Mating structure of the blue and red shrimp, Aristeus antennatus (Risso, 1816) characterized by relatedness analysis

Understanding life history variation and strategies is crucial for stock assessment and fisheries management due to the direct effects on population dynamics, effective population size, sex-ratios, levels of inbreeding, and relatedness among individuals. Aristeus antennatus (En ─ Blue and red shrimp; Fr ─ Crevette rouge; Sp ─ Gamba rosada) is one of the most exploited demersal resources in the Western Mediterranean Sea. However, information regarding the mating system and mate choice preferences remains largely unknown. Advances in molecular genetic markers and methods of inferring biological relationships among individuals have facilitated new insights into the reproductive dynamics of the species in the wild. Here, we used microsatellite markers to examine the A. antennatus mating system and putative mate choice preferences. Our results provided clear evidence of polyandry and polygyny. Relatedness analyses, together with FST and DAPC values showed females exhibited a mating bias towards unrelated males. Mating males were inferred from spermatophores and suggested males were sympatric with females and were also from other spawning grounds. Our findings provided the first description of the reproductive behavior of blue and red shrimp.


Materials and Methods
Sampling and DNA extraction. In August 2015, a total of 111 samples of mature A. antennatus adults (59♂: 52♀) were collected while on board the Nova Gasela trawling vessel at the spawning ground of Palamós (600 m depth), Spain (41°53′040″N and 3°23′777″E). Mature males were identified based on rostrum shorter than 12 mm in length 28 and the presence of petasma fusion 24 . Similarly, we followed Sardà and Demestre 29 to identify mature females: the cephalothorax was larger than 33 mm and one or more spermatophores were located in the thelycum. Individuals were quickly transported to the laboratory on ice to remove all the spermatophores from the females (61 spermatophores). Each spermatophore was stored separately in a 1.5 mL Eppendorf at −20 °C to extract DNA using the differential lysis protocol by Planella et al. 30 to attempt to genotype males successfully mating with females. In addition, a piece of muscle tissue from each adult individual was preserved in 95% ethanol until DNA extraction was performed using the genomic DNA extraction protocol outlined in Fernández et al. 31 .
Microsatellite loci. Genetic diversity at thirteen polymorphic microsatellite loci developed for A. antennatus 32 were analyzed using one singleplex and three multiplex PCR panels delineated to avoid primer-dimer and hairpin formation (assessed in silico with AUTODIMER v 1.0 program 33 ). For each locus, the forward primer was labeled with one fluorescent dye (PET, 6 -FAM, NED or VIC) and loci with overlapping allelic ranges were tagged with different dyes. Multiplex PCRs were conducted in a 10 µL final volume containing 1X GoTaq ® G2 Hot Start Colorless Master Mix (Promega Corporation), primer multiplex mix at primer-specific concentrations (Table 1), 0 or 1.5 mM MgCl 2 solution (adults and spermatophores, respectively), and ~40 ng of template DNA. The Aa1255 locus was analyzed in a singleplex panel performed in a 15 µL total reaction mixture containing 1X NH 4 Reaction Buffer, 1.5 or 4.6 mM MgCl 2 solution (adults and spermatophores, respectively), 1 mM of dNTP, 0.2 mM of each primer (forward and reverse), 0.375 units BIOTAQ TM DNA Polymerase (Bioline), and ~40 ng of template DNA. All PCR reactions proceeded as follows: initial denaturation step of 2 min at 94 °C, followed by 30 (multiplex) or 35 (singleplex) cycles of 30 s at 94 °C; annealing for 1 min 30 s at 50 °C or 60 °C; and elongation for 1 min 30 s at 72 °C; with a final extension of 30 min at 60 °C (Table 1). One µL of PCR product was mixed with 0.15 µL of GeneScan TM 500LIZ Size Standard (Applied Biosystems) and 10 µL of highly deionized (Hi-Di) formamide (Applied Biosystems) and denatured for 3 min at 95 °C. Resulting amplicons were run on an ABI PRISM 3130 Genetic Analyzer (Applied Biosystems) and genotypes scored using GENEIOUS R7.1 34 . www.nature.com/scientificreports www.nature.com/scientificreports/ Diversity within and between sexes. Genetic diversity levels within sexes (females and males) were estimated as the number of alleles per locus (N A ) and the observed (H O ) and expected (H E ) heterozygosity. In addition, at each sex and locus, the conformance of genotypic proportions to Hardy-Weinberg expectations (HW) was calculated using the program FSTAT v 2.9.3.2 35 and summarized using Wright's fixation index (F IS ). Loci exhibiting significant positive F IS values were evaluated for potential genotyping errors due to stuttering, allele dropout, or null alleles using MICRO-CHECKER v 2.2.3 software 36 . The Dempster et al. 37 algorithm was employed to estimate the frequency of null alleles; however modifications were implemented in the FREENA software 38 . Genetic diversity differences (N A , H O , H E , F IS ) between sexes were statistically tested in SPSS v 23 39 using the Wilcoxon signed rank test. Genetic differentiation between males and females was examined by F ST values using the program FSTAT v 2.9.3.2 35 .
Spermatophores indicated mature males effectively mating with females; therefore, we also examined these individuals for deviations from HW expectations and diversity levels, as detailed above for males and females. Diversity levels for the spermatophore group were compared with males and females and genetic differentiation from these two groups was estimated as F ST values. Furthermore, discriminant analysis of principal components (DAPC) using the package ADEGENET v 1.4-2 40 was performed in R v. 3.3.2 41 to establish an additional view of differentiation among the three groups (males, females and spermatophores). DAPC does not require any specific population structure model and it is free of assumptions regarding Hardy-Weinberg or gametic equilibrium 42 . Alleles are considered original variables with the largest between-group and smallest within-group variance.
Paternity and relatedness in the spawning ground. The exclusion probability of identity (PI) (probability that two randomly selected individuals matched their genotypes by chance) was applied to test the potential that the set of microsatellite loci was robust for individual identification using the software GIMLET v 1.3.3 43 . The regrouping genotype option was chosen to detect multiple paternity by comparing all spermatophore genotypes. This analysis was conducted to detect males mating multiple times with either the same female or different females. Similar comparisons were used to ascertain if some spermatophores originated from sampled males.
Relatedness analyses were used to determine if all specimens (females, males, and mating males inferred from spermatophores) in the spawning ground were derived from the same population. We first confirmed the efficiency of seven relatedness estimators available in COANCESTRY v 1.0.1.7 software 44 . We subsequently applied the related estimator to distinguish among low relatedness groups, which might be inherent in this species 45 ; 1,000 dyads were simulated for each of the following relationships: first cousins (expected relatedness: r xy = 0.125), second cousins (r xy = 0.031), and unrelated (r xy = 0). The best estimator was chosen for additional relatedness comparisons. Consequently, the distribution of observed pairwise relationship coefficients within each reproductive group (male, female, and spermatophores) was compared using COANCESTRY v 1.0.1.7 44 with the relationship coefficients expected for a sample of 1,000 unrelated simulated offspring generated by mating 1,000 simulated males and 1,000 simulated females from our male and female genotypes. All simulated individuals were obtained using HYBRIDLAB v 1.0 software 46 . In addition, comparisons of relatedness between all male, female, and spermatophore pairs were used to determine if all sampled individuals derived from the same population. Finally, average relatedness in observed female-spermatophore pairs was compared to all other potential female-spermatophore pairs (excluding observed female-spermatophore pairs) and female-male pairs to determine if mate choice was associated with kinship. For each of the above comparisons, 1,000,000 replicate iterations were computed in COANCESTRY v 1.0.1.7 44 and a genotyping error rate of 2% was allowed.

Results
Diversity analysis. All loci were polymorphic for each studied reproductive group (females, males, and spermatophores), with the number of alleles ranging from 2 to 19 (Table 2). Average observed heterozygosity (H O ) ranged from 0.3939 in adult females to 0.4668 in mating males, genotyped from the 61 spermatophores. In the three groups, larger average H E compared with H O was detected, ranging from 0.6116 in females, 0.6166 in males, and 0.6190 in spermatophores. Significant differences were not detected for these three diversity measures between males and females, but the Wilcoxon signed rank test indicated the average N A (Z = 2.072; P = 0.038) and H O (Z = 2.132; P = 0.033) were larger in spermatophores than in females. Several loci for each reproductive group showed significant positive F IS values and H O was often lower than H E . Following Bonferroni corrections, significant genotypic deviations from HW expectations were observed at eight loci in male and female groups and seven in spermatophores. MICRO-CHECKER v 2.2.3 revealed these significant differences were likely due to the presence of null alleles, but some biological processes also produced positive F IS values (e.g.: Wahlund effect from high migration rates). At loci where null alleles were detected, estimated frequencies ranged from 0.1017 at the Aa681 locus in the spermatophores group to 0.3572 at the Aa421 locus in males (Table 2). This latter locus (Aa421) was discarded from all subsequent analyses because its estimated null allele frequency was higher than 0.25 for the three groups.
Significant allele frequency differences were not detected between males and females (F ST = −0.0011; P = 0.5000) and males and spermatophores (F ST = 0.0016; P = 0.5500). Results did detect significant differentiation between females and spermatophores (F ST = 0.0091; P = 0.0167). Bidimensional DAPC scatterplots obtained from the 25 principal components retained more than 80% of the total genetic variance (Fig. 1a). The first component in the scatterplot depicted a distinction between females and their spermatophores. Males were located in a central position with an overlapping distribution at the left side of the central plot axis scattered with a portion of the females to the right of the main axis with some spermatophores. Spermatophores showed a somewhat bimodal distribution, scattered with females and males for the first component (Fig. 1b). Paternity and mating system. Excluding locus Aa421, a combined probability of identity exclusion higher than 0.9999 was observed in all three groups; males, females, and mating males represented by their spermatophores (Table 2). Therefore, the probability that two randomly selected individuals matched their genotypes by chance was lower than 0.0001. In order to check how null alleles could influence probability of identity we conducted an analysis with six loci (Aa138, Aa956, Aa496b, Aa123, Aa751, and Aa1195). We observed the same results we previously generated, suggesting null alleles have no influence on paternity. Multiple spermatophores were identified in six of fifty-two females (11.5%). Three females carried two spermatophores in their thelycum and another three female individuals carried three. In all cases, multiple spermatophores from the same female showed different genotypes. These results indicated multiple mating in females. Identity analysis on 12 microsatellite loci resulted in identical genotypes in one pair of spermatophores from different females, which was  www.nature.com/scientificreports www.nature.com/scientificreports/ consistent with the above estimated exclusion probability of identity, also supporting multiple mating in males. A genetic identity match was not found between spermatophores and males.
All relatedness coefficients produced small and similar estimates between simulated first and second cousin pairs, and more importantly, these values were not clearly differentiated from the values obtained between unrelated pairs of individuals (Table S1). A more robust relatedness estimator was not available for our data set. The triadic likelihood estimator was selected for all further analyses because it typically out performs other estimators in natural populations, where most dyads are unrelated or only loosely related 47 . The pairwise relatedness in our groups ranged from 0.0528 in males to 0.0657 in spermatophores ( Table 3). The average relatedness among   www.nature.com/scientificreports www.nature.com/scientificreports/ female individuals (0.0584) was not statistically different from the 1,000 simulated unrelated individuals (0.0591). Similar results were obtained for males, but the observed relatedness among spermatophores differed significantly from the simulated, indicating the spermatophores from males were more closely related than expected by chance alone ( Table 3). The average relatedness among females (0.0584) was not significantly higher than the average relatedness among males (0.0528) ( Table 3), but relatedness among spermatophores (0.0657) was higher than between all female pairs and male pairs (P < 0.01). In addition, the average relatedness between a female-spermatophore pair was similar to all other female-spermatophore pairs, but the value was slightly differentiated from all possible pairs involving a female and an adult male, suggesting some mating males were differentiated from the sampled males (P < 0.1) ( Table 3).

Discussion
The results of our study showed a mean number of alleles per locus (9.38) ( Table 2) in A. antennatus lower than the value reported in Italian blue and red shrimp samples (13.5) 48 . However, our results were consistent with those reported for other shrimp species, including whiteleg shrimp (Litopenaeus vannamei Boone, 1931; mean: 7.9) 49 and the deep-sea hydrothermal vent shrimp (Rimicaris exoculata Williams and Rona, 1986; mean: 8.73) 50 . Cannas et al. 48 reported higher mean H O (0.65) with an increased number of alleles compared with our data (mean females: 0.39; mean males: 0.42). In both studies, deviations from HW genotype expectations often resulted from heterozygote deficits in several microsatellite loci, a common result in the species of Penaeoidea 51-53 . Several explanations for the pattern have been proposed, including stutter bands, null alleles, and biological factors, e.g. mating tactics, recent population bottlenecks, and/or subpopulation structure. In fact, Wahlund effects resulting from subpopulation structure could not be rejected, as indicated by DAPC results on sampled spermatophores (Fig. 1b). Our results provided clear evidence for the A. antennatus mating system involves multiple mating for both sexes, suggesting that the reproduction in A. antennatus involved polyandry (11%) and polygyny (1.5%). We confirmed female multiple mating by conducting an extended genotyping of spermatophores from 20 additional females collected from the same spawning ground in 2016. Each female carried more than one spermatophore in her thelycum and in each sample different male genotypes were obtained (Table S2). Multiple female mating (polyandry) is common across crustacean species, particularly in species where females do not store sperm in specific structures 54 . Multiple mating in invertebrate females: (i) facilitates retention of genetic diversity and increased offspring fitness (e.g., the Pacific gooseneck barnacle 55 ); (ii) prevents the risks of choosing an infertile partner (e.g., polychaetes 56 ); (iii) favors the storage of enough sperm to fertilize the entire clutch (e.g., lepidopterans 57 ); (iv) and might prevent mate competition that causes injuries (e.g., the common rock shrimp 6 ). In the case of A. antennatus, the benefits of multiple mating could be a compendium of the first three factors, due to the high fecundity of females (≅600,000 oocytes produced in each vitellogenesis 26 ); the latter option is unlikely due to the sex ratio of the species (females seems to be much more abundant than males 23 ) and the small size and without hypertrophied weaponry of the males (major chelipeds and third maxillipeds are evident in aggressively dominant males 58 ). Nevertheless, multiple mating does not always translate into multiple paternity. For instance, females of some crustacean species show behavioral and physiological mechanisms to control male fertilization, such as selective sperm passage or spermatophore removal 59,60 . In the caridean rock shrimp (Rhynchocinetes typus H. Milne Edwards, 1837), females accept multiple mating to avoid injury by harassing males, but afterwards remove spermatophores from subordinate males to guarantee sperm from the highest quality males 61,62 . Thus, our A. antennatus findings demonstrated multiple mating, but did not ensure multiple paternity.
Our data demonstrated that males and females were not genetically differentiated, but mating males, inferred from spermatophores, were somewhat distinct from sampled males. In fact, the DAPC scatterplots showed a portion of spermatophores did not overlap with adult males (Fig. 1b) and females were clearly differentiated from their spermatophores. Females apparently bias mating towards unrelated males. Mating males inferred from spermatophores might include those sympatric with females, as well as those from other spawning grounds. Evidence for mate choice in favor of genetically dissimilar partners was reported in some invertebrates, interpreted as a way of inbreeding avoidance, which improves reproductive survival and individual fitness, and increases genetic diversity of local populations 63 , and references therein. Liu et al. 64 found that females of the cabbage beetle (Colaphellus bowringi Baly, 1865) preferred to mate with non-siblings rather than siblings, decreasing the potential for inbreeding depression. Simmons et al. 65 observed similar behavior in the Australian black field cricket (Teleogryllus oceanicus Le Guillou, 1841), where females fertilized their eggs with sperm from non-sibling males rather than full-siblings. Gherardi et al. 66 suggested that invertebrates make genetic decisions regarding their mating partners; however, relatively little is known about the mechanisms by which recognition is performed, especially in deep-water species, which are not amenable to laboratory conditions, as is the case for A. antennatus. One possibility for the observations made is that these species, use waterborne chemical cues (olfactory) to obtain information about conspecifics and consequently distinguish between kin and non-kin individuals 67 . The big-clawed snapping shrimp (Alpheus heterochaelis Say, 1818) employs this recognition mechanism to differentiate between familiar and unfamiliar individuals 67 . Another possibility is these taxa develop an olfactory and contact chemoreception combination mechanism, comparable to the eusocial coral-reef shrimp (Synalpheus regalis Duffy, 1996), where colony members discriminate between nest-mates and foreign conspecifics 19,68 . Often, marine Penaeoidean shrimps with pure searching mating strategy have small sized males, as does A. antennatus, and use contact chemical communication to search for receptive females in reproductive aggregates [21][22][23]58 . This strategy, for us the most plausible in A. antennatus was observed in terrestrial (e.g., mangrove tree crab 69 ) and freshwater decapoda (e.g., freshwater atyid shrimp 70 ).
In general, when individuals of one sex do not disperse far from the natal group, they are expected to have increased relatedness on average than those that disperse 71 . Our results showed that the spermatophores were differentiated from sampled males and females, suggesting that they were deposited in the female thelycum by A. antennatus migrating males. In contrast, using different set of loci Cannas et al. 48 analyzed the populations of www.nature.com/scientificreports www.nature.com/scientificreports/ A. antennatus of three Italian collections at different depths (Sant' Antioco, <800 m; Pesca Profonda-Sperimentale North, 1,464 m; Pesca Profonda-Sperimentale South, 1113 m) and suggest female-biased dispersal. The different depth sampling design might explain the lack of congruency between Cannas et al. 48 and present results (600 m). Oceanographic currents in the Mediterranean Sea are different among water masses at distinct depths (Surface Waters, <150 m; Intermediate Waters, 150-1,000 m; Deep Waters, >1,000 m) which could facilitate different dispersal patterns 72 . Contrasting results on dispersal patterns might also reflect temporal and spatial ecological variation affecting resource availability and population density [21][22][23]73,74 . Otherwise, the sex ratio obtained in Pesca Profonda-Sperimentale North (1:4) differed from that in the North -Western Mediterranean Sea 23,26,73 , which suggested that the former locality supports a unique population dynamic.
Our research demonstrated a polyandry and polygyny mating system in the blue and red shrimp. Additional research in different natural populations will be fundamental to fully elucidate the mating system in A. antennatus. Mating tactics might vary among populations depending on population density, mate quality, mate availability or variation in environmental conditions, as Gosselin et al. 75 reported in the American lobster (Homarus americanus). The apparent capacity of A. antennatus to select mating partners based on genetic dissimilarity provides the opportunity to pursue new research avenues and explore possible factors influencing mate choice in this species. Additionally, it will be important to include information regarding life history variation and strategies for stock assessment and effective fishery management of heavily exploited marine resources, such as A. antennatus, to ensure sustainability.

Data Availability
All data are included in the article.