In situ filtration rates of blue mussels (Mytilus edulis) measured by an open-top chamber method

Blue mussels, Mytilus edulis, form dense beds of both commercial and ecological importance, and many attempts have been made to determine their filtration rate. The total time in which mussels actually utilise their filtration capacity in nature varies greatly, making in situ methods for filtration rate measurements relevant. Further, it is being debated to what extend filtration rates measured in the laboratory using cultivated algal cells may apply for mussels in nature. In the present study, we have used an open-top chamber setup in order to allow repeated in situ filtration rate measurements of M. edulis using ambient natural phytoplankton and free-living bacteria. We found that the in situ measured filtration rates are comparable to filtration rates obtained in laboratory studies using different methods and controlled diets of cultivated algal cells. Further, we found that the retention efficiency of free-living bacteria was between 22.2% and 29.9%, in good agreement with values from laboratory studies. Our findings support the assumption that mussels in nature tend to use their filtration capacity when the phytoplankton concentration is above a certain lower trigger level.


Introduction
The filter-feeding blue mussel, Mytilus edulis (Linnaeus 1758) (Mollusca, Mytilidae), is widely distributed in the North and Mid-Atlantic regions [1] where it frequently forms dense mussel beds of both commercial and ecological impor-  [3]. Many attempts have over the years been made to determine the filtration rate of mussels using various methods [2] [3] [4], but it is currently being debated to what extend filtration rates measured in the laboratory using cultivated algae may apply for mussels in nature and whether the filtration rate is physiologically controlled [3] [5].
Laboratory observations of valve gap responses of mussels to absence or presence of added cultivated algal cells to the ambient water have revealed that they close their valves below a lower chlorophyll a (Chl a) trigger concentration of about 0.9 [6] to 0.5 µg Chl a l −1 [7]. Likewise, it has been shown that high algal concentrations > 8 µg Chl a l −1 also induce valve closure, reduce filtration rate [8] and subsequently reduce growth [9]. Under optimal conditions, with algal concentrations between the lower and upper trigger concentrations, mussels tend to filter the ambient water at a maximum rate. Because mussels are often living in dense beds, the ambient Chl a may frequently be strongly reduced [10] [11] [12] [13] [14] and likewise, during winter periods with no primary production [15] [16] [17]. The total time in which mussels actually utilise their filtration capacity in nature may therefore vary greatly, making in situ methods for filtration rate measurements relevant.
In the present study, we have slightly modified the design of the open-top chamber setup presented by Hansen et al. [18] in order to allow repeated filtration rate measurements using the clearance method and ambient natural phytoplankton and free-living bacteria. This approach has also become relevant because Cranford et al. [5] have claimed that experiments using added cultivated algal cells stimulate mussels to filter at anomalously high rates. Therefore, the aim of the present work was both to present a modified in situ method and to compare in situ measured filtration rates of mussels using natural phytoplankton with filtration rates measured in the laboratory using cultivated algal cells. Finally, we attempted to measure the in situ retention efficiency of free-living bacteria.

Open-Top Chamber Setup
The experimental open-top chamber setup used for in situ clearance measurements of mussels is depicted in Figure 1. A group of mussels attached to a PVC-plate is placed on the bottom and a transparent acrylic glass tube (d = 29 cm) is subsequently mounted, confining a certain known volume of natural seawater. The water volume in the chamber can be varied by vertical adjustment of the ladder holding the system, which can be submerged down to 90 cm (V max = 60 l; here we used 24.1 ± 5.0 l). Mussels were allowed to acclimate for some hours before the transparent plastic cylinder was mounted ( Figure 1) and sometimes kept submerged between the experimental sessions (here several days). Air stones placed at the periphery of the tube ensured efficient water mixing.
Water samples (1 l, 20 µm filtered and preserved in 5 ml Lugol's solution (6% iodine-potassium, 4% iodine solution) and 5 ml preserved in 1.5 ml 1% glutaraldehyde) were taken at discrete time intervals about 10 cm above the bivalves to follow the decrease in algal and bacterial concentrations (C, cells ml −1 ) as a function of time over a period of 60 min by means of cell enumeration (described below). When the concentration had decreased about 50%, the tube was lifted to allow new seawater to flow in to replace the particle diminished water.
Using this procedure, the mussels were minimally distributed.
Using the new open-top chamber method almost all disturbing side effects (e.g. changing current velocity, re-suspended algal cells from the sediment and epifaunal co-filtration) that may usually affect in situ clearance rate measurements, can be excluded.

Clearance Rate and Particle Retention Efficiency
The individual clearance rate (Cl ind , l·ind. −1 ·h −1 ) of a mussel was calculated as (Riisgård et al. [8]):  For quantification of the bacteria concentrations in water samples, DAPI staining and epifluorescence microscopy were applied according to Porter & Feig [19]. Thus, 5 ml samples (fixed with glutaraldehyde) were stored in a refrigerator until analysis. Samples were filtered through a 0.2 µm black polycarbonate sheet filter (Whatman Nuclepore Track Etch Membrane) and subse-  [20]. The clearance rate of mussels feeding on Ceratium spp. (mainly C. tripos and C. longipes and to a minor extend C. fusus, C. lineatum and C. furca) and free-living bacteria were measured simultaneously. After 60 min the tube was lifted and surrounding seawater replaced the mussel filtered water. This procedure was repeated up to 5 times (cf. Figure 2). After the last clearance rate was measured, the cultured flagellate Rhodomonas salina (diameter about 6 µm) was added to the experimental chamber to be grazed by the mussels (up to 3 repetitions). Mussels were observed for shell-opening degree and only open specimens were included in the calculations. data not shown), which demonstrated that the entire reduction in concentration could be attributed to the filter-feeding mussels. The height of the water in the chamber (which changed with the tide) was measured before each experiment in order to calculate the actual water volume. After the experiments, the shell length of mussels (L, mm) was measured using a calliper rule.
The retention efficiency (Re, %) of bacteria was expressed as the ratio of mean clearance rate on bacteria to mean clearance rate of Ceratium spp. [21].    (Table 1). Sub-figures C and D show clearance of bacteria measured simultaneously on these experimental days ( Table 2).

Discussion
From Table 1 it appears that the in situ measured clearance rates (= filtration rates) of Mytilus edulis are comparable to the filtration rates estimated from the shell length by means of the "model reference equation" presented by Riisgård et al. [23] based on available data on M. edulis obtained by the same research group using different methods and controlled diets of cultivated algal cells. However, it has been claimed by Cranford et al. [5] that experiments using added cultivated algal cells stimulate the mussels to filter at anomalously high clearance rates and that "a major methodological pitfall stems from the application   of artificial dietary conditions that stimulate a predetermined (e.g., maximal) feeding response". The presented new data (Table 1) clearly disprove this assumption. In the present study, the mussels were exposed to natural phytoplankton occurring in sufficiently high quantities to stimulate the mussels to filter at rates comparable to rates measured on completely open mussels fed cultivated algal cells in the laboratory. The findings support the assumption that mussels in nature tend to use their filtration capacity when undisturbed and the phytoplankton concentration is above the lower trigger level [6] [7] [24]. However, that is frequently not the case, especially in dense mussels beds [ [27]. Kreeger & Newell [28] showed in 14 C-prey-labeled ingestion and assimilation experiments that M. edulis ingested a significantly lower proportion (19%) of bacteria (< 1 µm diameter) than the larger (3 to 5 µm diameter) heterotrophic flagellates (58%). The bacteria (about 0.5 µm diameter) retention efficiency reported by Lucas et al. [29] was about 28%. In the present study, the retention efficiency of free-living bacteria was found to be between 22.2 and 29.9% (Table 2), which is in good agreement with earlier reported values.
A number of attempts have been made to study the filtration activity of mussels and scallops transferred to laboratory or near natural conditions [30] [31] [32] [33], and mussel behaviour has been recorded in situ over time in relation to changing Chl a [6] [11] [13]. More recently, Hansen et al. [18] used an open-top chamber method designed for in situ measurements of community clearance rate on bivalve populations at low water depths (< 0.5 m). In ambient natural seawater, Hansen et al. [18] found that an assemblage of bivalves (mussels and oysters) only realised 10% of their theoretical filtration capacity, but when cultivated algal cells were added to the experimental chamber this soon af-  [34]). Using the same technique developed by Hansen et al. [18], Vismann et al. [14] found that the filtration rate of Mytilus edulis in an intertidal bivalve bed was only about 13% of the theoretical. These findings indicate that the grazing impact of bivalves in shallow water locations may frequently result in depletion of phytoplankton and subsequently closure of the valves and cessation of filtering activity. This interpretation is supported by a recent study by Comeau et al. [35] who monitored the valve-opening behaviour of raft-cultivated M. galloprovincialis and found that valves were open 97.5% of the time. In agreement with this, our data (Table 1) show that M. edulis apparently utilises its filtration potential under natural conditions, as long as the Chl a level is above the lower trigger concentration.