| Collection Mémoires et thèses électroniques | |
| AccueilÀ proposNous joindre |
| Chapitre V. Significant inter-annual variation in seasonal-spatial differences in growth among populations of Dreissena polymorpha (Pallas) and D. bugensis (Andrusov) | ||
|---|---|---|
 |  | |
Table des matières
Des transplantations réciproques durant 2 ans ont été utilisées pour tester si les différences de croissance de la coquille étaient fixées ou plastiques entre les populations source des moules zébrées de l’estuaire et des masses d’eaux des Grands Lacs du fleuve St-Laurent. Les résultats des manipulations d’été et d’hiver indiquaient que la croissance de la coquille variait avec l’environnement (rivière > estuaire), mais il y avait aussi une composante forte de la population source à cette variance (rivière > estuaire). Les moules quagga (actuellement rares dans le système bas du St Laurent) ont été aussi transplantées du site riverain vers le site estuarien. La réduction dans la croissance de la coquille des moules quagga entre le site riverain et le site estuarien était plus importante que celle chez les moules zébrées du même lieu source. Les taux de croissance en été étaient consistants et plus importants, mais il y avait une différence dramatique inter-annuelle entre les moules quagga et zébrées du même lieu source comparée à celle de leur environnement natal. Nous pouvons conclure qu’il y avait une différenciation physiologique entre les populations des moules zébrées et entre les deux espèces à l’intérieur du même écosystème.
A two-year field study of the growth of zebra and quagga mussels (Dreissena polymorpha and D. bugensis) in the St. Lawrence Rivertested whether observed differences in species-specific distribution and dominance can be explained by differential growth and survival of adults. Zebra mussels were reciprocally transplanted between the tidally brackish and riverine portions of the St. Lawrence ecosystem, to examine whether site-specific physiological differentiation is occurring. Quagga mussels, which were only found at the river site, were also transplanted to the estuary. The results of both summer and winter manipulations indicated that while shell growth varied with environment (river > estuary), there is also a nested source population component to the variance (river > estuary). The summer growth rates were very different, though they reversed dramatically in consecutive years (quagga < zebra in 2002 versus zebra < quagga in 2003). The reduction in shell growth of quagga mussels between the river and estuary was greater than zebra mussels from same source location. Shell growth of both species at both sites was greatest in summer regardless of year, but there was large inter-annual difference in which site produced the largest zebra mussel growth rates (river in 2002 but estuary in 2003). The dramatic inter-annual change in growth corresponds to on-going changes in species dominance and distribution in the fluvial section of the river, but not the estuarine. We concludethat there is site-specific physiological differentiation among zebra mussels and between the two species within the same ecosystem.
The zebra mussel (Dreissena polymorpha) invaded the Great Lakes system in the mid-80’s and has since spread rapidly through many freshwater systems in eastern North America. A second dreissenid species, the quagga mussel (D. bugensis), was introduced shortly thereafter and has been displacing zebra mussels in many locations where they co-occur (Mills et al. 1999, Ricciardi and Whoriskey 2004, Wilson et al. 2006). The capacity to tolerate or adjust to novel environments is hypothesized to be a key to the success of dreissenid mussels and other exotic species (Kolar and Lodge 2001, Sakai et al. 2001). The combination of their rapid spread into a variety of different environments and the high level of gene flow via long-range dispersal of both larvae and adults makes the invasion of North America by these species a model for exploring questions rapid population differentiation in novel environments. Initial detection and subsequent pattern of spread are most consistent with a single introduction from northern and central European and Eurasian sources, but the high allelic diversity in North American populations suggest that genetic bottlenecks are unlikely to explain any trans-Atlantic or continental differences in population structure (Herbert et al. 1989, Stepien et al. 2002). Yet despite this, there is evidence of regional (Elderkin et al. 2001, Elderkin et al. 2004a, Elderkin et al. 2004b) and local scale (Lewis et al. 2000) population structuring consistent with either isolation by distance or stochastic processes, possibly related to river hydrology (Horvath and Lamberti 1999, Stoeckel et al. 1997, Schneider et al. 2003, Carr et al. 2004). Examinations of genetic diversity have, however, tended to describe population structure relative to either geographic distance (Lewis et al. 2000) or to an abiotic factor that varies strongly with geographic distance (Elderkin et al. 2001).
In contrast to the other large rivers of North America, the St. Lawrence River has divergent, yet seasonally and inter-annually stable regional water masses (Thorp et al. 2005). The physical-chemical characteristics of these different water masses include several less than optimal water chemistry characteristics for dreissenids: salinity (Chapter 3), low ionic content (Dietz et al. 1996, Wilcox and Dietz 1998), high turbidity and tidal turbulence (Alexander et al. 1994, Thorp et al. 1998, Payne et al. 1999).Natural selection that leads to rapid differentiation and local adaptation is an alternative to isolation by distance that is usually invoked in situations where gene flow among populations is restricted (Schluter 2000). The presence of adult Dreissena throughout the 600-km mainstem of the St. Lawrence River, an ostensibly contiguous population connected via unidirectional larval dispersal, initially suggests a situation in which gene flow limits localdifferentiation. However, the pattern of fixed spatial variation in the physiology of Dreissena polymorpha and D. bugensis seen in earlier chapters (Chapter 2, 3 and 4) is usually associated with restricted gene flow and local selection. Thus the St. Lawrence River provides a mosaic of divergent environments useful for examining the strength of either phenotypic plasticity or natural selection without the confounding effect of distance.Our research focus was to determine if the observed differences in dreissenids from the various water masses related to site-specific adult physiological adjustment (selection-differentiation) or a result of colonization and dispersal dynamics (stochastic-isolation by distance)? We used a reciprocal transplant experiment, moving individuals between the most physically-chemically divergent and geographically distant water masses, to test for differences in performance resulting from demographic causes (i.e., an effect of source) as opposed to differences resulting from a plastic response to differences between environments (i.e., an effect of the destination). The power of this approach lies in both the reliance on a response variable (growth) that synthesizes expression of multiple gene-based traits and the statistical ability to separate endogenous source population variation from that due to divergent exogenous conditions (Schluter 2000, Honkoop et al. 2003).This model system also provides a possible way of testing whether the displacement of the zebra mussel by the quagga mussel (Mills et al. 1999, Ricciardi and Whorisky 2004, Wilson et al. 2006) is more likely linked tospecies-specific differences in adult physiology or constraints on larval dispersal. Therefore we also included the quagga mussel, which is not yet found in the fluvial estuary, within the experimental design.
Reciprocal transplant experiments were conducted between two study sites selected for their contrasting physical-chemical characteristics. The river site “R” was located near Massena, NY, (44o59’ 19’’and 74o49’ 45’’) in an oligotrophic water mass originating from the Great Lakes, whereas the estuarine site “FE” was at Saint-Francois on the eastern tip of Île d’Orléans, Québec, (46o59’ 47’’ and 69o48’ 21”) in the more energetic and turbid waters of the freshwater portion of the tidally dominated fluvial estuary (Figure 5.1). We assumed that the FE site was more stressful for dreissenids due to the higher turbidity, currents, and brackish conditions each of which are physiologically challenging for adults and larvae of Dreissena, although they undoubtedly differed in other ways as well.
For each transplant experiment, approximately 200 adult zebra mussels (<10 mm) were collected from hard substrata at >2 m depth at both sites. These mussels were then held in aerated aquaria for 24-72 h during which time they were measured (shell length, nearest 0.01 mm) and tagged (numbered bee tags attached with cyanoacrylic glue). To obtain an estimate of measurement error for shell length, calliper measurements were repeated 10 times on 6 different individuals ranging in size from 14 to 26 mm in length. Our mean estimate of measurement error was 0.075 mm (STD = 0.048). 100 mussels were then returned to a pair of cages in the source environment as controls with another 100 transplanted into a pair of cages at the destination site as a treatment. Cages (20 x 25 cm) were made of 1-cm2-mesh plastic screen (Vexar™) and anchored on cinder blocks. Three separate reciprocal transplant experiments were conducted between these two sites, two in 2002 and one in 2003. The first ran from July to September 2002 (58 days) and was followed by the second from October 2002 to April 2003 (174 days) with different mussels used for each experiment. In 2003 a single transplant was done from August 2003 to June 2004 with growth measurements taken at three times during the experiment: August to October of 2003 (47 days), October 2003 to April of 2004 (199 days), April to June (64 days). The change in experimental protocol (i.e., from replacement of mussels between different periods to sequential measurements on the same individuals) was based on the results of the first experiments that indicated that there were significant seasonal differences between the two sites in glycogen energy reserves of the mussels (Chapter 2). This change in protocol was an attempt to reduce the potential influence of any source-specific differences in the initial energetic reserves (e.g. lipids, glycogen) of transplanted mussels, especially over the winter period. If left unaccounted for, this source-specific difference in internal mussel condition could potentially distort any comparisons of growth rates between transplant destinations. Additionally a logistic constraint in the transport of mussels between sites meant the two species were not mixed in the same cages during the winter 2002 experiment. Quagga and zebra mussels transplanted from the river site to the estuary (FE) location were placed in a single cage whereas the control zebra mussels, FE source mussels that remained at the FE site, were distributed between two cages (i.e. one cage contained 150 mussels whereas the second held 50 mussels). This deviation in protocol had little effect on the results as mussels at the FE site were lost to spring melt conditions and only certain data could be taken from the cage with the single group of mussels.A second change in the design of the experiment and subsequent statistical analysis was necessitated by unanticipated spring snow melt conditions at the FE site. During the winter of 2002 both replicate cages at the estuary site were anchored too close to the sediment surface and subsequently buried by sediment during the spring melt before they could be retrieved. In the winter of 2003 both cages were then anchored higher above the sediment however they were too close to the surface and subsequently destroyed by surface ice pack. Thus seasonal growth comparisons are limited to animals transplanted to the river site where no such problems occurred
While the use of two cages at each site technically functioned as replicates for site effects, they are not replicates for particular environmental conditions (e.g., high turbidity, low calcium). Our goal, however, was just to compare divergence in the physiological responses of mussels from widely separated and contrasting abiotic environments. Thus, for statistical purposes, the experimental unit was the individual mussel with the two cages being used as a hedge against disturbance or loss rather than as statistical replicates. Finally, while we transplanted quagga mussels (D. bugensis) from the river source to the estuary site, the reciprocal treatment between the estuary source and river site was not possible because of their rarity at the Île d’Orléans site and the estuary in general (Chapter 3).
After retrieval of both summer and winter 2002 experiments, the shell growth (mm.d-1) was estimated as from the change in shell length of surviving mussels. Because the 2003 experiment used only one set of mussels, shells were measured at the end of each season (Oct, Apr, and Jun) on the same mussels.Growth rate varies ontogenetically in mussels, and thus a range of mussel sizes (Table 5.2) was used in the experiments to encompass the natural demographic variability. To limit the potentially confounding effects of the range of mussel size, initial length was used as a covariate in the ANCOVAs (Tables 5.3 and 5.4). Because the difference in distribution of the two species prevented reciprocal transplants for quagga mussels, separate ANCOVAs were conducted for each species:zebra musseldaily growth = initial length * source population * transplant destination versus quagga mussel daily growth = initial length * transplant destination. These model statements allow us to interpret whether variation in growth was the result of population characteristics (source effect), environmental differences (destination effect), or an interaction between the two factors, but do not allow a direct statistical comparison of the two species. Data were checked for normality and homoscedasticity by visual examination for outliers, then checking for randomness using plots of residuals from preliminary ANCOVAs; only the least square (LS) means are presented in all figures and tables. Because the FE site has multiple abiotic factors that have been shown to be significant stressors of dreissenid mussels, we predicted lower growth rates at this site relative to the R location independent of source populations.
For both species mortality rates were less variable in summer (2-15%) than in winter (4-28%) and 5-fold higher in 2003 than in 2002 (Table 5.1 and 5.2). However this latter disparity could be attributable to the differences in experiment design between the two years (i.e., replacement after summer in 2002 vs. successive measurements of the same mussels for all seasons in 2003). Over the two years mortality was highest in spring 2003 (55-61%) and higher among mussels transplanted to the estuarine site in both years (53%, Tables 5.1 and 5.2). As noted earlier initial mussel size was an important covariable for the growth of each species-site combination, but it also had a seasonally dependent component. Daily winter zebra mussel growth at the R site was, however, negatively influenced by length in 2003 (Finitial length 1,78= 5.4, p = 0.02) but not 2002 (Finitial length 1,165= 0.95, p = 0.33).
Substantial temporal variation in growth was observed between both seasons and years. Due to a loss of cages from the estuary site (see above), we limit our statistical comparisons to river source zebra mussels held at the river site. Growth of these river mussels was consistently greater in summer than in winter as little or no growth occurred in winter (Figure 5.2). In addition to the seasonal variation in growth rate, there was also a substantial inter-annual difference: with a four-fold increase in 2003 compared to 2002 (Figure 5.2). At the estuary site (FE), quagga mussels were so rare that only the inter-annual variation in zebra mussel growth could be compared. Zebra mussels at the estuarine site followed the same pattern as observed for zebra mussels at the riverine site, i.e., higher in 2003 than in 2002.
The mortality of river source zebra mussels was different between sites; generally greater than estuary source zebra mussels regardless of where they were transplanted to, except for mussels transplanted from the FE site to the R site in October 2002 which suffered no mortality (Table 5.1 and 5.2). In addition to the effect of season on mortality there was also a pronounced difference related to source of mussels; the summer mortality of river source zebra mussels was low anddiffered little between transplant destinations.In contrast, mortality of estuary zebra mussels was lower at the R site in both years (0 in 2003 relative to 38% at the FE site, Tables 5.1 and 5.2). Summer versus winter comparison of the different zebra mussel source populations (i.e., R vs. FE) at the R site shows that, despite the substantial inter-annual growth differences, river source mussels outgrew the estuary mussels in summer but not winter (Figure 5.5). In spite of the large inter-annual difference in summer growth rates, winter growth rates were largely equivalent between the two years (Figure 5.5).
Reciprocal transplantation showed that the river source zebra mussels consistently out-grew estuary source individuals at both sites in both years (Table 5.3 and Figure 5.3). Growth rate varied also significantly between the two destination sites, but the differences switched between year by site combinations: slightly higher at the R site in 2002 but then becoming much greater at the FE site in 2003 (Figure 5.3).
In contrast to the pattern for zebra mussels, measurable winter growth occurred in quagga mussels. However as observed for zebra mussels, summer growth rates varied between the two years (an 8-fold higher in 2002). The inter-annual variation in summer growth was, however, the opposite of that observed in zebra mussels, being greater in the summer of 2002 compared to 2003 when summer growth was equivalent to winter rates. When compared directly, quagga growth went from being 5-fold greater than zebra mussels in 2002 to a fourth of that of the zebra mussel in 2003 (Figure 5.2). Spring growth, only recorded during 2003, was equivalent to 2002 and 2003 winter rates for both species (Figure 5.2). No consistent pattern of mortality in relation to year by transplant destination was seen for quagga mussels. Unlike river source zebra mussels, river quaggas always performed poorly at the FE site (Table 5.4 and Figure 5.4) and also grew best in the year zebra mussel growth was poorest (see above). Conversely, initial length was a significant negative factor for quagga mussel growth in the winter of 2002 (Finitial length 1,84= 20.48, p <0.01) but not 2003 (Finitial length 1,65= 1.08, p = 0.31).
Two aspects of the environmental physiology of dreissenid mussels are the focus of this study; whether thereis a seasonal component to the fixed geographic differences among zebra mussel populations and whether differences in the physiological traits of individual adults explains any of the patterns of distribution and dominance between them. Growth rate is size dependent (as seen in the significant initial size effects in the ANCOVA) with small (i.e. young) individuals growing faster than older ones. The large difference in quagga growth was seen despite the combination of small difference in initial quagga size between years and that both species were normalized by the ANCOVA procedures. However if this were the case, then we would expect the growth of quagga mussels to be similarly affected which it is not. Instead the growth rate of the quagga mussels also changed dramatically, though this was roughly a four-fold increase in the opposite direction. Thus, the potentially confounding factor of small differences in initial size is unlikely to explain the inter-annual differences in growth that we observed. All of these facts argue that patterns we demonstrate were real and not a size-dependant artefact. Thus the significant site and source specific variation demonstrates that a difference in growth between sites is due to more than environment alone. This also has broader implications for our understanding of the interaction of natural selection and/or acclimation of adult mussels in structuring population genetic/allelic structure.
Substantial variation in growth rate of mussels was observed between seasons, years, sites and species. Such variation is not surprising in for Dreissena in that the seasonal temperature hydrograph and its bioenergetic implications are an important factor in defining annual growth and production of this mussel (Thorp et al. 1998, Allen et al. 1999). For zebra musselsthesefield experiments confirm that both the season of peak of shell growth is earlier and that the maximum growth rate during that period increases latitudinally from north to south: 3-12 μm . day-1in August in New York (this study), 25-35 μm . day-1in June in Indiana (Garton and Johnson 2000), and 30-40 μm . day-1in April in Louisiana (Allen et al. 1999). If temperature alone were the controlling environmental factor, as suggested elsewhere (Thorp et al. 1998, Allen et al. 1999), then this disparity might be an expected response to a latitudinal delay in both the peak rates and period of length of primary production. An alternative explanation has been proposed based on work in the Rhine River where an ontogenetic difference in the season of maximum growth is seen: larger zebra mussels, such as the ones examined in this study, start growing later in the year than smaller ones (Walz 1978, Jantz and Neumann 1998). Thus in addition to the latitudinal difference, ontogenetic differences in either scope for growth or season of growth would potentially confound comparison of the populations. However because the mussels used in the Louisiana and Indiana studies were the same size as ours, differences in annual temperature of the two rivers likely outweigh any demographically related effects. A final alternative presented here would be the effects of any local acclimation or differentiation among water masses in the St. Lawrence River.
In general the greater zebra mussel summer growth rates in 2003 indicate both that abiotic constraints and the mussels’ physiology vary greatly from year to year. The difference in experimental design between years (separate summer and winter transplants in 2002 versus a single transplant carried through both seasons in 2003) does not appear to have been a confounding factor as the winter growth was both very low and similar between years. This inter-annual differentiation in growth illustrates how variable an individual site can be in terms of species-specific growth. A priori, the estuary should have been a poor environment for Dreissena based both on the biology (a freshwater animal) and on the low abundance of the mussels in field collections (Chapter 3), and indeed the negative growth was observed there in 2002. In contrast, growth rate was highest there in 2003, surpassing all other site-source-year combinations. Previous results (Chapter 2) have suggested that mussels found in the estuary, though sparse in abundance, were in better condition than those from the river. Combined with the differences in growth rates, this may be evidence for a stochastic temporal mechanism that could account for the irregular recruitment and abundance of the mussels (Chapter 3) but also higher performance. Regardless, the occurrence of strikingly higher shell growth is estuarine mussels, even if only on occasion, is surprising considering the brackish estuary is a consistently stressful habitat thought to represent the edge of this freshwater mussel’s physiological ability.
Quagga and zebra mussels are closely related species and possess many similarities in morphology and ecology (Mills et al. 1996, McMahon 2002). However, quagga mussels appear to have some key physiological attributes (Baldwin et al. 2002, Stoeckmann 2003) that may contribute to the observed shift in dominance at numerous locations in the Great Lakes (Mills et al. 1999, Ricciardi and Whoriskey 2004, Wilson et al. 2006). This study suggests the emerging dominance of quagga mussels in the river and their continued absence from the fluvial estuary may be related to physiological differences in adults of the two species. Summer was consistently the season of greatest growth for both quagga and zebra mussels, but there were inter-annual reversals in which species had the faster growth rates, demonstrating that at least in some years the quagga mussel can substantially outperform the zebra mussel in riverine environments. The striking reversal between years also emphasizes the danger of generalizing from limited data. Our conclusions would have been entirely different if we had examined this system in just one year. Thus, despite the many similarities between zebra and quagga mussels, the interaction with potentially stochastic environmental variability renders the outcome of competition context-dependent. This conclusion is significant in that it suggests that the success or failure by one invasive or expanding species should not imply that phylogenetically or ecologically related species would do similarly in a given environment. Moreover, any anthropogenic or natural environmental variations (e.g., temperature) may lead to rapid shifts in species dominance. Our results also suggest that periodic inter-annual variation in the riverine conditions that favour the performance advantages of quagga physiology (Baldwin et al. 2002, Stoeckmann 2003) is needed to cause the large shifts species dominance in some parts of the St. Lawrence system (Ricciardi and Whoriskey 2004).
Significant spatial variation in growth rate was also evident. Whereas zebra mussel abundance was low in the estuary, quagga mussels were a rarity. Thus, even though larvae of both species are probably dispersed to the fluvial estuary, there is differential larvae mortality or recruitment for quagga mussels. Transplants of zebra and quagga mussels from the same river site, neither of which had been previously exposed to estuarine conditions, show clear species-specific differences in their response to estuarine conditions: zebra mussels had either low or very high growth rates at the FE site depending on the year. In contrast, quagga mussels performed poorly there in both years. These results support the idea that physiological differences between adults of the species (Baldwin et al. 2002, Stoeckmann 2003, Chapter 3) strongly influence habitat use and geographic distribution. Whether quagga mussels are eventually able to become established in the estuary may depend on either their ability to acclimate (Fong et al. 1995, Dietz et al. 1996, Wilcox and Dietz 1998) or disperse and recruit effectively, but the consistently poor performance of naïve adult quagga mussels in comparison to either naïve or estuarine zebra mussels suggests that zebra mussels are more physiologically tolerant of estuarine conditions.
Dreissena’s successful expansion into a diversity of environments on a new continent poses questions about the role physiological plasticity in the rapid spread of invasive species. To gain a better understanding of the evolutionary response of species in heterogeneous environments, it is important to distinguish between physiological tolerance (plasticity) and physiological differentiation (fixed local adaptations) in relation to changing or novel conditions. Tolerance implies physiological plasticity while divergent population structure involves a rapid local population sub-division due to pre- and/or post-settlement local selection. Yet within the first decade of their arrival in North America, sub-population differences in stress tolerance were apparent (Kilgour and Baker 1994). If not the accumulation of genetic mutations, thenwhat is the process driving rapid development of the geographic differences? Hebert et al. (1989) and others have shown that despite the inferred single introduction event, Dreissena has very high levels of genetic diversity. Hence, there is also high potential for fixed differences to develop via either stochastic or selective processes. In addition to local acclimation and post-settlement selection, the spectrum of possibilities includessimple genetic-stochastic interactions (Lewis et al. 2000), developmental-environmental influences (Haag and Garton 1995, Schneider et al. 2003), or a mixture of the two.While the phenotypic results of these alternatives would appear similar (high geographic variability in phenotypic diversity), understanding the underlying processes would improve prediction of how successfully a future invasive species will adjust to a novel and/or changing environment.
Two of the most successful experimental designs for assigning patterns of phenotypic divergence to environmental, genetic, or a mixture of those influences are the common garden and reciprocal transplants (Schluter 2000). Of these two, only reciprocal transplants have a robust ability to summarize the ultimate effect of multiple environmental physiological constraints.This experimental design also has the added benefit of being able to contrast phenotypic variation due to environment alone with a mix of genetic (source population) and/or environmental influences. The reciprocal transplant approach is also complementary to both genomic and proteomic perspectives because it can help distinguish which cellular/molecular differences actually translate to differences in an organism’s physiology and phenotype. One limitation, however, to the transplant approach is that a single influential environmental factor can rarely be identified, but it is also true that a species is rarely ever confronted by only one constraint in nature.The major drawback in our use of the reciprocal transplantation was that it could not account for any maternal effects or environmental-developmental interactions during larval or post-settlement stages. From the reciprocal transplants we conclude that while the isolation by distance mechanism may be important (sensu Elderkin et al. 2001), rapid and irreversible changes in mussel performance can also be a result of localized natural selection pressure. Thus, the genetic diversity of this species permits the colonization of environmental diverse habitats and may permit rapid range expansion in this species.
In the Dreissena – St. Lawrence River model, the aspect particularly relevant to how population structure in a widespread species responds to environmental heterogeneity is whether local pre- and/or post-settlement selection pressures are strong enough to counter unidirectional gene flow (i.e., downstream pelagic dispersal of larvae) and thus allow local adult phenotypes to develop. Marine analogs to Dreissena, the acorn barnacle (Semibalanus balanoides)and the blue mussel (Mytilus edulis and M. galloprovicialis), have been used to address two different aspects of the population subdivision versus plasticity question, the genome/proteome linkage and source population effects (Levinton 1980). Acorn barnacles reciprocally transplanted between coastal zones and estuaries have shown that animals do better in their natal environment compared to transplants, suggesting local adaptation (Bertness and Gaines 1993). Yet subsequent proteomic and genetic analysis demonstrates that even though spatially divergent environments tend to promote protein polymorphism (Schmidt et al. 2000), the more likely explanation for the differences among transplants is some sort of population-level adjustment to an unpredictable environment such as regulation of protein expression during development as opposed to local genetic adaptation (Brown et al. 2001, Brind’Amour et al. 2002). In contrast to the interest in how local selection regimes may maintain allelic/genetic diversity in Semibalanus (see also Véliz et al. 2004), Mytilus research has examined the relationship between source environment and growth and condition. One pattern detected in many of these Mytilus studies is that initial source differences in protein, carbohydrate and lipid content between populations disappear after acclimation to a new transplant destination thus providing evidence of plasticity (Widdows et al. 1984, Freites et al. 2003). Yet even this clear link between biochemical condition and environment is countered by the equally strong evidence of a population source role in functional attributes such as clearance rate, absorption efficiency, growth rate, mass, and condition (Fuentes et al. 1994, Iglesias et al. 1996, Babarro et al. 2000a,b). The most conservative conclusion about the relative importance of source population or environmental plasticity in phenotypic expression from the Semibalanus, Mytilus, and the present Dreissena study ambivalent. Whereas this may at first seem to be intellectually unsatisfying, it is important to realize that this also constitutes an alternative conclusion: that the effects of population source and phenotypic plasticity are not mutually exclusive in these benthic invertebrates.
Large river ecosystems make a useful model for exploring the mechanism structuring populations because the strong unidirectional gene flow carries multiple geno-/phenotypes into a mosaic of divergent environments. Assessments of Mississippi River zebra mussel population structure based on paired gene sequence (AFLP) and gene expression (LAP allelic diversity) have found evidence of low levels of differentiation (Elderkin et al. 2001, 2004a). Originally, the authors invoked selection as the most likely explanation for the gradual downstream increase in one LAP allele over the others because the high level of potential gene flow (either as pelagically dispersed larvae or as adults attached to shipping) would the counter any geographic isolation among sample sites (Elderkin et al. 2001). The LAP locus is believed to be associated with the strong north-source temperature cline in the Mississippi River (Elderkin et al. 2001), but subsequent studies discovered low heritability for the LAP locus, confounding the authors’ original conclusion that this shift is due to natural selection (Elderkin et al. 2004b). Comparisons of AFLP (genomic) to allozyme (proteomic) results from the same animals are not in agreement making it difficult to reconcile the two different methodological perspectives (Elderkin et al. 2004a).AFLP analysis suggested population structure at regional scale for the mussels whereas LAP profiles implied a downstream genetic shift.Thus while we cannot reach a final conclusion about which mechanism or mixture of mechanisms (selection or isolation by distance) is the root cause or at what level it is occurring at (genetic sequence versus protein expression), these studies in large rivers do demonstrate that population structuring is occurring despite strong gene flow across a widespread but contiguous population.
Unlike the gradual latitudinal cline of the Mississippi, the St. Lawrence has divergent regional-scale environments (i.e., the different water masses) in close proximity, indeed even flowing side by side (Thorp et al. 2005). Larvae from the same gene pool are thus dispersed downstream into the contrasting physiological constraints of these environments (e.g. salinity, food, and hydrology). If spatial differences in physiological traits within this ostensibly contiguous population are associated with local water mass characteristics (Chapter 2 and 3), a plastic response (Ha: Environmental Plasticity, Figure 5.6) to transplantation would be expected. Alternatively if they are from divergent populations, where both are responding to differences in local environment but one population is better able to tolerate the conditions than the other, then (either Ha: simple genetic-environmental interaction or Ha: CoGV respectively, Figure 5.6) would be expected. Our results support localized irreversible co-gradient variation allowing us to conclude that there is at least some level of population structuring. This conclusion of phenotypic sub-division runs counter to predictions that, because of their recent colonization and rapid dispersal, Dreissena populations would not have had time for either selection or mutation mechanisms that promote population differentiation. Because differentiation due to either genetic-mutational mechanisms or local environmental selection is a dynamic process, the differences seen in large river Dreissena populations to date may limited to the level of protein expression and not yet fixed into the genetic architecture.
Table 5.1. Mean initial size and percent mortality of mussels transplanted between the Great Lakes (R) and estuary (FE) water masses during 2002 (“na” indicates mussel transplants destroyed by ice floes during the winter).
|
|
* Transplanted mussels recoveredfor this time period had been smothered by recent sedimentation during spring snowmelt but still contained soft tissue inside the shell. We concluded that the animals had died very recently and thus shell growth measurements were still valid though percent mortality was not.
Table 5.2. Mean initial size and cumulative percent mortality of mussels transplanted between the Great Lakes (R) and estuary (FE) water masses during 2003 (“na” indicates mussel transplants destroyed by ice floes during the winter).
|
|
Table 5.3. ANCOVA comparison of the shell growth of zebra mussels (D. polymorpha) reciprocally transplanted between the river and estuary sites during the summer of 2002 and 2003 (*p<0.05, **p<0.01).
|
|
Table 5.4. ANCOVA comparison of the shell growth of quagga mussels (D. bugensis) transplanted between the river and estuary sites during the summer 2002 and 2003 (*p<0.05, **p<0.01).
|
|
Figure 5.1. Map of the St. Lawrence river-estuary system with the general location of sample sites are indicated.The abbreviation R stands for the site in the Great Lakes water mass near Massena, NY and FE for the fluvial estuary at the eastern tip of Île d’Orléans, QC.
Figure 5.2. Comparison of both D. polymorpha and D. bugensis at the river site emphasizing inter-annual variability.LS means and SE for seasonal (summer, winter, and spring) daily shell growth of a 26 mm zebra mussel (D. polymorpha) from both source populations and a 21 mm quagga mussel (D. bugensis) from the river site at the river site during the summer of 2002 and 2003.
Figure 5.3. Comparison emphasizing source population differentiation for D. polymorpha despite inter-annual variability. LS means and SE for summer daily shell growth of a 21 mm zebra mussel (D. polymorpha) reciprocally transplanted between the river and estuary sites during the summer of 2002 and 2003. Arrows indicate the direction of the transplants.
Figure 5.4.Comparison of inter-annual differences in the response of D. bugensis at the river site but not the fluvial estuary. LS means and SE for shell growth of a 19 mm quagga mussel (D. bugensis) from the river source populations measured at both the river (R) and estuary (FE) site in 2002 and 2003.
Figure 5.5. LS means and SE for seasonal shell growth of a 26 mm the two zebra mussel (D. polymorpha) source populations measured at the river site in both 2002 and 2003.
Figure 5.6. Various hypothesized outcomes/interpretations for this reciprocal transplant experiment including the null model (no differentiation), environmental cline/plasticity, local adjustment adaptation by one source population, or positive covariance/co-gradient variation (CoGV) (sensu Conover & Schultz 1995, Schluter 2000). Circle colors indicate different source populations for mussels (black from the benign environment, white from the stressful environment), x-axis shows transplant destination, and the connecting arrows indicate the direction of the transplant
© Andrew F Casper, 2007
| | | |
| Chapitre IV. Contrasts in shell and soft tissue physiology of dreissenid mussels coincident with shifts in species dominance. |  | Chapitre VI. Conclusion Générale |
