Coupled biophysical global ocean model and molecular genetic analyses identify multiple introductions of cryptogenic species

Dawson et al. 10.1073/pnas.0503811102.

Supporting Information

Files in this Data Supplement:

Supporting Table 2
Supporting Figure 5
Supporting Figure 6
Supporting Figure 7
Supporting Figure 8
Supporting Figure 9
Supporting Figure 10
Supporting Figure 11
Supporting Table 3
Supporting Figure 12

Fig. 5. Schematic of the life history of Aurelia. Aurelia has a typical scyphozoan life history, consisting of a small, probably perennial, benthic polyp that reproduces asexually to produce other polyps and, usually during Spring, free-living planktonic medusae. Medusae reproduce sexually, and the planula larva is brooded for a short period by the female, then released into the water column where it spends probably <1 week before settling on the benthos and metamorphosing into a polyp. Thus, in contrast to many marine benthic invertebrates and fishes, the "adult" is the main dispersive phase. Medusae usually live <6 months in the wild, although medusae older than 1 year have been recorded, and they may live up to 2 years in captivity (1, 2).

1. Hamner, W. M. & Jensen, R. M. (1974) Am. Zool. 14, 817–818.

2. Arai, M. N. (1997) A Functional Biology of Scyphozoa (Chapman & Hall, London.)

Fig. 6. El Niño simulation. Changes in ocean circulation over at least the past seven millennia have been primarily limited to subtle shifts in the location and intensity of the subtropical gyres and to natural variability associated with phenomena such as the El Niño–Southern Oscillation (ENSO) and the North Atlantic Oscillation. The largest interannual change in ocean circulation is caused by the El Niño phenomenon. The effect of this change in circulation is missing in the current experiments because we use monthly advection fields averaged from 20 years of the Parallel Ocean Climate Model (POCM) integration. To investigate the effect of this changed circulation and thereby gauge the possible impact of interannual regional circulation variability on the simulated advection pathways, additional sensitivity experiments were carried out by using annually looped seasonal advection fields from the El Niño year of July 1982–June 1983. These artificial circulation scenarios represent an upper limit for the effects of interannual changes in Pacific Ocean conditions, providing a test of the robustness of our findings with respect to natural dispersion of Aurelia. The results from this set of experiments are shown in this figure. Although shifts occurred in the final distributions and maximum particle migrations, consistent with expectations of ENSO-altered circulation patterns, the results with respect to the absence of translocation between the Aurelia sites are robust. A further set of sensitivity experiments under perpetual La Niña conditions (data not shown) also indicates that our conclusions regarding particle dispersal patterns are unchanged by interannual variations in Pacific Ocean circulation fields.

Perpetual El Niño simulations (100 years) are depicted. Shown are final year cumulative occurrence distributions (colored marine areas and scale) for releases in the five primary zones of occurrence of Aurelia sp.1 (A) and Aurelia sp.4 (B) indicated by adjacent red land areas. Black contours represent estimated maximum extent of particles based on samples taken over the integration period. Gray shading (see scale bar) indicates the proportional effect of temperature on survivorship of medusae before reproduction.

Fig. 7. Parameterization of open ocean diffusivity. The resolution of the POCM permits a certain amount of eddy-mixing in prognostic mode. However, the temporal averaging of U = (u, v) will more or less eliminate any advection due to eddy-mixing in the present "off-line" Lagrangian model. In addition, there is a degree of mixing at subgrid scale resolutions not resolved by the POCM in prognostic mode. As a result, to incorporate mixing from unresolved open ocean diffusive processes, we add a random walk component to the particle displacements at each time step. This component is calculated based on a simple constant horizontal diffusivity, K_h = 1,000 m²•s^–1, i.e.,

where R_N is a normally distributed random number (mean = 0, standard deviation = 1) and Dt is the model time step. A similar relationship holds for Dy the north–south direction. A sensitivity analysis in this figure shows that the overall results are robust to changes in the value of the horizontal diffusivity. In particular, we assessed a series of experiments wherein the diffusivity is halved, doubled, and increased 5-fold, and we found that none of our principal conclusions regarding natural Aurelia dispersion were altered. Lower values, such as those used to investigate Lagrangian transport of krill in the Southern Ocean (1), would only act to further limit dispersal potential. Our experimental philosophy was to run with generous diffusivity so that the particles’ simulated range extent is an upper bound on natural dispersal. Final year cumulative occurrence distributions (COD), colored marine areas and scale, for standard releases (integrated for 100 years) and estimated maximum extent of particles (black contour) for eddy diffusivities set to K_h = 500 m²•s^–1 (Left) and K_h = 5,000 m²•s^–1 (Right).

1. Thorpe, S. E., Heywood, K. J., Stevens, D. P. & Brandon, M. A. (2004). Deep-Sea Res. I 51, 909–920.

Fig. 8. Coverage of island stepping stones. To assess the model’s ability to simulate migration by "island hopping," the POCM land mask was supplemented with data from the 5° ETOPO5 bathymetry data set. Ocean grid points in the POCM that contained islands in ETOPO5 were assigned a fractional number indicating the percentage of the POCM grid containing land. Islands were considered to be any area with an ocean depth of <10 m, i.e., areas with potential suitable benthic habitat for Aurelia scyphistomae. During the simulations, for each successive rerelease, particles were released from these island grid boxes in proportion to the number of particles that had impacted on that region during the previous year, scaled by the land fraction. An investigation of the 2° ETOPO2 data set showed no additional island sites at positions that would aid particle stepping. This figure shows additional grid points (red circles) containing land unresolved in ETOPO5/POCM (5 min) that is present in ETOPO2 (2 min) data sets. In all relevant positions, this just represents a minor change in size of already resolved islands. Data are superimposed on distributions from Fig. 2.

Fig. 9. Parameterization of unresolved coastal dispersion. A major deficiency in using global models to investigate biological pathways, especially for species that spend a large amount of their life cycle in coastal regions, is the failure to resolve coastal flows that can affect along-shore dispersal. In an attempt to parameterize this effect, we have included a term (C_mix) in the advection equation that acts on the coastal distribution of particles at the end of each set of life cycles (1 year) to diffuse the distribution along-shore. Characteristic tidal velocities are estimated by using a global inverse tidal model that incorporates TOPEX/Poseidon altimetry data (for further details, see www.oregonstate.edu/research/po/research/tide/global.html). A background speed of 1 m/s is added to the tidal velocity to account for nontidal flows (e.g., coastal-trapped waves, transient coastal currents, and shelf waves). The characteristic distance (L) that results from a random walk of N steps of distance l is

Here, we take l = U_coastal•Dt, where U_coastal is the sum of the tidal and the nontidal speeds, and Dt = 6 h is the time scale over which the most important tidal components vary (also corresponding to the model time-step). As the diffusion is applied only at the end of each life span segment (one year for a given particle),

= 1,460.

At the end of each set of life cycles, the cumulative occurrence distribution COD(0) at each coastal grid box is recalculated to spread the distribution along-shore with a Gaussian distribution, centered at that grid box, such that

where x is the along-shore distance from the grid box. A simple application of the C_mix term to a single point distribution on the Japanese coastline is shown in this figure. Example of the effect on a point distribution (Left) of coastal diffusion parameterization (C_mix) acting with COD(0) = 1000 (Right).

Fig. 10. Preliminary gene tree incorporating additional partial rDNA sequence data used to identify molecular phylogenetic species of Aurelia. 1 shortest tree (1,377 steps); Consistency index = 0.6398; 10,000 heuristic (TBR) searches; unweighted maximum parsimony; ITS1 plus partial 5.8 S; 606 positions, including gaps; 1,000 bootstrap replicates (10 heuristic searches per replicate, as above); values of >50% shown.

Fig. 11. Preliminary gene tree incorporating additional specimen from Kwajalein, for which only COI sequence is available, used to identify molecular phylogenetic species of Aurelia. 1 shortest tree (1,115 steps); consistency index = 0.3794; 10,000 heuristic (TBR) searches; unweighted maximum parsimony; COI; 658 positions, including missing; 1,000 bootstrap replicates (10 heuristic searches per replicate, as above); values of >50% shown.