Patterns and persistence of larval retention and connectivity in a marine fish metapopulation

Transcription

1 Molecular Ecology (2012) doi: /j X x Patterns and persistence of larval retention and connectivity in a marine fish metapopulation PABLO SAENZ-AGUDELO,* 1 GEOFFREY P. JONES, SIMON R. THORROLD and SERGE PLANES* *USR 3278 Laboratoire d excellence CORAIL, CNRS-EPHE, CRIOBE Centre de Biologie et d Ecologie Tropicale et Méditerrannéenne, Université de Perpignan, 66860, Perpignan Cedex, France, School of Marine and Tropical Biology, and ARC Centre of Excellence for Coral Reef Studies, James Cook University, Townsville, 4811, Qld, Australia, Biology Department, MS 50, Woods Hole Oceanographic Institution, Woods Hole, MA, 02543, USA Abstract Connectivity, the demographic linking of local populations through the dispersal of individuals, is one of the most poorly understood processes in population dynamics, yet has profound implications for conservation and harvest strategies. For marine species with pelagic larvae, direct estimation of connectivity remains logistically challenging and has mostly been limited to single snapshots in time. Here, we document seasonal and interannual patterns of larval dispersal in a metapopulation of the coral reef fish Amphiprion polymnus. A 3-year record of larval trajectories within and among nine discrete local populations from an area of approximately 35 km was established by determining the natal origin of settled juveniles through DNA parentage analysis. We found that spatial patterns of both self-recruitment and connectivity were remarkably consistent over time, with a low level of self-recruitment at the scale of individual sites. Connectivity among sites was common and multidirectional in all years and was not significantly influenced by seasonal variability of predominant surface current directions. However, approximately 75% of the sampled juveniles could not be assigned to parents within the study area, indicating high levels of immigrations from sources outside the study area. The data support predictions that the magnitude and temporal stability of larval connectivity decreases significantly with increasing distance between subpopulations, but increases with the size of subpopulations. Given the considerable effort needed to directly measure larval exchange, the consistent patterns suggest snapshot parentage analyses can provide useful dispersal estimates to inform spatial management decisions. Keywords: Amphiprion polymnus, connectivity, larval dispersal, microsatellites, parentage analysis, temporal series Received 21 March 2012; revision received 16 June 2012; accepted 20 June 2012 Introduction Many species have a patchy distribution that consists of spatially discrete local populations (Hanski 1998). The Correspondence: Pablo Saenz-Agudelo, Fax: +966 (0) ; pablo.saenzagudelo@gmail.com 1 Present address: Red Sea Research Center, King Abdullah University of Science and Technology, Thuwal, Kingdom of Saudi Arabia. persistence of these species has been viewed as a dynamic process of local extinction and recolonization events, historically encompassed in a metapopulation framework that was first formalized by Levins (1969). The dynamics and persistence of such metapopulations depend to a large degree on the demographic links among local populations through dispersal. Metapopulation models have proven to be useful for describing and predicting the dynamics of species in a wide range of systems (Hanski & Gaggiotti 2004; Sale et al. 2006). The behaviour of these complex systems is critically

2 2 P. SAENZ-AGUDELO ET AL. dependant on spatial and temporal variation in the exchange of individuals among constituent populations (Hanski & Gaggiotti 2004). In current metapopulation theory, particular attention has been given to specific factors such as the area and proximity of neighbouring populations. These metrics are widely used as predictors of dispersal rates and colonization-extinction dynamics in natural terrestrial populations but have been rather neglected in marine systems (Sale et al. 2006). Many marine species have complex life cycles with a benthic adult phase that occurs in discrete habitat patches and a pelagic larval phase that assures population connectivity (Sale et al. 2006; Jones et al. 2009). During the larval phase, marine larvae are subject to multiple sources of mortality and advective-diffusive processes that transport them varying distances away from the populations of origin. Recent evidence from a diverse range of approaches including biophysical modelling (James et al. 2002; Cowen et al. 2006; Paris et al. 2007; Treml et al. 2008), otolith or shell chemistry (Swearer et al. 1999; Carson et al. 2010) otolith tagging (Jones et al. 1999; Almany et al. 2007), population genetics (Hogan et al. 2012) and parentage analysis (Jones et al. 2005; Planes et al. 2009; Christie et al. 2010; Saenz- Agudelo et al. 2011; Berumen et al. 2012) have shown that at least some larvae return to their population of origin, while others may travel long distances. However, little is known about the levels of variability of larval connectivity, both in space and in time, or the underlying factors that dictate metapopulation dynamics. There is evidence that temporal and spatial variation in larval connectivity can be responsible to some extent for spatial and temporal patchiness in the genetic structure of coastal marine populations (Selkoe et al. 2006; Hogan et al. 2010). However, due to significant logistical challenges of tracking dispersing larvae in the ocean, long-term patterns of demographically relevant connectivity have rarely been measured in marine metapopulations. Certainly, no empirical studies have tested how well habitat metrics including patch size and distance may predict dispersal patterns. Assessing the relationship between dispersal and habitat metrics, as well as the magnitude of temporal variation of connectivity, is essential for determining the broader population consequences of habitat loss, fragmentation and exploitation in spatially complex systems (Hanski 1998; Clobert et al. 2001; Bowler & Benton 2005; Levin 2006; Sale et al. 2006; Ronce 2007). An understanding of connectivity is also seen as vital for long-term spatial planning for conservation and resource management (Sale et al. 2005, 2006). In theory, predictable patterns of connectivity could be explained by many factors, including consistent oceanographic processes (Galindo et al. 2006; Treml et al. 2008), local population size (Caley et al. 1996) and distance among neighbouring populations (Jones et al. 2007). There is also potential for significant fluctuations in dispersal dynamics as a result of large-scale oceanographic processes (e.g. seasonal wind patterns, upwelling) and small-scale interactions among turbulent flow, larval behaviour and settlement cues (Hamilton et al. 2006; Navarrete et al. 2008). To our knowledge, only one study has documented the seasonal and interannual variability of larval connectivity for a marine metapopulation (Carson et al. 2010). Using elemental composition of shells as natural tags of natal origins in two species of mussels, Carlson and co-workers found consistent seasonal larval exchange coinciding with the direction of near-surface currents during each season. Studies of larval connectivity have received considerable attention in coral reef systems, which are characterized by extreme habitat patchiness (Sale et al. 2005; Jones et al. 2009). Despite a growing literature on the dynamics of coral reef ecosystems, the influence of factors, such as patch size, distance and oceanography, on the dispersal of coral reef organisms remains poorly understood (Botsford et al. 2009). Only a few studies have linked regional scale genetic structure with dispersal patterns predicted from hydrodynamic models of the study areas (Galindo et al. 2006; Salas et al. 2010; Foster et al. 2012). Another study showed that for a coral reef fish with strong homing behaviour, genetic patterns were not consistent with oceanographic model predictions (Gerlach et al. 2007). Yet to date, there have been no studies in which direct estimates of individual dispersal trajectories in a metapopulation have been monitored over time to evaluate consistent patterns and their underlying causes. In a previous study we used a likelihood-based parentage assignment method to measure larval retention within and exchange among nine discrete anemone aggregations hosting the anemonefish Amphiprion polymnus encompassing an area of approximately 35 km of coastline near Port Moresby, Papua New Guinea. We found this metapopulation to be characterized by high levels of connectivity and low self-recruitment rates within sites with an average self-recruitment rate of ~10% (Saenz-Agudelo et al. 2011). However, the temporal stability of these patterns remains unknown. Here, we set out to provide the first estimates of interannual variability in patterns of larval dispersal in a coastal marine metapopulation. We report a three consecutive year record ( ) of measurements of larval exchange and retention in our focal A. polymnus metapopulation. The study region is characterized by two contrasting seasons (wet summer and dry winter) with contrasting wind patterns and associated surface currents (Wyrtki 1960; Dennis et al. 2001). We tested for the

3 PATTERNS AND PERSISTENCE OF LARVAL CONNECTIVITY 3 evidence of directionality in larval transport associated with dominant surface currents by comparing larval exchange patterns between both seasons over the 3 years. Finally, we evaluated the role of subpopulation size and distance among subpopulations in explaining both the magnitude and the temporal variation of dispersal patterns observed in this system. Materials and methods Study species, site and data collection The panda clownfish (Amphiprion polymnus) is a southeast Asian endemic that lives in close association with discrete aggregations of two species of anemones (Stichodactyla hadonni and Heteractis crispa) occurring in sandy habitats associated with coral reefs (Fautin & Allen 1992). Each anemone is usually occupied by one breeding pair and up to eight smaller nonbreeders and juveniles. The female (the largest individual) lays demersal eggs on the upper surface of shells or dead coral next to the anemone. The embryos develop over a period of 6 7 days before hatching (Fautin & Allen 1992) and postlarvae settle into anemones after a pelagic larval phase lasting 9 12 days (Thresher et al. 1989). We sampled a total of 1394 potential A. polymnus parents and 1412 juveniles over 3 years ( ) (Table 1) from nine discrete anemone aggregations (hereafter referred to as sites ) distributed across ~35 km of coastline around Bootless Bay (Papua New Guinea) (Fig. 1). With the exception of Fisherman Island (FI) anemones within each site were confined to a ~1 ha patch of shallow sand and seagrass. For a 2-week period, each year, an exhaustive search for all anemones colonized by A. polymnusat all sites was performed and tissue samples were collected from all fish present at all sites with one exception [Fishermen Island (FI)]. The anemone aggregation at Fishermen Island (FI) was spread over a much larger area and only a small proportion of the protected side of the island were randomly explored in 2008 and 2009 (44 and 41 anemones in 2008 and 2009, were found, respectively). In 2010, sampling effort in FI was doubled and 91 anemones were found. However, we estimated this last figure represents around 80% of the anemones present at this site. A small site (SE) with eight anemones was found in a search for additional sites in 2009, and therefore, no data are available for this site in At each site, all fish were captured on SCUBA using hand nets, measured (total length TL), fin clipped underwater in situ and then released back onto the same anemone they were captured from. Fish that were too small to be fin clipped (<30 mm) were collected. For all analyses, fish were divided in to three categories according to their size. The first category breeders consisted of the female and male (the two biggest individuals) of each anemone. The remaining fish were then divided in to nonbreeders (>50 mm) and juveniles (<50 mm). Sampling was conducted once each year, and therefore, we had to estimate settlement times for all sampled juveniles. We used a combined approach of otolith reading and multilocus genotype based individual identification to approximately determine the size range of individuals that settled during the dry or the wet season each year. Lapilli otoliths were dissected from a subsample of 245 fish collected in the field (up to 30 mm). The age of each fish was estimated by counting the number of daily increments from the nucleus along the longest axis of the otolith. Clear readings were obtained for 145 of these fish (ranging from 7 to 26 mm in total length), but age estimation was not always Table 1 Description of the nine subpopulations of Amphiprion polymnus sampled between 2008 and 2010 considered in this study. For each year, the number of anemones colonized by at least one A. polymnus (NA), number of adult and subadult A. polymnus (A + SA) and juveniles (J) of the two size categories considered are shown.

4 4 P. SAENZ-AGUDELO ET AL. Fig. 1 Map showing sites of the nine anemone aggregations hosting Amphiprion polymnus in Bootless Bay area (black filled circles) and prevailing surface currents during the summer monsoon (November March) and during winter (April October). Crosses indicate sites outside Bootless Bay with potential suitable habitat for A. polymnus host anemones that were explored but no anemones were found. Dashed lines indicate shallow reef limits. Inset: Location of Bootless Bay in Papua New Guinea. Site abbreviations are as follows: Fishermen Island (FI), Manubada Island (BE), Lion Island (LI), Taurama (TA), Motupore north patch reef (MN), Motupore Island (MO), Loloata Island (LO), Loloata South Bank (BA), SouthEast patch reef (SE). Photo: A. polymnus colony in a Stichodactyla hadonii anemone (credit Serge Planes). possible, particularly for fish larger than 26 mm. The maximum age at size 25 mm TL observed from these otoliths estimates was ~102 days (Fig. S1, Supporting information). Fish of ~100 days old sampled in February recruited around the last week of October in the previous year, which coincided with the end of winter season; therefore, 25 mm TL was used as a proxy to delineate between winter and summer. Fish of 25 mm TL or smaller were assumed to have settled in the summer season and fish larger than 26 mm TL were assumed to have settled before this point in time. Genetic analyses We screened 18 polymorphic microsatellite DNA markers previously reported (Quenouille et al. 2004; Beldade et al. 2009) (Table S1, Supporting information). In our previous study (Saenz-Agudelo et al. 2011), we found that all these loci satisfied Hardy Weinberg and linkage disequilibrium assumptions for this particular species and study site. In addition, we also found no evidence of significant genetic structure among sites, so we treated all sites as one single genetic pool for this study. We used the FAMOZ platform (Gerber et al. 2003) to assign juveniles (TL < 50 mm) back to sampled adults in the metapopulation. This procedure combines exclusion probabilities and maximum likelihood ratios to select the most likely parent for each offspring based on population allele frequencies, genotype matching among parent/offspring pairs and the distribution of LOD scores of true parent offspring pairs and false pairs (share one allele per locus by chance), allowing for the inclusion of genotype scoring errors. Details of parentage analysis procedure can be found in the supporting information. We also compared individual microsatellite multilocus genotypes from all juveniles fin clipped over the 3 years to identify juveniles that were clipped on more than one occasion and had not yet reached the 50 mm TL to avoid repeated assignments among years. We used the Genalex 6 package (Peakall & Smouse 2006) to identify pairs of juvenile samples among years that had the same multilocus genotype. Given the combined probability of identity and identity between sibs (Waits et al. 2001) for the 18 loci (given the sample size and allele frequencies) were small ( and , respectively), we assumed that perfect matches between two samples from different years corresponded to the same individual. The possibility of genotyping errors was accounted for by allowing up to 3 (of 36) allele mismatches per pair in a first run of pairwise comparisons. Genotypes

5 PATTERNS AND PERSISTENCE OF LARVAL CONNECTIVITY 5 that matched at all but 1, 2 or 3 alleles were rescored. If observed differences were maintained after re-examination, then samples were assumed to belong to different individuals. Juveniles that were only genotyped on one occasion were considered to have settled between sampling events. Juveniles genotyped over two consecutive years were only included in parentage analysis in the first year they were captured. In situ current measurement Current velocity and direction was measured at five sites (LI, MO, BE, TA and BA) from February 4th to April 18th in 2008 and from January 26th to April 6th in 2009 using an Aanderaa DCM12 acoustic Doppler current profiler (ADCP). The instrument was deployed at a depth of 8 10 m at each of the sites for a period of 10 days in 2008 and 15 days in 2009 by moving it from one site to the next after 10 or 15 days, respectively. The ADCP sampled in 30-minute intervals over 2.5 m depth bins. For the purpose of this study, only surface current measures were used (first 2.5 m). Deeper bins where excluded to eliminate complex current patterns at depth caused by the complex bathymetry near reefs where the instrument was deployed. sites were not included in this analysis due to incomplete observations (site SE: only 2 years of data) and incomplete sampling of the subpopulation (site FI). We used the number of anemones colonized by A. polymnus as a proxy for site size and the shortest over-water distance between pairs of sites. The stability of any given connection between two sites (e.g. between the i th and j th sites) was quantified as the proportion of seasons where at least one juvenile went from site i (source) to site j (settlement). The magnitude of each connection was quantified as the cumulated number of juveniles that went from source-site i to settlement-site j over the 3-year period. Analysis of deviance via general linear models (GLMs) in R (R Development Core Team 2007) was performed to estimate the proportion of variance in stability and magnitude of connections that was explained by the size of source sites (measured as the number of anemones hosting A. polymnus) and the distance between source and settlement sites. As both stability and magnitude of connections were measured as count data (number of juveniles and number of seasons), GLMs were fitted using a log link (to ensure that fitted values are bounded below) and quasi-poisson errors (to account for non-normality and over-dispersion) (Crawley 2007). Analysis of temporal patterns Prevailing winds and associated surface currents in this region flow from northwest to southeast in the wet summer season (November to March) and from southeast to northwest during the dry winter season (April October) (Wyrtki 1960; Dennis et al. 2001) (Fig. 1). We investigated if there was a distinct seasonal pattern in the direction of A. polymnus larval transport along the coast that was correlated with the seasonal variability in surface currents. We tested for significant differences in the proportion of juveniles travelling southeast or northwest at each season. If the influence of larval behaviour and swimming speed was less important than predominant currents, then one should be able to detect directionality in the seasonal larval connectivity patterns reflecting dominant current flows. Under this assumption, it would be expected that during summer, the proportion of larvae transported from the southeast to northwest should be greater than transport in the opposite direction, and the reversed patterns should be observed during winter. Role of size and distance between sites We investigated the role of subpopulation size and distance among subpopulations in the stability and magnitude of connections in the focal metapopulation. Two Results Interannual and seasonal variability of larval dispersal Our estimates of per cent self-recruitment within the metapopulation (including all sites) in 2008, 2009 and 2010 were surprisingly similar. Of the 490 juveniles collected in 2008, 88 (18%) were identified as being progeny of adults from this metapopulation, compared with 128 of 507 (21.5%) in 2009 and 94 of 417 (22.5%) in The number of self-recruits within sites varied from 0 to 19 individuals. One site (TA) consistently had higher self-recruitment than all the others (16, 19 and 13 selfrecruits in 2008, 2009 and 2010, respectively) (Fig. 2). The average per cent self-recruitment at the level of sites was low in all seasons, ranging from 4.1% to 9.7% (Table 2). All sites, with one exception (site TA), showed low and steady levels of self-recruitment (0 15%) regardless of the year or the season considered. Self-recruitment at site TA was consistently higher (22 42%) in five of six seasons but was zero in one (summer 2009) (Fig. 3A). Despite this drastic difference in self-recruitment for one season at site TA, differences in self-recruitment among seasons within sites were much smaller than differences in self-recruitment among sites within each season. An analysis of variance demonstrated that while differences among sites within seasons explained 64% of total variation, differences

6 6 P. SAENZ-AGUDELO ET AL. among sites was higher than self-recruitment in all seasons ( %) (Table 2). In terms of the contribution to recruitment of each site to the metapopulation (all nine sites together), none of the individual sites contributed more than 10% of juveniles to the metapopulation, regardless of the season or year analysed (Fig. 3B). Patterns of contribution to the recruitment of each site were consistent among different seasons. Analysis of variance indicated that most of the variation in the local contribution of each site to total recruitment was explained by differences among sites within seasons (~48%), while differences among seasons within sites explained <10% regardless of whether the analysis was performed with or without site TA (Table S2A, B, Supporting information). Larval transport and dominant surface currents Fig. 2 Maps of Bootless Bay Area showing inferred individual trajectories (arrows) of A. polymnus juveniles among sampled sites over three consecutive years (2008, 2009 and 2010) based on parentage analysis. Black circles represent self-recruitment. Thickness of arrows and diameter of circles are proportional to the number of juveniles with similar trajectories. A zoomed map of bootless bay area is shown as an inlet at the top-right corner of each map. A detailed matrix with the data used to generate this figure is available as supplementary material. among seasons explained only 4.5% and did not differ from random expectations (Table S2A, Supporting information). The same pattern was observed when site TA (highest contribution to spatial variation) was dropped from the ANOVA. (Table S2B, Supporting information). A major proportion of juveniles (~75%) could not be assigned to any of the sampled parents, indicating high levels of migration from sites outside the study area. The dispersal distances within the study area inferred from parental analyses (~25% of juveniles) ranged between 1.2 km (TA-MN) and 35.5 km (FI-SE). Connectivity was observed between most sites in all three consecutive years analysed (Fig. 2). Mean local connectivity Acoustic Doppler current profiler measurements confirmed that for both periods measured (wet summer season in 2008 and 2009) dominant surface current directions were mostly to the southeast (see Fig. S2, Supporting information), but with frequent reversals of their direction (See Fig. S3, Supporting information). We found no evidence of directionality in larval transport for any of the 3 years or six seasons, with longdistance dispersal trajectories in both directions (Fig. 2). None of the 2-sample binomial tests for the equality of proportions were significant at a = That is, larvae from both size categories dispersed northwest and southeast in similar proportions regardless of the season or the year considered (Table 1). (A detailed connectivity matrix for each season in each year is available in Table S3, Supporting information). Connectivity, site size and distance between sites The size of the source site and the distance between source and settlement sites had significant but opposite effects on the magnitude of connectivity ( Source size deviance = 29.76, F 1,47 = 5.74, p = 0.021; source-settlement distance deviance = 91.94, F 1,46 = 17.72, p < Dispersion parameter for quasi-poisson errors = 5.18) (Figs. 4A, B). The size of the source site explained 9.5% of the variation in the magnitude of connections, while the distance between source and settlement sites explained 29.3%. Similarly, the stability of connections, that is, the proportion of times (seasons) a particular connection was observed, was also associated with the size of the source and the distance between source and settlement sites ( Source size deviance = 4.72, F 1,47 = 5.49, p = 0.023; source-settlement distance deviance = 7.20, F 1,46 = , p = Dispersion parameter for quasi-poisson errors = 0.858) (Fig. 4C, D). The size of

7 PATTERNS AND PERSISTENCE OF LARVAL CONNECTIVITY 7 Table 2 Number of Amphiprion polymnus juveniles that dispersed southwest and northeast, mean self-recruitment and local connectivity per site for Bootless Bay area for each juvenile size category as a proxy for recruitment in different seasons. Last column corresponds to the proportion of juveniles of each size category assigned by parentage analysis, all sites confounded (metapopulation), corresponding to the overall self-recruitment for each season. Year Size category (Season) Total juveniles Assigned by parentage Juveniles dispersing SW NE % Selfrecruitment mean ± SD % Local connectivity mean ± SD Overall selfrecruitment (%) 2008 >25 mm <50 mm (winter) <25 mm (summer) 2009 >25 mm <50 mm (winter) <25 mm (summer) 2010 >25 mm <50 mm (winter) <25 mm (summer) ± ± ± ± ± ± ± ± ± ± ± ± (A) (B) Fig. 3 Distribution of relative frequencies of (A) self-recruitment and (B) contribution of the site to metapopulation among the nine anemone aggregations hosting A. polymnus in Bootless Bay for each year (2008 circles, 2009 triangles and 2010 diamonds) and seasons: <25 mm (summer) grey filled symbols and >26 mm (winter) empty symbols. (A) (C) (B) (D) Fig. 4 Effects of subpopulation source size (measured as the number of anemones per site) and distance between source and sink subpopulations on the magnitude: cumulated number of juveniles produced over 3 years (A and B) and the stability of connections: number of seasons that a particular connection was observed in the Bootless Bay Amphiprion polymnus metapopulation system (C and D). GLMs fitted lines are: (A) y=e x (B) y=e x (C) y=e x (D) y=e x

8 8 P. SAENZ-AGUDELO ET AL. source sites explained 7.7% of the variation in the stability of connections, while the distance between source and settlement sites explained 11.9%. There was no evidence of interaction between source size and distance between source and settlement subpopulations. In both cases, model simplification allowed for the removal of the interaction term without any significant consequences. Discussion This study provides the first empirical, multiyear description of the magnitude and direction of larval connectivity in a marine fish metapopulation. We confirmed that local replenishment in this system is dominated by connectivity, with low levels of selfrecruitment at the scale of small subpopulations. Sufficient juveniles could be assigned to adults in the nine subpopulations to test three distinct hypotheses. First, our 3-year survey provided strong evidence that patterns of local replenishment and connectivity can be surprisingly stable even at small geographic scales of individual reefs. Second, we found connectivity to be multidirectional and among all subpopulations, with no evidence for directional larval dispersal related to dominant surface current patterns. Finally, we showed that both the size of the source sites and distance between source and settlement sites are reasonably good predictors of both the magnitude and stability of larval connectivity. Temporal stability was observed in terms of the per cent of recruitment explained by self-retention within and local connectivity among individual sites. It seems that even at the scales of individual reefs larval retention patterns can be highly conserved over several years. Three other empirical larval connectivity studies have incorporated temporal data. Berumen et al.(2012) showed that two coral reef fish species in a small isolated Island in Papua New Guinea displayed consistently high self-recruitment rates over 3 years, while dispersal to adjacent areas was more variable. Similarly, Hogan et al.(2012) reported fluctuating annual connectivity patterns for a another coral reef fish in the Caribbean. Temporal stability has been also described for two temperate mussel species, with connectivity following regular seasonal patterns and rather constant selfrecruitment rates (Carson et al. 2010). Together with this study, these empirical time series suggest that temporal stability is more general at demographic timescales than previously thought (but see Hogan et al. 2012), and that geographic settings may play a significant role in the shape and magnitude of these patterns (Jones et al. 2009; Pinsky et al. 2012). However, only the addition of more and longer time series will help to validate this idea and to develop a mechanistic understanding of larval connectivity in marine fishes. We found no evidence of seasonal associations between directionality in larval dispersal and average surface currents. Direct evidence for a correlation between currents and the directionality of larval connectivity exists only for the two previously mentioned species of mussels (Carson et al. 2010). While average surface currents in the study area had opposite dominant directions between seasons, we also documented several current reversal events in summer months. We therefore cannot exclude the possibility that a significant proportion of larvae were released in the water column during these current reversals as Amphiprion polymnus reproduces all year long in this region. Overall, our observations appear consistent with Carson et al. s conclusion that connectivity may be more diffusive than advective at small spatial scales. However, the origin of approximately 75% of larvae remains unknown in our study and whether directionality driven by major oceanographic settings is more evident at larger scales remains unresolved. Despite the lack of predictable directionality of larval dispersal, the magnitude and the stability of connections among sites were significantly correlated with both the size of the source and the distance between source and settlement sites. While increasing distance between sites was associated with decreasing magnitude and stability of connections, increasing source size was associated with increasing magnitude and stability. Both variables had similar effects on the magnitude and stability of connectivity. However, the distance between sites explained near twice the variance compared with the size of source sites for both magnitude and stability of connectivity. These results are encouraging for the implementation of metapopulation models that use simple connectivity formulations based on this kind of information when empirical estimates are not available (Moilanen & Nieminen 2002). There are, however, two major limitations that need to be addressed. First, more empirical studies for a wide range of taxa and locations are warranted before this pattern can be generalized. Second, recent metapopulation modelling studies have highlighted that local demography might be more important than connectivity for the persistence of the metapopulation (Figueira 2009). This has been supported recently by one empirical study (Carson et al. 2011). Time series, like the one presented here, combined with life history-based population projections will definitely help to assess to what extent the concept of marine metapopulations can be generalized and will provide managers with a strong solid baseline to target the processes that may have a significant effect on metapopulation persistence.

9 PATTERNS AND PERSISTENCE OF LARVAL CONNECTIVITY 9 Both the high temporal consistency in self-recruitment and local connectivity and the lack of association between larval transport and dominant current directionality seem to indicate that processes other than passive dispersal may be dominant at this scale. One possible explanation comes from the suggestion that the magnitude of stochasticity in recruitment patterns should diminish for species that spawn throughout the year and have larvae with short pelagic durations and high swimming capacities (Siegel et al. 2008), which are all characteristic of A. polymnus. Another possibility is that, for organisms like coral reef fish whose larval stage has excellent swimming capacities and sensory abilities (Kingsford et al. 2002), larval behaviour might play an important role in buffering stochastic patterns of turbulent circulation at small spatial scales (Paris et al. 2007). For example, active homing behaviour will probably change expectations based on passive dispersal (Gerlach et al. 2007). Such behaviour has been suggested for A. polymnus in a different location (Jones et al. 2005). However, the rather consistent low self-recruitment rates in Bootless Bay suggest that homing was not predominant in this location. It has been shown that clownfish can detect and discriminate several different environmental cues (Munday et al. 2009). Unfortunately, the full extent of environmental cues and the extent to which they can be used to predict consistent larval behaviour remain unresolved. Our results are encouraging from the point of view of conservation and models of marine population dynamics. First, they support the applicability of models of marine protected areas which assume that dispersal kernels and recruitment patterns remain fairly constant over relevant timescales (e.g., Lockwood et al. 2002; Kaplan et al. 2009), at least at small spatial scales. Second, they suggest that factors such as subpopulation size and distance between subpopulations do play a more important role in larval connectivity than does local hydrodynamic settings, at least at small spatial scales. Also, compared with previously reported selfrecruitment for the same species in a different location (Jones et al. 2005), the low self-recruitment rates found here reinforces the idea that distance between subpopulations (or geographic settings) can be an important factor determining the degree of population openness (Pinsky et al. 2012). This information may be used to inform management decisions when no other information on larval dispersal is available. Finally, our results show that single-year estimates using this kind of approach provide reliable information of the dynamic processes that occur over longer timescales in marine populations and therefore might be considered as appropriate guidelines for marine conservation strategies and MPA network design (Botsford et al. 2009; Kaplan et al. 2009). It is important to highlight that the full extent of this metapopulation remains to be determined. At the spatial scale of this study, the parental origin of ~75% of the juveniles samples remains unknown. Therefore, conclusions about temporal stability in connectivity refer only to local connectivity within the 35 km study area. We can only speculate about the location and size of subpopulations outside our study that produced the juveniles that were not assigned to parents. Reefs and potential suitable habitat for the anemones that host A. polymnus extend over 50 km northwest and ~100 km southeast from the study area based on satellite images, after which the reefs are interrupted by large river deltas. Given the rather short pelagic larval duration for this species (~12 days), we hypothesize that most of the incoming larvae replenishment our study system are from populations within this range, but this remains to be tested. Finally, our results suggest that simple metrics such as subpopulation size and distance among them may be good predictors of connectivity when no other information is available. Long-term empirical estimates of larval connectivity on organisms with different life history traits at similar spatial scales are required to confirm the generality of this pattern. However, empirical estimation of dispersal kernels using available methods such as parentage analysis remains logistically overwhelming over larger spatial and temporal scales. Therefore, it will probably be necessary to combine direct methods of larval dispersal estimation, biophysical modelling and larval behaviour to provide estimates of connectivity at scales relevant to spatial management approaches in reef ecosystems (James et al. 2002; Cowen et al. 2006). Combining these approaches will also help to estimate the relative importance of larval behaviour and oceanographic processes in dispersal patterns. Acknowledgements We thank Chris McKelliget, Vanessa Messmer, Juan David Arango, Jennifer Smith, Agnes Rouchon and the Motupore Island Research Centre staff for assistance in the field. Nuria Raventos assisted with otolith analyses. The ARC Centre of Excellence for Coral Reef Studies, the National Science Foundation (OCE ), the Coral Reef Initiatives for the Pacific (CRISP), the TOTAL Foundation, Populations Fractionees et Insulares (PPF EPHE) and the Connectivity Working Group of the global University of Queensland World Bank Global Environmental Facility project, Coral Reef Target Research and Capacity Building for Management provided financial support. Special thanks to Motupore Island Research Centre, Dik Knight and Loloata Island Resort for logistic support.

10 10 P. SAENZ-AGUDELO ET AL. References Almany GR, Berumen ML, Thorrold SR, Planes S, Jones GP (2007) Local replenishment of coral reef fish populations in a marine reserve. Science, 316, Beldade R, Holbrook S, Schmitt R, Planes S, Bernardi G (2009) Isolation and characterization of eight polymorphic microsatellite markers from the orange-fin anemonefish, Amphiprion chrysopterus. Conservation Genetics Resources, 1, Berumen ML, Almany GR, Planes S, Jones GP, Saenz-Agudelo P, Thorrold SR (2012) Persistence of self-recruitment and patterns of larval connectivity in a marine protected area network. Ecology and Evolution, 2, Botsford L, White J, Coffroth M et al. (2009) Connectivity and resilience of coral reef metapopulations in marine protected areas: matching empirical efforts to predictive needs. Coral Reefs, 28, Bowler DE, Benton TG (2005) Causes and consequences of animal dispersal strategies: relating individual behaviour to spatial dynamics. Biological Reviews, 80, Caley MJ, Carr MH, Hixon MA, Hughes TP, Jones GP, Menge BA (1996) Recruitment and the local dynamics of open marine populations. Annual Review of Ecology and Systematics, 27, Carson HS, López-Duarte PC, Rasmussen L, Wang D, Levin LA (2010) Reproductive timing alters population connectivity in marine metapopulations. Current Biology, 20, Carson HS, Cook GS, Lopez-Duarte PC, Levin LA (2011) Evaluating the importance of demographic connectivity in a marine metapopulation. Ecology, 92, Christie MR, Johnson DW, Stallings CD, Hixon MA (2010) Selfrecruitment and sweepstakes reproduction amid extensive gene flow in a coral-reef fish. Molecular Ecology, 19, Clobert J, Danchin E, Dhondt AA, Nichols JD (2001) Dispersal. Oxford University Press, Oxford, UK. Cowen RK, Paris CB, Srinivasan A (2006) Scaling of connectivity in marine populations. Science, 311, Crawley MJ(2007) The R Book. Wiley & Sons Ltd, Chichester, UK. Dennis DM, Pitcher CR, Skewes TD (2001) Distribution and Transport Pathways of Panulirus ornatus (Fabricius, 1776) and Panulirus spp. larvae in the Coral Sea Australia. Marine and Freshwater Research52, Fautin DC, Allen GR (1992) Field Guide to Anemone Fishes and their Host Sea Anemones. Western Australia Museum, Perth. Figueira WF (2009) Connectivity or demography: defining sources and sinks in coral reef fish metapopulations. Ecological Modelling, 220, Foster NL, Paris CB, Kool JT et al. (2012) Connectivity of Caribbean coral populations: complementary insights from empirical and modelled gene flow. Molecular Ecology, 21, Galindo HM, Olson DB, Palumbi SR (2006) Seascape genetics: a coupled oceanographic-genetic model predicts population structure of Caribbean corals. Current Biology, 16, Gerber S, Chabrier P, Kremer A (2003) FAMOZ: a software for parentage analysis using dominant, codominant and uniparentally inherited markers. Molecular Ecology Notes, 3, Gerlach G, Atema J, Kingsford MJ, Black KP, Miller-Sims V (2007) Smelling home can prevent dispersal of reef fish larvae. Proceedings of the National Academy of Sciences, 104, Hamilton SL, White JW, Caselle JE, Swearer SE, Warner RR (2006) Consistent long-term spatial gradients in replenishment for an island population of a coral reef fish. Marine Ecology Progress Series, 306, Hanski I (1998) Metapopulation dynamics. Nature, 396, Hanski I, Gaggiotti O (2004) Ecology, Genetics and Evolution of Metapopulations. Academic Press, Burlington. Hogan JD, Thiessen R, Heath D (2010) Variability in connectivity indicated by chaotic genetic patchiness within and among populations of a marine fish. Marine Ecology Progress Series, 417, Hogan J, Thiessen R, Sale P, Heath D (2012) Local retention, dispersal and fluctuating connectivity among populations of a coral reef fish. Oecologia, 168, James MK, Armsworth PR, Mason LB, Bode L (2002) The structure of reef fish metapopulations: modelling larval dispersal and retention patterns. Proceedings of the Royal Society B: Biological Sciences, 269, Jones GP, Milicich MJ, Emslie MJ, Lunow C (1999) Self-recruitment in a coral reef fish population. Nature, 402, Jones GP, Planes S, Thorrold SR (2005) Coral reef fish larvae settle close to home. Current Biology, 15, Jones GP, Srinivasan M, Almany GR (2007) Population connectivity and conservation of marine biodiversity. Oceanography, 20, Jones GP, Almany G, Russ G et al. (2009) Larval retention and connectivity among populations of corals and reef fishes: history, advances and challenges. Coral Reefs, 28, Kaplan DM, Botsford LW, O Farrell MR, Gaines SD, Jorgensen S (2009) Model-based assessment of persistence in proposed marine protected area designs. Ecological Applications, 19, Kingsford M, Leis J, Shanks A, Lindeman KC, Morgan SG, Pineda J (2002) Sensory environments, larval abilities and local self-recruitment. Bulletin of Marine Science, 70(Suppl), Levin LA (2006) Recent progress in understanding larval dispersal: new directions and digressions. Integrative and Comparative Biology, 46, Levins R (1969) Some demographic and genetic consequences of environmental heterogeneity for biological control. Bulletin of the ESA, 15, Lockwood DR, Hastings A, Botsford LW (2002) The effects of dispersal patterns on marine reserves: does the tail wag the dog? Theoretical Population Biology, 61, Moilanen A, Nieminen M (2002) Simple connectivity measures in spatial ecology. Ecology, 83, Munday PL, Dixson DL, Donelson JM et al. (2009) Ocean acidification impairs olfactory discrimination and homing ability of a marine fish. Proceedings of the National Academy of Sciences, 106, Navarrete SA, Broitman BR, Menge BA (2008) Interhemispheric comparison of recruitment to intertidal communities: pattern persistance and scales of variation. Ecology, 89, Paris CB, Chérubin LM, Cowen RK (2007) Surfing, spinning, or diving from reef to reef: effects on population connectivity. Marine Ecology Progress Series, 347,

11 PATTERNS AND PERSISTENCE OF LARVAL CONNECTIVITY 11 Peakall ROD, Smouse PE (2006) GENALEX 6: genetic analysis in Excel. Population genetic software for teaching and research. Molecular Ecology Notes, 6, Pinsky ML, Palumbi SR, Andrefouet S, Purkis SJ (2012) Open and closed seascapes: where does habitat patchiness create populations with high fractions of self-recruitment? Ecological Applications, 22, Planes S, Jones GP, Thorrold SR (2009) Larval dispersal connects fish populations in a network of marine protected areas. Proceedings of the National Academy of Sciences, 106, Quenouille B, Bouchenak-Khelladi Y, Hervet C, Planes S (2004) Eleven microsatellite loci for the saddleback clownfish Amphiprion polymnus. Molecular Ecology Notes, 4, R Development Core Team(2007) R: A Language and Environment for Statistical Computing. R Foundation for statistical computing, Vienna. Ronce O (2007) How does it feel to be like a rolling stone? ten questions about dispersal evolution. Annual Review of Ecology, Evolution, and Systematics, 38, Saenz-Agudelo P, Jones GP, Thorrold SR, Planes S (2011) Connectivity dominates larval replenishment in a coastal reef fish metapopulation. Proceedings of the Royal Society B: Biological Sciences, 278, Salas E, Molina-Ureña H, Walter R, Heath D (2010) Local and regional genetic connectivity in a Caribbean coral reef fish. Marine Biology, 157, Sale PF, Cowen RK, Danilowicz BS et al. (2005) Critical science gaps impede use of no-take fishery reserves. Trends in Ecology & Evolution, 20, Sale PF, Hanski I, Kritzer JP (2006) Chapter 1 The merging of metapopulation theory and marine ecology: establishing the historical context. In: Marine Metapopulations (eds. Jacob PK and Peter FS), pp Academic Press, Burlington. Selkoe KA, Gaines SD, Casselle JE, Warner RR (2006) Current shifts and kin aggregation explain genetic patchiness in fish recruits. Ecology, 87, Siegel DA, Mitarai S, Costello CJ et al. (2008) The stochastic nature of larval connectivity among nearshore marine populations. Proceedings of the National Academy of Sciences, 105, Swearer SE, Caselle JE, Lea DW, Warner RR (1999) Larval retention and recruitment in an island population of a coralreef fish. Nature, 402, Thresher RE, Colin PL, Bell L (1989) Planktonic duration, distribution and population structure of Western and Central Pacific Damselfishes (Pomacentridae). Copeia, 1989, Treml E, Halpin P, Urban D, Pratson L (2008) Modeling population connectivity by ocean currents, a graph-theoretic approach for marine conservation. Landscape Ecology, 23, Wyrtki K (1960) The surface circulation in the Tasman and Coral seas. CSIRO Division of fisheries and oceanography technical paper No 8, 44pp. All authors participated in the fieldwork, P.S.A. performed analyses and wrote the first draft of the manuscript; all authors contributed substantially to revisions. This study was conducted as part of P.S.A. PhD dissertation research. Data accessibility Microsatellite individual genotype per locus data: DRYAD entry doi: /dryad.cq700. Supporting information Additional Supporting Information may be found in the online version of this article. Appendix S1 Details of DNA extraction and parentage analysis. Table S1 Microsatellite loci used in this study. Table S2 Summary of analysis of variance. Table S3 Representation of the seasonal connectivity matrices for both inferred seasons. Fig. S1 Relationship between total length and age at the time of collection for Amphiprion polymnus in Bootless Bay area estimated from Otoliths. Fig. S2 Frequency of surface current direction measured at each of five sites. Fig. S3 Time series of v (A) and u (B) current components in Bootless Bay in 2008 and Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.