Study sites
We selected two field sites for our comparative investigations: wave protected mussel beds on the soft sediment tidal flats near the island of Sylt (55°02′N, 8°26′E) and the exposed rocky offshore island of Helgoland (54°11′N, 07°53′E) (Fig. 1). Both sites are located in the south-eastern North Sea and linked by currents enabling exchange of pelagic larvae of benthic organisms (Giménez and Dick 2007). They represent fundamentally different environments with respect to abiotic and biotic conditions.
At Sylt, mean annual water temperature is about 9°C, with a summer average of 15°C and a winter average of 5°C. Tides are semi-diurnal and average tidal range is about 2 m. Average salinity is 30. The area is protected from prevailing westerly (onshore) winds and waves by sand dunes. Detailed descriptions of the geology, hydrography and sediments of the study area are given by Reise (1985), Austen (1992, 1994a, b), Bayerl and Higelke (1994) and Reise et al. (1994). In the Wadden Sea, periwinkles L. littorea reach highest densities on mixed epibenthic aggregations of the non-indigenous Pacific oyster Crassostrea gigas (Thunberg) and the native blue mussel Mytilus edulis L. which represent preferred habitats for periwinkles because bivalve aggregations provide an appropriate attachment substratum and suitable food conditions.
The island of Helgoland represents the only rocky shore in the southeastern part of the North Sea. The island is located <100 km south–west of the island of Sylt. Here, average tidal range is about 2.3 m, mean water temperatures are 13°C in summer and 7°C in winter at an average salinity of 32. Blue mussels and the Pacific oyster occur at a low density on the macroalgae-dominated intertidal at Helgoland. More detailed descriptions of the ecology, geology and of abiotic factors are given by Hagmeier (1930), Wurster (1962), Janke (1986, 1990), Bartsch and Tittley (2004) and Reichert and Buchholz (2006).
Size and abundance of L. littorea
Periwinkles were collected between July and August 2006 during low tide using identical methods at both study sites. To investigate periwinkle densities and their size distribution in different tidal zones a stratified random sampling design was used. Snails were collected in the mid intertidal zone (emersion time about 4 h), low intertidal (emersion time about 2 h) and in the shallow subtidal zone. A plastic ring with a diameter of 20 cm (corresponding to an area of 0.03 m2) was placed randomly on the bottom. All snails inside the ring were counted and grouped into 5 size categories (measured from aperture to apex, Fig. 2): (1) 1–8 mm, (2) >8–12 mm, (3) >12–18 mm, (4) >18–22 mm, (5) >22 mm. At each tidal level six replicate samples were taken.
Shell strength
In August 2006, we collected snails of the same size range (15–19 mm) at both study sites. Before investigating shell strength snails were inspected for infection with P. ciliata. This shell boring polychaete may strongly decrease periwinkle shell strength (Buschbaum et al. 2007). Thus, only non-infested snails were used for shell strength comparisons (n = 45 snails per site). To determine the force necessary to break periwinkle shells, investigations were performed with the same methods and measuring device used by Buschbaum et al. (2007), which allows estimation of general shell strength by measuring the force required to crack shells in one particular plane.
Shell thickness
To determine shell thickness of periwinkles originating from Sylt and Helgoland, 20 periwinkles from 17 to 19 mm shell height were collected at each site and boiled in order to remove the soft body from the shell. A moulding cutter was used to bisect the shells. Afterwards, snail shell thickness was measured to the nearest 0.01 mm using a stereomicroscope fitted with an ocular micrometer (see Fig. 2).
Snail growth experiments
In a reciprocal transplantation experiment, we measured shell height increment of L. littorea at both sites from April/May until October 2006. At both Sylt and Helgoland, juvenile snails of a size range from 9.5 to 12 mm shell height were collected. This size class was used because length increments of larger organisms might be too small to detect effects of periwinkle origin and location of growth on L. littorea shell increment. Furthermore, shells of larger specimens were often eroded, potentially confounding length measurements with shell degradation.
To test whether snail origin and site where periwinkles matured affects growth rate in L. littorea, snails collected at Sylt and Helgoland were marked with differently coloured nail varnish and transplanted to the mid intertidal zone of both sites. We corralled the snails in closed cages (20 cm in diameter, 25 cm in height) made of polyethylene mesh wire (mesh size 0.5 cm) and a polyvinyl chloride (PVC) bottom plate. Altogether, we used 16 cages per study site. At both sites, eight cages each with ten snails originating from Helgoland and eight cages each with ten snails collected near Sylt were installed.
On the rocky island of Helgoland cages were fixed using dowels and screws while at Sylt cages were fixed with iron rods on mixed beds of C. gigas and M. edulis. Snail size was measured at the beginning and at the end of the experiment to the nearest 0.01 mm using callipers, and shell increment was calculated as the difference between both measurements.
At the beginning of the experiment, mean shell height of L. littorea from both habitats did not differ significantly from each other.
Predator densities
In the soft sedimentary environment of the Wadden Sea, periwinkles are mainly consumed by green crabs (Carcinus maenas L.), while on the rocky shore of the island of Helgoland both green crabs and edible crabs (Cancer pagurus L.) prey on L. littorea (Scherer and Reise 1981; Buschbaum et al. 2007; own observations).
We used the same method to determine crab abundances on intertidal beds of oysters and mussels on the sandy shore and in the intertidal zone of the rocky environment. In August 2006, all crabs inside a randomly selected area of 1 m2 (n = 12) were collected, counted and their carapace width was measured to the nearest 1 mm using calliper.
Infection with shell boring P. ciliata and trematode parasites
At each site, 56 snails were randomly collected to study the infection rate with shell boring polychaetes P. ciliata; we also collected 75 periwinkles at each site to investigate the prevalence of digenean trematodes. Since both, P. ciliata and trematodes occur exclusively in periwinkles ≥12 mm shell height (Lauckner 1984; Buschbaum et al. 2007; Thieltges and Buschbaum 2007), only individuals of this size range were sampled.
To assess the prevalence of shell boring polychaetes, we determined the presence of living P. ciliata for each snail. In the laboratory, periwinkles were placed in separate Petri dishes for 1 h prior to visual inspections with a stereo microscope; the 1 h in water allowed live P. ciliata to extend from shells. Periwinkles hosting living polychaetes were considered to be infested.
Tissue-inhabiting trematodes utilise L. littorea as first intermediate hosts and are commonly associated with periwinkles (Lauckner 1984). To determine the prevalence of trematodes, periwinkle shells were carefully cracked with a hammer and the entire soft body of each snail was checked for trematodes under a stereo microscope.
Mean prevalence of trematodes and P. ciliata (% infected snails from total sample) was calculated for the rocky and sedimentary shores.
Statistical analyses
Results are presented as arithmetic mean ± standard deviation. Data of the growth experiment were analysed by using a two-factorial ANOVA, with location of growth (two levels) and snail origin (two levels) as fixed factors. The Tukey-honest-significant-difference-test (HSD) was used for pairwise multiple comparisons of the growth experiment. Homoscedasticity of untransformed data was confirmed by Cochrans’s test. Comparisons of periwinkle abundance and shell size, snail shell strength and thickness and predator abundance between both sites were analysed by Mann–Whitney U test. Differences in prevalence of trematodes and P. ciliata were analysed using Chi-square tests. Effects were considered to be statistically significant if P < 0.05.