MARINE ECOLOGY PROGRESS SERIES l Vol. 113: 257-269, 1994 I Published October 27 Mar. Ecol. Prog. Ser. Effects of flow-habitat on body size and reproductive patterns in the sea anemone Metridium senile in the Gullmarsfjord, Sweden Kenneth R. N. Anthony, Ib Svane The Royal Swedish Academy of Sciences, Kristineberg Marine Research Station, Kristineberg 2130, S-450 34 Fiskebackskil. Sweden ABSTRACT: In the Gullmarsfjord, on the west coast of Sweden, a variety of morphs of Metridium senile (L.) are found. A small form (pedal disc diameter 0.3 cm) is found in dense clones (90% coverage) in high-flow habitats at 2 to 5 m depth, and larger forms (pedal disc diameter = 2.5 to 3.5 cm) are found abundantly in subtidal low-flow habitats where body size increases with depth. In order to investigate the adaptive significance of body size and laceration in M senile from different current regimes, indi- viduals of a large and small morph were reciprocally transplanted between a low-flow (<5 cm S-') and a high-flow (>65 cm S-') subtidal habitat, and incidents of pedal laceration were monitored from August 1992 to January 1993. For comparison, 2 experiments (analogous to the field transplants) were carried out in laboratory flumes. The relative importance of sexual reproduction in the 2 habitats was investigated by comparing gonad sizes of mature females. The results of field and laboratory experi- ments showed that individuals from the high-flow habitat decreased their rate of laceration in low flow, but maintained a higher rate of pedal laceration than the low-flow population irrespective of site of transplantation, suggesting that properties of asexual proliferation possess both habitat-related and intrinsic components. Sexual reproductive allocation was positively correlated with body size and was significantly greater among females from the low-flow population When adjusted for a common body weight, however, differences in gonad sizes were insignificant. Frequent asexual replication of the genet and fertility at a small size (1 g) are argued to be important features in the high-flow habitat which compensate for the reduced reproductive output of individual anemones and eventually maxi- mize clonal fitness. KEY WORDS: Sea anemone Metridium senile. Flow . Body size - Pedal laceration . Reproduction INTRODUCTION includes several colour morphs, whereas the small form is mainly monomorphic, indicating differences in The sea anemone Metridiurn senile (L.) is a conspic- clonal diversity (Hoffmann 1976, 1986, Purcell 1977). uous and quantitatively dominant species in hard bot- A similar size-distribution pattern has been tom assemblages in the Gullmarsfjord on the west described for Metridium senile in the British Isles and coast of Sweden. It is particularly abundant at depths along the west and east coasts of North America, and from 0 to 15 nl but may be found in dense patches or attempts have been made to divide morphs into new clones in deeper waters. M. senile occurs in popula- species or subspecies (Stephenson 1935, Fautin et al. tions with different size-frequency distributions and 1989).I n the British Isles, 2 varieties of M. senile, dis- colour variations. Small, red-brown individuals occur tinguished by size, degree of lobing of the oral disc and at shallow sites in the fjord and dominate at certain shape while contracted, were recognized by Stephen- sites where the current velocity periodically reaches son (1935). According to Bucklin (1985), the large 100 cm S-'. Larger individuals are commonly found in morph (M. senile var. dianthus) is found in shallow sheltered localities in the Gullmarsfjord proper, pre- water while the small morph (M,s enile var. pallidurn) dominantly below 15 m depth, or at deeper sites in the is typically infralittoral and is only occasionally found archipelago at the mouth of the fjord. The large form sublittorally. Bucklin (1985) found no genetic evidence, O lnter-Research 1994 Resale of full article not permitted 258 Mar. Ecol. Prog. Ser. 113: 257-269, 1994 however, permitting a systematic division of M. senile, MATERIALS AND METHODS but concluded that the varieties in the Bntish popula- tions were ecophenotypes or morphological forms Study sites. GAsklAvan is a cove situated in the cen- determined by habitat conditions. tral part of the Gullmarsfjord with vertical rock walls As for numerous other species of sea anemone, facing NNE and reaching a depth of 70 m (Fig. 1).T he Metridium senile reprod.uces both sexually and asexu- cove is sheltered by tall emergent rock walls, and ally (Chia 1976). Sexual reproduction occurs annually wave exposure and current velocity are generally low. by shedding eggs and sperm into the water which From 0 to 15 m depth, Metndium senlle is the domi- results in free-swimming planulae (Gemmill 1920), nant hard bottom species, covering 20 to 50% of the while asexual reproduction occurs regularly by frag- substratum when unexpanded. At increasing depth, mentation of the pedal disc, a process known as pedal M. senile is replaced, with respect to actiniarian domi- laceration (Hammatt 1906, Chia 1976). nance, by Protanthea simplex, but dense patches and Several authors have noted the relationship between clones of M. senile are found. M. senile is most abun- flow velocity and clonal proliferation by pedal lacera- dant in the upper 15 m at a salinity of about 20%0 tion in populations of Metridium senile on the east (Baltic water). Below 15 m, the salinity increases to coast (Shick et al. 1979, Shick & Hoffmann 1980) and about 32%a (North Sea water). the west coast (Bucklin 1987) of North America. Two Strommarna is an area of narrow channels connect- general hypotheses have been proposed: populations ing the Gullmarsfjord with the Koljofjord (Fig. 1).T he of M. senile are structured either through differential channel chosen as an experimental slte (Ribergssund, indeterminate growth and rate of laceration in Fig. 1) is 15 to 20 m wide and has nearly vertical rock response to flow (Shick & Hoffmann 1980. Bucklin walls to depths of 2 to 8 m. The tidal amplitude on the 1987), or through selection of well-adapted genotypes Swedish west coast is small and generally does not (Hoffmann 1986). Shick et al. (1979) and Shick & Hoff- exceed 30 cm, but in the channel the current velocity is mann (1980) suggested that different size-frequency usually high. The current is irregular, bidirectional and distributions is an accommodation to morphological generated by tides, winds and changing air pressures constraints brought about by differential food avail- and may periodically reach 100 cm S-'. The rock walls ability associated with the position of the tentacle are dominated by the blue mussel Mytilus edulis, crown within the boundary layer (Koehl 1977, Shick et numerous ascidians and large patches of sponges al. 1979),a nd furthermore proposed that genotype is a (Halichondna panicea and Haliclona sp.). A dense determinant of success in a given flow environment. population of small (3 to 4 cm high), red-brown Metnd- When comparing populations of small intertidal M. ium senile is found on the rock walls and on mussel senile with harbor float populations of large anemones, beds on the bottom, covering almost 90 % of the latter. Bucklin (1987) supported the hypothesis that size-fre- Field observations. The size distribution of Metnd- quency distributions of M, senile reflect different phys- ium senile at the 2 sites and at different depths was ical constraints and food availability in different flow estimated by quadrat sampling. At GAsklbvan, a 20 m regimes, as well as adaptive responses to these habitat transect line was deployed horizontally at each depth conditions. In his review, Shick (1991) provides a level (2 to 5, 10, 15 and 20 m) and 3 quadrats each model considering both the mechanical and nutritional measuring 50 cm by 50 cm were haphazardly selected. effects of current speed on body size and clonal growth In each quadrat, the anemones were gently touched in M. senile. to provoke contraction, the number of specimens According to Bucklin (1987), gonad size is positively counted, and the pedal disc diameter (pdd, mean of correlated with body size in Metridium senile. If flow- largest and smallest basal diameter) measured using habitat affects the size-frequency distribution and pop- plastic calipers. At Strommarna, the size distribution ulation structure of M. senile, either by stimulating was recorded only at 2 to 5 m depth, since this IS the pedal laceration or by favouring well-adapted geno- actual bottom depth range. types with high intrinsic rates of laceration, it may Transplantation experiments in situ. In July 1992, a influence both the sexual and asexual components of 3 m long aluminium bar was mounted horizontally genet fitness. The purpose of this study is to compare approximately 10 cm from a NNE facing vertical rock the patterns of clonal growth (laceration) and sexual wall at 5 m depth in RBbergssund (Strommarna) and at reproduction in 2 populations of M. senile in habitats of 15 m depth at Gdskldvan (the Gullmarsfjord; Fig. 1). high and low current regimes, and furthermore, to The rod was kept in position by means of 3 dowels investigate whether frequencies of pedal laceration plugged into holes drilled into the rock walls (see are regulated by the flow habitat, indicating pheno- Svane 1988). Individuals of Metridium senile were typic plasticity, or due to intrinsic factors within popu- haphazardly collected at each site by diving. The lation~. anenones were removed from the substratum using a Anthony & Svane. Flow-habitat and reproductive patterns in Metridlurn senile 259 (1984). The pseudo-replicates were, however, considered to be independent due to a space of minimum 10 cm between the panels. Random and independent deployment of individual panels at the 2 sites would have been prefer- able but was not practically possible. Before transplantation, an additional dowel was mounted 3 nl from the end of the panel .,,., rack at each site along the same isobath, which provided support for an additional aluminium rod for stereophotography. Below the rod, 6 contiguous quadrats, each measuring 50 by 50 cm, were selectively and loosely cleared of protruding epibioses, leaving spaced individu- als of Metridium senile visible on photographs. At monthly intervals, all panels and quadrats were stereophotographed using the technique described by Svane (1988).O nly the side of the panels facing away from the rock wall was photographed. The stereophotographs were analyzed in 2 aligned stereomicroscopes Fig. 1. Overview of the Gullmarsfjord on the west coast of Sweden (Svane 1988), one equipped with a calibrated and the study sites. The sites at Gdsklavan and Strommarna are ocular micrometer. indicated by arrows. The inset shows the exact location of the moni- toring site at Strutsholm in Strommarna The transplantation experiment was initiated in mid July 1992 and terminated 7 October (Expt I), and repeated from 13 October to 9 thin-bladed, blunt spatula in order to minimize tissue December 1992 (Expt 11).S elective clearing of quadrats damage. The individuals were collected from as many was not repeated, but monitoring of quadrats contin- clones within each population as possible by sampling ued during part of Expt 11. widely spaced individuals (Shick et al. 1979, Hoffmann Laboratory experiments: laceration at constant lev- 1986). In the laboratory, specimens from each popula- els of low and high flow. In mid May 1992, approxi- tion were transplanted onto 12 asbestos panels each mately 100 individuals from each of the populations at measuring 40 by 40 cm. The anemones were allowed GBsklAvan and Strommarna (using the sampling tech- 48 h to establish themselves on the panels submerged nique and considerations mentioned earlier) were col- on seawater tables. The panel populations were con- lected, established on acrylic panels and subsequently structed to reflect size-frequency distributions equal transferred to 4 laboratory flow tanks supplied with to those of the natural populations (Kolmogorov-Smirnov 2-sample test, p > 0.05).A fter 48 h, 6 panels each contained Strijmmarna a population of small M, senlle from RBbergssund and 6 panels each carried a population of large M. senlle from GBsklAvan. The panels, submerged in seawater, were then transferred to the experimental sites and mounted aligned and rigidly below the rods in randomized G.sklavan order. At each site, 3 panels with approx- imately 20 individuals from GAsklAvan and 3 panels with approximately 30 indi- viduals from Strommarna were sus- pended from the aluminium rods (Fig. 2). Since the panels were arranged on the same rod and were used in Fig. 2. Experimental design of transplantations of panel populations of Metri- the each panel a dium senile at Gdsklavan and StrBmmarna. Panels were mounted on the alu- pseudo-replicate according to Hurlbert minium bar in random order. Each panel measured 40 by 40 cm 260 mar Ecol. Prog. SE seawater in a closed circuit. About 50 individuals from tory experiments were similar, the data were pooled, each population were transplanted onto four 90 by whereas the field expenments were treated separately. 30 cm acrylic panels and were allowed 48 h to establish Sexual reproductive cycle: oocyte size-frequencies. themselves prior to deployment in the flow tanks. Size- From the populations at Strommarna and Gbsklbvan, frequency distributions of panel populations were 20 and 50 individuals respectively were collected at determined as in the field. Separate panels with monthly intervals from May to October 1992. The same attached anemones from GAsklbvan and Strommarna sampling technique and considerations mentioned were thereafter mounted in four 40 1 flow tanks where earlier were applied in order to maximize the number 2 tanks were preset at high (>25 cm S-') and 2 at low of clones representing each population. The specimens (<5 cm S-') flow, respectively. The flow through the were transferred to beakers where an isotonic solution tanks was maintained with powerful circulation pumps of MgC1, was gradually added in order to relax the (Sacem 1200),p roducing an even (though not laminar) musculature and allow fixation of the anemones while flow. The low-flow tanks shared a circulation pump set fully extended (Manuel 1988).W hen the anemones no at ?4 the power of the high-flow tanks. All tanks were longer responded to mechanical stimuli, they were aerated and the water filtered through gravel filters transferred to Bouin's fixative, and a proportional (see Moe 1982). In all tanks, the panels were kept ver- amount of the fixative was injected into the coelen- tical and parallel to the flow. The animals were fed live teron. Approximately 72 h later, the anemones were Artemia nauplii every 48 h. and every 2 wk 'h of the transferred successively to increasing concentrations water was renewed. Asexual reproduction (laceration) of ethanol and were finally stored in a 70% solution. of the 4 cross-transplanted populations was monitored From approximately mid-column, a 3 to 5 mm thick weekly for 2 mo by photographing the panels. The transversal disc was dissected from each female, since size-frequency distributions (0.5 cm intervals of pedal gonad size is largest in this region (Bucklin 1982).A fter disc diameter) of the panel populations before trans- dissection, individual preserved wet wt (ww) was plantation were compared to distributions after trans- determined after draining the anemone (including the plantation in order to examine effects of exposure. By dissected disc) on paper towels for 1 to 2 min. The regular census of lacerates and potential parents, transversal discs were transferred to 2,2-dimethoxy- mean monthly rates of asexual reproduction were propane to reduce shrinkage of the tissues during the determined analogous to the transplantations in situ. succeeding paraffin-imbedding (Prentn 1979). The The trial was repeated during late October 1992 to discs were subsequently sectioned 8 vm on a micro- early January 1993 with new panel populations. tome and stained with Ehrlich's Hematoxylin-Eosin. In Data analysis. The monitored panels and quadrat order to estimate the developmental stage and time of populations were scored for parent size (pedal disc spawning, the largest and smallest diameters of 100 diameter), number of parents and number of new indi- oocytes were measured from each female using a cali- viduals produced by laceration between dates of brated ocular micrometer (see Giese & Pearse 1974, photographing. Only lacerated pedal disc fragments Bucklin 1982, Grant & Tyler 1983a). The measure- no longer connected to the parent were counted. Mean ments were done on different sections and gonads monthly rates of laceration were calculated as: from each specimen as a precaution against heteroge- nous development within an individual (Grant & Tyler Laceration rate = -3 0 (1) 1983~11a,n d only oocytes with a nucleolus in the section Npar At were measured (Grant & Tyler 1983b). Oocyte size- where AN,,, = number of lacerations produced frequencies were determined based on females from between days of monitoring, Npar= mean number of each sample. Only individuals larger than 2 cm pdd parents on the panels during that period, and At = were included in order to exclude immature individu- number of days between monitoring. Calculating the als from the analysis (Grant & Tyler 1983a). Median mean rate of laceration based on intervals between oocyte sizes were tested for differences in location monitoring reduces biases from lacerate mortality and between the populations by the non-parametric parent migration compared to a calculation based on a Kruskal-Wallis test (Sokal & Rohlf 1981). census before and after the experiment. Sexual reproductive allocation. To estimate the rel- At the termination of each experiment, the rates of ative contribution to sexual reproduction of the 2 pop- laceration of the transplanted parent populations were ulation~,tr ansverse sectional gonad areas (TSGA) of compared reciprocally by an analysis of variance female gonads as a function of preserved wet body wt (ANOVA) followed by the Student-Newman-Keuls (see Bucklin 1987) were used. Measurements of TSGA (SNK)m ultiple comparison test (Sokal & Rohlf 1969).A were made by means of a calibrated ocular micrometer significance level of a = 5 % was used throughout the on sections from the mid-column of 10 mature females study. Since the conditions for the 2 successive labora- from Gdskldvan and 8 mature females from Strom- Anthony & Svane: Flow-habitat and r- eproductive patterns in A4etr1diurn senile 261 marna collected prior to spawning. To exclude irnma- of plaster balls was used as a measure of the mean cur- ture females, only individuals with a pedal disc diame- rent velocity irrespective of direction of flow. Plaster ter larger than 2 cm (ww -0.3 g) were included. balls moulded on metal wires were weighed dry before Largest and smallest diameter of each ovary in each exposure, mounted at a distance of approximately section were measured and the total cross-sectional 10 cm from each panel, and collected after 24 to 96 h of gonad area was determined from the sum of gonad exposure. The plaster balls were then dried for 72 h at areas. This method provided a relatively coarse esti- 60°C and the weight loss per 24 h of exposure calcu- mate of ovary area because not all transverse discs lated. Six batches of 3 balls each were exposed for 6 to were intact, and in these latter cases total ovary area 24 h in a flume at various, constant current velocities was estimated by extrapolation. Since gonad size is not measured by thermistor probes to estimate weight loss a linear function of adult size in Metridium senile as a function of flow velocity. The resulting calibration (Bucklin 1987), sexual reproductive allocation could curve was used to determine the approximate mean not be compared using gonad indices based on body current velocity from the weight loss of plaster balls size (e.g. Bucklin 1982, Giese & Pearse 1974).I nstead, exposed in situ. TSGAs were compared between sites using an analy- sis of covariance (ANCOVA) on log-transformed data with preserved wet wt (ww) as the covariate (TSGA = RESULTS a wwb; Sokal & Rohlf 1981, Grant & Tyler 1983a). Conversion of preserved wet weight to pedal disc Field observations area. For use in interconverting pdd and ww, a conver- sion curve was constructed. The attached pdd of 95 The densities of the natural populations of Metrid- individuals from harbor pilings at the Kristineberg ium senile at the 2 sites were estimated to be 470 & 21 * Marine Research Station were measured individually (SE,n = 3) ind. m-2 at 5 m depth, decreasing to 104 19 (mean of largest and smallest pedal disc diameter) (n = 3) ind. m-2 at 20 m depth, at GBskldvan and 25 000 before collecting. The specimens were grouped into i 2200 (n = 3) ind. m-2 at Strommarna. Estimates are pdd size classes of 0.5 cm intervals, preserved in 10 % based on a photogrametric analysis at GasklBvan (see formalin and washed in 70 % ethanol48 h later. Corre- Svane 1988) and in situ recordings of anemones in sponding values of pdd and ww were plotted in a log- 0.25 m2 quadrats, haphazardly selected on the bottom log graph (pdd = awwb), and the relationship was of the Rdbergssund channel in Strommarna. tested by linear regression analysis (Fig. 3). The size-frequency distributions of pedal disc diam- Flow measurements in situ. In order to quantify the eters in the 2 studied populations, at 2 to 5 m depth in accumulated flow exposure experienced by the panel Stromrnarna and at 5, 10, 15, and 20 m depths at populations in the field, a technique developed by GBskldvan, are shown in Fig. 4. Comparisons of distri- Muus (1968) was applied. Weight loss per unit of time butions between sites at 5 m (2 to 5 m) were signifi- cantly different (Kolmogorov-Smirnov 2-sample test, p < 0.01). Median pdd at Strommarna was only 0.3 cm, with more than 80% of the population consisting of individuals <0.5 cm pdd, which were mainly lacerates not yet regenerated. Only 4 % had a pdd >l cm, and the largest size class (2.0 cm + pdd 12.5 cm) was only represented by 2 individuals. At GBskldvan, a marked difference in size-frequency distribution in relation to depth was also observed (Kolmogorov-Smirnov 2- sample test, p < 0.01, Fig. 4). Small anemones domi- nated at depths of 2 to 5 m and 10 m (median pdd 0.7 and 1.3 cm, respectively), whereas larger ane- mones became more dominant with increasing depth (median pdd 2.5 and 3.5 cm at 15 and 20 m, respec- l tively). Flow velocity at the panels was considerably -I,- 0.1 1 - 7 1 --cmr+. 10 I I l l l.,1:0,0 higher at Strommarna than at GBsklAvan (15 m, Table Wet weight (g) 1). Additional flow measurements along the vertical Fig. 3. Metridium senile. Log-log regresson of pedal disc transect at GBskldvan showed no significant difference diameter and wet weight of 95 individuals. Attached pdd was in flow velocity between the 4 water depths (not measured before collecting the animals. The expected power shown). At Strommarna, the mean flow velocity was function relating pdd to preserved wet wt (ww)f itted signifi- cantly to the data (Fs= 1921.4, p < 0.001) higher than 65 cm S-' at all measurements, which was 262 Mar. Ecol. Prog. Ser. 113. 257-269, 1994 Gisklivan l0 m Gisklivan 15 m Gisklivan 20 m Pdd size classes (cm) Fig. 4 Metridium senile. Size-frequency dstributions of pedal disc diameter (pdd)i n natural populatlons from Strommarna at 2 to 5 m and from Gaskldvan at 5, 10, 15, and 20 m, respectively. Body size was larger at GAsklAvan (5 m) compared to Strommarna, and increased with depth at GAsklAvan (all significant by the Kolmogorov-Smirnov 2-sample test, p < 0.01) the maximum obtainable velocity in the flume during patches and heavy colonization by colonial ascidians calibration of the plaster balls. During 2 surveys at and sponges, eventually occupying more than 50 % of Strommarna, flow velocity was estimated to be higher the quadrat areas after 2 mo and consequently ob- than 100 cm S-'. Mean flow velocity at GAsklbvan was scuring the view. 3 cm S-', never exceeding 6 cm S-'. Transplantation experiments in situ Laceration on natural substratum The results of the 2 field transplantation experiments On the 6 quadrats of selectively cleared natural are shown in Fig. 5A, B. Strommarna individuals con- substratum at GAsklbvan, pedal laceration was virtu- sistently produced more lacerates. When Strommarna ally absent. Only 2 lacerates among 62 potential individuals were transfered to GAsklBvan, the rate of parental individuals were observed from August 19 to laceration decreased significantly, whereas laceration October 27, giving a final ratio of produced pedal lac- rates were unaffected among GAsklbvan individuals erates per potential parental individual of 0.03. In when transplanted to Strommarna. The ANOVA Strommarna, it was unfortunately not possible to mon- showed a significant effect of treatment in both exper- itor individual pedal laceration on natural substratum iments (Table 2). In all panel populations, it was due to the migration of mussels into the cleared observed that predominantly the larger anemones were undergoing frequent laceration. Differences in size-specific amount of laceration within panel popula- Table 1. Weight loss per 24 h of plaster balls deployed at the tions could not be quantified, however, as locomotion 2 sites and the corresponding mean current velocity Flow speeds higher than 65 cm S-' could not be measured, since and frequent crossing of trails of lacerates obscured this was the maxlmum veloc~tyIn the flume dunng calibration the analysis. At both sites, fragmentations of the pedal disc were Exposure Gdskldvan Strommarna discrete, periodical events. Jt was observed that an Mean Current Mean Current anemone produced up to 15 lacerations (new modules) weight loss velocity weight loss velocity in less than 2 wk. Individuals typically extended the (g L4 h ') (cm S-') (g 24 h-') (cm ss') pedal disc to cover a large area (compared to the area - of the tentacle crown), and detached fragments In a 1 3 65 5 7 21.4 > 65 2 3 29 4.7 18.9 > 65 ring around the pedal disc. Wlthin 1 wk, the ring sepa- 3 3.08 4.2 18.5 > 65 rated into a larger number (10 to 15) of regeneratmg 4 2.60 3.1 15.2 > 65 units. Some anemones showed extensive locomotion, 5 2 11 2 1 leaving a trail of fragments on the panel, but other Anthony & Svane. Flow-habltat and reproductive patterns ~nM etnd~ums enile 263 A - Experiment I B Experiment II . Fig 5. Metrjdium senile. Two recipro- j cal transplants at Str6mmarna (S) and Gdskldvan (G). The figure shows 2 4 monthly mean rate of laceration per * 5 individual 95% confidence limits of 3 n = 3 panels. Underlined transplants are not significantly different by the 5' 2 SNK test. Expt I (A) was run July to U October 1993 and Expt I1 (B) was run October to December 1993. Transplant initials denote site of origin and site of exposure, respecbvely Transplant Transplant anemones were stationary throughout the period of the Sexual reproduction: period of spawning and experiment and were not seen lacerating. Size fre- reproductive allocation quency distributions of any panel population did not change significantly when transplanted to an altered A total of 140 individuals were collected at Gbsklb- flow habitat (Kolmogorov-Smirvov 2-sample test, p > van and 350 at Strommarna for sexual reproductive 0.05). analysis. More animals were collected from Strom- marna, as only a few clones apparently dominated Laboratory experiments: laceration at constant levels there, and the first samples indicated a skewed sex of low and high flow ratlo with a preponderance of males. Despite exten- sive sampling at Strommarna, it was not possible to Individual anemones from Strommarna produced obtain more than 2 females in most samples and no more than twice the amount of lacerations per parental females were found in the October sample. The sex individual during the experimental period than did ratio at Strommarna was found to be 1:8 compared to individuals from Gbskltivan, irrespective of flow 1: 1.4 at Gbskltivan. However, the estimated sex-ratio regime (Fig. 6). The analysis of vanance showed a sig- at Strommarna was based on large individuals only. nificant treatment effect (Table 2), and the SNK-test Sampling at Gaskldvan was more representative with revealed that Strommarna individuals produced signif- a mean preserved wet wt of 10 g and a pedal disc icantly more lacerates than Gaskldvan individuals diameter of 4 cm, as this size class dominated the (Fig. 6). population. Table 2. Metridium senile. One-way ANOVA of laceration data. 'p < 0.05; ''p > 0.01 Source of variation df SS MS Fs p Expt I Transplant 3 16.13 5.38 7.63 0.0.0. 9 8 Error 8 5.63 0.70 Total 11 21.76 Expt I1 Transplant 3 9.11 3.04 10.16 0 0.0. 4 2 1 Error 8 2.39 0.30 Treatment Total 11 11.50 : Fig. 6. Metndium senile. Laboratory expenments. GiSskldvan Laboratory (G) and Stromnlarna (S) panel populations exposed to high Transplant 31.45 10.48 9 06 0.0295 (H) and low (L) flow velocity in flow tanks. Mean monthly Error 4.63 116 rates of lacerabon per individual k 95 % confidence lim~tosf 2 Total 36.08 successive expenments are shown Groups with a common underl~nea re not slgnlficantly d~fferenbt y the SNK test 264 Mar. Ecol. Prog. Ser. 113: 257-269, 1994 G Askl Avan B Strommarna Fig. 7. Metridium sen~leO. ocyte size-frequency distnbut~ons of monthly samples from (A) GAS- kldvan and (B) Strommarna, May-October 1992. The number on top of each polygon is the number of females In the sarn- ple. Arrows indicate spawning. Median ooryte diameter was significantly larger at Gaskldvan (Kruskal-WaUis test, p = 0.046) In the GBsklAvan population, mature oocytes were occurred 1 mo earlier (July).T he median oocyte diame- present from the begining of June until September ter of monthly samples was significantly larger at (Fig. ?), and spawning occured in late August and early GAsklBvan compared to Strommarna females (Kruskal- x'~ September. At Strommarna, spawning apparently Wallis test, = 3.98, p = 0.046; Figs. 7 & 8). Unfortu- Fig. 8. Metridium senile. Transverse section of ovary from a GasklAvan female (left, wet wt = 15 g) and a Stromrnarna female (right, wet wt = 2.6 g),c ollected prior to spawning (July and August, respectively). The sectional area of the ovary and percent age of mature oocytes (> 100 pm) is larger for the GAskIavan female. Photos: Jsrgen Liitzen Anthony & Svane: Flow-habitat and reproductive patterns in Metridiurn senile 265 Table 3. Metridiurn senile. ANCOVA of transverse sectional 1.9, respectively, indicated that transverse sectional gonad areas. ns: not significant gonad area nearly doubled with increasing increments of preserved wet wt. Mean gonad index (based on Source of variation df SS MS Fs p TSGA/pedal disc area) was therefore significantly l larger among the Ggsklbvan females compared to Log (adjusted means) 1 0.12 0.12 1.47 0 2461"' Strommarna females (Fig. 8). Error [deviat~ons 15 0.27 0.08 I from a common slope) l Strommarna GBsklAvan DISCUSSION Estimated* log (adjusted 0.97 i 0 l1 0.65& 0.21 The size of Metl-idium senile can be related to its means) SE habitat. The small morph, or variety, previously named M. senile pallidurn (<2.5 cm pdd; expanded column height <-5 cm), is found in areas of high current nately, no sampling was done in Strommarna in late regimes but also at shallow depths along the sheltered June or early July. In that period, samples may have rocky shores on the Swedish west coast (Fig. 4). The contained females with a greater proportion of larger large morph, previously named M. senile dianthus oocytes and the onset of spawning may consequently (>2.5c m pdd; expanded column height >5 cm), which have occured earlier than at the end of July. Only 2 fe- may reach an expanded individual column height of males (small individuals <4 g) were scored as infertile in more than 30 cm, is found at greater depths. No indi- the GbsklAvan samples, while a large proportion of the viduals are found intertidally since tides are negligible samples from Stronlmarna were infertile, particularly along the Swedish west coast. individuals smaller than 0.5 g. Spent females were rec- The observed differences in body size may be geneti- ognized in the August samples from Strommarna and in cally determined (Fautin et al. 1989). Genotypic differ- the October samples from GBsklAvan. ences may interact with environmental factors, how- The 2 populations of females were not different in ever, to produce the observed differences in body size terms of transverse sectional gonad area (TSGA)w hen and population structure (Shick et al. 1979). Several adjusted for a common body weight (Table 3, Fig. 9). hypotheses, not necessarily mutually exclusive, have Linear regression of log(TSGA) on log(ww) was highly been suggested to explain habitat-related differences in significant for both samples (Gbskldvan: FS(~=, ~63). 2, size-distributions and have been reviewed by Shick p < 0.001; Strommarna: Fs,,,~=, 19.80, p = 0.004), and (1991):( 1)I n fast current habitats, size is constrained by the slopes did not differ significantly (FS,~=, 0~.0~53), mechanical limitations of the anemone's body plan p = 0.82), thus meeting the assumptions for the (Shick et al. 1979, Shick & Hoffmann 1980). (2) Asexual ANCOVA. The power functions (TSGA = a wwb) pro- reproduction (laceration rate) increases with current vided a good fit to the data (Strommarna: r = 0.88, speed resulting in a small body size (Shick & Hoffmann : GgsklBvan: r = 0.94; Fig. 9) and exponents of 1.8 and 1980, Bucklin 1987). (3) Differential predation by the nudibranch Aeolidia papillosa on small individuals - affects the size-structure of anemone populations "E - // (Harns 1973, 1986, Shick et al. 1979). ,, - G& TSGA - 0.35 W , R -0.94 The results of the transplantation experiments m F! Str o TSGA 0.63 W l.'' R '0 .88 in situ indicated that Strommarna animals pro- 'Dm 10 - duced consistently more lacerations than GBskli- 8 van individuals in high flow, but reduced their as rates of laceration when transplanted to the low- - 1 flow habitat. Conversely, Ggsklbvan individuals 5 I 4 did not increase laceration rates when trans- 0.1 4 m O planted to the high-flow site in Strommarna (Fig. zV) 5A, B). A relatively high mean laceration rate was C 0.01 ! ,.. ., maintained by the transplanted Strommarna indi- 0.1 1 10 100 viduals but due to the large variation among Wet weight (g) (pseudo-) replicates this was not significantly dif- Fig. 9. Metridiurn senile. Log-log regression of transverse sectional ferent from ~ ~ ~ indkividulals, ;=he~ labo~ra- ~ gonad area (TSGA) of 10 females from GBsklAvan and 8 females tory experiments showed a slightly different pat- from Strommarna, sampled prior to spawning, on preserved wet ~ v t tern where Gbskldvan individuals maintained low (WW) Difference in TSGA was insignificant between populations when adjusted for a common ww (see Table 3) laceration rates while Strommarna individuals 266 Mar. Ecol. Prog. Ser. 113: 257-269, 1994 maintained high laceration rates irrespective of flow als of M. senile by A. papilJosa should influence the size- regime (Fig. 6). However, within the results of the lab- distribution of anemones by increasing the mean size as oratory experiments there is a tendency, although not suggested by Harris (1973, 1986),s ince the predator is statistically significant, for the Strommarna individuals rarely found among larger anemones. to decrease their laceration activity when kept in low The observed patterns of size-distribution in Metrid- flow. The conclusion is, accordingly, that laceration ium senile in the Gullmarsfjord are to some extent in rates are influenced to some extent by current speed, accordance with observations made in North America but that Strornmarna individuals maintain a higher lac- and the British Isles. However, these patterns have eration rate indicating an intrinsic difference between been related to distribution in intertidal versus subtidal populations. Whether this difference is genetic or due environments (Bucklin 1987) or high versus low cur- to phenotypic plasticity coupled with a time lag is, rent regimes (Shick et al. 1979, Shick & Hoffmann however, not conclusive from our results. If rate of lac- 1980). In the sheltered Gullmarsfjord, a pattern of size eration is only a function of flow exposure on the distribution, where small individuals occur in the high anemone body plan, Gdskldvan anemones would have subtidal, and where the mean body size increases with been expected to produce more lacerates in high flow depth, can be observed (Fig. 4). Tidal currents are vir- than Strommarna individuals in low flow, due to the tually absent and turbulence by wave action is limited proportionately greater drag in large individuals to the surface thereby reducing the plausibility of cur- (Wainwright & Koehl 1976. Koehl 1977, 1982). In addi- rent velocity as the sole factor explaining the observed tion to the different responses to flow, there seemed to size-distribution pattern. Since current velocity and be a marked effect of transplantation to a new substra- other physical factors (temperature, salinity), as well as tum per se on the rate of laceration, at least within the fouling rate (Svane pers. obs.), are influenced by site G5skldvan population. GAs+GAs panel populations and depth, a likely general explanation for the produced a significant number of lacerates relative to observed size-distribution pattern is that stress stimu- the almost negligible laceration rate exhibited by the lates pedal laceration, subsequently reducing the undisturbed anemones on the rock wall (only 2 lacer- mean body size. This is supported by the fact that ates per 62 adults during 2 mo), suggesting that the manipulation and transplantation to an artificial sub- change of substratum produced a larger increase in stratum (a type of physical stress or perturbation) had a laceration activity than the change of flow habitat. marked effect on laceration, and by the results of Wahl In the Gullmarsfjord area, a marked abundance of (1985) who showed that individuals of M. senile trans- small sized Metndium senile is associated with fast planted to horizontal panels initially migrated, leaving current habitats. However, small individuals are also trails of lacerates prior to reestablishment. found abundantly in shallow sheltered sites where the Based on its colour-monomorphic appearance and mean body size increases with depth (Fig. 4). Wave a sex ratio of 1 female to 8 males, indicating a low exposure of some degree may cause a higher mean genotypic diversity, the Strommarna population pre- flow velocity in these habitats compared to deeper sumably consist of only a few clones proliferating sites inhabited by large morphs, but only close to the extensively by clonal growth (see Williams 1975, Hoff- surface (1 to 2 m). In the fast current sites at Strom- mann 1976, 1986, Shick 1991). Populations of antho- marna, current velocities above 100 cm S-' are com- zoans with predominantly sexual reproduction are mon and the small morph dominates. Large morphs expected to have high genotypic diversities and a sex transplanted to Strommarna managed to expand their ratio of 1:l (Shick 1991). This seems to be the case for tentacle crowns, apparently feeding, but the column the Gdskldvan population at 15 m depth where the sex was kept short and within the height range of the small ratio was found to be 1: 1.4 with a slight dominance of morphs, indicating some mechanical constraints. The males. At this site several colour morphs are found, experiments were performed in August to October but although the brown morph dominates at 15 m of depth the panels with the transplanted anemones were left and above. and revisited during May the following year. At Strom- Individual Gdskldvan females produced significantly marna, the individual animals were still on the panels larger amounts of mature oocytes than Strommarna and no reduction in size of the large GdsMdvan females (Figs. 7 & 8). The size at which oocytes of anemones was evident. At Gdskldvan, only few ivletridium senile have been reported to be ripe is 100 anemones were still located on the panels in May. to l73 pm (Cernrnill 1920, Rawlinson 1934, Bucklin The predatory nudibranch Aeolidia papillosa is found 1982, Shick 1991).O ocytes larger than 100 pm in diam- abundantly in the high current site in Strommarna, eter were found only in late July at Strommarna and where it feeds vigorously on Metridium senile. At the constituted less than 2% of the oocytes in the sample sheltered Gdskldvan it is rarely found. It seems there- (Fig. ?B), whereas oocytes above that size made up 5 to fore unlikely that differential feeding on small individu- 14 % of the samples from July to September at Gdskla-