Particle-tracking Estimates of Larval Connectivity
Virtual particles representing larvae were released from 438 distinct release sites distributed equidistantly along the Irish Sea coastline, at 10 km intervals and 1 km offshore (Fig. 1B). By selecting release sites 1 km offshore, all particles were released within the hydrodynamic model grid cell adjacent to the coast, appropriate for simulating spawning of near-shore species. Within a 100 m radius of each release site, 625 release locations were randomly spaced to allow for some stochasticity within the simulated currents and resulting particle advection that represents unresolved sub-grid-scale and sub-time-scale processes. Cohorts of 625 particles were released daily (at 12:00 noon) from each of the 438 sites. To encompass releases during the full spring-neap tidal cycle, particles were released for the first 16 days of June. To capture dispersal variability over seasonal timescales, this procedure was repeated for July, August and September, which covers the spawning season of S. umbilicalis (Williams 1964; Underwood 1972; Garwood & Kendall 1985). Adopting a similar experimental design to this study, Robins et al. (2013) showed that release of 10,000 particles was sufficient to represent dispersal and connectivity patterns within the Irish Sea. Here, 40,000 individual particles were released from each site, totalling 17.52 million particles (625 particles × 438 sites × 16 days × 4 months) within the Irish Sea. Particles were allowed to propagate for 10 days from release, based on the maximum pelagic larval duration of S. umbilicalis (Keith et al. 2011). During each 10-day simulation, individual particles were advected, and their locations were recorded hourly.
Given that larvae have been shown to settle passively in areas of low current velocity (Kendall & Lewis 1986), the particles were parameterized as passive, neutrally buoyant, and released at the surface to simulate maximum dispersal. No criteria were applied to account for habitat type, and larvae were able to settle at any of the 438 sites. Sexual maturity in S. umbilicalis can occur between 1-2 years of age (Williams 1964; Bode et al. 1986; Garwood & Kendall 1985), but for model simplicity we assumed sexual maturity occurred within 1 year. We did not consider mortality/fecundity of larvae so that our results represent the maximum dispersal potential.
Pairwise site connectivity for the 12 sampled sites was determined by calculating the proportion of particles released from a source site that arrived within a 5 km radius of a settlement site. This was calculated at each model time-step, between 2-10 days after their release, producing an overall averaged connectivity. The 5 km radius was set to minimize overlap between release and settlement areas, but also to account for dispersal potential over a tidal cycle based on the average tidal excursion of the Irish Sea (Robins et al. 2013). Multi-generational connectivity was estimated between all 438 sites by accounting for the connectivity of each site with all other sites to quantify the year-on-year spread, over 100 years (Fig. S1). In this way, the intermediary sites implemented between our 12 sampled sites facilitated stepping-stone spread along coastal sites of the Irish Sea. Here, we have focused on potential larval spread via stepping-stone connectivity and have thus assumed the metapopulation size remains constant, with population fecundity equalling mortality (also see Giménez et al. 2020b). Pairwise connectivity estimates were averaged over the four months that the model was run to create a connectivity matrix between each pair of sites, and this year-on-year spread was calculated for 100 years. The matrix used in subsequent analyses was selected based on the maximum number of years it took for any pair of sites to become connected, excluding sites that never became connected over the 100-year timeframe (Fig. S2, Table S3).