Field study finds rival arrival predicts higher female song in house wrens

Female song rates in northern house wrens rose sharply when a second female arrived at a male’s territory, according to a three-year field study of this facultatively polygynous songbird. The pattern was most pronounced on the newcomer’s territory: recordings assigned to second females contained an average of 26.36 songs per hour, compared with 5.26 per hour on a first female’s territory when she arrived in a monogamous pairing. The result supports the interpretation that female song is associated primarily with competition between females rather than with attracting a mate, although the passive acoustic design could not always identify which individual produced each song.

The study centered on an established population of northern house wrens, Troglodytes aedon, at Lackawanna State Park in northeastern Pennsylvania. The birds breed in forest-edge habitat around 66 wooden nest boxes distributed across approximately 0.95 square kilometers. Males generally establish territories and begin nest construction before females arrive. Depending on the year, between 6% and 35% of breeding males in this population attracted more than one female. That variation supplied the central comparison: a female could arrive at a male who remained monogamous, become the first female at a territory that later became polygynous, or arrive as the second female at a male who already had a mate.

Researchers used automated recorders to capture the early-morning vocal activity of 23 identified females represented by 54 recordings. Sixteen recordings came from 11 first females whose pairings remained monogamous, 26 came from 14 first females whose mates later acquired a second female, and 12 came from seven second females. A territory could contribute recordings from both the day an arrival was detected and the following day, so some females appeared in more than one social context. The researchers defined polygyny through overlapping successful breeding stages before hatching, treating the overlap as evidence that females could compete over a shared male or associated resources. Every second-female recording occurred while the first female was still in a pre-incubation stage.

To make recordings comparable, the team standardized the number of songs per hour and limited most analyses to the period from sunrise through 194 minutes after sunrise. This window represented the longest recording interval shared by nearly all files; three shorter recordings were analyzed over the approximately 1.5 hours available. A ResNet-18-based deep-learning system initially classified house wren vocalizations by sex, after which researchers manually reviewed its predictions in Raven Pro. The model had been trained on 1,277 male and 1,368 female songs and had reported detection rates of 82–87% for female songs and 87–88% for male songs. Only vocalizations judged to be genuine song were retained, while high-frequency introductory calls were excluded.

The researchers then fitted linear mixed-effects models to the right-skewed song-rate data, applying a natural-log transformation and including female identity as a random effect to account for repeated observations. They compared alternative specifications, including square-root-transformed and Tweedie mixed models, and found that the transformed models satisfied key residual assumptions. A Tweedie model produced qualitatively identical results. Model estimates were back-transformed to the song-rate scale, where they represent geometric rather than arithmetic means. The main model did not include year or recorder type because adding those covariates did not change the findings; a separate analysis within 2022 likewise found that recorder brand did not improve model fit.

Song rates on second-female territories were significantly greater than those on first-female territories in monogamous pairings, a Tukey-adjusted comparison with p = 0.0210 and Cohen’s d = 0.994. They were also greater than rates on the first female’s territory after the arrival of a rival, with an adjusted p = 0.0051 and Cohen’s d = 1.44. By contrast, the first female’s song rate did not differ significantly between monogamous and polygynous contexts, with p = 0.73. In other words, the decisive change was not simply that a female sang more after acquiring a mate. The high rate was concentrated where a second female had entered the breeding arrangement, while the incumbent female’s rate remained comparatively low before and after the rival appeared.

That timing is central to the researchers’ argument. A signal used primarily to attract a mate would be expected to peak when a female arrives, before pairing, and then decline once she has secured a male. A signal associated with female-female competition would instead be expected to rise when a rival enters and begins contesting a male, territory, or nest site. The study did observe elevated singing during second-female arrival, but not a comparable increase in the first female’s singing after the male became polygynous. The authors therefore regard the results as inconsistent with mate attraction being the primary function of female song in this population. Because the two proposed functions can overlap, the evidence does not establish that song has no role in mate attraction under any circumstances.

The passive recording system imposed an important limitation. Recorders were positioned beside nest boxes and could detect most vocal activity within a typical territory, but they did not visually track individual singers. Songs were therefore assigned to territories rather than always to known females. Northern house wrens have individually distinctive songs, and the researchers were able to assess five of the seven second-female territories by comparing recordings with known focal songs. In those five cases, the vocalizations appeared to come predominantly from the second female; two territories could not be assigned confidently because the females’ songs were similar or suitable focal recordings were unavailable. At least one polygynous-territory recording clearly contained two females singing. The high territorial rate may thus reflect newcomer defense, incumbent intrusion, countersinging, or some combination of these interactions.

Several checks argued against simpler explanations for the pattern. Elevated song on second-female territories was not significantly associated with the first female’s nesting stage when the rival arrived, and including breeding stage did not improve model fit. The result also remained when the researchers added year to the model and tested recorder type within 2022. Song rates were similar on the day of arrival and the following day, supporting the decision to combine those observations. Nevertheless, the second-female sample was small—seven territories and 12 recordings—because polygyny is relatively uncommon, so additional breeding seasons would help establish the strength and generality of the pattern.

What the two females are competing over remains unresolved. Secondary females in house wrens can receive less male provisioning and defense than primary females and may have lower reproductive success even after accounting for territory quality, giving both rivals a stake in securing male care or resisting mate sharing. Yet females in this population sometimes choose already-mated males over unmated males, suggesting that paternal care alone may not explain the attraction to a polygynous pairing. Nest-site quality, cavity availability, and the male’s ability to defend or control a territory may therefore be central. Those resources can overlap with mate quality, making it difficult to assign the behavior cleanly to either sexual selection, in which competition concerns access to mates, or social selection, in which females contend for resources affecting reproduction. The study establishes which context predicts more singing, not whether that singing changes mating success, prevents egg destruction, secures nestling care, or ultimately alters fitness.

Subject of Research: Female song rates in facultatively polygynous northern house wrens were examined in relation to mate attraction, female-female competition, and variation in breeding context.

Article Title: Female-female competition, rather than mate attraction, predicts high female song rates in a facultatively polygynous songbird

Article References: Gleason, M. G., Krieg, C. A., & Odom, K. J. (2026). Female-female competition, rather than mate attraction, predicts high female song rates in a facultatively polygynous songbird. Frontiers in Bird Science, 5, 1961683. https://doi.org/10.3389/fbirs.2026.1961683

Image Credits: AI Generated

DOI: 10.3389/fbirs.2026.1961683

Keywords: female birdsong, female-female competition, mate attraction, house wrens, polygyny, sexual selection, social selection, animal communication, field study, northern house wren, Female-female, competition