Marine Biology

, 156:39

Differences in morphology and habitat use among the native mussel Mytilus trossulus, the non-native M. galloprovincialis, and their hybrids in Puget Sound, Washington

Authors

    • Department of BiologyUniversity of Puget Sound
    • Department of BiologyUniversity of Puget Sound
  • Kathleen Holmes
    • Department of BiologyUniversity of Puget Sound
  • Rachel Chambers
    • Department of BiologyUniversity of Puget Sound
  • Kate Leon
    • Department of BiologyUniversity of Puget Sound
  • Peter Wimberger
    • Department of BiologyUniversity of Puget Sound
Original Paper

DOI: 10.1007/s00227-008-1063-3

Cite this article as:
Elliott, J., Holmes, K., Chambers, R. et al. Mar Biol (2008) 156: 39. doi:10.1007/s00227-008-1063-3

Abstract

Mytilus galloprovincialis (Mg), the Mediterranean blue mussel, is sympatric with the native M. trossulus (Mt) throughout much of the north Pacific, likely as the result of human introduction. We investigated the distribution of the two species and their hybrids (Mgt) in Puget Sound, Washington, to determine whether differences occur in habitat preference between the two species and hybrids. In addition, we investigated whether there were consistent size and shape differences between the native and introduced mussels and hybrids. Measurements of over 6,000 mussels from 30 sites, of which 1,460 were genotyped for a species-specific genetic marker, revealed that Mg and Mgt can be found throughout Puget Sound. Mg and Mgt were larger and exhibited a greater height:length ratio than Mt. Frequencies of Mg and Mgt were higher in subtidal habitats, such as docks, than on intertidal rocks, walls or pilings. Within intertidal habitats, Mg and Mgt were more frequent than Mt in the lower reaches of the intertidal. At slightly more than half the sites the frequency of the three genotypes accorded with random mating expectations suggesting no consistent barriers to gene flow between species. The standardized random sampling methods and simple morphometric identification techniques described here can be used to test whether the frequency of invasive mussels changes over time and space in Puget Sound.

Introduction

The introduction of non-indigenous species (NIS) in marine environments has been recognized as an important scientific and environmental issue for the past two decades (Carlton 1989, Carlton and Geller 1993; Ruiz et al. 2000, Grosholz 2002). In marine systems, NIS are usually introduced through ship ballast-water exchange, hull fouling, and/or aquaculture, and the rate of these introductions into marine communities has increased exponentially over the past century (Ruiz et al. 2000). In some instances these exotic species have become invasive and had dramatic impacts on native species through direct or indirect competition, predation, parasitism, disease, and hybridization (Grant and Cherry 1985; Rhymer and Simberloff 1996; Carlton et al. 1999). As more NIS are detected in aquatic ecosystems it becomes increasingly important to understand the dynamics of introductions and what characteristics successful invaders have in common so that we can predict the impact these NIS have on native communities and ecosystems (Carlton et al. 1999, Miller et al. 2007). Predicting these characteristics and impacts requires that we understand the ecology of the NIS and the native species in relation to the local environmental conditions because the effects of an invader can be condition-specific (e.g., Krassoi et al. 2008). Ecological studies must employ sampling techniques that allow quantitative comparisons of the distribution and abundance of the NIS and native species in space and time. These studies should provide a comparison of the interacting species’ responses to the physical and biotic environment, and their relative growth rates, reproductive rates and ecological interactions. Habitat “preference” can result from the interplay of these factors, and understanding whether native and NIS partition habitats or competitively exclude each other under different environmental conditions is a first step to predicting potential impacts of NIS.

Mussels of the Mytilus species complex, M. edulis (Me), M. trossulus (Mt), and M. galloprovincialis (Mg) are geographically widespread in temperate marine waters and are important members of nearshore benthic communities (Seed 1992; Wonham 2004). Me is native to the cold-temperate areas of the Atlantic, Mt is native to the north Pacific and the north Atlantic, the northern form of Mg is native to the Mediterranean Sea and western Europe, and the southern form of Mg is found on the coasts of Argentina, Chile, New Zealand, Australia and some islands such as the Kerguelens and Falklands (McDonald et al. 1991; Daguin and Borsa 1999; Hilbish et al. 2000). The three species maintain distinct genetic identities over large parts of their ranges, but where their ranges overlap they often hybridize (Skibinski 1983; Gardner and Skibinski 1988; Sarver and Foltz 1993; Gardner 1994, 1997; Riginos and Cunningham 2005; Riginos et al. 2006). Of the three species, only the northern form of Mg has significantly expanded its geographic range through human agency (Wonham 2004). The northern form of Mg has been introduced accidentally via shipping and/or intentionally via aquaculture to South Africa (Grant and Cherry 1985), the Pacific coast of North America, where it occurs locally from southern California to British Columbia (McDonald and Koehn 1988; Sarver and Foltz 1993; Geller et al. 1994; Heath et al. 1995; Suchanek et al. 1997; Rawson et al. 1999; Anderson et al. 2002; Wonham 2004; Braby and Somero 2006a), and the Pacific coast of Asia in Japan, China, Hong Kong, Korea and Russia (McDonald et al. 1991; Inoue et al. 1995; Kartavtsev et al. 2005). Mg has appeared on the Pacific coast of North America in the past century, first on the coast of southern California, likely as the result of shipping (Geller 1999). South of Monterey, Mg is the predominant mussel. Further north, Mg occurs locally, mostly in protected waters such as San Francisco Bay, Tomales Bay, Bodega Bay, Coos Bay, Puget Sound and inland British Columbia waters (reviewed in Wonham 2004; Braby and Somero 2006a). Mg has appeared in Puget Sound, Washington during the past 30 years (Brooks 1991; Chew 1996; Suchanek et al. 1997; Anderson et al. 2002; Wonham 2004) and hybridizes with the native Mt wherever the two species are sympatric (Suchanek et al. 1997; Rawson et al. 1999; Anderson et al. 2002; Wonham 2004). Mg, Mg × Mt (Mgt) hybrids and backcross individuals are now patchily distributed throughout most of Puget Sound, and have been reported to be most common near aquaculture farms and areas that receive international ship traffic (Anderson et al. 2002; Wonham 2004).

The different Mytilus species and their hybrids are morphologically similar and to definitively identify them as Mg, Mt, or Mgt has typically required genetic testing. Shell shape is phenotypically plastic and affected by abiotic and biotic environmental factors (Seed 1968; Leonard et al. 1999; Akester and Martel 2000; Reimer and Harms-Ringdahl 2001; Steffani and Branch 2003; Freeman and Byers 2006), and may change during development (Seed 1973). A number of studies have used multivariate analysis of a suite of internal and external shell characters to distinguish the three species (Beaumont et al. 1989; McDonald et al. 1991; Innes and Bates 1999; Gardner 2004); however, there is overlap between the species even when using multiple characters because of plasticity, size variation and the possible presence of hybrids in the samples (McDonald et al. 1991; Seed 1992; Gardner 1994, 1997). Despite these difficulties, in some areas it is possible to fairly accurately discriminate species with simple measures (Skibinski 1983; Penney and Hart 1999). On the Pacific coast, it appears that Mg are typically longer than Mt (Anderson et al. 2002, Wonham 2004; Braby and Somero 2006a). Further study of morphological differences between Mg and Mt on the Pacific coast would be beneficial if it led to quick and cost effective species identification, which would then allow for more extensive ecological studies and rapid assessments of the spread of Mg in space and time.

Many studies have revealed habitat differences between Mytilus congeners in the Atlantic where species partially segregate in relation to depth, salinity, temperature and wave exposure (Gardner and Skibinski 1988; Hilbish et al. 2002; Kenchington et al. 2002; Riginos and Cunningham 2005; Schneider et al. 2005; Coghlan and Gosling 2007). In the Pacific, McDonald et al. (1991) and Wonham (2004) reported that Mt occurred higher in the intertidal zone than Mg. Schneider and Helmuth (2007) have shown that in San Francisco Bay Mg were more abundant in sun-exposed intertidal habitats and that Mt were more abundant subtidally on docks. Braby and Somero (2006a) found that the proportion of Mg living subtidally on docks was positively correlated with average salinity and negatively correlated with temperature. Mt showed the opposite pattern. In physiological studies, Braby and Somero (2006b) found that Mg was more resistant to high temperatures than Mt and functioned better at high salinities. Mt was more tolerant of low temperatures and salinities than Mg. It is unknown whether these correlations between habitat/environmental conditions and genotype frequencies occur in other geographic areas of the Pacific Coast, such as Puget Sound. Ecological studies of invasive species and their effects on closely related native congeners are relatively rare, especially when hybridization also occurs.

Studies of Mytilus genotypes on the Pacific coast have typically documented deficits of hybrids relative to expectations from random mating, suggesting that intrinsic or extrinsic barriers prevent random mating (Heath et al. 1995; Anderson et al. 2002; Braby and Somero 2006a). However, Springer and Heath (2007) report that selection against hybrids is environment-dependent, and varies among locations. Braby and Somero (2006a) showed that the distribution of hybrids in the San Francisco Bay region did not correlate with environmental patterns in temperature and salinity, but Shields et al. (2008) have shown that hybrid mussels on Vancouver Island have a growth and survival advantage over parental species in certain environments. Further research is needed to identify and explain the environmental-specific distribution of genotypes in mussel hybrid zones.

In this study we examined the distribution of Mt, Mg and their hybrids throughout Puget Sound to: (1) determine whether there were simple morphometric measures (i.e., length, height and width) that could be used to distinguish genotypes (Mt, Mg and Mgt) in Puget Sound to aid the monitoring and rapid assessment of mussel communities, (2) compare the relative frequencies of Mt, Mg and Mgt among habitats (docks vs. pilings vs. intertidal) to facilitate predicting the potential impact of introduced Mg in Puget Sound marine communities, and (3) determine the frequency of hybrids in random samples to test for the presence of barriers to interbreeding in Puget Sound.

Materials and methods

Sampling locations

Mussels were collected from 30 locations in Puget Sound from 2000 to 2004 (Fig. 1, Tables 1, 2). Detailed distribution and abundance data were collected from four locations where high frequencies of Mg occurred. Two of these locations were in south Puget Sound (Totten Inlet and Oakland Bay) and two were in central Puget Sound (Quartermaster Harbor and Dyes Inlet). At every location random samples were taken from both docks and pilings. On pilings exposed to full sunlight, mussels were typically observed approximately 1–3 m above MLLW on the north-facing side of pilings. Few to no mussels were observed on the south side of these pilings. All of the individuals within a 0.25 m2 quadrat were scraped from the substrate at the low, mid, and high sections of the mussel zone on pilings. On docks, 0.25 m2 quadrat samples were taken from randomly chosen locations. Samples were collected from the middle of the side of the dock, with the top of the quadrat approximately 0.5 m below the surface of the water. At Dyes Inlet we also collected mussels from the top of the sides of the dock (just below the waterline) and from the bottom of the dock to examine differences among microhabitats. We collected data on the distribution and abundance of mussels on rocks in the intertidal zone by collecting all of the mussels from randomly placed 0.25 m2 quadrats in the upper, mid and lower intertidal zones.
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-008-1063-3/MediaObjects/227_2008_1063_Fig1_HTML.gif
Fig. 1

Map of collection sites in Puget Sound with inset showing location on the west coast of North America. Numbers for locations given in Tables 1 and 2

Table 1

Locations where samples of mussels were collected in different habitat types

Site

Locality

Habitat type

Random sample (n)

Median length (mm)

Range (mm)

Total genotyped

Size genotyped

Mt

Mg

Mgt

HW deviation

Mt

Exotic

1

Oak Harbor

Piling

91

50.5

9.9–72.1

68

11

1

0

0

 

1

0

2

Penn Cove

Rocky intertidal

100

39.2

12–70.8

54

5

1

0

0

 

1

0

3

Coupeville

Rocky intertidal

90

30.3

9.5–69.0

59

10

1

0

0

 

1

0

4

Utsalady

Rocky intertidal

78

31.5

11–61.6

49

0

1

0

0

 

1

0

5

Madrona Beach

Rocky intertidal

92

25.5

10.3–67.1

56

8

1

0

0

 

1

0

6

Freeland

Piling

67

36.3

9.4–67

56

9

1

0

0

 

1

0

7

Dye’s Inlet

Dock

841

42.4

5.3–125.9

25

10

0.82

0.09

0.09

*

0.94

0.06

Piling

442

32.7

5.0–79.7

16

16

NC

   

0.97

0.03

10

Dockton

Dock

135

56

19.7–130.8

68

6

0.53

0.10

0.37

NS

0.49

0.51

Piling

1,263

26.2

4.8–98.5

6

6

NC

   

0.99

0.01

11

Thea Foss

Dock

20

36.8

8.8–71.5

41

21

0.96

0

0.04

NS

0.95

0.05

12

Commencement Bay

Piling

254

34.5

6.4–57.0

13

13

NC

   

0.99

0.01

16

Steilacoom

Dock

102

48.2

30–133.1

35

0

0.89

0.02

0.09

NS

0.9

0.1

19

Hunter Point

Piling

124

22.8

4.7–65.2

61

10

0.982

0

0.018

NC

0.99

0.01

20

Carlyon

Piling

71

34.6

9.6–66.8

55

11

0.978

0.004

0.018

NC

1

0

23

Totten middle

Piling

638

27.1

5.9–99.2

144

8

0.982

0

0.018

NS

0.99

0.01

24

Totten dock

Dock

114

33.6

5.6–95

42

0

0.785

0.055

0.16

NS

0.78

0.22

25

Totten wall

Intertidal

508

34.4

8.8–102.2

41

0

0.86

0.04

0.1

*

0.88

0.12

26

Totten South 1

piling

94

32.5

6.2–59.4

51

0

0.989

0

0.011

NC

0.99

0.01

27

Totten South 2

Piling

88

31.5

8.2–78.6

51

3

0.909

0.026

0.065

NS

0.9

0.1

28

Arcadia Point

Piling

79

38.9

6–70.6

45

0

0.948

0.027

0.025

*

0.99

0.01

29

Shelton Marina

Dock

118

50.3

10.1–117.3

66

17

0.717

0.142

0.140

**

0.78

0.22

Piling

291

43.3

7.4–112.2

90

0

0.938

0.028

0.034

**

0.96

0.04

30

Oakland Bay

Piling

263

39

5.3–100.5

88

0

0.992

0

0.008

NC

0.99

0.01

Total

5,963

34.55

4.7–125.9

1,280

164

      

The sites are numbered from north (1) to south (30). The Random Sample column gives the number of mussels collected in random samples and measured for morphometrics. The Size Genotyped column gives the number of large mussels genotyped from size-selected samples. The first set of genotype frequencies were determined by genotyping mussels using the Me 15/16 marker and then applying the size-corrected genetic estimate described in the text. The second set of genotype frequencies were generated using the morphometric estimation method (see text)

NC not calculated, NS not significant

P < 0.05

** P < 0.005

Table 2

Locations where size-selected samples of mussels were collected in different habitat types

Site

Locality

Habitat type

Sample size (n)

Median length (mm)

Range (mm)

Number genotyped

Mt

Mg

Mgt

1

Oak Harbor

Piling

13

68.4

60–82.0

12

1.00

0.00

0.00

2

Penn Cove

Rocky intertidal

7

76.4

71.8–77.4

7

1.00

0.00

0.00

3

Coupeville

Rocky intertidal

13

67.4

61.4–73.4

10

1.00

0.00

0.00

4

Utsalady

Rocky intertidal

5

54.9

52.7–58.2

0

Nc

  

5

Madrona Beach

Rocky intertidal

8

60.6

57.8–63.7

8

1.00

0.00

0.00

6

Freeland

Piling

14

73.6

64–89.2

9

0.22

0.33

0.44

7

Dye’s Inlet

Dock

10

64.7

50.2–99.1

9

0.20

0.50

0.30

  

Piling

16

65.8

54.3–82.3

16

0.13

0.06

0.81

8

Elliott Bay

Dock

11

39.7

34.0–60.4

11

0.64

0.00

0.36

9

West Seattle

Dock

10

48.4

34.28–62.9

10

0.70

0.00

0.30

10

Dockton

Dock

6

50.3

46.7–59.0

6

1.00

0.00

0.00

11

Thea Foss

Dock

21

81.6

50.7–109.5

21

0.43

0.05

0.52

12

Commencement Bay

Piling

7

64.9

57.3–74.2

7

0.14

0.00

0.86

13

Point Defiance

Dock

10

80.4

74.5–97.3

9

0.00

0.44

0.56

14

Fox Island

Piling

12

45.0

40.0–75.0

12

0.83

0.08

0.08

15

Allyn

Dock

6

110.3

88.8–136.2

4

0.00

0.25

0.75

17

Ketron Island

Piling

26

76.2

7–135

14

0.50

0.00

0.50

18

Johnson Point

Piling

16

49.8

43.8–59.9

13

1.00

0.00

0.00

19

Hunter Point

Piling

11

62.0

56–80.1

10

0.90

0.00

0.10

20

Carlyon

Piling

11

24.3

20.2–39.6

11

0.82

0.09

0.09

21

Hope Island

Rocky intertidal

16

54.1

45–74.0

15

1.00

0.00

0.00

22

Totten North

Rocky intertidal

20

61.5

51.5–77.2

20

0.45

0.15

0.40

23

Totten Middle

Piling

8

48.6

37.6–79.4

8

0.88

0.00

0.13

Subtidal rope

89

30.0

15.8–86–8

77

0.00

1.00

0.00

26

Totten South 1

Piling

5

81.0

68.5–82.2

5

0.00

0.00

1.00

27

Totten South 2

Piling

3

66.3

64.4–69.9

3

0.00

0.33

0.67

29

Shelton Marina

Dock

17

100.2

50.7–140.0

17

0.06

0.59

0.35

Genotype frequencies were determined using the Me 15/16 marker

In addition to the random samples collected at each location, we also searched for large mussels, which ranged in length from 50 to 140 mm (size-selected samples). Where possible, we collected 5 or more of these large individuals to provide an indication of whether Mg or hybrids were present at the site because larger mussels usually carried Mg alleles (see below). Mussels were also obtained from Taylor Shellfish Farms, an aquaculture company that cultures Mg in Puget Sound, to examine the genotype and shell characteristics of these cultured mussels. All mussels from each sample were placed in labeled plastic bags and transported to the University of Puget Sound on ice in coolers. They were then stored at −20°C until processed for morphometric and genetic analyses.

Morphometric and genetic analyses

Shell length (anterior–posterior axis), height (dorsal–ventral) and width (left–right) of the mussels were measured to the nearest 0.1 mm using Vernier calipers as in McDonald et al. (1991). We used easily-made measures because we were seeking a method that would allow quick identification of live mussels with Mg alleles in Puget Sound. Shape measures were ratios of the three measurements made. All mussels were individually marked and archived. To account for differences in genotype frequencies with size as has been observed by Anderson et al. (2002) and Braby and Somero (2006a), each random sample of mussels was sorted into 20 mm increment shell length classes: <20, 20–40, 40–60, 60–80 mm, up to the largest 20 mm size class available. We genotyped ten randomly selected mussels from each size class from most samples when sample size permitted. Typically, larger size classes included relatively few individuals, thus these samples were supplemented with those from size-selected samples from the site for genotyping.

DNA extraction was performed on mussels from each size class according to Geller et al. (1994). A 0.5 g piece of mantle tissue was incubated at 55°C for >1 h in 700 μL 2× CTAB extraction buffer (1 M Tris pH 8, 5 M NaCl, 0.5 M EDTA, 2 g CTAB, 14 M β-mercaptoethanol) and 5 μL proteinase K (20 mg ml−1). DNA was extracted once with phenol-chloroform-isoamyl alcohol (25:24:1) and once with chloroform–isoamyl alcohol (24:1). DNA was precipitated in 100% isopropanol, frozen at −20°C for 12 h, and microcentrifuged at 14 K for 7 min to pellet the DNA. Pellets were rinsed twice with 70% ethanol/30% TE, air dried, and re-suspended in 200–400 μL TE (1 M Tris, 0.5 M EDTA).

We genotyped the mussels for the codominant nuclear marker Me15/16, which distinguishes the three Mytilus species and allows the detection of hybrids (Inoue et al. 1995). This marker is a fragment of an adhesive protein that makes up part of the byssal thread. We used a single genetic marker because previous studies using multiple nuclear markers of Mt and Mg in the Pacific and in Puget Sound showed very high linkage between markers and little evidence of backcrossing (Rawson and Hilbish 1995; Rawson et al. 1999; Anderson et al. 2002; Braby and Somero 2006a, although see Suchanek et al. 1997). The oligonucleotide primers Me-15 [CCA GTA TAC AAA CCT GTG AAG A] and Me-16 [TGT TGT CTT AAT AGG TTT GTA AGA] result in amplification products of 168-bp in Mt, 126-bp in Mg, and 180 bp in Me, although we did not encounter Me (Inoue et al. 1995). The size of the amplified gene fragment in Mg is shorter because it lacks 18 amino acids in the nonrepetitive domain of the adhesion protein gene sequence (Inoue and Odo 1994; Inoue et al. 1995). F1 hybrids between Mt and Mg have one copy of each the 168 and 126-bp DNA fragments (Suchanek et al. 1997). For amplification we used 1–2 μL template DNA, 2.5 μL Qiagen 10X PCR buffer, 25 mM MgCl2, 10 mM dNTPs, 10 pmol primers, 0.85 µm Taq, and ddi water. The reaction mixture was heated to 95°C for two min and then subjected to 30 cycles of 15 s at 95°C, 40 s at 54°C, and 80 s at 72°C on a Perkin Elmer 2400 Gene Amp PCR system. The PCR products were visualized on 2.5% agarose gels buffered in 1× TBE run at 100 V for 60 min. Gels were stained in 5 μg ml−1 ethidium bromide for 20 min and visualized over UV light.

Estimation of relative frequencies of Mt, Mg, and hybrids

We used two methods to estimate the relative frequencies of mussel genotypes in each sample. In the first method (here termed size-corrected genetic estimate), for each location we estimated the genotype frequencies (Mt, Mg, Mgt) for each of the size classes, then multiplied each size class’ genotype frequencies by the relative frequency of that size class in the random sample. In the second method (here termed “morphometric estimate”), we used the correspondence between genotype and a set of morphometric criteria described below (see Results) to estimate the relative frequencies of native (Mt) and exotics (Mg and Mgt) in three different size classes (0–40, 40–80, and >80 mm) from the randomly sampled mussels.

Statistical analysis

Statistical analyses were conducted using the programs Excel and SPSS. Comparisons of mean morphometric characters (e.g., length, width and height) among mussel genotypes were conducted using One-way ANOVA. Kolmogorov–Smirnov tests were used to examine differences between size frequency distributions. Analysis of covariance (ANCOVA) tests were used to determine if significant differences existed between slopes and between intercepts of regression lines for shell length, height and width. Since slopes were homogeneous in ANCOVA tests, i.e., the interaction term was not significant, the interaction term was removed to obtain the difference among regression line intercepts. Random mating among mussel genotypes (Mt, Mg and Mgt) among the different habitats were compared using Fisher Exact tests as implemented in Genepop (Raymond and Rousset 1995).

Results

Distribution in Puget Sound

We measured a total of 6,354 mussels from 30 locations in Puget Sound (Fig. 1). Most of the mussels were from random samples (n = 5,963) and the remainder were from size-selected samples (Tables 1, 2). A subset (n = 1,460) of these measured mussels were genotyped using the diagnostic Me15/16 marker.

Frequencies of Mt, Mg and their hybrids based on the Me15/16 marker varied widely among the geographic locations and habitats sampled (frequencies reported are the size-corrected genetic estimates unless otherwise stated). Most genotyped mussels in the random samples were Mt (overall 92.3%). Hybrids (overall 5.3%) were more than twice as frequent as Mg (overall 2.3%). At the northern Puget Sound locations in Whidbey Basin (sites 1–6) all of the randomly sampled, genotyped mussels were Mt (N = 342). At the most southern site in Whidbey Basin (Freeland, site 6), Mt made up 97.5% of the mussels when using the size-corrected genetic estimate; however, over half of the size-selected mussels (larger than any in the random sample) had Mg alleles. Most areas sampled in central and south Puget Sound had mussels of all three genotypes. The highest frequencies of Mg and Mgt were found on docks at Dockton (Quartermaster Harbor, Vashon Island—0.47) and in south Puget Sound in Totten Inlet and Oakland Bay (Shelton Marina—0.28). The frequency of mussels with Mg alleles from other areas ranged from 0 to 0.14 (Table 1). All of the cultured mussels acquired from Taylor Shellfish in Totten Inlet were Mg (Table 2).

Size and shape

Mt mussels were significantly smaller than either Mg or Mgt (Fig. 2, One-way ANOVA, P < 0.0001). All Mt mussels were <80 mm in length, thus, all mussels > 80 mm in length were either Mg or Mgt. Genotyped mussels 0–40 mm were predominantly Mt (99.8% of mussels on pilings and 93% on docks), with only 7/437 genotyped mussels in this size range being Mg (2) or Mgt (5). Thus, small Mg or Mgt mussels were very rare at the times we sampled them (primarily summer and fall).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-008-1063-3/MediaObjects/227_2008_1063_Fig2_HTML.gif
Fig. 2

Size distributions of randomly sampled mussels that were genotyped using the Me-15/16 marker: aMytilus trossulus = Mt, bM. galloprovincialis = Mg, c hybrids = Mgt. The mean length of Mt was significantly smaller than Mg and Mgt (P < 0.0001, ANOVA) but there was no significant difference between Mg and Mgt

The height:length and width:length ratios of different genotypes were indistinguishable at lengths between 0 and 40 mm, but in mussels from 40 to 80 mm long the height of Mt mussels was significantly less (3–5 mm) than Mg and Mgt (Fig. 3; P < 0.0001, ANCOVA on log-transformed data). Mg mussels obtained from Taylor Shellfish were used to determine the height/length relationship for small Mg individuals. The height:length and width:length ratios did not differ between Mg and Mgt at any size. There was no significant difference in the width of the different mussel genotypes (P > 0.05, ANCOVA on log transformed data, data not shown).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-008-1063-3/MediaObjects/227_2008_1063_Fig3_HTML.gif
Fig. 3

Height plotted against length for genotyped mussels 40–80 mm in length with line generated by the morphometric estimation technique for discriminating between native (Mt) and exotic (Mg or Mgt) mussels. See text for explanation

We used shell length and height of the genotyped mussels to construct an easily used predictive model for identifying mussels as either native (Mt) or exotic (Mg or Mgt). Because all genotyped mussels over 80 mm long (n = 173) were either Mg or Mgt, we inferred that all mussels over 80 mm had Mg alleles. Based on genotype frequencies obtained from random field samples, mussels < 40 mm long were inferred to be 98.6% Mt and 1.4% exotic. Mussels 40–80 mm long were classified based on their height:length relationship. We used an iterative process to determine a line that best discriminated between native and exotic genotypes in this size class by maximizing the percentage of genotyped mussels correctly classified (Fig. 3). The relationship between shell height and length for each genotype was best described by a polynomial equation. Mussels with
$$ {\text{Height}} > 0.6{\text{\;Length}} + ( - 0.0025{\text{\;Length}}^{2} ) + 4 $$
(1)
were considered to be exotics (Mg or Mgt). This method (morphometric estimate) successfully assigned 759/797 (95%) mussels genotyped between 40 and 80 mm in length. The method was most successful in identifying Mt mussels (97%), and slightly less successful for Mg (89%) and Mgt (86%).
Using this model to classify each mussel as either Mt or exotic (Mg or Mgt), we estimated the frequencies of both mussel types from random samples taken in different sites (Table 1). We compared these morphometric estimate frequencies to the frequencies generated by the size-corrected genetic estimates. There was strong agreement between the observed size-corrected genotype frequencies of Mt and the morphometric estimated frequencies (Spearman rank correlation r = 0.90, P < 0.001). A lower agreement between the two methods typically occurred when the sample size for genotyping was relatively small (25), or when there was a higher relative frequency of mussels with exotic alleles in the sample. The model was unintentionally weighted to correctly identifying Mt due to their higher frequency among the genotyped sample. In general, for the different habitat types, the morphometric estimated Mt frequencies were within 1–2% of the actual genotype frequencies for intertidal habitats and within 5% for the dock samples (Tables 1, 3).
Table 3

Mussel genotype frequencies in different habitats

Habitat type

Total Sample size (n)

Median length (mm)

Range length (mm)

No. genotyped

Mt freq.

Mg freq.

 

Mgt freq.

Dock

1,417

43.8

5.3–140.0

305

0.76

0.07

 

0.17

0.80

 

0.20

 

Piling

3,934

32.0

30.5–135.0

737

0.96

0.01

 

0.03

0.98

 

0.02

 

Rocky Intertidal

368

30.6

9.5–70.8

281

0.97

0.01

 

0.02

0.99

 

0.01

 

Concrete wall

508

34.4

8.8–102.2

41

0.85

0.05

 

0.10

0.86

 

0.14

 

The genotype frequencies in the first row were determined by genotyping mussels for the Me15/16 marker and using the size-corrected estimation method. Genotype frequencies for Mt and exotics (Mg and Mgt) in the second row for each location were determined using the morphometric estimation (see text for detailed description). There was no significant difference between the two estimation methods for any of the habitat types (Contingency Table Analysis, P > 0.05). Mussels were significantly larger and there were significantly greater frequencies of Mg and Mgt on docks than in intertidal habitats

Habitat preference

Genotype frequencies varied significantly among habitat types, with the frequency of Mg and Mgt mussels being highest on docks and in the low intertidal zone (Table 3). The difference in genotype frequencies among habitats was reflected in significant differences in mean mussel size (One-way ANOVA, P < 0.0001). Mussels reached largest sizes on docks with individuals up to 140 mm long (Table 3, Fig. 4), followed by the concrete wall, rocky intertidal zone and pilings. Pairwise comparisons of length from each of the habitats were significant for all comparisons except the wall/intertidal comparison and the intertidal/piling comparison (One-way ANOVA, Bonferroni Corrected P < 0.05). Dock mussels were, on average, 8.4, 10.5 and 12.4 mm larger than their counterparts on wall, intertidal and piling habitats, respectively. One result of this larger size on docks was that the most frequent size class on docks was mussels from 40 to 60 mm, whereas in the other habitats the most frequent size class was 20–40 mm. In addition, a greater proportion of 40–60 mm mussels on docks harbored Mg alleles than on the other habitats. The largest size classes were typically dominated by Mgt mussels.
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-008-1063-3/MediaObjects/227_2008_1063_Fig4_HTML.gif
Fig. 4

Length frequency distributions of randomly sampled mussels from different habitats. Genotype frequencies for each size class are given above each bar (n > 10 mussels)

On pilings, mussels in the lowest part of the mussel zone reached larger sizes compared to the mid and high intertidal (Fig. 5, ANOVA, P < 0.0001). The size distribution of mussels on docks also varied with their vertical location. There was typically a distinct band of small mussels near the surface of the water (swash zone) on the sides of docks. The mussels in this top section of the dock were significantly smaller than those on the middle and bottom sections of the dock (Fig. 6, ANOVA, P < 0.0001). Applying the morphometric model to the mussel size distributions for these microhabitats we estimated that the frequency of exotic mussels was low on the top section of docks (0.06), and increased on the middle (0.25) and bottom sections (0.24).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-008-1063-3/MediaObjects/227_2008_1063_Fig5_HTML.gif
Fig. 5

Length frequency distributions of randomly sampled mussels from different tidal heights on pilings. Genotype frequencies for each size class are given above each bar (n > 10 mussels)

https://static-content.springer.com/image/art%3A10.1007%2Fs00227-008-1063-3/MediaObjects/227_2008_1063_Fig6_HTML.gif
Fig. 6

Length frequency distributions of randomly sampled mussels collected from top, middle, and low bottom sections of a floating dock at Dye’s Inlet. Genotype frequencies (Mt and Exotic) for each size class are given above each bar (n > 10 mussels)

Random mating

Random mating among genotypes was detected in slightly over half of the locations where the data allowed analysis (6/11, Table 1). In all locations where non-random mating was detected, hybrid genotypes were less frequent than predicted and the degree of heterozygote deficiency was positively correlated with the frequency of Mg alleles in the population.

Discussion

Mussel morphometrics and population sampling

This study has shown that a simple analysis of mussel length and height measures can be used to successfully identify mussel genotypes with Mg alleles for individuals > 40 mm long in Puget Sound. Mg and Mgt were relatively taller than Mt. The morphometric relationship (Eq. 1) provides a simple quantitative method for differentiating native from exotic Mytilus genotypes in Puget Sound. It worked best at identifying mussels with Mt genotypes (98%), but was less successful, although still very useful, for those with Mg (89%) and Mgt genotypes (86%). This difference in assignment effectiveness between genotypes occurred because through maximizing the total number of correctly classified mussels, the preponderance of Mt mussels in the analysis weighted the analysis towards classifying a greater proportion of those correctly. Few studies have reported the ability of simple external measures to discriminate genotypes so effectively within a hybrid zone. The relationship between height:length ratio and species appears to apply only to mussels within Puget Sound. We examined a sample of 102 mussels collected from the southwest shore of San Juan Island and found that the slope of the height length regression was similar, but height was significantly lower in the San Juan mussels (unpublished data).

Our result is similar to other studies that use a suite of external and internal shell characters, in that the morphometric characteristics of species overlap, which is often considered to be caused by hybrids in the population (Skibinski 1983; McDonald et al. 1991; Gardner 1996; Penney and Hart 1999). Mussels with Mgt genotypes were most difficult to identify using height and length. Their shape was intermediate between Mt and Mg, but was most similar to Mg. Mg mussels were most distinct, having the greatest height:length ratio. Some studies have used shell height as one of the measures to differentiate among mussel species (e.g., Beaumont et al. 1989; McDonald et al. 1991; Matson et al. 2003). Mg mussels typically have a taller shell height than other species, which may be an adaptation to increased wave exposure in their native habitat (Innes and Bates, 1999).

Our morphometric method for identifying native or exotic individuals provides a relatively accurate estimate of genotype frequencies in the field. This method should prove very useful in conducting rapid assessments of introduced species in Puget Sound and assessing whether frequencies of exotic mussels are increasing in different areas. If high frequencies of exotics are indicated by morphometric analysis, then further genetic analyses could be used to accurately determine the frequencies of Mg and Mgt genotypes at a given site. Because we used a single genetic marker we were unable to determine if some of the largest mussels we identified as Mt were actually introgressed individuals.

Size differences among genotypes

Species and hybrid-specific genotype frequencies of mussels in Puget Sound varied with length and shape. Mg and Mgt genotypes had longer shells, as was reported by Anderson et al. (2002) for a small sample of mussels collected at Silverdale, and suggested by Wonham (2004) as the reason her Mg and Mgt genotype frequencies were higher than those reported by Anderson et al. (2002). Wonham (2004) collected the largest mussels at her sampling sites, biasing her results to higher estimates of Mg and Mgt frequencies. The size difference we report for Puget Sound also occurs further south along the Pacific; Braby and Somero (2006a) reported that mussels over 60 mm in their California and Oregon study sites were nearly all Mg and Mgt.

In Puget Sound there is a distinct shift in genotype frequencies from predominantly Mt in the smaller size classes to Mg and Mgt in the larger size classes (Figs. 2, 3, 4, 5, 6). Anderson et al. (2002) also found that Mg and Mgt were more abundant in the larger size classes at a sampling site in Silverdale, WA. The pattern is similar to that reported for some European hybrid populations of Mg and Me in which the frequency of Mg-specific alleles increases in larger (probably older) size classes (Gardner and Skibinski 1988, Hilbish et al. 2002). However, in California, Braby and Somero (2006a) found a more uniform distribution of Mg and Mgt across size classes.

The near total absence of Mg and Mgt in smaller size classes in our samples may be the result of either faster growth and larger maximum size in Mg and Mgt, or the result of a strong recruitment of Mg and Mgt in previous years with little recent recruitment, or the result of strong natural selection on the different genotypes at smaller sizes. These hypotheses are not mutually exclusive. Rensel et al. (2005) showed that Mg grow faster and have higher survival rates than Mt in Puget Sound. Furthermore, the consistency of the pattern across a diversity of sites (intertidal and subtidal, average high salinity and low salinity, etc.) suggests that selection is a less likely explanation than growth rate and/or recruitment differences for the nearly complete absence of small Mg and Mgt in summer and fall-collected Puget Sound mussels. Gardner and Skibinski (1988) suggest that differences in growth rates between Me and Mg do not explain the size differences they observed in a Me/Mg hybrid zone. However, Gardner et al. (1993) showed that Mg and Mg x Me hybrids are larger at a given age than Me, and that they typically live longer than Me. These are the kinds of differences that may result in the observed pattern of a higher frequency of Mt alleles in the smaller size classes and a higher frequency of Mg alleles in larger size classes. Finally, observations from cultured Mg show that they can attain ~ 60 mm length in a single year (G. King, personal communication), consistent with our findings that Mg are infrequent at lengths below 50–60 mm. Thus, within 12 months, newly recruiting Mg and Mgt may achieve a length of 60 mm; their rapid growth would mean that they are rare among smaller mussels.

The largest mussels observed in our samples were hybrids, and they were more similar to Mg than Mt. Although there was no significant difference in the average sizes of Mg and Mgt from the random samples (Fig. 2), the most abundant mussels in the largest size classes and the largest mussels from the size samples were hybrids. Gardner et al. (1993) found that hybrids between Mg and Me have intermediate growth between the parental species at one site (Whitsand), although they were more similar to Mg, whereas at a second site (Croyde) hybrids grew like Mg through 6 years. Comesaña et al. (1999) showed that hybrids between Mt and Me were intermediate in size, and Mt backcrossed hybrids were generally smaller than the parental species. Wilhelm and Hilbish (1998) found that Mg × Me hybrids had lower survival than Mg but greater than Me. Our results suggest that hybrids may grow faster or survive longer than either of the parental species in Puget Sound, especially in those environments where they are most frequent (e.g., Dockton, Oakland Bay). Shields et al. (2008) found that introgressed mussels (includes F1 and backcrosses) grew better and had higher fitness than the native Mt, but the results varied with environmental conditions at two different study sites. Environment-dependent selection on different mussel genotypes is likely important in maintaining hybrid zones in different geographic locations.

Random mating and genotype frequencies

An unusual finding was that mussels from over half the sites appeared to be mating randomly between species, even without the application of a multiple comparison correction. This pattern is surprising given the reported difference in spawning time between the species (Mg from ca. November to March, Mt from ca April to July; Curiel-Ramirez and Caceres-Martinez 2004), which should lead to a deficiency of hybrids in the population. Most studies of Mytilus hybrid zones document hybrid or heterozygote deficiencies [for Mt/Mg zones—Anderson et al. (2002), Braby and Somero (2006a), Springer and Heath (2007)]. However, Springer and Heath (2007) report that patterns in heterozygote deficiencies may be environment-specific. It was surprising that even in some populations with high frequencies of Mg alleles (e.g., Dockton) the populations appeared to be randomly mating. The difference between our study and others on the Pacific coast could be the result of our size-corrected genetic estimate approach to calculating genotype frequencies, greater overlap in spawning time in Puget Sound than elsewhere or differences in the selective regimes among study areas (Springer and Heath 2007; Shields et al. 2008). These patterns requires further investigation.

Distribution among habitats

Another major finding of this study was that mussel genotype frequencies in Puget Sound varied depending on the habitat type (dock, piling, rocky intertidal) and even microhabitat (height on docks and pilings). Mg and Mgt mussel genotypes were most common on the sides of floating docks and least common in the rocky intertidal. McDonald et al. (1991) noted that in some localities different species occur most commonly in different habitats, such as Mt in the intertidal and Mg on floating docks. In contrast, Schneider and Helmuth (2007) reported that in San Francisco Bay Mg were more abundant than Mt in sun-exposed intertidal habitats. Wonham (2004) reported that only one intertidal hybrid (Mgt) has been reported for Washington state; however, our results show that pilings have significant numbers of Mg and Mgt in certain locations. Holloway and Connell (2002) showed that urban structures such as docks and pilings provide novel habitats for some organisms. Urban structures are typically more prevalent in port areas with high levels of shipping traffic, which could facilitate the invasion of exotic species (e.g., Bax et al. 2002). These urban structures appear to provide more suitable habitat for the introduced Mg than natural intertidal habitats in Puget Sound, thus facilitating the survival of Mg and hybrids in this region. Holloway and Connell (2002) also found that Me mussels were most common on floating docks in comparison to pilings and the rocky intertidal. They suggest that floating structures differ in a number of physical characteristics from fixed structures such as pilings (e.g., constant vs variable depth, light, water flow, access by benthic predators), which may influence recruitment, survival or growth of sessile organisms.

Rensel et al. (2005) examined whether differential survival, growth, or competition caused the differences in mussel genotype frequencies among habitats and microhabitats in Puget Sound. Competition between mussel genotypes for limited space or food resources did not limit the survival or growth of Mt in the subtidal zone on docks. However, Mg and Mt mussels had higher growth and survival rates on floating docks than in the intertidal on pilings. Similarly, Schneider and Helmuth (2007) found that Mytilus spp. had higher survival in subtidal habitats compared to sun-exposed intertidal habitats. The harsher and more variable environmental conditions (higher temperatures and desiccation stress) in the intertidal zone in comparison to the constant temperatures and submersion in the subtidal are likely causes. The submersion of mussels on floating docks likely provides more constant conditions for feeding than in the intertidal zone where a mussel spends significant time periods out of water. A number of studies have found that growth rates decrease at higher tidal heights (Gardner and Skibinski 1988; Kenchington et al. 2002). Habitat and microhabitat differences among the different mussel genotypes are likely the result of a complex interplay of abiotic and biotic factors acting during pre and post-settlement, and further studies are needed to determine which of these is most important in causing the observed distribution and abundance patterns.

Mytilus distribution within Puget Sound

Despite the differences in sampling protocols and frequencies reported in this and previous studies (Suchanek et al. 1997; Anderson et al. 2002; Wonham 2004), similar geographic patterns in genotype frequencies were observed in our study as in previous studies. There were few to no Mg alleles present in Whidbey Basin, low frequencies in the main basin, and highest frequencies of Mg alleles in sheltered bays such as Dyes Inlet, Quartermaster Harbor, Totten Inlet, and Oakland Bay. Our random samples indicate lower frequencies of Mg and Mgt at all locations where we sampled the same localities as Anderson et al. (2002) and Wonham (2004). The random sampling protocol that we employed probably resulted in fewer large mussels being included in our samples leading to lower frequencies of Mg and Mgt (see below).

The highest concentrations of Mg and Mgt we encountered were on docks in shallow bays, such as Quartermaster Harbor (Dockton), Dye’s Inlet and Oakland Bay (Shelton). These bays are characterized by low amounts of freshwater input, small variations in salinity, and higher summer temperatures. These bays are unlike others that had relatively low Mg and Mgt genotype frequencies (e.g., Elliot Bay and Commencement Bay) and have high levels of freshwater input and high variation in salinity. We found no Mg or extremely low frequencies at all sites in the Whidbey basin, which typically has high levels of freshwater input and high variation in salinity. However, beyond these general geographic patterns we did not find a significant statistical correlation between the frequency of Mg mussels and temperature or salinity (means, maximum, minimum) at different locations in Puget Sound (data not shown, Dept. of Ecology marine water quality monitoring data; http://www.ecy.wa.gov/apps/eap/marinewq/mwdataset.asp). This is not consistent with the significant relationship observed between the frequency of Mg mussels and temperature and salinity at different locations by Braby and Somero (2006a) in the San Francisco region. A variety of other studies have also described Mg as being intolerant of low salinities (Sarver and Foltz 1993; Hilbish et al. 1994; Geller et al. 1994; Hofmann and Somero 1996; Matson et al. 2003; Rensel et al. 2005; Braby and Somero 2006a, b). Environmental conditions (e.g., water flow, temperature, salinity) can vary dramatically with differences in tidal height, depth, and aspect over spatial scales of only centimeters (e.g., Rensel et al. 2005; Schneider and Helmuth 2007). Given that the WA Dept. of Ecology water quality data that we used in our analysis were not collected at the exact sites or depths of the mussels studied, it is not surprising that we did not find a correlation between temperature/salinity and mussel genotype frequencies.

One pattern that demonstrates small spatial scale variation is that Mt were more abundant near the water surface on docks (swash zone), whereas Mg were most abundant on the lower sides and bottoms of docks. Salinity is lowest near the water surface and typically increases to near full seawater levels within the top 0.5–1 m depth at most sites in Puget Sound. Thus, if Mg mussels are intolerant of low salinities then they could survive better on the deeper sections of docks. Holloway and Connell (2002) also found that Mytilus mussels were most abundant on floating structures with a swash zone. Which environmental factors cause this pattern were not determined, but it is clear that more detailed research is needed that measures salinity/temperature/water flow at the exact locations where mussels are sampled to better determine whether physiological tolerances/preferences to certain environmental conditions cause the spatial variation in distribution and abundance observed in nature.

Implications for monitoring introduced species and hybrid zones

Our results have important implications for sampling protocols in Mytilus hybrid zones. Many previous studies of Mg on the Pacific coast have sampled mussels “haphazardly” (Sarver and Foltz 1993; Anderson et al. 2002, Wonham 2004) or no sampling protocol has been specified. We have demonstrated that both the size of collected mussels and the specific habitat in which the mussels were collected will bias the outcome of any study in Puget Sound and should provide a caution to others working in Mt/Mg hybrid zones and, probably, in any Mytilus hybrid zone. In Puget Sound, mussels sampled from the sides and bottoms of docks will have higher frequencies of Mg and Mgt than mussels from the intertidal. If large mussels are preferentially sampled, a higher frequency of mussels with Mg alleles will be sampled than are actually present. Thus, if the goal or the future potential use of a study is to compare genotype frequencies of mussels among locations over time (an important objective of introduced species and hybrid zone monitoring), we recommend a standardized sampling protocol be developed that involves collecting random samples from the same habitat/microhabitat types (e.g., middle sides of docks). In addition, in assessments of random mating it is critical that a truly random sample of the population be taken. If there are size differences among species in a hybrid zone and the sample is size biased, assessments of random mating will likely be flawed because one of the parental species will be under-sampled relative to its actual frequency. To track the distribution and abundance of mussels over time in areas with introduced Mg, a consistent and representative sampling regime should be adopted. If a study’s goal is to obtain accurate genotype frequencies for a location, we recommend that a random and stratified sampling regime be incorporated that also takes into account the relative proportion of habitats in an area.

Acknowledgments

Many thanks to the University of Puget Sound Summer Research Committee and the University Enrichment Committee for financial assistance for this research, G. King and J. Davis at Taylor Shellfish Farms for supplying Mt and Mg mussels and information, M. Morrison, L. Rudensey, E. Young, J. Garamella, K. Haley, R. Coon, and all of the others who have helped us in numerous ways with this research. We thank J.P.A. Gardner and A. Wood for insightful comments on the manuscript and Victoria University, Wellington for space and hospitality to PHW. The methods of the study comply with the current laws of Washington State and the USA.

Copyright information

© Springer-Verlag 2008