Interspecific, Spatial and Temporal Variability of Self-Recruitment in Anemonefishes

Polymorphic microsatellite DNA parentage analysis was used to investigate the spatio-temporal variability of self-recruitment in populations of two anemonefishes: Amphiprion ocellaris and A. perideraion. Tissue samples of A. ocellaris (n = 364) and A. perideraion (n = 105) were collected from fringing reefs around two small islands (Barrang Lompo and Samalona) in Spermonde Archipelago, Indonesia. Specimens were genotyped based on seven microsatellite loci for A. ocellaris and five microsatellite loci for A. perideraion, and parentage assignment as well as site fidelity were calculated. Both species showed high levels of self-recruitment: 65.2% of juvenile A. ocellaris in Samalona were the progeny of parents from the same island, while on Barrang Lompo 47.4% of A. ocellaris and 46.9% of A. perideraion juveniles had parents from that island. Self-recruitment of A. ocellaris in Barrang Lompo varied from 44% to 52% between the two sampling periods. The site fidelity of A. ocellaris juveniles that returned to their reef site in Barang Lompo was up to 44%, while for A. perideraion up to 19%. In Samalona, the percentage of juveniles that returned to their natal reef site ranged from 8% to 11%. Exchange of progeny between the two study islands, located 7.5 km apart, was also detected via parentage assignments. The larger Samalona adult population of A. ocellaris was identified as the parents of 21% of Barrang Lompo juveniles, while the smaller adult population on Barrang Lompo were the parents of only 4% of Samalona juveniles. High self-recruitment and recruitment to nearby island reefs have important implications for management and conservation of anemonefishes. Small MPAs, preferably on every island/reef, should ensure that a part of the population is protected to enable replenishment by the highly localised recruitment behaviour observed in these species.


Introduction
Self-recruitment is defined as the proportion of larvae returning to and settling in their natal population, whereas population connectivity is the linking of distinct populations by individual dispersal or migration [1]. These two aspects are fundamental for the management and conservation of living marine resources [2], management of highly harvested species [3], understanding the population dynamics of marine organisms [4], and improving the design of marine reserves [5]. Sufficient self-recruitment and connectivity among populations in marine reserves are believed to prevent local extinction that might otherwise occur as a result of anthropogenic disturbances such as fishing pressure [6]. However, directly measuring the degree of self-recruitment and connectivity in populations of marine organisms is challenging due to the large number and small size of the propagules, the time spent in the dispersive pelagic larval stages and the high mortality. Although the pelagic larval duration (PLD), which varies from days to weeks in fish [7], affects dispersal capability [8], dispersal distances are also potentially influenced by oceanographic processes [9], geographic location and flow variability of ocean currents [10], as well as larval behaviour, such as vertical positioning, swimming and olfactory reef-sensing [11][12][13][14].
Genetic markers that can be used for determining parentage and relatedness offer an indirect method for measuring selfrecruitment and connectivity, thus providing important information on population dynamics. These markers are also widely used for addressing wildlife management issues in a variety of organisms [15][16][17][18][19]. A commonly used genetic marker are microsatellites, simple repetitive sequences located throughout the eukaryote nuclear genome [20]. Because of their high variability they are useful for fine-scale ecological studies, such as parentage analysis [21]. Parentage analysis uses data from polymorphic microsatellites for relationship reconstruction based on the maximum likelihood method, where juveniles are assigned to the most likely parent from a data set of potential parents [22]. This method has been proven a powerful tool for investigating self-recruitment in marine fishes [23][24][25][26], identifying connectivity among fish populations [27][28], and determining whether larvae of marine organisms remain close to their origin over small scales (e.g., among groups within a population) [29].
In this study microsatellites are used to study self-recruitment in two species of anemonefish. Spatial patterns of recruitment in anemonefishes are interesting in part due to their unusual symbiosis with anemones, social structure and breeding biology, but also critically important due to the high level of exploitation of these species and their host anemones by the global ornamental fish trade [30]. Anemonefishes have two very different phases in their lifecycle: sedentary adults live in close association with host anemones, while larvae are planktonic. Metamorphosing juveniles recruit to a species-specific host anemone, usually joining a mixedage group of conspecifics. Within that group, the largest individual is the reproductive female, the second largest the reproductive male, while the remaining individuals are non-reproductive subadults and juveniles [31] [32].
This study focuses on two species of anemonefish: Amphiprion ocellaris and A. perideraion. With an estimated 145,000 individuals collected from the wild during 1997-2002, A. ocellaris is the most frequently traded marine ornamental fish in the global market [30]. The research was conducted in Spermonde Archipelago (Indonesia), where anemonefishes, especially A. ocellaris, are intensely collected and overexploitation is indicated [33]. A recent study showed limited connectivity of A. ocellaris populations across Indonesia and among shelf areas in Spermonde Archipelago, predicting high self-recruitment in the mid-shelf area of the archipelago [34]. In this study polymorphic microsatellite DNA parentage analysis was used to investigate the degree of selfrecruitment, site fidelity, and genetic relatedness of A. ocellaris and A. perideraion populations of two small islands in Spermonde Archipelago. Understanding the degree of self-recruitment in these populations and their connectivity to neighbouring populations could directly support the design and implementation of effective Marine Protected Area (MPA) networks, as well as the sustainable management and conservation of these species.

Study Areas
Spermonde Archipelago (South Sulawesi, Indonesia) ( Fig. 1) comprises about 150 islands [35] and is situated at the southwestern tip of Sulawesi in the centre of marine biodiversity, the so-called ''Coral Triangle''. This archipelago is affected by the very strong Indonesian Throughflow (ITF) current, which connects the Pacific Ocean with the Indian Ocean [36]. This setting potentially enhances the dispersal of marine organisms in Spermonde Archipelago, though interactions between oceanographic processes and larval behaviour may enable larvae to stay close to their natal population [37]. About 50,000 people live in Spermonde Archipelago and coral reef resources form an important part of their livelihoods. Therefore, these reefs are under threat from a variety of anthropogenic activities, including destructive fishing practices and land-based pollution [38]. The present study was conducted at two small islands, Barrang Lompo and Samalona (Fig. 1), located in the mid-shelf region of Spermonde Archipelago. Barrang Lompo (5u02952.070S, 119u19945.250E), located 13 km west of Makassar, is 19 ha in size and inhabited by about 5,000 people. Its fringing coral reefs have been impacted by dynamite-fishing and local sewage pollution [39]. Samalona (5u07930.480S, 119u20936.480E), located 5 km west of Makassar, is 2 ha in size, and inhabited by about 80 people. Samalona has been developed by local people for smallscale tourism and is therefore relatively protected from destructive fishing activities. However, Samalona9s reefs have been impacted by anchor damage and pollution from Makassar [39]. On both study islands, the fringing reefs extend from the shore to depths of 2-10 m, where the substrate changes to soft sediment. The outer circumferences of the fringing reefs are 2.5 km and 1.48 km at Barrang Lompo and Samalona, respectively.

Field Sampling Methods
In order to completely sample the populations of A. ocellaris and A. perideraion on Barrang Lompo and Samalona, scuba divers systematically searched the entire area of the fringing reefs for host anemones. To facilitate this process and provide more information on location of the anemonefishes, the reef in Barrang Lompo was subdivided into five sites and the reef in Samalona into four. At each site, all host anemones were located and associated anemonefishes identified and counted. To obtain tissue for genetic analysis of the anemonefishes, two small aquarium nets were used to carefully capture each individual fish and a small fin clip of the caudal fin was collected. The length of the fish was measured and then it was immediately released back to the host anemone. Finclipped individuals could be readily identified, so resampling was not a problem and it was possible to visually ensure that all individuals associated with a particular anemone were sampled. Each tissue sample was put into a separate tube and all associated data (fish species, size, date, location, and anemone species) was recorded immediately. Tissue samples were preserved in 96% ethanol after the dive and stored at 4uC in the laboratory until DNA extraction.
The sex and reproductive status of individuals within each group on a host anemone were determined by body size. The largest fish was assumed to be the reproductive female, the second largest the reproductive male, and all others were assumed to be non-breeding individuals [31] [40]. Non-breeding individuals will be referred to as ''juveniles'' henceforth.
A total of 364 tissue samples of A. ocellaris and 105 tissue samples of A. perideraion were collected at the two islands (

Ethics Statement
Fin-clipping is a non-destructive, minimally invasive and the most commonly used method to obtain tissue from living fishes in the wild (e.g. [24] [26] [28]) and in aquaculture (e.g. [41]). We took great care to minimise harm, and ensure survival by safely releasing the fishes back into their host anemones. Tissue sampling of these anemonefishes was permitted within the framework of the German-Indonesian SPICE project (Science for the Protection of Indonesian Coastal Ecosystems), in cooperation with the Hasanuddin University, Makassar, Indonesia. Genomic DNA from Amphiprion ocellaris and A. perideraion was extracted with the NucleoSpin tissue extraction kit (Macherey-Nagel), following the manufacturer's guidelines. All DNA extracts were analysed by gel electrophoresis to monitor DNA quality prior to polymerase chain reaction (PCR) amplification of microsatellite loci. DNA extracts were stored at 220uC.
PCRs were carried out in a total volume of 25 ml, containing 2.5 ml 10x PCR buffer, 3 ml 25 mM MgCl 2 , 1 ml 2 mM each dNTP, 1 ml each 10 mM primer forward and reverse, 0.1 ml (5 unit/ml) Taq polymerase (F100L Taq DNA), 1 ml (1-10 ng) genomic DNA. PCRs were performed in a TProfessional Thermocycler (Biometra) or a Mastercycler ep (Eppendorf) with the following thermo-profile: 94uC for 2 minutes, followed by 35 cycles of 94uC for 30 seconds as the denaturing step, 50-65uC for 30 seconds as the annealing step (the optimal annealing temperature varies between primers, see Table 2), 72uC for 1 minute for the polymerisation, and finally 72uC for 2 minutes. Twenty microsatellite loci were amplified, using primers from other Amphiprion species [29] [42] [43]. Of these, seven were polymorphic in A. ocellaris and five in A. perideraion. These loci ( Table 2) were amplified by PCR with a labelled forward primer containing a 59-fluorescent dye (FAM or HEX). PCR products were diluted in pure water prior to fragment analysis. Dilution factors were determined empirically for each locus, and ranged from 1:5 to 1:30. For fragment analysis, 1 ml of diluted PCR product was combined with 8.85 ml HiDi TM formamide and 0.15 ml GENESCAN LIZ-500 size standard (Applied Biosystems). Microsatellite fragments were size fractioned using an ABI 3730 48 capillary sequencer with a capillary of 50 cm length (Applied Biosystems). Allele sizes were determined and corrected with PEAK SCANNER v1.0 (Applied Biosystems) and GENEMAR-KER v1.85 (SoftGenetics GeneMarker). The program MICRO-CHECKER was used to detect null alleles and to identify irregularities in the data, including mistyped allele sizes, typographic mistakes, as well as scoring errors [44].

Summary Statistics, Test of Hardy-Weinberg Equilibrium, and Linkage Disequilibrium
The total number of alleles per locus, allele frequencies, observed and expected heterozygosities [45], and the Polymorphic Information Content (PIC) of each locus [46] were calculated with the program CERVUS 3.0 [47]. The levels of polymorphism at each microsatellite locus was ranked as: (1) highly informative (PIC .0.5), (2) reasonably informative (0.5$PIC $0.25), or (3) slightly informative (PIC ,0.25), following Botstein et al. [46]. PIC values are determined based on the frequency of alleles at a given locus. Hardy-Weinberg equilibrium (HWE) exact tests and loci combinations for linkage disequilibrium with the Markov chain methods were conducted using GENEPOP on the web [48] [49]. In order to test the null hypothesis of HWE, the probability test was conducted and the alternative hypothesis of heterozygote deficiency was tested. The null hypothesis of linkage disequilibrium for the diploid case was tested through pairwise comparisons of loci. For all Markov chain methods, parameters used were the default settings for dememorisation number (1000), number of batches (100), and iterations per batch (1000). Significance levels were adjusted with sequential Bonferroni corrections for multiple tests with P#0.05. The coefficient of inbreeding (F IS ) was calculated with the program FSTAT 2.9.3 [50] in order to detect nonrandom mating within populations [51]. The F IS -value ranges

Parentage Analysis
Microsatellite DNA parentage analysis was conducted with FAMOZ [53]. Using a likelihood-based approach [54], juveniles were assigned to a single parent or parent pair in order to select the most likely parent from a pool of potential parents [22]. Suitability of this program for such analyses in fish populations was shown in several studies [23] [26] [27]. In this program, the exclusion probability [55] is generated, which uses incompatibilities between parents and offspring to reject particular parentoffspring hypotheses [22]. The log-likelihood ratios or logarithm of odds (LOD) scores were calculated with this program for each parent/offspring association, using microsatellite allele frequencies calculated by CERVUS 3.0. LOD score threshold values for each error type were taken from the intersection of offspring with genotyped parents and offspring generated according to allele frequencies. Simulations have to be performed to determine suitable thresholds, error levels, and the impact of scoring errors for each population [56]. Therefore, five different error rates were evaluated for all data sets (Table S1). Each error rate was evaluated with 10,000 replicates of simulated offspring. Finally, to compensate for error in scoring parents or offspring genotypes, the presence of null alleles, and marker mutation [47], an error rate of 0.01 was chosen for parentage analysis for all populations of A. ocellaris, and an error rate of 0.001 for A. perideraion populations (Table S1). LOD score threshold values for A. ocellaris for single parent and parent pair were 1.9 and 5.9 for the Barrang Lompo population, respectively. In the population from Samalona, the LOD score threshold values were 2.5 and 6.7, respectively. For A. perideraion, LOD score threshold values for single parent and parent pair were 2.8 and 7.9, respectively. Unassigned juveniles of A. ocellaris to a single parent or parent pair in Barrang Lompo were assigned to potential parents in Samalona, and vice versa. These two data sets were also simulated to obtain suitable LOD thresholds (Table S1). All tests showed high cumulative exclusion probabilities (.0.9). The parameters for the parent/offspring assignment decision were as follows [57]: (1) individuals were assigned to the most likely single parent if the LOD score was equal or larger than the single parent threshold and (2) individuals were assigned to the most likely parent pair if the LOD score was equal or larger than both the single parent and parent pair threshold. No parent assignment was made if the LOD score was less than the single parent threshold.

Dispersal Distances and Spatial Patterns of Recruitment
The results of parentage analysis were used to calculate the proportions of juveniles that: (1) recruited to the same anemone inhabited by their parents; (2) recruited to the natal site; (3) recruited to an adjacent site on their natal island reefs; (4) recruited to a non-adjacent site on their natal island reefs; (5) recruited from the other study island, or (6) had no parent identified from either study island. Satellite images were used to measure distances among sites both within and between islands. This information was used to estimate a minimum dispersal distance based on the locations of parent and offspring.

Genetic Relatedness
In addition to parentage analysis, a genetic relatedness index was calculated to determine whether individuals sharing an anemone were related to one another. Genetic relatedness among the individuals inhabiting each anemone (''anemone group'') was conducted using KINGROUP v2 [58]. In this program, the method ''kinship pairwise'' [59] was chosen to construct the coefficient of relatedness (r), which estimates patterns of kinship in natural populations. An r value less than zero means that individuals from the same anemone are unrelated, and an r value greater than zero means that individuals within anemones are related. Relatedness in anemone groups was calculated for three populations: A. ocellaris from Barrang Lompo (n = 53; 17 groups), A. ocellaris from Samalona (n = 276; 83 groups), and A. perideraion from Barrang Lompo (n = 100; 35 groups). Allele frequencies from each of the three populations were estimated separately to calculate relatedness. A permutation test was used to compare the relatedness values obtained between two individuals from the same anemone-group and individuals within the same island. The statistical analysis was conducted in BASE SAS 9.3 [60]. The mean coefficient of relatedness within reef-sites and within island were also calculated for both species.

Summary Statistics, Hardy-Weinberg Equilibrium, and Linkage Disequilibrium
All markers used in the analysis were ranked as highly informative ( Table 3). The average PIC value was 0.76660.161 (mean 6 SD) and ranged from 0.465 (locus 120) to 0.925 (locus Cf42) in the Amphiprion ocellaris populations, and 0.79960.154 (mean 6 SD) with a range of 0.601 (locus AC915) to 0.969 (locus Cf42) in the A. perideraion population. The number of alleles varied between five (locus 120) and 30 (loci Cf42 and AC137) in the A. ocellaris populations, and between five (loci AC915 and AC1578) and 57 (Cf42) in the A. perideraion population. The observed heterozygosity (H o ) ranged from 0.5 (120) to 0.977 (Cf29), and the expected heterozygosity (H e ) ranged from 0.539 (120) to 0.931 (Cf42) in the A. ocellaris populations. In the A. perideraion population, H o ranged from 0.267 (AC1578) to 0.952 (Cf42), and H e ranged from 0.653 (AC915) to 0.975 (Cf42). The probability test indicated that locus AC1578 was not in HWE (P,0.01; Table 4). However, the alternative hypothesis of heterozygote deficiency was rejected (P = 0.087). Therefore, this locus remained in the dataset for further analysis. In the A. perideraion population, HWE tests indicated two loci (Cf42 and AC1578) deviating from equilibrium. Only locus 45 showed evidence of null alleles. However, these loci remained in the dataset for further analysis as well. No significant linkage disequilibrium was found for any loci pair, indicating that all loci could be considered as independent.

Parentage Analyses and Spatial Patterns of Recruitment
Single-parent or parent-pair assignments were made for 94 of the 169 Amphiprion ocellaris juveniles and 28 of the 64 A. perideraion juveniles (Table 4). Of these, the majority represented withinisland recruitment (44-65%), with between-island recruitment quite low for A. ocellaris on Samalona (4%) and higher on Barrang Lompo (23%). Within-island recruitment patterns showed that juveniles were most likely to be found on reef-sites that were adjacent to their natal site and least likely to be found on nonadjacent sites (Fig. 2).

Site Fidelity
Juveniles of A. ocellaris and A. perideraion were staying at their natal site in different proportions (Fig. 2). The percentage of juveniles of A. ocellaris that returned to their natal site in Barang Lompo range from 0 to 44%, while A. perideraion ranged from 0 to 19%. In the Samalona, the percentage of A. ocellaris juveniles that returned to their origin site ranged from 8% to 11%. However, most of them settled and dispersed close to their natal site within their island.

Genetic Relatedness
The mean coefficient of relatedness of Amphiprion ocellaris individuals within anemone groups at Barrang Lompo was 0.10860.162 (mean 6 SD; n = 17 groups), which was significantly higher than the mean within-island relatedness of 0.04760.111 (SD) (P = ,.0001, Fig. 3A, Table 5). Thus, at Barrang Lompo, A. ocellaris sharing an anemone were more closely related to one another than to other individuals on the island. In contrast, at Samalona, the mean coefficient of relatedness of A. ocellaris individuals within anemone groups was 0.00160.114 (mean 6 SD; n = 83 groups), not significantly different than the mean of 0.00460.056 (SD) for within-island relatedness (P = 0.515, Fig. 3B). The value for within-anemone group relatedness of A. perideraion at Barrang Lompo was 0.05160.151 (mean 6 SD; n = 35 groups) significantly higher than the mean of 0.00860.164 (SD) withinisland relatedness (P = ,.0001, Fig. 3C, Table 5). Within-reef relatedness ranged from 20.0460.24 to 0.3060.01 (mean 6 SD) for Amphiprion ocellaris populations, while for the A. perideraion population it ranged from 20.0660.15 to 0.1560.12 (mean 6 SD).

Self Recruitment
The present study revealed high self-recruitment of anemonefishes within reefs surrounding small islands, with 47-65% of Amphiprion ocellaris and A. perideraion progeny staying on their natal island. Self-recruitment of A. ocellaris (65%) at Samalona is higher than the 42% reported from a previous study on the sibling species A. percula [26]. The high self-recruitment in Samalona and Barrang Lompo is in agreement with evidence of restricted gene flow revealed in Spermonde Archipelago and across the Indo-Malay Archipelago [34]. Restricted dispersal might be triggered by the sheltered environment within the mid-shelf of Spermonde Archipelago, where the study islands are located, compared to outer shelf of the archipelago, which are strongly affected by the Northwest Monsoon [35].
The pelagic larval duration (PLD) varies from days to weeks in different species of coral reef fish [7] and thus may influence their dispersal distance. Due to pelagic dispersal of eggs and larvae, most marine species have been considered as open populations [63], even though this is under discussion [64] [65]. In an open marine population, fish larvae are assumed to be transported by Table 4. Recruitment within-island (self-recruitment) and between-island of anemonefishes in Barrang Lompo and Samalona, (Spermonde Archipelago. Indonesia).  [68]). The findings mentioned above are supported by the current study. Therefore, it seems as if the dispersal of pelagic fish larvae may be more restricted and marine systems may not be as open as previously assumed.

Site Fidelity and Spatio-temporal Patterns of Recruitment
Larval dispersal in marine organisms can vary from less than 1 km to 100 s km [27] [28] [69][70][71], which may affect their spatial and temporal recruitment. This study shows that most anemonefish larvae settle less than 2 km from their natal reef site. The populations of Barrang Lompo and Samalona were connected by a moderate exchange rate of 10 to 19%. A study on self-recruitment in Amphiprion polymnus showed that even though no individuals settle into the same anemone as their parents, most of them settled in other anemones close to them [23], which is in concordance with the findings in A. ocellaris and A. perideraion of the current study. Recruitment of A. ocellaris in Barrang Lompo was high but showed a slight difference between the two sampling periods. This could be natural variation or due to the collection for marine ornamental trade, as explained above. With a proper estimation of connectivity and degree of self-recruitment in marine populations, it might be possible to improve the design of marine reserves. For example, a series of small Marine Protected Areas on each island with short distance to each other can serve to maintain local populations both by self-recruitment and through larval dispersal from nearby reserves [1] [26].
The connectivity within and among populations might be also influenced by the behaviour of the planktonic larvae, as shown in the coral reef fish Amblyglyphidodon curacao that has the capability to swim against a current, control its vertical position, and locate a reef [72]. The early inception of the active larval movement is important to mediate the dispersal potential [70], such as olfactory sensing that might also influence larval movement [73]. Planktonic larvae of marine fish are assumed to be able to recognise and return to their natal site. It was shown that Amphiprion ocellaris imprints itself to its species-specific host sea anemone using olfactory cues, which are genetically inclined towards olfactory recognition of their host anemone [12]. However, this study found  that juveniles come back to their natal reef but not to their natal anemone. In order to facilitate retention, fish larvae may use odour recognition [74]. In addition, the connectivity within and among populations in anemone fish is limited by their relatively short PLD of about 8-12 days. However, exploitation of species, leading to decreased population density and body size [33] [75] [76], could also reduce larval abundance and reduce the dispersal kernel and effective connectivity distances [77].

Genetic Relatedness
The average genetic relatedness in Barrang Lompo populations showed that A. ocellaris and A. perideraion individuals within an anemone were more closely related to one another than to other individuals on the island, indicating that fish larvae do not disperse far from their parents. The values of genetic relatedness are in concordance with the coefficient of inbreeding. The current study observed low but positive values of the inbreeding coefficient for A. ocellaris and A. perideraion at Barrang Lompo, meaning that there is an indication to inbreeding in these populations.
The close relationship between individuals within a site at Barrang Lompo might be explained by several mechanisms. Many marine fishes have the ability to recognise their relatives in order to avoid inbreeding and competition [78]. However, due to a low abundance of anemones as a result of removal by the ornamental fishery [33], the pelagic larvae might not be able to avoid settling to their natal anemone. Close relatedness between individuals in an anemone may lead to inbreeding, which increase homozygosity, causes deleterious alleles in the first generation and reduces the adaptive capacity of species, and could lead to increased mortality [79] [80].
In contrast, individuals inhabiting an anemone were unrelated in A. ocellaris at Samalona. These results agree with findings in A. percula [29] and Dascyllus aruanus [81], forming groups consisting of unrelated individuals.

Implications for Management and Conservation
Anemonefishes have been exploited for marine ornamental fishery and traded globally for many years. Most marine ornamental fish species are collected from the wild, with Indonesia and the Philippines as the major exporters [30]. In order to avoid overexploitation and to reduce the pressure on natural populations, some efforts have been made to rear ornamental fish species [82][83][84]. Anemone fish such A. ocellaris have been successfully bred in captivity [85], which is potentially a good solution to meet the high demand. However, mariculture needs comprehensive knowledge of the biology of the reared fish species and is expensive [83] [86]. Therefore, it seems that mariculture would be difficult to be implemented for many ornamental species, especially in developing countries. In Spermonde Archipelago, A. ocellaris is the most collected marine ornamental fish species and this fishery has a negative impact [33]. However, there is no quota for anemonefishes in Indonesia so far.
The two focal species (A. ocellaris and A. perideraion) have not yet been included in the CITES list (Convention on International Trade in Endangered Species of Wild Fauna and Flora), which aims to prevent overexploitation by controlling international trade [87]. Considering the high level of exploitation, a proper management plan and conservation strategy should be implemented for this species in Spermonde Archipelago [33]. Implementation of MPAs as a tool to manage fisheries and marine biodiversity has been used to protect coral reefs from overexploitation, increase fish populations, restore ecosystem health, and prevent local extinctions [88][89][90][91]. However, determining the optimal size and location of self-sustaining MPAs is essential for promoting population persistence [6] [92][93][94].
The high amount of self-recruitment of Amphiprion ocellaris and A. perideraion found in the current study gives valuable information for management and conservation strategies within the region. High levels of self-recruitment imply that the populations are more vulnerable to local fishing activity [95]. While further studies on other marine organisms are needed, the current study suggests that single marine protected areas (MPAs) are not suitable as sources for the replenishment of exploited populations. Small MPAs, preferably at every island or reef, should ensure that a part of the population is protected to allow for replenishment by selfrecruitment. In addition, the population size should be estimated in order to establish appropriate catch quotas.

Supporting Information
Table S1 Error rates on parentage assignment and gene flow for different populations of parents and juveniles.