Introduction

Predicting dispersal among marine populations is an important aspect of modelling population dynamics and levels and patterns of genetic diversity (Coleman et al. 2011). Such information is of particular importance for marine ecosystems such as estuaries due to major declines in fishery stocks and rapid degradation of natural coastal habitat (Cowen et al. 2006). Studies measuring metapopulation dynamics and connectivity are common for many terrestrial organisms but are comparatively rare for marine environments (Bradbury et al. 2008a), particularly for benthic invertebrates. Identifying patterns of connectivity in the marine realm can assist in the identification of sources and sinks of larval dispersal (Coleman et al. 2011), and is crucial for optimising the design of effective marine protected areas (MPAs). There are, however, long-standing assumptions about long-distance dispersal ability for species with planktonic larval stages (Bradbury et al. 2008a, b) that can lead to the overestimation of dispersal potential (Cowen and Sponaugle 2009) and can thus misinform the design of protected areas.

Estuaries provide a number of ecosystem services including water treatment such as filtering by suspension feeders and detoxification by submerged vegetation and wetlands (Barbier et al. 2010). Estuaries are also important nursery habitats for many organisms and a source of fishery stocks (Bradbury et al. 2008a). Estuarine health can suffer from harbour and coastal development, pollution, overfishing and overharvesting, freshwater diversions and dredging (Kennish 2002). Projected sea level rise with a changing climate will cause the structure and composition of estuarine communities to change as species disperse or disappear (Jenkins et al. 2011). Understanding the factors influencing the connectivity of estuarine species—especially those towards the base of the trophic chain—can help in mitigating anthropogenic damage to estuaries by identifying whether specific areas should be protected.

In this study, we used phylogeographic approaches to test for evidence of restricted gene flow in two species of polychaete worms from the family Nephtyidae (Aglaophamus australiensis and Nephtys longipes) commonly found in estuaries along the coast of New South Wales (NSW), Australia. These species are found only in estuaries and are absent from the NSW open coast. Dispersal of polychaetes can be influenced by abiotic and biotic factors such as ocean currents, larval survival (Sherman et al. 2008) and bathymetry (Piggott et al. 2008; Bors et al. 2012). Polychaetes with planktonic larvae are dependent on environmental factors for dispersal and have been variously found to show both high (Jolly et al. 2004; Barroso et al. 2010) and low (Kesäniemi et al. 2012) levels of connectivity among populations. Although both of our study species have a planktonic larval phase, they also have discontinuous distributions and might not readily disperse among populations, particularly as the predominantly southward flow of the East Australian Current (EAC) is often interrupted by eddies (Coleman et al. 2013) that could isolate some populations (Fig. 1). Estuaries can also represent eco-physiological boundaries to marine organisms due to their somewhat restricted connection to the ocean (Bilton et al. 2002; Kennish 2002); this intermittent connection to the sea could promote population differentiation. We hypothesised that populations of the same species might therefore show strong genetic differences, indicating limited connectivity for these benthic invertebrates among estuaries along the southeastern coast of Australia.

Fig. 1
figure 1

Schematic of flow directions of major currents in southeastern Australia in winter (left) and summer (right) (modified and generalised from Figure 1 in Coleman et al. 2013). The EAC is stronger and extends further south in summer, whereas in winter, coastal flows in southern NSW can be dominated by eddies. The spatial extent of sites used in this study is indicated by a thick black line along part of the southeastern Australian coast

Full size image

Methods

Sampling

In February and March, 2013 and 2014, specimens of A. australiensis and N. longipes were collected from estuaries along the coast of NSW, southeastern Australia. For each species, samples were collected from estuaries separated by at least 30 km (and up to 590 km). Samples of N. longipes were collected from four estuaries, and A. australiensis from ten estuaries (Table 1; Figs. 2 and 3). To allow assessment of fine scale, within-estuary structure, samples of A. australiensis were also collected from four sites within the connected Pittwater/Hawkesbury estuaries (Table 1; Fig. 4). Intertidal benthic sediment was collected at low tide using a spade and then gently sieved using a 0.5-mm mesh. All samples were collected towards the mouth of estuaries in full marine salinities. N. longipes was generally found closer to the ocean than A. australiensis, and only in sites with cleaner sand, whereas A. australiensis was found in muddier sand usually associated with Zostera seagrasses. N. longipes was only found in the four estuaries listed for that species in Table 1. Polychaetes were identified to species level and preserved in 95 % ethanol within approximately an hour of collecting, and alcohol was changed at least twice in the 24 h following collection.

Table 1 List of field sites with species collected and GPS coordinates, and numbers of individuals sequenced for each marker
Full size table
Fig. 2
figure 2

Sites (a) and haplotype networks for COI, (b) and 16S (d) for Aglaophamus australiensis (pictured in c [Photo by K. Atkinson])

Full size image
Fig. 3
figure 3

Sites (a) and haplotype networks for COI, (b) and 16S (d) for Nephtys longipes (pictured in c [photo by S. Lindsey]. Note the much longer anterior chaetae than for A. australiensis)

Full size image
Fig. 4
figure 4

Sites (top) and COI haplotype network (bottom) for Aglaophamus australiensis samples from four sites in the Pittwater/Hawkesbury estuaries, north of Sydney

Full size image

DNA amplification

We sequenced parts of the mitochondrial cytochrome c oxidase subunit I (COI), mitochondrial ribosomal (16S) and nuclear ribosomal (28S) genes. Although more slowly evolving than microsatellites, these markers can nonetheless shed light on phylogeographic structure (and hence on dispersal capacity over long time frames) in polychaetes. For example, COI was used by Carr et al. (2011) to determine phylogeographic structure in polychaetes in the Pacific, Arctic and Atlantic Oceans, and by Meissner et al. (2014) to infer dispersal among polychaetes North Atlantic seamounts. COI and 16S have also been used to infer past connectivity for polychaetes in Europe (Jolly et al. 2006) and the New Zealand region (Bors et al. 2012), and Schüller and Hutchings (2012) used 16S to infer dispersal for a trichobranchid polychaete in the Southern Atlantic.

DNA extraction of a small section of worm (approximately 1 mm) was carried out using a standard Chelex extraction procedure (Walsh et al. 1991). All samples used were required to have a head to avoid sampling the same individual twice. PCR amplification of the three markers was achieved using published primers (Table 2). Various other primer pairs were tested for each marker, but the greatest success was achieved with those shown in Table 2. PCRs were carried out using an Eppendorf Flexlid Mastercycler Nexus thermocycler in 20 µl volumes containing 1.0 µl of DNA, 1.0 µl of 10 µM of each the forward and reverse primer, 2.0 µl buffer, 2.0 µl of 8 mM dNTPS, 1.0 µl of 25 mM MgCl2, 0.2 µl of 5 U/µl Taq and 11.6 µl water, using the following program: 94 °C for 120 s, then 40 cycles of 94 °C for 15 s, 48 °C for 30 s and 72 °C for 60 s, finishing with 72 °C for 240 s. Amplified PCR product was purified using IllustraTM ExoProStar ‘enzymatic PCR and sequencing clean-up’ kit and sequenced by Macrogen (Korea).

Table 2 List of primers used for amplifying COI, 16S and 28S markers in A. australiensis and N. longipes
Full size table

Genetic analyses

Sequences were aligned using Geneious 6.1.6 (Biomatters), and any ambiguities were assessed by eye. Haplotype networks were constructed using TCS (Clement et al. 2000). Isolation by Distance (IBD) analyses were carried out for each species for COI and 16S by paired Mantel tests using 999 permutations in GenAlEx (Peakall and Smouse 2012). For the Mantel test, the geographic distances among sites were calculated using a Coordinate Distance Calculator (http://boulter.com/gps/distance/), and mean uncorrected pairwise distances between sites were calculated using MEGA (Tamura et al. 2011). AMOVA analyses were used to compare within versus between population diversity for each species and marker, implemented in Arlequin version 3.5.1.3 (Excoffier and Lischer 2010). Directionality in gene flow between populations of the most widespread taxon, A. australiensis, was investigated using COI and 16S data via maximum likelihood in Migrate-n 3.6.6 (Beerli 1998; Beerli and Felsenstein 2001). The transition/transversion ratios were set to 16.70 for COI and 1.88 for 16S as calculated by Arlequin. The inheritance scalar was set to 0.25 for both loci. The four populations from the Pittwater/Hawkesbury estuaries were combined. The DNA sequence model was run with 10 short chains, each sampling 100,000 genealogies, and three long chains, each sampling 1000,000 genealogies, with a burn-in of 10,000 per chain. Three runs for both a full migration (asymmetric dispersal) model and a symmetric migration model were carried out, and a likelihood-ratio test (LRT) was used to assess which model best fit the data [LRT = 2* (lnL(asymmetric model) − lnL(symmetric model)], where lnL is the natural log of the likelihood).

Results

A fragment of COI 594 bp long was successfully amplified for 63 samples of N. longipes (yielding 39 haplotypes), and a fragment 340 bp long was amplified for 258 samples of A. australiensis (yielding 56 haplotypes). A fragment of 16S 397 bp long was amplified for 38 samples of N. longipes (yielding seven haplotypes), and a fragment 423 bp long was amplified for 149 samples of A. australiensis (yielding 26 haplotypes). The differences in lengths of the COI and 16S fragments amplified for each species were the result of differences in sequencing success; unreadable ends due to poor sequence quality in some samples were trimmed from all alignments, resulting in shorter fragments for some data sets. 28S only amplified for eleven samples in A. australiensis, and none in N. longipes. The 11 sequences obtained for 28S (eight from Shadracks Creek, one from Tuross Heads and two from Wallaga Lake), each with a length of 372 base pairs after trimming of ambiguous ends, showed no variation. All unique sequences from this study have been deposited with GenBank (accessions: A. australiensis COI: KP836357-412; 16S: KP836413-438; 28S: KP860235-245; and N. longipes COI: KP792237-275: 16S: KP836439-445).

No convincing pattern of IBD was found for either species (P > 0.05); for N. longipes, neither marker showed a significant IBD trend (COI: P = 0.479; 16S: P = 0.095), and although a slight IBD trend was found for A. australiensis for 16S (R 2 = 0.192, P = 0.031), this pattern was not observed for the more informative COI marker (P = 0.273). IBD plots are shown in supplementary figures S1-S4. Network analysis (Figs. 2, 3, 4) also did not indicate any clustering of haplotypes according to geography; indeed, several haplotypes were shared between multiple sites in each species. A single haplotype shared by one individual of A. australiensis from Wallaga Lake, and one from Dangar Island, could not be joined to the COI network at the 95 % confidence limit (Fig. 2). Network analysis indicated considerable diversity within sites; for example, 19 distinct COI haplotypes were recovered from 20 samples of N. longipes from Cuttagee Beach (Fig. 3). Haplotype diversity values were between 0.78 and 0.99 for both species and all sites (supplementary tables S1–S2). No evidence of directional dispersal was found, with Migrate-n analysis of A. australiensis sequences indicating in all three runs that a symmetric dispersal model fits the data better than an asymmetric model (P < 0.001).

AMOVA analyses showed that within population variation was far greater than between population variation (COI: A. australiensis: within 99.05 %, between 0.95 %, P > 0.05; N. longipes: within 96.24 %, between 3.76 %, P = 0.025; 16S: A. australiensis: within 93.34 %, between 2.92 %, P = 0.047; N. longipes: within 88.29 %, between 11.71 %, P = 0.045). Significant population differentiation was detected for both species for both markers. For N. longipes, population differentiation was entirely driven by the Dolls Point population differing from the other three estuaries. This population did, however, share haplotypes for COI and 16S with all of the other estuaries sampled for this species (Fig. 3). For A. australiensis, most population pairwise differences in the 16S marker were between Wallaga Lake and other sites (Table 3), but more pairwise differences were detected for COI (Table 4). For example, the two most southern sites, Wallaga Lake and Shadracks Creek, were different to most sites from Taren Point northward. Nonetheless, as for N. longipes, several of the most common haplotypes for A. australiensis were found across almost all sites, including the most northern and most southern sites (Fig. 2), and differences therefore appear to be largely the result of diversity in rare (e.g. singleton) haplotypes at each site. Within the connected Pittwater and Hawkesbury estuaries, no fine-scale genetic structure was observed among sites, with several of the same haplotypes found at all four sites (Fig. 4) and no significant population differentiation found by AMOVA (P ≫ 0.05).

Table 3 Population pairwise FST values for A. australiensis for 16S. Pairwise differences that were significant (P < 0.05) are in bold and underlined
Full size table
Table 4 Population pairwise FST values for A. australiensis for COI. Pairwise differences that were significant (P < 0.05) are in bold and underlined
Full size table

Discussion

Our results suggest that dispersal of planktonic polychaete larvae can readily occur among many estuaries in southeastern Australia. Although AMOVA analyses indicated differentiation among a few sites, these differences were apparently driven by relatively rare haplotypes in a few individuals, whereas the finding that several common haplotypes were shared among most sites for both species and markers suggests widespread inter-estuarine dispersal. Our hypothesis that populations might be poorly connected was thus not supported by our results.

Although several other molecular studies on polychaete worms have identified the presence of unrecognised species (e.g. Barroso et al. 2010; Nygren and Pleijel 2011; Borda et al. 2013; Glasby et al. 2013), the close relationships of haplotypes from each marker for each species in this study suggest their taxonomy is well resolved across the study region. Indeed, although Nephtyidae worms in southeastern Australia have recently been the focus of taxonomic revision, the two species used in this study appear valid (Dixon-Bridges et al. 2014).

Polychaetes are abundant in marine and estuarine environments (Hutchings 2004), and the widespread distributions of some polychaete species are probably due to their long pelagic larval stages. There is no literature on the larval duration of N. longipes and A. australiensis; however, the duration of larval stages in other members of the Nephtyidae is estimated to be 11–42 days (Caron et al. 1995). This is a relatively long planktonic phase, consistent with the potential for widespread dispersal suggested by our study. The average distance among estuaries in NSW (south of Sydney) is approximately 10 km (estimated from Google Earth path measurements), and the EAC can have coastal flows of ~1 ms−1 (Roughan and Middleton 2004). Larval dispersal among adjacent estuaries could therefore potentially occur in a matter of a few hours or days, once the larvae have been flushed by tides into the sea.

Several other studies have likewise found evidence of high dispersal capacity in polychaete worms. Barroso et al. (2010) found that populations of Eurythoe complanata appeared to be well connected along 2500 km coast of the Caribbean and Brazil and proposed that the teleplanic larvae of E. complanata were being transported by the westward South Equatorial Current and the eastward South Equatorial Countercurrent (Brown 1990). Widespread larval dispersal facilitating population connectivity was also inferred for the polychaete Pectinaria koreni along the north coast of France (Jolly et al. 2004), with evidence for past divergence due to historical processes but contemporary dispersal with ocean currents. Polychaetes with pelagic larvae do not, however, always show evidence of connectivity among populations. Kesäniemi et al. (2012) found, for example, that many European populations of the spionid polychaete Pygospio elegans were genetically distinct. Life history alone is inadequate to predict dispersal capacity (Johannesson 1988), and in the case of our study, tidal flushing, ocean currents and anthropogenic influences are likely to play a major role in the transport of larvae.

Despite estuaries potentially acting as eco-physiological barriers (Bilton et al. 2002; Kennish 2002), ballast water from ships can transport marine organisms long distances, particularly among ports (Hutchings 1992; Ruiz et al. 1997). At least one of the sites in our study was close to a busy industrial port (Dolls Point, near Botany Bay), and all other sites, while not major ports, are estuaries that are open to the sea and experience considerable tidal influences. Coleman (2013) found that estuarine and nearby open-ocean populations of the kelp Ecklonia radiata in southeastern Australia were not genetically differentiated, suggesting that well-flushed estuaries are not isolated environments for benthic marine organisms. Indeed, for many of the estuaries studied, the environment is almost entirely marine, with relatively little freshwater input from streams and rivers, and strong tidal flushing via fast-flowing currents around sand banks. The entrance to the Pittwater/Hawkesbury River estuaries (location of sites Dangar Island, Patonga, Careel Bay and Bayview), Sydney Harbour (Fig Tree Bridge site) and Botany Bay (Taren Point site) are all well-flushed estuaries with wide, deep river mouths. However, sites such as Wallis Lake and Wallaga Lake have shallow entrances, partially blocked by sand bars that limit flushing which varies between neap and spring tides (Roy et al. 2001). Future research could assess whether less well-flushed estuaries, such as those that are often completely closed by sand movements at the river mouth, or by anthropogenic sea walls, house more differentiated polychaete populations than ‘open’ estuaries.

The high dispersal potential of these two estuarine worm species has positive implications for their resilience in the face of ongoing anthropogenic environmental change, as damaged populations may be able to be ‘rescued’ by colonists from elsewhere (Bradbury et al. 2008a). Which populations would be most likely to act as sources of such rescues could not, however, be determined by this research. Although the EAC flows southward along the eastern coast of Australia, and thus could be expected to connect from north to south only, bidirectional dispersal may also be facilitated by inshore, north-flowing counter currents and eddies due to seasonal variation of strength and positioning of the EAC (Coleman et al. 2011, 2013; see Fig. 1). Indeed, our Migrate-n analyses suggest that bidirectional, rather than asymmetric, dispersal of polychaete larvae is occurring along the NSW coast. Furthermore, if north–south dispersal were the norm, genetic diversity could be expected to decrease towards the south, but this pattern was not supported by haplotype diversity values (see supplementary tables S1, S2).

In concordance with our results, Piggott et al. (2008) did not find a significant IBD relationship in abalones (Haliotis coccoradiata) along the south coast of NSW. There was, however, additional evidence of fine-scale genetic structure, provided by microsatellite analysis, which suggested some degree of local larval retention and local recruitment (Piggott et al. 2008). The authors were able to conclude that the most likely scenario was that abalones have the ability to infrequently disperse over long distances, although recruitment occurs primarily on a small spatial scale. Comparable results were found using microsatellite loci for populations of the kelp Ecklonia radiata along the eastern coast of Australia, with no IBD pattern but mosaic genetic differentiation apparently driven by eccentricities in ocean current flow (Coleman et al. 2011). Rapidly evolving molecular markers such as microsatellites are ideal for analysing patterns of contemporary connectivity among marine populations (Sherman et al. 2008; Chust et al. 2013). Unfortunately, polymorphic microsatellite loci have not yet been identified for our study species (nor for any species in the polychaete family Nephtyidae), and our research therefore relied on the more slowly evolving, but nonetheless phylogeographically informative, mitochondrial markers COI and 16S. Although we were able to infer connectivity among estuaries, we cannot be sure whether this connectivity is ongoing or is simply a reflection of processes in the recent past. Future research using microsatellite markers could shed light on the finer-scale population structure and connectivity of Nephtyidae worms in eastern Australia. Research on the life history of these polychaetes, such as their larval duration and length of spawning, would also help us to understand the likely extent of their dispersal capacity. Nonetheless, our results indicate that benthic polychaetes with planktonic larvae may have little trouble dispersing among estuaries connected by strong ocean currents. For less dispersive estuarine species, such as those without a planktonic stage, and for organisms in poorly flushed estuaries, dispersal is likely to be more of a challenge. Research on a wide range of taxa with differing dispersal mechanisms is critical if the design of protected areas such as marine park networks is to be effective.