Translator Disclaimer
27 February 2015 Physiology and Biogeography: The Response of European Mussels (Mytilus spp.) to Climate Change
Author Affiliations +

To understand how ecological communities may respond to climate change we have adopted the approach of determining the response of major ecosystem engineers that determine community composition and function. We utilize two approaches, correlative and mechanistic, to understand the current and future distributions of the marine mussels Mytilus edulis Linnaeus, 1758 and M. galloprovincialis Lamarck, 1819 in Europe. Both are dominant space-occupying species that control biodiversity in many coastal ecosystems and are the basis of the largest aquaculture production in Europe. A mechanistic analysis of physiological energetic response to temperature of the two species indicates that M. edulis cannot sustain a positive energy balance for sustained periods when sea surface temperature (SST) is greater than 23 °C, while M. galloprovincialis can maintain a positive energy balance at SST up to 30 °C. There is no difference in energetic response of the two species at cold temperatures (5–10 °C). The upper temperature threshold of positive energy balance in each species corresponds closely to the distribution of SST at their respective southern range limits in Europe. Alternatively, the northern range limit of M. galloprovincialis coincides with areas where winter SST is less than 9 °C, but there is no evidence of an energetic limit to this species at the cold end of its geographic range. Presently there is no mechanistic explanation for the difference between species in their northern range limits; however, as indicated by Random Forest modeling, M. galloprovincialis appears to be limited by cold temperatures during winter, suggesting the hypothesis of failure in reproductive development. These approaches allow for the ability to forecast changes in the distributions of these two species in Europe as SST continues to increase.

It is now well established that organisms are responding to the unprecedented rate of climate warming, including changes in the timing of biological activities (Philippart et al. 2003, Edwards and Richardson 2004, Burrows et al. 2011) and biogeography (Burrows et al. 2011, Hilbish et al. 2012, Jones et al. 2012). To understand current and potential distributions of species, many biologists rely on species distribution models (SDMs) that use spatial environmental data relevant to a species to predict its distribution (Guisan and Thuiller 2005, Elith and Leathwick 2009, Peterson et al. 2011). There are two types of SDMs: correlative and mechanistic, though the two are not mutually exclusive. Correlative SDMs create predictions based on correlations between species distribution records and environmental variables (Robertson et al. 2003, Buckley et al. 2010). These types of models can be generated fairly quickly and can often, particularly for a species for which little is known, provide more information on what specific environmental variables may be important for the distribution of that species (Robertson et al. 2003, Elith and Leathwick 2009). Mechanistic SDMs create predictions based on knowledge of underlying physiological and/or life history responses to environmental variables (Robertson et al. 2003, Kearney and Porter 2009, Buckley et al. 2010, Woodin et al. 2013). Mechanistic SDMs are more time consuming to develop, since the important physiological mechanisms that potentially regulate the species distribution must first be determined (Kearney and Porter 2009), but may provide greater insight into the factors limiting a species' distribution than a purely correlative model.

The current challenge is to move beyond modeling changes in the range limits of single species to develop an understanding of the biological mechanisms responsible for these changes and, importantly, the responses of communities and the functions of ecosystems. We have taken the approach of focusing on the response of key ecosystem engineers that both control the composition and diversity of ecological communities and are primary determinants of the “function” of ecosystems with respect to the goods and services provided to human societies, using both correlative and mechanistic methods. Mechanistic SDMs have been suggested as a more responsive model to climate change and invasive species scenarios (Kearney and Porter 2009) as they are better able to extrapolate to novel environmental conditions due to an understanding of the causative mechanisms behind species distributions (Helmuth et al. 2005, Kearney and Porter 2009, Woodin et al. 2013). However, these two types of approaches are not mutually exclusive and we show that incorporating both may provide a more robust model of a species' complete biogeographic range (Lawler et al. 2006, Elith and Leathwick 2009, Sara et al. 2011).

In this study our focal system is marine mussels in the genus Mytilus across Linnaeus, 1758 a European scale. We selected mussels because they are major determinants of space and, therefore, biodiversity within rocky intertidal marine communities (Suchanek 1992). Also, mussels are farmed commercially, and mussel aquaculture is a multibillion dollar industry (FAO 2012). Mytilus edulis Linnaeus, 1758 and M. galloprovincialis Lamarck, 1819 are sister species that inhabit the coastal waters of Europe. Mytilus edulis is a cold-temperate species, inhabiting northern European waters to the French/Spanish border in the Bay of Biscay (Hilbish et al. 2012). Mytilus galloprovincialis is native to the Mediterranean and is found as far north as the British Isles (Skibinski et al. 1983, McDonald et al. 1991, Seed 1992). The overlap of these species has created a mosaic hybrid zone, with alternating patches of hybrids and pure parental populations from the Bay of Biscay to northern Scotland (Skibinski et al. 1983, Coustau et al. 1991, Bierne et al. 2003, Gosling et al. 2008, Hilbish et al. 2012). We sought to understand what controls the biogeographic distributions of these economically and ecologically important species in Europe, and how their functional performance differs within their ranges. This work provides a basis for forecasting future changes in the distribution of these species as the climate continues to change.


European Mytilus distributions

Mytilus spp. have been well studied for many decades in Europe (Skibinski et al. 1983, Coustau et al. 1991, Gardner and Skibinski 1991, Rawson and Hilbish 1998, Daguin et al. 2001, Bierne et al. 2003), and their distribution in the past decade in various regions has been documented quite thoroughly. Much of this distribution was reported in Hilbish et al. (2012), who sampled 97 locations along the Atlantic coast of Spain and France. However, we developed an even more extensive range database by collating data from several studies (Lionetto et al. 2003, Gosling et al. 2008, Pisanelli et al. 2009, Banni et al. 2011, Kijewski et al. 2011, Hilbish et al. 2012, Wethey unpubl. data, Appendix A) as well as sampling in certain regions that were missing recent data (particularly the United Kingdom). Although these species do hybridize, for the purposes of this study we focused on “pure” species populations (populations that contained > 95% allele frequency of one species) because the physiological studies of Fly and Hilbish (2013), described below, focused on “pure” species rather than hybrids. In 2008 and 2009, we sampled mussel populations at 43 sites in England, Scotland, and Wales that were originally sampled in 1976 and 1977 by Skibinski et al. (1983). To gain higher spatial resolution of the hybrid zone in southwest England, we sampled an additional 24 sites in 2010 that were previously sampled in 1996 by Hilbish et al. (2002). DNA extraction and genotyping at the Glu-5′ locus followed the protocols in Hilbish et al. (2012).


We used two separate, yet complementary, modeling approaches to identify key variables determining the range limits of Mytilus edulis and M. galloprovincialis. The correlative approach was used primarily to classify known sites of “pure” populations of one of the two species, and then identify temperature variables that could be used to differentiate between M. edulis sites and M. galloprovincialis sites. The mechanistic approach used laboratory estimates of scope for growth of the two species to compare regions where scope for growth is estimated to be zero with the geographic limits of the two species.

Correlative approach

To estimate the importance of sea surface temperature (SST) on the distribution of Mytilus spp. in Europe, we analyzed the distribution of “pure” Mytilus populations using Random Forest (RF) modeling, which is becoming increasingly popular in ecological studies (Cutler et al. 2007). This machine learning model generates a summary of many classification trees to determine the best predictor variables for a dataset (Breiman 2001). To build the model, we classified sites from both our field studies and the literature (see Appendix A) as either M. galloprovincialis or M. edulis based on allele frequency with a 95% threshold; sites with ≥ 95% of alleles for a particular species were classified as a population of that species. We used the temperature parameters of yearly mean SST, monthly mean SST (for each month), and seasonal mean SST, as in Hilbish et al. (2012). These parameters were calculated using National Oceanic and Atmospheric Administration (NOAA) Optimum Interpolation ¼ Degree Daily SST (OISST) data derived from Advanced Very High Resolution Radiometer (AVHRR) satellite sensors (Reynolds et al. 2007), acquired from the National Climatic Data Center archive at NetCDF/AVHRR/catalog.html. The SST value for each sample site was estimated by finding the pixel on each SST grid nearest to the sampling location. We determined SST means listed above for each population for the five years prior to each mussel sample date. This time period of five years was chosen to encompass on average the life span of the mussels (Hilbish et al. 2012).

The RF model was built using the randomForest package (Liaw and Wiener 2002) in R version 2.7.1 (R Development Core Team 2010) with 250,000 individual trees, in which one-third of the cases were left out of each tree to be used as an estimate of classification error as the trees were added to the forest. Classification error is the frequency of test cases (those not included in model generation) incorrectly classified by the final model and was used to evaluate the ability of the model to differentiate between M. galloprovincialis and M. edulis sites. The relative importance of each temperature variable was determined for the RF model to identify which variables were most useful for classifying the sites, thus, identifying temperature differences between the ranges of the two species. The importance of the predictor temperature variables was determined using the mean decrease in accuracy calculated during the RF analysis. This metric is calculated by identifying the decrease in the accuracy of the model when the values of a single predictor variable are randomly permuted. Thus, variables with the larger mean decrease in accuracy are more important for data classification.

Mechanistic approach

The physiological energetics of Mytilus edulis and M. galloprovincialis were measured by Fly and Hilbish (2013). They integrated measurements of filtration rate, absorption efficiency, and metabolic rate to derive an estimate of the scope for growth (SFG) (sensu Widdows and Bayne 1971 the SFG measures the energy above maintenance available for growth and reproduction) of mussels at water temperatures ranging from 5 °C to 30 °C. We used their estimate of the critical SST at which SFG = 0 to determine the geographic regions in Europe in which each species can potentially maintain a permanent population. In regions that exceed the critical SST at which SFG = 0, populations of each species are expected to be absent or ephemeral (Sokolova et al. 2012, Woodin et al. 2013). We compared the distribution of M. edulis and M. galloprovincialis to the monthly SST along the European coastline and to the critical temperature at which SFG = 0 (Fly and Hilbish 2013). SFG data were collected from one population of each species (Fly and Hilbish 2013). While local adaptation may be possible, Mytilus populations have the potential for very high gene flow, with dispersal distances of up to 100 km (Gilg and Hilbish 2003). The data presented below suggest that there is not local adaptation occurring in genes relevant to setting Mytilus range limits, since the physiological energetics measured using animals from one population appear to well explain the southern range edge of both M. edulis and M. galloprovincialis. If there were local adaptation occurring, we would expect to see these data fail to explain distribution limits.

Ecological forecasting

Forecasts of future environmental conditions were made using model output from the EURO-CORDEX project (Jacob et al. 2014). This project uses regional circulation models (RCM) to downscale global circulation models (GCM) in order to predict monthly average SST on a 44 km grid. We chose to use the 4.5 Wm-2 (rcp4.5) global energy imbalance scenario of the IPCC 5th Assessment Report, which simulates moderate future warming and is similar to the IPCC 4th Assessment Report scenario A1B. To date, 8 GCMs have been downscaled with one RCM (Appendix B), and the results have been posted on the Earth System Grid ( EURO-CORDEX 44-km data were on a curvilinear grid, so they were interpolated to a 25 km Mercator grid, using the R package “akima” (Akima et al. 2013). Since there is disagreement among the models in terms of the spatial distribution of sea surface temperatures, we made maps of the fraction of models that predicted SST of at least 20 °C in summer (August) and at least 9 °C in winter (February). The temperature of 20 °C in summer is a proxy for the southern geographic limit of M. edulis (Fly and Hilbish 2013) and the temperature of 9 °C is a proxy for the northern geographic limit of M. galloprovincialis (Hilbish et al. 2012), as indicated by the mechanistic and correlative models. We used the median of the model predictions as our estimate of the expected geographic distribution of the Mytilus species over the periods 2046–2050 and 2096–2100. For comparison to the present, we used the daily NOAA OISST product (Reynolds et al. 2007) to make maps of monthly average SST over the period 2002–2009. We then determined for each pixel in the map, the fraction of years (2002–2009) with temperatures higher than 9 °C in winter and 20 °C in summer.


The distribution of Mytilus spp. in Europe ranges from northern Africa, through the Mediterranean, along the Atlantic coastline into Scandinavia and on the British Isles (Fig. 1). Contiguous populations of M. galloprovincialis extend from the Mediterranean north to the border between Spain and France in the Bay of Biscay (Fig. 1). Two disjunct populations of M. galloprovincialis occur in Brittany, France, and in Cornwall, England (Fig. 1). Thus, there are three northern limits of M. galloprovincialis in Europe. The North Sea, Irish Sea, and the eastern English Channel are composed almost exclusively of populations of M. edulis. A disjunct population of M. edulis occurs in a certain area of the Bay of Biscay in France (Fig. 1) where SST is generally cooler than the rest of the bay (Fig. 5A). We plotted these populations with respect to winter and summer SST, using February and August SST, respectively, to see how the distributions were related to the warmest and coldest environmental temperatures these populations experienced (Fig. 2). Populations of M. galloprovincialis in Europe never occur in areas where mean monthly SST falls below 9 °C in the winter, while populations of M. edulis occur in areas down to 6 °C (Fig. 2). In the summer, most populations of M. edulis occur where summer temperatures are below 20 °C, and no populations persist where summer SST is greater than 23 °C. Conversely, populations of M. galloprovincialis inhabit much warmer locations (Figure 2) including sites where summer SST regularly exceeds 23 °C. The warmest average summer temperature experienced by M. galloprovincialis in our dataset was 26.4 °C; however, the distribution of this species in northern Africa is not well-documented, so populations likely experience even warmer temperatures.

Figure 1.

Sample locations used for biogeographic modeling of Mytilus spp. in Europe. These data were collated from several published studies (see Appendix A) as well as additional sampling in certain regions that were missing data (in particular the United Kingdom). For the purposes of this study we focused on “pure” species populations, and we defined “pure” as populations that contained > 95% genotype of either M. edulis or M. galloprovincialis. Although the data points are sparse in the Mediterranean and North Sea, these areas are nearly exclusively M. galloprovincialis and M. edulis, respectively. The current southern limit of M. edulis occurs at the France/Spain border in the Bay of Biscay, while the current northern limit of M. galloprovincialis is on France's Brittany coast on Europe's mainland and in southwest England.


The Random Forest classification model, a correlative model, performed well, correctly classifying the out of bag samples (30% of data excluded when building the model) 97.14% of the time. Relative importance evaluation identified winter and spring variables (mean winter temperature, mean spring temperature, mean January temperature, mean March temperature, and mean April temperature) as the most important variables (ranked by mean decrease in model accuracy) for correctly classifying a site as Mytilus edulis or M. galloprovincialis. These results align nicely with a smaller-scale RF model based on M. galloprovincialis distribution in France and Spain, and explain historical changes in the distribution of M. galloprovincialis along the Normandy coast in France (Hilbish et al. 2012). Based on this correlative model and the known distribution of M. galloprovincialis with respect to SST, M. galloprovincialis is likely precluded from inhabiting areas where winter SST is routinely below 9 °C, while M. edulis is more cold-adapted (Fig. 2). However, this correlative approach provides no explanation for the mechanism involved in this northern range limit of M. galloprovincialis. Previous work discerning the energetics of these two species (Fly and Hilbish 2013) provides evidence that the mechanism is not adult mussel energetics, as there are no significant differences at cold temperatures (Fig. 3).

Differences in energetics can, however, be ascertained as the mechanism for the differences in the species' southern limits. As stated previously, populations of Mytilus edulis do not inhabit areas where summer temperatures routinely exceed ∼23 °C (Fig. 2). Fly and Hilbish (2013) determined the critical temperature at which SFG = 0 for M. edulis and M. galloprovincialis. While there were seasonal differences, the critical temperature for M. edulis was regularly around 23 °C (Fig. 3, Fly and Hilbish 2013), which corresponds with the environmental temperatures at the southern range edge of this species (Fig. 2, Fig. 4). The critical temperature for M. galloprovincialis was between 29–30 °C (Fig. 3, Fly and Hilbish 2013), and M. galloprovincialis inhabits regions in Europe with SST that does not exceed this threshold (Fig. 2, Fig. 4).

Figure 2.

Distribution of European Mytilus populations with respect to winter and summer sea surface temperatures (SST). We plotted the populations shown in Figure 1 with respect to winter and summer SST, using A, February and B, August SST, respectively, to see how the distributions were related to environmental temperatures. A, Populations of M. galloprovincialis never occur in areas where mean monthly SST falls below 9 °C in the winter, while populations of M. edulis occur in areas down to 6 °C. B, In the summer, most populations of M. edulis occur where summer temperatures are below 20 °C, and no populations persist where summer SST is greater than 23 °C. Conversely, populations of M. galloprovincialis inhabit much warmer locations.


In the past decade, the 9 °C winter isotherm has been in the western part of the English Channel in the south (Fig. 5A). During the same period, the 20 °C summer isotherm has traversed the Bay of Biscay from just south of the mouth of the Gironde (45 °N) to NW Spain (Fig. 5B). This pattern is consistent with the exclusion of Mytilus edulis from the lower Bay of Biscay (Fig. 1). The EURO-CORDEX forecast predicts both winter and summer warming (Fig. 5C–F). The result of this warming is the northward movement of the conditions for survival of M. edulis and M. galloprovincialis. The median of the EURO-CORDEX rcp4.5 forecasts predicts that summer maximum monthly SST of 20 °C will move northward to just south of the Brittany peninsula in France by 2050 (Fig. 5D), and move into the western English Channel and southern North Sea to 53°N by 2100 (Fig. 5F). This 20 °C isotherm is a proxy for the expected southern geographic limit of M. edulis, and predicts that the only region of the English Channel that will be habitable by M. edulis by 2100 will be along the Normandy coast of France (between Cherbourg and Calais). While a similar permissible area appears on England's side of the English Channel (between Bournemouth and Dover), mussels are not readily found in this area due to a predominance of sandy beaches rather than hard substrates. The median of the rcp4.5 forecasts suggest that winter monthly SST higher than 9 °C will move eastward into the English Channel and northward into the Irish Sea by 2050 (Fig. 5C), with only slightly further movement into the northern North Sea by 2100 (Fig. 5E). This prediction suggests that M. galloprovincialis will be much more prevalent in the British Isles and English Channel than it currently is.

Figure 3.

Physiological energetics of Mytilus edulis and M. galloprovincialis. Measurements of filtration rate, absorption efficiency and metabolic rate were used to derive an estimate of scope for growth (SFG) (sensu Widdows and Bayne 1971 the SFG measures the energy above maintenance available for growth and reproduction) of mussels at water temperatures ranging from 5 °C to 30 °C in A, spring, B, summer, and C, fall. No mussels survived at 30 °C in the fall. Estimates of the critical temperatures at which SFG = 0 (dashed line) were used to determine the geographic regions in Europe in which each species can potentially maintain a permanent population. This figure was adapted from Fly and Hilbish (2013).


Figure 4.

Average temperature profile of European Mytilus edulis (gray) and M. galloprovincialis (black) populations in each season. The transparent rectangles (gray for M. edulis and black for M. galloprovincialis) indicate the critical temperature at which SFG = 0 in spring, summer, and fall, as determined by Fly and Hilbish (2013). SFG data were not collected for winter (A). In regions that exceed this critical temperature, populations of mussels are expected to be ephemeral or absent and, indeed, very few populations experience temperatures higher than that critical temperature. Summer (C) and fall (D) appear to be more physiologically-limiting than spring (B). This figure was adapted from Fly and Hilbish (2013).



Species distribution models are useful in understanding current and potential distributions of species; however, oftentimes one type of SDM is incapable of accurately predicting all of a species' range limits (e.g., north and south, east and west) (Woodin et al. 2013). This is because a given SDM applies a single mechanism or environmental profile across an entire species' range, when, it is often the case that different mechanisms or stressors control different portions of a species' geographic range or its distribution in different geographic regions (Sokolova et al. 2012, Woodin et al. 2013). This appears to be the case with Mytilus edulis and M. galloprovincialis. As suggested by mechanistic modeling, the southern limit of M. edulis is likely controlled by physiological constraints on the adult populations: specifically, populations cannot persist in areas where summer SST routinely exceeds 20 °C, due to rapidly declining SFG above this temperature. It is likely that the southern limit of M. galloprovincialis is also controlled by energetic limitation; however, the southern limit of this species on the northwest African coast is not well-documented, so we cannot currently determine temperatures experienced at the very southern range edge of this species. Interestingly, while the southern limits of Mytilus spp. in Europe appear to be controlled by sublethal temperature effects of adult energetics, the southern limit of M. edulis on the east coast of the U.S. is controlled by lethal temperature effects on adult populations (Jones et al. 2010). This is likely because the east coast of the U.S. has a more continental climate (and, therefore, greater temperature extremes) than the west coast of Europe, which has a more coastal climate (Wethey 1985, Jenkins et al. 2008, Bailey 2009). Thus, mechanisms can differ within the same species in different regions.

The mechanism limiting the northern range of Mytilus galloprovincialis in Europe is still unknown; however, limitations due to adult energetics are very unlikely. Blue mussels are physiologically quiescent during winter (Bayne 1976), and there are no major differences in the SFG of M. edulis and M. galloprovincialis at cold temperatures in other seasons that would explain the inability of M. galloprovincialis populations to persist where winter temperatures are below 9 °C (Fly and Hilbish 2013). This cold-water limitation has been noted previously for European M. galloprovincialis populations (Hilbish et al. 2012), as well as populations on the west coast of the U.S. (Hilbish et al. 2010). There are several hypotheses on the mechanism of this limitation that warrant further investigation, including tolerance to cold-shock and reproductive abilities.

The tolerance of species to cold temperatures depends on a variety of factors, including the thermal regimes and acclimation times experienced (Renault et al. 2004, Terblanche et al. 2011). Jansen et al. (2007) found that Mytilus galloprovincialis populations were much more sensitive to cold shock than M. edulis populations. This suggests that M. galloprovincialis might not have the same capability to recover from cumulative cold damage as M. edulis. Another possibility is that the reproductive capabilities of M. galloprovincialis are impeded at temperatures below 9 °C (Hilbish et al. 2010, Hilbish et al. 2012). Most populations of mussels spawn in the spring after having undergone gametogenesis over the winter (Bayne 1976). The fact that the RF model indicated winter or early spring as a key time for determining the distribution of these mussels suggests a hypothesis that some aspect of gametogenesis and/or spawning is temperature dependent and different between species.

Figure 5.

A, Fraction of winters 2002–2009 with temperatures above 9 °C in winter (February) and B, above 20 °C in summer (August), derived from NOAA Optimally Interpolated Sea Surface Temperature data. The contour line is the 50th percentile of the values throughout the figure. C, Fraction of EURO-CORDEX rcp4.5 models indicating temperatures above 9 °C in winter (February) and D, 20 °C in summer (August) in the period 2046–2050. E, Fraction of EURO-CORDEX rcp4.5 models indicating temperatures above 9 °C in winter (February) and F, 20 °C in summer (August) in the period 2096–2100. The temperature of 20 °C in summer is a proxy for the southern geographic limit of Mytilus edulis and the temperature of 9 °C is a proxy for the northern geographic limit of M. galloprovincialis. The contour lines for the forecasts (C-F) indicate the median of the model predictions. (Color shown in electronic version only).


Understanding the mechanism controlling this northern limit will be useful in making better predictions for the expansion of these populations as winter SSTs warm. The RF correlative model used monthly mean SST, while the mechanistic model considered distributions on a daily scale. Hilbish et al. (2012) ran a similar model using maximum entropy modeling (Maxent,, that considered the number of days below temperature thresholds as variables. The two variables contributing most to mussel distribution were days below 9 °C and days below 10 °C (Hilbish et al. 2012). As hypotheses for mechanisms explaining this temperature distribution are examined, the temporal scale of the temperature distribution should be considered.

Warming of the ocean is not expected to occur uniformly (Xie et al. 2010), which suggests that the northern range limit of Mytilus galloprovincialis and the southern range limit of M. edulis will likely change at heterogeneous rates over the remainder of the 21st century. During the first half of the current century (present to 2050) the northern range limit of M. galloprovincialis is expected to shift substantially into the English Channel and into the Irish Sea and North Atlantic Ocean (Fig. 5C). However, during the second half of the century further range expansion is expected to be comparatively small (Fig. 5E). Conversely, the anticipated contraction of the southern range edge of M. edulis is expected to remain south of France's Brittany coast in the Bay of Biscay through 2050 (Fig. 5D), but shift dramatically north to 53°N by 2100, leaving only a small portion of French coastline, mainly the Normandy coast, suitable for M. edulis (Fig. 5F). France produces about 73,000 tons of mussels each year, 22% of which are produced in Normandy (Eurostat 2009). These potential future SST shifts indicate the bulk of the commercial mussel farms on the rest of the French coastline may no longer be able to cultivate M. edulis by the end of the century.

Mytilus edulis and M. galloprovincialis hybridize extensively, and the differing rates in their range expansions and contractions have implications for the hybrid zones formed between them. With a rapid northward expansion of M. galloprovincialis and minimal contraction of M. edulis by 2050, the hybrid zones between these two species could expand substantially. However, by 2099, many current regions of hybridization should be displaced by M. galloprovincialis and new hybrid zones may form in the Irish and North Atlantic Seas, regions currently inhabited primarily by M. edulis. This brings to light the vulnerability of M. edulis to warming environmental conditions: those populations exposed to summer SSTs greater than 20 °C are highly vulnerable to being unable to persist.

Although we understand an important mechanism controlling the southern range edge of Mytilus edulis in Europe, we currently have still used 20 °C as our proxy for the physiological limit of this species. While the temperature at which SFG = 0 for M. edulis is ∼23 °C, the southern range edge of this species maps more closely to a 20 °C SST isotherm, with only several populations persisting where summer temperatures range between 20–23 °C. It is likely that the populations living in this summer temperature range are physiologically stressed (Sokolova et al. 2012) and heavily dependent on other factors such as food availability. We should be able to develop an even stronger model to predict not only the range edge of this species with changing SST based on where SFG = 0, but to begin to quantify the productivity of these mussel beds based on the physiological rates measured (Fly and Hilbish 2013). These mechanistic models, as stated earlier, require much more input and background knowledge than correlative models. Thus, several issues must be resolved to produce a meaningful mechanistic model based on physiological productivity. A more detailed model will better take into account any differences between submerged and intertidal mussels. Mytilus edulis and M. galloprovincialis reduce their aerial rate of oxygen consumption to 4–17% that of their rate of aquatic oxygen consumption (Widdows et al. 1979), and incur only a small cost in terms of oxygen debt recovery (Widdows and Shick 1985). However, intertidal mussels can only feed while submerged, thus, affecting food availability and energy intake. A robust mechanistic model will take into account food availability and duration of exposure at low tide conditions.

The most outstanding question is how to incorporate food availability into projections of secondary production. Mussels consume food particles in the size range of 2–20 µm (Bayne 1976), but algal biomass within this size range is only, at best, weakly correlated with chlorophyll a or other pigments that can be assessed with remote sensing (Alpine and Cloern 1985, Han and Furuya 2000, Arin et al. 2002). Additionally remote sensing products are often unreliable in near-shore coastal environments where materials of terrestrial origin and reflection off of the ocean bottom interfere with sensor readings (Hellweger et al. 2004, Moses et al. 2009). The result is that it is very difficult to predict present day secondary production of mussels (and many other marine species) on a regional scale. It will be even more difficult to forecast changes in primary production, specifically of the nanoplankton as a consequence of climate change and to incorporate these changes into models of secondary production of mussels and other coastal species (Sommer and Lengfellner 2008). Until meaningful projections of changes in coastal primary production become available the most productive approach may be to forecast changes in local and regional potential for secondary production and provide mechanistic models that will allow the incorporation of local primary production to assess the management of coastal species with respect to climate change.


This study was supported by grants from NSF (OCE1039513, OCE1129401), and NASA (NNX07AF20G, NNX11AP77G). We acknowledge the World Climate Research Programme's Working Group on Regional Climate, and the Working Group on Coupled Modelling, former coordinating body of CORDEX and responsible panel for CMIP5. We also thank the climate modelling groups (listed in Appendix B of this paper) for producing and making available their model output. We also acknowledge the Earth System Grid Federation infrastructure, an international effort led by the U.S. Department of Energy's Program for Climate Model Diagnosis and Intercomparison, the European Network for Earth System Modelling and other partners in the Global Organisation for Earth System Science Portals (GO-ESSP). P. M. Brannock, A. O. Hollis, and S. J. Jones assisted with field and genetic work. We are grateful to P. B. Marko and B. A. Seibel, the American Malacological Society, and the University of the Azores for their organization of the World Congress of Malacology 2013 conference and the symposium on “Climate Change and Molluscan Ecophysiology.” This is publication number 73 in Ecological Forecasting from the University of South Carolina.


  1. H. Akima , A. Gebhardt , and T. Petzoldt . 2013. Akima: Interpolation of irregularly spaced data. R. package version 0.5–10. Internet resource: Scholar

  2. A. E. Alpine and J. E. Cloern . 1985. Differences in in vivo fluorescence yield between three phytoplankton size classes. Journal of Plankton Research 7: 381–390. Google Scholar

  3. L. Arin , X. A. G. Moran , and M. Estrada . 2002. Phytoplankton size distribution and growth rates in the Alboran Sea (SW Mediterranean): Short term variability related to mesoscale hydrodynamics. Journal of Plankton Research 24: 1019–1033. Google Scholar

  4. R. G. Bailey 2009. Ecoregions of the United States. In R. G. Bailey , ed., Ecosystem Geography: From Ecoregions to Sites. 2nd edition. Springer-Verlag. New York, New York. Google Scholar

  5. M. Banni , A. Negri , F. Mignone , H. Boussetta , A. Viarengo , and F. Dondero . 2011. Gene expression rhythms in the mussel Mytilus galloprovincialis (Lam.) across an annual cycle. PLoS ONE 6: e18904. Google Scholar

  6. B. L. Bayne 1976. Marine Mussels: Their Ecology and Physiology. Cambridge University Press, Cambridge, United Kingdom. Google Scholar

  7. N. Bierne , P. Borsa , C. Daguin , D. Jollivet , F. Viard , F. Bonhomme , and P. David . 2003. Introgression patterns in the mosaic hybrid zone between Mytilus edulis and M. galloprovincialis. Molecular Ecology 12: 447–461. Google Scholar

  8. L. Breiman 2001. Random forests. Machine Learning 45: 5–32. Google Scholar

  9. L. B. Buckley , M. C. Urban , M. J. Angilletta , L. G. Crozier , L. J. Rissler , and M. W. Sears . 2010. Can mechanism inform species' distribution models? Ecology Letters 13: 1041–1054. Google Scholar

  10. M, T. Burrows , D. S. Schoeman , L. B. Buckley , P. Moore , E. S. Poloczanska , K. M. Brander , C. Brown , J. F. Bruno , C. M. Duarte , B. S. Halpern , J. Holding , C. V. Kappel , W. Kiessling , M. I. O'Connor , J. M. Pandolfi , C. Parmesan , F. B. Schwing , W. J. Sydeman , and A. J. Richardson . 2011. The pace of shifting climate in marine and terrestrial ecosystems. Science 334: 652–655. Google Scholar

  11. C. Coustau , F. Renaud , and B. Delay . 1991. Genetic characterization of the hybridization between Mytilus edulis and M. galloprovincialis on the Atlantic coast of France. Marine Biology 111: 87–93. Google Scholar

  12. D. R. Cutler , T. C. Edwards , K. H. Beard , A. Cutler , K. T. Hess , J. Gibson , and J. J. Lawler . 2007. Random forests for classification in ecology. Ecology 88: 2783–2792. Google Scholar

  13. C. Daguin , F. Bonhomme , and P. Borsa . 2001. The zone of sympatry and hybridization of Mytilus edulis and M. galloprovincialis, as described by intron length polymorphism at locus mac-1. Heredity 86: 342–354. Google Scholar

  14. M. Edwards and A. J. Richardson . 2004. Impact of climate change on marine pelagic phenology and trophic mismatch. Nature 430: 881–884. Google Scholar

  15. J. Elith and J. R. Leathwick . 2009. Species distribution models: Ecological explanation and prediction across space and time. Annual Review of Ecology, Evolution, and Systematics 40: 677–697. Google Scholar

  16. Eurostat, 2009. Eurostat statistics in focus. Agriculture and Fisheries, 83/2009. Google Scholar

  17. FAO. 2012. Cultured Aquatic Species Information Programme. Mytilus edulis. Text by P. Goulletquer In: FAO Fisheries and Aquaculture Department [online]. Rome, Italy, 6 February 2012. Google Scholar

  18. E. K. Fly and T. J. Hilbish . 2013. Physiological energetics and biogeographic range limits of three congeneric mussel species. Oecologia 172: 35–46. Google Scholar

  19. J. P. A. Gardner and D. O. F. Skibinski . 1991. Biological and physical factors influencing genotype-dependent mortality in hybrid mussel populations. Marine Ecology Progress Series 71: 235–243. Google Scholar

  20. M. R. Gilg and T. J. Hilbish . 2003. The geography of marine larval dispersal: Coupling genetics with fine-scale physical oceanography. Ecology 84: 2989–2998. Google Scholar

  21. E. Gosling , S. Doherty , and N. Howley . 2008. Genetic characterization of hybrid mussel (Mytilus) populations on Irish coasts. Journal of the Marine Biological Association of the UK 88: 341–346. Google Scholar

  22. A. Guisan and W. Thuiller . 2005. Predicting species distribution: Offering more than simple habitat models. Ecology Letters 8: 993–1009. Google Scholar

  23. M.-S. Han and K. Furuya . 2000. Size and species-specific primary productivity and community structure of phytoplankton in Tokyo Bay. Journal of Plankton Research 22: 1221–1235. Google Scholar

  24. F. L. Hellweger , P. Schlosser , U. Lall , and J. K. Weissel . 2004. Use of satellite imagery for water quality studies in New York Harbor. Estuarine, Coastal and Shelf Science 61: 437–448. Google Scholar

  25. B. Helmuth , J. G. Kingsolver , and E. Carrington . 2005. Biophysics, physiological ecology, and climate change: Does mechanism matter? Annual Review of Physiology 67: 177–201. Google Scholar

  26. T. J. Hilbish , P. M. Brannock , K. R. Jones , A. B. Smith , B. N. Bullock , and D. S. Wethey . 2010. Historical changes in the distributions of invasive and endemic marine invertebrates are contrary to global warming predictions: The effects of decadal climate oscillations. Journal of Biogeography 37: 423–431. Google Scholar

  27. T. J. Hilbish , E. W. Carson , J. R. Plante , L. A. Weaver , and M. R. Gilg . 2002. Distribution of Mytilus edulis, M. galloprovincialis and their hybrids in open-coast populations of mussel in southwestern England. Marine Biology 140: 137–142. Google Scholar

  28. T. J. Hilbish , F. P. Lima , P. M. Brannock , E. K. Fly , R. L. Rognstad , and D. S. Wethey 2012. Change and stasis in marine hybrid zones in response to climate warming. Journal of Biogeography 39:676–687. Google Scholar

  29. D. Jacob , J. Petersen , B. Eggert , A. Alias , O. B. Christensen , L. M. Bouwer , A. Braun , A. Colette , M. Déqué , G. Georgievski , E. Georgopoulou , A. Gobiet , L. Menut , G. Nikulin , A. Haensler , N. Hempelmann , C. Jones , K. Keuler , S. Kovats , N. Kröoner , S. Kotlarski , A. Kriegsmann , E. Martin , E. van Meijgaard , C. Moseley , S. Pfeifer , S. Preuschmann , C. Radermacher , K. Radtke , D. Rechid , M. Rounsevell , P. Samuelsson , S. Somot , J.-F. Soussana , C. Teichmann , R. Valentini , R. Vautard , B. Weber , and P. Yiou . 2014. EURO-CORDEX: New high-resolution climate change projections for European impact research. Regional Environmental Change 14: 563–578. Google Scholar

  30. J. M. Jansen , S. W. Bonga , and H. Hummel . 2007. Differential coldshock resistance among acclimated European mussel populations. Marine and Freshwater Behaviour and Physiology 40: 233–245. Google Scholar

  31. S. R. Jenkins , P. Moore , M. T. Burrows , D. J. Garbary , S. J. Hawkins , A. Ingólffson , K. P. Sebens , P. V. R. Snelgrove , D. S. Wethey , and S. A. Woodin . 2008. Comparative ecology of North Atlantic shores: Do differences in players matter for process? Ecology 89: S3–S23. Google Scholar

  32. S. J. Jones , F. P. Lima , and D. S. Wethey . 2010. Rising environmental temperatures and biogeography: Poleward range contraction of the blue mussel, Mytilus edulis L., in the western Atlantic. Journal of Biogeography 37: 2243–2259. Google Scholar

  33. S. J. Jones , A. J. Southward , and D. S. Wethey . 2012. Climate change and historical biogeography of the barnacle Semibalanus balanoides. Global Ecology and Biogeography 21: 716–724. Google Scholar

  34. M. Kearney and W. Porter . 2009. Mechanistic niche modeling: Combining physiological and spatial data to predict species' ranges. Ecology Letters 12: 334–350. Google Scholar

  35. T. Kijewski , B. Śmietanka , M. Zbawkicka , E. Gosling , H. Hummel , and R. Wenne . 2011. Distribution of Mytilus taxa in European coastal areas as inferred from molecular markers. Journal of Sea Research 65: 224–234. Google Scholar

  36. J. J. Lawler , D. White , R. P. Neilson , and A. R. Blaustein . 2006. Predicting climate-induced range shifts: Model differences and model reliability. Global Change Biology 12: 1568–1584. Google Scholar

  37. A. Liaw and M. Wiener . 2002. Classification and regression by random Forest. R News 2/3: 18–22. Google Scholar

  38. M. G. Lionetto , R. Caricato , M. E. Giordano , M. F. Pascariello , L. Marinosci , and T. Schettino . 2003. Integrated use of biomarkers (acetylcholinesterase and antioxidant enzymes activities) in Mytilus galloprovincialis and Mullus barbatus in an Italian coastal marine area. Marine Pollution Bulletin 46: 324–330. Google Scholar

  39. J. H. McDonald , R. Seed , and R. K. Koehn . 1991. Allozymes and morphometric characters of three species of Mytilus in the Northern and Southern Hemispheres. Marine Biology 111: 323–333. Google Scholar

  40. W. J. Moses , A. A. Gitelson , S. Berdnikov , and V. Povazhnyy . 2009. Estimation of chlorophyll-a concentration in case II waters using MODIS and MERIS data — successes and challenges. Environmental Research Letters 4: 045005, doi:10.1088/1748– 9326/4/4/045005. Google Scholar

  41. A. T. Peterson , J. Soberón , R. G. Pearson , R. P. Anderson , E. MartínezMeyer , M. Nakamura , and M. B. Araújo . 2011. Ecological Niches and Geographic Distributions. Princeton University Press, Princeton, New Jersey, U.S.A. Google Scholar

  42. C. J. M. Philippart , H. M. van Aken , J. J. Beukema , O. G. Bos , G. C. Cadée , and R. Dekker . 2003. Climate-related changes in recruitment of the bivalve Macoma balthica. Limnology and Oceanography 48: 2171–2185. Google Scholar

  43. B. Pisanelli , M. Benedetti , D. Fattorini , and F. Regoli . 2009. Seasonal and inter-annual variability of DNA integrity in mussels Mytilus galloprovincialis: A possible role for natural fluctuations of trace metal concentrations and oxidative biomarkers. Chemosphere 77: 1551–1557. Google Scholar

  44. R Development Core Team . 2010. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Google Scholar

  45. P. D. Rawson and T. J. Hilbish . 1998. Asymmetric introgression of mitochondrial DNA among European populations of blue mussels (Mytilus spp.). Evolution 52: 100–108. Google Scholar

  46. D. Renault , O. Nedved , F. Hervant , and P. Vernon . 2004. The importance of fluctuating thermal regimes for repairing chill injuries in the tropical beetle Alphitobius diaperinus (Coleoptera: Tenebrionidae) during exposure to low temperature. Physiological Entomology 29: 139–145. Google Scholar

  47. R. W. Reynolds , T. M. Smith , C. Liu , D. B. Chelton , K. S. Casey , and M. G. Schlax . 2007. Daily high-resolution-blended analysis for sea surface temperature. Journal of Climate 20: 5473–5496. Google Scholar

  48. M. P. Roberston , C. I. Peter , M. H. Villet , and B. S. Ripley . 2003. Comparing models for predicting species’ potential distributions: a case study using correlative and mechanistic predictive modeling techniques. Ecological Modelling 164: 153–167. Google Scholar

  49. G. Sará , M. Kearney , and B. Helmuth . 2011. Combining heat transfer and energy budget models to predict thermal stress in Mediterranean intertidal mussels. Chemistry and Ecology 27: 135–145. Google Scholar

  50. R. Seed 1992. Systematics evolution and distribution of mussels belonging to the genus Mytilus: An overview. American Malaco logical Bulletin 9: 123–137. Google Scholar

  51. D. O. F. Skibinski , J. A. Beardmore , and T. F. Cross . 1983. Aspects of the population genetics of Mytilus (Mytilidae; Mollusca) in the British Isles. Biological Journal of the Linnean Society 19: 137–183. Google Scholar

  52. I. M. Sokolova , M. Frederich , R. Bagwe , G. Lannig , and A. A. Sukhotin . 2012. Energy homeostasis as an integrative tool for assessing limits of environmental stress tolerance in aquatic invertebrates. Marine Environmental Research 79: 1–15. Google Scholar

  53. U. Sommer and K. Lengfellner . 2008. Climate change and the timing, magnitude, and composition of the phytoplankton spring bloom. Global Change Biology 14: 1199–1208. Google Scholar

  54. T. H. Suchanek 1992. Extreme biodiversity in the marine environment: Mussel bed communities of Mytilus californianus. Northwest Environmental Journal 8: 150–152. Google Scholar

  55. J. S. Terblanche , A. A. Hoffmann , K. A. Mitchell , L. Rako , P. C. le Roux , and S. L. Chown . 2011. Ecologically relevant measures of tolerance to potentially lethal temperatures. Journal of Experimental Biology 214: 3713–3725. Google Scholar

  56. D. S. Wethey 1985. Catastrophe, extinction and species diversity: A rocky intertidal example. Ecology 66: 445–456. Google Scholar

  57. J. Widdows and B. L. Bayne . 1971. Temperature acclimation of Mytilus edulis with reference to its energy budget. Journal of the Marine Biological Association of the UK 51: 827–843. Google Scholar

  58. J. Widdows , P. Fieth , and C. M. Worrall . 1979. Relationships between seston, available food and feeding activity in the common mussel Mytilus edulis. Marine Biology 50: 195–207. Google Scholar

  59. J. Widdows and J. M. Shick . 1985. Physiological responses of Mytilus edulis and Cardium edule to aerial exposure. Marine Biology 85: 217–232. Google Scholar

  60. S. A. Woodin , T. J. Hilbish , B. Helmuth , S. J. Jones , and D. S. Wethey . 2013. Climate change, species distribution models, and physiological performance metrics: Predicting when biogeographic models are likely to fail. Ecology and Evolution 3: 3334–3346. Google Scholar

  61. S.-P. Xie , C. Deser , G. A. Vecchi , J. Ma , H. Teng , and A. T. Wittenberg , 2010. Global warming pattern formation: Sea surface temperature and rainfall. Journal of Climate 23: 966–986. Google Scholar


Appendix A

Data used for Europe-wide Random Forest model using distributions of “pure” populations of Mytilus galloprovincialis and M. edulis based on published accounts or genetic analyses done for this study where M. galloprovincialis allele frequencies were greater than 95% or less than 5%, respectively. The sample sites are listed by latitude from North to South.





Combinations of global and regional circulation models used in ensemble predictions of ocean climate change. C4I, Community Climate Change Consortium for Ireland, CNRM, Météo France, DMI, Danish Meteorological Institute, ETHZ, Swiss Institute of Technology Zurich, GKSS, Helmholz Center Geesthacht Institute for Coastal Research, ICTP, KNMI, Royal Netherlands Meteorological Institute, HC, UK Met Office Hadley Centre, MPI, Max Planck Institute, OURANOS, Consortium on Regional Climatology and Adaptation to Climate Change, SMHI, Swedish Meteorological and Hydrological Institute, UCLM, Universidad Castilla, La Mancha, VMGO, Voeikov Main Geophysical Observatory.

Elizabeth K. Fly, Thomas J. Hilbish, David S. Wethey, and Rhiannon L. Rognstad "Physiology and Biogeography: The Response of European Mussels (Mytilus spp.) to Climate Change," American Malacological Bulletin 33(1), 136-149, (27 February 2015).
Received: 10 December 2013; Accepted: 1 August 2014; Published: 27 February 2015

Back to Top