Sloppy feeding in marine copepods: prey-size-dependent production of dissolved organic carbon

Abstract

An equation predicting dissolved organic carbon (DOC) production by sloppy feeding was developed from literature data on apparent gross growth efficiency and copepod-to-prey size ratio based on the assumption that real gross growth efficiency is independent of relative prey size. The equation suggests that, for copepod-to-prey size ratios <55, DOC production by sloppy feeding can be quantitatively important. It can be calculated as: Q = 0.714 – 0.013 × (ESD_copepod/ESD_prey), where Q is the fraction of carbon removed from suspension that is lost as DOC and ESD is equivalent spherical diameter. Predictions by the equation were corroborated by actual measurements of copepod DOC production from the literature.

Received May 28, 2004; accepted in principle July 16, 2004; accepted for publication September 14, 2004; published online September 30, 2004

INTRODUCTION

Marine copepods may constitute up to 80% of mesozooplankton biomass (Verity and Smetacek, 1996). Thus, it is essential to understand their trophic position in order to describe the fluxes of matter and energy in marine food webs. Copepods are the key link from primary producers towards fish production. Recent studies have, however, shown that their activity can also contribute considerably to microbial food webs through the production of dissolved material from sloppy feeding, excretion and leakage from fecal pellets (Hasegawa et al., 2001; Møller et al., 2003; Steinberg et al., 2004). Thus, not all of the material grazed by copepods is necessarily transferred up the food chain.

The production of dissolved organic carbon (DOC) by sloppy feeding has been suggested to depend on the relative size of the prey (Lampert, 1978; Møller and Nielsen, 2001). Thus, the blooms of large sized phytoplankton, traditionally regarded as most important prey organisms for copepod production and, therefore, fish production, may also result in the highest production of DOC by copepods. However, actual quantitative measurements of DOC production by copepods are scarce (Roy et al., 1989; Strom et al., 1997; Møller and Nielsen, 2001) and do not allow prediction of when this DOC release may be of significance.

When the prey is large relative to the copepod, some of the material cleared is lost as DOC by sloppy feeding (Møller and Nielsen, 2001). Therefore, the actual ingestion by the copepod may be less than that calculated from clearance experiments. In contrast, when the prey is small, the difference between estimates of ingestion based on the amount of food cleared and the actual ingestion will be insignificant. Gross growth efficiency of copepods is, however, usually calculated assuming that ingestion calculated from clearance equals the actual ingestion. Consequently, when the relative size of the prey is above a certain level, there will be a discrepancy between the apparent gross growth efficiency (GGE_apparent) based on ingestion calculated from clearance and real gross growth efficiency (GGE).

In the present study, DOC production by sloppy feeding is estimated from literature data on apparent gross growth efficiency and copepod-to-prey size ratio using the assumption that real gross growth efficiency is independent of relative prey size.

METHOD

When ingestion calculated from clearance is higher than the actual ingestion, i.e. GGE_apparent is lower than GGE, a linear relationship between the GGE_apparent and the ratio between copepod and prey size is predicted:

\[\mathrm{GGE}_{\mathrm{apparent}}\ =\ a\ {\times}\ \frac{\mathrm{ESD}_{\mathrm{copepod}}}{\mathrm{ESD}_{\mathrm{prey}}}\ +\ b\]

(1)

where a and b are constants and ESD_copepod and ESD_prey are the size of the copepod and its prey, respectively, measured as the equivalent spherical diameters (ESD). At higher copepod-to-prey size ratios, when ingestion calculated from clearance equals actual ingestion then GGE_apparent will equal GGE.

GGE_apparent can also be expressed as a function of GGE, the gross growth efficiency of the actually ingested (I) and Q, the fraction of the carbon cleared/removed from suspension (RFS) that is lost as DOC:

\[\begin{array}{lll}\mathrm{GGE}_{\mathrm{apparent}}&=&\frac{\mathrm{Growth}}{\mathrm{RFS}}\ =\ \frac{\mathrm{I}\ {\times}\ \mathrm{GGE}}{\mathrm{RFS}}\\&=&\frac{(\mathrm{RFS\ {-}\ DOC})\ {\times}\ \mathrm{GGE}}{\mathrm{RFS}}\\&=&\frac{(\mathrm{RFS}\ {-}\ Q\ {\times}\ \mathrm{RFS})\ {\times}\ \mathrm{GGE}}{\mathrm{RFS}}\\&=&(1\ {-}\ Q)\ {\times}\ \mathrm{GGE}\end{array}\]

(2)

By combining equations (1) and (2), a relationship for estimating the fraction of the carbon removed from suspension that is lost as dissolved organic matter (DOM) (Q) for a given copepod/prey ratio is established:

\[Q\ =\ 1\ {-}\ \left(\frac{a}{\mathrm{GGE}}\ {\times}\ \frac{\mathrm{ESD}_{\mathrm{copepod}}}{\mathrm{ESD}_{\mathrm{prey}}}\ +\ \frac{b}{\mathrm{GGE}}\right)\]

(3)

Data were compiled from the literature when GGE_apparent and size of the copepod and its prey were included or could directly be calculated. In papers where size measurements of copepod or prey were not provided, the data were not included in the current study. This was necessary as the regressions carried out here are sensitive to relatively small deviations in size. For the same reason, studies providing only the size of single cells in a chain but not the total size of the chain could not be used. If available, information on the C:N ratio, prey concentration, carbon-to-volume ratio of the prey, temperature and methods of determination of GGE_apparent were recorded (Table I). Grazing was estimated as the number of cells or amount of carbon removed from suspension and growth by egg production or somatic growth. Values for GGE_apparent were calculated based on carbon measurements. Copepod and prey volume was converted to ESD, defined as the diameter of sphere with equal volume as the organism. If copepod or prey size was expressed as carbon, it was converted to volume using 0.13 pg C μm⁻³ (Berggreen et al., 1988). Dry weight was converted to carbon assuming a factor 0.5.

Table I:

Published apparent gross growth efficiencies (GGE_apparent) for copepods feeding different prey

Copepod species	Size (ESD, μm)	Prey species	Size (ESD, μm)	Concentration (μg C L−1)			C vol−1 (pg C μm−3)	C:N (mass)	Copepod:prey size ratio (ESD:ESD)	GGEapparent	Temperature (°C)	Method	Reference
				Minimum	Maximum
Calanus	264	Isocrysis galbana	5	291	364	0.15	7.0	57	0.59	15	Somatic growth	(Rey et al., 2001)
helgolandicus nauplii	260	Rhodomonas baltica	8	334	520	0.16	7.0	35	0.42
	247	Pleurochrysis carterae	10	280	505	0.21	6.4	26	0.12
	238	Thalassiosira weissflogii	13	576	768	0.12	4.8	18	0.29
	255	Prorocentrum micans	27	229	429	0.28	7.3	10	0.13
C. helgolandicus	1015	T. weissflogii	13	120	480	0.28	4.3	78	0.12	15	Egg production	(Huskin et al., 2000)
	1015	P. micans	30	202	1680	0.14	8.0	34	0.09
	295	P. micans	27	77.5	77.5	0.28	7.3	11	0.40	15	Somatic growth	(Rey-Rassat et al., 2002)
	375	P. micans	27	77.5	77.5	0.28	7.3	14	0.44
	480	P. micans	27	77.5	77.5	0.28	7.3	18	0.50
	645	P. micans	27	77.5	77.5	0.28	7.3	24	0.26
	813	P. micans	27	77.5	77.5	0.28	7.3	31	0.14
	295	P. micans	27	278	278	0.28	7.3	11	0.15
	394	P. micans	27	278	278	0.28	7.3	15	0.26
	517	P. micans	27	278	278	0.28	7.3	19	0.31
	751	P. micans	27	278	278	0.28	7.3	28	0.28
	986	P. micans	27	278	278	0.28	7.3	37	0.28
Paracalanus parvus	353	Thalassiosira fluviatilis	12	20a	510a	0.10	10.2	29	0.18	18	Egg production	(Checkley, 1980)
	354	T. fluviatilis	14	12a	291a	0.08	5.8	26	0.27
	353	Ditylum brightwelli	11	12a	306a	—	6.1	31	0.29
	353	Gonyaulax polyedra	37	22a	560a	—	11.2	9	0.13
	353	Peridinium trochoideum	18	7a	176a	—	3.5	20	0.45
Acartia tonsa	388	R. baltica	7	55	1280	0.31	—	55	0.49	20	Egg production	(Møller and Nielsen, 2001)
	388	Ceratium lineatum	25	60	640	0.39	—	16	0.11
	388	D. brightwelli	30	130	2570	0.14	—	13	0.06
	388	R. baltica	7	86	1737	0.15	4.1	56	0.39	20	Somatic growth	(Berggreen et al., 1988)
	375	R. baltica	8	0	1700	0.13	4.9	46	0.33	20	Egg production	(Kiørboe et al., 1985)
	319	I. galbana	5	170	4320	0.45	8.7	66	0.24	18	Egg production	(Støttrup and Jensen, 1990)
	319	Dunaliella tertiolecta	7	20	510	0.08	4.9	45	0.40
	319	R. baltica	8	40	1780	0.16	3.8	42	0.34
	319	T. weissflogii	14	10	1800	0.12	9.3	22	0.16
	319	P. micans	27	10	1610	0.08	6.3	12	0.20
Pseudocalanus elongatus	477	Rhodomonas sp.	7	508	580	0.25	5.9	69	0.21	15	Egg production	(Koski et al., 1998)
	546	Gymnodinium simplex	7	290	420	0.14	5.0	74	0.12
	510	Tetraselmis suecica	8	690	780	0.26	9.4	67	0.007
	364	T. weissflogii	11	1710	2200	0.35	8.1	34	0.04
P. elongatus nauplii	184	Rhodomonas sp.	7	508	580	0.25	5.9	27	0.28	15	Somatic growth	(Koski et al., 1998)
	184	Rhodomonas sp.	7	710	1010	0.38	7.7	26	0.29
	184	G. simplex	7	290	420	0.14	5.0	25	0.32
	167	Rhodomonas sp.	7	1100	1100	0.25	5.8	24	0.16
	167	T. weissflogii	11	1710	2200	0.35	8.1	16	0.07
	167	Amphidinium sp.	9	930	1790	0.23	4.5	18	0.05
	167	Amphidinium sp.	9	930	1790	0.23	4.5	18	0.08
P. elongatus copepodite	341	Rhodomonas sp.	7	508	580	0.25	5.9	50	0.17	15	Somatic growth	(Koski et al., 1998)
	341	Rhodomonas sp.	7	710	1010	0.38	7.7	47	0.17
	341	G. simplex	7	290	420	0.14	5.0	46	0.20
	267	Rhodomonas sp.	7	1100	1100	0.25	5.8	39	0.15
	267	Rhodomonas sp.	7	1620	1620	0.35	7.7	36	0.12
	267	T. weissflogii	11	1710	2200	0.35	8.1	25	0.03
Temora longicornis	533	T. weissflogii	11	25	600	0.15	5.3	50	0.09b	14–17	Egg production	(Dam and Lopes, 2003)
	533	Heterocapsa triquetra	13	25	600	0.20	5.1	42	0.09b
	533	Uronema sp.	10	25	600	0.29	3.8	52	0.09b

ESD, equivalent spherical diameter.

^a

Readings from figure.

^b

Shared values for different prey.

Although egg production has been shown to not always be equivalent to total growth of females (Hirst and McKinnon, 2001), the majority of GGE_apparent reported used this method and no discrimination was made here between the methods for estimating GGE_apparent. The authors of two studies (Huskin et al., 2000; Dam and Lopes, 2003), however, indicated that they believed their egg production rates to be unusually low. Both studies used females from the field captured late in their growth season, when egg production can be expected to be low. Therefore, data from these studies (although they are identified in Table I) were not included in the statistical analyses performed. It has been suggested that egg-carrying copepods have a different life strategy than free spawners including lower fecundity (Kiørboe and Sabatini, 1994). Hence, data from the one experiment with Pseudocalanus elongatus females were also excluded from the statistical analyses (Koski et al., 1998).

GGE_apparent is, inevitably, influenced by other factors than the copepod-to-prey size ratio. Prey concentration is an important factor to consider. However, prey concentration during the egg production experiments reported in the literature covers a sufficient range of concentrations for calculating GGE as a regression between ingestion and egg production, typically from ∼0 to >1000 μg C L⁻¹, not allowing for analysing the impact on GGE. The studies reported in the literature measuring somatic growth generally used rather high food concentrations––well above 200 μg C L⁻¹ (Table I). The only exception was Rey-Rassat et al. (2002), who found higher GGE_apparent at low (78 μg C L⁻¹) than at high (278 μg C L⁻¹) food concentrations. To decrease the impact of food concentration, this one exception (78 μg C L⁻¹) was excluded from the dataset when performing the statistical test.

To evaluate the impact of different factors potentially affecting GGE_apparent, Pearson correlation coefficients were calculated between GGE_apparent and copepod-to-prey size ratio, temperature, C:N ratio of the prey, the absolute size of the prey and the carbon-to-volume ratio of the prey. To further test for the influence of diatoms possibly being a low quality food source (Kleppel, 1993), correlation coefficient excluding diatom data were also calculated. The Statistical Package of Social Science (SPSS version 10.0) for Windows was used for all statistical tests.

To validate the predictions of DOC production from copepod-to-prey size ratio, data on real DOC measurements were obtained from the literature (Table II). Only a few studies give well-defined copepod/prey ratios and measured DOC production from copepods, simultaneously. Most of these studies do not distinguish between DOC sources; i.e. they include DOC produced by sloppy feeding, excretion and leakage from fecal pellets. The equation only predicts DOC production by sloppy feeding. However, the potential contribution of DOC production through leakage from fecal pellets was evaluated assuming constant assimilation efficiency of 60% (Conover, 1966), and an immediate leakage of DOC from fecal pellets of 28% of total fecal pellet production (Møller et al., 2003). The results of DOC production presented by Copping and Lorenzen (1980) were not included here as no attempts were made in that study to decrease bacteria during the 48 h incubation. Thus, it is likely that bacteria had taken up a substantial part of the produced DOC.

Table II:

Published dissolved organic carbon (DOC) production by copepods

Copepod species	Size (ESD, μm)	Prey species	Size (ESD, μm)	Copepod-to-prey size ratio (ESD:ESD)	DOC production RFS−1	DOC production by	Comment	Reference
Temora longiconis	612			2.4	0.79	S, E	Dissolved and particulate debris	(Roy et al., 1989)
Calanus helgolandicus	1391	Coscinodiscus wailesii	257	5.4	0.95	S, E	Dissolved and particulate debris	(Roy et al., 1989)
Calanus pacificus	1060a	Oxyrrhis marina	18	58	0	S, E, F		(Strom et al., 1997)
	1060a	Thalassiosira weissflogii	12	87	0	S, E, F
	1060a	Prorocentrum minimum	14	75	0.16	S, E, F
Acartia tonsa	388	Rhodomonas baltica	7	55	0	S, E, F		(Møller and Nielsen, 2001)
	388	Ceratium lineatum	25	16	0.59	S, E, F
	388	Ditylum brightwelli	30	13	0.62	S, E, F

E, excretion; ESD, equivalent spherical diameter; F, leakage from fecal pellets; RFS, removed from suspension; S, sloppy feeding.

^aSize from Runge (1980).

RESULTS

Despite the scatter caused by the different variables unavoidably influencing GGE_apparent, a significant increase was found with increasing copepod-to-prey size ratio (Fig. 1).

\begin{eqnarray*}&&\mathrm{GGE_{apparent}}\ =\ 0.0045\ {\pm}\ 0.0012\ {\times}\ \frac{\mathrm{ESD_{copepod}}}{\mathrm{ESD_{prey}}}\ +\ 0.1014\\&&{\pm}0.0416,\ r^{2}\ =\ 0.26,\ P\ =\ 0.0009,\ n\ =\ 37\end{eqnarray*}

(4)

Fig. 1.

Apparent gross growth efficiency (GGE_apparent) as a function of the copepod-to-prey size ratio. Dotted lines indicate the 95% confidence intervals for the filled line. References in Table I.

No significant correlation was found between GGE_apparent and any factor other than the copepod-to-prey size ratio (Table III).

From the relationship between GGE_apparent and the copepod:prey size ratio, a prediction for DOC production was obtained. To produce this equation an estimate of GGE (the growth efficiency when there is no DOC production by sloppy feeding) was obtained from equation (4) using copepod:prey size ratio of 55. This is assumed to be the best estimate of GGE available. Thus, the fraction of prey carbon removed from suspension, which is lost, as DOC (Q) for copepod:prey size ratio below 55 can be calculated as:

\[Q\ =\ 0.714\ {-}\ 0.013\ {\times}\ \frac{\mathrm{ESD}_{\mathrm{copepod}}}{\mathrm{ESD}_{\mathrm{prey}}}\]

(5)

The literature provides only a few reports of DOC production by copepods and all of these were carried out with copepods feeding either on small or very large prey. None of the studies were carried out for copepod-to-prey size ratios between 16 and 55. Nevertheless, these literature measurements corroborate the predictions of high DOC production when the prey is large relative to the copepod, and low DOC production when the prey is small relative to the copepod (Fig. 2). A deviation between the model and the laboratory measurements was expected because most of the actual experiments included DOC produced by excretion and leakage from fecal pellets. An estimate of the additional DOC production from leakage from fecal pellets fitted well with the observed DOC production.

Fig. 2.

Model prediction of DOC production (solid and dotted lines) and actual measurements (dots) of DOC production (Q) as a function of copepod-to-prey size ratio. References in Table II.

DISCUSSION

The present study illustrates that when the prey is large relative to the copepod, e.g. during a bloom of large cells, copepods are not an efficient link to higher trophic levels, but lose significant amounts of what they clear to the surroundings as dissolved material. In contrast, the link between copepod feeding and energy flow to higher trophic levels is tighter when the prey is small, e.g. during oligotrophic periods when small cells dominate the phytoplankton prey. The DOC produced by copepods may in some situations contribute significantly to the DOC pool. For instance, if 50 μg copepods L⁻¹ are grazing 100% their own bodyweight per day on large prey only 10 times smaller than themselves, the prediction is that the DOC production would be 85 μg C L⁻¹ day⁻¹. On the other hand, if the prey were so small that no DOC is predicted to be produced by sloppy feeding, the DOC production by leakage from fecal pellets from the same 50 μg copepods L⁻¹ would be 6 μg C L⁻¹ day⁻¹.

The validation of the prediction is complicated by the fact that most of the laboratory studies (Table II and Fig. 2) included in this study did not separate excretion and leakage of DOC from fecal pellets from the contribution from sloppy feeding. However, there is no reason to believe that excretion and fecal pellet DOC production should be dependent on the copepod-to-prey size ratio. The relative importance of the DOC excretion and leakage from fecal pellets will be greatest when the contribution to DOC production from sloppy feeding is small, because more prey carbon is then actually ingested. An estimate of fecal pellet DOC production is provided in Fig. 2. No estimate of C excretion as a function of grazing was found in the literature. However, a contribution from excretion should result in a parallel displacement of the indicated dotted line. The 16% DOC loss of carbon cleared found by Strom et al. (1997) when prey was small (Table II) could possibly be explained by leakage from fecal pellets (Fig. 2). That no DOC production was found in the other experiments with high copepod-to-prey size ratio, even though the contribution from excretion and leakage from fecal pellets was included, could be due to the detection limits of the experiments. Thus, overall the developed equations predicting DOC production by sloppy feeding are supported by laboratory results on DOC production by copepods carried out for small and large sized prey (Fig. 2).

The feeding ecology of cladocerans differs from that of copepods. However, laboratory experiments performed with freshwater cladocerans have also shown a relationship between the relative size of the prey and DOC production although at a lower level. Up to 17% of the ingested carbon was lost as DOC due to sloppy feeding when the prey was large, while only 4% was lost when cells were small enough to be swallowed directly (Lampert, 1978).

During diatom blooms, sloppy feeding was, by far, found to be the most important contributor to the DOC production by Calanus spp., and 49% of the carbon removed from suspension by the copepods was returned to the water column as DOC (Møller et al., 2003). Hasegawa et al. (2001) did not discriminate experimentally between sources of dissolved material produced by Acartia spp., Pseudocalanus spp. and Paracalanus spp. Nevertheless, during blooms of large diatoms, they found a total release of up to 91% of dissolved organic or inorganic nitrogen that had been removed from suspension by the copepods and assigned it primarily to catabolism and sloppy feeding.

When grazing by copepods is estimated from clearance and egg production and/or gut fluorescence, an inconsistency is sometimes found. The loss of DOC through sloppy feeding provides an additional possible explanation to those already proposed for this discrepancy, e.g. female weight changes and pigment destruction. If the copepod-to-prey size ratio is known, the production of DOC can be estimated by equation (5), thereby enabling an evaluation of the potential magnitude of this fate for the carbon. How much of the cleared carbon that is, potentially, lost as DOC should also be considered before applying literature values of GGE_apparent to obtain growth or production from ingestion calculated from measurements of clearance or the grazing impact on lower trophic levels from egg production. GGE_apparent may be significantly lower when phytoplankton is dominated by large cells than by small (Fig. 1).

Straile (1997) examined the impact of copepod-to-prey size (weight) ratio on GGE_apparent, but using a more heterogeneous dataset than that used here. Straile’s study included copepods from both marine and freshwater environments. He did not observe any significant relationship but did find low GGE_apparent both when the copepod-to-prey ratio was low and when the copepod-to-prey ratio was high; he found the highest GGE_apparent at intermediate ratios within the range 22–46 (recalculated from weight).

Factors such as food quality (Checkley, 1980; Koski et al., 1998; Jones et al., 2002), food concentration (Checkley, 1980; Rey-Rassat et al., 2002) and temperature (Straile, 1997) have all been shown to affect GGE_apparent. None of these other factors correlated significantly with GGE_apparent in the present study (Table III). Furthermore, the impact of these factors, ideally, is equally distributed along the regression and therefore only contributes to the scatter in Fig. 1, but does not change the slope of the linear regression. However, the intercept of the regression between copepod-to-prey size ratio and GGE_apparent would be higher if any other factor influencing GGE was not optimal. This would influence the determination of Q, the fraction of the cleared food that is lost as DOC. A conservative estimate, assuming the intercept to be two times higher than what was actually found (i.e. ∼80% of the actual measurements would then be below the regression line), results in Q = 0.555 – 0.010 × (ESD_copepod/ESD_prey). Thus, although this model predicts lower DOC production, the prediction is still considerable and emphasizes the importance of DOC production by sloppy feeding when the prey is large. Moreover, this latter model fits far worse with the actual laboratory measurements.

Table III:

Pearson correlation coefficients between GGE_apparent and variables

		P	n
Copepod-to-prey size ratio	0.518	0.001	37
Copepod-to-prey size ratio (excluding diatom data)	0.495	0.005	31
C:N	−0.328	0.058	34
Temperature	0.215	0.200	37
Prey size	−0.250	0.136	37
C vol−1	−0.302	0.083	34

An alternative explanation, for an increase in GGE_apparent with increasing copepod-to-prey size ratio, could be that small cells are better food for the copepods than large cells, either because small cells contain more carbon per volume (Verity et al., 1992) or because the large cells are diatoms, which in some studies have been suggested to be low quality food sources for copepods. However, neither the absolute size of the prey nor the carbon-to-volume ratio correlated significantly with GGE_apparent (Table III). In fact, the correlation between the carbon-to-volume ratio and GGE_apparent was negative, contrary to what would be expected to support this alternative explanation. Furthermore, when diatom data were excluded from the correlation, a significant correlation was still found (Table III). Thus, it is the copepod-to-prey size ratio that explains the variation and not the absolute size of the prey itself.

Using ESD as a proxy for size may also itself create some variation. It definitely makes a difference for the copepod if the prey of a given size measured as ESD is elongated or round. Concerning the size of the copepod, it is probably the size of the copepod mouth that determines how large a prey the copepod can handle without breaking it. However, there is probably not a close correlation between copepod ESD and the size of its mouth.

Focussing on subsets of the data (Table I and Fig. 1) limited to one species/stage indicates that differences between species in feeding or life strategy could be important contributors to the variation observed in Fig. 1. For Acartia tonsa, the copepod-to-prey size ratio alone can explain 58% of the variation in GGE_apparent. The data for P. elongatus copepodites cover a smaller range of copepod-to-prey size ratios (25–50) than those for A. tonsa (12–56) (Table I). Still, the copepod-to-prey size ratio alone can explain 88% of the variation. The other species/stages vary more, but also generally cover a smaller range in copepod-to-prey size ratio.

The present study suggests that prey should be ∼50–60 times smaller on a biomass (i.e. biovolume) basis than the copepod for the predator to achieve the highest GGE_apparent. Hansen et al. (1994) compiled data to estimate the optimum size ratio between predator and prey for obtaining maximum clearance. For copepods, they found the optimum ratio to be 18:1. Compared to the present dataset and to Straile’s (1997), who found the highest GGE_apparent of copepods within the range 22–46, there is no match between the preferred prey size and high GGE_apparent. Small food particles that, according to Hansen et al. (1994) should be cleared with low efficiency, can support growth with much higher GGE_apparent than the larger cells that, according to Hansen et al., are preferentially cleared.

The present study supports the emerging picture of a more diverse trophic role of copepods, and the approach developed [equation (3)] offers a possibility to indirectly estimate the potential fuelling of the microbial food web through DOC production by sloppy feeding. However, more laboratory measurements of DOC production are certainly needed, especially at the medium range of prey sizes. The data on GGE_apparent and copepod-to-prey size ratio suggest that a strategy to enable more precise predictions of DOC production might be to investigate DOC production by a single copepod species representative of groups with the same morphology or feeding strategy.

The constructive criticism by B. W. Hansen, T. Kiørboe, M. Koski, T. G. Nielsen and K. Richardson is greatly acknowledged. The work was financially supported by Danish Natural Science Research Council Grant Number 9700196.