Homophily by morphological and behavioral traits has been described in several species of vertebrates, but its functional consequences remain poorly studied. Homophily by plurally breeding females may improve direct fitness by enhancing reproductive success. Female mammals may exhibit phenotypical masculinization due to exposure to androgens during early development, a condition that is associated with maternal performance during subsequent breeding. Our goal was to assess whether female composition (in terms of masculinization) of plurally breeding groups influences female fitness in a natural population of degus (Octodon degus). We assessed if plurally breeding female degus assort themselves by anogenital distance (AGD), an accurate measure of masculinization level. We also quantified if homophily by AGD phenotype affects female reproductive success and the reproductive output of the group. Plurally breeding groups typically included similarly masculinized (i.e., long AGD) females or similarly feminized (short AGD) females, indicating a strong degree of homophily. Females weaned more offspring in plurally breeding groups with more masculinized females. Additionally, standardized variance in the number of offspring weaned decreased in plurally breeding groups with mostly masculinized females, indicating greater reproductive equality in these groups. We conclude that female degus organize into homophilic social groups of similar AGD, and that social groups of masculinized females exhibit higher reproductive success.

Study population: Data come from a long-term study conducted between 2009 and 2015 (7 generations) on a natural degu population located at Estación Experimental Germán Greve Silva (33°23′ S, 70°31′ W, altitude 495 m), a field station of the Universidad de Chile.

Number of females examined: Between 2009 and 2015, we recorded AGDs on 190 adult females. However, we only include 150 females in the analysis, because they were part of 57 multifemale groups.The remaining 40 females were excluded from the analysis because they were alone (n=15), were in female-male pair (n=16), were in two male-one female trios (n=5), or in a three male-one female quartet. Year 2010 was discarded from our analyses because there were no social groups with two or more females.

Individual attributes: female AGD and female body weight: Female phenotype in terms of masculinization was assessed through anogenital distance (AGD), the distance between the ventral anus commissure to the base of the genital papilla. We measured the AGD of adult females exhibiting a non-perforated vagina with a digital caliper (precision 0.1 mm) at every capture event. We calculated the average AGD from 17.8 ± 8.6 measurements per female, resulting in a single AGD estimate per female. For all statistical analyses, AGD was used as a continuous predictor. Repeatability of AGD from the 359 females examined during 2009-2015 was 0.73 (95% confidence interval: 0.70–0.77, n = 3168 measurements; 822 measures from autumn, 2346 measures from spring), suggesting that AGD is a stable measurement within and between individuals and therefore, an efficient trait to estimate the masculinization level of females

Social group attributes: We considered four attributes of social groups. These variables were (i) the number of females, representing the total number of adult females in the social group; (ii) the mean group AGD, representing the average AGD of all female group members, (iii) the within group coefficient of variation of AGD, an estimate of how different the females’ AGD are in a group (the degree of heterogeneity in masculinization within the group). Additionally, we calculated (iv) the mean AGD of groupmates after excluding the AGD value of the focal female (mean AGD of group - AGD of focal female). These variables were considered to test the effect of social group composition (in terms of female groupmates) on litter size at weaning of each focal female.

Social group determination: The main criterion used to assign individuals to social groups was the sharing of burrow systems at night. The sharing of burrow systems was determined by (i) burrow trapping during early morning activity and (ii) night-time telemetry.

Measures of reproductive success: For all females, reproductive success was quantified as the number of offspring per litter (hereafter LS) at weaning. Variation in reproductive success among females in the same plurally breeding groups was indexed by the standardized variance in litter size (S²) at weaning. S² was calculated as: variance in litter size of adult females in a group/mean in litter size of adult female group members²

First model: mean AGD of groupmates = AGD of focal female + number of females + female body weight + (1+ AGD of focal female |Year). 

Second model: Litter size at weaning = AGD of focal female + mean AGD of groupmates + mean AGD of group + coefficient of variation of AGD of group + number of females + female body weight + (1|Year)

Third model: Standardized reproductive variance = AGD of focal female + mean AGD of groupmates + mean AGD of group + coefficient of variation of AGD of group + number of females + female body weight + (1|Year)

Marginal and conditional R² values were reported as measures of goodness-fit. Posterior analysis of outlier influence and qq-plots revealed no strong departures from model assumptions. All statistical analyses were implemented in R 4.0.3. Generalized mixed models were fitted using functions in library lme4 1.1–26, and inflation factor values were estimated using the library car ver. 3.0-10. Poisson dispersion factor was estimated using functions in library blmeco 1.4. All libraries were accessed during May 2021.

Between 2009 and 2015, we recorded AGDs on 190 adult females. However, we only include 150 females in the analysis, because they were part of 57 multifemale groups.The remaining 40 females were excluded from the analysis because they were alone (n=15), were in female-male pair (n=16), were in two male-one female trios (n=5), or in a three male-one female quartet. Year 2010 was discarded from our analyses because there were no social groups with two or more females.

ND: undetermined. Female without tissue sampling, then without DNA analyses. Litter size, was not determinated. 

Correa et al. 2021 Dataset Readme attached.