Differences in coloration between the sexes (sexual dichromatism) can increase or decrease in a species through evolutionary changes in either or both sexes diverging or converging in their colors. Few previous studies, however, have examined the relative rates of such changes, particularly when dichromatism is lost. Using reflectance data from 37 species of the New World blackbird family (Icteridae), we compared evolutionary rates of plumage color change in males and females when dichromatism was either increasing (colors diverging) or decreasing (colors converging). Increases in dichromatism involved divergent changes in both sexes at approximately equal rates. Decreases in dichromatism, in contrast, involved changes in females to match male plumage colors that were significantly more rapid than any changes in males. Such dramatic changes in females show how selection can differ between the sexes. Moreover, these evolutionary patterns support the idea that losses of dimorphism involve genetic mechanisms that are already largely present in both sexes, whereas increases in dimorphism tend to involve the appearance of novel sex-specific traits, which evolve more slowly. Our results have broad implications for how sexual dimorphisms evolve.

We collected reflectance measurements from male and female representatives of 37 New World blackbird species (Icteridae) in the grackles-and-allies clade. For each species, we measured 22 feather patches on study skins of three males and three females using an S-2000 spectrometer (Ocean Optics, Dunedin, Florida) with an R200-7- UV /VIS reflectance probe (fiber diameter = 200 µm) and a PX-2 pulsed xenon light source oriented perpendicular to the measured surface. Data consisted of the percentage of light reflected at each wavelength from 300 to 700 nm, calibrated against a Spectralon white reflectance standard and averaged into 10 nm bins, collected using OOIBASE32 software. We averaged the reflectance data across three individuals for each feather patch, separately for each sex within each species. We calculated color differences (ΔS) by first calculating receptor quantum catch values for each feather patch using the equation Q_i = ∫_λ R_i(λ)S(λ)dλ, where λ denotes wavelength, R_i(λ) is the spectral sensitivity of each cone cell of type i, S(λ) is the reflectance spectrum of a given feather patch, and integration is over the entire range of avian visual sensitivity (300-700 nm). Q₁–Q₄ therefore represented the receptor quantum catch of the ultraviolet sensitive cone, the short-wave-sensitive cone, the middle-wave-sensitive cone, and the long-wave-sensitive cone, respectively. We used Maximum Likelihood reconstructions of continuously valued characters on a molecular phylogeny for the grackles and allies to estimate ancestral values at each node on the phylogeny for each of the four quantum catches (Q₁-Q₄), separately for both males and females. We summed the ∆S values for all 22 feather patches to calculate all-patch-∆S differences between the sexes, reflecting overall levels of sexual dichromatism, and changes in males and in females along each phylogenetic branch.