. However, there have been no comparative evolutionary studies elucidating the origin of UV signals in birds. Here, I used a comparative phylogenetic approach to investigate the evolution of chromatic UV plumage colors in the grackles-and-allies clade of the New World blackbirds (Icteridae). On the basis of reflectance data collected from museum study skins, I have determined that UV plumage signals have evolved multiple times from an ancestral condition that lacked UV plumage signals, with very few unambiguous reversals. Although UV plumage has evolved in both males and females, there have been significantly more evolutionary changes in male UV plumage characters. Concentrated changes tests and correlations of independent contrasts reveal evidence for sexual selection of some male UV plumage characters, as well as an increase in UV plumage coloration for species found in open habitats. These results support the use of objective assessments of avian colors (i.e. spectrophotometry) to properly interpret patterns of plumage evolution generally, and they suggest the need for behavioral studies on the function of chromatic UV signals in several blackbird species. Received 11 September 2004, accepted 2 July 2005.

Key words: comparative method, New World Blackbirds, ultraviolet (UV) plumage.

Una Perspectiva Filogentica sobre la Evolucin de la Coloracin Ultravioleta en los Changes y Chamones (Icteridae)

RESUMEN.-Los caracteres del plumaje, en especial los colores brillantes y los rasgos exagerados, han sido estudiados intensamente durante ms de un siglo. Existe un cmulo amplio de evidencia en distintos taxa de aves que apoya la hiptesis de que la seleccin sexual es una importante fuerza evolutiva que acta sobre la coloracin del plumaje. El descubrimiento de la coloracin ultravioleta (UV) en el plumaje de las aves conllev a la extension de la hiptesis de seleccin sexual para explicar la evolucin de los caracteres UV del plumaje. Sin embargo, no existen estudios evolutivos comparativos que hayan elucidado el origen de las seales UV en las aves. En este estudio, emple un enfoque filogentico comparativo para investigar la evolucin de la coloracin UV del plumaje en el clado de los changos y chamones, que forma parte de la familia Icteridae. Con base en datos de reflectancia obtenidos a partir de pieles de museo, determin que las seales UV han evolucionado repetidamente a partir de una condicin ancestral que careca de seales UV, con muy pocas reversiones no ambiguas. Aunque la coloracin UV del plumaje ha evolucionado tanto en los machos como en las hembras, han existido significativamente ms cambios evolutivos en los caracteres UV de los machos. Pruebas de cambios concentrados y correlaciones de contrastes independientes proveen evidencia de la existencia de seleccin sexual sobre algunos de los caracteres UV de los machos, y de un incremento en la coloracin UV en las especies que habitan ambientes abiertos. Estos resultados apoyan el uso de mediciones objetivas de los colores de las aves (i.e. espectrofotometra) para interpretar los patrones de evolucin del plumaje en general, y sugieren la necesidad de estudios de comportamiento que permitan elucidar la funcin de las seales cromticas UV en varias especies de ictridos.

THE THEORY OF sexual selection (Darwin 1871) has played an important role in explaining why birds have evolved particular plumage patterns and colors (e.g. Baker and Parker 1979, Butcher and Rohwer 1989, Johnson 1991, M011er and Birkhead 1994, Gray 1996, Price and Birch 1996, Owens and Hartley 1998, Kose et al. 1999, Badyaev and Hill 2000, Badyaev et al. 2002, McNaught and Owens 2002, Safran and McGraw 2004). The vast evidence supporting sexual selection of plumage traits in general (e.g. see references in Andersson 1994), combined with avian abilities to see ultraviolet (UV) wavelengths (Kreithen and Eisner 1978, Chen et al. 1984, Shi et al. 2001, Odeen and Hastad 2003) and the prevalence of UV plumage reflectance among birds (Burkhardt 1989, Finger and Burkhardt 1994, Eaton and Lanyon 2003, Hausmann et al. 2003), has turned attention to investigations of sexual selection operating on UV signals (e.g. Bennett et al. 1997, Hunt et al. 1998, Siitari et al. 2002, Doucet and Montgomerie 2003). Although growing evidence supports the maintenance of UV plumage signals via sexual selection, only comparative studies can reveal the evolutionary origins of such traits.

Several comparative analyses found associations between plumage characteristics and both mating system and habitat, respectively, which suggests causal evolutionary relationships (e.g. Marchetti 1993, Figuerola and Green 2000, Dunn et al. 2001, McNaught and Owens 2002, Gomez and Thery 2004). Thus, evolution of UV plumage signals may be subject to similar influences from these ecological variables and can be tested in a comparative context. For example, if males of polygynous species are under more intense sexual selection (Webster 1992, Andersson 1994), associated changes between polygyny and male UV plumage is evidence that sexual selection may be driving the evolution of UV plumage traits. Different habitats may select for certain plumage colors, because of the particular light environment available for signaling (Endler 1993). For example, UV plumage signals should be found in woodland or forest shade habitats because UV contrasts most with ambient light, whereas open habitats should select for darker plumage overall (i.e. reduced UV; Endler 1993, Marchetti 1993). Correlation analyses of these characters in the phylogenetic framework of grackles and allies (Icteridae) can test these evolutionary predictions.

The grackles and allies of the New World blackbirds provide a model group for studying UV plumage evolution in a comparative phylogenetic context. The species form a monophyletic group (Lanyon and Omland 1999), for which a well-supported, well-resolved specieslevel phylogeny exists (Johnson and Lanyon 1999). A well- supported phylogeny is essential for reliably reconstructing character evolution (Harvey and Pagel 1991). Species within this clade are characterized by a wide range of habitat types and mating systems (Orians 1985, Jaramillo and Burke 1999), as well as variation in plumage coloration. These differences in ecology and coloration between species provide the necessary clade-wide variation for investigating associated evolutionary changes among these characters.

Evolutionary changes in mating system and habitat may not be mutually exclusive as sources of selection on UV plumage signals in grackles and allies. The potential for both sexual selection and the environment to influence plumage evolution has previously been shown for this group (Searcy et al. 1999, Johnson and Lanyon 2000). However, little is known about UV coloration in grackles and allies, and even less about the evolutionary history or behavioral ecology of such plumage characters. The main goals here are to (1) reconstruct the historical changes in UV plumage signals in grackles and allies by mapping UV plumage characters of both males and females onto the molecular phylogeny and (2) compare patterns of plumage evolution for each sex with historical changes in mating system and habitat use to investigate the selective forces of these ecological variables on the evolution of UV coloration.

MATERIALS AND METHODS

Phylogenetic reconstruction.-I obtained molecular sequence data for 36 species of grackles and allies (Icteridae; nomenclature follows the Howard and Moore checklist [Dickinson 2003], as modified by Lowther et al. 2004), plus six outgroup taxa (Amblycercus holosericeus, Cacicus cela, Gymnostinops montezuma, Icterus galbula, Psarocolius angustifrons, and Sturnella neglecta), from S. M. Lanyon and F. K. Barker (see Johnson and Lanyon 1999 for extraction and sequencing methods). Outgroup taxa were chosen from among the other four major New World blackbird lineages (Lanyon and Omland 1999). I re-analyzed molecular data from Johnson and Lanyon (1999) with the addition of Hypopyrrhus pyrohypogaster (Cadena et al. 2004) and Agelaius assimilis to obtain a robust phylogeny for subsequent character mapping. Phylogenetic analyses were performed using PAUP*, version 4.0 (Swofford 2002) and MRBAYES (Huelsenbeck and Ronquist 2001) on 2,184 base pairs of mitochondrial DNA (mtDNA) from the ND2 (NADH dehydrogenase subunit 2) and cytochrome-b gene regions. Johnson and Lanyon (1999) determined that these two gene regions did not produce significantly different phylogenies when analyzed independently; thus, both gene regions were combined for phylogenetic analyses here. I constructed a maximumlikelihood (ML) tree with PAUP*, under a general time-reversible model of sequence evolution, with rates at variable sites following a gamma distribution \and with invariant sites (GTR + G + I; gamma-shape parameter = 1.3816; proportion of invariant sites = 0.5297), as determined by the likelihood-ratio tests in MODELTEST (Posada and Crandall 1998). I constructed a Bayesian tree with MRBAYES (GTR + G + I model of sequence evolution, as determined by MODELTEST) run for 1 10^sup 6^ generations (4 chains; T = 0.2) with Markovchain Monte- Carlo (MCMC) sampling every 100 generations to yield 10,000 trees. The first 100 of these sampled trees were discarded as "burn-in" (i.e. likelihood values reached stationarity) before construction of the Bayesian consensus tree. Both trees were rooted by the designated outgroup taxa. Four additional taxa (Dives atroviolacea, D. dives, Curaeus forbesi, and Macroagelaius subalaris), for which no molecular data were available, were then added to the topology for purposes of character reconstruction in MACCLADE, and all were placed sister to congeners. For the Dives group, rearrangement of the trichotomy produced identical subsequent character state reconstructions.

The primary goal here was to examine evolutionary patterns of UV coloration in plumage characters across the grackles-and-allies clade. Therefore, I did not include plumage characters when inferring phylogenic relationships, so as to avoid circularity (de Queiroz 1996, Omland and Lanyon 2000). Several studies have cautioned against the use of plumage characters for reconstructing phylogenies (Hackett and Rosenberg 1990, Omland 1997, Burns 1998). For example, work on oriole plumage (Icteridae) has shown that plumage patterns and colors are extremely labile and have high levels of homoplasy (Omland and Lanyon 2000). Ultraviolet coloration is the result of the microstructural arrangement of feather keratin (Prum et al. 1999, 2003), yet its presence as defined here potentially results from different evolutionary pathways (e.g. in association with carotenoid pigments or melanin pigments or differing microstructural arrays). Mapping nonhomologous UV coloration onto an independently derived phylogeny can still be useful for uncovering correlations with ecological variables, thus testing alternative hypotheses of selection for UV plumage signals regardless of their proximate cause.

To reduce potential bias in choosing which plumage characters to emphasize, I measured the entire external appearance of male and female conspecifics (Omland and Lanyon 2000). I collected reflectance data on museum study skins, using the ornithology collections at the Field Museum of Natural History (Chicago, Illinois) and the American Museum of Natural History (New York, New York), from all feather patches (later referred to as plumage characters) that varied in coloration for each of three male and three female specimens from all 40 species of the grackles-and- allies clade of the New World blackbirds. (Deviations from these sample sizes are shown in Tables 1 and 2.) A feather patch was defined as an area of similar continuous human-visible coloration greater than ~4 mm^sup 2^, and one reflectance spectrum was saved for each of 22 feather patches from each individual specimen (i.e. 22 spectra per specimen). Ultraviolet visual sensitivity in birds suggests the existence of UV feather patches that are not visually discernible to humans; thus, these would be missed under my definition of a feather patch. However, to date, there is no evidence of hidden UV markings that solely define feather patches (Burkhardt 1989, Finger and Burkhardt 1994, Andersson and Amundsen 1997, Eaton and Lanyon 2003). Use of museum specimens for studying coloration may be biased, because of fading or differences between live birds and research skins. McNett and Marchetti (2005) found that the UV component of carotenoid colors was different in live birds than in museum study skins for several warbler species (Parulidae), but this study did not include primarily UV-reflecting colors, which is the main concern here. In the present study, noncarotenoid colors represent most of the plumage patches found in the grackles and allies; thus, any biases introduced by potential differences of carotenoid colors with live birds will have a minimal effect on the overall conclusions. Furthermore, several icterid colors showed no effect of specimen age on the UV portion of coloration, from study skins spanning the 188Os to freshly collected specimens (M. D. Eaton unpubl. data). Fading in museum specimens with respect to UV coloration has been addressed in several other studies, all of which concluded that specimen age had no effect on UV coloration (Andersson 1996, Endler and Thery 1996, McNaught and Owens 2002, Parker et al. 2003). Nonetheless, I chose specimens with the freshest-looking breeding plumage; and within each species, I chose specimens of approximately the same age (i.e. year of collection) and same general locality when possible.

Spectral reflectance data were collected with an S-2000 spectrometer (Ocean Optics, Dunedin, Florida) equipped with an R200- 7UV/VIS reflectance probe (fiber diameter = 200 m) and a PX-2 pulsed xenon light source. Data collection was calibrated against a Spectralon white reflectance standard, with the following settings: ms = 100, average = 10. (These settings designate the pulse rate of the light source, and number of scans averaged per spectrum saved, respectively.) The reflectance probe was housed in a black rubber tube that blocked ambient light, kept the distance from the probe to the feather surface constant (~2 mm), and achieved a 90 measurement angle in relation to the feather surface. Area of measurement for each scan was ~4 mm^sup 2^. The spectrometer was recalibrated after all measurements were taken for each individual specimen. Raw reflectance data for every measurement were averaged to yield a percentage of light reflected every 10 nm, between 300 and 700 nm, using language written for the SAS statistical package. These processed data were then averaged within species for each feather patch, separately for each sex.

Character coding and comparative analyses, -I scored each feather patch on males and females for the presence and absence of both peak reflectance in the UV range of the spectrum ("PeakUV") and maximum reflectance in the UV ("MaxUV"). Both PeakUV and MaxUV were chosen to represent chromatic UV coloration, and thus potential color signals used in communication and subject to selection (Andersson 1996,1999). Chromatic color is typified by sharp changes (peaks and troughs) in the reflectance spectrum of an object and has been shown to be important in discriminating object color under variable illumination (Osorio et al. 1999). PeakUV was defined as present if the total UV reflectance (summed percentage of reflectance from 320 to 400 nm) was greater than the total blue reflectance (summed percentage of reflectance from 401 to 480 nm) plus 5%. Thus, PeakUV was scored conservatively because of the additional 5% required for its presence. Wavelength ranges used to compare UV reflectance to blue reflectance represent equal segments of the light spectrum that roughly correspond to the UV- and violet-sensitive cones of the passerine retina (review by Cuthill et al. 2000). MaxUV was defined as present if the maximum value of percentage of light reflectance across the entire visual range (300-700 nm) occurred in the UV portion of the spectrum. Thus, PeakUV and MaxUV represent discrete binary character states for each plumage character across all 40 study taxa. I reconstructed evolutionary changes in UV coloration separately for males and females using MACCLADE (Maddison and Maddison 2000), with unordered parsimony over the rooted ingroup phylogeny for the grackles and allies. Transformations were polarized using the root indicated by the molecular phylogeny. I calculated ensemble retention indices (RI; Farris 1989) and consistency indices (CI; Kluge and Farris 1969) for the whole tree, excluding uninformative characters (Maddison and Maddison 2000), for both male and female plumage, independently. I also calculated RIs and CIs for each individual plumage character for both males and females, independently. Both indices measure the amount of homoplasy in the character data set; however, the RI has an advantage in comparative studies, because it indicates the observed number of steps in relation to the maximum number of steps. (Consistency index values are reported here to allow comparison with other studies that do not include RI values.) A value of 1.00 for either index represents perfect congruence with the phylogeny, whereas a value of 0.00 represents maximum homoplasy and a lack of fit between the data and the tree.

TABLE 1. Matrix of male sample sizes (n), UV plumage coloration characters, habital, and mating system for grackle and allied species (Icteridae). Plumage character descriptions are listed in Table 3. Character state definitions: PeakUV: 1 = present , 0 = absent; MaxUV: 1 = present, 0 = absent; habitat: 1 = woodland and forest shade, 0 = open (Endler 1990); mating system: 1 = polygynous, 0 = monogamous. Lack of data is indicated by a question mark(?). See text for definitions of "PeakUV" and "MaxUv."

TABLE 1. Matrix of male sample sizes (n), UV plumage coloration characters, habital, and mating system for grackle and allied species (Icteridae). Plumage character descriptions are listed in Table 3. Character state definitions: PeakUV: 1 = present , 0 = absent; MaxUV: 1 = present, 0 = absent; habitat: 1 = woodland and forest shade, 0 = open (Endler 1990); mating system: 1 = polygynous, 0 = monogamous. Lack of data is indicated by a question mark(?). See text for definitions of "PeakUV" and "MaxUv."

Each species was scored for predominant habitat use and mating system on the basis of descriptions and data from Jaramillo and Burke (1999) and Orians (1985). Habitat was coded as one of three states, corresponding approximately to the open, f\orest-shade, and woodland-shade light environments as defined by Endler (1993). The concentrated changes test and regression of independent contrasts require binary character states; thus, woodland-shade and forest- shade were combined as one character state for these analyses. Mating system was coded as monogamous or polygynous. If different populations of a species were described as both monogamous and polygynous, that species was coded as polygynous for correlation analyses. I reconstructed evolutionary changes in habitat use and mating system using MACCLADE (Maddison and Maddison 2000), with unordered parsimony over the rooted ingroup phylogeny for the grackles and allies.

TABLE 2. Matrix of female sample sizes (n), UV plumage coloration characters, habitat, and mating system for grackle and allied species (Icteridae). Plumage character descriptions are listed in Table 3. Character state definitions: PeakUV: 1 = present, 0 = absent; MaxUV: 1 = present, 0 = absent; habitat: 1 = woodland and forest shade, 0 = open (Endler 1990); mating system: 1 = polygynous, 0 = monogamous. Lack of data is indicated by a question mark (?). See text for definition of "PeakUV" and MaxUV."

TABLE 2. Matrix of female sample sizes (n), UV plumage coloration characters, habitat, and mating system for grackle and allied species (Icteridae). Plumage character descriptions are listed in Table 3. Character state definitions: PeakUV: 1 = present, 0 = absent; MaxUV: 1 = present, 0 = absent; habitat: 1 = woodland and forest shade, 0 = open (Endler 1990); mating system: 1 = polygynous, 0 = monogamous. Lack of data is indicated by a question mark (?). See text for definition of "PeakUV" and MaxUV."

I tested whether evolutionary changes in chromatic UV coloration (i.e. PeakUV and MaxUV) were consistently associated with predominantly open habitat use or non-monogamous mating system (i.e. strength of sexual selection) on the molecular tree using the concentrated changes test in MACCLADE (Maddison 1990). Ambiguous reconstructions of all character states were resolved with both ACCTRAN and DELTRAN options, as implemented in MACCLADE, and the concentrated changes tests were performed for all possible combinations of ambiguous reconstruction resolutions. All P-values <0.100 are reported, though a sequential Bonferroni adjustment for multiple comparisons was applied when interpreting statistical significance (Rice 1989).

For each plumage patch for every species, I also calculated the visual input signal for the UVsensitive cone, separately for males and females, on the basis of the following formula modified from Vorobyev et al. (1998): Q1 = ∫^sub λ^ R^sub 1^(&955;, where λ denotes wavelength, R^sub 1^(λ) is the spectral sensitivity of the UV-sensitive cone (data taken from the Blue Tit [Cyanistes caeruleus], provided by N. Hart [Hart et al. 2000]) and is assumed to be the same for all species, S(λ) is the reflectance spectrum of a given feather patch, and integration is over the UV portion of the light spectrum (300-400 nm). Thus, Q1 values are continuous variables representing relative UV coloration (i.e. relative UV signal potentially detected by the eye) for each feather patch. I tested for clade-wide correlations between increases (or decreases) in relative UV coloration and both changes in habitat use and changes in male and female body size (natural log transformed), respectively, with independent contrasts (Felsenstein 1985). Tarsus length was used as an indication of relative body size for each sex, and these data were taken from Jaramillo and Burke (1999) and Webster (1992). I examined evolutionary associations between changes in relative UV coloration and body size, rather than mating system directly, for two reasons. First, the interpretation of a correlation between a continuous trait and a multistate categorical trait can either be ambiguous (Purvis and Rambaut 1995) or provide a limited number of comparisons of independent contrasts, or both. second, mating system is an aspect of life history that is not well known for several of the taxa here (Orians 1985, Jaramillo and Burke 1999). Comparisons with changes in body size for each sex independently provided more comparisons of independent contrasts (Hormiga et al. 2000). In the New World blackbird family species with larger average harem sizes generally have larger differences in body size between males and females; thus, size dimorphism is closely associated with mating system (see below; Webster 1992, Price and Lanyon 2004).

Standardized independent linear contrasts were calculated using CAIC (Purvis and Rambaut 1995) for both male and female body size, and Ql values for each male and female feather patch, whereas habitat was treated as a categorical variable. I tested for a relationship between the independent variables, body size and habitat, and the dependent variables, Ql for each male and female feather patch, by regression of each of the latter onto each of the former, forcing the regressions through the origin (Purvis and Rambaut 1995). For two continuous variables, if the slope of the regression is significantly different from zero, there is a true relation between the two variables in the absence of phylogenetic effects (Purvis and Rambaut 1995). When the independent variable is categorical (i.e. habitat), the expected mean of the contrasts in the dependent, continuous variable is zero, if there is no relationship between the two traits. A positive mean significantly greater than zero indicates that evolution of the state in the independent variable that was given a higher value is correlated with evolution of a larger value in the dependent variable (Purvis and Rambaut 1995). All P-values <0.100 are reported below, though a sequential Bonferroni adjustment for multiple comparisons was applied when interpreting statistical significance (Rice 1989). All statistical analyses on contrasts were performed with JMP statistical software.

RESULTS

Maximum-likelihood and Bayesian analyses recovered the same two major clades (groups 1 and 2) found by Johnson and Lanyon (1999), with nearly identical resolution of relationships within each clade (Fig. 1). In addition, H. pyrohypogaster was reconstructed as the sister taxon to Lampropsar tanagrinus, and A. assimilis was placed as the sister taxon to A. phoeniceus (Cadena et al. 2004). The two analyses here produced slightly different topologies overall; however, that did not alter character reconstructions in MACCLADE nor interpretation of correlations between independent contrasts. Most of the conflict was in the placement of Nesopsar nigerrimus and D. warszewiczi. In the ML analysis, N. nigerrimus was sister to group 1 and D. warszewiczi was sister to group 2; whereas the Bayesian analysis placed D. warszewiczi sister to both groups 1 and 2 and N. nigerrimus sister to the entire ingroup. The latter topology is congruent with Johnson and Lanyon (1999) in the placement of these two taxa; thus, all trees shown throughout this paper represent the Bayesian inference of phylogeny (Fig. 1).

Among the 40 grackle and allied taxa, I identified 22 plumage patches (Table 3), all of which varied in at least one aspect of UV coloration for males (Table 1), and 17 of which varied in at least one aspect of UV coloration for females (Table 2). Intraspecific variation was such that for MaxUV and PeakUV, all individuals of the same sex were scored identically for each feather patch, respectively. For males, three patches were invariant and lacked PeakUV among all taxa: upper tail coverts, base retrices, and outer retrices (characters 9, 21, and 22, respectively). For females, tertials, secondaries, lower rump, upper tail coverts, belly, thigh, base retrices, and outer retrices (characters 1, 2, 8, 9, 13, 16, 21, and 22, respectively) were invariant and lacked MaxUV among all taxa; tertials, secondaries, scapulars, middle back, upper tail coverts, base retrices, and outer retrices (characters 1, 2, 5, 7, 9, 21, and 22, respectively) were invariant and lacked PeakUV among all taxa. Many plumage characters showed repeated convergence both within and among groups 1 and 2 (Fig. 1), when mapped onto the molecular phylogeny, for both PeakUV and MaxUV. There were only two instances of unambiguous reversal in male plumage characters (loss of MaxUV in characters 12 and 13 for Quiscalus major), and no unambiguous reversal in female plumage characters (i.e. once PeakUV or MaxUV has evolved in a lineage, it is rarely lost).

Plumage characters for males had an overall RI of 0.11 and an overall CI of 0.18 when PeakUV was reconstructed on the molecular tree. Individual plumage-character RIs ranged from 0.0 to 0.50, with most having RIs of 0.0, and individual plumage character CIs ranged from 0.0 to 1.0, with most between 0.11 and 0.33 (Table 3). MaxUV reconstruction for male plumage characters resulted in an overall RI of 0.20 and an overall CI of 0.16. Individual plumage-character RIs ranged from 0.0 to 0.57, and individual plumage-character CIs ranged from 0.09 to 1.00 (Table 3). Plumage characters for females had an overall RI of 0.21 and an overall CI of 0.35, when PeakUV is reconstructed on the molecular tree. Individual plumage-character RIs ranged from 0.0 to 0.67, with all but four characters having RIs of 0.0, and individual plumage-character CIs ranged from 0.0 to 1.0, with most between 0.20 and 0.50 (Table 3). MaxUV reconstruction for female plumage characters resulted in an overall RI of 0.0 and an overall CI of 0.40. All individual plumage-character RIs were 0.0, and individual plumage-character CIs ranged from 0.00 to 1.00 (Table 3). For all female plumage characters, changes in MaxUV were restricted to terminal taxa when reconstructed on the molecular tree.

FIG. 1. Bayesian inference of phylogeny for grackles and allies (Icteridae) on the basis of 2,184 base pairs of DM\A combined from two mitochondrial gene regions, cytochrome b and ND2. Numbers above each branch indicate the Bayesian posterior probability of that node derived from 9,900 MCMC sampled trees. A similar phylogeny was obtained with a maximum-likelihood (ML) optimality criterion (-In L = 16392.85), with the main topological differences being the placements of Nesopsar nigerrimus and Dives warszewiczi sister to groups 1 and 2, respectively. The phylogeny shown here is used for all subsequent character reconstructions (Figs. 2-6), which were identical on the ML tree.

TABLE 3. Homoplasy statistics for male and female plumage characters in grackles and allies for two character states, PeakUV and MaxUV (see text for definition of character states). Retention indices (RI) and consistency indices (CI) are given for individual plumage characters mapped onto the molecular phylogeny. Plumage characters are named as in Jaramillo and Burke (1999) and arranged in the order in which reflectance data were collected from each individual.

Very few plumage characters unambiguously define clades with more than two taxa. For both males and females, PeakUV in the breast and belly (characters 12 and 13, respectively) unites Pseudoleistes virescens, P. guirahuro, and Xanthopsar flavus (Fig. 2). Additionally for males, PeakUV in the vent (character 14) defines this clade. These characters have relatively high RI values for the data set (see Table 3). For male plumage characters, MaxUV in the posterior auricular, posterior supercilium, and crown (characters 17, 18, and 19, respectively) unites the clade of four Quiscalus species that is sister to Q. quiscula (Fig. 3). MaxUV in the breast and nape (characters 12 and 20, respectively) also define this clade, with a subsequent loss for both in Q. major. Overall, many of the patches seem to have evolved independently within males and females; however, many adjacent feather patches for each sex have similar patterns of evolution, yet rarely are they identical.

As reconstructed onto the molecular phylogeny, polygyny has evolved at least twice, and possibly as many as six times, from a monogamous plesiomorphic condition (Fig. 4, top; RI = 0.40, CI = 0.14). After sequential Bonferroni adjustment, very few associations between gains or losses in UV coloration and the evolution of a polygynous mating system are statistically significant (Table 4). Of the three statistically significant associations with the evolution of a polygynous mating system, all were in male plumage characters; MaxUV in the anterior flank (character 10) has been gained once and lost three times, MaxUV in the breast (character 12) has been gained twice and lost three times, and MaxUV in the belly (character 13) has been gained twice and lost three times. These gains and losses were significantly associated with the evolution of polygyny when parallelisms (DELTRAN) were favored in the reconstruction of a polygynous mating system. When reversals (ACCTRAN) were favored in reconstructing the evolution of polygyny, numbers of gains and losses of MaxUV were equal for these plumage characters (see below). Notably, for all female plumage patches, there was no association between changes in UV coloration and branches of the phylogeny on which polygyny had evolved (Table 4). Reconstruction of predominant habitat use on the molecular phylogeny revealed that the use of open habitats has evolved at least seven times independently in the grackles and allies from the plesiomorphic condition of forested habitats (Fig. 4, bottom; RI = 0.47, CI = 0.10). The concentrated changes tests indicated that changes in several male (both PeakUV and MaxUV) and female (PeakUV only) plumage characters were associated with open habitat, though none of these was statistically significant after sequential Bonferroni adjustment (Table 4). For all female plumage patches, changes in MaxUV were not associated with an open habitat.

Regression of standardized independent linear contrasts indicated that Ql values (see above) for several plumage characters were significantly positively correlated with body size, after sequential Bonferroni adjustment (Table 5). Increases in relative UV coloration for the male breast, belly, middle back, nape, scapulars, and upper back (characters 13, 12, 7, 20, 5, and 6, respectively) were correlated with larger male size, whereas an increase in relative UV coloration for the female outer retrices (character 22) was correlated with larger female size. Meta-analysis of the significant P-values in Table 5 indicated that significantly more (χ^sup 2^ = 4.25, df = 1, P < 0.05) male plumage characters than female plumage characters were correlated with an increase in body size. Furthermore, six male plumage characters was a significant number of positive correlations (sign test, P = 0.03). There were no significant negative correlations between Q1 values and body size for either sex, and there were no significant positive or negative correlations between Ql values and habitat for either sex.

FIG. 2. Ancestral-state reconstructions for PeakUV coloration of the male breast (top: character 12) and belly (bottom: character 13).

DISCUSSION

Chromatic UV plumage signals have evolved multiple times in the plumage of grackles and allies. Many individual plumage characters, both male and female, showed high levels of homoplasy for these species. Overall RIs ranged from 0.00 to 0.21 when plumage characters were mapped onto the molecular tree (Table 3). These values are much lower than values reported for morphological data in a few other studies that provide RIs (e.g. Omland and Lanyon 2000, Price and Lanyon 2002). Consistency indices are more often reported for studies mapping morphological characters and, again, the overall CIs here (range: 0.16-0.40) are much lower than those reported elsewhere (Sanderson and Donoghue 1989). However, on the basis of a similar number of taxa, Ornland and Lanyon (2000) reported CIs for plumage color evolution in orioles (Icteridae) comparable to those found here. Only a few characters unambiguously define nodes above the species level (see Figs. 2 and 3), and many of the changes in PeakUV and MaxUV were autapomorphic (uniquely derived), especially for female plumage characters. Reconstructions of UV plumage evolution for both males and females suggest that the divergence of UV coloration between closely related taxa can occur rapidly and are congruent with the rapid evolution of plumage characters shown in other studies (Burns 1998, Omland and Lanyon 2000, Hill and McGraw 2004, Lovette 2004). High levels of homoplasy, and potential for rapid divergence of UV coloration in the grackles and allies, reinforces the growing consensus that when using plumage characters in phylogenetic reconstruction, caution is warranted. Although some authors have found plumage characters to be congruent with phylogeny (Livezey 1991, Prum 1997), the present study indicates that phylogenetically informative plumage characters are in the minority and cannot be identified a priori. Their identification requires the reconstruction of plumage evolution on an independently derived hypothesis of evolution.

FIG. 3. Ancestral-state reconstructions for MaxUV coloration of the male posterior auricular (top: character 17), posterior supercilium (middle: character 18), and crown (bottom: character 19).

FIG. 4. Ancestral-state reconstructions for mating system (top) and habitat (bottom) for grackles and allies. Character state assignments for extant taxa taken from Orians (1985) and Jaramillo and Burke (1999).

TABLE 4. Concentrated changes statistics for correlated evolution of plumage characters (arranged alphabetically) and predominal habitat or mating system or both in the grackles and allies (Icteridee), for both PeakUV and MaxUV. Values indicate that changes in that plumages character are strongly concentrated (P ≤ 0.050) or marginally concentrated (0.050 ≤ P ≤ 0.100) on branches of the molecular phylogeny reconstructed to have open habitat or a phylogenous mating system, respectively, given the indicated resolutions of ambigous character reconstruction. Plumage characters not a appearing in the table, or empty cells, indicate P- values for the concentrated changes test ≥0.100. If neither ACC nor DEL follows a plumage character in the table, there was no ambigouty in its evolutionary reconstruction

TABLE 5. Regression statistics for independent linear contrasts of relative UV coloration (i.e. Ql values; see text) of plumage patches on both male and female body size (natural log transformed) for grackles and allies (Icteridae). All regressions were forced through the origin. Independent contrasts were calculated on the Bayesian tree (Fig. 1). A positive or negative correlation is indicated under "association." For all analyses, degree of freedom = 1.

The extant distribution of chromatic UV coloration in grackles and allies has resulted from changes, since the common ancestor of the group, to ancestral male and female plumage characters that lacked UV signals. Overall, males have gained more UV colored plumage patches than females. Female PeakUV and MaxUV were invariant in 7 and 8 of the 22 plumage characters (31.8% and 36.4%), respectively; whereas only 3 of the 22 (13.6%) male plumage characters were invariant for PeakUV, and none was invariant for Max UV (Tables 1 and 2). This contrasts with Irwin's (1994) conclusion that changes in female plumage were more labile across the icterines when considering aspects of human-visible coloration. One possible explanation for this discrepancy is that UV coloration may be lumped into Irwin's (1994) categorization of black plumages, especially for males. Thus, the number of male evolutionary changes would be underestimated. Although gains in UV plumage coloration have occurred across the lineage of grackles and allies in both sexes, only \two unambiguous examples of reversal in UV plumage were found, both in males, which suggests that losses of UV coloration may be more constrained evolutionarily than gains. The potentially costly microstructural aspects of the keratin that result in UV reflectance from the feather (Andersson 1999, Prum et al. 2003) may have secondary benefits to an individual, such as resistance to feather wear, and thus could be under strong selection to be retained once gained. However, because almost nothing is known about the genetic control of UV coloration, whether early gains (ACCTRAN) or parallelisms (DELTRAN) should be more likely in the evolution of PeakUV and MaxUV is hard to determine, and the interpretation of gains and losses relies heavily on the method of reconstruction chosen for many of the plumage characters. Regardless of these two extreme scenarios, the large number of changes reconstructed for many plumage characters suggests that UV coloration, in general, is evolutionarily labile for grackles and allies (e.g. Fig. 3).

In spite of high levels of homoplasy for individual plumage characters, several plumage patches showed strong or marginal associations with evolutionary changes in both habitat and mating system, as indicated by the concentrated changes tests (Table 4). Although few of these are statistically significant after adjusting for multiple testing, many of these associations may still indicate important biological relationships. Statistical insignificance does not necessarily equal biological insignificance (Johnson 1999). A clear trend indicated by the P-values from the concentrated changes tests (Table 4) is the relatively large number of associations between the evolution of MaxUV in many male plumage characters and a polygynous mating system (e.g. compare Fig. 3 [middle] with Fig. 4 [top]). This can be contrasted with no associations between UV evolution (MaxUV or PeakUV) in female plumage characters and polygyny. Meta-analysis of the P-values in Table 4 indicates that significantly more male plumage characters had changes in UV coloration concentrated on branches of the phylogeny where polygyny had evolved as compared with female plumage characters (MaxUV: χ^sup 2^ = 13.87, df = 1, P < 0.0005; PeakUV: χ^sup 2^ = 5.11, df = 1, P < 0.025). These data are consistent with a hypothesis of sexual selection for male UV coloration, because sexual selection is often considered to be more intense in polygynous species, thus driving the exaggeration of male traits (M011er and Pomiankowski 1993, Andersson 1994, Badyeav and Hill 2003). Studies have shown female preference for UV coloration (Hunt et al. 1999, Keyser and Hill 2000, Siitari et al. 2002), which suggests a role for female choice in its evolution. However, many of the associations between changes in male UV coloration and polygyny found here were very sensitive to the method of resolution chosen for ambiguous character-state reconstruction (Table 4). Furthermore, depending on the method of reconstruction used for interpreting the evolution of polygyny, some male plumage characters showed more losses of UV on branches that had evolved polygyny, when sexual selection would predict more gains.

Although character-state reconstruction is somewhat ambiguous regarding the role of sexual selection on UV plumage signals, independent contrasts provide support for a sexual selection hypothesis. Independent contrasts identified seven male plumage characters that had strong positive correlations between increases in relative UV coloration (Ql values) and larger body size; six of these were statistically significant after sequential Bonferroni adjustment (Table 5). This was a significant number of positive correlations (sign test, P = 0.03), because there were no negative correlations between relative UV coloration and male body size (all P > 0.100 for negative associations). Again, meta-analysis of Table 5 that compares the number of significant correlations for males with those for females indicated significantly more correlations between increases in male UV coloration and increases in body size (χ^sup 2^ = 4.25, df = 1, P < 0.05). Notably, females had no clear trend in the directionality of UV evolution in relation to changes in body size (Table 5). The number of strong negative as compared with positive correlations (P < 0.100 in Table 5) of changes in UV and changes in female size were not statistically different (sign test, P = 0.44). The above results combined would be consistent with a hypothesis of sexual selection for male UV plumage signals, assuming that size changes in males resulted in sexual size- dimorphism. Sexual size-dimorphism has been shown to be closely associated with increased polygyny in the blackbird family (Webster 1992, Price and Lanyon 2004), and polygynous species are believed to experience more intense sexual selection (Andersson 1994). Visual inspection of the evolution of sexual-size dimorphism in the grackles-and-allies phylogeny indicated that gains of UV coloration for these seven male plumage characters occurred on branches with an increase or no change in sexual size-dimorphism. Furthermore, the concentrated changes tests identified the evolution of UV coloration in all of these seven male plumage characters to be associated with the evolution of a polygynous mating system in at least one method of character-state reconstruction (i.e. ACCTRAN, DELTRAN, or both; Table 4). The agreement between the two methods of data analysis strengthens the evidence for sexual selection of UV plumage signals in grackles and allies. Future studies of mate choice on the basis of UV signals in blackbirds are warranted, given that several plumage patches in males have evolved chromatic UV coloration.

On the basis of analyses of the light available in different environments, Endler (1993) predicted animals to use shorter wavelength colors (i.e. blue and UV-possibly in combination with red) for signaling in woodland-shade and forestshade habitats and black for signaling in open habitats. Marchetti (1993) has been more general in predicting closed-habitat species to be brighter overall than open-habitat species. Marchetti (1993) found support for these predictions in a group of Phylloscopus warblers, where species living in dark habitats had brighter plumage patches. However, the influence of habitat on plumage coloration within grackles and allies may be confounded by an interaction between habitat and mating system. Searcy et al. (1999) found evidence that polygyny was associated with marsh nesting (i.e. open habitat) for blackbirds, and Table 4 shows that UV evolution in several plumage characters was associated with the evolution of both open habitat and polygyny. Decoupling the influence of habitat and mating system on the evolution of UV plumage signals is difficult, because both could cause selection for similar coloration. Furthermore, the concentrated changes test only tests for associations between binary characters. Here, the broad categorization of habitats may, in addition to lumping variations in mating system, mask or inflate true evolutionary correlations between these characters and chromatic UV coloration.

Regardless of limitations of the character coding here, changes in UV coloration for several plumage patches, mainly in males, were marginally concentrated on branches of the phylogeny where preference for open habitat had evolved, to the exclusion of associations with mating system (Table 4). Johnson and Lanyon (2000) concluded that the ancestral plumage of grackles and allies was all black. Given this, the evolution of several UV plumage signals (i.e. increased brightness), together with preference for open habitat, contradicts predictions that (1) UV signals should be found predominantly in species using woodland and forest shade and (2) brighter species should be found in darker habitats (Endler 1993, Marchetti 1993). Although this conclusion ignores the evolution of other visual colors, previous work found the evolution of carotenoid coloration to be associated with open habitat (Johnson and Lanyon 2000), which also contradicts the above predictions. Taken together, results from the present study combined with Johnson and Lanyon's data are consistent with the results from a study across more than 60 species of birds that found open-habitat species to have higher overall brightness than closed-habitat species (i.e. forest shade; McNaught and Owens 2002). Ultraviolet signals are believed to be more conspicuous at close range but harder to see at longer ranges (Andersson 1996). In open habitats, UV signals may provide a way for blackbirds to communicate with conspecifics while not increasing conspicuousness to distant predators. In fact, UV coloration may enhance crypsis at long ranges in open habitats, because open-light environments have relatively high levels of UV (as well as other wavelengths; Endler 1993). Thus, increased UV reflectance may help birds blend in with the surrounding ambient light. Unfortunately, comparisons of independent contrasts provide no evidence for the effect of different habitats on UV coloration. I found no strong correlations, positive or negative, between changes to an open habitat and changes in relative UV coloration (i.e. the UV visual signal, Ql) for male or female plumage characters (all P > 0.100). A possible explanation is the small number of contrasts generated from the categorical habitat variable for comparison with the Ql contrast values.

Examples of all feather colors, with and without a UV contribution to color, have been found (Burkhardt 1989, Finger and Burkhardt 1994, Eaton and Lanyon 2003), which suggests that UV can evolve independently of human visual colors. However, Andersson (1996) has suggested that much of avian UV coloration is a passive byproduct of feather pigmentation for other colors (e.g. reds, yellows, a\nd browns). (For example, many red and yellow feathers also have peaks of light reflectance in the UV; M. D. Eaton pers. obs.) Yet there is growing evidence that feather color is often the result of interactions between two or more pigment types (McGraw et al. 2004), or pigments and the keratin microstructure (Mays et al. 2004, Shawkey and Hill 2005). Mays et al. (2004) concluded that the pigment lutein and the keratin structure interacted to produce the yellow color of the throat and breast feathers of Yellowbreasted Chat (Icteria virens). They found that the UV color of these feathers was more different between the sexes than the visible color wavelengths (Mays et al. 2004). Bridge and Eaton (2005) also found larger differences in the UV component of tern (Sterna spp.) coloration when comparing new and old (with respect to molt sequence) primary flight feathers within individuals. These two studies suggest that the differing proximate causes for different components of feather coloration can respond to selection independently.

In the absence of data on the exact proximate cause(s) of UV reflectance in grackles and allies, I defined PeakUV and MaxUV in an attempt to isolate UV as a distinct color evolving independently. Unfortunately, neither one may be completely free of interacting effects from other mechanisms of color production. For example, the greater coverts (character 3), which are equivalent to the epaulets, have been shown to have evolved carotenoid coloration in grackles and allies several times in males (see figure 2 in Johnson and Lanyon 2000). The reconstruction of PeakUV for the greater coverts shows a pattern of evolution strikingly similar to that of carotenoid coloration in the epaulets, which suggests that UV is not evolving independently of the other color-producing mechanisms of the epaulets (McGraw et al. 2004), or that all of the different proximate causes of color in the epaulets have responded similarly to historical selection pressures. Nonetheless, PeakUV and MaxUV each represent aspects of chromatic UV coloration that can potentially respond to selective pressures from changes in habitat, mating system, or both, and thus were treated separately throughout the present study. A definition of UV coloration that requires both maximum spectral reflectance in the UV (e.g. MaxUV), as well as a sharp change in spectral reflectance into the UV (e.g. PeakUV), may be advantageous for future research. This would be a more conservative definition of chromatic UV coloration (than those used in the present study), and it would potentially decouple UV coloration from all other feather colors. This approach faces the added challenge of designating a priori what constitutes an adequately sharp change in spectral reflectance. I am presently exploring this methodology to reconstruct the evolution of pure UV signals in a subset of the grackles and allies.

In conclusion, this is the first study to reconstruct the evolution of UV plumage coloration in a comparative phylogenetic context. Chromatic UV coloration has evolved multiple times in grackles and allies, and it is not restricted to particular regions of the plumage. Character-state reconstructions suggest that chromatic UV plumage can evolve rapidly, and the high levels of homoplasy for the data set caution against the use of UV plumage characters for reconstructing evolutionary relationships. The results presented here do not clearly favor either habitat changes or mating-system changes as selective forces acting on the evolution of UV plumage. Yet the many strong associations of male UV plumage signals with the evolution of polygyny (as compared with no female associations), and significant correlations between increased relative UV coloration and larger male size for a significant number of plumage characters, suggest a role for sexual selection in the evolution of male UV coloration. In addition, associations for several plumage characters between an open habitat and UV signals suggest a role for light environment in explaining extant patterns of UV coloration across the grackles and allies. The multiple gains of chromatic UV coloration in this group warrant future behavioral studies on many of these species to determine the function of UV signals, and ultimately differentiate between alternative evolutionary hypotheses. Finally, the evolution of UV signals would be entirely missed if the human visual system were used to assess plumage colors. Spectral reflectance data provide a way to objectively measure and quantify plumage colors, and additional methods of character coding of spectral data are needed to facilitate its use in future comparative analyses of plumage evolution.

ACKNOWLEDGMENTS

I thank S. Lanyon, R. Zink, S. Weller, A. Simons, K. Johnson, and two anonymous reviewers for reviewing drafts of this manuscript and providing helpful comments. I thank F. K. Barker and S. Lanyon for molecular data. I thank F. K. Barker, B. Barber, and P. Berendzen for assistance with computer programs. Funding for this research was provided by the Bell Museum of Natural History through the Dayton- Wilkie Natural History Funds.

LITERATURE CITED

ANDERSSON, M. 1994. Sexual Selection. Princeton University Press, Princeton, New Jersey.

ANDERSSON, S. 1996. Bright ultraviolet colouration in the Asian whistling-thrushes (Myiophonus spp.). Proceedings of the Royal Society of London, Series B 263:843-848.

ANDERSSON, S. 1999. Morphology of UV reflectance in a whistling- thrush: Implications for the study of structural colour signalling in birds. Journal of Avian Biology 30:193-204.

ANDERSSON, S., AND T. AMUNDSEN. 1997. Ultraviolet colour vision and ornamentation in Bluethroats. Proceedings of the Royal Society of London, Series B 264:1587-1591.

BADYAEV, A. V., AND G. E. HILL. 2000. Evolution of sexual dichromatism: Contribution of carotenoid- versus melanin-based coloration. Biological Journal of the Linnean Society 69:153-172.

BADYAEV, A. V., AND G. E. HILL. 2003. Avian sexual dichromatism in relation to phylogeny and ecology. Annual Review of Ecology, Evolution, and Systematics 34:27-49.

BADYAEV, A. V., G. E. HILL, AND B. V. WECKWORTH. 2002. Species divergence in sexually selected traits: Increase in song elaboration is related to decrease in plumage ornamentation in finches. Evolution 56:412-419.

BAKER, R. R., AND G. A. PARKER. 1979. The evolution of bird coloration. Philosophical Transactions of the Royal Society of London, Series B 287:63-130.

BENNETT, A. T. D., I. C. CUTHILL, J. C. PARTRIDGE, AND K. LUNAU. 1997. Ultraviolet plumage colors predict mate preferences in starlings. Proceedings of the National Academy of Sciences USA 94:8618-8621.

BRIDGE, E. S., AND M. D. EATON. 2005. Does ultraviolet reflectance accentuate a sexually selected signal in terns? Journal of Avian Biology 36:18-21.

BURKHARDT, D. 1989. UV vision: A bird's eye view of feathers. Journal of Comparative Physiology A: Sensory, Neural, and Behavioral Physiology 164:787-796.

BURNS, K. J. 1998. A phylogenetic perspective on the evolution of sexual dichromatism in Tanagers (Thraupidae): The role of female versus male plumage. Evolution 52:1219-1224.

BUTCHER, G. S., AND S. ROHWER. 1989. The evolution of conspicuous and distinctive coloration for communication in birds. Pages 51-108 in Current Ornithology, vol. 6 (D. M. Power, Ed.). Plenum Press, New York.

CADENA, C. D., A. M. CUERVO, AND S. M. LANYON. 2004. Phylogenetic relationships of the Redbellied Crackle (Icteridae: Hypopyrrhus pyrohypogaster) inferred from mitochondrial DNA sequence data. Condor 106:664-670.

CHEN, D., J. S. COLLINS, AND T. H. GOLDSMITH. 1984. The ultraviolet receptor of bird retinas. Science 225:337-340.

CUTHILL, I. C., J. C. PARTRIDGE, A. T. D. BENNETT, S. C. CHURCH, N. S. HART, AND S. HUNT. 2000. Ultraviolet vision in birds. Advances in the Study of Behavior 29:159-214.

DARWIN, C. 1871. The Descent of Man, and Selection in Relation to Sex, 2nd ed. Murray, London.

DE QUEIROZ, K. 1996. Including the characters of interest during tree reconstruction and the problems of circularity and bias in studies of character evolution. American Naturalist 148:700-708.

DICKINSON, E. C., ED. 2003. The Howard and Moore Complete Checklist of the Birds of the World, 3rd ed. Princeton University Press, Princeton, New Jersey.

DOUCET, S. M., AND R. MONTGOMERIE. 2003. Multiple sexual ornaments in Satin Bowerbirds: Ultraviolet plumage and bowers signal different aspects of male quality. Behavioral Ecology 14:503-509.

DUNN, P. O., L. A. WITTINGHAM, AND T. E. PITCHER. 2001. Mating systems, sperm competition, and the evolution of sexual dimorphism in birds. Evolution 55:161-175.

EATON, M. D., AND S. M. LANYON. 2003. The ubiquity of avian ultraviolet plumage reflectance. Proceedings of the Royal Society of London, Series B 270:1721-1726.

ENDLER, J. A. 1993. The color of light in forests and its implications. Ecological Monographs 63:1-27.

ENDLER, J. A., AND M. THRY. 1996. Interacting effects of lek placement, display behavior, ambient light, and color patterns in three Neotropical forest-dwelling birds. American Naturalist 148:421- 452.

FARRIS, J. S. 1989. The retention index and the rescaled consistency index. Cladistics 5: 417-419.

FELSENSTEIN, J. 1985. Phylogenies and the comparative method. American Naturalist 125: 1-15.

FIGUEROLA, J., AND A. J. GREEN. 2000. The evolution of sexual dimorphism in relation to mating patterns, cavity nesting, insularity and sympatry in the Anseriformes. Functional Ecology 14:701-710.

FINGER, E., AND D. BURKHARDT. 1994. Biological aspects of bird colouration and avian colour vision including ultraviolet range. Vision Research 34:1509-1514.

GOMEZ, D., AND M. THERY. 2004. Influence of ambient light on the evolution of colour signals: Comparative analysis of a Neotropical rainforest bird community. Ecology Letters 7:279-284.

GRAY, D. A. 1996. Carotenoids and sexual dichromatism i\n North American passerine birds. American Naturalist 148:453-480.

HACKETT, S. J., AND K. V. ROSENBERG. 1990. Comparison of phenotypic and genetic differentiation in South American antwrens (Formicariidae). Auk 107:473-489.

HART, N. S., J. C. PARTRIDGE, I. C. CUTHILL, AND A. T. D. BENNETT. 2000. Visual pigments, oil droplets, ocular media and cone photoreceptor distribution in two species of passerine bird: The Blue Tit (Parus caeruleus L.) and the blackbird (Turdus merula L.). Journal of Comparative Physiology A 186:375-387.

HARVEY, P. H., AND M. D. PAGEL. 1991. The Comparative Method in Evolutionary Biology. Oxford University Press, Oxford.

HAUSMANN, F., K. E. ARNOLD, N. J. MARSHALL, AND I. P. F. OWENS. 2003. Ultraviolet signals in birds are special. Proceedings of the Royal Society of London, Series B 270:61-67.

HILL, G. E., AND K. J. McGRAw. 2004. Correlated changes in male plumage coloration and female mate choice in cardueline finches. Animal Behaviour 67:27-35.

HORMIGA, G., N. SCHARFF, AND J. A. CODDINGTON. 2000. The phylogenetic basis of sexual size dimorphism in orb-weaving spiders (Araneae, Orbiculariae). Systematic Biology 49:435-462.

HUELSENBECK, J. P., AND F. RONQUIST. 2001. MRBAYES: Bayesian inference of phylogeny. Bioinformatics 17:754-755.

HUNT, S., A. T. D. BENNETT, I. C. CUTIIILL, AND R. GRIFFITHS. 1998. Blue Tits are ultraviolet tits. Proceedings of the Royal Society of London, Series B 265:451-455.

HUNT, S., I. C. CUTHILL, A. T. D. BENNETT, AND R. GRIFFITHS. 1999. Preferences for ultraviolet partners in the Blue Tit. Animal Behaviour 58:809-815.

IRWIN, R. E. 1994. The evolution of plumage dichromatism in the New World Blackbirds: Social selection on female brightness? American Naturalist 144:890-907.

JARAMILLO, A., AND P. BURKE. 1999. New World Blackbirds: The Icterids. Princeton University Press, Princeton, New Jersey.

JOHNSON, D. H. 1999. The insignificance of statistical significance testing. Journal of Wildlife Management 63:763-772.

JOHNSON, K. P., AND S. M. LANYON. 1999. Molecular systematics of the grackles and allies, and the effect of additional sequence (cyt b and ND2). Auk 116:759-768.

JOHNSON, K. P., AND S. M. LANYON. 2000. Evolutionary changes in color patches of blackbirds are associated with marsh nesting. Behavioral Ecology 11:515-519.

JOHNSON, S. G. 1991. Effects of predation, parasites, and phylogeny on the evolution of bright coloration in North American male passerines. Evolutionary Ecology 5:52-62.

KEYSER, A. J., AND G. E. HILL. 2000. Structurally based plumage coloration is an honest signal of quality in male Blue Grosbeaks. Behavioral Ecology 11:202-209.

KLUGE, A. G., AND J. S. FARRIS. 1969. Quantitative phyletics and the evolution of anurans. Systematic Zoology 18:1-32.

KOSE, M., R. MAND, AND A. P. MQLLER. 1999. Sexual selection for white tail spots in the Barn Swallow in relation to habitat choice by feather lice. Animal Behaviour 58:1201-1205.

KREITHEN, M. L., AND T. EISNER. 1978. Ultraviolet light detection by the homing pigeon. Nature 272:347-348.

LANYON, S. M., AND K. E. OMLAND. 1999. A molecular phylogeny of the blackbirds (Icteridae): Five lineages revealed by cytochrome-b sequence data. Auk 116:629-639.

LIVEZEY, B. C. 1991. A phylogenetic analysis and classification of recent dabbling ducks (Tribe Anatini) based on comparative morphology. Auk 108:471-508.

LOVETTE, I. J. 2004. Molecular phylogeny and plumage signal evolution in a trans Andean and circum Amazonian avian species complex. Molecular Phylogenetics and Evolution 32:512-523.

LOWTHER, P. E., R. FRAGA, T. S. SCHULENBERG, AND S. M. LANYON. 2004. Nomenclatural solution for a polyphyletic Agelaius. Bulletin of the British Ornithologists' Club 124:171-177.

MADDISON, D. R., AND W. P. MADDISON. 2000. MACCLADE: Analysis of Phylogeny and Character Evolution, version 4.0. Sinauer and Associates, Sunclerland, Massachusetts.

MADDISON, W. P. 1990. A method for testing the correlated evolution of two binary characters: Are gains or loses concentrated on certain branches of a phylogenetic tree? Evolution 44:539-557.

MARCHETTI, K. 1993. Dark habitats and bright birds illustrate the role of the environment in species divergence. Nature 362:149-152.

MAYS, H. L., JR., K. J. McGRAw, G. RITCHISON, S. COOPER, V. RUSH, AND R. S. PARKER. 2004. Sexual dichromatism in the Yellow-breasted Chat Icteria virens: Spectrophotometric analysis and biochemical basis. Journal of Avian Biology 35:125-134.

McGRAW, K. J., K. WAKAMATSU, A. B. CLARK, AND K. YASUKAWA. 2004. Red-winged Blackbirds Agelaius phoeniceus use carotenoid and melanin pigments to color their epaulets. Journal of Avian Biology 35:543- 550.

McNAUGHT, M. K., AND I. P. F. OWENS. 2002. Interspecific variation in plumage colour among birds: Species recognition or light environment? Journal of Evolutionary Biology 15:505-514.

McNETT, G. D., AND K. MARCHETTI. 2005. Ultraviolet degradation in carotenoid patches: Live versus museum-based specimens of wood warblers (Parulidae). Auk 122:793-802.

MLLER, A. P., AND T. R. BiRKHEAD. 1994. The evolution of plumage brightness in birds is related to extrapair paternity. Evolution 48: 1089-1100.

MLLER, A. P., AND A. PoMiANKOwsKi. 1993. Why have birds got multiple sexual ornaments? Behavioral Ecology and Sociobiology 32: 167-176.

ODEEN, A., AND O. HASTAD. 2003. Complex distribution of avian color vision systems revealed by sequencing the SWSl opsin from total DNA. Molecular Biology and Evolution 20:855-861.

OMLAND, K. E. 1997. Examining two standard assumptions of ancestral reconstructions: Repeated loss of dichromatism in dabbling ducks (Anatini). Evolution 51:1636-1646.

OMLAND, K. E., AND S. M. LANYON. 2000. Reconstructing plumage evolution in orioles (Icterus): Repeated convergence and reversal in patterns. Evolution 54:2119-2133.

ORIANS, G. 1985. Blackbirds of the Americas. University of Washington Press, Seattle.

OSORIO, D., A. MIKLOSI, AND Zs. GONDA. 1999. Visual ecology and perception of coloration patterns

Source: Auk, The

More News in this Category