Marine Biology

, 154:919

Growth and survival differences among native, introduced and hybrid blue mussels (Mytilus spp.): genotype, environment and interaction effects

Authors

  • Jody L. Shields
    • Lehrstuhl für Zoologie und Evolutionsbiologie, Fachbereich BiologieUniversität Konstanz
    • Great Lakes Institute for Environmental ResearchUniversity of Windsor
  • Penny Barnes
    • Centre for Shellfish ResearchMalaspina University College
    • Great Lakes Institute for Environmental ResearchUniversity of Windsor
Original Paper

DOI: 10.1007/s00227-008-0985-0

Cite this article as:
Shields, J.L., Barnes, P. & Heath, D.D. Mar Biol (2008) 154: 919. doi:10.1007/s00227-008-0985-0
  • 152 Views

Abstract

The Mytilus species complex consists of three closely related mussel species: Mytilus trossulus, Mytilus edulis, and Mytilus galloprovincialis, which are found globally in temperate intertidal waters. Introduction of one or more of these species have occurred world-wide via shipping and aquaculture. Stable hybrid zones have developed in areas where these species have come into contact, making the invasion process complex. On the east coast of Vancouver Island (VI), British Columbia (BC), Canada, the native (M. trossulus) and introduced species (M. edulis and M. galloprovincialis), as well as their hybrid offspring, occur sympatrically. This study used a common environment experiment to quantify growth and survival differences among native, introduced, and introgressed mussels on VI. Mussels were collected from an area of known hybridization and reared in cages from May to August 2006. The cages were deployed at a local site as well as a remote site (approximately 150 km apart), and the mussels were genotyped at two species-specific loci. Growth and survival, as fitness measures, were monitored: native, introduced, and introgressed individuals were compared between and within sites to determine whether growth and survival were independent of site and genotype. Overall, mussels reared at Quadra Island performed better than locally-reared mussels at Ladysmith. Specifically, introgressed mussels reared at Quadra Island performed better than all genotypes reared at Ladysmith, as well as better than native mussels reared at Quadra Island. Differences in survival and growth among the native, introduced and introgressed mussels may serve to explain the complex hybridization patterns and dynamics characteristic of the VI introgression zone.

Introduction

Historically, natural hybridization among most animal species was thought to be an evolutionary dead end (Mayr 1942). However, hybridization has been observed in animal taxa, often across environmental gradients creating transition zones between genotypes, these are known as hybrid zones (Barton and Hewitt 1985). Genotype–environment interactions may influence the structure and stability of hybrid populations, as environment-dependent selection can result in differential survival of genotypes (Slatkin 1973; Moore 1977; Springer and Heath 2007). Thus, spatial and temporal variability in the distribution and abundance of hybrid individuals will depend on the fitness of hybrid, relative to parental, genotypes (Albert et al. 2006). Characterization of such hybrid zone dynamics has primarily relied on three models which differ in their assumptions of the relative fitness of hybrid genotypes: the ecotonal (Endler 1977), tension (Barton and Hewitt 1985) and mosaic (Harrison and Rand 1989) models. The tension and mosaic models suggest that hybrids are uniformly less fit than parental genotypes due to their mixed ancestry and resulting genetic incompatibility, and their distribution and abundance is maintained by a balance between dispersal and selection against hybrid individuals (Arnold 1997). The ecotonal model, on the other hand, suggests that hybrid genotypes exhibit higher fitness relative to either parental genotype at some point along an environmental gradient (e.g., Rawson et al. 1999). Assessing the relative fitness of hybrid individuals is thus essential for understanding the maintenance of hybrid zones and the ability of the parental taxa to maintain genetic integrity in the face of hybridization (Day and Schluter 1995), although direct estimates of fitness are generally logistically difficult.

Marine blue mussels of the Mytilus species complex (Mytilus trossulus, Mytilus edulis, and Mytilus galloprovincialis) occur globally in temperate waters and hybridize where they come into contact (Hilbish et al. 2000). Thus they form an excellent system, in which, to investigate the interaction of genotype and environment and the breakdown of interspecific reproductive barriers. Several studies of Mytilus spp. have shown that environmental effects are large determinants of both growth and survival (Dickie et al. 1984; Mallet and Carver 1989; Johannesson et al. 1990; Kautsky et al. 1990; Stirling and Okumus 1994). Environment-dependent effects on indirect measures of fitness such as physiological performance, reproductive investment, fecundity, strength of attachment to the substrate and susceptibility to parasitic infections have been shown in mussels (e.g., Rawson et al. 1999; Riginos and Cunningham 2005; Gardner 1994). Water temperature, salinity and wave exposure appear to be the most important environmental factors driving fitness differences in Mytilus hybrid zones (Riginos and Cunningham 2005). Such variance in fitness measures among genotypes may reflect underlying genetic incompatibilities, which are conditionally expressed in an environment-dependent manner (Rundle and Whitlock 2001; Springer and Heath 2007).

Since divergent taxa tend to become fixed for different alleles at multiple genes, their hybrid offspring are expected to be highly heterozygous, which may result in the masking of hybrid breakdown in early generations due to heterosis (Rhode and Cruzan 2005). This effect would play a critical role in the establishment of hybrid zones, and perhaps in the subsequent introgression between the two parental taxa (Arnold 1997; Burke and Arnold 2001), since it will serve to create novel genotypes that perform better than their parents in certain environments. However, local adaptation may contribute to fundamental genetic incompatibilities, perhaps masked in F1 hybrids, which result in variation in the fitness of backcross individuals that will depend unpredictably on the environment (Whitlock et al. 2000; Burke and Arnold 2001; Rundle and Whitlock 2001).

If increased heterozygosity in early generation hybrids serves to mask deleterious alleles due to heterosis, a measurable performance advantage relative to parental individuals is expected. Hybrids have been found to be equally or more fit than their parents in both laboratory and field studies among several taxa (reviewed in Arnold and Hodges 1995; Bombina: Nurnberger et al. 1995; Geospiza: Grant and Grant 1992; Mytilus: Gardner et al. 1993). Attempts to relate growth (a fitness related trait) and heterozygosity (a metric of hybridization), independent of environmental factors, have produced inconsistent results. Several studies of hybrid populations of bivalves failed to find a positive correlation between heterozygosity and growth due to problems with replication (Skibinski and Roderick 1989) and with sampling from a limited number of parental populations (Beaumont et al. 1983; Gaffney and Scott 1984; Adamkewicz et al. 1984). Studies showing a positive growth–heterozygosity correlation in several natural populations of mussels (Koehn and Gaffney 1984; Diehl and Koehn 1985; Hawkins et al. 1986) have focused on the well established hybrid zones of northwestern Europe. To date, little is known of the structure and dynamics of the relatively young hybrid zones on the Pacific coast of North America (Braby and Somero 2006).

All the three sibling Mytilus species are found on Vancouver Island (VI), British Columbia (BC), Canada where M. trossulus is the native species. The three species form hybrids, with an unstable abundance that does not appear to fit any one particular model (Yanick 2002; JL Shields, unpublished data). This lack of stability makes it difficult to define the VI introgression areas as a true “hybrid zone”, as the abundance of hybrid mussels on VI does not appear to be self-sustaining, but rather is likely maintained by repeated introductions (Yanick 2002). The hybrid zone found on VI has a localized structure where parental individuals are found intermixed with hybrid, backcrossed, and higher order introgressed individuals, whose relative fitness is unknown. The VI hybrid zone differs from those in California, where environmentally dependent selection influences the hybrid zone structure, as Mytilus spp. genotypes correlate with both temperature and salinity along an estuarine–oceanic gradient (Rawson et al. 1999; Braby and Somero 2006). Although Yanick et al. (2003) demonstrated local adaptation effects in the native M.trossulus on VI using a transplant study, there is no known environmental gradient that can explain the dynamics of the VI hybridization and introgression. Thus, the Mytilus hybrid zone off the coast of VI (Heath et al. 1995) provides a unique opportunity to examine the contribution of genetics, environment and genotype–environment interactions to the distribution and abundance of hybrid and backcross offspring. A first step is to estimate components of relative performance of parental taxa and their hybrids in various environments since such data are key to uncovering the mechanisms of hybrid zone stability, or lack thereof (Rolan-Alvarez et al. 1997). In this study, we used growth and survival measures to estimate relative performance of Mytilus genotypes in the VI hybrid zone. We predicted environmentally-dependent performance differences between parental and hybrid groups, and that the pattern of those differences would explain the distribution and dynamics of the Mytilus hybrid zone on VI. A clearer understanding of the complex dynamics of the VI Mytilus hybrid zone will provide valuable insight into how reproductive barriers are eroded, as well as inform management efforts aimed at minimizing the potential impact of non-native species introduction.

Materials and methods

Site selection and sample collection

In May of 2005 and 2006, a genetic survey of Mytilus spp. distribution on the east coast of Vancouver Island indicated that Ladysmith, BC had an abundance of introduced and hybrid mussels. Ladysmith was chosen as the collection site for this study to increase the likelihood of sampling native, introduced and introgressed mussels. A total of 864 small [average shell length (±1 SEM) = 18.6 ± 1.0 mm] mussels were collected from the underside of the Ladysmith Harbour Maritime Society dock. Half of the collected mussels (n = 432) were reared locally at Ladysmith, while the other half were transported to a site on Quadra Island, BC (Fig. 1). All the mussels were placed in a container of seawater and shell measurements (length, width, and depth) were taken at the largest part of the shell on each mussel. Length was measured from the tip of the beak to the base of the posterior edge; width was measured across the widest portion of the mussel from the umbo to the byssal opening; and depth was measured at the center of the valve at the peak curvature when held dorso-ventrally. In addition, a hemolymph sample was taken for each mussel following the protocol of Yanick and Heath (2000). The mussels were then placed in individual compartments in the cages (see below). To standardize for the travel time to Quadra Island (approximately 3 h), the mussels deployed at Ladysmith were held in the cages (out of direct sunlight) for 3 h prior to final submergence.
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-008-0985-0/MediaObjects/227_2008_985_Fig1_HTML.gif
Fig. 1

Map of Vancouver Island, British Columbia showing the deployment sites for the mussel cages. The source of the experimental mussels was the Ladysmith (local) site which is a commercial dock in the Ladysmith harbor, while the Quadra Island (transplanted) site is located at the Yellow Island Aquaculture Ltd fish farm site on the west central coast of Quadra Island

Rearing

Growth and survival of mussels at the local (Ladysmith, BC) and remote (Quadra Island, BC) sites were evaluated in a common environment cage experiment carried out from 26 May to 27 August 2006. The choice of the timing for this experiment was predicated on the known high mussel growth and mortality associated with the warmer water temperatures of the summer months. The cages used were as described in Yanick et al. (2003), and consisted of a 70 × 70 × 5.0 cm polyethylene sheet, 0.5-mm mesh screen on either side of the center sheet and 1.0-cm polyethylene sheets on both outer sides. A total of 144 5.0-cm holes (in a 12 × 12 grid) held individual mussels. Each cage thus housed 144 mussels in individual cells designed to keep mussels separate for identification purposes while allowing sufficient water flow for obtaining food and expelling wastes. Fouling organisms were removed from the cages and screens regularly throughout the experiment. Three cages (housing a total of 432 mussels) were deployed at each site, suspended at approximately 1.0 m below the surface. Temperature loggers (StowAway Tidbit; TBI32-05+37; Onset Computer Corporation) were attached to the rope of one cage at each site. Temperature was recorded at 15-min intervals from May to August 2006 and was compared between sites using a t test.

Survival and growth

Survival and growth of the mussels were monitored monthly for the experimental period, 26 May 2006 to 27 August 2006. Survival was assessed by retrieving the cages and identifying all mussels as either alive or dead. The shells from the dead mussels were removed from the cages and preserved in 50 mL tubes containing 95% EtOH. To determine growth, digital calipers (±0.1 mm) were used to take shell measurements (length, width, depth) at the starting and end of the experiment. Mussels that died during the experiment were not used in growth comparisons, since we could not identify the date on which mussels died.

Species identification

DNA was extracted from the hemolymph samples (see Yanick and Heath 2000) to determine the proportion of introgressed and introduced mussels in the cages. When the experiment was terminated in August 2006, mantle tissue was taken from all surviving mussels, stored in 95% EtOH, and transported to the laboratory for DNA extraction (following Elphinstone et al. 2003). DNA fragments were amplified via PCR at two species-specific co-dominant marker loci: the internal transcribed spacer region of ribosomal DNA (ITS—diagnostic between native and alien mussels and their hybrids) following the protocol described in Heath et al. (1995), and the adhesive byssal thread protein (Glu-5′—diagnostic among all three mussel species and their hybrids) following a modified protocol of that described in Rawson et al. (1996). Individual mussels were genotyped at each locus via diagnostic restriction fragment length polymorphisms (RFLPs) for ITS (Heath et al. 1995) and using an automated DNA analyzer (LiCor 4300) to determine PCR fragment length polymorphisms for Glu-5′ (Rawson et al. 1996). Mussels were genetically identified as homozygote [pure native Mytilus trossulus (MT), introduced M. galloprovincialis (MG), or introduced M. edulis (ME)] or heterozygote (MT × MG, MT × ME or MG × ME) at the Glu-5′ marker locus. The ITS marker locus identified mussels as homozygous (native MT or pure introduced ME or MG), or as heterozygote (MT × ME or MT × MG). Mussels were classified as F1 hybrids when heterozygous at both ITS and GLU. Backcross (BC) and higher-order hybrid crosses were identified whenever the markers disagreed. With only two diagnostic markers, error in hybrid classification by genotype is unavoidable, hence only the backcross (BC) classification is certain.

Analyses

Since fitness is logistically difficult to measure in mussels, we estimated fitness indirectly using survival and growth as indicators (see Skibinski and Roderick 1989; Bert and Arnold 1995; David 1998; Arnold and Hodges 1995). Survival was calculated as counts (dead or alive) for each genotype, where dead mussels were identified to species using DNA from the hemolymph sample taken before deployment. For survival comparisons among genotypes, percent survival was calculated as the total number of mussels of a particular genotype alive at the end of the study divided by the total number of mussels of that genotype at the start of the study. To determine if survival was independent of site and genotype, survival was calculated for the duration of the experiment (May–August 2006) for each site and genotype and a 2-way contingency table (χ2) was used to determine if there was a significant difference in survival between the local and transplanted mussels. To determine if the much larger sample size of native (MT) mussels was affecting the results, we adjusted the MT sample size by randomly selecting 10 individuals, from those surviving at each site, using bootstrap re-sampling techniques with 1,000 iterations at a 90% confidence interval in PopTools (www.cse.csiro.au/poptools). Of the 1,000 subsamples (n = 10), 50 were randomly chosen to repeat the analyses. Due to small sample sizes in some genotype classes, all individuals of mixed ancestry (putative F1 and backcross) were grouped into a single “introgressed” category. Contingency tables were used to determine differences in survival among genotypes within and between sites. Log-likelihood ratio tests (G test for independence) were used when small sample sizes among mussel genotypes precluded contingency table analyses (Sokal and Rohlf 1995).

Relative growth rate (RGR) was calculated for each mussel at each site and compared between locally reared (Ladysmith) and transplanted (Quadra Island) mussels, as well as among genotypes [native (MT), introduced (MG), introgressed (F1 and backcrossed)] to determine if growth was independent of site and genotype. Only those mussels that survived the experiment were used in growth comparisons, since we could not identify the exact date on which mussels died. RGR was calculated as:
$$ {\text{RGR}} = \frac{{({\text{SL}}_2 - {\text{SL}}_1 )}}{{({\text{SL}}_1 \times t)}} $$
where SL1 = shell length at time 1, SL2 = shell length at time 2, and t = number of days between the two dates. This equation was applied to each shell measure (length, width and depth), resulting in three RGR values for each mussel. Although shell length is generally the standard measure used in growth studies of bivalves, here we measured length, width and depth of each mussel, thereby allowing the estimation of volumetric growth. Principal components analysis (PCA factor analysis, Statistica 6.0) reduces variables that may be autocorrelated into a series of orthogonal variables; the first few principal components usually account for the majority of the observed variation in the original variables. RGR of the three shell measurements were combined into a single “volumetric growth” variable using PCA, as length, width and depth are correlated due to isometric growth in mussels. This reduced the data set from three variables to a single variable for each mussel, which was used in comparisons of growth among individuals. A 2-Factor ANOVA (Statistica 6.0) was used to determine if there was a significant difference (P < 0.05) in growth (PC1) between sites and among genotypes, as well as genotype–site interaction effects. Post hoc Tukey tests were used to discern significant differences among genotypes. To determine if the much larger sample size of native (MT) mussels was affecting the results, we adjusted the MT sample size by randomly selecting 10 individuals, from those surviving at each site, using bootstrap re-sampling techniques with 1,000 iterations at a 90% confidence interval in PopTools (www.cse.csiro.au/poptools). Of the 1,000 subsamples (n = 10), 50 were randomly chosen to repeat the 2-Factor ANOVA (Statistica 6.0) to determine if there was a significant difference in volumetric growth (PC1) between sites and among genotypes, as well as genotype–site interaction effects. Due to small sample sizes in some genotype classes, all individuals of mixed ancestry (putative F1 and backcross) were grouped into a single “introgressed” category for growth analyses.

We estimated total mussel volume (cm3) as the volume of an ellipsoid [(3/4π) × (L)(W)(D)] and, because fecundity is highly correlated with shell volume in marine bivalves (Jablonski 1996; David 1998), total mussel volume provides an important fitness variable. A linear relationship between shell volume and survivorship was found by plotting the average shell volume of each genotype at each site against the average survivorship for each genotype at each site. Volume and survival were used to calculate a fitness metric [average volume (cm3) of a genotype × average survival of a genotype] from which relative fitness was estimated among genotypes. It should be noted that our fitness metric is not true lifetime fitness, but does reflect relative fitness during a critical stage in the mussel life history. Native MT mussels reared at Ladysmith were set as the reference to which all other genotype fitness estimates were compared, such that genotype-specific fitness was relative to native mussels in their home environment. Relative fitness was calculated as the fitness metric of one genotype divided by the fitness metric of the native MT genotype reared at Ladysmith (FMn/FMMT). An ANOVA was used to partition the variance in genotype-specific fitness metrics between sites and among genotypes within sites. All the individuals of mixed ancestry (putative F1 and backcross) were grouped into a single “introgressed” category for survival analyses for consistency of methods. However, results for such analyses should be interpreted with caution as combining genotypes in such analyses can be inappropriate and can generate misleading results (Arnold 1997) due to expectations of outbreeding depression among introgressed genotypes.

Results

Of the 432 mussels reared at Ladysmith (local) and 432 mussels transplanted to Quadra Island, 307 (71.1%) and 347 (80.3%) mussels, respectively, survived until the end of the experiment. Of the original 864 mussels used in this experiment, 705 (alive and dead) were successfully identified to species, with the majority being pure native (Table 1). The proportions of genotypes sampled deviate significantly from Hardy–Weinberg Equilibrium (HWE) expectations (GLU: χ2(2) = 325.07, P < 0.001; ITS: χ2(2) = 296.67, P < 0.001). No pure M. edulis were identified, perhaps not surprisingly as according to HWE, only 2 of 100,000 are expected. Genotype proportions were similar at the two sites (Table 1). Average water temperature (at 1-m depth) at Quadra Island (11.5 ± 0.0165°C, Fig. 2) was significantly cooler than Ladysmith (19.7 ± 0.0178°C, Fig. 2) for the duration of the experiment (t = 1.66, P < 0.0001; Fig. 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-008-0985-0/MediaObjects/227_2008_985_Fig2_HTML.gif
Fig. 2

Mean daily water temperature (°C) profile for Ladysmith (black) and Quadra Island (grey) sites at 1 m depth

Table 1

Survival of individually caged mussels at two sites (Ladysmith, BC; Quadra Island, BC) by genotype

Site

Genotype

Classification

Alive

Dead

Total

GLU

ITS

LS

MT

N

MT

282

35

317

MG

A

MG

6

2

8

MT × MG

H

F1

7

0

7

MG × ME

H

F1

7

1

8

MG × ME

A

BC

1

0

1

MT

A

BC

4

3

7

Site total

   

307

41

348

QI

MT

N

MT

329

10

339

MG

A

MG

6

0

6

MT × MG

H

F1

2

0

2

MG × ME

H

F1

9

0

9

MG × ME

A

BC

0

0

0

MT

A

BC

1

0

1

Site total

   

347

10

357

Total

   

654

51

705

At the Glu-5′ locus: MT, Mytilus trossulus; MG, Mytilus galloprovincialis; MT × MG, M. trossulus × M. galloprovincialis hybrid; MG × ME, M.galloprovincialis × M. edulis hybrid. At the ITS locus: N Native (M. trossulus), A alien (either M. edulis or M. galloprovincialis), H hybrid (either M. trossulus × M. galloprovincialis or M. trossulus × M. edulis)

Survival

At Ladysmith, survival varied with genotype as follows: 88.9% (282/317) for the native, 75.0% (6/8) for the introduced, 82.6% (19/23) for the introgressed (Fig. 3). Genotype-specific survival was less variable at Quadra Island: 97.1% (329/339) for the native, 100% (6/6) for the introduced, 100% (12/12) for the introgressed (Fig. 3). Overall survival (all mussels regardless of genotype) was significantly higher at Quadra Island (χ2(1) = 9.58, P < 0.05; Fig. 3) than at Ladysmith. There was a significant 3-way interaction among site, genotype, and survival (G(2) = 7.26, P < 0.05), suggesting that genotype-specific survival is environmentally dependent. Survival among mussel genotypes (both sites combined) did not differ significantly, (G(2) = 1.75, P > 0.05). There were also no significant differences among genotype-specific survival within sites; Ladysmith G(3) = 2.19, P > 0.05, and Quadra Island G(3) = 1.42, P > 0.05 (Fig. 3). However, the lack of statistical significance within sites may be due to small sample sizes in some size classes. We found the results of the 50 randomly selected subsamples (n = 10) were not significantly different from the results of the full sample at both Ladysmith and Quadra Island. LS sub-sample: mean 87.8% MT survival (variance 0.46%); QI sub-sample: mean 95.2% MT survival (variance 1.83%). All of the 50 randomly selected subsamples (n = 10) indicated the same significant G test results as the full MT samples among site, genotype, and survival (average result: G2 = 7.35; P < 0.05).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-008-0985-0/MediaObjects/227_2008_985_Fig3_HTML.gif
Fig. 3

Percent survival of caged Mytilus mussels from 27 May to 30 August 2006 calculated as the number of mussels of a genotype alive at the end divided by the number of mussels of that genotype to start. MT, Mytilus trossulus; Int introgressed (includes F1 and backcrosses); MG, Mytilus galloprovincialis

Growth

Principal component 1 (PC1) accounted for 95% of the variance in shell measures. Using PC1, volumetric growth was compared between sites and among genotypes [native MT, introduced MG, and introgressed (putative F1 and backcrossed combined)]. Overall, mussels reared at Quadra Island exhibited faster growth than those reared at Ladysmith (Fig. 4a; F(1,57) = 12.59, P < 0.0001). These results held true regardless of adjustment for unequal sample size. We found the mean and variance of the 50 randomly selected subsamples (n = 10) were not significantly different from the mean and variance of the full sample at both Ladysmith and Quadra Island (LS full: mean 0.771, var 0.859; LS subsample: mean 0.737, var 0.862; QI full: mean 1.231, var 0.895; QI subsample: mean 1.219, var 0.788). All of the 50 randomly selected subsamples (n = 10) indicated the same significant ANOVA results as the full MT samples between sites (average result: F(1,57) = 12.49; P < 0.0001), site–genotype interaction (average result: F(3,57) = 3.93, P = 0.013) as well as the specific comparisons among genotypes. Significant site–genotype interactions were observed (F(3,55) = 3.95, P = 0.013), indicating that growth patterns among genotypes differed between sites. Specifically, introgressed mussels reared at Quadra Island showed significantly greater volumetric growth than the native MT reared at the same site (Fig. 4a) as well as all genotypes reared at Ladymith (Fig. 4a). Additionally, native MT reared at Quadra Island showed significantly greater volumetric growth than native MT reared at Ladysmith (Fig. 4a). Within sites, at Ladysmith, where the water temperature was warmer and more variable (Fig. 2), post hoc Tukey tests revealed that growth among genotypes was not significantly different (Fig. 4a). However, at Quadra Island, where water temperature was cooler and more stable (Fig. 2), Tukey tests revealed growth among genotypes was significantly different. Specifically, introgressed mussels exhibit greater volumetric growth than native MT genotypes at this site (Fig. 4a). There were no significant differences in shell length among genotypes at Ladysmith or Quadra Island (Fig. 4b).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-008-0985-0/MediaObjects/227_2008_985_Fig4_HTML.gif
Fig. 4

Volumetric growth and shell length. a Relative growth rate as measured by PC1 from May to August 2006 at Ladysmith and Quadra Island among MT, MG, Introgressed (F1, BC) genotypes. PC1 combines length, width and depth into a single variable accounting for 95% of the variance in these metrics. Overall, Quadra Island had significantly higher growth rate than Ladysmith (P < 0.05). b Final (August 2006) shell length (mm). Error bars represent standard error. a, b, c and d designate significant differences

A linear relationship between mean final shell volume (cm3) and survival (arcsine square root transformed for percentage data) explained 84% of the variance in survival across all rearing site–genotype combinations (Fig. 5a). The fitness metric (volume × survival) was significantly higher at the Quadra Island site (F(7,1) = 26.27, P < 0.05; Fig. 5b). Relative fitness among genotypes within sites varied significantly (Fig. 5b). At Ladysmith, F1 mussels had higher fitness than MT (F(7,3) = 6.51, P < 0.001) and MG (F(7,3) = 5.86, P < 0.001; Fig. 5b). At Quadra Island F1 mussels had higher fitness than MT (F(7,3) = 5.92, P < 0.001. At both Ladysmith and Quadra Island, the F1 genotype had the highest fitness relative to the native MT in its home environment (Fig. 5b).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-008-0985-0/MediaObjects/227_2008_985_Fig5_HTML.gif
Fig. 5

Relationship between size and survival and genotype relative fitness of mussels reared at Ladysmith and Quadra Island. a A linear relationship between survival and volume. Survival was arcsine square root transformed for percentage data. b fitness calculated relative to Ladysmith MT, indicated by the dashed line at 1. Error bars represent standard error. a, b, c and d designate significant differences

Discussion

Our results indicate that growth and survival in VI Mytilus varies among genotypes and between environments. While the reproductive success of mussels cannot be directly measured through growth and survival estimates, many studies use such parameters as proxies for fitness (reviewed in Arnold 1997; Gardner et al. 1993; Gardner and Skibinski 1990; David 1998). Furthermore, body size (generally measured as shell length, volume, or mantle weight) is positively correlated with fecundity (Mzighani 2005; Ehrlich 1989; Duncan et al. 2001; Rodhouse et al. 1986). In marine bivalves, Mzighani (2005) found a strong correlation between shell length and fecundity (R2 = 0.9509) in their study of cockles, and suggested that the best parameter for estimating the number of eggs is in fact shell length. Therefore, since growth (fecundity) and survival are primary determinants of relative fitness, they are likely critical hybrid zone parameters as the structure and maintenance of hybrid zones is defined by the relative fitness of hybrid individuals. Where hybrids exhibit a fitness advantage over the parental species, introgression is likely and may lead to the breakdown of reproductive barriers between the parental species, resulting in extensive hybrid zones (Burke and Arnold 2001). In contrast, where hybrid fitness is reduced relative to parental types, the cohesion of taxa is reinforced via ecological selection or strict genetic incompatibilities (Barton and Hewitt 1985). Alternatively, hybrid fitness may be expressed in an environment dependent manner (Rundle and Whitlock 2001), resulting in transient and unstable hybrid zones. Our data demonstrate varying patterns of fitness-related parameters, differentially expressed in an environment-dependent manner, and indicate that hybrid mussels are not universally less fit than the pure parental types due to inherent genetic incompatibilities limiting growth or survival (Burke and Arnold 2001; Springer and Heath 2007). Furthermore, the existence of many backcross and other higher-order introgressed genotypes argue strongly against intrinsic genetic incompatibilities (Springer and Heath 2007), making growth and survival reasonable fitness estimators.

Native, introduced, hybrid and introgressed mussels differed in growth and survival patterns over the summer months May–August 2006, a time in which environmental conditions, including water temperature, vary considerably. The high survival and greater relative growth of introgressed mussels reared at Quadra Island suggests a fitness advantage, at least over the experimental period. Elevated hybrid fitness is not consistent with the observed maintenance of the pure native species over much of VI; however, spatial and temporal variation in growth and survival among genotypes may address this apparent anomaly. Introgressed genotypes exhibit greater relative fitness over the study period than all other genotypes at both Ladysmith and Quadra Island; however, genotype-specific fitness patterns were not consistent between sites, suggesting a genotype–environment interaction (growth showed a significant genotype–environmental interaction effect in the ANOVA). Environment-dependent selection has been shown for Pacific Mytilus zones in California (Rawson et al. 1999) and recently in Chemainus, BC and San Francisco, CA (Springer and Heath 2007). Earlier studies also demonstrate genotype-dependent mortality and selection within Mytilus populations in both southwest England (Gardner et al. 1993; Gardner & Skibinski 1991; Secor et al. 2001) and Atlantic France (Coustau et al. 1991). Thus, environment-dependent variation in mussel growth and survival likely contributes to the incomplete reproductive isolation between Mytilus species and the spatial limitation of the VI hybrid zone.

Genotype-specific growth and survival patterns in the VI Mytilus hybrid zone suggests that selection against hybrids is limited, at least during the summer months, and indicates that non-native genotypes (including pure introduced, F1 and backcross) have the potential to thrive in northern habitats. However, the VI hybrid zone has not expanded, persisting with essentially the same geographic range for over 10 years (Heath et al.1995). We transplanted mussels from an area of high temperature to an area of lower temperature at the beginning of summer when water temperature may approach the upper thermal tolerance of Mytilus off southern VI. Thus the transfer in this study may have served to alleviate thermal stress, allowing the transplanted mussels to invest more energy into growth. This would explain the lower growth and survival of all genotypes at Ladysmith during the summer: perhaps mussels are released from the upper thermal limit when transplanted to the cooler water temperatures of Quadra Island, resulting in 100% survival and intermediate growth. Furthermore, our study estimated growth and survival during the summer months only, and relative fitness relationships may change through the winter months. Indeed, previous studies have documented dramatic seasonal declines in the abundance of introduced and introgressed genotypes from summer to winter months in the VI hybrid zone (Yanick 2002, Shields 2007). Such a seasonal decline suggests variable selection against introduced and introgressed individuals which may be related to temperature as thermal stress has been implicated as a cause of mortality in mussels (Tremblay et al. 1998). Thus, seasonal fluctuations in hybrid and introgressed proportions at Ladysmith may be related to temperature: the winter months may be too cold to allow the establishment of a self-sustaining population. In southwest England, M. galloprovincialis is also reported to have a northern limit to its dispersal due to cooler water temperatures, despite having a small but significant growth and survival advantage over other genotypes in the area (Gardner et al. 1993; Gardner and Skibinski 1991; Skibinski and Roderick 1989; Skibinski and Roderick 1991).

Although the apparent growth and survival advantage exhibited by hybrid mussels helps explain the presence and sometimes high abundance of hybrid mussels on VI, it does not explain the lack of hybrid zone expansion beyond southern VI (Yanick 2002). A number of possible factors may contribute to the limited spatial distribution of the VI hybrid zone. Perhaps the southern VI hybrid zone is driven by specific introduction vectors that do not exist further north, and dispersal is limited by unknown physical barriers (e.g., Southwest England—Gilg and Hilbish 2003; Shields 2007). Alternatively, the localized distribution of hybrids on southern VI may be due to hybrids favoring marginal habitats (Choler et al. 2004); Ladysmith is highly impacted by industry and urbanization and, hence, hybrid mussels may experience early-life habitat advantages. Variation in selection pressures, therefore, may be critical in the establishment and maintenance of introduced and introgressed genotypes.

Our data indicate that there is no survival or growth advantage to introduced or introgressed genotypes at Ladysmith. This, coupled with the observed seasonal decline in non-native genotype abundance, suggests that considerable immigration is required to maintain the southern VI mussel hybrid zone. The lack of an established natural population of MG north of California (Heath et al. 1995; Suchanek et al. 1997), suggests that intermittent seeding of M. galloprovincialis from aquaculture tenures in Puget Sound and the Strait of Georgia, as well as shipping, likely play a critical role in maintaining hybrid and introgressed genotypes on southern VI (Heath et al. 1995). The lack of an established, stable introgressed population suggests some inherent genetic incompatibilities that may be conditionally expressed in various environments (Rundle and Whitlock 2001; Springer and Heath 2007). While the persistence of introduced and introgressed genotypes on VI is likely to continue (Heath et al. 1995), the abundance of hybrids is highly unstable and more information is required on environmental factors that may explain the lack of an established self-sustaining hybrid zone.

This study adds to the growing body of literature documenting examples of elevated hybrid performance, suggesting hybridization may play an important role in the evolution of locally-adapted populations or species (Seehausen 2004; Whitlock et al. 2000). Hybrid fitness is habitat specific and assessment of hybrid performance is essential in understanding the mechanisms of hybrid zone maintenance. Our data suggest that introduced and introgressed genotypes should expand their range to northern VI and yet, this has not occurred. While the temporal instability in abundance of introduced and introgressed genotypes may be related to temperature variation, the mechanism maintaining the localized distribution of this hybrid zone is not fully understood. As a result, further investigation into potential dispersal barriers is warranted. Understanding the habitat-mediated variation of genotype-specific fitness may help uncover the dynamics required to create and maintain hybrid zones.

Acknowledgments

We thank B. Dixon, V. A. Heath, J. W. Heath, W. Callery, A. Burgoyne and M. J. Shields for assistance in the field; R. P. Walter and D. W. Kelly for valuable critical reviews; and J. Muirhead for statistical assistance. Field work was supported by Yellow Island Aquaculture Ltd., the Ladysmith Maritime Society, and the Centre for Shellfish Research. This project was funded by the Department of Fisheries and Oceans, Canada, and by Natural Science and Engineering Council of Canada Discovery and Canada Research Chair grants to DDH. This study complies with the current laws of Canada.

Copyright information

© Springer-Verlag 2008