Introduction
In many highly fecund marine species reproduction and recruitment dynamics can be highly stochastic, involving extreme year-to-year variation in the number of surviving offspring per parent. As a result, in a given year nearly all the recruits at a specific location may be the offspring of just a few “lucky winner” individuals out of orders of magnitude larger adult population, which has been termed “sweepstakes reproductive success” (SRS, (D. Hedgecock 1994). Since in terms of genetics this implies very small effective population size, SRS dramatically accelerates genetic drift relative to the expectations based on the census sizes, and thus can be the major force shaping genetic diversity and adaptive potential of wild populations. Adaptive capacity of reef-building corals is of particularly high interest, due to worldwide reef decline driven by the effects of climate change (Global Coral Reef Monitoring Network 2021; Eddy et al. 2021). Existing models of coral adaptation (e.g. (Bay et al. 2017; Matz, Treml, and Haller 2020; Kleypas et al. 2016) assume that equally fit corals are equally likely to produce surviving offspring, which may be an oversimplification if coral recruitment actually follows the SRS pattern. Here, we studied adult and juvenile cohorts of the broadcast-spawning coral Acropora hyacinthus across reef sites at the island of Yap, Micronesia (Fig. 1 A), to look for the following three predictions of the SRS hypothesis (Dennis Hedgecock and Pudovkin 2011). First, since “lucky winners” in a given year are different among locations, SRS is expected to generate transient (or “chaotic”) genetic differentiation among recruits at different sites (Johnson and Black 1982). This variation may not necessarily translate into differentiation of adult populations since these assemble over multiple years, integrating over yearly variation in recruit sources (Broquet, Viard, and Yearsley 2013). Second, since under the SRS the recruits are produced by far fewer parents than the whole adult population, their genetic diversity should be lower than in adults. Third, SRS predicts that at least some recruits at a given site will be close relatives. We found evidence for the second and, most notably, the third of these predictions. We then modeled the effect of SRS on coral persistence in the Indo-West Pacific metapopulation model we have developed previously (Matz, Treml, and Haller 2020).