Mutations in microsatellite loci arise due to changes in repeat numbers and nucleotide substitutions. In total, 274 different colonies were successfully sequenced from the above six microsatellite loci of four Anguilla eels. Ranges of allelic sizes at the six loci from four Anguilla eels (Figure 2) indicated overlap across these species. Patterns of repeat regions at the six loci are listed in Table 1. The AJ-1 locus from 49 colonies was sequenced, with repeat numbers of five to 13 in the major region. An interrupted sequence from one individual was observed in A. marmorata. TG repeat numbers at the AJ-8 locus ranged 12–22. A substitution of TG/TA was only found in A. anguilla. The AJ-9 locus from 40 colonies was examined from the four Anguilla species. Interrupted nucleotide mutations, including TG/AG, TG/TA, and TG/CG, appeared in three of these freshwater eels, but not in A. marmorata. Allelic size variations were fully displayed by differences in numbers of repeat units. The repeat number of the dinucleotide, TG, at AJMS-3, ranged from seven to 11. Two interrupted sequences were only found in A. anguilla. Perfectly repeated sequences of the AJMS-6 locus were present in three species, but not in A. bicolor pacifica. Perfect GA repeat numbers of eight to 39 occurred at the AJMS-10 locus in the four species. Substitutions consisting of three mutations of GA/GG, GA/CA, and GA/AT in major regions only occurred in A. bicolor pacifica and A. anguilla. After sequencing, we observed that several types of mutations appeared in homologously-sized alleles with the same nucleotide number of the AJ-9 and AJMS-6 loci from A. bicolor pacifica specimens. Some alleles at the AJ-1 locus of A. anguilla also showed an identical size to those of other alleles, which resulted from the occurrence of one dinucleotide deletion in the flanking region.

Consequently, compositions of the major regions at the six microsatellite loci of Anguilla eels were generally categorized into two types: perfect- and interrupted-repeat sequences. No single-nucleotide indel was found in the major region of any of the six microsatellites from the four studied eel species. As to genetic variations in major regions of the six loci, no specific mutants appeared in the two temperate or tropical freshwater eels.

In each species, six consensus sequences were built up from all of the homologous sequences of the six loci, and they were then combined together. Lengths of the four consensus sequences ranged 458–469 bp. Mean A + T components (51.16%) were slightly higher than G + C ones (48.84%). In total, 68 point mutation sites were found among the four Anguilla eels with variations including 61.97% substitutions and 38.03% indels. All 27 indel sites, including 23 monomorphic and 4 polymorphic sites, and 44 substitution sites, were present at the consensus sequences of the six loci among the four Anguilla species. Some particular dinucleotide mutations were only present in certain species. For example, an insertion of CG was only observed at loci AJMS-3 (positions 31 and 32) and AJ-8 (positions 45 and 46) in A. anguilla, and at locus AJMS-6 (positions 25 and 26) in A. marmorata (Figure 3). Single-nucleotide substitutions were extensively distributed at the six microsatellite loci in the four species. A K substitution was found at position 20 of the AJ-1 locus in A. marmorata. More substitutions were found at alleles of the AJ-8 locus in A. anguilla than in other species. For example, a substitution of A/G was specifically observed at positions 11, 12, and 15, and C/A at position 40. A specific G/C substitution (position 23) at locus AJ-9 was only found in the two tropical species but not in the temperate eels. There were three specific substitutions at positions 5, 7, and 9 of the AJMS-10 locus in A. bicolor pacifica. In summary, the mutation patterns in the flanking regions primarily resulted from indels and substitutions. Indels in the flanking regions are believed to be one cause that results in allelic size variations.

Numbers of transitions (Ts) between species in the combined consensus sequence ranged from eight (A. japonica vs. A. marmorata and A. marmorata vs. A. bicolor pacifica) to 11 (A. anguilla vs. other species). Numbers of transversions (Tv) ranged from 12 (A. japonica vs. A. anguilla and A. marmorata vs. A. bicolor pacifica) to 16 (A. anguilla vs. A. marmorata). These results implied that fewer substitutions appeared in tropical A. marmorata vs. A. bicolor pacifica than in temperate A. japonica vs. A. anguilla. The interspecific genetic distances obtained from the K2P genetic model ranged from 0.044 (A. marmorata vs. A. bicolor pacifica) to 0.061 (A. marmorata vs. A. anguilla) (Table 2). The NJ topology among the four Anguilla eels indicated that A. marmorata had a closer evolutionary relationship with A. bicolor pacifica than with temperate eels, and A. japonica was not clustered with A. anguilla (Figure 4). These results implied that interspecific allopatric evolution existed between these two temperate eels.

Changes in repeat numbers of major regions, substitutions in entire sequences, and indels in flanking regions, are principle reasons resulting in polymorphisms within microsatellite loci. Previous reports indicated two mutation models that may cause changes in repeat numbers and lead to microsatellite instability. One model is unequal crossing-over that is the result of a recombination between homologous chromosomes that are misaligned. The alternative model is slip-strand mispairing errors that occur during DNA replication [31]. When major regions of perfect microsatellites produce a single-nucleotide substitution, it may result in interrupted microsatellites. In this study, several identical single-nucleotide substitutions were present in these Anguilla species (Table 1). For example, an interrupted major region, (TG)_n AG (TG)_m, at locus AJ-9, was found in all of these Anguilla species except A. marmorata. In contrast, some specific, interrupted mutations were only discovered in particular species. For instance, a TA substitution in the major region of the AJMS-6 locus was specifically present in A. bicolor pacifica, but not in other species. As a result, this sequence character can be considered a good genetic marker to identify A. bicolor pacifica. However, these mutations have induced high genetic divergences within and among Anguilla species.

In this research, most homologous alleles of these microsatellites sequenced across species revealed that the major mutational event was a change in the repeat number, but these interrupted sequences from the same or different species of Anguilla also produced homologous allelic sizes. For example, the major regions at locus AJ-1 with TC (TG)₆ and (TG)₇ expressed an identical allelic size in A. marmorata. The sequence of the interrupted repeat at locus AJ-9, (TG)₈ AG (TG)₁₀, in A. japonica had the same sequence length as (TG)₁₉ in A. marmorata. Most of the interrupted sequences were found in A. bicolor pacifica, A. anguilla, and A. marmorata rather than in A. japonica (Table 1), which suggests that greater complexities were present in those three species, in contrast to a rather simple form in A. japonica. The lack of a significant relationship existing between genetic distances and DNA complexity in these microsatellites suggests it is possible that substitutions randomly occurred within major regions of these microsatellites. In this study, indels were not found in major regions of the six microsatellites from these Anguilla eels. Consequently, random mutations are another factor which can cause variations in complexity among different species. Stephan reported that a single-nucleotide indel did not occur in any major region of microsatellite loci, which indicated that principal mutational patterns of microsatellites involved changes in the repetition number by the two cardinal mechanisms of slippage-strand mispairing and unequal crossing-over, and random substitutions within major regions [32]. Those inferences of evolutionary mechanisms are consistent with the results of this study. All these results imply that parallel evolution is present in these Anguilla eels. Kuittinen et al. described a parallel pattern of microsatellite sequence variations within and between populations of Arabidopsis thaliana [33]. A similar result was also revealed at the Anguilla species level (Figure 2, Table 1).

In fish, indels and substitutions are fairly frequent in flanking regions of microsatellite loci. Blankenship et al. observed point mutations and typed 668 different microsatellite flanking-sequence haplotypes from Chinook salmon [34]. In addition, microsatellite size variants due to indels in flanking regions were previously described [18,35]. In this study, indels were also found in flanking regions of five microsatellite loci (but not locus AJ-9) (Figure 3) and produced greater divergence of allelic sequences. Our results implied that mutational processes at these loci are probably far more complex than expected from the simple model of changes in the number of repeat units. There were no unique mutant patterns present in temperate or tropical freshwater eels.

Patterson estimated that Anguilla originated approximately 50–60 million years ago (Mya) during the Cretaceous-Eocene period, which is compatible with an Eocene Anguilla fossil collected from the Ypresian Stratum at Montevorca, Italy [36]. Based on the “Tethys Corridor hypothesis”, ancestral eels entered the Atlantic Ocean before the closure of the Tethys Sea (in the Oligocene, ca. 20~30 Mya) [37]. Using molecular dating, Aoyama and Tsukamoto determined that the two Atlantic Anguilla eels and A. mossambica diverged from a common ancestor at least 20~30 Mya [38,39]. However, we conjecture that the existence of these loci within Anguilla genomes must exceed 30 Mya, when the genus Anguilla may have first begun to diversify. In this study, six cross-specific microsatellite loci were successfully amplified from Anguilla eels. The conservation of a basic structure revealed by the sequence analysis among all species confirmed the homology of these loci within the genus Anguilla (Figure 3). Thus, the flanking sequences of the six microsatellite loci can also be used to clarify phylogenetic relationships of freshwater eels.

Ege subdivided Anguilla eels into four distinct groups, and the largest group with a long dorsal fin and uniform coloration included A. rostrata, A. anguilla, A. japonica, A. mossambica, A. dieffenbachii, and A. borneensis [3]. It was inferred that all six species within this group were more primitive than the others. They dispersed into different oceans before the closure of the Tethys Sea, with the subsequent formation of particular migration loops and speciation models once they were established in various oceans. In this study, we analyzed the phylogenetic relationship of four Anguilla eels using flanking sequences from six microsatellite loci. Although A. anguilla and A. japonica have similar adult-phase morphological characteristics and temperate habitats, our results suggested that A. japonica was not clustered with A. anguilla in the phylogenetic tree (Figure 4). The result is very similar to that inferred from mtDNA [10].

Most Anguilla species inhabit the Indo-Pacific Ocean, and it is firmly believed that this ocean is the center of their speciation. Marble eels were the next group to occupy the Indo-Pacific Ocean. Anguilla marmorata is considered to be the most primitive type among the marble eels, and it has the broadest species range which extends from southern Japan to southeastern Africa. However, if sympatric evolution seems logical, we can clearly imagine that A. bicolor pacifica, A. marmorata, and A. japonica have more-similar genetic components than those of the allopatric A. anguilla. These results seem to better conform to earlier results.

Ten specimens of A. japonica were caught in the estuary of the Tanshui River in northern Taiwan (N25°15′, E121°25′); 10 specimens of A. anguilla originally sourced from Europe were donated by a fish farm in northern Taiwan; and 10 specimens each of A. bicolor pacifica and A. marmorata were collected from Hueá, Vietnam (N16°45′, E107°30′).

Muscle tissues of all individuals were immediately preserved in 95% ethanol after being caught until DNA extraction. Ethanol was removed from the tissues by evaporation before further treatment. The tissue (500 mg) was digested overnight in 1 mL lysis buffer (10 mM Tris-HCl (pH 8.0), 2 mM EDTA, and 10 mg/mL dithiothreitol) and 55 μL of proteinase K (0.5 mg/mL) at 55 °C before DNA extraction. DNA was extracted following standard procedures [40]. Genomic DNA was quantified and diluted to a working concentration of 1 ng/μL. We chose six conserved GA/GT dinucleotide microsatellite loci cloned from the A. japonica genome. The AJMS-3, -6, and -10, and AJ-1 and -8 microsatellite sequences in EMBL (with the respective accession numbers of AJ297601, AJ297603, AJ297605, AJ845112, and AJ845113) were described previously [41,42]. The AJ-9 microsatellite locus was freshly cloned from the A. japonica genome for this study, and its accession number is AJ844913.

Six microsatellite loci from the four Anguilla species were amplified via a polymerase chain reaction (PCR). The PCR consisted of approximately 5 ng genomic DNA, 50 pmol of a reverse primer, 50 pmol of a forward primer, 25 mM dNTP, 0.05~0.1 mM MgCl₂, 10× buffer, and 5 U Taq polymerase (Takara, Tokyo, Japan), brought to a 125-μL volume with Milli-Q water. We ran one cycle of 4 min at 95 °C; 8 cycles of 30 s at 95 °C, 30 s at 50 °C, and 30 s at 72 °C; and 35 cycles of denaturing for 30 s at 95 °C, primer annealing for 30 s at 52~56 °C (the temperature varied depending on the primers and species), and a further extension of 30 s at 72 °C. We evaluated 10 μL of each product on a 2% agarose gel to check the PCR success and confirm the product sizes. In an initial survey of these four species of Anguilla eels, primers for these loci were annealed, and PCR products for 70%~90% of the samples were produced.

The remaining PCR-amplified products were run on 1% agarose gels. Bands visualized using ethidium bromide were purified from the gel. Subcloning was used to isolate haplotypes for sequencing purposes. Purified DNA was cloned into a pGEM-T easy vector (Promega, Madison, WI, USA) for each subclone, five colonies on each plate were randomly selected, and plasmid DNA was isolated using a mini plasmid kit (Geneaid, Taichung, Taiwan). In total, 274 different sequences were determined on an Applied Biosystems (ABI, Foster City, CA, USA) automated DNA sequencer 377 (vers. 3.3) using a Bigdye sequencing kit (Perkin-Elmer, Wellesley, MA, USA). A T7 or SP6 primer was used in the sequencing reaction each time. PCR cycle parameters for sequencing were 35 cycles of 30 s at 95 °C, 30 s at 50 °C, and 60 s at 72 °C.

Sequences of the six microsatellite loci from four Anguilla eels were aligned using the BioEdit software program [43]. Inter- and intraspecific variations in sequences were determined using DNASP software [44]. Mutant patterns at major regions of the six microsatellites from the four species were separately visualized with the naked eye. Microsatellites were divided into three categories of perfect, interrupted, and compound types, based on the composition of their major regions [19]. All flanking regions of the microsatellite sequences were freshly aligned and analyzed. The consensus sequences were generated by combining information from point mutations of various intraspecific flanking regions at each locus with the WebLogo program [45]. All consensus sequences from homologous loci were combined and aligned, and interspecific variations were analyzed with the BioEdit [43] and MEGA programs [46]. Numbers of transitions (Ts) and transversions (Tv) were calculated with DAMBE software [47]. Genetic distances were computed, based on the Kimura two-parameter (K2P) model [48]. The topology was constructed using the Neighbor-joining (NJ) [49] method and bootstrap values were obtained by 1000 replicates [50].