Joe Woodman gives us information on his paper, “disentangling the causes of age-assortative mating in bird populations with contrasting life-history strategies“, which was shortlisted for this year’s Elton Prize. We also hear a little about his story.

About the paper:

Age is a fundamental trait which affects almost all aspects of life, and the importance of age in influencing the ecological processing of wild populations is becoming increasingly well established (Nussey et al., 2013; Siracusa et al., 2023). Age effects on reproductive success can be particularly profound, so when animals of the same age mate (age-assortative mating), these effects can be magnified or mitigated depending on the age make-up within the pair. Age-assortative mating is widespread in animal populations (Jiang et al., 2013), but surprisingly little is known about the underlying drivers of this process, particularly given that both passive and active mechanisms might lead to age-assortative mating, and that these processes may vary between species.

So, what drives age-assortative mating, and does this depend on features of species-specific demography and ecology? To answer this question, we first defined the mechanisms which might drive age-assortative pairing. Firstly, when partners remain together over many reproductive attempts across their lifespan (pair fidelity), then this might passively drive age-assortative mating if they first mate when they are the same age. Secondly, when a large proportion of a population is made up of a single age-cohort (for example, when lots of individuals are born during a previous breeding season), then this might drive age-assortative mating simply because there is a high chance that pairs will consist of two individuals that fit within this age-cohort. Finally, if all individuals in a population want to enhance their own reproductive success by mating with an older, more experienced partner, then this might lead to active age-assortative pairing where pairs made up of two older individuals commonly arise, forcing younger unpaired individuals to mate with one another. In our research, we wanted to test the relative contributions of these processes in driving age-assortative mating and its variation.

However, to really understand the importance of these processes in producing age-assortative mating, we thought it would be important to take a comparative approach by comparing their consequences in two species with different demographic strategies. For this, we used long-term breeding data from two species with contrasting life-histories. Great tits Parus major are a short-lived passerine bird, with breeding ages ranging 1 to 9, averaging 1.8 (Bouwhuis et al., 2009). They exist at the ‘fast’ end of the fast–slow continuum of life-history, as they start breeding early with high fecundity, but die earlier (Stearns, 1992). In contrast, mute swans Cygnus olor are at the ‘slow’ end, with breeding ages ranging from 2 to 24, averaging 7.6 years (Perrins et al., 1994). However, although these species have very different life-histories, they are linked in that they both undergo annual breeding attempts, display age-specific reproductive success, form socially monogamous pairs, and have been shown to demonstrate age-assortative mating (Harvey et al., 1979; McCleery et al., 2008). We therefore endeavored to see whether the seemingly similar patterns of age-assortment within these two species are produced by the same mechanisms, or whether different aspects of the species’ life-histories mean that age-assortment is driven by different processes.

Our analysis centred around randomisation and permutation analysis, which allowed us to compare the observed number of pairs assorted by age to that expected from a process of random pairing. For this, we took each year of data which represents a single breeding season (51- and 31-years for the tits and swans respectively), and randomly paired the individuals together that were found breeding in that year, repeating this 1000 times for both populations. This led to 1000 permuted populations for each species per year, from which we could build a distribution of the number of age-assorted pairs that form from a process of random pairing, against which we could compare the actual number of age-assorted pairs.

This shows the general strength of age-assortative mating, but how can we test the contributions of our three proposed mechanisms? To assess the role of pair fidelity, we restricted the data to only include newly-formed pairs, therefore removing the effect that individuals which remain together have on producing age-assortment. To evaluate the role of incidental age-assortment due to variation in age structure, we looked at the correlation between the proportion of young breeders in an annual population and the proportion of pairs assorted by age. We then ran our permutation analysis, therefore distorting the proportion of age-assorted pairs over 1000 permuted populations, allowing us to assess whether the relationship between these two population-level variables is consistent with random pairing. Finally, we looked for evidence for active age-assortative mating by running our permutation analysis on a restricted subset of the data that we defined as ‘newly-formed experience pairs’. This only includes pairs where both partners have bred before, but not with each other, therefore these pairs could not be assorted by age as a consequence of being the same age at first breeding, or through subsequent pair retention.

We found age-assortative mating in both species, but this shows variation between years and is likely explained by different underlying mechanisms which depend on their different life-histories and resulting demographic structures. In both the tits and swans, we found that pair fidelity contributes to age-assortative mating, but to a much greater extent in mute swans. In great tits, the amount of age-assortment seems to be largely driven by fluctuations in the age-structure, in that when the proportion of young breeders in the population is higher, we see more age-assortative mating. Importantly, this relationship would be expected if pairing was random with respect to age, and so the age-assortment among young breeders occurs because they occupy a large cohort within the population and so are likely to pair with each other through chance. We found that there is no age-assortment above that expected through random pairing among newly-formed experienced great tit pairs, further suggesting that age-assortment in this species is passive and related to the age composition of available mates, as opposed to active age-based mate selection. In contrast, in the swans, we see much weaker association between variation in population age structure and age-assortative mating. Interestingly, among the newly-formed experienced pairs, we find some evidence of age-assortative mating above what would be expected through random pairing, potentially suggesting that a process of active age-assortative mate selection is at play (Figure 2).

Our research highlights a really key point in that population-level patterns that seem superficially similar across species might not be driven by the same underlying processes. In the case of age-assortative mating, we can see that the high mortality rates and resulting age distribution of great tit populations allows for purely passive age-assortment, as the younger individuals may dominate the age structure, a trend not observed in longer-lived species like mute swans. Considering this, there is greater contribution of pair fidelity in mute swan pairs as partners who pair together at the same age will live longer and therefore remain age-assorted for longer, but there is also evidence of active age-assortative pairing, potentially influenced by greater age-dependent fitness in this species. Gaining a better understanding of the proximate mechanisms that allow for active age-assortative mating in species such as mute swans will allow us to understand the consequences of individual variation in discerning age and its potential importance in driving age-assortative mating in other species with varying life-history strategies.

This study is also one that highlights the value of long-term population studies. The work would not be possible were it not for the collective 82-years of data collection from the great tits at Wytham Woods, Oxford, and mute swans at Abbotsbury Swannery in Dorset. It is the continual efforts of literally hundreds of fieldworkers which means that key questions about the ecology of wild populations can be answered. The maintenance of data collection from these populations as new themes in ecology arise will continue to prove incredibly valuable. Combined with advances in technological tracking and genetic approaches, this will allow for even more integration of individual-level understanding across the lifespan with heaps of continuous long-term data. It is exciting to see what themes these long-term population studies will contribute to in the future!

About the author:

Joe is a final year PhD student at the Edward Grey Institute of Field Ornithology, University of Oxford. His research focuses on how variation in age structure arises in natural populations, and the consequences of this on ecological and social functioning.