Dispersal and population connectivity in the deep North Atlantic estimated from physical transport processes

Highlights

• We estimate dispersal and population connectivity in the deep western North Atlantic.

• A circulation model was used to simulate dispersal of neutrally buoyant larvae.

• Cross-slope dispersal occurred despite the strong along-slope southwestward flow.

• Cross-slope population connectivity does not appear to be precluded by hydrodynamics.

• Depth-related genetic divergence likely reflects ecological or evolutionary forces.

Abstract

Little is known about how larvae disperse in deep ocean currents despite how critical estimates of population connectivity are for ecology, evolution and conservation. Estimates of connectivity can provide important insights about the mechanisms that shape patterns of genetic variation. Strong population genetic divergence above and below about 3000 m has been documented for multiple protobranch bivalves in the western North Atlantic. One possible explanation for this congruent divergence is that the Deep Western Boundary Current (DWBC), which flows southwestward along the slope in this region, entrains larvae and impedes dispersal between the upper/middle slope and the lower slope or abyss. We used Lagrangian particle trajectories based on an eddy-resolving ocean general circulation model (specifically FLAME – Family of Linked Atlantic Model Experiments) to estimate the nature and scale of dispersal of passive larvae released near the sea floor at 4 depths across the continental slope (1500, 2000, 2500 and 3200 m) in the western North Atlantic and to test the potential role of the DWBC in explaining patterns of genetic variation on the continental margin. Passive particles released into the model DWBC followed highly complex trajectories that led to both onshore and offshore transport. Transport averaged about 1 km d⁻¹ with dispersal kernels skewed strongly right indicating that some larvae dispersed much greater distances. Offshore transport was more likely than onshore and, despite a prevailing southwestward flow, some particles drifted north and east. Dispersal trajectories and estimates of population connectivity suggested that the DWBC is unlikely to prevent dispersal among depths, in part because of strong cross-slope forces induced by interactions between the DWBC and the deeper flows of the Gulf Stream. The strong genetic divergence we find in this region of the Northwest Atlantic is therefore likely driven by larval behaviors and/or mortality that limit dispersal, or local selective processes (both pre and post-settlement) that limit recruitment of immigrants from some depths.

Introduction

Understanding how populations are connected through dispersal is of fundamental importance in ecology, evolution, conservation and management, and will play a crucial role in predicting how organisms might respond to contemporary anthropogenic stresses. From an evolutionary perspective, dispersal influences gene flow among populations, which in turn affects genetic diversity, phylogeographic patterns, adaptation to local selective pressures, and ultimately the likelihood of speciation (reviewed in Nosil, 2012). At an ecological scale, dispersal can influence demographic processes (Roughgarden et al., 1985), source–sink dynamics (Holt, 1985), metapopulation (Hanski, 1999) and metacommunity persistence (Leibold et al., 2004), the maintenance of biodiversity (Mouquet and Loreau, 2003, Hubbell, 2001), the spread of invasive species (Neubert and Caswell, 2000, Byers and Pringle., 2006, Caswell et al., 2011) and play a vital role in management and conservation (Palumbi, 2003, Botsford et al., 2003, Botsford et al., 2009, Gaines et al., 2003, Gaines et al., 2010). Connectivity links dynamics at different scales and integrates local heterogeneity affecting regional-scale dynamics and long-term persistence. Quantifying the scale over which populations are connected is vital to understand the relative importance of various ecological processes, identify the appropriate scales of environmental influence in driving population dynamics and determine how local communities might respond to environmental change.

The extent to which populations are connected is determined by the scale, intensity, direction and frequency of dispersal among populations as well as post-settlement processes that influence the fitness of recruits. For many sessile marine invertebrates, dispersal occurs during the larval stage. Larvae (or gametes) are released into the water column where they are dispersed by the currents due to both advective and diffusive processes. The speed, variability and direction of the currents have a strong effect on where larvae move, especially if they exhibit little behavior and disperse essentially as passive particles. However, the interaction between larval behavior and physical transport processes (Paris et al., 2007, North et al., 2008, Sakina-Dorothée et al., 2010, Morgan, 2014) as well as environmental heterogeneity in productivity, local fecundity and larval mortality (White et al., 2014) can profoundly influence the patterns and magnitude of connectivity among populations. Post-settlement processes can also alter patterns of connectivity if recruits do not survive to reproduce or have reduced fitness. Connectivity thus represents the integration of both biological and physical processes and involves complex interactions between benthic and pelagic forces.

Measuring connectivity in marine organisms with small propagules that drift for various lengths of time in the ocean is extremely difficult. Although the length of time larvae drift in ocean currents has been estimated, how that translates to distance and direction traveled and thus connectivity among populations is not well understood. Because of its obvious importance in ecology, evolution and conservation, a number of techniques have been developed to estimate dispersal based on a variety of complementary approaches including larval ecology, hydrographic models, coupled bio-physical models and empirical estimates based on genetics and geochemistry (reviewed in Levin, 2006, Levin, 2006, Thorrold et al. (2007), Cowen and Sponaugle (2009), Lowe and Allendorf (2010), Leis et al. (2011) and Kool et al. (2013)). Each technique has inherent advantages and disadvantages such that a combined approach, when possible, provides a more accurate estimate of connectivity and a more complete understanding of the forces that shape patterns of connectivity (Levin, 2006, Levin, 2006, Lowe and Allendorf, 2010, Leis et al., 2011).

While considerable advances have been made in estimating dispersal in shallow-water ecosystems (e.g. Kinlan and Gaines, 2003; Bradbury et al., 2008; Shanks, 2009; Selkoe and Toonen, 2011; López-Duarte et al., 2012; Riginos et al., 2014), much less is known about dispersal in the deep ocean, except perhaps around hydrothermal vents (e.g. Marsh et al., 2001; Mullineaux et al., 2002, Mullineaux et al., 2010, Mullineaux et al., 2013; Adams and Mullineaux, 2008, Adams and Mullineaux, 2008; McGillicuddy et al., 2010). Recent estimates of the scale of dispersal in deep-sea organisms based on Planktonic Larval Duration (PLDs – e.g. Young et al., 2012) or genetic patterns of isolation by distance (Baco et al., 2014) suggest larvae can disperse 100s of km and are quite similar to shallow-water organisms in dispersal distance, despite lower temperatures that likely extend PLDs (e.g. O’Connor et al., 2007; Peck et al., 2007; Kelly and Eernisse, 2007) and weaker current velocities typically associated with increasing depth. In the deep sea, few have estimated dispersal based on physical processes (e.g. Yearsley and Sigwart, 2011; Young et al., 2012; Sala et al., 2013) and rarely have the predictions from physical transport models and/or PLDs been compared to inferences derived from patterns of genetic variation (e.g. Henry et al., 2014). Such a combined approach provides a powerful framework to test the validity of various hypotheses and provides a more complete understanding of the potential explanations for observed patterns of genetic variation (e.g. Siegel et al., 2003;Sotka et al., 2004; Baums et al., 2006; Galindo et al., 2006, Galindo et al., 2010; Weersing and Toonen, 2009, Weersing and Toonen, 2009; White et al., 2010; Alberto et al., 2011; Sunday et al., 2014). Here we use simulated Lagrangian particle trajectories based on an eddy-resolving ocean general circulation model (specifically FLAME – Family of Linked Atlantic Model Experiments) to estimate the nature and scale of dispersal at bathyal depths in the western North Atlantic and to test the potential role of the Deep Western Boundary Current (DWBC) in explaining patterns of genetic variation on the continental margin.

In particular, we test the hypothesis that the strong genetic divergence at bathyal depths in the western North Atlantic might be due to the DWBC impeding gene flow between upper and lower bathyal depths. Population genetic analyses of protobranch bivalves have repeatedly indicated strong genetic breaks along depth gradients in the western North Atlantic such that populations above and below approximately 3000 m are highly divergent (Etter et al., 2005, Zardus et al., 2006, Jennings et al., 2013, Glazier and Etter, 2014). Populations separated by as little as 40 km distance and 100 m depth exhibited pronounced divergence at multiple loci (see Fig. 1). The strong divergence at such small scales is very surprising because dispersal over that distance is likely within the dispersal window of their lecithotrophic larvae (Zardus, 2002, Scheltema and Williams, 2009) and no obvious topographic features exist in this region that would impede gene flow.

The present day DWBC flows equatorward along the continental slope between 700 and 4000 m (Fig. 2) with mean flows of 5–10 cm s⁻¹ (Pickart and Watts, 1990, Pickart and Watts, 1990, Toole et al., 2011). The genetic break occurs within the DWBC, so one possible explanation for the observed divergence is that the relatively strong mean southwestward flow of the DWBC entrains passively (or weakly swimming) dispersing demersal larvae advecting them equatorward, preventing cross slope dispersal between the upper/middle slope and the lower slope or abyss. Despite the relatively strong mean flows oriented along the isobaths, neutrally buoyant free-drifting floats released at depth and Lagrangian simulations of particle trajectories in the DWBC suggest that circulation is highly complex and departs considerably from time-averaged mean southwestward flow (Bower and Hunt, 2000a, Bower and Hunt, 2000b, Bower et al., 2009, Bower et al., 2011, Bower et al., 2013, Lozier et al., 2013). Both empirical and simulated trajectories indicate considerable mesoscale variability with a high potential of cross-slope movement (i.e. between depth regimes), especially where the DWBC interacts with the Gulf Stream and its associated eddies. The highly complex mesoscale flows and the high probability of floats and simulated particles to move across-slope raises questions about whether the DWBC impedes larval exchange between upper and lower bathyal populations. To test the hypothesis that connectivity between populations from upper and lower bathyal depths might be precluded by the DWBC, we estimated dispersal trajectories of simulated passive larvae (neutrally buoyant particles) released along a depth gradient adjacent to our genetic samples in the western North Atlantic (Fig. 1). Specifically, we examined whether particles would move across slope and the distance they moved given certain PLDs.