1. Age shapes fundamental processes related to behaviour, survival, and reproduction. Where age influences reproductive success, non-random mating with respect to age can magnify or mitigate such effects. Consequently, the correlation in partners’ age across a population may influence its productivity. Despite widespread evidence for age-assortative mating, little is known about what drives this assortment and its variation. Specifically, the relative importance of active (same-age mate preference) and passive processes (assortment as a consequence of other spatial or temporal effects) in driving age-assortment is not well understood.
2. In this paper, we compare breeding data from a great tit and mute swan population (51- and 31-year datasets respectively) to tease apart the contributions of pair retention, cohort age-structure, and active age-related mate selection to age-assortment in species with contrasting life-histories.
3. Both species show age-assortative mating, and variable assortment between years. However, we demonstrate that the drivers of age-assortment differ between the species, as expected from their life-histories and resultant demographic differences. In great tits, pair fidelity has a weak effect on age-assortative mating through pair retention; variation in age-assortment is primarily driven by fluctuations in age-structure from variable juvenile recruitment. Age-assortative mating is therefore largely passive, with no evidence consistent with active age-related mate selection. In mute swans, age-assortment is partly explained by pair retention, but not population age-structure, and evidence exists for active age-assortative pairing.
4. This difference is likely to result from shorter life-spans in great tits compared to mute swans, leading to fundamental differences in their population age-structure, whereby a larger proportion of great tit populations consist of a single age-cohort. In mute swans, age-assortative pairing through mate selection may also be driven by greater age-dependent variation in fitness.
5. The study highlights the importance of considering how different life-histories, and demographic differences arising from these, affect population processes that appear congruent across species. We suggest that future research should focus on uncovering the proximate mechanisms that lead to variation in active age-assortative mate selection (as seen in mute swans); and the consequences of variation in age-structure on the ecological and social functioning of wild populations.

Data Collection

Data used here are from a long-term study population of great tits in Wytham Woods, Oxford (51°46’N, 1°20’W). This population has been monitored since 1947, where breeding adults and their chicks have been marked with unique British Trust for Ornithology (BTO) rings since the 1960s. All chicks reared in nest-boxes are ringed at 14 days of age, while adults are trapped at boxes during the nestling phase, and identified by ring number or marked with a new ring if they are immigrant birds. Parent age is based on the year of hatching for local birds, or plumage characteristics for immigrants. Although immigration rates are high (53%), most are first caught as yearlings (76%) and therefore can be aged accurately.

We also used data from Abbotsbury Swannery and Chesil Fleet (50°35’N, 2°30’W), where swans have been breeding since at least the fourteenth century. Since 1977, all individuals that breed and hatch at the colony have been marked with a unique BTO ring. Additionally, since 1989, a roundup has occurred every 2-years after the breeding season, where all birds in the colony are captured and marked, thus identifying non-breeding individuals present in the population. Age is based either on the date of hatching for resident birds or colouration of plumage and beak for immigrant birds, which can discriminate between birds of 1-, 2- and 3-years or older. Most breeding individuals hatched in the swannery (80% males and 93% females), and therefore age is known. Additionally, of the 13% that immigrate and breed in the population, 57% do so at an age when they can still be aged accurately.

Data collection from both populations adhered to local guidelines for the use of animals in research, and all fieldworkers involved over the many years of fieldwork have held BTO ringing licenses (author BTO permit number: C / 6987). Additionally, the long-term monitoring in Wytham has been subject to review by the Department of Zoology ethics committee.

Statistical Analyses

We calculated the correlation of age in mates using Spearman’s rank correlation as a nonparametric test of association that calculates the direction and strength of correlation in partner ages.

We used permutation analysis to simulate random pairing. We compared the observed number of pairs assorted by age against a distribution expected from random pairing with respect to age. We created this distribution through permutation analysis by randomly assigning partners together, depending on the birds breeding in a given year, and calculating the frequency of absolute age-differences in these randomly-paired populations. This was repeated 1000 times. From these 1000 iterations, we constructed a frequency distribution of the number of age-assorted pairs, against which we compared our observed number of age-assorted pairs to evaluate whether the number of pairs assorted by age would be expected through random pairing.

We evaluated whether age-assortative mating is age-dependent and driven by assortative pairing within certain age-cohorts. We ran a Spearman’s rank correlation between the within-year proportion of ‘first-time breeders’ and the proportion assorted by age in both species. We also ran a Spearman's rank between the within-year proportion of ‘young breeders’ and the proportion assorted by age in both species. Further, we used permutation analysis to assess whether the relationship we see between these two variables would exist if the population were to pair randomly. We assigned partners randomly, depending on those breeding in the given year, and calculated the relationship between the proportion of young breeders and age-assorted pairs. We repeated this 1000 times, yielding a distribution of 1000 Spearman’s rank correlation coefficients when within-year pairing is random.

We used Spearman’s rank correlation to assess the relationship between the within-year proportion of the population that consists of 'newly-formed experienced pairs', and the proportion of the population assorted by age.

This includes:

1. Age-assortative_Mating.RData (all dataframes required to run the analyses can be found here)

2. Disentangling_the_causes_of_age-assortative_mating_-_ANALYSIS.R (R code to perform the analyses)