Introduction

Connectivity is a broad term used to describe the extent to which populations in different parts of a species’ range are linked by the exchange of larvae, recruits, juveniles or adults (Palumbi, 2003; Cowen and Sponaugle, 2009; Weersing and Toonen, 2009; Lowe and Allendorf, 2010), and it can be generally divided into genetic connectivity and demographic connectivity. Genetic connectivity refers to the degree to which gene flow affects evolutionary processes within populations, whereas demographic connectivity refers to the degree to which population growth and vital rates are affected by dispersal and recruitment (Lowe and Allendorf, 2010). Understanding connectivity is important because it controls the following: (1) a population’s buffering potential from local catastrophes and therefore its extinction risk (Allison et al., 2003); (2) a population’s potential as a source of new individuals to other populations; (3) the level of genetic mixing between populations (Bell and Okamura, 2005; Bell, 2008); and (4) a population’s susceptibility to disease or pollution, the so-called ‘dark-side’ of connectivity (Hughes et al., 2010).

Although measuring genetic and demographic connectivity provides insights into patterns of dispersal in the marine environment, it is usually demographic connectivity that is of most use to resource managers, as it provides information relevant to the typically short time scales at which management operates (years to decades). This is particularly the case for spatially explicit management strategies such as protected areas networks, where protection success is usually dependent on whether the design accurately reflects ecologically relevant patterns of connectivity (Palumbi, 2003; Shanks et al., 2003; Berumen et al., 2012).

Despite the importance of connectivity in marine systems, it is poorly understood in most species, as it is difficult to estimate (Kinlan and Gaines, 2003; Cowen et al., 2007; Selkoe and Toonen, 2011). Demographic connectivity is particularly difficult to measure directly because the majority of marine species (particularly invertebrates) have a bi-partite life cycle comprising a relatively sessile adult stage and a dispersing larval stage. The duration of the larval stage ranges from several hours to years, making dispersal technically and logistically challenging to track (Leis et al., 2011; Selkoe and Toonen, 2011). Although genetic data are more easily collected, the nature of data often makes it difficult to interpret in a management context (Lowe and Allendorf, 2010; Selkoe and Toonen, 2011). A better approach to understanding connectivity patterns is to provide connectivity estimates using a combination of different approaches (Cowen et al., 2007). Although a large number of population genetic studies have been conducted across the world for marine species to estimate genetic connectivity (see Kinlan and Gaines, 2003; Cowen et al., 2006; Bradbury et al., 2008; Gardner et al., 2010; Kelly and Palumbi, 2010 who consider hundreds of species), there are far fewer estimates of demographic connectivity based on either direct or indirect observation (but see Swearer et al., 1999; Jones et al., 2005; Becker et al., 2007 for individual species studies). Further, there have been few studies that have validated connectivity models through complementary approaches (but see Thorrold et al., 2002; Jones et al., 2005).

Importantly, the development of new genetic tools and analysis methods, particularly highly polymorphic microsatellite markers (Selkoe and Toonen, 2006), is enabling genetic data to be used to estimate demographic connectivity, particularly through assignment testing (Berry et al., 2004; Manel et al., 2005; Underwood et al., 2007; Saenz-Agudelo et al., 2009; Polato et al., 2010). These developments provide opportunities for using genetic tools to estimate both the genetic and demographic connectivity and present opportunities to compare such data with mathematical models.

Jasus edwardsii supports valuable commercial fisheries in both New Zealand (NZ) and Australia, worth over New Zealand Dollar (NZD) 230 million per annum to the New Zealand economy alone. It is a keystone predator of subtidal rocky reef systems (Shears and Babcock, 2003) with a range that encompasses 10 degrees of latitude and stretches over 5500 km from the Chatham Islands in New Zealand to Western Australia. Importantly, J. edwardsii is thought to have one of the longest pelagic larval durations of any marine species and can remain in offshore oceanic waters for up to 24 months, where it undergoes 11 developmental stages with 17 instars before developing into the nektonic puerulus that settles in coastal environments (Booth and Phillips, 1994).

Previous analysis of mitochondrial DNA has shown that J. edwardsii populations are genetically indistinguishable between and within Australia and New Zealand (Ovenden et al., 1992). This suggests that populations were isolated too recently to allow genetic divergence, that there is sustained gene flow between Australia and New Zealand homogenizing populations or that it is the result of the conserved nature of mitochondrial DNA that may have prevented the identification of genetic structure (Selkoe and Toonen, 2006). However, although the existing genetic data suggest homogeneity, extensive oceanographic modeling studies for J. edwardsii suggest that some source-sink dynamics exist. These models have estimated New Zealand-scale patterns of connectivity (Chiswell and Booth, 2008) and also the supply of recruits from Australia to New Zealand (Chiswell et al., 2003). Chiswell and Booth (2008) used a Lagrangian modeling approach, applying time varying ocean currents (1993–2003) derived from a satellite altimeter to track the paths of individual J. edwardsii larvae released from different locations around New Zealand. The aim of their study was to identify source-sink relationships between the different rock lobster management areas (CRA zones of which there are currently nine). The results from the model showed significant exchange among most regions, and the authors identified four major geographic areas associated with rock lobster larval dispersal around New Zealand: (1) the far north, (2) the east coast of the North Island, (3) the South Island and (4) the Chatham Islands. In addition to these four geographic regions, Chiswell and Booth (2008) identified high levels of local settlement at the bottom of the South Island and suggested that populations in this region are likely to be maintained by self-recruitment. Using a similar modeling approach, Chiswell et al. (2003) investigated trans-Tasman dispersal rates and estimated that 9–14% of J. edwardsii larvae originating from the south east coast of Australia would be able to reach the west coast of New Zealand, which was used to explain the genetic homogeneity for J. edwardsii on the basis of mitochondrial DNA data (Ovenden et al., 1992).

In this study, we reassessed the genetic structure of J. edwardsii using a recently developed panel of eight microsatellite markers (Thomas and Bell, 2011) to determine levels of genetic differentiation and gene flow between populations at the New Zealand scale and also the genetic relationships between Australia and New Zealand. Specifically, we used our genetic data to provide estimates of contemporary rates of gene flow and estimates of demographic connectivity to test the findings of the earlier oceanographic models that predicted dispersal patterns of J. edwardsii across the Tasman Sea (Chiswell et al., 2003) and throughout New Zealand (Chiswell and Booth, 2008). Given that the results from the New Zealand scale model (Chiswell and Booth, 2008) revealed high levels of larval exchange between management areas throughout New Zealand, we hypothesized that the genetic structure of J. edwardsii would reflect those findings and reveal a relatively homogenous population. One possible exception to this might be found at the bottom of the South Island, for which there is evidence for high levels of self-recruitment. Furthermore, we propose that the resolving power of highly polymorphic microsatellites would identify a significant trans-Tasman genetic structure that has previously remained undetected using more conservative mDNA markers (Ovenden et al., 1992).

Materials and methods

Sampling

A total of 336 adult lobsters from eight populations were collected by commercial fishers between January and November 2011. Sample sizes ranged from 45–48 individuals per population and were all from adult individuals of minimum legal size. Samples were collected from five of the nine management zones (termed ‘CRA’ zones) in New Zealand (CRA 2, 4, 5, 6 and 8), one population in the Southern Zone (SZRLF) of the South Australia Rock Lobster Fishery and one population from the Tasmania Fishery zone (Figure 1). Geographical coordinates and the corresponding management zone from where the samples were collected are provided in Table 1. Samples covered roughly 10 degrees of latitude and over 3500 km across the Tasman Sea to South Australia.

Figure 1
figure 1

Sampling locations for J. edwardsii throughout New Zealand and on the Australian coast.

Full size image
Table 1 Sampling locations with standard genetic diversity indices: (N) number of individuals sampled; (NA) mean number of alleles per locus; (HO) observed and (HE) expected heterozygosity; and (FIS) inbreeding coefficients with corresponding significant values (PHWE)
Full size table

Laboratory protocol and PCR amplification

Tissue samples for DNA extraction were taken from the pereiopods and antennae preserved in 99% ethanol, and the remains were stored in 50 ml polypropylene test tubes at −80 °C. DNA for PCR amplification was extracted using a modified phenol–chloroform protocol and genotyped across eight microsatellite loci (for details see Thomas and Bell, 2011). Samples were screened using a Touchdown PCR protocol (Korbie and Mattick, 2008). Amplifications were performed on a GENEAMP 2700 (Applied Biosystems, Foster City, CA, USA) thermocycler in a final reaction volume of 12.5 μl, containing 6.75 μl MyTaq Red Mix 2X Bioline, 0.75 μl of the forward and reverse primers, 50–100 ng of DNA template and ddH20 to reach the final volume. PCR products were viewed on a 2% agarose gel. Genotyping was conducted on an ABI 3730XLs sequencer (Applied Biosystems) and loci were scored by eye using 5′ fluorescent labels with a 500LIZ size standard in GENEMARKER v 1.97 (www.softgenetics.com).

Genetic analyses

MICRO CHECKER V 2.2.3 (van Oosterhout et al., 2004) was used to identify deviations from Hardy–Weinberg equilibrium resulting from large-allele drop out, scoring errors due to stuttering or the presence of null alleles. Genetic diversity of sampling locations was quantified using the mean number of alleles per locus and total number of alleles by FSTAT v 2.9.3 (Goudet, 1995) and number of private alleles and Nei’s unbiased heterozygosity (Nei, 1987) by GENEALEX 6.3 (Paekall and Smouse, 2006). We also preformed exact tests for deviations from Hardy–Weinberg equilibrium with GENEALEX v 6.3 and tested for any evidence of linkage equilibrium with ARLEQUIN v 3.5.1.2 (Excoffier et al., 2005) P-values were adjusted for multiple comparisons using standard Bonferroni corrections (Rice, 1989). Markov-chain parameters were 10 000 dememorization steps, 1000 batches and 10 000 iterations per batch.

Population genetic structure

To determine the proportion of genetic variation that could be attributed to differences between sampling sites, hierarchal analysis of molecular variance (n=10 000 permutations) was performed by GENEALEX v 6.3. Estimates of population differentiation (FST; Weir and Cockerham, 1984) were also measured across all loci with ARLEQUIN v 3.5.1.2 using sampling locations as population units. Significance values were based on 10 000 permutations. To account for the high levels of within-population diversity, values were standardized to a scale of 0-1 (F′ST) according to the method described by Meirman (2006) using RECODEDATA (Meirman, 2006). There is much debate as to whether the stepwise mutation model and associated RST index are appropriate for population genetic studies using microsatellite loci (Meirmans and Hedrick, 2011); therefore, we avoided using the RST index as it is unlikely that the loci used in this analysis adhered to a strict stepwise mutation model, given the imperfect nature of the repeats (see Thomas and Bell, 2011). A Mantel test for isolation by distance was conducted using linearized values of F′ST (F′ST/(1−F′ST)) and oceanographic distance (km).

To visualize the genetic relationships between sampled populations, a principle coordinate analysis was conducted by GENEALEX 6.3 using the pairwise matrices of FST, F′ST, and the genotype likelihood ratio distance, DLR. This measurement of genetic distance (DLR: Paetkau et al., 1995) was developed from assignment tests in which the likelihood of complete multi-locus genotypes are compared across individuals from discrete populations and was determined to be a better predictor of statistical power than FST (Paetkau et al., 1997).

Bayesian clustering analysis in STRUCTURE v 2.3.2 (Pritchard et al., 2000) was used to infer population structure using the admixture model with correlated allele frequencies among clusters and informed priors with a burn-in period of 104 iterations and 105 Markov Chain Monte Carlo repetitions and with the K value ranging from 1–8. LOCPRIOR models (Hubisz et al., 2009) consider sampling locations to be informative about ancestry and are best suited for instances when genetic structure is present but at relatively weak levels (Pritchard et al., 2010). Each individual in the data set is represented by a single vertical line, which is partitioned into K segments that represent that individual’s estimated membership fraction in each of the K-inferred clusters. The appropriate K value for the data set was determined by plotting the log probability (L(K)) and ΔK across multiple runs (Evanno et al., 2005) as implemented in STRUCTURE HARVESTER (Earl and vonHoldt, 2012). Results from 10 runs were merged with CLUMPP (Jakobsson and Rosenberg, 2007) and visualized using DISTRUCT (Rosenberg, 2004). Analysis of molecular variance (n=10 000 permutations) was then used to test the results of STRUCTURE using GENEALEX v 6.3 and determine the proportion of genetic variation that was attributed to differences between clusters. In addition, standardized fixation indices were calculated between the clusters.

Trans-tasman dispersal

To investigate Trans-Tasman connectivity and determine the likely proportion of the New Zealand lobster to have originated in Australia, assignment tests were conducted using a Bayesian approach (Rannala and Mountain, 1997) in GENECLASS v 2.0 (Piry et al., 2004) on the basis of the clustering scenario identified by STRUCTURE. An analysis of first generation migrants was conducted using an exclusion threshold approach where individuals were excluded from their corresponding sampling site when probability of assignment to the reference population was less than 0.05 (Type I error; Berry et al., 2004). This exclusion approach is likely to be more accurate than the ‘leave one out’ methodology by Paetkau et al. (2004) because it does not require that all true source populations be sampled. Excluded individuals were then reassigned to a source population when the probability of assignment was greater than 10%. When an excluded individual was reassigned to more than one population (P>0.10), it was left unassigned. Those individuals that could not be reassigned to any of the other populations were considered to have originated from a non-sampled location.

Trans-Tasman migration rates were determined using the coalescent approach implemented in MIGRATE v 3.2 (Beerli, 2010). Migration rates between clusters as well as effective population sizes were determined under the maximum likelihood strategy (Beerli and Felsenstein, 1999) with variable mutation rates using a Brownian motion model. Results are expressed as M (m/μ), which is the mutation-scaled immigration rate that represents the importance of variability brought into the population by immigration compared with the variability by mutation, and Θ (4Neμ), which is the mutation-scaled effective population size. The parameters were set as the following and averaged across two independent runs: 10 short chains with 1 000 000 genealogies sampled; 3 short chains with 10 000 000 genealogies sampled; and a burn-in of 10 000 genealogies for each chain. Results are presented with 95% confidence intervals.

Selection pressure and effective population sizes

Although it is generally assumed that microsatellite loci are selectively neutral, linkage to regions of the genome that are under selection can often lead to skewed estimates of connectivity (Selkoe and Toonen, 2006). As a result, screening markers to test for neutrality are important for any connectivity study. To identify any outlier loci that may be affected by selection, we applied the method by Beaumont and Nichols (1996) in the selection detection workbench LOSITAN (Antao et al., 2008). This method plots FST values against heterozygosity and tests for conformity to the expected null distribution of Wrights (1951) Island Model. Analysis was performed with 50 000 simulations under both infinite allele and stepwise mutation models with a forced and ‘neutral’ mean FST.

Results

Genetic diversity

MICRO CHECKER found no evidence of large-allele drop out and detected no evidence of scoring error due to stutter or the presence of null alleles. There was no evidence of linkage disequilibrium between any pair of loci across all sample sites following standard Bonferroni corrections. All populations exhibited a significant global heterozygote deficiency (Table 1) with inbreeding coefficients (FIS) ranging from 0.047 to 0.101. Deviations from Hardy–Weinberg equilibrium at the locus level varied across sampling locations, although no population showed a significant deficiency at all loci (Table 1: Supplementary Material). The eight microsatellite loci exhibited a wide range of polymorphism: the number of alleles per locus ranged from 10 to 76 and the number of alleles per population ranged from 109 to 161 (Table 2). Genetic diversity was lowest at Stewart Island across all measures and showed significant heterozygote deficiencies at seven out of eight loci (Table 1; Supplementary Material) but did not exhibit a very strong overall heterozygote deficit relative to other populations. Heterozygosity can often be relatively insensitive to population bottlenecks and can remain high despite a significant loss of allelic diversity (Allendorf, 1986).

Table 2 Allele size variation (standardized allelic richness) and total number of alleles per locus (N) at 8 microsatellite loci
Full size table

Population genetic structure

Our analysis revealed significant population structure of J. edwardsii stock across the Tasman Sea and within New Zealand (FST of 0.018, P=0.010). Although analysis of molecular variance indicated that only a small proportion (2%) of genetic variation could be attributed to differences between sampling sites, significant F′ST values were detected at 19 out of 26 (73%) pairwise comparisons following standard Bonferroni corrections (Table 3). Stewart Island and South Australia populations were significantly different from all other populations sampled. The highest values of genetic differentiation occurred between Kaikoura and South Australia (F′ST=0.118). Pairwise comparisons between the SouthWest Coast, Wellington and Kaikoura were low and non-significant. Interestingly, only two of the sampled locations from New Zealand (Stewart Island and Kaikoura) were significantly different from the Tasmania sampling location. Isolation by distance plots (Figure 1: electronic Supplementary Material) showed a weak but significant positive correlation between F′ST and oceanographic distance (R2=0.16, P=0.037). When the Stewart Island sampling location was excluded from the analysis, the isolation by distance plot showed a much stronger correlation (R2=0.480, P<0.001).

Table 3 Pairwise fixation index values FST (above diagonal) and standardized values F′ST (below diagonal)
Full size table

Principle coordinate analysis plots (Figure 2) show the genetic relationships between the sampled locations, and results were largely consistent across the three measures of differentiation (FST, F′ST and DLR). All plots showed a close relationship among the New Zealand sample sites, except for Stewart Island, which was as different to the New Zealand grouping as South Australia. The Tasmania sample location appeared as an intermediate between the New Zealand and South Australia populations, which is consistent with its geographical location. The main discrepancy between the three measures was the level of separation within the New Zealand group, particularly relating to the Hauraki Gulf sample site. Principle coordinate analysis using standardized values revealed a more tightly clustered New Zealand group than produced by the FST matrix. The separation of the Hauraki Gulf population from the New Zealand cluster was most evident in the principle coordinate analysis based on the DLR distance matrix.

Figure 2
figure 2

Principle coordinate analysis (PCA) implemented in GENEALEX v 6.3 of pairwise (a) FST, (b) F′ST and (c) DLR values between populations.

Full size image

The genetic divisions identified above were further supported by STRUCTURE. The optimal number of clusters according to Evanno et al. (2005) as implemented in STRUCTURE HARVESTER was determined to be K=4 (Figure 3, Figure 2; electronic Supplementary Material): a central New Zealand cluster (Wellington, Kaikoura, Chatham Islands and SouthWest Coast ), a northern New Zealand cluster (Hauraki Gulf), a southern New Zealand cluster (Stewart Island) and an Australian cluster (Tasmania, South Australia). Despite the evidence suggesting that the sampling locations represented four genetically discrete populations, analysis of molecular variance indicated that less than 2% of genetic variation could be partitioned between clusters using this scenario. Standardized fixation indices between clusters were all significant and ranged from 0.036 to 0.106 (Table 4).

Figure 3
figure 3

Bayesian clustering analysis using STRUCTURE 2.3.2 for J. edwardsii sample locations. Sampling locations are along the x axis and membership coefficient in each predefined cluster (K) is along the y axis. K=4 clustering scenario according to STRUCTURE HARVESTER. Results were averaged across 10 runs with CLUMP and visualized with DISTRUCT.

Full size image
Table 4 Hierarchal analysis of molecular variance was used to estimate levels of genetic differentiation between sampling sites (n=8) and clusters (K=4) identified by STRUCTURE
Full size table

Trans-Tasman dispersal

Assignment tests were performed by Geneclass v 2.0 to identify the relative contribution of the Australian stock to the New Zealand fishery. FST values, sample size and the number of markers can provide a measure of confidence to the results of assignment tests, particularly when marker variability is standardized (Cornuet et al., 1999). Although population differentiation was relatively low between sample sites, we used the clustering scenario identified by STRUCTURE (K=4, F′ST=0.072) for the assignment tests, increasing the likelihood of correct exclusion from groups of samples (Cornuet et al., 1999; Berry et al., 2004). As a result, we are confident that this analysis provided an accurate representation of immigration across the Tasman Sea.

Under an exclusion approach, roughly nine percent of the lobster sampled from the central New Zealand locations (SouthWest Coast, Wellington, Kaikoura and Chatham Islands) were identified as first generation migrants on the basis of an a threshold of P<0.05 (Table 5). Three of those individuals were determined to have originated from Australia and one individual near Stewart Island. The majority of excluded individuals, however, originated from non-sampled locations. Two individuals sampled in the far north of New Zealand (Hauraki Gulf) were also identified to have come from Australia. There was no evidence from assignment tests that the far north of New Zealand acts as a larval source for populations to the south.

Table 5 Assignment tests conducted in GeneClass v. 2.0 to determine rates of trans-Tasman connectivity
Full size table

Maximum likelihood trans-Tasman migration rates (M) determined in MIGRATE were low and ranged from 1.54 (to Stewart Island) to 3.13 (to central New Zealand) (Table 6). Migration rates were consistent with results from assignment tests and showed that the central New Zealand cluster received relatively high amounts of gene flow from Australia compared with the other clusters; migration rates to New Zealand were more than two times higher than to Stewart Island. Estimates of the mutation scaled effective population size (Θ) were lowest for Stewart Island (4.76) and highest for Australia (10.87). Hauraki Gulf and the central New Zealand clusters showed similar values. Assuming a mutation rate of 10−4 (Gonzalez and Zardoya, 2007), we were able to translate these values into estimates of Ne that ranged from 1189 (Stewart Island) to 2717 (Australia).

Table 6 Effective population sizes and migration rates determined by MIGRATE
Full size table

Selection pressure and effective population sizes

In order to identify other evolutionary processes other than gene flow that could be influencing the patterns of population structure, an outlier loci analysis to test for selection pressure was conducted in LOSITAN. On the basis of the results from this analysis, there was no indication that any loci developed by Thomas and Bell (2011) were experiencing selection.

Discussion

Despite J. edwardsii having one of the longest larval durations known of all marine species, having a wide distribution around New Zealand and Southern Australia, and being previously found to have genetic homogeneity throughout its range (Ovenden et al., 1992), we found significant genetic structuring and evidence of restricted gene flow between certain regions, suggesting that J. edwardsii is not a single panmictic population. Although the levels of genetic differentiation were low, consistent patterns of differentiation were identified between Australia and New Zealand and also within New Zealand using multiple analysis methods. We also detected a strong pattern of isolation by distance, suggesting that distance is a barrier to gene flow, even though J. edwardsii has one of the longest known pelagic larval durations. Our analyses highlight the need to re-evaluate the management of J. edwardsii stocks and also the need to re-examine other species that are presumed to form panmictic populations on the basis of traditional molecular markers. The failure to accurately identify the biological boundaries of a harvested species can have dramatic effects on the sustainability of that resource (Allendorf et al., 2008).

Consistency with oceanographic dispersal models in New Zealand

We aimed to use our genetic data to test the consistency between molecular and oceanographic modeling approaches for measuring connectivity. By comparing our results with an earlier study by Chiswell and Booth (2008), we are able to determine whether the prevailing current regimes in the region are likely to be the predominant forces driving the population structure and also explore the significance to the levels of genetic differentiation we reported.

As we hypothesized, there is little evidence to suggest that the four geographic regions responsible for larval dispersal proposed by Chiswell and Booth (2008) represented genetically discrete stocks. Rather, our results provide genetic evidence for high levels of larval exchange among all regions across New Zealand (Table 2; electronic Supplementary Material). With the exception of sample sites in the far north and far south of the country, all of the sample locations showed high levels of genetic homogeneity. This includes sample sites on either side of the Cook Straight, as well as the geographically isolated Chatham Island location, of which we found no evidence to support genetic divergence from the mainland. These findings are consistent with the majority of genetic connectivity studies on marine species in New Zealand that have reported no genetic break between the North and South Island across the Cooke Strait, although other North-South breaks have been reported (see Gardner et al., 2010 for a review).

Chiswell and Booth (2008) determined that nearly 95% of the larvae that recruit to CRA 8 are hatched within the CRA 8 boundary, and, although CRA 8 acts as a source of larvae to other areas around New Zealand, very few larvae that recruit to this region were determined by the model to have originated from CRA areas to the north. Our results support these findings with Stewart Island exhibiting signs of a self-recruiting population, as it had the lowest levels of genetic diversity (mean number of alleles, total alleles, private alleles and unbiased heterozygosity) and the smallest effective population size. Assignment tests also revealed that no individual sampled at Stewart Island was likely to have originated from northern sampling locations. On the basis of results from both studies, we propose that populations around Stewart Island (and likely other areas of CRA 8) are not reliant on outside larval sources, including recruits from Australia or Tasmania, to maintain healthy stock levels.

Trans-Tasman connectivity

The SouthWest coast of New Zealand supports a productive J. edwardsii fishery with high levels of annual recruitment (Booth and Breen, 1992); however, there are no apparent oceanographic retention mechanisms in the region to trap larvae and facilitate self-recruitment (Heath, 1980; Chiswell et al., 2003). This implies that the larvae recruiting to this coast are coming from another source. It has been previously suggested that because of the lack of genetic differentiation between New Zealand and Australian populations (Ovenden et al., 1992), and the detection of mid-late stage phyllosoma larvae in Tasman Sea (Booth et al., 1990; Booth and Ovenden, 2000), gene flow is occurring across the Tasman Sea at levels high enough to maintain west coast lobster populations in New Zealand.

The oceanographic model developed by Chiswell et al. (2003) tracking passive drifting particles across the Tasman Sea determined that dispersal from the west was ‘likely to be a viable mechanism maintaining lobster populations in New Zealand.’ The model showed that roughly 8–14% of J. edwardsii larvae released from southeast Australia can successfully arrive in New Zealand within the pelagic larval durations of the species. Despite identifying significant levels of genetic differentiation across the Tasman Sea, the results from our analyses are consistent with these findings and support the possibility of larval exchange between the two countries. In particular, our results revealed low levels of differentiation between Tasmania and the New Zealand populations, particularly the SouthWest Coast sample site, which is likely to be the main region of recruitment for Australian-derived larvae (Chiswell et al., 2003). The high level of genetic differentiation between South Australia and New Zealand, however, suggests that populations from South Australia are unlikely contributors to the New Zealand fishery. Rather, trans-Tasman larvae recruiting to New Zealand originate along the southeast coast of Australia and Tasmania.

Trans-Tasman migration rates from this analysis suggest that dispersal across the body of water occurs at low levels. Although the dispersal model indicated that ∼9–14% of the larvae originating on the east coast of Australia could reach New Zealand, the true rate of demographic connectivity is likely to be much lower as a result of pre- and post-settlement mortality. According to the first-generation migrant analysis, less than 2% of adult rock lobster in New Zealand originates from Australia. We therefore conclude that Australian populations are likely only a small contributor to the New Zealand fishery.

Considering that recruitment patterns of marine species can be highly sporadic across time (for example, Wooster and Bailey, 1989), it is possible that dispersal across the Tasman Sea for J. edwardsii may occur during some years but not others. It is important to note that this study did not address temporal variation, and the data here possibly reflect only a single recruitment event (assuming that all sampled individuals are from the same cohort). Discrete cohorts often have markedly different genetic signatures (Planes and Lenfant, 2002) resulting from sweepstakes reproductive success, whereby a small proportion of adults account for the majority of the recruitment (Hedgecock et al., 2007). Consequently, quantifying the dependence of the New Zealand rock lobster fishery on recruitment from Australia with the levels of accuracy needed to inform and benefit management practices would require further temporal sampling.

Population structure of J. edwardsii in New Zealand

Our microsatellite analysis revealed significant population structure within New Zealand and contrasts with a previous analysis by Ovenden et al. (1992), which found populations across New Zealand and Australia to be panmictic based on mitochondrial DNA data. The results from this analysis suggest that there are likely three genetically discrete populations throughout New Zealand: (1) the far north, (2) central (both islands) and (3) far south.

Despite identifying significant levels of differentiation between a number of New Zealand sample sites, results indicate that centrally located populations in New Zealand form a single-homogenous population characterized by high levels of gene flow between the regions. The consistencies with the oceanographic model developed by Chiswell and Booth (2008) suggest that the genetic homogeneity reflects contemporary patterns of connectivity and the exchange of larvae via coastal surface currents. The Westland Current flows north along the east coast of the South Island and eventually meets the D’Urville current that sweeps through the Cook Strait. The Straight was formed between 16 000 and 6000 years ago as a result of sea level rise (Lewis et al., 1994), and a lack of any evidence for an east–west genetic break suggests that it has had a major role in homogenizing populations of this species on opposing coasts. Furthermore, the Waiarapa Eddy is the largest retentive mechanism along the east coast of New Zealand and is thought to be the most significant oceanographic feature influencing recruitment in the region (Lesser, 1978; Chiswell and Roemmich, 1998; Chiswell and Booth, 2008). It has been well documented that J. edwardsii larvae become entrapped in the Wairarapa Eddy along the east coast of New Zealand (Lewis et al., 1994; Chiswell and Booth, 1999) and are kept within successful recruitment distance from the coastline for the duration of their pelagic larval stage. This oceanographic feature may not only be an important retention mechanism to maintain high levels of recruitment along the east coast of the North Island but also have an important role in maintaining genetic homogeneity for this species, by trapping phyllosoma larvae from a number of sources and providing a homogenized supply of recruits to east-coast populations, including the Chatham Islands.

One of the most consistent results from our analysis was the strong genetic break between Stewart Island and New Zealand populations to the north. Contrary to these findings, there appears to be no major oceanographic mechanisms near Stewart Island that would prevent gene flow to the north. However, Chiswell and Booth (2008) showed that although CRA 8 is largely self-seeding, 70% (data not shown) of larvae hatched in this region are widely distributed throughout the country, traveling up the west coast and through the Cook Straight, and also up the east coast of the South Island (Table 2: electronic Supplementary Material). These dispersal patterns, however, are based on the release of larvae across the entire CRA 8 region and because the prevailing current that flows through the Foveaux straight moves in an easterly direction, any larvae released near Stewart Island would likely be transported up the east coast of the South Island in the Southland Current (Heath, 1985). The Southland Current eventually hits the Chatham Rise at∼45 S and is diverted offshore. Any larvae trapped in this current would be removed from the system entirely, which would explain the levels of differentiation detected with other regions of New Zealand. Further sampling, particularly along the south east coast, would clarify this matter and determine whether restricted gene flow is the primary evolutionary force driving the patterns of differentiation.

Selection pressure and other considerations

The extremely long larval duration of J. edwardsii combined with the wide-spread dispersal of larvae, as determined by the oceanographic model (Chiswell and Booth, 2008), suggests that the population structure of this species in New Zealand may not be exclusively a product of restricted gene flow. J. edwardsii’s range extends over 10 degrees of latitude and consequently across a gradient of environmental variables (for example, sea surface temperature; Chiswell, 1994). If loci were acting under selection for such traits, then local adaptation to abiotic factors would inflate measures of population differentiation (Whitlock and McCauley, 1999; Neilson et al., 2006; Coyer et al., 2011). Results from the outlier loci analysis, however, revealed no evidence of selection pressure on any of the markers used in this analysis, and therefore it is unlikely that the patterns we report are the result of local adaptation to environmental conditions.

Despite having a large census population size capable of supporting high catch levels for more than half a century (Table 3; National Rock Lobster Management Group, (2010)), J. edwardsii populations in the CRA 8 region are characterized by low levels of genetic diversity and a small effective population size relative to other regions in New Zealand. Although the model (Chiswell and Booth, 2008) indicated that these results are likely a product of self-recruitment and subsequent isolation from the larger gene pool, it is also possible that harvesting may be responsible for the patterns observed. The conventional wisdom that marine species with large census population sizes, high fecundity and pelagic larvae are safe from the effects of intensive fishing pressure (from a conservation genetic diversity perspective) has proven to be a coarse misconception. Recent studies have shown that fisheries exploitation can lead to marked declines in genetic diversity and effective population sizes of the stock (Hauser et al., 2002; Hutchinson et al., 2003), although the biological or ecological effects of these declines (if any) are poorly understood.

Harvest levels for J. edwardsii across New Zealand have been variable across regions historically, but more consistent in recent years since the implementation of the Quota Management System. Fishing pressure in CRA 8 over the last 20 years (since the implementation of the Quota Management System) has been more intensive than anywhere else in New Zealand. From 1990–2010, CRA 8 has made up 30% (+/−4.6% s.d.) of the total national catch each year with average annual landings of 776 tonnes (+/−140 s.d.), exceeding all other regions by at least 50% (Table 3: electronic Supplementary material). Historical catch data show that harvest pressure before the 1990s was well above these levels, peaking at 4500 tonnes in the early 1950s and slowly declining to roughly 900 tonnes by 1990 (National Rock Lobster Management Group, (2010)). Further, the region that extends from the Banks Peninsula through to western Foveaux Straight (CRA 7 and part of CRA 8) is the only part of the country where female size at maturity is greater than the minimum legal size (National Rock Lobster Management Group, (2010)), meaning that lobsters are likely to have been removed from the system before reaching reproductive maturity.

The degree to which intensive fishing pressure has influenced the genetic diversity and effective population size of lobster stock in this region remains unknown given that there are no historic samples from which we could directly compare our results. Nevertheless, these results add to the cumulating evidence that exploited marine species with large populations sizes may be in danger of losing genetic diversity as a result of harvesting (Hauser et al., 2002). Given the economic importance of this species, this should be a focus of future research.

Conclusions

Applying the results of genetic connectivity studies to the management actions can be problematic for widely dispersing marine species, as it is often difficult to interpret the ecological importance of low levels of genetic differentiation (Lowe and Allendorf, 2010; Hedgecock et al., 2007). However, by combining our genetic data with earlier modeling approaches, we have been able to considerably increase our confidence in the demographic connectivity patterns for J. edwardsii throughout New Zealand and southern Australia, an approach we advocate for other commercial and ecologically important species. Although it is possible that selection and harvesting may have influenced the genetic patterns that we have reported, it is likely that oceanographic processes have the most influence in determining genetic structure in J. edwardsii. Finally, we examined the genetic structure of a species at the extreme end of the dispersal spectrum and found genetic structure and evidence for source-sink dynamics, which demonstrate that homogenous genetic structure cannot be assumed for species with a long-lived planktonic phase.

Data archiving

Data can be obtained from the Dryad repository doi:10.5061/dryad.3639h. Data files: Jasus edwardsii microsatellite analysis.