Oecologia (2005) 143: 402–411 DOI 10.1007/s00442-004-1819-5
P L AN T A N IM A L I NT E R AC TI O NS
Petri Kemppainen Æ Solveig van Nes Æ Christofer Ceder Kerstin Johannesson
Refuge function of marine algae complicates selection in an intertidal snail
Received: 16 September 2004 / Accepted: 20 December 2004 / Published online: 15 February 2005 Springer-Verlag 2005
Abstract Species with restricted gene flow often show trait-shifts from one type of environment to another. In those rock-dwelling marine gastropods that lack larval dispersal, size generally decreases in wave-exposed habitats reducing risk of dislodgement, while increases in less exposed habitats to resist crab-crushing. In Littorina fabalis, however, snails of moderately exposed shores are generally much larger (11–14 mm) than snails of sheltered shores (5–8 mm). Observations from the White Sea (where crabs are not present) indicate that in the absence of crabs snails are small (6–7 mm) in both habitats. We assumed that the optimal size for L. fabalis in the absence of crabs is less than 8 mm, and thus that increased size in moderately exposed habitats in areas with crabs might be a response to crab predation. In a crab-rich area (Sweden) we showed that crab predation is an important mortality factor for this snail species in both sheltered and moderately exposed habitats. In sheltered habitats, snails were relatively more protected from crab-predation when dwelling on their habitual substrate, fucoid algae, than if experimentally tethered to rocks below the algae. This showed that algae function as snail refuges. Snail dislodgement increased, however, with wave exposure but tethering snails in moderately exposed habitats showed that large snails survived equally well on rocks under the algae as in the canopy of the algae. Thus in sheltered habitats a small snail size is favored, probably due to life-history reasons, while increased risk of being dislodged from the algae refuges promotes a large size in moderately exposed habitats. This study shows an example of selection of a trait depends on complex interactions of different factors (lifehistory optimization, crab predation, wave induced dislodgement and algal refuges). P. Kemppainen Æ S. van Nes Æ C. Ceder Æ K. Johannesson (&) Department of Marine Ecology, Tja¨rno¨ Marine Biological Laboratory, Go¨teborg University, 452 96, Sweden E-mail:
[email protected] Tel.: +46-52-668600 Fax: +46-52-668607
Keywords Littorina fabalis Æ Size polymorphism Æ Habitat heterogeneity Æ Hydrodynamic force Æ Size-selective predation
Introduction As a result of divergent natural selection, species living in heterogeneous environments often show habitatlinked morphological variation in adaptive traits (Levins 1968; Hedrick et al. 1976; Endler 1977; Hedrick 1986). Such shifting traits seem particularly common in species with a restricted gene flow among populations (Johannesson et al. 1993; Nevo 1998; Filchak et al. 2000; Schluter 2000; Hendry et al. 2002). In rocky-shore snails, trait-shifts have been extensively studied, both with respect to patterns and mechanisms, and it is evident that many traits change in parallel in different species, given the same circumstances of environmental variation (Boulding and van Alstyne 1993; Kirby et al. 1994; Johannesson 2003; Rochette et al. 2003), supporting natural selection as the main component of variation for these traits. Adult shell-size is one trait that in general varies with habitat in direct developing rocky-shore gastropods with limited dispersal and this variation is largely inherited in several species (Janson 1982; Johannesson and Tatarenkov 1997; Tatarenkov and Johannesson 1998; Rochette et al. 2003). For instance, in rock-dwelling littorinid gastropods shell size is generally negatively correlated with wave exposure (Littorina saxatilis: Sundberg 1988; Reid 1996; Hull 1998; Johannesson et al. 1993; L. arcana: Reid 1996; L. compressa: Heller 1976; Reid 1996; L. obtusata: Goodwin and Fish 1977; Reimchen 1982; Trussell et al. 1993; Reid 1996; L. sitkana: Behrens Yamada 1992; L. subrotundata: Reid 1996 [less complete evidence]). This trend is commonly explained by intense crab predation in sheltered and moderately exposed habitats promoting large and robust shells (Elner and Raffaelli 1980; Johannesson 1986; Boulding
403
and Van Alstyne 1993; Boulding et al. 1999), while in wave-exposed habitats, small shell-size enables snails to escape friction from waves by hiding in crevices (Underwood and McFadyen 1983; Denny et al. 1985; Trussell et al. 1993). In the direct developing epiphytic littorinid, L. fabalis, size variation is different; shell-size decreases with wave-exposure as expected from moderately exposed shores to heavily exposed rocky shores (in areas where a rare exposed-shore morph is present; Rola´n and Templado 1987), while comparing sheltered shores and moderately exposed shores, shell-size instead increases (by 50–100%) with wave-exposure (Goodwin and Fish 1977; Reimchen 1981; Tatarenkov and Johannesson 1994; Reid 1996). This is despite the fact that L. obtusata, a sibling species of L. fabalis, shows a slight decrease in size from sheltered to moderately exposed habitats (Goodwin and Fish 1997; Reimchen 1982; Reid 1996). Both species dwell mostly in the canopies of fucoid algae where they graze directly on the algal thallus (L. obtusata: Reid 1996) or on epiphytic micro-algae (L. fabalis: Norton et al. 1990; Reid 1996) in contrast to most other littorinids, which predominantly stay on rocky surfaces (Reid 1996). The ‘‘Large-Moderate’’ morph (hereafter LM-morph) of L. fabalis present in moderately exposed habitats, and the ‘‘Small-Sheltered’’ morph (hereafter SS-morph) present in sheltered habitats, have a wide distribution over the shores of western Europe (Reimchen 1981, 1982; Reid 1996; Tatarenkov and Johannesson 1999). The small morph of heavily exposed shores is however only common in NW Spain (Rola´n and Templado 1987; Reid 1996) and will not be considered further here. Besides the disparity in size trends between L. obtusata and L. fabalis, there is also a general size difference between these species, such that L. obtusata is even larger than the large morph of L. fabalis (Goodwin and Fish 1977; Reimchen 1981, 1982; Reid 1996). This is most likely explained by different life-history optimizations (Williams 1992); in the North Sea, L. obtusata takes up to 2 years to reach sexual maturation and lives the subsequent 2–3 years to reproduce even more (Fretter and Graham 1980; Williams 1992) while L. fabalis usually only reproduces one season and few survive more than 1 year (Williams 1992, 1996). The shorter lifespan of L. fabalis probably promotes earlier maturation and thus a small size compared to L. obtusata. The latter instead delay maturation to increase female fecundity by growing large females. Crab predation and wave exposure are viewed as the two most important selective forces shaping the morphology and size of marine gastropods (e.g., Reimchen 1982; Underwood and McFadyen 1983; Etter 1989; Williams 1992; Boulding et al. 1999). The White Sea, where shore crabs are not present, makes an interesting comparison to the North Sea. Similar to elsewhere, L. obtusata is larger and lives longer than L. fabalis in the White Sea (Mikhailova, personal communication). On the other hand, L. fabalis is similar in size (average
adult size 6–7 mm) to the SS morph of Western Europe in both sheltered and moderately exposed habitats (M. Fokin, personal communication). This explicitly suggests that crab predation is a major factor structuring the size variation of L. fabalis in Western Europe, and that an adult size of 6–8 mm is probably favored for lifehistory reasons. An earlier study showed that the adults of the SS morph may easily be crushed by medium-sized (40– 60 mm carapace width) shore crabs (Carcinus maenas), while adults of the LM-morph are much more resistant to these attacks (Reimchen 1982). As crab predation seems to be an equally important mortality factor in both sheltered and moderately exposed habitats (this study) we propose that, the presence of a third factor, fucoid algae (besides crab predation and wave exposure), changes the effect of the two other factors by functioning as a refuge from crab predation. We predict that snails of L. fabalis are more protected from crab predation in the canopies of fucoid algae compared to the situation when, after being dislodged from the algae, they dwell on the rocks under the algae, and that dislodgement due to wave action increases with wave exposure. Thus, the overall risk of predation should increase with wave exposure and select for a larger shellsize in moderately exposed habitats, while in sheltered habitats shell-size should be small, reflecting early maturation of a relatively short-lived intertidal snail species. In the present study we test the predictions of differential survival and dislodgement of snails in habitats of different exposure and our results support the explanatory model outlined above.
Materials and methods Study area and collection of animals Field experiments were conducted on three small islands (Lo¨kholmen, La˚ngholmen and Salto¨; 0.3–1.5 km across), near Tja¨rno¨ Marine Biological Laboratory on the Swedish West-coast (5852¢N; 1109¢E). Snails used in the laboratory experiments were also obtained from these islands. Sampling-sites and experimental sites are referred to as either ‘‘moderately exposed’’ or ‘‘sheltered’’ to wave action (in still more exposed shores no fucoid algae are present and no snails of L. fabalis). Both habitat types are boulder shores with dense stands of fucoid macroalgae such as Fucus spiralis, F. vesiculosus and F. serratus, but sheltered sites were also occupied by the fucoid algae Ascophyllum nodosum, which is particularly sensitive to wave-exposure and an indicator of wave-protection (Lewis 1972). Indeed, we used the presence/absence of A. nodosum to define habitats as sheltered or moderately exposed, and we found that in all sites this distinction also separated habitats with large and small adults of L. fabalis. From the abundance of Ascophyllum and from geomorphological characteristics of the shore, we actively avoided sampling from within
404
in, or close to intermediate zones between the two morphs (Tatarenkov and Johannesson 1999). We used the width of the aperture (the distance from the junction between the lip and the body whorl to the furthest point of the anterior lip) as an indicator of overall size, as this measurement is more easily defined than shell height in this flat-spired species. Crab predation Old and repaired shell injuries of gastropod shells are thought to stand in proportion to the amount of shellbreaking predation on the populations (Raffaelli 1978; Reimchen 1982). Reimchen (1982) found repaired shell injuries to be very common in populations of L. fabalis in tidal shores of Britain and Ireland ranging up to 50% of the populations. Reimchen (1982) argued that the injuries were predominantly caused by unsuccessful attacks from C. maenas, since (1) the damages were indistinguishable from those seen in laboratory experiments with this organism, (2) other potential (rare) predators, such as intertidal fish, do not crush but instead swallow the entire shells, and (3) damages caused by the physical forces of waves and rolling boulders were not likely to have caused the damages since shell injuries were more common on sheltered shores than on exposed. To show that C. maenas is an important predator of L. fabalis, also at the Swedish west-coast, we conducted a similar survey of shell damages as Reimchen (1982). We collected 150 adult snails from four sheltered and four moderately exposed shores and calculated the proportion of shells with repaired shell injuries. Shell damages between sheltered and moderately exposed shores were tested with a one-way ANOVA. In addition, we estimated crab densities and sizes in both sheltered and moderately exposed areas. Crabs were captured at equal effort in two moderately exposed and two sheltered shores on two of the islands (Lo¨kholmen and 1 week later La˚ngholmen) at about 0.5 m depth. Width of the carapace was used as an estimate of crab size. We used traps that were able to catch crabs larger than approximately 33 mm in carapace width (smaller crabs were able to move through the netting). The traps were baited with 85 g of thawed herring that were placed in perforated tubes that allowed crabs to sense, but not eat, the bait. After 20 h, the trapped crabs were counted and measured. We used a three-way ANOVA to test for differences between moderately exposed and sheltered shores of the two islands. In this test, island was treated as a random factor orthogonal to the fixed factor exposure. The random factor site was nested under the interaction of exposure and island. Tethering experiments In two consecutive experiments, the consequences of snail dislodgement from the plant fronds in terms of
altered predation risk from crabs was simulated by tethering snails to the bottom substratum (stones) and as a comparison to canopies of fucoid algae. Tethering experiments have successfully been applied to investigate crab predation pressure in earlier experiments on marine snails (e.g. Williams 1992; Behrens-Yamada and Boulding 1996; Boulding et al. 1999). In this experiment, we predicted that the frond structure of fucoid algae constitutes a refuge from crab predation and thus that snails would survive better in the fronds than on the bottom. In the first experiment LM-morphs and SS-morphs were tethered to algae and stones beneath the algae in their respective habitat, that is, LM-morph in moderately exposed sites and SS-morph in sheltered sites. This test investigated the effect of crab attacks, while remaining in the canopy as compared to being dropped on to the rocky bottom in the snail’s own habitat. Moderately exposed and sheltered habitats are likely to differ in several aspects, and, in particular, in the degree of wave exposure. A difference in crab attack rate between the morphs in this experiment could therefore be due either to differences between the habitats or to differences between the two snail morphs, or both. To separate effects of morph and habitat differences, we conducted a second experiment in which we tethered both morphs in the same habitat. We used moderately exposed shores for this experiment, because it seemed more relevant to tether snails to the bottom substratum in the habitat where dislodgement was most likely to happen. We sampled 120 individuals of each morph for the first experiment. Only adult snails were used in the experiment. Adults (sexually mature individuals) were distinguished from juveniles on the basis of their morphology; juveniles have thinner shells than adults and the lip of the aperture extends outside the rest of the whorl (Reid 1996). The snails were individually marked using numbered, 2-mm-diameter, plastic tags that were glued to the shell. The tether, consisting of nylon monofilament fishing line, was attached under the plastic tag. Three moderately exposed and three sheltered test sites were chosen on Lo¨kholmen and La˚ngholmen. In each test site, four groups of five snails each were tethered to the canopy of different individual algae of F. vesiculosus and four groups were attached to large stones. From the five snails, the proportion of attacked snails was calculated and used as the dependent variable in the analysis. We assumed that all snails that were either partly damaged or completely gone had been attacked. One of the groups at each site was placed under a cage excluding predators from access to the snails. This was to estimate the rate at which snails disappeared by other factors than predation. The cages may, however, also have constrained some physical factor of the environment such as waves. Physical effects were, on the other hand, probably not that important since fewer snails had disappeared or been damaged in exposed than sheltered habitats, even though exposed shores were subjected to heavy wave action during the experiment. The snails were checked after 1 day and 5 days of exposure to crab predation. The
405
results were analyzed in a three-way ANOVA with the factors morph, position (fixed orthogonal factors) and site (nested under position). In the second experiment we changed the mode of tethering. Holes were drilled through the lips of the shells with a 0.6-mm drill, through which the tether (fishing line) was tied to the snails. The tags were tied to the fishing line. We changed the mode of tethering since this method was less time-consuming than the first one. The snails were checked after 2 days and 5 days of tethering, instead of 1 day and 5 days. We also changed the experimental setup so that site would include only one combination of morph and position, which made the interpretation more general. To get enough sites, we included a close-by island, Salto¨, in this experiment. No control group was used in this experiment, the reason being that only one of the control animals (1.7%) in the first experiment was missing after 5 days of tethering. In addition, a tethering experiment, where the method of drilling holes also was used, showed that all control animals were untouched after 74 days of tethering, except for one (Williams 1992). In both experiments, the snails that obviously had been damaged by the experimental setup were discarded. Some snails were, for instance crushed by the stone to which they were tethered. In addition, some replicates could not be found, and therefore the analyses were made on unbalanced data. Since we had measured size of all the tethered snails, we could analyze the relationship between size and the probability of being attacked by a crab. Here we used linear regressions that were applied both for separate morphs (data from the first and second experiments), and for pooled morphs (data from the second experiment only; these data could be pooled since the combinations involved the same times and habitats). The snails were divided into different size-classes in a way that each class had approximately the same number of snails. The dependent variable was the proportion of snails being attacked, that is, the number of attacked snails divided by the total number of snails in each size-group. Wave simulation and dislodgement of snails We hypothesized that snails living in our moderately exposed sites were at a higher risk of being dislodged than snails living in our sheltered sites. We initially attempted to test this in a field experiment, but it proved to be technically difficult due to the strong physical forces acting on the moderately exposed shores, and we were thus constrained to use a laboratory test. The effect of waves was simulated in a tank (105·550·40 cm) using a sled constructed from reinforcement rods and plywood that could run smoothly back and forth in the tank. Shock cords were attached to one side of the sled and the other side was anchored with a rope. As the sled was pulled back by means of the rope and released, both the backwash and the swash of a breaking wave could be
simulated in a consistent way. We consider this a realistic way of simulating the effects of waves as the force imposed on an organism by a breaking wave and postbreaking flows is due to both the water’s velocity and its acceleration (Denny et al. 1985). Three different levels of exposure were used a, b, and c by varying the number of shock cords, a being the lowest levels and c being the highest. Five plants in three rows (15 plants total) were attached to the sled at a distance of 15 cm between each plant. Five snails were allowed to attach to two plants (replicates), which were positioned as number two and four of the middle row. Eight additional plants thus surrounded each replicate plant. This was done to simulate field conditions where a snail most likely may be dislodged not only by drag from water currents but also by contact with other branches of the plant or adjacent plants. Plants from moderately exposed habitats were used for LM-morph snails and plants from sheltered shores were used for SS-morph snails. In the experiment, every trial (consisting of one level of exposure and one morph) was initiated by slowly moving the sled, back and forth, ten times so that the snails would become accustomed to water movements. After this the sled was pulled backwards and released at increasing distances. After eight of these pre-pulls, the sled was pulled back and released ten times at its maximum distance (=120 cm). Thereafter, the replicate plants were removed from the sled and the number of snails that had been dislodged was counted. Exactly when snails had been dislodged was not assessed. New replicate plants and snails were used for every trial. The whole experiment was repeated twice (during 2 days) and time of the experiment was used as an additional random orthogonal factor (besides morph and exposure) in a three-way ANOVA. A Student–Newman–Keuls test (SNK) was performed to distinguish how snails of different morphs were dislodged at different exposures. The probability of snail dislodgement is dependent in part on the maximum acceleration of the water. This can be quantified with a modified dynamometer originally designed by Jones and Demetropoulos (1968) and further improved by Bell and Denny (1994). Two of these modified dynamometers (hereby referred to as maximum velocity recorders [MVRs]) were built according to the design of Bell and Denny (1994). We used the MVRs to roughly calibrate the exposure scale generated by the water flow in the tank by comparing the recordings from the tank with those occurring naturally on exposed and sheltered shores. As the deflections on the MVR is just a measure of the maximum extension of a spring during a trial, we recalculated our MVR data to flow-speed (m s2) using the equation: E 10 þ 3:16 ; Flow speed ¼ 4:8 where E is the maximum extension of the spring in the MVR during a trial. This equation is from a study by Gamfeldt (unpublished) in which he dragged MVRs
406
behind a boat and registered the speed of the boat with a GPS. Speed of the boat was then plotted against the maximum extension of the spring to which the above equation was fitted. Recordings in the tank were obtained by attaching two MVRs (replicates) to a sled with algae using the same procedure as in the experiment with snails. Every level of exposure was replicated two times to see if the water flows that were generated in the tank were stable. We analysed the differences between the three levels of exposure with a two-way ANOVA in which trial was a random factor nested under the fixed factor exposure. Prior to this we also checked that the choice of plants (exposed or sheltered) did not have any effect on the level of exposure. This test was conducted by dragging and releasing the sled only once at each trial. Every trial was, in turn, replicated three times for three levels of exposure. A mean from the two attached MVRs was used as the dependent variable in the following two-way ANOVA (factors plant, fixed, and exposure, random). Recordings in the field were made during a day with strong wind (‡10 m s1) of a direction (SW) that generates maximum wave action in the study area. The MVR was left in the water fixed 90 towards the waves, at a depth of approximately 10 cm below the average water level, for 5 min at each time. Five recordings were made on sheltered shores and seven on moderately exposed shores and these were compared with a one-factor ANOVA. Statistics Prior to ANOVA, we checked all data for homogeneity of variances with Cochran’s test (Underwood 1997) and data-sets with heterogeneous variances were transformed. To increase the statistical power of the analysis, all factors were tested against pooled estimates of mean squares each time the original error terms proved to be highly insignificant (P>0.25; Underwood 1997). A posteriori tests (Student–Newman–Keuls) were performed when appropriate following Underwood (1997).
Table 1 Crab (Carcinus maenas) abundances and average sizes in wave-exposed and sheltered habitats Source of variation
Exposure (E) Island (I) E·I Site (I, E) Residual (R) Pooled (P)
df
1 1 1 4 16 20
Crab abundance
Crab size
MS
P
MS
P
459 805 1110 268
0.64 0.19 0.13 0.7 483 440
4.68 17.3 11.5 8.8
0.64 0.16 0.25 0.4 8.11 8.25
Error term
E·I P P R
Dependent: number of trapped crabs and carapace width
than 35, but crabs of this size could pass through the netting of the trap and the presence of small crabs could not be estimated. Mean crab carapace width for captured crabs ranged from 44.5 mm to 49.7 mm for the different sites, and mean numbers trapped varied between 19 and 57. After 20 h of trapping, no factor or interaction between factors, had significant effects on average captures of crabs or on average carapace widths (Table 1). That is, equally many crabs of about equal size distributions were trapped at moderately exposed and sheltered sites (Fig. 1). Tethering experiments For 25% of the snails that had been attacked shell fragments were still left attached to the tether, indicating a crushing predator (i.e., crabs). The remaining 75% were completely removed and tethers broken. It is most likely that also these snails were lost by predator attacks, as in our control cage excluding predators, only one snail (1.7%) was lost over 5 days of experiment. As argued earlier, crabs are the dominating snail predators, although we cannot exclude some impact by fishes and birds.
Results Crab predation Repaired shell injuries varied from 43% to 63% in the eight sampled sites, indicating a considerable effect of crab predation on different populations of L. fabalis. In addition, shell injuries were significantly higher on moderately exposed compared to sheltered shores (49% vs. 58%; F1,6=8.0; P=0.030) indicating a stronger impact of crab predation in moderately exposed shores. A total of 749 C. maenas were trapped. Of these, 2.9% (22) had carapaces that were 60 mm or wider, the largest measuring 68 mm. About 20% (146) of the crabs were smaller than 40 mm and only 2% (15) were smaller
Fig. 1 Mean number and size (carapace width) of captured crabs (C. maenas) from moderately exposed (e) and sheltered (s) sites (mean ± SE). No significant differences were found between habitats or between islands (Table 1)
407 Table 2 Differences in attack rates between snails tethered to algae and stone Source of variation
df
After 1 day MS
After 5 days MS
P
Error term P
Result of the first experiment in which we tethered LE-morph in exposed habitats and SS-morph in sheltered habitatsa Position (P) 1 0.15 0.023 0.09 0.14 Habitat (H) 1 0.51 Ccrit=0.392 after 2 days and 5 days, respectively) we power-transformed the data and this reduced
Fig. 2 a Attack rates when LM- and SS-morphs of L. fabalis were tethered to their own habitats. The interaction between habitat and position was significant after 1 day of tethering but not after 5 days (Table 2). b Attack rates when LM- and SS-morph snails of L.
fabalis were tethered in moderately exposed habitats. LM-morphs were less attacked after both 2 days and 5 days of tethering (Table 2). Numbers above bars indicate how many days the snails were tethered. Bars indicate SE
408
0.028x+0.45, r2=0.033, F1,6=0.21, P=0.67; b SS-morph: 0.024x+0.40, F1,6=0.18, P=0.69, r2=0.028; LM-morph: 0.0048x+0,079, r2=0.0035, F1,6=0.021, P=0.89; pooled: 0.072x+0.67, r2=0.35, F1,15=7.6, P=0.016)
Fig. 3 Proportion attacked snails as a function of snail size in the first (a) and second (b) tethering experiment. Each point represents a size-group consisting of eight to sixteen individuals (a SS-morph: y= 0.10x+1.4, F1,5=11, P=0.021, r2=0.69; LM-morph:
y= y= y= y=
the heterogeneity below the critical values (0.272 and 0.360, respectively). In this experiment, the effect of both site and the interaction between position and site were highly insignificant and we pooled mean squares of these factors as before (Table 2). The position of a snail (on stone or in algae) had no effect on the proportion of attacked snails after 2 days or 5 days and we found no significant interaction between morph and position (Table 2). After both 2 days and 5 days of tethering, however, a significant effect of morph was found, with the SS-morph being more frequently attacked than the LEmorph in both positions (Table 2, Fig. 2b). There was a large difference in attack rates between the first and the second experiment. Only 14% had been attacked in the second experiment compared to 56% in the first experiment. This was most likely an effect of weather conditions creating different water levels in this almost atidal area (maximum tidal range 0.3 m). During the first experiment, the fucoid belt was submerged most of the time as the daily mean water level was above normal during a period of southwest wind and low air pressure. During the second experiment, however, a high air-pressure and north to east winds resulted in an extremely low water level and the fucoids remained emerged most of the time of the experiment, resulting in fewer crab attacks, which most likely also resulted in a low power of the experiment. The probability of being attacked decreased with increasing size for snails of the SS-morph while not for the LM-morph in the first experiment (Fig. 3a). In the second experiment, none of the morphs indicated a sizeeffect when analysed separately, but after pooling data for the two morphs to get a wider size interval, we found a significant effect of size (Fig. 3b). This effect might, however, be confounded by differences between morphs other than size.
water flow (plant: F1,14=0.85, P=0.16 [pooled error term]; exposure*morph: F1,12=0.16, P=0.85) and thus an effect of snail morph would not be confounded by type of plant. The three different levels of exposure a, b and c that were used in the dislodgement experiment corresponded to a flow velocity of 0.90±0.06 m s1, 1.26±0.07 m s1 and 1.46±0.09 m s1, respectively
Dislodgement of snails The choice of plant (moderately exposed or sheltered) in the dislodgement experiment had no significant effect on
Fig. 4 a Water flows in the tank used in the wave simulation experiment (open circles). Level ‘‘a’’ is the lowest flow and ‘‘c’’ is the highest. Filled circles represent values that were obtained from the field during a moderately windy day of SW winds. b Dislodgement of LM and SS-morph snails of L. fabalis from plants of F. vesiculosus at different water flows (see (a)). There was a significant interaction between morph and exposure (Table 3). Bars indicate SE
409
(±95% C.I.; Fig. 4a). There was no difference in flow velocity between the two different trials (F2,6=0.47, P=0.71), thus water flows that were generated in the tank were consistent. The lowest level, a, was somewhat higher than the water flow recorded on the sheltered shore during a windy day (0.78 m s1), whereas the average flow on the moderately exposed shore during the same day (1.64 m s1) was higher than level c (Fig. 4a). That is, exposures generated in the tank were within the limits of the exposure on sheltered and moderately exposed L. fabalis habitats during a windy day. When we tested dislodgement rate of snails in different wave flows, time of experiment had no significant effect on dislodgement rate between morphs, but as expected, the overall rate of dislodgement increased with exposure though depending on an interaction between morph and exposure (Table 3; Fig. 4b). An SNK test revealed that more LM-morphs were dislodged at high exposure, c, compared to medium (b: Dobs=0.4> Dcrit=0.29) and low exposure (a: Dobs=0.35> Dcrit=0.24), whereas no difference was found between high (a) and medium (b) levels (Dobs=0
