Avian flush-pursuit insectivores typically use contrasting white plumage patches in their tails or wings to startle potential prey. Although experimental evidence indicates that the extent of white has been fine-tuned by natural selection to optimize foraging performance, the hypothesis that within-population plumage variation directly influences survival or lifetime reproduction and is subject to stabilizing selection has not been tested. Here I provide such a test using data collected as part of a 14-year study of a colour-ringed breeding population of the hooded warbler (Setophaga citrina), a migratory flush-pursuit insectivore that shows inter-individual variation in the extent of white in the tail that is highly repeatable across molts and likely heritable. As expected under stabilizing selection, warblers with average-sized white tail patches achieved significantly higher long-term apparent survival than individuals with either a lesser or greater extent of white in the tail. Evidence of stabilizing selection was especially strong in males, an observation that is likely related to pronounced sexual habitat segregation on the wintering range. My results provide infrequently observed evidence of stabilizing selection operating in a natural population and also illustrate how stabilizing selection can act on avian plumage traits outside the context of sexual and social signaling.

Field data were collected as part of a long-term study of hooded warblers at Hemlock Hill Field Station (41.8°N, 79.9°W), 2010-2023. Intensive efforts were made May-August each year to completely census the breeding population of 45–70 pairs, capture and individually colour-ring all breeding adults, and locate and monitor the reproductive success of all nests. Complete details on general field methods and hooded warbler breeding biology are available elsewhere [Mumme 2018, Mumme et al. 2023, Lignac and Mumme 2023]. 

When birds were captured for ringing, I took a standard series of morphometric measures, including wing chord, tail length, body mass, and the maximum linear extent of white in each of the outer 3 tail feathers, rectrices 4–6 (figure 1a). I then calculated a tail white index (TWIndex) as the sum of the 3 measures divided by 3 times the total tail length, an index that approximates the proportion of the outer rectrices that is white. TWIndex is strongly correlated with the extent of white in each rectrix (electronic supplementary material, figure S3) and constitutes an effective summary index of the extent of white in the tail that simultaneously controls for variation in tail length. About 28% of males and 14% of females have an additional small white spot on rectrix 3. However, because this tail spot is absent from most individuals, is considerably smaller in both length and width, and is only weakly correlated with the extent of white in the other rectrices (electronic supplementary material, figure S3), rectrix 3 was not included in the calculation of TWIndex. For individuals captured and measured in multiple years (electronic supplementary material, figure S1), I used their mean measurements in statistical analyses. 

For all colour-marked breeding individuals I calculated their apparent long-term survival, defined as the difference between the last year an individual was observed or captured in the study area and the year it was initially captured and marked. Thus, apparent long-term survival would be 0 years for a bird present only 1 year. Survival data were available only for breeding birds rung 2010–2022; mean apparent long-term survival was 1.05 (range 0–8, n = 317) for males and 0.73 (range 0–7, n = 308) for females. A small number of marked individuals (8 males, 14 females) were absent from the study area (or present but undetected) 1–2 years before returning in a subsequent year, but their apparent long-term survival was calculated in the same way as birds present continuously.

(b) Statistical analysis: apparent long-term survival

Apparent long-term survival was modeled as a Poisson variable (log link function) in a Generalized Linear Mixed Model (GLMM). I chose this approach over alternatives, such as Cormack-Jolly-Seber mark-recapture models [Schaub and Royle 2013] or Cox proportional hazards models [Moore 2016], because the primary analytical focus for stabilizing selection is the effects of variation in tail white on the observed lifespan of individual birds, not on population-level parameter estimates of either annual survival or hazard. First- and second-order (quadratic) polynomials of TWIndex, and their interactions with sex, were modeled as fixed factors; including the quadratic terms allows statistical testing of the most likely forms of selection acting on variation in tail white, including positive and negative directional selection, disruptive selection, and stabilizing selection [Kingsolver et al. 2012]. Initial year on the study site was included in the model as a random effect. 

Apparent long-term survival clearly underestimates true long-term survival, for two reasons. First, birds rung later in the study had fewer years of potential survival available to them than birds from earlier cohorts, an issue that can be controlled statistically by including initial year on the study site as a random effect in the GLMM. Second, apparent long-term survival confounds actual mortality with undetected emigration off the study area [Schaub and Royle 2013], a problem that is especially prevalent in females, which show less breeding-site fidelity than males [Howlett and Stutchbury 2003, Mumme et al. 2023]. However, because it is unlikely that the extent of white in the tail has any effect on emigration or breeding site fidelity, apparent long-term survival is likely an unbiased underestimate of true long-term survival in relation to tail white. All statistical analyses were conducted using the Fit Model platform of JMP Pro 17.0 (SAS Institute Inc., Cary, North Carolina, USA).

(c) Statistical analysis: reproductive performance and body condition

Reproductive performance for each individual hooded warbler was calculated as the total number of young fledged from nests throughout its reproductive life on the study area. Mean ± SD lifetime reproductive output was 4.6 ± 4.8 (range 0–25, n = 317) for males and 4.2 ± 3.9 (range 0–27, n = 308) for females. Because of high rates of extra-pair fertilizations [Stutchbury et al. 1994, Chiver et al. 2008, Mumme et al. 2023], conclusions about male reproductive performance should be interpreted cautiously. I investigated the effects of tail white on breeding-season body condition by analyzing data on body mass after controlling for the effects of wing chord—a strong indicator of overall body size—and sex. Data on body mass were available for only a subset of individuals in the complete data set; mean ± SD body mass was 11.3 ± 0.5 g (n = 266) for males and 11.9 ± 0.8 g (n = 186) for females.

Reproductive performance and body mass were both modeled as normal (Gaussian) variables in General Linear Mixed Models. Initial year on the study site was included as a random effect, and the first- and second-order polynomials of TWIndex, and their interactions with sex, were modeled as fixed factors. The model for lifetime reproductive performance included number of years present on the study site as a fixed factor, while the body mass model included wing chord as a fixed factor. Examination of normal quantile plots indicated that residuals of both models approximated a normal distribution.

References

Chiver I, Stutchbury BJM, Morton ES. 2008 Do male plumage and song characteristics influence female off-territory forays and paternity in the hooded warbler? Behav. Ecol. Sociobiol. 62, 1981–1990. (doi:10.1007/s00265-008-0629-x)

Howlett JS, Stutchbury BJM. 2003 Determinants of between-season site, territory, and mate fidelity in hooded warblers (Wilsonia citrina). Auk 120, 457–465. (doi:10.1093/auk/120.2.457)

Kingsolver JG, Diamond SE, Siepielski AM, Carlson SM. 2012 Synthetic analyses of phenotypic selection in natural populations: lessons, limitations and future directions. Evol. Ecol. 26, 1101–1118. (doi:10.1007/s10682-012-9563-5)

Lignac C, Mumme RL. 2023 Brood parasitism of hooded warblers by brown-headed cowbirds: severe impact on individual nests but modest consequences for seasonal fecundity and conservation. Ornith. Appl. 125, duac041. (doi:10.1093/ornithapp/duac041)

Moore DF. 2016 Applied survival analysis using R. Springer International Publishing, Cham, Switzerland.

Mumme RL. 2018 The trade-off between molt and parental care in hooded warblers: simultaneous rectrix molt and uniparental desertion of late-season young. Auk: Ornithol. Adv. 135, 427–438. (doi:10.1642/AUK-17-240.1)

Mumme RL, Chiver I, Evans Ogden LJ, Stutchbury BJ. 2023 Hooded warbler (Setophaga citrina), version 2.0. In Birds of the World (Rodewald PG, Ed.). Cornell Lab of Ornithology, Ithaca, NY, USA. (doi:10.2173/bow.hoowar.02)

Schaub M, Royle A. 2014 Estimating true instead of apparent survival using spatial Cormack–Jolly–Seber models. Meth. Ecol. Evol. 5, 1316–1326. (doi: 10.1111/2041-210X.12134)

Stutchbury BJ, Rhymer JM, Morton ES. 1994 Extra-pair paternity in hooded warblers. Behav. Ecol. 5, 384–392. (doi:10.1093/beheco/5.4.384)