Shell morphology and length of the nonrepetitive region of the foot protein 1 (adhesive protein) amplified by the polymerase chain reaction (PCR) using oligonucleotide primers, Me 15 and Me 16, were examined in mussels collected from Asamushi water, northern Japan, and compared to those of M. galloprovincialis Lamarck from Maruishi, Hiroshima, Japan. The Asamushi mussels were collected from a suspended rope substrate and M. galloprovincialis of Maruishi from a floating dock. The mussels of the Asamushi area have been referred to as M. edulis Linnaeus or M. edulis galloprovincialis Lamarck, and were recently supposed to be M. galloprovincialis and/or M. trossulus Gould based on the geographical distributions of these species. Although 14 of 19 variables of shell characters differed significantly between mussels of Asamushi and M. galloprovincialis from Maruishi, both canonical variates analysis using eighteen shell characters and PCR product analysis showed that the mussels from Asamushi were M. galloprovincialis. The result of PCR product analysis also indicated no hybridization between M. galloprovincialis and M. trossulus in the Asamushi area.
Mussels of the genus Mytilus (family Mytilidae) are widely distributed throughout temperate coastal waters. Because they have impressive abilities to dominate a variety of habitats such as rocks, embankments, and floating docks, they are an ecologically important component of shallow water benthic communities as dominant space occupiers (Suchanek, 1985), habitat providers for associated organisms (Tsuchiya and Nishihira, 1985, 1986; Nishihira, 1992a, b; Matsumasa, 1994; Matsumasa and Nishihira, 1994), and/or ecosystem engineers (Jones et al., 1994; Crooks, 1998). They are used as a nutritious human food source. Moreover, their wide geographical distribution makes them suitable for biomonitoring organisms to assess coastal water quality (Widdows and Donkin, 1992; Tanabe, 1994). Many scientific studies have been conducted on numerous aspects of the biology of this interesting coastal marine species.
Among the five species of the genus Mytilus, three closely related species of so-called “Mytilus edulis complex” are morphologically similar and their shell shapes are often influenced by the local environment. Therefore, it has been difficult to identify these species (M. edulis Linnaeus, M. galloprovincialis Lamarck and M. trossulus Gould) based only on morphological characters (see, Gosling 1992a), and the confusing taxonomy has hampered the development of studies in biology and ecology of the Mytilus edulis complex. Recently, however, various powerful methods such as electrophoretic (Gosling, 1984, 1992a, b for reviews), mitochondrial DNA (e.g., Skibinsky, 1985; Koehn, 1991; Rawson and Hilbish, 1995), and PCR (polymerase chain reaction) product analyses (Inoue et al., 1995) have been proposed in addition to a fairly sophisticated morphological method, canonical variates analysis of eighteen morphometric characters of shells (McDonald et al., 1991).
In the course of our ecological study on the roles of mussel species in marine benthic community organization at Asamushi, northern Japan, we had doubts about the taxonomy of the mussel. The mussels of the Asamushi area had been referred to as M. edulis or M. edulis galloprovincialis (Matsumasa and Nishihira, 1994 and references therein; see also, Huq and Nishihira 1997) and have been recently supposed to be M. galloprovincialis based on its geographical distribution (McDonald et al., 1991; Gosling, 1992a; Nishikawa, 1997 and references therein). However, some characters of shells of Asamushi mussels are apparently distinct from M. galloprovincialis collected from other sites (see, Figs 1–3). Besides, since Asamushi is close to Hokkaido where both native M. trossulus and introduced M. galloprovincialis are distributed (Inoue et al., 1995) and hybridization often occurs between species of the M. edulis complex (Skibinski et al., 1983; McDonald and Koehn, 1988; Skibinski and Roderick, 1991; Inoue et al., 1995), there is a possibility that the mussels of Asamushi were M. trossulus or represented a mixed population including hybrids between M. galloprovincialis and M. trossulus. To clarify this, we examined the mussels using (1) canonical variates analysis of 18 morphometric characters of shells (McDonald et al., 1991), and (2) PCR product analysis which discriminates interspecific differences in length of the nonrepetitive ‘variable region’ of the adhesive protein gene using a set of primers (Me 15 and Me 16; Inoue et al., 1995). In addition to the 18 morphometric characters used for canonical variates analysis, another shell character (distance between the anterior end of posterior retractor muscle scar and pallial line) was also examined, because it was reported that this distance was narrower in M. trossulus than in M. galloprovincialis (see, Nishikawa, 1997 and references therein). Using specimens of M. galloprovincialis from a ‘pure’ site (Maruishi, Hiroshima, Japan) for the sake of contrast, our results shed light on the geographical distributions of the Mytilus species and provide a test of the usefulness of morphological and PCR methods of identifying species at the boundary of geographical distributions of the Mytilus edulis complex.
MATERIALS AND METHODS
Specimens were collected in the vicinities of the Asamushi Marine Biological Station of Tohoku University at Asamushi, Aomori Prefecture (sampling date, 8 Nov. 1996), and of the National Research Institute of Fisheries and Environment of Inland Sea, at Maruishi, Hiroshima Prefecture (sampling date, 25 Feb. 1997), Japan (Fig. 1). The Asamushi mussels were collected from a rope that had been artificially suspended as the substrate and the Maruishi ones from a floating dock (sampling depth, ca 50 cm). On these substrates, both Asamushi and Maruishi mussels have formed their crowding beds and were densely packed. Eighteen shell characters used by McDonald et al. (1991) were measured on mussels from Asamushi and Maruishi. We measured an additional character (i.e., the distance between the anterior end of posterior retractor muscle scar and pallial line: character ‘e’ in Fig. 2) but did not use this for the canonical variates analysis based on McDonald et al. (1991). Characters were measured in millimeters using either electronic digital calipers (Mitutoyo, DC-20) or an ocular micrometer fitted to a stereo microscope (Nikon, SMZ-U). To standardize the variables for size, all characters but number of teeth were log-transformed (using base 10 logs) and divided by the log-transformed shell length. The shells used for morphometrics ranged from 25.01 to 63.66 mm and from 27.25 to 71.37 mm in length for specimens from Asamushi and Maruishi, respectively. These ranges in shell length were within that of mussels used by McDonald et al. (1991) (i.e, 23–80 mm).
Differences in the distribution of each morphometric variable were examined between Asamushi and Maruishi mussels by the Kolmogorov-Smirnov two sample test using SYSTAT (Wilkinson, 1992). First and second canonical variates for each individual were calculated using the raw canonical coefficients given in McDonald et al. (1991). The canonical variate for each individual is the sum of each morphometric variable (log-transformed and length-standardized) multiplied by its raw canonical coefficient. See McDonald et al. (1991) for details in the canonical variate analysis.
PCR products analysis
The lengths of the ‘variable region’ in the nonrepetitive domain of the adhesive protein gene (Inoue et al., 1995) were examined for 28 individuals from the Asamushi site (sampling date, 8 Nov. 1996) as well as M. galloprovincialis (n=30) from the Maruishi site (sampling date, 25 Feb. 1997) and M. trossulus (n=32) from Akkeshi, Hokkaido (sampling dates, 10 Jan. and 25 Apr. 1997) for the sake of contrast. The shell length of individuals from Asamushi ranged from 12.80 to 62.30 mm. A piece of the gill from each mussel, about 0.5 cm2, was incubated in 500 μl lysis buffer containing 50 mM Tris-HCL (pH 7.5), 10 mM EDTA, 0.5% SDS, 500 μg/ml Proteinase K at 55°C for 2–4 hr. Samples were then extracted twice with equal volumes of saturated phenol and twice with phenol chloroform: isoamyl alcohol (25:24:1). The aqueous phase was precipitated with ethanol and dissolved in 50 μl TE (10 mM Tris, 1 mM EDTA). About 100 ng of DNA was dissolved in 100 μl 1×Tth buffer (TOYOBO, Japan) containing 6 μg sense primer, 6 μg antisense primer, and 200 μM dNTP. After preheating to 95°C, 1 unit of Tth DNA polymerase (TOYOBO, Japan) was added and 35 cycles of amplification were performed. Each cycle consisted of 30 sec at 94°C, 30 sec at 56°C, and 90 sec at 72°C, using a set of oligonucleotide primers, Me 15 and Me 16 (Inoue et al., 1995). Ten microliters of PCR product was mixed with the loading dye solution containing bromophenol blue (BPB) and xylene cyanol and subjected to agarose gel electrophoresis [4.8% NuSieve GTG agarose gel (FMC)]. Electrophoresis was continued until BPB reached the end of the gel. The shell morphology was also examined for all individuals from the Asamushi site, in addition to the above PCR products analysis.
The shells of mussels from Maruishi (Mytilus gallprovincials), particularly of large individuals, exhibited the typical morphological characteristics of M. gallprovincials (see Gosling, 1984, 1992), but individuals from Asamushi only vaguely exhibited these characteristics (Fig. 1). The anterior ends of the shells of the Maruishi mussels were distinctly beaked or incurved, while those from Asamushi were not incurved. The shells of the Maruishi mussels tended to be higher and flatter than those from Asamushi.
Fourteen of 19 characters examined showed the significant differences in length-standardized size distribution between Asamushi and Maruishi specimens (Fig. 2). In particular, the largest level of significance (Kolmogorov-Smirnov two sample tests; p <0.0001) in the four characters (Fig. 3a–d) quantitatively indicated that the mussels from Maruishi were higher and grew longer anterio-dorsally than those from Asamushi. The distance between the anterior end of posterior retractor muscle scar and pallial line (character e in Fig. 2) was significantly narrower in Asamushi than in Maruishi mussels (p <0.0001; Fig. 3e).
All 46 mussels from Maruishi were certified as M. galloprovincialis by canonical variates analysis for morphometric data (Fig. 3f). Of the 59 individuals from Asamushi, 51 mussels (86.4%) were identified as M. galloprovincialis, 6 (10.2%) as M. trossulus, and 2 (3.4%) as M. edulis.
The sizes of amplified fragments of a part of the nonrepetitive region including the primer sequences (primers, Me 15 and Me 16) are 180, 168, and 126 bp for M. edulis, M. trossulus, and M. galloprovincialis, respectively (Inoue et al., 1995; see also Filpula et al., 1990). In addition, it is considered that the hybrid between M. galloprovincialis and M. trossulus exhibits two bands, 126 and 168 bp (Inoue et al., 1995). All 28 individuals of Asamushi mussels, including 5 individuals which were not morphologically considered as M. galloprovincialis, showed a single band of 126 bp, which was consistent with that of M. galloprovincialis from the Maruishi site (Fig. 4). This indicated that mussels from Asamushi were M. galloprovincialis, and that hybrids between M. galloprovincialis and M. trossulus were not included in the tested samples.
The results of both the PCR product and the canonical variates analyses indicated that the mussels from Asamushi were M. galloprovincialis (Figs. 3f and 4). Therefore, among the ecological and biological studies in the Asamushi area, the species name given to the mussel in Matsumasa and Nishihira (1994) must be changed from ‘Mytilus edulis L.’ to ‘M. gallprovincialis Lamarck’ at the very least. Both mussels in Matsumasa and Nishihira (1994) and in this study were collected from the same substrate at a similar depth from the same site. However, re-examination is needed for mussels dealt with in the other works of the Asamushi area (e.g., Hoshiai, 1958, 1960; Tsuchiya, 1979; Tsuchiya and Nishihira, 1985, 1986), because those mussels were from other habitats (i.e., intertidal natural rocks and embankments). It has been reported that beaches exposed to wave action are dominated by M. galloprovincialis, while more protected areas a short distance away are dominated by M. edulis in the British Isles (Gosling and Wilkins, 1981; Skibinski et al., 1983; Skibinski and Roderick, 1991). Moreover, mussels collected from an intertidal beach at Posjet Bay, USSR, contained only M. trossulus, while those from a floating dock a few meters away were all M. galloprovincialis (see, McDonald et al., 1991). Since the closely related mussels would appear to segregate in microhabitats within an area where their geographical distributions overlap, samples from a variety of habitats should be examined in northern Japan, including the Asamushi area, in the future.
As stated in previous studies (Gosling, 1984, 1992a; Koehn, 1991; McDonald et al., 1991), no single morphological character is reliable for identifying species within the mussel group. Fourteen of 19 shell characters of Asamushi mussels differed significantly from individuals of Maruishi which exhibited the typical characteristics of M. galloprovincialis (Fig. 2). This led to the erroneous presumption that the mussels of Asamushi were not M. galloprovincialis. The distance between the anterior end of posterior retractor muscle scar and pallial line is believed to be narrower in M. trossulus than in M. galloprovincialis, but the results of this study revealed that M. galloprovincialis had a great deal of intraspecific variation in this character (Fig. 3e). On the other hand, the canonical variates analysis based on 18 shell characters (McDonald et al., 1991) would be a good morphological method for field ecologists and biologists studying the ‘M. edulis complex.’ The results of this study from the canonical variates analysis indicated that 86.4% of Asamushi mussels were the same species as those from Maruishi (i. e., M. galloprovincialis), even though 13 of 18 characters differed statistically in their standardized sizes between individuals from the two sites. However, when this method indicates a mixed population, other methods such as electrophoretic (Gosling, 1984, 1992a, b for reviews), mitochondrial DNA (e.g., Skibinsky, 1985; Koehn, 1991; Rawson and Hilbish, 1995), and PCR product analyses are indispensable.
Although the recently proposed method of PCR product analysis by Inoue et al. (1995), has rarely been tested for wild mussels, its reliability was supported by our result which was consistent with that of canonical variates analysis method that had been made in conjunction with the careful examination in allozyme characters (McDonald et al., 1991). Inoue et al. (1995) have pointed out two advantages of their method: (1) PCR requires only a small amount of DNA as a template, so that the method may be used for larvae or young individuals. (2) The hybrids within the ‘M. edulis complex’ can be identified by detecting haplotypes of different species. The first point is very important for understanding a variety of phenomena in the field. Patterns of larval dispersion and recruitment are essential in determining distributions and population dynamics of benthic organisms (e.g., Hall et al., 1994), but taxonomic difficulties obstruct ecological studies of larval stages. Moreover, relationships (e.g., intraspecific competitions for space and food) between adults and younger individuals may affect the stability of populations (Seed, 1969; Kautsky, 1982; Seed and Suchanek, 1992). These interesting problems can not be examined without testing whether larger ‘adults’ and smaller ‘young’ individuals are members of the same species or not. Matsumasa and Nishihira (1994) reported a morphological difference between larger (shell length≥40 mm) and smaller (shell length<40 mm) shells of mussels collected from a rope substrate at Asamushi, and suggested that the growth pattern of the mussel shell changes with size in this gregarious animal. In this study, all Asamushi mussels ranging from 12.80 to 62.30 mm in shell length were identified as M. galloprovincialis by PCR product analysis. Therefore, the morphological difference in shell shape between two size groups would appear to be due to the change in growth pattern with size in a single species population, rather than a mixture of different species. Secondarily, no hybrids were detected for mussels of Asamushi by PCR product analysis in this study. This finding should be examined in more detail using other methods such as morphological, electrophoretic, and mitochondrial DNA analyses. An interesting future problem may be to determine whether the difference in shell morphology between Asamushi and Maruishi sites is due to genetic variation, environmental variation, or both.
We thank S. Takeda, S. Tamura, and M. Washio, Asamushi Marine Biological Station of Tohoku University for their help in our ecological and taxonomic works. We are also indebted to Dr. M. Nakaoka, University of Tokyo, for providing specimens of M. trossulus. Space and equipment were provided by National Research Institute of Fisheries and Environment of Inland Sea. Dr. Keiji Wada, Nara Women's University, Dr. Patricia R. Y. Backwell and Dr. Michael D. Jennions, Smithsonian Tropical Research Institute, and anonymous reviewers gave us helpful comments. This study was partly supported by a Grant-in-Aid for Scientific Research in Priority Area (Creative Basic Research Program, DIVER) from the Japan Ministry of Education, Science, Sports and Culture.