Atmospheric-ocean coupling drives prevailing and synchronic dispersal patterns of marine species with long pelagic durations

Abstract

Dispersal shapes population connectivity and plays a critical role in marine metacommunities. Prominent species for coastal socioecological systems, such as jellyfish and spiny lobsters, feature long pelagic dispersal phases (LPDPs), which have long been overlooked. Here, we use a cross-scale approach combining field surveys of these species with a high-resolution hydrodynamic model to decipher the underlying mechanisms of LPDP patterns in northwestern Mediterranean shores. We identified basin-scale prevailing dispersal routes and synchronic year-to-year patterns tightly linked to prominent circulation features typical of marginal seas and semienclosed basins, with an outstanding role of a retentive source area replenishing shores and potentially acting as a pelagic nursery area. We show how the atmospheric forcing of the ocean, a marked hydrological driver of the Mediterranean Sea, modulates dispersal routes and sources of LPDP at interannual scales. These findings represent a crucial advance in our understanding of the functioning of metapopulations of species with LPDP in marginal seas and may contribute to the effective management of coastal ecosystem services in the face of climate change.

Figure 1

Average sea surface circulation in the western Mediterranean and biological time series locations. The lobster settlement index (LSI, red circles) belongs to different marine protected areas (MPAs) at two sites, one in the northernmost basin (NCat) and three in the Balearic Islands (North, NW and South Mallorca, green area and figure inset) (a). The jellyfish stinging index (JSI, blue circles) covered three tourist hot-spot locations in the Balearic Islands with different orientations and exposed to different water masses (E Ibiza, N Mallorca and S Menorca). Squares denote FAO Geographical Statistical SubAreas (GSAs) used to interpret the connectivity of LSI lobster series (see “Methods”). The Gulf of Lions (GoL, light orange area) and Balearic Sea (BS, light green area) indicate the areas where ocean-atmosphere fluxes were extracted for comparison against field records. A 10-year climatology of drifter sources (%) reaching either NCat (b) or Mallorca (c) is displayed as the particle location eight months before the observations (origin). Source geographic locations were aggregated in monthly density maps of $0.25\times 0.{25}^{\circ }$. NC, Northern Current in red. LS, Ligurian Sea. SoG, Strait of Gibraltar. The figure and superimposed maps were created using “M Map: A mapping package for MATLAB”, v.1.4m (www.eoas.ubc.ca/~rich/map.html) in MATLAB v.R2010b (www.mathworks.com).

Figure 2

Spatiotemporal connectivity patterns and dispersal kernels. Comparison of observed LSI anomalies for NCat and Mallorca (red line, Panels a and c, errors are 1 SD) vs. percentage of individuals arriving from adult lobster habitat GSAs (0–200 m) (blue bars) (Fig. 1). The percentage of individuals was averaged over the settlement season (May to August) to compare with settlement field records. Correlations LSI vs. total % individuals from GSAs (8, 9, 10, 11.2): NCat ($r=0.89$; p-value $=0.0003$; $n=10$); Mallorca ($r=0.81$; p-value $=0.004$; $n=9$). Dispersal kernels were calculated for drifting times spanning from 1 to 9 months (panels b, d).

Figure 3

Prevailing sources and interannual patterns of LPDP arriving at the target areas after eight months. Temporal (a,c) and spatial (b,d) representation of the first principal component (PC1) of empirical orthogonal function (EOF) analysis conducted on source areas of individuals based on 120 monthly origin maps (2003–2014) (more details in “Methods”). The results are for N. catalonia (a,b) and Mallorca (c,d). The total explained variance was 29% for N. Catalonia and 44% for Mallorca. Positive PC and EOF values are in red, and negative values are in blue. Original source maps can be reconstructed using the climatology (Fig. 1b,c) and adding the product of PC1 by EOF values. Therefore, intensified influence as source areas emerges when blue/red areas (from EOF value) and PC value match (blue/red). The figure and superimposed maps were created using “M Map: A mapping package for MATLAB”, v.1.4m (www.eoas.ubc.ca/~rich/map.html) in MATLAB v.R2010b (www.mathworks.com).

Figure 4

Atmosphere-ocean coupling driving dispersal patterns and proposed mechanisms explaining LPDP interannual variability. Density distribution of the Pearson correlation coefficients between ocean-atmosphere fluxes and PC1 (Fig. 3a,c) for summers/settlement season of jellyfish and lobster, which was obtained by bootstrap resampling (10,000 times) for the two sites: (a) Heat loss (HL) vs. PC1 for North Catalonia (NCat) (Fig. 3a). (c) (Evaporation-precipitation, E−P) vs. PC1 for Mallorca (Fig. 3c). Linear relationship of the atmosphere-ocean fluxes and LPDP field records for both sites, Lobster Settlement Index (LSI) and Jellyfish Stinging Index (JSI). (b) For NCat, HL vs. LSI ($R^2=0.33$; p-value $=0.021$, $n=16$). (d) In Mallorca, (E−P) vs. LSI ($R^2=0.33$; p-value $=0.003$, $n=14$) and (E-P) vs. JSI ($R^2=0.66$; p-value $=0.005$, $n=10$). More details in Supplementary Table S2, Supplementary Information). We found synchronic patterns for both LPDP: LSI vs. JSI ($R^2=0.65$; p-value $=0.005$, $n=10$).