Evaluating the use of exhalent siphon area in estimating feeding activity of blue mussels, Mytilus edulis.

ABSTRACT We evaluated the technique of measuring the exhalant siphon area (ESA) as an indicator of feeding activity in the blue mussel, Mytilus edulis. To accomplish this, we established the relationships between ESA measured using video and image analysis and clearance rate (CR) measured simultaneously in mussels exposed to various concentrations of microalgae as a food source in the laboratory. Two size classes of mussels (30 and 60 mm shell length) were fed 6 and 7 concentrations of the unicellular alga Isochrysis galbana., respectively (0, 0.6, 1.2, 1.8, 2.4, 3.0, 6.0 mg [L.sup.-1]). ESA and the variation in feeding activity associated with individual mussels significantly influenced CR in 60mm mussels, whereas effects of algal concentration were not significant within the ran[TM] tested. Individual variation in feeding activity significantly influenced CR in 30 mm mussels. However, unlike the 60 mm mussels, ESA did not significantly influence CR, whereas algal concentration did have a significant effect in 30 mm mussels. We did observe significant relationships between ESA and CR in some groups of mussels suggesting it may be a useful indicator of feeding activity under certain circumstances. However, the high degree of variation observed in our laboratory-based experiments on the relationship between CR and ESA leads us to conclude that measurements of ESA may be better used as an estimate of general behavior trends in feeding rather than a quantitative measure of clearance rate.

KEY WORDS: exhalent siphon area, Mytilus edulis, clearance rate, feeding activity, individual variability

INTRODUCTION

The bivalve, Mytilus edulis, is an ecologically and economically valuable species distributed throughout the world's cool temperate oceans (Seed & Suchanek 1992). Research has often focused on the feeding activity of this bivalve to understand its requirements for culture, its effects on the surrounding environment and its function as an indicator of ecosystem health (Seed & Suchanek 1992). Mussel feeding rates are usually determined by calculating clearance rate (CR; 1 [h.sup.-1]) and/or filtration rate (FR; mg [h.sup.-1]). Clearance rate is defined as the volume of water cleared of particles per unit time, whereas FR is the product of CR and particle concentration often defined as the number or weight of particulate matter removed from the suspension over time (Bayne 1985). There are often different definitions for these terms throughout the literature so it is often necessary to establish the specific definition for each study.

Several different methods have been used to estimate mussel CR, and thus FR, including: Coughlan static method, the flow-through method, and the bio-deposition method. Each of these methods typically has their own advantages and disadvantages. Coughlan (1969) developed a chamber method, which measures the decrease in food particles over time in a static chamber, thus providing a quantitative measurement of feeding rate. The flow-through method measures the decrease in food particles in a flow-through chamber, containing the bivalve, over time, compared with an empty container (Bayne et al. 1977, Widdows 1985). Both of these indirect methods are practical in a controlled laboratory setting, although the flow-through method may be somewhat more representative of natural conditions (Hildreth & Crisp 1976). The third method (bio-deposition) estimates feeding rate through measurements of suspended particles and the bio-deposit product (Iglesias et al. 1998). Basically, this technique uses inorganic particles as a tracer in the gut of the bivalve; by comparing the POM/PIM ratio of the seston in the water to the POM/PIM ratio in the bivalve feces, or the amount of feces, the feeding rate can be determined (Iglesias et a1. 1998). Two inherent problems with this method are that: (1) it assumes that the POM/PIM ratio is the same for the seston and the ingested particles (there is no preferential selection or difference in retention times), and (2) biodeposit collection must be precise for accurate quantitative analysis (biodeposits from one individual cannot be mixed with another's and the whole sample must be collected, which can be difficult when water currents or sedimentation processes are occurring) (Iglesias et al. 1998). This method can be practical in the laboratory when the diet is properly controlled over time and when the bio-deposition products are isolated. However, as an indicator of feeding in the field, this method is less practical because mussel bio-deposition products cannot be accurately isolated without disturbing the mussel. As well, strict diet control is not possible, thus it is unknown if the bio-deposition product is a true reflection of the environmental conditions at that time or of past conditions. Therefore gut retention times, which may vary according to a multitude of conditions Dame (1996) must be considered when using this method.

These methods for estimating feeding rates have been recently reviewed by Navarro and Velasco (2003) and Petersen et al. (2004). Navarro and Velasco (2003) found that Coughlan's static method and the bio-deposition method gave similar values for FR of the bivalves Mulinia edulis and Mytilus chilensis when measured gravimetrically, but not when measured volumetrically. Petersen et al. (2004) found that CR of Mytilus edulis differed significantly as a function of method used (e.g., flow-through method, bio-deposition method or Coughlan's static method) and with the origin of the mussels (e.g., genetic predisposition for gill surface area and weight). They also reported that CR, as determined gravimetrically by the bio-deposition method, was significantly lower than those measured by the flow-though or static chamber methods. It has also been argued that conditions in the laboratory may not accurately reflect in situ filtration (Cranford 2001, Riisgfird 2001), as a suite of constantly changing environmental attributes may influence basic physiological parameters and thereby render comparisons difficult (Petersen et al. 2004). With the observation that many feeding rates may have been underestimated by the bio-deposition method (Petersen et al. 2004), the potential benefits of a reliable method that could be used in the field, without disturbing the specimens and that would more accurately reflect the mussel's true feeding activity are apparent and are worth pursuing. Yahel et al. (2005) have described an alternative nonintrusive, but technically demanding method for estimating feeding rates of some species of active suspension-feeders in the field, including bivalves, by sampling inhalant and exhalent water masses using fairly precise techniques under water using SCUBA.

Several recent studies have suggested that the opening state of the valves and siphons may be closely related to food availability, and could potentially be useful as an indicator of feeding activity. Dolmer (2000) found that mean valve gap size of M. edulis was correlated with the concentration of chlorophyll a. Newell et al. (2001), and MacDonald and Nodwell (2003), found a significant positive effect of particle concentration on mussel exhalant siphon area (ESA). Both studies found that ESA was a potentially good qualitative indicator of feeding activity in the mussel Mytilus edulis. However, quantitative relationships between ESA and CR have not been established.

Riisgard et al. (2003) found a correlation between the opening state of the valves with filtration rate (= CR by our definition), at algal concentrations above a threshold level of ~800 cells [mL.sup.-1] (~0.8 X [10.sup.6] [micro][m.sup.3] particle volume) of Rhodomonas sp., in three species of bivalves, including M. edulis. However, the relationships between the opening state of the valves and siphons, with clearance rate (1 [h.sup.1]), were not studied at concentrations higher than 10,000 cells [ml.sup.1] (~10 x [10.sup.6] [micro][m.sup.3] particle volume) of Rhodomonas sp. Recent studies by Riisgard et al. (2003) and (2006) have emphasized that the underlying physiological regulatory mechanism for the opening-closing behavior of mussels is very important for many physiological processes, but has largely been ignored in studies. They found valve-gape response time to be strongly influenced by preceding feeding conditions which should be taken into consideration when planning and evaluating laboratory studies. This is an area that requires further research to better understand the relationship between activity of the siphons/valves, feeding rates and algal concentration.

Jorgensen and Ockelmann (1991) suggested there is a coordination of activity of the lateral cilia of the gill (ctenidium) with the adductor and retractor muscles, as well as, the exhalant siphonal sphincter. This suggests that, when the filtration response of the mussel decreases, a nervous system response occurs whereby the exhalant siphonal musculature contracts and thus closes the exhalant siphon. Mantle extension is limited by gape but it may facilitate filtration rates by increasing the siphon area and elongating the gill axis (Jorgensen 1990). Considering this, ESA may prove to be a more useful indicator of filtration rate than valve gape, as the siphons must be open for a mussel to feed, regardless of degree of valve opening (Newell et al. 2001). Maire et al. (2007) found that pumping rate was much better correlated with ESA than % valve gape. However, Frank et al. (2007) and (2008) using new laser imaging techniques for measuring both valve gape in oysters and mussel siphon area, respectively, did not record convincing relationships with clearance rate to support these earlier findings.

The objectives of this study were to measure CR using a standard in-direct method and ESA using time lapse video techniques simultaneously for the same mussels exposed to various concentrations of microalgae in the laboratory. We will estimate predictive relationships between CR and ESA and determine whether this technique can be used to remotely monitor feeding activity in commercially and ecologically important shellfish species such as Mytilus edulis.

MATERIALS AND METHODS

Mussels of two size classes, 30 and 60 mm in length ([+ or -] 0.5), were collected from the intertidal zone of Irving Nature Park, in Saint John, N B (45[degrees]12' 50" N, 66[degrees]07' 31 "W), and transported to the University of New Brunswick, in Saint John, NB. Mussels were scraped clean of fouling organisms and acclimated to laboratory conditions (8[degrees]C to 12[degrees]C and 32[per thousand]), for a minimum of 2 d to a maximum of 7 d. At least 24 h before experimentation, mussels had Velcro attached to their outer shell using cyano-acrylate glue. Previous studies by MacDonald and Nodwell (2003) have shown this technique secures the mussel, but does not impede its ability to open the valves and feed. Different mussels were used for each experiment.

The unicellular alga Isochrysis galbana (Parke), strain T-ISO, was grown at a temperature of 25[degrees]C by the method of Guillard (1984). The average cell diameter of 1. galbana used in this study was ~5 [micro]m. The experimental set-up is pictured in Figure 1. Algae were delivered to the 20 1 supply container via the control of a peristaltic pump thus ensuring constant flow was maintained. Filtered seawater entered the main supply container directly through a hose until water volume reached the overflow valve and maintained constant pressure. The controlled diet of I. galbana and seawater was thoroughly mixed in the main supply container with a submersible stir bar.

[FIGURE 1 OMITTED]

The diet mixture was then gravity-fed at a constant rate, via individual plastic tubing leaving the bottom of the main supply container and regulated by a calibrated flow restrictor, to individual 11 plastic feeding chambers that housed each individual mussel. Five feeding chambers were set-up: a control chamber with a mussel shell (to correct for any possible effects the experimental set-up may have on the results, such as settlement or algae sticking to the shell), and 4 test chambers, which had digital video cameras suspended overhead. The chambers were designed so that in-flowing water entered the bottom of the chamber, passed over a baffle and post (with Velcro[TM] attached) in the center of the chamber then exited the chamber from the top of the opposite end through a plastic standpipe thereby preventing recirculation. Flow rates of 200 ml [min.sup.-1]were obtained using flow restrictors, having the appropriate opening diameter, fitted inside the tubing. The flow rate of 200 ml [min.sup.-1] corresponded to approximately 6.1 cm [s.sup.-1] determined using a Flow-mate model 2000 portable flowmeter, with an accuracy of [+ or -] 2% of the reading. This flow rate was used as the standard because preliminary experiments revealed that CR became independent of flow for mussels in these chambers at flows above this level, but it was not too high to inhibit feeding activity (Wildish & Miyares 1990).

[FIGURE 2 OMITTED]

The digital video cameras were suspended directly above the feeding chambers to record the area of the mussel's exhalant siphon. Eight mussels (4 mussels per trial with two replicates; n = 8) were attached to the vertical posts in the 4 feeding chambers using Velcro[TM] so that their exhalant siphons were oriented toward the camera lenses, with the inhalant siphon facing into the flow of water and the exhalant siphon facing downstream to prevent any recirculated water from entering the inhalant siphon (Fig. 2). Mussels were allowed lh to acclimate to the flow rate and the addition of the algal diet to the filtered seawater. Table 1 shows the approximate concentrations of the algae added to the seawater: 0 (background filtered water), 0.6, 1.2, 1.8, 2.4, 3.0 and 6.0 mg [1.sup.-1]. The weight (mg [1.sup.-1]) of I. galbana offered to the mussels was determined using the dry-weight and ashing technique described by Strickland and Parsons (1972). A 11 sample of the diet mixture (in triplicate) was filtered onto a preashed, pre-weighed 25 mm GF/C Whatman filter and rinsed with 10 ml 3% ammonium formate. The filter was then dried for 24 h at 80[degrees]C and weighed to determine total particulate matter (TPM).

Mussels were exposed to each diet mixture for a minimum of 3 h up to a maximum of 5 h. Water samples were collected every 45 rain from the plastic standpipe outflows of the control chamber and feeding chambers to determine particle concentrations and estimate clearance rates. Particle concentrations and size distributions were determined using a Coulter Multisizer II fitted with a 100 gm aperture. Clearance rate (CR; 1 [h.sup.-1]) for each individual mussel was calculated as follows:

CR = ([C.sub.1] - [C.sub.0]/[C.sub.1]) x flow rate (1 [h.sup.-1])

where, [C.sub.1] is the particle concentration of the water flowing through the control chamber and Co is the concentration of the water after it has passed over the mussel. To compare CR it was necessary to correct for differences in dry tissue weight. This was done by standardizing all rates to a 1.0 g, total dry weight mussel using the following allometric equation:

CR = [([W.sub.s]/[W.sub.o]).sup.b] x [CR.sub.o],

where CR (1 x [h.sup.-1]) is the clearance rate for a standard bivalve of dry tissue weight [W.sub.s] (1.0 g), [CR.sub.o] (1 x [h.sup.-1]) is the observed clearance rate for a bivalve of dry tissue weight [W.sub.o] (g), and b is the exponent (b = 0.68) relating feeding to body weight. After completion of the experiments, soft tissues for each bivalve were isolated, dried to a constant mass at 80[degrees]C, and final mass recorded

In addition to CR, filtration rate (FR; mg [h.sup.-1]) has been used in other studies to estimate feeding activity. For comparative purposes we will also calculate FR using our estimates of CR and the following equation as outlined in Bayne (1985):

FR = CR ([lh.sup.-1]) x particle concentration (mg [1.sup.-1] )

Exhalant siphon area (ESA) was recorded continuously for all the mussels during the CR experiments using time-lapse 2 s/30 s intervals. Every 45 min (corresponding to the determination of particle concentration and CR), 3 still images of mussel ESA (taken 1 min before, 1 min during and 1 min after the 45 min mark) were collected. The digital images were downloaded to the program Image J (2001) (NIH public domain Java image processing program-URL: http://rbs.info.nih.gov/ij). Exhalant siphon area ([mm.sup.2]) was calibrated using a 1 cm reference mark on the vertical posts in the chambers (Fig. 2). The mean area of the 3 images taken per 45 min sample was used to determine the ESA of each mussel/sample time. In total, 4-6 mean values of ESA were used to calculate a single value for each mussel in this experiment.

A model investigating the influence of ESA and algal concentration on CR was determined using a nested analysis of covariance (ANCOVA) design in SPSS (2002) where algal concentration was the fixed factor, mussel ID was the random factor (nested within algal concentration) and ESA was the covariate (Zar 1999). Data were tested for normality using the Kolmogorov-Smirnov normality test, and for homogeneity of variance using the Levene's test of equality of variance, in Minitab (2005) 14.1 statistical software for Windows.

RESULTS

CR typically increased as ESA increased in individual 60 mm mussels exposed to the six different algal concentrations (Fig. 3A). However, there was considerable variability among individuals exposed to the same concentration and between different groups of mussels exposed to slightly different concentrations. For example, some mussels exposed to the 1.2 mg [1.sup.-1] concentration had high ESA's and low CR's, suggesting large siphon opening, but little feeding whereas the opposite was observed for some mussels in the 6.0 nag [1.sup.-1] trial. A nested ANCOVA for 60 mm mussels revealed that not all slopes of CR and ESA were equal (F = 3.06, P < 0.01, Table 2). Further analysis established that the slope from the 0 mg [1.sup.-1] concentration group of mussels was significantly different from the slopes for other algal concentration (T = -3.11, P < 0.01, Table 3) and was thus removed from subsequent analysis (Table 4). Using only the algal concentrations ranging from 0.6-6.0 mg [1.sup.-1] a value of F = 1.98 and P = 0.084 was obtained indicating there were no significant differences in slopes of CR and ESA for mussels in these remaining five groups of mussels (Table 4). We can see from Table 4 that individual mussel variability (Mussel ID; (F = 5.82, P < 0.0001 ) and ESA (F = 99.12, P < 0.000l) had highly significant effects on CR whereas algal concentration did not (F = 0.92, P < 0.469) at least for the range of concentrations tested here.

Based on the above mentioned statistical test it is a reasonable approach to pool the mussels from the remaining five experimental concentrations and recalculate an overall relationship between CR and ESA. However, we express some concern with this approach and suggest caution when interpreting the results for the following reasons: We selected a traditional P value of 0.05 for determining significance of the various components in the predictive model and it could have a major influence on the applicability of the model. Hendrix et al. (1982) suggested that only P values greater than 0.25 be used to help control against a type II error when assessing the heterogeneity of slopes in ANCOVA. If we had adopted this more conservative approach we could have concluded that the slopes were significantly different and calculated the relationship between CR and ESA individually for each of the five remaining trial concentrations. Whereas it is certainly possible that CR changes with ESA at different rates for various concentrations between 0.6-6.0 mg [1.sup.-l] it is not very practical to plot them individually given the small sample size (n = 8) per concentration and the high individual variability observed.

[FIGURE 3 OMITTED]

CR increases with increasing ESA in individual 60mm mussels (0.6-6.0 mg [1.sup.-1]), however only 42% ([R.sup.2] = 0.42, n = 56) of the variation in CR can be explained by ESA alone, which further increases to 53% ([R.sup.2] = 0.53, n = 48) when the mussels from the 1.2 mg [1.sup.-1] concentration are omitted (not shown). The same pattern of CR and ESA seen for 60mm was not observed for the 30mm mussels at similar trial concentrations of 0.0-3.0 mg [1.sup.-1] (Fig. 3B). Again we observed some high ESA's and low CR's (1.2 mg [1.sup.-1]) and some low ESA's with corresponding high CR's, the latter type being especially associated with the zero concentration group. With these two exceptions the variability between individuals exposed to the same trial did not seem to be as high as that observed for the 60mm mussels. A nested ANCOVA for 30 mm mussels could not distinguish any difference in slopes of CR and ESA (F = 0.62, P < 0.687) or for ESA alone (F = 3.25, P < 0.073) for the different trial concentrations (Table 5). Like the 60 mm mussels individual variability was also was highly significant in the 30 mm (Mussel ID; (F = 4.29, P < 0.0001), but unlike their larger 60 mm counterparts algal concentration had a significant effect on the smaller mussels (F = 7.22, P < 0.001).

If we perform a similar regression analysis for the 30mm mussels that we did for the 60 mm mussels, it is not until we remove the 0.0 mg [1.sup.-1] and the 1.2 mg [1.sup.-1] groups that we see a positive relationship and ESA and it only then explains 13% of the variation in CR (not shown). However, when we plot the relationship between another indicator of feeding activity, filtration rate (FR; mg [h.sup.-1]), and ESA for 30mm mussels we typically see an increase in FR with ESA with the exception of a few outliers again from the 1.2 mg [1.sup.1] group (Fig. 4). If we perform a similar regression analysis on FR for the 30mm where we remove the 0.0 mg [1.sup.-1] and the 1.2 mg [1.sup.-1] groups we find a significant positive relationship with ESA but it only explains 33% of the variation in CR (not shown).

DISCUSSION

The feeding activity of mussels is thought to be largely dependent on food concentration and quality (e.g., organic content) of the food (Widdows 1978, Widdows et al. 1979), and to a lesser extent on ambient flow velocity (Wildish & Miyares 1990). Particle concentrations tested in this experiment were relatively low (ranging from 0-6.0 mg [1.sup.-1]), but comparable to concentrations measured from locations where Mytilus edulis is commonly found in nature including our local populations in the Bay of Fundy (Table 6). Seston values from Canadian waters are in the range of 0.5-15 mg [1.sup.-1] TPM (0.4-9.0 mg [1.sup.-1] POM) (Jones & Iwama 1991, Taylor et al. 1992, Cranford & Hill 1999, MacDonald 2000, Barrington et al. 2002), whereas values from some European waters have been observed to have a wider range of 5-90 mg [1.sup.-1] TPM (2.2-9.0 mg [1.sup.-1] POM) (Stirling & Okumus 1995, Hawkins et al. 1996, Hawkins et al. 1997, Smaal & Haas 1997, Smaal et al. 2001).

Clearance rates observed in our laboratory experiments ranged between 0.1 and 4.2 1 x [h.sup.-1][g.sup.-1] (dry weight of soft tissue) and are very comparable to other published values for Mytilus edulis (Fig. 3A). Examples of studies reporting standardized CR's for M. edulis include values between 1.3 and 2.6 1 x [h.sup.-1][g.sup.-1] (Okumus & Stirling 1994), and ranged between 1.1 and 5.0 1 x [h.sup.-1][g.sup.-1] i for several different mussel populations exposed to a wide range of particle concentrations during summer months and primarily studied in the UK (their Table 4 and citations therein). Riisgard et al. (2003) recorded rates between 1.2 and 4.31 x [h.sup.-1][individual.sup.-1] for mussels ranging in size from 1.9-4.9 cm in length. Using the flow through method, Petersen et al. (2004) reported a large variation in CR's for different populations of M. edulis ranging between 5 and 10 1 x [h.sup.-1][g.sup.-1], possibly related to mussel condition index and gill area. Newell et al. (200l) reported mean mussel CR's between 2.9 and 4.0 l x [h.sup.-1][g.sup.-1] comparable to other studies, but observed different response whether using flow through, biodeposition or video techniques and exhalent siphon area.

[FIGURE 4 OMITTED]

Our findings that mussels exposed to very low concentrations greatly reduced their feeding activity or intermittently open and close their siphons are not novel observations, but they could contribute significantly to the high variation we recorded. It has been shown in both the field (Riisgatrd & Randlov 1981, Riisgard 1991) and laboratory investigations (Newell & Shumway 1993, Newell et al. 1998) that under conditions of low food levels mussels may close their shells and reduce filtration rate. The purpose of filtration reduction and shell closure may be to reduce oxygen uptake and thus respiration to save energy at low algal concentrations (Jorgensen 1990). However, once it becomes energetically profitable to begin filtering again (e.g., when algal concentrations reach a threshold level) mussels open their shells and siphons, and begin to feed. This study, as did Newell et al. (2001) and Riisgard et al. (2003), finds that M. edulis may regulate its feeding activity by altering its feeding rate (e.g., CR or FR), or opening and closing siphons when food concentrations reach threshold levels.

Several studies have observed feeding threshold levels (both low and high) for M. edulis. Newell et al. (2001) found that ESA became more predictable (qualitatively) when particle volume reached a similar particle volume between 1.0-2.0 x [10.sup.6] [micro][m.sup.3] of Isochrysis-Dunaliella-mixture. This finding is in agreement with those of Riisgard and Randlov (198 l) and Riisgard et al. (2003). Riisgard and Randlov (1981) found that M. edulis reduces valve opening and filtration rate, and may even close the shell when algal concentrations are low ([less than or eaul to] 1000 cells [ml.sup.-1] Phaeodactylum tricornutum, 4 x 12 [micro]m cell diameter, 0.15 x [10.sup.6] [micro][m.sup.3] particle volume). Riisgard et al. (2003) also found that when bivalves, including M. edulis, experience very low algal concentrations (< 630 cells [ml.sup.1] Rhodomonas sp., ~6.2 [micro]m spherical diameter, < 0.6 x [10.sup.6] [micro][m.sup.3] particle volume) the bivalve will reduce siphon opening and valve gape, or completely close, until the algal concentration is elevated to a threshold level (> 800 cells [ml.sup.-1] Rhodomonas sp., > 0.8 x [10.sup.6] [micro][m.sup.3] particle volume). MacDonald and Nodwell (2003) observed a positive relationship between ESA and particle concentration until concentrations increased to 25,000-30,000 cells [ml.sup.-1] Pavlova lutheri or Tetraselmis suecica (~10 x 12 [micro]m cell diameter) (24-28 x [10.sup.6] [micro][m.sup.3] particle volume) after which ESA appeared to decrease with further increase of particle concentration, implying a potential upper threshold level for mussels may exist. (N.B. in this experiment < 0.6 mg [1.sup.-1] I. galbana particle concentration = < 0.5 x [10.sup.6] [micro][m.sup.3] particle volume).

The statistical analysis for the General Linear Model revealed that as much as 81% (adjusted [R.sup.2]) of the variation in CR's of 60mm mussels could be explained by the combination of exposure to experimental algal concentrations (0.6-6.0 mg [1.sup.-1]), ESA and individual variability of mussels tested. In contrast, only 76% the variation in CR's of 30mm mussels could be explained by the same model (0.0-3.0 mg [1.sup.-1]). When we pooled the data for the various algal concentrations and removed the 0.0 mg [1.sup.-1] group we observed that approximately 42% of the variation in CR can be explained by ESA alone in the 60 mm mussels. If a second group of 60 mm mussels are removed (1.2 mg [1.sup.-1]) from the analysis then ESA alone may explain as much as 53% of the variation in CR. ESA alone did not appear to be a significant factor (F = 3.25, P < 0.073) for the smaller 30 mm mussels (Table 5). Note that individual variability of the mussels was highly significant in the model for both 60 mm and 30 mm mussels.

There may be several reasons why ESA was less correlated with CR in the smaller 30 mm mussels than their larger 60 mm counterparts. First, unlike the larger mussels the smaller mussels had a smaller sample size and were only exposed to 6 treatments of microalgae which did not include the highest (6.0 mg [1.suibub.-1]) concentration where feeding activity may have been expected to be higher and possibly more consistent over time. Secondly, some technical limitations to our video techniques where it may have been more difficult to consistently and accurately digitize the smaller siphon areas of the 30 mm mussels from the videotape. Thirdly, the smaller 30 mm mussels may have remained feeding at a lower threshold concentration than their larger counterparts as indicated by the higher than expected CR values for the 0 mg [1.sub.-1] concentration observed in Figure 3B. Plus an artefact may be introduced with the way CR is calculated, perhaps more so for the smaller mussels feeding at the 0 mg [1.sup.-1] concentration than the larger mussels which had almost ceased feeding. CR is calculated on the basis of the percentage or particles removed so the removal of 200 cells [ml.sup.-1] for example at this extremely low concentration of 858 cell [ml.sup.-1] would produce a CR of 2.81 [h.sup.-1] whereas the removal of 200 cells from the next highest concentration of 6611 cells [ml.sup.-1] would estimate a CR of only 0.36 1 [h.sup.-1] (Table 1). So, if they remained open and feeding at all, it would produce a relatively high estimate of CR. We also recognize that some of the poor correlation between ESA and CR are not just associated with estimating ESA, but also with the difficulties in consistently and accurately measuring CR using the indirect particle removal technique we used.

There have been very few attempts to quantitatively link feeding rates with either valve gape or ESA. We observed higher variation in the relationship between CR and ESA than we had anticipated possibly making it a less reliable predictor of feeding rate than earlier expected. There is a possibility than the quality of the microalgae may have varied between the experiments despite our attempts to control for any variation. In particular, results for the 1.2 mg [1.sup.-1] trials with a few exceptions gave unusual results of apparently very high ESA values and very low CR values for both groups of mussels. Riisgard et al. (2006) have shown that preceding feeding conditions can influence valve gape response in Mytilus edulis. It possible that some of the variation we observed for CR and ESA in both groups of mussels may be related to potential differences in feeding conditions in the field before collection or while held in the laboratory.

Frank et al. (2007) described an innovative technique using fiber optic sensors to monitor valve gape in Crassostrea virginica and measure feeding rates simultaneously. They measured valve gape instead of ESA however we noted that CR and valve gape in oysters (their Figure 6) looked very similar to the relationship we observed for CR and ESA in mussels. They did occasionally find significant relationships between CR and valve gape in a few individuals, but they also observed high variability between individuals and only once did they observe an [R.sup.2] exceeding 0.5. They concluded after studies with oysters that valve gape alone is not a good measure of feeding activity in bivalve molluscs in general. It is interesting to note that one of their objectives was to design an instrument to measure valve gape in bivalves 3 cm or larger, however they actually ended up using oysters 6.9-8.4 cm in length. Perhaps they also had some technical issues on working with small bivalves as we did in this study. In a subsequent study Frank et al. (2008) did not find a consistent relationship between cross sectional area (= ESA) and CR, but they did report that their estimates of CR were made under less than ideal conditions.

Our findings and those of Frank et al. (2007) showing highly variable relationships between ESA/shell gape and feeding activity in bivalves are in contrast to the results of Maire et al. (2007) working on filtration activity and ESA in the Mediterranean mussel Mytilus galloprovincialis from France. Maire et al. (2007) used an automated system to measure ESA, valve gape and pumping activity repeatedly in single mussels to establish the relationships between these three variables. They observed much higher feeding rates than have typically been reported for this species and extremely high [R.sup.2] values (0.93-0.99). Some of the explanation for the very high feeding rates may be attributable to working at very low particle concentrations. However, the extremely high [R.sup.2] values are likely related to a large number of pseudo-replicates, and violation of the assumption of independence of data required for standard regression which omits variability associated with each individual mussel that we have seen is highly significant.

CONCLUSION

Bivalves respond to a combination of various environmental conditions by altering the gape of the shell or area of the siphon and are thought to be a reasonable indicator of the rate of water processing and/or feeding rates. Some studies, including this one, have examined the quantitative relationships between mussel shell gape and/or features of the siphon and an estimate of feeding activity. Whereas significant relationships have been found, we would recommend caution when attempting to apply these relationships, which are often accomplished under specific laboratory conditions or using small numbers of animals, to quantitatively predicting the rate of feeding for a population based solely on a single predictor (e.g., ESA or shell gape). Rather, we would conclude that measurements of ESA may be better used as an estimate of general behavior trends in feeding rather than a quantitative measure of clearance rate. To go further, we need to identify the major factors influencing the particular feeding behavior of interest, exposure to prior conditions, recognize the potential significance of high individual variation typical of bivalves and apply the appropriate analysis including possibly repeated measures designs.

ACKNOWLEDGMENTS

We thank Wayne Armstrong for technical support and Ian Butts for supplying cultures of microalgae. Statistical advice was provided by Chris Blanar and Maureen Tingley, and Dr.s R. Rochette and T. Chopin provided valuable input to the project and commented on earlier versions of the manuscript. This research was conducted under certification through the University of New Brunswick and the Canadian Council for Animal Care. Funding for the project was provided by grants from "AquaNet", a National Centre of Excellence funded by the Canadian government and the Natural Sciences Research Council of Canada (NSERC). We appreciate their support.

LITERATURE CITED

Barrington, K. A., B. A. MacDonald & S. M. C. Robinson. 2002. Feeding behaviour of blue mussels living within an Atlantic salmon aquaculture site. Bull. Aquacult. Assoc. Can. 102-3:109. (abstract).

Bayne, B. L. 1985. Physiological measurements. In: B. L. Bayne, D. A. Brown, K. Burns, D. R. Dixon, A. Ivanovici, D. R. Livingstone, D. M Lowe, M. N. Moore, A. R. D. Stebbing, & J. Widdows, editors. The effects of stress and pollution on marine animals. New York: Praeger Scientific. pp. 3-45.

Bayne, B. L., J. Widdows & R. I. E. Newell. 1977. Physiological measurements on estuarine bivalve molluscs in the field. In: B. F. Keegan, P. O. Ceidigh, & P. J. S. Boaden, editors. Biology of benthic organisms 11th European Symposium on Marine Biology, Galway. Ireland. pp. 57-68.

Coughlan, J. 1969. The estimation of filtering rate from the clearance of suspensions. Mar. Biol. 2:356-358.

Cranford, P. J. 2001. Evaluating the 'reliability' of filtration rate measurements in bivalves. Mar. Ecol. Prog. Ser. 215:303-305.

Cranford, P. J. & P. S. Hill. 1999. Seasonal variation in food utilization by the suspension- feeding bivalve molluscs Mytihts edulis and Placopecten magellanicus. Mar. Ecol. Prog. Ser. 190:223-239.

Dame, R. F. 1996. Ecology of marine bivalves: an ecosystem approach. Boca Raton, FL: CRC Press. 254 pp.

Frank, D. M., J. F. Hamilton, J. E. Ward & S. E. Shumway. 2007. A fiber optic sensor for high resolution measurement and continuous monitoring of valve gape in bivalve molluscs. J. Shellfish Res. 26:575-580.

Frank, D. M., J. E. Ward, S. E. Shumway, B. A. Holohan & C. Gray. 2008. Application of particle image velocimetry to the study of feeding in marine invertebrates. Mar. Freshwat. Behav. Physiol. 41:1-18.

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

Guillard, R. R. L. 1984. Culture of phytoplankton for feeding invertebrates. In: C. J. Berg, editor. Culture of marine invertebrates. Stroudsbury, Pennsylvania: Hutchinson Ross. pp. 108-132.

Hawkins, A. J. S., R. F. M. Smith, B. L. Bayne & M. Heral. 1996. Novel observations underlying the fast growth of suspension feeding shellfish in turbid environments: Mytilus edulis. Mar. Ecol. Prog. Ser. 131:179-190.

Hawkins, A. J. S., R. F. M. Smith, S. Bougrier, B. L. Bayne & M. Heral. 1997. Manipulation of dietary conditions for maximal growth in mussels, Mytilus edulis, from the Marennes-Oleron Bay, France. Aqua. Liv. Res 10:13-22.

Hendrix, L. J., M. W. Carter & D. T. Scott. 1982. Covariance analyses with heterogeneity of slopes in fixed models. Biometrics 38: 641-650.

Hildreth, D. I. & D. J. Crisp. 1976. A corrected formula for calculation of filtration rate of bivalve Molluscs in an experimental flowing system. J. Mar. Biol. Ass. U.K. 56:111-120.

Iglesias, J. I. P., M. B. Urrutia, E. Navarro & I. Ibarrola. 1998. Measuring feeding and absorption in suspension-feeding bivalves: an appraisal of the biodeposition method. J. Exp. Mar. Biol. Ecol. 219:71-86.

Image, J. 2001. Version 1.25, National Institutes of Health, USA, URL: http://rbs.info.nih.gov/ij.

Jones, T. O. & G. K. Iwama. 1991. Polyculture of the Pacific oyster, Crassostrea gigas (Thunberg), with Chinook salmon, Oncorhynchus tshawytscha. Aquaculture 92:313-322.

Jorgensen, C. B. 1990. Bivalve filter feeding: hydrodynamics, bio-energetics, physiology and ecology. Denmark: Olsen and Olsen. 140 pp.

Jorgensen, C. B. & K. Ockelmann. 1991. Beat frequency of lateral cilia in intact filter feeding bivalves: effect of temperature. Ophelia 33: 67-70.

MacDonald, B. A. 2000. Potential impacts of increased particle concentrations on scallop feeding and energetics. In: Alaska Department of Fish and Game and the University of Alaska Fairbanks. A workshop examining potential fishing effects on population dynamics and benthic community structure of scallops with emphasis on the weathervane scallop Patinopecten caurinus in Alaskan waters. Alaska Department of Fish and Game, Division of Commercial Fisheries, Special Publication 14, Juneau. pp. 20-26.

MacDonald, B. A. & L. M. Nodwell. 2003. A portable and practical method to monitor bivalve feeding activity in the field using timelapse video technology. J. Shellfish Res. 22:209 212.

Maire, O., J. M. Amouroux, J. C. Duchene & A. Gremare. 2007. Relationship between filtration activity and food availability in the Mediterranean mussel Mytilus galloprovincialis. Mar. Biol. 152:1293-1307.

Minitab, 2005. Minitab Release 14.1 Statistical software for Windows.

Minitab, Inc.

Navarro, J. M. & L. A. Velasco. 2003. Comparison of two methods for measuring filtration rate in filter feeding bivalves. J. Mar. Biol. Ass. U.K. 83:553 558.

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

Newell, C. R. & S. E. Shumway. 1993. Grazing of natural particulates by bivalve molluscs: a spatial and temporal perspective. In: R. F. Dame, editor. Bivalve filter feeders in estuarine and Coastal Ecosystem Processes. Berlin: Springer-Verlag. pp. 85-148.

Newell, C. R., D. J. Wildish & B. A. MacDonald. 2001. The effects of velocity and seston concentration on the exhalant siphon area and valve gape and filtration rate of the mussel Mytilus edulis. J. Exp.

Mar. Biol. Ecol. 262:91-111.

Okumus, I. & H. Stirling. 1994. Physiological energetics of cultivated mussel (Mytilus edulis L.) populations in two Scottish west coast sea lochs. Mar. Biol. 119:125-132.

Petersen, J. K., S. Bougrier, A. C. Smaal, P. Garen, S. Robert, J. E. N. Larsen & E. Brummelhuis. 2004. Intercalibration of mussel Mytilus edulis clearance rate measurements. Mar. Ecol. Prog. Ser. 267: 187-194.

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

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

Riisgard, H. U. & A. Randlov. 1981. Energy budgets, growth and filtration in Mytilus edulis at different algal concentrations. Mar. Biol. 61:227-234.

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., J. Lassen & C. Kittner. 2006. Valve gape response times in mussels (Mytilus edulis) Effects of laboratory preceding-feeding conditions and in situ tidally induced variation phytoplankton biomass. J. Shellfish Res. 25:901-911.

Seed, R. & T. H. Suchanek. 1992. Population and community ecology of Mytilus. In: E. Gosling, editor. The mussel Mytilus: ecology, physiology, genetics and culture. B.V., Amsterdam: Elsevier Press Science Publishers. pp. 87-169.

Smaal, A. C. & H. A. Haas. 1997. Seston dynamics and food availability on mussel and cockle beds. Estuar. Coast. Shelf Sci. 45:247-259.

Smaal, A. C., M. van Stralen & E. Schuiling. 2001. The interaction between shellfish culture and ecosystem processes. Can. J. Fish. Aquat. Sci. 58:991-1002.

SPSS. 2002. SPSS for Windows, version 11.5, LEAD Technologies, Inc.

Stirling, H. P. & I. Okumus. 1995. Growth and production of mussels (Mytilus edulis L.) suspended at salmon cages and shellfish farms in two Scottish sea lochs. Aquaculture 134:193-210.

Strickland, J.D.H. & T.R. Parsons. 1972. A practical handbook of seawater analysis, Bulletin 167, 2nd ed. Fisheries Research Board of Canada, Ottawa. 310 pp.

Taylor, B. E., G. Jamieson & T. H. Carefoot. 1992. Mussel culture in British Columbia: the influence of salmon farms on growth of Mytilus edulis. Aquaculture 108:51-66.

Widdows, J. 1978. Combined effects of body size, food concentration and season on the physiology of Mytilus edulis. J. Mar. Biol. Ass. UK. 58:109-124.

Widdows, J. 1985. Physiological Procedures. In: B. L. Bayne, D. A. Brown, K. Burns, D. R. Dixon, A. Ivanovici, D. R. Livingstone, D. M. Lowe, M. N. Moore, A. R. D. Stebbings & J. Widdows, editors. The effects of stress and pollution on marine animals. New York: Praeger Scientific. pp. 161-178.

Widdows, J., P. Fieth & C. M. Worrell. 1979. Relationships between seston, available food and feeding activity in the common mussel Mytilus edulis. Mar. Biol. 50:195-207.

Wildish, D. J. & M. P. Miyares. 1990. Filtration rate of blue mussels as a function of flow velocity: preliminary experiments. J. Exp. Mar. Biol. Ecol. 142:213-221.

Yahel, G., D. Marie & A. Genin. 2005. UnEx--a direct method to measure filtration rates, nutrition, and metabolism of active suspension feeders. Limnol. Oceanogr. Methods 3:46-58.

Zar, J. H. 1999. Biostatistical Analysis, 4th ed. Prentice-Hall, New Jersey. 663 pp.

BRUCE A. MACDONALD, (1) * SHAWN M. C. ROBINSON (2) AND KELLY A. BARRINGTON (1)

(1) Department of Biology, University of New Brunswick, Saint John, New Brunswick, Canada E2L 4L5;

(2) Biological Station, Department of Fisheries and Ocean, 531 Brandy Cove Road, St. Andrews, New Brunswick, Canada, E5B 2L9

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

TABLE 1.
Concentrations of Isochrysis galbana added to filtered seawater
comprising the seven different diets offered to mussels in this study.
Samples were measured in triplicate and algal weights were determined
using the Strickland and Parsons dry-weight and ashing technique.

    Theoretical                 Actual
   Concentration            Concentration
   (mg [L.sup.-1)           (mg [L.sup.-1)

         0                       0.10
        0.6                      0.58
        1.2                      1.15
        1.8                      1.73
        2.4                      2.30
        3.0                      2.88
       6.0 *                     5.76

    Theoretical                 Actual
   Concentration            Concentration
 (cells [L.sup.-1])       (cells [L.sup.-1])

         0                  858 [+ or -] 80
       6,000              6,611 [+ or -] 2010
       12,000            12,374 [+ or -] 3526
       18,000            18,724 [+ or -] 4290
       24,000            24,780 [+ or -] 6127
       30,000            32,231 [+ or -] 6098
       60,000            60,769 [+ or -] 9363

Theoretical Particle       Actual Particle
Volume X [10.sup.6]      Volume X [10.sup.6]
 ([micro][m.sup.3])       ([micro][m.sup.3])

         0               0.062 [+ or -] 0.006
        0.44              0.48 [+ or -] 0.15
        0.88              0.91 [+ or -] 0.26
        1.31              1.36 [+ or -] 0.31
        1.75              1.81 [+ or -] 0.45
        2.19              2.35 [+ or -] 0.44
        4.38              4.44 [+ or -] 0.68

TABLE 2.
Results from the nested ANCOVA for clearance rate of 60 mm
mussels using all algal concentrations (0-6.0 mg [L.sup.-1] Isochrysis
galbana), where algal concentration was the fixed factor,
mussel ID was the random factor nested within algal
concentration and ESA was the covariate.

Source                  df      Seq SS     Adj SS

Algal concentration       6    197.8788     3.6459
Mussel ID                49    110.0799    83.8348
  (Algal conc.)
ESA                       1     36.5734    24.5055
Algal cone. * ESA         6      7.6151     7.6151
Error                   200     82.9986    82.9986
Total                   262    435.1459

Source                  Adj MS        F        P

Algal concentration      0.6077     1.18    0.320
Mussel ID                1.7109     4.12    0.000
  (Algal conc.)
ESA                     24.5055    59.05    0.000
Algal cone. * ESA        1.2692     3.06    0.007
Error                    0.4150
Total

TABLE 3.
Results from the posthoc Tukey pairwise comparisons on the
slopes of all algal concentrations tested (0-6.0 mg [L.sup.-1]
Isochrysis galbana) for 60 mm mussels determined using nested
ANCOVA.

Term (Algal conc.
   in mg 1-1)        Coefficient    SE Coefficient      T       P

       0              -0.06634         0.02133        -3.11   0.002
       0.6             0.03363         0.01906         1.76   0.079
       1.2            -0.04593         0.02397        -1.92   0.057
       1.8             0.00529         0.01449         0.37   0.715
       2.4             0.03425         0.01798         1.91   0.058
       3.0             0.01995         0.01838         1.09   0.279

TABLE 4.
Results from the revised nested ANCOVA for clearance rate of 60 mm
mussels, where the zero algal concentration was dropped from analysis
based on posthoc tests. Algal concentration was the fixed factor,
mussel ID was the random factor nested within algal concentration and
ESA was the covariate.

Source                   df      Seq SS     Adj SS     Adj MS

Algal concentration       5    182.5346     1.9996     0.3999
Mussel ID                42    101.8942    77.2482     1.8392
  (Algal conc.)
ESA                       1     41.0181    31.3496    31.3496
Algal cone. * ESA         5      3.1315     3.1315     0.6263
Error                   177     55.9794    55.9794     0.3163
Total                   230    384.5579

Source                     F        P

Algal concentration      0.92    0.469
Mussel ID                5.82    0.000
  (Algal conc.)
ESA                     99.12    0.000
Algal cone. * ESA        1.98    0.084
Error
Total

TABLE 5.
Results from the nested ANCOVA for clearance rate of 30 mm mussels
using all algal concentrations (0-3.0 mg [L.sup.-1] Isochrysis galbana),
where algal concentration was the fixed factor, mussel ID was the
random factor nested within algal concentration and ESA was the
covariate.

Source                   df      Seq SS     Adj SS    Adj MS

Algal concentration       5     138.242    17.8995    3.5799
Mussel ID                42     80.6766    73.0312    1.7388
  (Algal conc.)
ESA                       1      1.2413     1.3182    1.3182
Algal cone. * ESA         5      1.2515     1.2515    0.2503
Error                   172     69.7837    69.7837    0.4057
Total                   225    291.1952

Source                    F        P

Algal concentration     7.22    0.000
Mussel ID               4.29    0.000
  (Algal conc.)
ESA                     3.25    0.073
Algal cone. * ESA       0.62    0.687
Error
Total

TABLE 6.
Examples of in situ seston concentrations from sites in the geographic
range of Mytilus edulis.

Location                      TPM (mg [L.sup.-1])   POM (mg [L.sup.-1])

Trinity Bay, NL, Canada              1-10                  0.6-4
Mahone Bay, NS, Canada               0.5-8                0.4-1.2
Bocabec Bay, NB, Canada               1-5                  0.5-4
Georgia Strait, BC, Canada           2-15                  0.7-9
Oosterschelde Estuary,               2-12              Not reported
  The Netherlands
Loch Etive, Scotland                 5-9.1                2.2-4.5
Marennes-Oleron. France              10-90                 4.8-9

Location                       Source

Trinity Bay, NL, Canada        MacDonald (2000)
Mahone Bay, NS, Canada         Cranford and Hill (1999)
Bocabec Bay, NB, Canada        Barrington et al. (2002)
Georgia Strait, BC, Canada     Taylor et al. (1992);
                                 Jones and Iwama (1991)
Oosterschelde Estuary,         Smaal et al. (2001);
  The Netherlands                Smaal and Haas (1997)
Loch Etive, Scotland           Stirling and Okuinus (1995)
Marennes-Oleron. France        Hawkins et al. (1996, 1997)