Invasion trajectory of Pacific oysters in the northern Wadden Sea

Since the annual imports of Pacific oysters for farming at Sylt had started, 15 years elapsed before high (>1000 m⁻²) feral densities occurred. Diederich et al. (2005) could show that successful reproduction was bound to a few summers with sufficiently high temperatures ≥18 °C. However, after 4 years larvae dispersed and settled on wild mussel beds. Regular stocking of oysters in the farm circumvented the hurdle of unpredictable temperature conditions for establishment by maintaining propagule pressure. One million densely packed oysters at the farm site may have also facilitated the fertilization rate in the tidal waters. Furthermore, only 50% of the high tide volume of water is exchanged in the tidal basin per semi-diurnal cycle (Gätje and Reise 1998), allowing for retention of planktonic larvae. Finally, abundant mussel beds up- and downstream of the oyster farm offered ample settling sites for oyster larvae at suitable depth in this sedimentary environment.

This combination of favorable prerequisites made the establishment of a feral population around Sylt almost inevitable, and could have been predicted from previous invasions elsewhere in cold waters. The time lag between first introduction and first establishment varied considerably between invaded coastal regions. At the American NW-coast, Pacific oyster introductions commenced in 1912 and first settlement in the wild happened 13 years later (Quayle 1988). In Tasmania, introductions since 1947 gave rise to a feral population 9 years later (Bourne 1979). At the Atlantic coast of southern France, introductions since 1968 were followed by wild settlement already 3 years later (Maurin and LeDantec 1979), while further north at Brittany this took more than 20 years (Lejart and Hily 2011). In the Oosterschelde estuary (SW Netherlands), introductions commenced in 1964 and first settlement happened after 11 years (Smaal et al. 2009). Only at the Swedish coast, where spat was introduced in 1973–1976, establishment took much longer with feral populations observed in 2007 (Wrange et al. 2010). However, it remains questionable whether these feral populations are direct descendents of farmed oysters or rather represent the invasion front from more southern populations unless population genetic studies will be performed.

It seems that first establishment in the wild after the start of sea-based farming was primarily temperature driven. Given the inherent temporal and spatial variability in this parameter, it was like playing ‘ecological roulette’ with oyster establishment bound to happen sooner or later (compare with Carlton and Geller 1993). Once the temperature hurdle was overcome, the almost global distribution of C. gigas after a century of introductions suggests that this species with its combination of invasive traits (wide tolerance to physical factors, fast growth, high fecundity, large size and longevity) has encountered little resistance by recipient ecosystems. Nevertheless, establishment phases with low abundances over several years (11 around Sylt) seem to be common, and in several cases have not yet developed further or showed signs of exponential growth. On rocky shores, abundances have generally remained rather low (i.e., Ruesink 2007; Robinson et al. 2005; Krassoi et al. 2008; Kochmann et al. 2013), although localized reefs have developed in parts of southern England (Herbert et al. 2016). This also applies to cold sedimentary coasts in Scandinavia (Wrange et al. 2010; Dolmer et al. 2014; Holm et al. 2015). On British coasts, field trials with C. gigas derived from hatcheries in British Columbia, commenced in 1967 and yet high abundances are confined to just a few estuaries (Herbert et al. 2016).

Repeated introductions may have circumvented extinctions after initial establishment. However, also longevity could overcome recruitment failures over many years. At the coast of Argentina, a singular introduction occurred in 1982 because farming was soon abandoned. Nevertheless, an established feral population was observed 10 years later (Escapa et al. 2004), which must be attributed to survivors of the initial introduction. In England, C. angulata, which later was recognized as a Taiwanese strain of C. gigas (Boudry et al. 1998), was introduced in 1890 and survived in small populations until 1970 (Humphreys et al. 2014; Herbert et al. 2016). Again, this can only be explained by long life spans of a few individuals because no regular spat falls and high abundances were reported.

It can be concluded that ready establishment of C. gigas on so many coasts has been made easy by sea-based farming practices with regular introductions of spawning individuals compared to irregular yachts and ships as vectors of introductions. The invasive traits of C. gigas made an almost universal establishment inevitable and coastal ecosystems resistant to the invasion of this oyster are few. However, expansion phases leading to exponential growth of feral populations are less ubiquitous.

A transition from less than one hundred oysters per m² to well over 1000 oysters per m² is not common at all. It has not occurred at exposed rocky shores but is a phenomenon of sheltered sedimentary shores where there are larger particles such as gravel, pebbles and shell, rocky outcrops, boulders, harbor walls or biogenic reefs such as mussel beds. Comparisons between coasts are hampered by different methods of quantification in the field, i.e., as percentage of area covered by oysters, as total wet weight or individual numbers per unit area. These are not convertible. At Sylt, abundance of 100 individuals m⁻² usually amounts to about 25% cover, >400 m⁻² to 100% cover, and 1000 m⁻² or more comprise large oysters with shells often partly fused to one another or to empty shells, and with attached younger ones on top, forming a complex reef structure above and below sediment surface, weighing >30 kg m⁻² (Fig. 6).

Open image in new window

The development of such high-density reefs is made possible by larval gregariousness during settlement and by circumventing the larviphagy of adult oysters (Diederich 2006; Tamburri et al. 2007; Troost et al. 2008a, b; Troost 2010). We could show that the proximity of contiguous oysters facilitates larval settlement by conducting an experiment with settlement plates in the centre of artificial rings composed of densely packed mussels, oysters or both species mixed in approximately equal proportion (for further description see Kochmann et al. (2008), and see circular patches at the seaward side of the wild bed in 2008 shown in Fig. 2). Panels surrounded by oysters had significantly more oysters spat than panels surrounded by mussels (unpublished data). Such gregariousness of larval settlement facilitates crowding on oyster reefs and also leads to stable bed areas.

More positive feedbacks may play a role in exponential population growth. The elevated three-dimensional reef structure offers firm substrate for settlement in good feeding positions, and reefs protect individual oysters against dislocation by waves or currents, and probably against predators. Lejart & Hily (2011) mention abundances >1000 m⁻² for estuarine habitats in Brittany, Cognie et al. (2006) wet weight up to 45 kg m⁻² south of Brittany, and Laugen et al. (2015) >1000 m⁻² as upper range for shallow water sites at Swedish fjords. These high abundances may be rather patchy or insular due to limiting attachment for settling oyster larvae. However, in the Wadden Sea and presumably in the nearby Oosterschelde estuary, abundances >1000 m⁻² on former mussel beds have become a common feature since the 2000s (Markert et al. 2010, 2013; Nehls et al. 2011; Dolmer et al. 2014; Walles et al. 2015 and this study). Apart from harbor walls and low-lying coastal revetments at low-to-moderate wave exposure, such high abundances occur on mussel beds or sites of former mussel beds with abundant empty shells. In the western Dutch Wadden Sea, oysters also covered sites where mussel beds had been overexploited and devastated by a previous fishery (van Stralen et al. 2012; van den Ende et al. 2016). Mussel beds provide the substratum for large and dense Pacific oyster reefs in the Wadden Sea at positions ideal for epibenthic suspension feeders. On a global scale, the Wadden Sea has developed into a hot spot of feral C. gigas thanks to its extensive mussels beds.

There seems to be a close match between the ecological niches of oysters and mussels in this coastal region. However, the match is not perfect. Oysters are few and small at the highest parts of mussel beds (own unpubl data; Waser et al. 2016), and there are still regions with mussel beds in the Wadden Sea where oysters have not achieved high abundances (Nehls et al. 2011; van Stralen et al. 2012; van den Ende et al. 2016). Also at the coast of SE-England, in Limfjorden (Denmark) and Kattegat (Sweden), Pacific oysters occur on sediment together with mussels but oyster abundances were one order of magnitude lower than in the Wadden Sea (Groslier et al. 2014; Holm et al. 2015, 2016; Norling et al. 2015; Hollander et al. 2015; Herbert et al. 2016). Insufficient food supply has been suggested as an explanation but the exact mechanisms preventing high oyster densities in these areas are still unclear.

The presence of mussel beds alone is, however, not sufficient to explain the development towards abundances >1000 per m². Temperature is crucial for the global distribution of C. gigas (Carrasco and Barón 2010), and for spawning a wide range of 16 to 34 with an optimal range of 20 to 25 °C has been suggested (Shatkin et al. 1997). The role of high summer temperatures for spawning and larval development has been shown to trigger recruitment success in C. gigas around Sylt (Diederich et al. 2005). From 2001 onwards, mean July–August water temperatures exceeded 17 °C (except in 2011) and recruitment success became rather regular. Particularly strong spat fall events coincided with peaks of temperature maxima (Fig. 3). Regular recruitment success may be a precondition for crowding to occur with oysters settling upon oysters, generating reefs which become independent of the underlying mussel beds. In conclusion, the transition from establishment phase with low abundances to expansion phase with abundances up to two orders of magnitude higher in the Wadden Sea was aided by common mussel beds on sheltered sediments and high regional summer temperatures (Wiltshire and Manly 2004; van Aken 2008; Witte et al. 2010) that allowed regular recruitment with oysters settling upon oysters.

The rare occurrence of C. gigas reefs with abundances >1000 m⁻² on global scale needs to be investigated further. High-density reefs also occur on mud flats in the native region but individuals tend to remain smaller than in the Wadden Sea (Fig. 6). In the Yellow Sea, these reefs are not grounded on mussel beds but on scattered boulders, often laid out by artisanal fisherman. Boulders may become buried by sedimentation while the oysters grow upwards and subsequent generations settle upon their predecessors. So far it seems that C. gigas reefs are confined to sedimentary environments at low tide level with firm substrate for attachment. Further prerequisites are frequent recruitment and probably eutrophic conditions with strong tidal flushing.

After peak density in the feral oyster population around Sylt in 2007, decline commenced with two consecutive summers without recruitment and a series of ice winters starting 2009/2010 when many oysters died. Comparing autumn and spring abundances, Büttger et al. (2011) estimated winter mortality to about 90% with all size classes of C. gigas affected. However, most shells remained in place and long-lasting effects on the oyster population were not to be expected. For the same winter, Strand et al. (2012) estimated 66% winter mortality in the adjacent Danish Wadden Sea, while no enhanced mortality was noted further south in the Dutch Wadden Sea (van Stralen et al. 2012). As can be expected, winter mortality at Scandinavian coasts increased with latitude with up to 100% in the course of that unusually severe winter (Strand et al. 2012). Thus, Pacific oyster populations at higher latitudes may experience occasionally dramatic crashes when hit by harsh winter conditions.

However, the long-term ice record for the German North Sea coast shows a general decline in frequency since the mid 1980s (Schmelzer et al. 2016). When oyster abundance was still low and most individuals young, the similarly harsh ice winter of 1995/1996 caused only a mortality of 34% (Reise 1998). Also Büttger et al. (2011) recorded lower winter mortality at sites with lower oyster abundances. They suggested that the high-density population at northern Sylt might have been more vulnerable to sheets of ice moving back and forth with the tides. The following two winters (2010 to 2012) brought some ice cover as well, and recovery of the feral population was delayed. In view of the long period with mild winters (1996 to 2009), the strong winter incursion in the invasion trajectory of Pacific oysters around Sylt should be regarded as a singular event but could have lasting effects by selecting for cold resistant oysters. During the establishment phase, the number of farmed and feral oysters remained in the same order of magnitude and extended gene flow from hatcheries to the feral population could have prevented selection for local adaptations (Moehler et al. 2011). After the expansion phase, however, strong selective pressure such as ice winters or disease on the numerically dominant feral population may facilitate local evolution and influence the extent of future frost and disease-associated mortalities (Sussarellu et al. 2015; Wendling and Wegner 2015; Wendling et al. 2017).

Recruitment failure in 2007, 2008 and 2015 coincided with particularly high abundances of resident oysters. Three years are not yet sufficient to prove this case; however, future research should address density dependence as a driver in the population dynamics of C. gigas. According to our own qualitative observations in adjacent areas with lower oyster density, these recruitment failures were a phenomenon confined to the tidal basins of Sylt. Since oyster spat was not only lacking within reefs but also on scattered clumps of oysters, we assume that food shortage or larviphagy is unlikely to explain recruitment failure. More likely could be epidemic diseases arising at high population density, affecting larvae or oysters soon after metamorphosis in particular. Size-frequency distributions (Fig. 4 and see Schmidt et al. 2008; Walles et al. 2015) suggest high mortality in the first year of benthic life. Although shore crabs Carcinus maenas and starfish Asterias rubens have been observed to prey on small oysters (own unpubl. experiments), predation is unlikely to account for complete recruitment failure in the intertidal zone. Crabs preferred young mussels over young oysters and starfish were almost absent from intertidal flats around Sylt.

However, pathogenic bacteria and viral infections resembling the widespread summer disease syndrome (SDS) are known to affect populations in adjacent regions (Watermann et al. 2008; Thieltges et al. 2013; Wendling et al. 2014; Mortensen et al. 2016) and the recent recruitment failure and population decline (2015/2016) were associated with increased loads of oyster herpes virus (OsHv1, unpublished data). This may suggest that disease can also play an important role for the population dynamics of feral oysters on Sylt. With global warming and continued crowding in a large host population, it is to be expected that C. gigas around Sylt will become more frequently exposed to disease agents (see Fey et al. 2015; Sweet and Bateman 2015; Travers et al. 2015). Oyster diseases can affect different life stages but mortalities are usually highest in early life (Le Roux et al. 2016), where also evolutionary responses are most efficient and can further contribute to establishment success (Wendling and Wegner 2015).With the development of a large feral oyster population around Sylt, we predict an adjustment phase with diseases taking a higher toll, causing strong density-dependent fluctuations around a mean density of about 1000 oysters m⁻² on the reefs, thus resembling the type II trajectory in Fig. 1.