Valve-gape response times in mussels (Mytilus edulis)--effects of laboratory preceding-feeding conditions and in situ tidally induced variation in phytoplankton biomass.

ABSTRACT Many physiological processes in bivalves are influenced by the valve opening sate and thus the valve opening-closing response times of the experimental animals. Controlled laboratory studies using underwater video camera recording of valve-gape responses of mussels (Mytilus edulis) to presence or absence of algal cells in the ambient water have revealed that valve opening and closing responses are strongly influenced by the preceding feeding conditions. The observations indicate that the period during which mussels may "learn" or "forget" to respond to the presence or absence of algal cells can last for weeks in the laboratory. Further, the critical algal concentration below which mussels close their valves has been identified to be about 700 Rhodomonas sp. cells [mL.sup.-1], or 0.9 [micro]g chl a [L.sup.-1]. Such knowledge is essential for making allowance for possible deviation from natural-response times in mussels kept in the laboratory previous to experimental physiological studies. In situ simultaneous registrations of water flow, chl a and valve-opening degree of mussels were made within a dense mussel bed in a shallow, tidally dominated inlet to a Danish fjord. Phases of strong tidal flow brought ample supply of phytoplankton to the mussel bed, and it was observed that concentrations higher than about 1 [micro]g chl a [L.sup.-1] in the near-bed layer stimulated the mussels to keep their valves wide open, whereas tidal slack with slow flow resulting in near-bed concentrations below 1 [micro]g chl a [L.sup.-1] caused by grazing by the mussels, after some time led to shell closure, or reduced valve gapes of the mussels. The valve opening time from opening degree 50% to 100% and the valve closing time from opening degree 100% to 50% in response to variation in near-bed chl a concentration were on average 59 [+ or -] 22 and 50 [+ or -] 19 rain, respectively, which is considerably faster than observed in unfed mussels in the laboratory.

KEY WORDS: feeding activity, response to phytoplankton, valve gape, feeding activity, grazing impact

INTRODUCTION

Many physiological processes in bivalves are strongly influenced by the valve opening state and thus the valve opening-closing response times of the experimental animals (Riisgard 2001b). Filter-feeding bivalves are able to sense the presence of suspended food particles in the ambient water, and this sensitivity is reflected in the gape of the valves (Jorgensen 1975). The behavioral valve-opening repertoire ranges from closed valves and retraction of mantle edges and siphons to fully open valves accompanied by extended mantle edges and siphons (Jorgensen et al. 1988). Under optimal conditions, bivalves filter the ambient water at a maximum rate and under sub optimal environmental conditions, including low or (very) high concentrations of phytoplankton, the valve gape is reduced and the mantle edges retracted (Riisgard & Randlov 1981, Jorgensen et al. 1988, Riisgard, 1991; 2001a, Clausen & Riisgard 1996, Newell et al. 1998; 2001, Dolmer 2000a; 2000b, Riisgard et al. 2003, van Duren et al. 2005, Lassen et al. 2006). Although mussels in nature may often experience phytoplankton concentrations below the threshold level, the valve gape opening-closing response to natural variations in phytoplankton biomass in nature has so far been unobserved.

In algal clearance experiments with blue mussels, Mytilus edulis, we have often over the years observed that when mussels are fed algal cells for some hours every day during a period of up to one week, as has been the case during many student laboratory exercises, both the opening and closing response apparently become faster than observed on mussels starved for some days--or sometimes weeks--previous to the first clearance experiments. These observations draw attention to the importance of the previous feeding history of the bivalves. More knowledge on the response times in mussels will improve our understanding of the role of mussels for the energy flow in highly dynamic systems. In such systems, the opening-closing-response times to presence/absence of food may potentially limit ingestion and the grazing impact on phytoplankton. Awareness of response times is also important when interpreting (e.g., observed correlations between low food concentrations and clearance rates in mussels). On shorter time scales (minutes to hours) the mussel response times may cause hysteresis and mask the underlying functional response of clearance rate to food concentration.

In the present work we first in controlled laboratory experiments examined if mussels are able to lose and regain the ability to respond promptly to absence and presence of suspended algal cells in the ambient water.

Such knowledge about "memory/ learning/forgetting" in filter-feeding bivalves may be essential for making allowance for possible deviation from natural-response times in animals kept in the laboratory for shorter or longer times previous to physiological experiments.

Next, in field studies we examined the interactions between water flow, near-bed phytoplankton biomass and valve gape opening-closing response times of mussels in a dense bed. The study site, the narrow inlet to a shallow fjord, was chosen because of a predictable water-flow regimen created by the tide. Because of a high area-specific population filtration rate, a considerable variation in the chlorophyll a concentration could be expected within a half-tidal period. Therefore, the mussels are likely to be wide open during strong-flow periods with ample supply of phytoplankton, in contrast to periods with temporary tidal slacks and low supply rates of phytoplankton, where reductions in valve-gapes could be expected.

MATERIALS AND METHODS

Laboratory Studies

Blue mussels, Mytilus edulis, were collected in the southeastern part of Kerteminde Fjord (Fyn, Denmark) in August 2003, January and May 2004, and brought to the nearby Marine Biological Research Centre where all experiments were conducted. In feeding experiments, mussels in aerated aquaria were offered suspended algal cells Rhodomonas sp. (almost spherical, about 6.3 [micro]m in diameter) that are 100% efficiently retained by the gills of M. edulis (Mohlenberg & Riisgard 1978). The algal cell concentration was measured by means of an electronic particle counter (Elzone 5380) by taking 15 mL water samples for analysis. All experiments were conducted at naturally low algal concentrations between 1,000-4,000 Rhodomonas sp. cells [mL.sup.-1], equivalent to about 1-5 [micro]g chl a [l.sup.-1] (Riisgard et al. 2003). In short-term laboratory experiments, the valve gapes of mussels were monitored by means of a digital camera taking a picture every 10 min. In long-term laboratory experiments, the valve gapes were monitored by means of an underwater video camera (SONY TRV25) recording the mussels for 2 s every 5 min. The individual valve gape was measured as the distance between the shells at the center of the exhalant opening and subsequently expressed as percentage of maximum distance recorded.

To study whether the opening-closing responses of the mussels and their duration are dependent on preceding feeding conditions a number of feeding/starvation experiments were performed. In one series of experiments, Mytilus edulis (Group # 1, Table 1) were placed in an aerated aquarium with 1.2-[micro]m filtered seawater (20 psu) and the temperature was kept constant (Hetofrig cooler/ heater) at the natural (field) acclimation temperature of the mussels (19.6[degrees]C). Afterwards the mussels were fed about 3,500 Rhodomonas sp. cells [mL.sup.-1] for 2 h. The following day (Day 1) Rhodomonas sp. cells were repeatedly added to the aquarium and filtration rate measurements were performed under simultaneous monitoring of the valve gape of the mussels during the opening phase (0% to 80% of maximum registered valve gape during the whole experiment). After 5 h feeding, the mussels were separated into two groups (Group #2A and Group #2B, Table 1) and placed in two aerated aquaria with filtered seawater and constant temperature (19.6[degrees]C). For the next 4 days (Day 2 to Day 5) filtration rate measurements were performed while simultaneously monitoring the valve gapes of mussels in Group #2A, which were fed Rhodomonas sp. cells for about 5 h [d.sup.-1]. Group #2B was starved in the same period. On Day 6 the valve-gape opening response was recorded in both Group #2A and Group #2B. Rhodomonas sp. cells were repeatedly added to the two aquaria with the two groups, and the valve gapes were monitored until all mussels were maximally open. Then the algal supply was subsequently stopped and the valve-gape closing response to decreasing algal concentration (<1000 cells [mL.sup.-1]) was recorded. When the mussels had reduced the algal concentration below the detection level of the electronic particle counter (about 100 Rhodomonas sp. cells [mL.sup.-1]; Riisgard et al. 2003) further water sampling for particle analysis was ceased whereas the monitoring of valve gape continued until all mussels had closed their valves. During the following 13-14 days both groups of mussels starved whereupon, on Day 20 and Day 21 for Group #2A and Group #2B respectively, the valve-gape opening closing response was again registered.

In another series of experiments, the valve gape opening-closing response of mussels caused by presence/absence of algal cells was also recorded during periods with continuous algal supply followed by starvation periods without algal supply. The experimental period lasted for 68 and 50 days for Group #3 and Group #4 respectively (Table 1). Supply of algal cells was established from Day 1 to Day 38 for Group #3, and from Day 1 to Day 28 for Group #4, whereas long-term starvation was established from Day 38 to Day 68 for Group #3, and from Day 28 to Day 50 for Group #4.

Filtered seawater was continuously flowing through a strongly aerated aquarium with mussels. A dosing pump ensured constant addition of Rhodomonas sp. so that a well-defined steady-state level, corresponding to a natural phytoplankton concentration of 1.3-2.5 [micro]g chl a [L.sup.-1] was established (cf. Riisgard & Randlov 1981, Clausen & Riisgard 1996). The valve opening-closing response to presence/absence of algal ceils was studied by stopping the dosing pump and the through-flow of seawater. The subsequent decrease in algal concentration caused by the filtering mussels was measured until the concentration was reduced to about 500 cells [mL.sup.-1]. Then, by means of the dosing pump, this concentration level was maintained until all mussels had noticeably closed or reduced their valves. The duration of this closing response was defined as the time from when the concentration was 800 Rhodomonas sp. ceils [mL.sup.-1] (fully open mussels) to the time when the mean valve gape was reduced to 50% of the maximum registered during the preceding steady-state period. Then subsequently, the feeding conditions were resumed by switching on the algal dosing pump and the through-flow of filtered seawater and the immediate valve gape opening response recorded. The duration of the opening response was defined as the time from addition of algal cells after a short starvation period until the mean valve gape of the mussels reached 80% of the valve gape registered during the preceding steady-state. The long-term feeding period was succeeded by a starvation period to reveal if the mussels had changed the duration of the opening/ closing response. To check this, the algal dosing pump was switched on for some hours to re-establish steady-state before the valve gape opening/closing response was recorded as previously described.

The experimental periods with added algal ceils resulted in some growth of the mussels (increase in shell length of 0.2 and 0.7 mm in Group #3 and Group #4 respectively) and thus in an increased filtration rate, but a constant steady-state algal concentration was ensured by adjusting the dosing pump. Shell length and flesh dry weight (24 h, 110[degrees]C) were determined for all mussels after the experiments (Table 1).

Field Studies

The mussels, Mytilus edulis, studied in the present work formed a dense bed in the narrow inlet of Kerteminde Fjord, where a semidiurnal tide (amplitude [+ or -] 0.2 m) forces a strong, alternating in- and outgoing current (Fig. 1). In 2002 the mussel bed was overgrown by dense macro algae (Fucus sp.).

[FIGURE 1 OMITTED]

During the summer of 2002, simultaneous registrations of water elevation, current speed, chlorophyll a (chl a) concentration and valve gapes of mussels were carried out as 10 one-day studies at different selected locations on the shallow north-western part (Site 1: field-study days #1 to #6, and #10), and in the somewhat deeper middle part (Site 2: field-study days #7, #8, and #9, see Fig. 1) of the mussel bed. The water elevation was registered in the fjord-inlet about 400 m from the mussel bed by means of a CTD (Aquamat). In all cases, the current speed, chl a and valve gapes were registered on the same spot (with half a square meter) on the selected locations. The current speed was registered by means of a current propeller (Hydrobios) fixed at 20-cm height Site 1 and 50-cm height above the mussel bed at Site 2. The chl a (chl a) concentration was registered simultaneously at 2 to several heights above the mussel bed with sampling intervals from 10-60 min. The lowest sampling height always equaled 1-cm height above the mussel bed. Additional heights were up to 70 cm above the bed, depending on the study location. Sampling heights and intervals have been specified for each field-study day on the prepared figures of chl a contours. The water samples (50 mL) were taken manually by means of syringes attached to silicone pipes with endings through small holes in a rigid plastic pipe fixed in the bottom. Before sampling, the tubes were emptied of the "old" water (water from the prior sampling) to prevent mixing of new with old water. The water samples were immediately filtered on a GF/C filter (pore size = 1.2 [micro]m) followed by 20 h of chl a extraction in 10 mL 96% ethanol. The fluorescence of the supernatant was measured on a fluorometer (Turner, model 450) and the fluorescence values were converted into concentration of chl a by means of a regression function determined for each day of study. The measured chlorophyll a concentrations represent time-averages over 5-6 min (= the water-sampling time).

In each field study a group of undisturbed mussels (from 4-10 individuals) was photographed every 5 min by means of a time-lapse video camera (Sony TRV 25) installed in a housing (Ikelite) fixed on a heavy tripod. The pictures (resolution: 480 x 640 [pixels.sup.2]) were analyzed for the valve gapes (i.e., the distance between the valves) of the mussels. For each particular mussel the valve gapes were standardized to the observed maximal valve gape of the mussel during the particular time-series. Hence, a valve-opening degree of 100% represents the maximal valve gape of the mussel. An uncertainty test making repeated measurements of the same distance revealed a measuring uncertainty for the valve gape analysis between 6% and 28% (the larger the distance measured, the less the uncertainty). Consequently, the results from the deeper part of the mussel bed, where the mussels were registered with less camera zoom, are expected to be more uncertain. Minor uncertainties in valve gapes are caused by smaller variations in the orientations of the mussels to the camera during recording.

Prior to the field observations, mussel density (D, ind. [m.sup.-2]), mean shell length (L, mm), and area-specific population-filtration rate of the mussel bed ([F.sub.pop], [m.sup.3] [m.sup.-2] [d.sup.-1]) were determined and the patchy distribution of mussels was mapped using Global Positioning System (GPS). [F.sub.pop] was estimated as: [F.sub.pop] = [summation][F.sub.ind]/A, where [F.sub.ind] (1 [h.sup.-1]) = [0.0012L.sup.2.14] (Kiorboe & Mohlenberg 1981, corrected by Riisgard 2001a) and A = the area investigated for mussels. Knowing the water depth (H, m) and assuming complete mixing of the water column, the half-life time of phytoplankton above the mussel bed was calculated as: [t.sub.1/2] = H x ln2/[F.sub.pop].

The test statistics (significance level [alpha] = 0.05) have been calculated in Microsoft Excel 2003 and compared with critical values from Zar (1999).

RESULTS

Laboratory Studies

Figure 2 shows an example of valve opening response (during the first 15 min) to an addition of algal cells to a group of starved mussels, and further, after a period with a maintained algal concentration between 1000 and 5000 cells [mL.sup.-1] to ensure maximum valve gape, the figure shows the closing response to a decreasing algal cell concentration. Figure 3 shows similar valve closing responses caused by low algal concentrations in two experiments. In the first experiment (Fig. 3A), a group of mussels was initially fed for 6 days before recording of the valve-gape closing response and then subsequently starved during the following 20 days. Initially the mean valve gape decreased nearly linearly, but the closing response was faster in fed than in starved mussels. In the second experiment (Fig. 3B), the group of mussels was starving during the whole experimental period. Both after 6 and 21 days of starvation the mean valve gape initially decreased nearly linearly, but the closing response was slower in the mussels starved for 21 days.

[FIGURES 2-3 OMITTED]

Figure 4 shows the opening response of mussels to addition of algal cells at time zero. It is seen that the opening time depends on the preceding feeding conditions in such a way that starved mussels seem to have a longer opening time than previously fed mussels. The opening response time for mussels fed daily with algal cells is shown in Figure 5. During the algal exposure period of 6 days, the duration of the opening response is decreasing, whereas, after a subsequent starvation period of 20 days, the opening response time had increased. The observed valve gape responses to presence/absence of algal cells in the ambient water shows that the valve opening/closure responses are strongly influenced by the preceding feeding conditions.

[FIGURES 4-5 OMITTED]

Valve-closing response were also studied in steady-state feeding experiments and Figure 6 shows that the duration of the closing response initially declined with algal exposure time to become near constant in the remaining period, whereas on the other way round, the duration of the closing response was increasing during the subsequent starvation period. Figure 7 shows a similar closing response, although less pronounced. The closing response is slowly decreasing with the algal exposure time from Day 1 to Day 18 to be constantly low (about 10 min) during the rest of the feeding period. In the subsequent starvation period (from Day 29 to Day 50), the duration of the closing response increased apparently linearly as a function of time at a rate of about 10 min [d.sup.-1]. The duration of the opening response (defined as the time from readdition of algal cells after a short period of valve closure caused by stoppage of the dosing pump until the mean valve gape had increased to maximum) is shown on Figure 8. It is seen that during the period in which the mussels were exposed to a constant steady-state algal cell concentration (from Day 1 to Day 38), the duration of the opening response was about 90 min, but in the subsequent starvation period (from Day 39 to Day 68) the opening-response time decreased to about 20 min. The previous observations indicate that the period during which mussels may "learn" or "forget" to respond to the presence/absence of algal cells can last for weeks.

[FIGURES 6-8 OMITTED]

Mussels that were maximally open and actively filtering at a low steady-state concentration of about 700 Rhodomonas sp. cells [mL.sup.-1] responded to a slight reduction in the algal concentration by a nearly immediate reduction in valve gape (Fig. 9). From this observation it can be concluded that the critical algal concentration below which the mussels close their valves is only slightly lower than 700 Rhodomonas sp. cells [mL.sup.-1], or 0.9 [micro]g chl a [L.sup.-1].

[FIGURE 9 OMITTED]

Field Studies

The density, mean shell length and calculated area-specific population-filtration rate ([F.sub.pop]) of the mussel bed are shown in Figure 1. [F.sub.pop] ranged between 38 and 166 [m.sup.3] [m.sup.-2] [d.sup.-1], with a local maximum of 255 [m.sup.3] [m.sup.2] [d.sup.-1] (not shown). Assuming a maximal water depth of 1 m and complete water mixing, the calculated half-life time of phytoplankton above the mussel bed is between 6 and 26 min. Because complete water mixing may never be attained in the benthic boundary layer, the actual half-life of phytoplankton in the near-bed water is likely to be even shorter. Hence, a considerable depletion in near-bed chl a concentration was expected on the mussel bed around tidal slacks.

The time-series of water depth, current speed (averaged over 5-10 min), chl a concentration, and valve-opening degrees of the mussels from the 10 field-study days are depicted in Figure 10. The water depth and current speed both followed the semidiurnal tide. The tidal range was up to about 0.4 m and the current speed ranged between tidal maxima of up to 0.4 m [s.sup.-1] at 20 cm above the mussel bed and zero when the tide changed between in- and outgoing flow. In general, the measured variables showed tidal variation. The transition from strong to zero tidal flow was recorded on field-study days #1 to #3 (Fig. 10). It is seen that the mussels were wide open in the initial phase of strong tidal flow with near-bed chl a concentrations above 1 [micro]g [L.sup.-1], whereas around tidal slack, where the near-bed phytoplankton biomass was grazed below about 1 [micro]g chl a [L.sup.-l], the mussels reduced their valve gapes. A time lag in the response to low phytoplankton concentrations may be expected, which may explain why (e.g., the mussels in the first 15-30 min of field-study day #3) are fully open even though the chl a concentration is extremely low. Likewise, maintained maximal valve-opening degrees are observed during strong tidal flow on field-study day #4 for most of the mussels. During field-study days #5 and #6 it was observed, as the tidal flow reincreased after tidal slack, that the near-bed chl a concentration increased above 1 [micro]g [L.sup.-1], whereas at the same time the mussels reincreased their valve-opening degrees to maximum. The same pattern with temporary reductions in mussel valve gapes during tidal slack appears from the field-study days #7, #8, and partly field-study day #9, carried out on the deeper part of the mussel bed (Fig. 1, Study site 2). More scatter in the depicted valve-opening degrees is expected from these 3 field-study days because of a higher uncertainty in the valve-gape analysis. However, on field-study day #9 the high degree of scatter in valve gapes and resultant low mean valve-opening degree seem to be markedly pronounced (even at high tidal flow rates) coincident with a constant exposure to extremely low near-bed chl a concentrations (below 1 [micro]g [L.sup.-1]). Hence, short-term individual variability in valve valve-gape may be an indicator of shorter starvation periods. The extremely low chl a concentrations registered on field-study day #9 may be explained partly by incoming low chl a concentrations from the seaside and/or partly by limited replenishment of phytoplankton above the mussel bed because of a small tidal range this day. To this end, the results from field-study day #10 show an atypically valve-gape behavior. In particular, for one of the mussels a prolonged (150 min) valve-gape reduction phase is initiated at a high tidal flow rate and chl a concentrations well above 1.5 [micro]g [L.sup.-1]. It is not clear whether the chl a concentration of the inhalant water of the particular mussel has been lower than the registered one at 1 cm above the mussel bed.

[FIGURE 10 OMITTED]

Figure 11 shows the ratio of near-bed to surface chl a concentration ([C.sub.b]/[C.sub.s]) versus current speed (pooled raw data and data-fit). An analysis of variance (F-test, Zar 1999) showed a significant (P < 0.0005; [r.sup.2] = 0.25) data-fit to the following asymptotic regression model: [C.sub.b]/[C.sub.s] = 1.794 - [1.147e.sup.-0.00104]/[F.sub.pop], where U (m [s.sup.-1]) = the current speed and [F.sub.pop] (the area-specific population-filtration rate of the mussel bed) = 1.16 x [10.sup.-3] [m.sup.3] [m.sup.-2] [s.sup.-1]. Hence, the vertical chl a gradient above the mussel bed became steeper as the current speed decreased.

[FIGURE 11 OMITTED]

Figure 12 (left) shows the means of the time-averaged (1 h intervals were used to approach independence between pooled valve-gape data) valve-opening degrees in various tidal flow phases. On the basis of a nonparametric analysis of variance by ranks (Kruskal-Wallis test; Zar 1999) it is found that the mean valve-opening degrees were different between flow phases (P < 0.0005). A subsequent nonparametric multiple comparison test using rank sums (Dunn test; Zar, 1999) shows that the mean valve-opening degrees observed in the flow phases t = -3, -2, -1, and 2 were not significantly different from each other, but the mean valve-opening degrees in the flow phases t = -3, -1, and 2 were significantly (0.005 < P < 0.01) different from the reduced ones observed around tidal slack at t = 0 and t = 1. The mean valve-opening degrees have also been plotted versus the corresponding mean time-averaged near-bed chl a concentrations in Figure 12 (right). It is seen that the mean near-bed chl a concentrations were generally low (<1.6 [micro]g [L.sup.-1]) and that reduced valve-opening degrees associated with chl a concentrations below 1.0 [micro]g [L.sup.-1]. Table 2 shows a calculated "grazing index," chl [a.sub.b]/chl [a.sub.a], where the underscores "b" and "a" refer to the time just before and right after a valve-gape opening-closing response, respectively. The index is significantly lower (0.005 < P < 0.01) for the valve-closing phase compared with the valve-opening phase, which shows that the closing and opening phase occurred simultaneously with a decrease (grazing index <1) and increase (grazing index >1) in near-bed chl a concentration, respectively. The proposed causal relation between near-bed phytoplankton biomass (chl a) and mussel valve gape was investigated further in a shallow tidal lagoon close to Kerteminde Fjord. Here, 4 mussels settled on the bottom and 2 mussels settled on a stone 16 cm above the bottom, respectively, were video-registered simultaneously. The current speed was below the detection level (= 0.05 cm [s.sup.-1]) of the current propeller, but changes in the tidal flow was estimated visually by following small particles in the flow. The results have been shown in Figure 13. It is seen that the benthic concentration-boundary layer increased in thickness during the recording interval, along with a decrease in tidal flow rate (not shown). As the near-bottom chl a concentration became below 0.8 [micro]g [L.sup.-1], the 4 mussels on the bottom reduced their valve gapes, whereas the 2 mussels on the stone higher up in the concentration boundary layer remained fully open.

[FIGURES 12-13 OMITTED]

Finally, the distributions of the valve-opening and closing times (taken from reduction of the valve-opening degree from 100% to 50%, and opening from 50% to 100%, respectively) have been shown in Figure 14. Table 2 shows that the mean duration of the mussel closure phase were almost of same duration: 50 and 59 min, respectively.

[FIGURE 14 OMITTED]

DISCUSSION

Laboratory Studies

In a recent work Riisgard et al. (2003) studied the valve opening-closing phenomenon and response times in three species of filter-feeding bivalves (Cardium edule, Mytilus edulis, Mya arenaria) in the presence and absence of algal cells in controlled laboratory experiments. Opening state and correlated filtration rate at varying or maintained levels of algal concentration was studied in filtration rate experiments combined with video observation. When initially unfed bivalves were offered algal cells, the animals soon opened their siphons/valves simultaneously with a pronounced increase of the filtration rate. On the other hand, when open and actively filtering bivalves experienced decreasing algal concentrations below a certain level, this lead within a few hours to a reduced opening state and cessation of filtration activity. However, a considerably shorter valve-closure response time of only 15 min was recorded by Riisgard & Randlov (1981). The short response time may be caused by the experimental conditions because the valve-gape observations were made as part of a long term (47 d) feeding experiment imitating realistic, natural food concentrations. This suggestion is in agreement with the present observations.

This study shows that the duration of the well-known valve-gape response of mussels to the presence or absence of algal cells in the ambient water (Riisgard et al. 2003) is strongly influenced both by the preceding feeding conditions and the length of starvation. The period during which mussels may "learn" or "forget" to respond to the presence or absence of algal cells has been found to last for weeks in mussels kept in the laboratory. This knowledge is relevant for a correct understanding of the frequently used term "acclimation" (e.g., acclimation of experimental mussels to different laboratory conditions) such as changes in temperature, salinity or feeding conditions in experimental set-ups or storage thanks with or without through-flowing seawater, see also Kittner & Riisgard (2005). Many parameters (e.g., oxygen uptake rate, filtration rate, elimination- and uptake rates of pollutants) measured in different studies using mussels as experimental animals are strongly dependent on the valve-opening degree, and thus--although indirectly--the preceding feeding conditions or length of starvation after being collected in the field. Newly collected mussels may react very differently to experimental conditions compared with long-term laboratory "acclimated" mussels.

Valve closure and cessation of filter feeding in mussels at algal concentrations below a certain trigger level has been designated "lower trigger-level cessation" by Riisgard (2001b). In the present work, this critical algal concentration has been determined to be about 0.9 [micro]g chl a [L.sup.-1] in quite good agreement with a suggested lower trigger-level of about 0.5 [micro]g chl a [L.sup.-1] (Clausen & Riisgard 1996, Dolmer 2000a, 2000b). The valve-closing phenomenon, and subsequent reduction of the metabolism (Jorgensen et al. 1986), seems to represent an adaptation to filter feeding in extremely meager situations as discussed by Riisgard (2001b). From the present laboratory findings it seems likely that the valve opening/ closing-response times in the field may be considerably faster than usually observed in the laboratory, a suggestion that has been confirmed by the present field studies.

Field Studies

Filter-feeding mussels are important sinks for phytoplankton in shallow fjords and coastal waters. In such places dense mussel beds with area-specific population filtration rates of 100-200 [m.sup.3] [m.sup.-2] [d.sup.-1] may exert a considerable grazing impact (Jorgensen 1984, 1990), as also confirmed by the present work (Fig. 1). The phytoplankton biomass in areas with mussel beds may be significantly reduced, and a pronounced near-bottom concentration-boundary layer may develop (Frechette et al. 1989, Prins et al. 1998, Noren et al. 1999, Dolmer, 2000a, Tweddle et al. 2005). The steepness of horizontal and vertical gradients in the phytoplankton biomass in such a boundary layer is determined by current speed and turbulent mixing of the ambient water (van Duren 2005, Lassen et al. 2006). Consequently, mussels in dense beds may experience a dynamic nutritional environment as also demonstrated in the present work. The results shown in Figure 11 agree with earlier findings that mussels within a mussel bed become increasingly limited in their supply of phytoplankton at decreasing current speeds caused by steeper horizontal and vertical phytoplankton gradients above the mussel bed (Frrchette et al. 1989, Dare 1993, Butman et al. 1994, van Duren et al. 2005, Lassen et al. 2006, Tweddle et al. 2005). The tidal flow rate clearly had an effect on the valve-gape behavior of the mussels (Fig. 11), but the present study suggests this effect to be an indirect one, because the mussels (located between dense macro-algae, Fucus sp.) were never exposed to such high current speeds reported to affect the valve gape of mussels (see e.g., Newell et al. 2001). As pointed out by Yu & Culver (1999), local zones of depleted water may be found inside dense mussel beds and this phenomenon is likely to have caused the observed "atypical" behavior of the mussels on field-study day #10 (Fig. 10).

Prevailing phytoplankton concentrations in noneutrophicated waters usually lie between 1 and 5 [micro]g chl a [L.sup.-1], and in this concentration interval the blue mussel is wide open and filtering at maximum rate (Riisgard 2001a, Riisgard et al. 2003). However the phytoplankton biomass in the near-bottom concentration boundary layer is often substantially lower. Thus, Dolmer (2000a) measured the vertical distribution of phytoplankton above Mytilus edulis beds in Limfjorden, Denmark. In one case, the chl a concentration measured 2 m above the mussel bed (3.2 [micro]g [L.sup.-1]) declined towards the bottom to about 0.3 [micro]g [L.sup.-1], which is less than the threshold value (0.7-1.0 [micro]g [L.sup.-1]) below which the blue mussel closes its valves (Clausen & Riisgard 1996, Dolmer 2000a; b, Newell et al. 2001, Kittner & Riisgard 2005). As stated by Riisgard (2001b) the valve-closing phenomenon, which has also been clearly demonstrated in the present work, represents a physiological adaptation to filter feeding in extremely meager situations, when a reduced valve gape reduces the metabolism of the mussel (Jorgensen et al. 1986).

Tidally induced variation in bivalve filtration activity (valve gape and/or siphon opening degree) has earlier been observed in situ in Mya arenaria (Roegner 1998, Thorin et al. 1998), Mytilus edulis (Newell et al. 1998, 2005), and Venerupsis corrugatus (Stenton-Dozey & Brown 1992). Newell et al. (1998) registered coincidence between reduced valve gapes in M. edulis and low particle concentrations in during the tidal period, but the observed responses in valve gape/siphon opening were not directly correlated to tidally driven variation in phytoplankton biomass in the near-bottom water. Recently, Newell et al. (2005) registered two peaks in exhalent opening diameter of M. edulis in a dense bed: a peak when new water with food particles was brought in by the tide, and a peak during ebb, coincident with a high settling flux of marine snow. In the remaining tidal period the mussels bad reduced valve gapes. The present study shows that the valve gape of mussels is maximum during tidal phases with sufficiently high flow and near-bed chl a concentrations above roughly 1 [micro]g [L.sup.-1], but at slower flow rates and chl a concentrations <1 [micro]g [L.sup.-1] the mussels reduce their valve gape. The present observations indicate that the mussel in situ valve gape opening-closing response to variation in phytoplankton biomass is short (50 min) compared with the response times in unfed mussels in the laboratory (up to 180 min) (Riisgard et al. 2003), but compare well with fed mussels in the present laboratory studies and with results from a field-study conducted by Newell et al. (1998). It can be read from Figure 6 in Newell et al. (1998) that the closure response of M. edulis in Mud Cove (Maine, USA) was about 60 min, whereas the subsequent opening phase was slightly shorter (about 40 min).

CONCLUSION

Laboratory studies of valve-gape response of mussels to presence or absence of algal cell have demonstrated that valve opening and closing response times are strongly influenced by the preceding feeding conditions. The period during which mussels may "learn" or "forget" to respond to the presence or absence of algal cells can last for weeks in the laboratory. Field studies have revealed that phytoplankton concentrations higher than about 1 [micro]g chl a [L.sup.-1] stimulate the mussels to keep their valves wide open, whereas lower near-bed concentrations cause valve closure, or reduced valve gape. The in situ valve opening and closing times in response to variation in near-bed chl a concentration are considerably faster than in unfed mussels in the laboratory. This knowledge is essential for making allowance for deviation from natural response times in mussels kept in the laboratory previous to experimental physiological studies. Further, the opening/closing response times to presence/absence of phytoplankton may influence the actual grazing impact of mussels, and therefore, awareness of response times may be of importance for future modeling because of hysteresis that can mask the connection between filter-feeding activity and the concentration of phytoplankton.

ACKNOWLEDGMENTS

The authors thank Camile Saurel and four anonymous reviewers for constructive comments on an earlier draft of the manuscript. This study has been supported by a grant from the Danish Natural Science Research Council (H.U. Riisgard; Grant No. 21-03-0481).

LITERATURE CITED

Butman, C. A., M. Frechette, W. R. Geyer & V. R. Starczak. 1994. Flume experiments on food supply to the blue mussel Mytilus edulis L. as a function of boundary--layer flow. Limnol. Oceanogr. 39:1755-1768.

Clausen, I. & H. U. Riisgard. 1996. Growth, filtration and respiration in the mussel Mytilus edulis: no evidence for physiological regulation of the filter-pump to nutritional needs. Mar. Ecol. Prog. Ser. 141:37-45.

Dade, W. B. 1993. Near-bed turbulence and hydrodynamic control of diffusional mass transfer at the sea floor. Limnol. Oceanogr. 38:52-69.

Dolmer, P. 2000a. Algal concentration profiles above mussel beds. J. Sea Res. 43:113-119.

Dolmer, P. 2000b. Feeding activity of mussels Mytilus edulis related to near-bed currents and phytoplankton biomass. J. Sea Res. 44:221-231.

Frechette, M. 1989. The importance of boundary-layer flows in supplying phytoplankton to the benthic suspension feeder, Mytilus edulis L. Limnol. Oceanogr. 34:19-36.

Jorgensen, C. B. 1975. On gill function in the mussel Mytilus edulis L. Ophelia. 13:187-232.

Jorgensen, C. B. 1984. Effect of grazing: metazoan suspension feeders. In: J. E. Hobbie & P. J. Williams, editor. Heterotrophic activity in the sea. New York: Plenum Press, pp. 445-464.

Jorgensen, C. B., F. Mohlenberg & O. Sten-Knudsen. 1986. Nature of relation between ventilation and oxygen consumption in filter feeders. Mar. Ecol. Prog. Ser. 29:73-88.

Jorgensen, C. B., P. S. Larsen, F. Mohlenberg & H. U. Riisgard. 1988. The mussel pump: properties and modelling. Mar. Ecol. Prog. Ser. 45:205-216.

Jorgensen, C. B. 1990. Bivalve filter feeding: hydrodynamics, bioenergetics, physiology and ecology. Fredensborg, Denmark: Olsen & Olsen.

Jorgensen, C. B., P. Famme, H. S. Kristensen, P. S. Larsen, F. Mohlenberg & H. U. Riisgard. 1986. The bivalve pump. Mar. Ecol. Prog. Ser. 34:69-77.

Kittner, C. & H. U. Riisgard. 2005. Effect of temperature on filtration rate in the mussel Mytilus edulis--no evidence for temperature compensation. Mar. Ecol. Prog. Ser. 305:147-152.

Kiorboe, T. & F. Mohlenberg. 1981. Particle selection in suspension-feeding bivalves. Mar. Ecol. Prog. Ser. 5:291-296.

Lassen, J., M. Kortegard, H. U. Riisgard, M. Friedrichs, G. Graf & P. S. Larsen. 2006. Down-mixing of phytoplankton above filter-feeding mussels--interplay between water flow and biomixing. Mar. Ecol. Prog. Ser. 314:77-88.

Mohlenberg, F. & H. U. Riisgard. 1978. Efficiency of particle retention in 13 species of suspension-feeding bivalves. Ophelia 17:239-246.

Newell, C., D. E. Campbell & S. C. Gallagher. 1998. Development of mussel aquaculture lease site model MUSMOD: a field program to calibrate model formulations. J. Exp. Mar. Biol. Ecol. 219:143-169.

Newell, C., D. J. Wildish & B. A. MacDonald. 2001. The effects of velocity and seston concentration on the exhalant siphon area, valve gape and filtration rate of the mussel, Mytilus edulis. J. Exp. Mar. Biol. Ecol. 262:91-111.

Newell, C. R., C. H. Pilskaln, S. M. Robinson & B. A. MacDonald. 2005. The contribution of marine snow to the particle food supply of the benthic suspension feeder, Mytilus edulis. J. Exp. Mar. Biol. Ecol. 321:109-124.

Noren, F., J. Haamer & O. Lindahl. 1999. Changes in the plankton community passing a Mytilus edulis mussel bed. Mar. Ecol. Prog. Ser. 191:187-194.

Prins, T. C., A. C. Smaal & R. F. Dame. 1998. A review of the feedbacks between bivalve grazing and ecosystem processes. Aquat. Ecol. 31: 349-359.

Riisgard, H. U. 1991. Filtration rate and growth in the blue mussel, Mytilus edulis Linnaeus, 1758: dependence on algal concentration. J. Shellfish Res. 10:29-35.

Riisgard, H. U. 2001a. On measurement of filtration rates in bivalves--the stony road to reliable data: review and interpretation. Mar. Ecol. Prog. Ser. 211:275-291.

Riisgard, H. U. 2001b. Physiological regulation versus autonomous filtration in filter-feeding bivalves: starting points for progress. Ophelia 54:193-209.

Riisgard, H. U., C. Kittner & D. F. Seerup. 2003. Regulation of opening state and filtration rate in filter-feeding bivalves (Cardium edule, Mytilus edulis, Mya arenaria) in response to low algal concentration. J. Exp. Mar. Biol. Ecol. 284:105-127.

Riisgard, H. U. & A. Randlov. 1981. Energy budgets, growth and filtration in Mytilus edulis at different algal concentrations. Mar. Biol. 67:227-234.

Roegner, G. C. 1998. Hydrodynamic control of the supply of suspended chlorophyll a to infaunal estuarine bivalves. Est. Coast Shelf Sci. 47: 369-384.

Stenton-Dozey, J. M. E. & A. C. Brown. 1992. Clearance and retention efficiency of natural suspended particles by the rock-pool bivalve Venerupis corrugatus in relation to tidal availability. Mar. Ecol. Prog. Ser. 82:175-186.

Thorin, S., H. Bourdages & B. Vincet. 1998. Study of siphon activity in Mya arenaria (L.) in the intertidal zone by means of an underwater video camera. J. Exp. Mar. Biol. Ecol. 224:205-224.

Tweddle, J. F., J. H. Simpson & C. D. Janzen. 2005. Physical controls of food supply to benthic filter feeders in the Menai Strait, UK. Mar. Ecol. Prog. Ser. 289:79-88.

van Duren, L. A., P. M. J. Herman, A. J. J. Sandee & C. H. R. Heip. 2005. Effects of mussel filtering activity on boundary layer structure. J. Sea Res. 55:3-14.

Yu, N. & D. A. Culver. 1999. Estimating the effective clearance rate and refiltration by zebra mussels, Dreissena polymorpha, in a stratified reservoir. Freshw. Biol. 41:481-492.

Zar, J. H. 1999. Biostatistical analysis. Fourth edition. Prentice Hall International.

HANS ULRIK RIISGARD, * JOHAN LASSEN AND CHRISTINA KITTNER

Marine Biological Research Centre, University of Southern Denmark, Hindsholmvej 11, DK-5300 Kerteminde, Denmark

* Corresponding author. E-mail: [e-mail]

TABLE 1.
Mytilus edulis. Data for groups of mussels at the end of experiments.
Number of individuals (n), shell length (L), body wet- and dry
weight ([W.sub.w], [W.sub.d]).

Group #   n     L (cm [+ or -] SD)    [W.sub.w] (g [+ or -] SD)

  1       13   3.997 [+ or -] 0.106
  2A       7   3.990 [+ or -] 0.080     1.958 [+ or -] 0.213
  2B       6   3.960 [+ or -] 0.139     2.036 [+ or -] 0.475
  3       11   4.191 [+ or -] 0.112     2.671 [+ or -] 0.500
  4       19   2.702 [+ or -] 0.195     0.414 [+ or -] 0.071

Group #        [W.sub.d] (g [+ o r-] SD)

  1
  2A             0.334 [+ or -] 0.029
  2B             0.342 [+ or -] 0.104
  3              0.462 [+ or -] 0.110
  4              0.070 [+ or -] 0.015

TABLE 2.
Mytilus edulis. Mean duration [+ or -] SD of valve gape closing phase
(taken as the duration from valve-opening degree 100% to 50%) and
opening phase (taken as duration from valve opening degree 50% to
100%). Mean grazing index: [(chl [a.sub.b]/chl [a.sub.a]).sub.mean],
where chl [a.sub.b] = the near-bed chlorophyll a concentration right
before initiation of the valve-gape response and chl [a.sub.a] = the
near-bed concentration right after initiation of the valve-gape
response. N = sample size.

                           Closure Phase            Opening Phase

Mean duration         59 [+ or -] 22           50 [+ or -] 19
  [+ or -] SD (min)     ([N.sub.mussel] = 26)    ([N.sub.mussel] = 13)

Mean (chl [a.sub.b]/
  chl [a.sub.a])      0.8 [+ or -] 0.2         1.2 [+ or -] 0.1
  [+ or -] SD           ([N.sub.chp a] = 6)      ([N.sub.chl a] = 4)