Differential reproductive investment, attachment strength and mortality of invasive and indigenous mussels across heterogeneous environments
- First Online:
- Cite this article as:
- Nicastro, K.R., Zardi, G.I. & McQuaid, C.D. Biol Invasions (2010) 12: 2165. doi:10.1007/s10530-009-9619-9
- 128 Views
Environmental heterogeneity challenges both indigenous species and invaders and can play a defining role in the dynamics of their interactions. We compare bay and open coast habitats to show how environmental heterogeneity and seasonality affect survival and physiological performances of invasive (Mytilus galloprovincialis) and indigenous (Perna perna) intertidal mussels. P. perna had significantly higher attachment strength than M. galloprovincialis. Attachment was strongly correlated with hydrodynamic stress and was lower for both species in bays. Both species had a major spawning event when wave action was weakest. In bays, there was no correlation between gonad index (GI) and attachment strength for either species, but on the open coast GI was negatively correlated with attachment. In bays, maximum GI of M. galloprovincialis was 64% higher than for P. perna, while on the open coast values did not differ between the two. Thus, on the open coast, both species invest more energy in attachment but P. perna can accommodate energetic demands of increased byssal production without altering gonad production, while M. galloprovincialis cannot. Mortality was significantly correlated to sand stress, while the correlation with wave action was very weak in bays and non-significant on the open coast probably because sand stress peaked during periods of low wave action. The success of the invader and thus the outcomes of its interaction with the indigenous species are governed by habitat-to-habitat variability. In this case the invasive species is likely to prove a weaker competitor on the more stressful and energetically demanding open coast.
KeywordsHabitat variabilityEnvironmental stressMytilus galloprovincialisPerna perna
Organisms inhabit a highly heterogeneous world. By acting on different temporal and spatial scales, environmental conditions vary in intensity, frequency and suddenness shaping species dynamics, creating a mosaic of structures, modifying interactions between organisms and affecting competition outcomes (e.g. Levins 1968; Paine and Levin 1981; Chesson and Huntly 1997; Zardi et al. 2008). Short- or long-term stressful changes in abiotic conditions, which impair or threaten to impair homeostasis, can trigger physiological and/or behavioural responses by an individual (Sapolsky 1992; Broom and Johnsen 1993; Buchanan 2000). However, a sudden or intense stress event can have lethal consequences and is known to affect space allocation, recruitment and subsequent inter-specific competition and species composition (e.g. Connell and Keough 1985; Chesson 1994; Sousa 2001).
By interacting with the diverse biological attributes of species, environmental heterogeneity plays a defining role in invasion and coexistence dynamics (Hu and Tessier 1995; Gerlach and Rice 2003; Leicht-Young et al. 2007). As both indigenous species and invaders respond to environmental variations, it is the difference in their responses that determines the success of the invader and how it interacts with native species (Chesson 2000; Shea and Chesson 2002). If environmental conditions change, the well adapted indigenous species may lose its prior advantage over non-indigenous species (Byers 2002; Pranovi et al. 2006). However, changes of environment can also challenge established invasive species and possibly limit their distribution (Sheley et al. 1997; Shea and Chesson 2002; Zardi et al. 2008).
Here, we examined the responses of competing indigenous and invasive mussels to environmental factors that show strong seasonality and spatial variability and we discuss how this affects the potential for co-existence. The Mediterranean mussel Mytilus galloprovincialis is a successful invader worldwide. It is the most successful marine invasive species in South Africa (Robinson et al. 2005; Hanekom 2008) where it has very high recruitment rates and reproductive output (van Erkom Schurink and Griffiths 1991; Harris et al. 1998) and it is a better exploitation competitor, occupying freed space more effectively than indigenous species (Erlandsson et al. 2006). On the south coast of South Africa, M. galloprovincialis co-exists with the indigenous mussel Perna perna with partial habitat segregation in the lower eulittoral zone (Bownes and McQuaid 2006). Among the most important physical factors shaping rocky intertidal populations are wave exposure and the effects of scouring or burial by sand (e.g. Marshall and McQuaid 1989; Gaylord et al. 1994; Denny 1995; Zardi et al. 2006a). Both abiotic components can cause a temporary impoverishment of the biota by selective species elimination (Devinny and Volse 1978; McQuaid and Branch 1985; Petes et al. 2007), explain the habitat segregation of some intertidal species (Marshall and McQuaid 1989; Zardi et al. 2006a, b; Petes et al. 2007) and, when intense and sudden, they can be responsible for non-selective mass mortality (Denny and Wethey 2000; Denny et al. 2003). M. galloprovincialis and P. perna show different tolerances to wave and sand stress. P. perna has higher attachment strength than the invasive species and consequently it is able to withstand wave action better (Zardi et al. 2006b). In contrast, M. galloprovincialis is less vulnerable than the indigenous species to the scouring and anoxic stresses induced by sand inundation and burial (Zardi et al. 2006a).
Attachment strength of both species will fluctuate seasonally and will be higher on the open coast where hydrodynamic stress is greater.
Reproductive output of both species will be reduced under more wave exposed conditions because of conflict with the need for stronger byssal attachment, but the consequences of this trade off will be greater for the invasive species because of its higher investment in gamete production.
Because the two species show different tolerances to wave action and sand stress, their mortality rates will differ spatially and seasonally, with higher mortality for M. galloprovincialis when and where hydrodynamic stress is high, and for P. perna during periods and at sites of maximum accumulation of sand.
Materials and methods
Along the south coast of South Africa, rocky headlands bound large “half heart” bays. The measurements were run monthly between November 2006 and October 2007 in two of these bays (Plettenberg Bay 34°00′17″S, 23°27′17″E; Algoa Bay 33°58′47″S, 25°39′30″E) and at two open coast sites at headlands (Robberg 34°06′14″S, 23°23′07″E; Cape Recife 34°02′27″S, 25°32′01″E).
Mytilus galloprovincialis and Perna perna individuals (3.5–4.5 cm; n = 12 each month for each species at each location) were tested in situ for attachment strength as previously described in Zardi et al. (2006b). All dislodged mussels were at least 20 cm from each other so that attachment strength measurements were not influenced by previous ones. Data fulfilled the pre-requisites for parametric analysis (Cochran’s Test) thus were analysed using nested ANOVA (GMAV5 software) to investigate the effects of species (P. perna or M. galloprovincialis, fixed factor) and time (12 months, fixed factor) on attachment strength with habitat (bays or open coast, fixed factor), sites nested within habitat (random factor) and locations nested within sites (random factor). All statistical interactions were considered in the analysis.
Because reproductive effort is strongly linked to size in mussels, we restricted ourselves to animals of a narrow size range. The wet mantle was dissected from each mussel (shell length 3.5–4.5 cm; n = 25 each month for each species at each location) and both body and mantle were dried at 60°C to a constant weight. Samples were weighed to the nearest 0.001 g and Gonad Index (GI) was then calculated as the dry mantle weight divided by the dry body weight (Carrington 2002a). For the cross-correlations (see below) seasonal fluctuations were used in the analysis. For the comparison between species in the two habitats only peaks of GI were used because the aim was to compare maximum reproductive output rather than fluctuations of gonad tissue development. These peaks occurred in January 2007 for both species. These values fulfilled the pre-requisites for parametric analysis (Cochran’s Test) and were analysed using nested ANOVA (GMAV5 software) to investigate the effects of species (P. perna or M. galloprovincialis, fixed factor) on maximum GI with habitat (bays or open coast, fixed factor), sites nested within habitat (random factor), locations nested within sites (random factor). All statistical interactions were considered in the analysis.
At each site, digital pictures of 12 quadrats (20 × 20 cm) were taken monthly, and new quadrats were selected each month (i. e. each quadrat was photographed twice only). In each quadrat, 20 individual mussels were identified and mortality was estimated every month as disappearance or death of identified individuals between consecutive photographs. Data were analysed as for attachment strength, with mortality as the dependent factor. Note that because mortality is based on differences between successive months, the number of time period is one less than attachment strength and GI (i. e. from December 2006 to October 2007).
Seasonal sand depth
Digital photographs of mussel beds populating vertical rocks were taken every month at four locations 200 m apart at each of the four sites (Plettenberg Bay, Robberg, Algoa Bay, Cape Recife). Seasonal sand depth was determined by analysing digital images of 3 m wide transects running c. 15 m horizontally across a vertical rock face at each site. Sand depth tended to be uniform across the vertical width of each transect and the lowest level recorded was the benchmark from which all other levels were measured. The data fulfilled the pre-requisites for parametric analysis (Cochran’s Test) and were analysed using nested ANOVA (GMAV5 software) to investigate the effects of habitat (bays or open coast, fixed factor) and time (12 months, fixed factor) on sand depth, with sites nested within habitat (random factor). All statistical interactions were considered in the analysis.
Off-shore wave heights
Off-shore wave heights were obtained from a virtual buoy located at 34°51′S, 23°53′E, as recorded by the USA National Data Buoy Centre (www.ndbc.noaa.gov). A virtual buoy gives a wave model prediction based on a real buoy report. Wave height, wave direction, wave period, wind speed and wind direction were extracted from the NOAA WAVEWATCH III model (Tolman 1999). Estimated mean daily wave heights were calculated from hourly values and then used to calculate mean monthly values.
Mean attachment strengths of M. galloprovincialis and P. perna were cross-correlated with GI, using GI as the lagged variable. Wave forces were cross-correlated with mean attachment strengths of each species, using attachment strength as the lagged variable. Mortality rates of each species were cross-correlated with seasonal sand elevation and wave force, using mortality as the lagged variable. Cross-correlations had a lag interval of 1 month and were performed with the computer program Statistica 7. Seasonal bay and open coast wave force data were obtained by converting offshore wave height values according to the regression equation derived in Nicastro et al. (2008).
While in bays M. galloprovincialis had a significantly higher (64%; Fig. 2b) maximum GI than P. perna, the difference between species was not significant on the open coast (species × habitat, P < 0.01). There was also a species × location (site, habitat) interaction because at one location on the open coast maximum GI for M. galloprovincialis > P. perna (P < 0.05).
M. galloprovincialis had two spawning events. The highest GI values were reached in January followed by a sudden decrease indicating strong summer spawning. A weaker, more protracted spawning event was observed in spring between August and October. There was only a single clear spawning event in summer for P. perna, with maximum gonad index values in January followed by spawning till February.
There was a significant time × habitat interaction (P < 0.05). In general, the open coast locations had higher monthly mussel mortality rates than bays but in June this pattern was reversed and in January no differences between habitats were recorded. A time × location (habitat, site) interaction is explained by location differences recorded in December, March, September and October (P < 0.001).
In bays, two peaks in mortality occurred at the end of summer (February) and in winter (June), with much lower rates for the rest of the year. On the open coast there was a peak in mortality again in late summer (February), but no such peak in June. Instead there was a protracted period of relatively high mortality over winter/spring (June–October).
Seasonal sand depth
In February and in June, bays had a significantly greater range between maximum and minimum sand depth than did the open coast (P < 0.01; Fig. 3a). There was a time × site (habitat) interaction due to high variability among sites for 9 out of 11 months (P < 0.05). In both habitats, sand accumulation increased towards the end of summer (February) and the beginning of winter (June), followed by massive removal in the subsequent months. Sand levels then remained low through spring/early summer and autumn. The lowest sand level (y = 0) occurred in March for the open coast and in April in bays and fluctuation of sand coverage showed mean vertical changes of up to 122 cm.
Wave force measurements
Mean monthly offshore wave height increased in late autumn–winter with a peak in July, and decreased in spring and summer with a minimum in January (Fig. 3b).
Environmental heterogeneity challenges invasive and native coexisting species and can alter the exploitative ability of the invader (Melbourne et al. 2007). Here, we show that physiological performances of co-existing mussels are affected by spatial and temporal fluctuations of environmental conditions. Because these species respond differently to abiotic stresses, the invasiveness of Mytilusgalloprovincialis and thus the outcomes of interactions between the two mussels are likely to vary along environmental gradients.
Previous studies have described M. galloprovincialis and Perna perna as species with different evolutionary strategies. M. galloprovincialis emphasises reproduction over attachment strength, and adopts a more dynamic response to disturbance, seeking aggregation or a safer arrangement as a mechanism of coping with hydrodynamic stress. In contrast, P. perna has higher attachment strength at the expense of lower reproductive output (van Erkom Schurink and Griffiths 1991; Harris et al. 1998; Zardi et al. 2007). Interestingly, our results only partially confirm these findings. Previous findings show that GI in both species is proportionate to the number of eggs, so that a higher GI value indicates greater egg production (Zardi et al. 2007). While in bay habitats the maximum GI of M. galloprovincialis was 64% higher than that of P. perna, on the open coast, maximum values for GI were not significantly different between the two species. M. galloprovincialis has a second spawning peak in August, however, the strong negative effect of a more stressful environment is evident in both spawning peaks. Attachment strength also varied spatially, mirroring the degree of wave action (see Nicastro et al. 2008 for measurements of maximum wave force at these sites), with higher values on the open coast than at bay sites for both species, and higher values for the indigenous species than M. galloprovincialis. The open coast habitat reduces M. galloprovincialis’ traits of high fecundity probably because of the need to channel energy into attachment strength. Thus gonad development of the invasive species is affected by coastal topography, while the spawning rate of the indigenous species is preserved. Therefore abiotic stress in the form of wave action appears to reduce the ability of M. galloprovincialis to be invasive as it loses its advantage over the native species in terms of reproductive output.
This correlates well with the observation of von der Meden et al. (2008) that the two species have significantly greater % cover within bays than on the open coast, and the effect of bay is stronger for M. galloprovincialis than for P. perna.
Besides spatial disparities, attachment strength also showed strong temporal variation, generally decreasing in summer and increasing in winter, and was strongly positively correlated (with little lag) with seasonal fluctuations in wave force in the two habitats. While this appears to be yet another manifestation of the ability of mussels to sense and respond to their flow environment, they are not always precise in their response, for example low attachment strength values relative to wave height were observed in September and October 2007 and it is well known that other factors affect the strength of attachment. These include sea surface temperature, food supply, predator cues, endolithic infestation and the molecular structure of the byssus (Lucas et al. 2002; Ishida and Iwasaki 2003; Moeser et al. 2006; Babarro et al. 2008, Zardi et al. 2009). For example, a previous study at Pletettenberg Bay (Zardi et al. 2007) showed that the attachment strength of both M. galloprovincialis and P. perna is negatively correlated with sea surface temperature. In addition, rocky shores are regularly and extensively inundated by sand increasing the amount of sediment (sand, shell fragments etc.) within a mussel bed and making the substratum less stable and mussels more prone to dislodgement (Zardi et al. 2008). The hydrodynamic stress experienced by intertidal organisms depends partly on their shape (Denny 1995). Previous studies on the west coast of South Africa showed that the shell of M. galloprovincialis tends to be lower and narrower at exposed sites, perhaps reducing the effect of hydrodynamic forces (Steffani and Branch 2003). It is possible that this allows mussel populations on the open coast to reduce higher hydrodynamic stress. However, regardless of intraspecific differences in shell shape, our data showed that mortality rates are higher on the open coast than in bays.
At bay sites, there was no correlation between GI and attachment strength for either species. This lack of correlation suggests that the relatively low hydrodynamic stress in bays allows mussels to invest less energy in attaching to the substratum and to maintain high levels of GI and survival simultaneously. In contrast, on the open coast, GI was negatively correlated with attachment strength at a lag of +1 and (only for P. perna) with zero lag, indicating that, for both species, attachment strength was high when, or 1 month after, GI was low. Alternatively, high GI coincided with or preceded (P. perna) low attachment strength. Note that the analysis refers to correlations between attachment strength and GI, which appear despite the effects of other energetically demanding factors such as shell growth, which may also show seasonality (Griffiths and King 1979). Thus, major spawning events occurred during periods of low wave action, when limited investment was required in attachment strength. In addition, the minor spawning for M. galloprovincialis coincided with periods of intense wave action and very high attachment strength. Hence, it seems that channelling more energy into attachment strength limits gonad tissue development, indirectly supporting the concept of trade off between reproduction and attachment strength (Carrington 2002a, b; Moeser et al. 2006; Zardi et al. 2007). Moeser and Carrington (2006) found that seasonal variation in attachment strength is not related to increased thread production in response to hydrodynamic stress, and that the material properties of byssal threads and thread decay rates could be determinants of attachment strengths; whatever the mechanism a trade off between gamete production and attachment strength is always maintained. However, this is not true for Mytilus edulis in suspended culture (Lachance et al. 2008). Intertidal mussels experience a more heterogeneous environment than when cultured in suspension especially in terms of temperature, food availability and hydrodynamic stress. Differences in environmental conditions could explain these contrasting results.
Wave action (Nicastro et al. 2008) and sand fluctuation are greater on the open coast and in bays, respectively. Cross-correlations between morality rates and sand elevation showed a high positive correlation at bay sites, indicating that sand inundation is a major stress affecting survival in mussel populations in bays. The lag period here was zero, underlining the immediate effects of sand, as mortality and sand accumulation peaked simultaneously. Although wave induced hydrodynamic stress often plays a crucial role in regulating mortality rates of intertidal organisms, the correlation between wave action and mortality rates of P. perna and M. galloprovincialis was very weak in bay habitats and non-significant on the open coast. Probably, this was due to the masking effect of sand stress. Our results show that these two stresses operate out of phase with one another. In particular, in February sand level was extremely high, coinciding with periods of relatively low wave action. Consequently, a sand-induced peak in mortality decreases the correlation between wave and mortality. In addition, wave stress is more prolonged over the year and is usually characterized by a relatively gradual seasonal increase. Contrary to sand stress, mussels have the potential to adapt to higher wave action by increasing their attachment to the substratum. This could also form part of the explanation for mortality being less well correlated to wave force than to sand stress. It is also possible that mussels on the open coast adjust to the continuously intense hydrodynamic stress typical of this habitat, giving them higher resistance to sudden increases in wave action during storms. Thus, the consequence of chronically higher wave action is a reduction in gonad output for M. galloprovincialis, rather than mortality. Unexpectedly, in February, despite greater sand accumulation in bays, more mussels died on the open coast. Carrington (2002a) showed that M. edulis on Rhode Island shores is slow to build up attachment strength following summer, perhaps due to energetic constraints on thread production during the reproductive season. Waves of equivalent magnitude arriving later in the storm season pose less risk of dislodgment because attachment strength steadily increases. In our study, the effect of greater wave action on the open coast could have been further enhanced by the concurrent sand inundation and a sudden increase in hydrodynamic stress acting on mussels weakened by a major spawning event in January. Regardless of the causes, the high mortality rates experienced by mussel populations on the open coast increase population turnover rates in this habitat, freeing space for new colonizers and allowing more frequent and intense gene pool replenishment (Nicastro et al. 2008).
To become a successful invader, an introduced species must have some advantage over an indigenous species, at least at particular times or places, or in a certain life-history trait, such as colonising ability (Shea and Chesson 2002; Hastings et al. 2005). A high reproductive output (van Erkom Schurink and Griffiths 1991; Harris et al. 1998; Zardi et al. 2007), together with fast growth (Griffiths et al. 1992; Hockey and van Erkom Schurink 1992) and the capacity to colonise free space very rapidly, make M. galloprovincialis a very strong competitor in terms of re-colonization of free space (Erlandsson et al. 2006). However, disturbance often prevents local competitive exclusion by dominant competitors because species usually exhibit trade-offs between competitive ability and colonization ability or between competitive ability and stress tolerance (Wilson 1984; Petraitis et al. 1989; Chesson and Huntly 1997). Our results partially confirm the initial hypotheses. P. perna is attached to the substratum more strongly than M. galloprovincialis. In both species, strength of attachment fluctuates spatially and during the year in response to wave action changes. However, probably because of the confounding effect of other environmental stressors (especially sand accumulation) a significant correlation between hydrodynamic stress and mortality rates was not observed. In addition, our results show that the reproductive output of M. galloprovincialis is generally lower on the open coast than in bays, and that on the open coast maximum GI values for the two species are not significantly different. Less stressful conditions in terms of wave action in bay habitats (Nicastro et al. 2008), allows M. galloprovincialis to maintain a low strength of attachment and invest more energy in gonad development.
The observed inter- and intra-specific differences in responses to the environment stress highlight the need of M. galloprovincialis and P. perna to optimize resource allocation and help explain patterns of adult abundance in these two habitats (von der Meden et al. 2008). This study shows that, by observing the responses of coexisting species to a wide range of environmental conditions, it is possible to identify characteristic differences in relative performance and ecological plasticity which are known to be crucial factors in dynamics of coexistence between competing indigenous and invasive species (Davis et al. 2000; Mack et al. 2000; Pranovi et al. 2006).
This work is based upon research supported by the South African Research Chairs Initiative of the Department of Science and Technology and National Research Foundation with additional funding from Rhodes University and a Claude Harris Leon postdoctoral research fellowship Foundation awarded to GIZ. We thank Liesl Knott for technical support.