JOURNAL OF PLANKTON RESEARCH j VOLUME 27 j NUMBER 10 j PAGES 987–1001 j 2005 Copepod foraging and predation risk within the surface layer during night-time feeding forays ANDREW W. LEISING1*, JAMES J. PIERSON2, SCOTT CARY2 AND BRUCE W. FROST2 1 NOAA, ENVIRONMENTAL RESEARCH DIVISION, SOUTHWEST FISHERIES SCIENCE CENTER, 1352 LIGHTHOUSE AVENUE, PACIFIC GROVE, CA 93950, USA AND 2 SCHOOL OF OCEANOGRAPHY, UNIVERSITY OF WASHINGTON, SEATTLE, WA *CORRESPONDING AUTHOR: 98150, USA [email protected] Received June 21, 2005; accepted in principle July 26, 2005; accepted for publication October 7, 2005; published online November 22, 2005 Communicating editor: R.P. Harris Vertical distribution data seem to indicate that certain species of diel vertical migrating copepods avoid the surface high chlorophyll (Chl) region within coastal and estuarine environments, even during the night. Copepods may avoid this layer to reduce predation mortality, avoid advective loss or to avoid consuming too much toxic algae. We hypothesize that copepods make several intermittent feeding ‘forays’ into shallow surface layers during the night, returning to intermediate depths between forays. Using an individual-based model (IBM) of Calanus pacificus, we examined the implications of this behavior on feeding success and mortality risk, and tested whether a practical field-sampling scheme would be able to detect foray-like behavior. In some cases, mortality of the foray-foraging copepods was up to 50% less than that of randomly behaving controls, for a given amount of food ingested. The trapping scheme devised should be able to detect the occurrence of foray behavior (FB) in the field and should show differences in the gut contents of copepods entering and leaving the uppermost food-rich layer. The presence or absence of foray-like behavior significantly altered the relative concentration of copepods within various surface strata and thus could influence the temporal availability of copepods as prey for the larvae and juveniles of several important managed fish species. INTRODUCTION Copepods exhibit a wide array of foraging behaviors across many spatial scales. The largest scale (for the copepod) includes well-studied behaviors such as diel vertical migration, whereby a copepod typically resides at depth during the day and may swim hundreds of meters to the surface to feed during the night. The adaptive value of this behavior is generally considered to be avoidance of mortality caused by visual predators that are highly abundant in surface layers and are thought to feed during the daytime (Zaret and Suffern, 1976; Ohman, 1990; Lampert, 1993). Such predators primarily include vertebrates such as small planktivorous fish and the larvae or juveniles of certain fish species. At the smallest scale relevant to a copepod’s foraging behavior (e.g. the microscale; mm and s), copepods exhibit moderately complex behaviors that vary by copepod species and prey type. Such behaviors have been documented through various laboratory studies utilizing high-speed cinematography (Alcaraz et al., 1980; Paffenhöfer et al., 1982) and other video-based observation systems (Landry and Fagerness, 1988; Tiselius and Jonsson, 1990). At more intermediate scales (e.g. the fine scale; 10s of m and h), much less is known about the foraging behavior of copepods, since it is currently impossible to track individual copepods in situ over time periods greater than a few seconds and technically difficult if not impossible to set-up and track individuals within tanks simulating water columns of sufficient size to allow unrestricted travel. Because of this, little is known about how individual copepods forage on these scales, particularly for species that undergo diel vertical migration and feed within the surface layers during the night. What field data there are at these intermediate scales suggest complex and temporally varying interactions between copepods and their food. For example, once This paper is one of six on the subject of the role of zooplankton predator–prey interactions in structuring plankton communities. doi:10.1093/plankt/fbi084, available online at www.plankt.oxfordjournals.org Published by Oxford University Press 2005. j VOLUME vertically migrating copepods reach the surface, they are typically not distributed evenly throughout the upper layers. Copepods may sometimes be found associated with subsurface chlorophyll (Chl) maxima (Sameoto, 1984; Townsend et al., 1984; Harris, 1988; Napp et al., 1988; Castro et al., 1991), which increases ingestion and subsequently growth or reproduction rates. However, numerous other studies have found copepods at higher abundances below the layer of maximum Chl concentration (Mullin and Brooks, 1976; Sameoto, 1984; Simard et al., 1985; Napp et al., 1988; Bjørnsen and Nielsen, 1991; Bollens et al., 1993; Tarling et al., 2002). Why would a copepod avoid the region with maximum food availability? Three obvious possibilities are that (i) such behavior would minimize time within high food regions if those regions also expose the copepod to higher predation risk, (ii) surface layers may be highly advective relative to deeper strata and displace the copepod to a less favorable region (Peterson et al., 1979, Peterson, 1998) and (iii) high-phytoplankton regions may contain toxic diatoms or other harmful algae (Ban et al., 1997; Miralto et al., 1999; Ianora et al., 2004; Pierson et al., 2005a). There is significant evidence supporting the first hypothesis, particularly when considering copepods that undertake diel vertical migrations. Whereas the surfacemost layer may often contain high food levels, this layer may also contain the highest number of predators, visual or otherwise. Although fish have traditionally been thought of as primarily daytime feeders, since they depend on visual means to locate prey, many larval and juvenile fish can feed quite readily under low-light conditions typical of night-time levels (Neilsen and Perry, 1990, and references therein). Indeed, the larvae of many fish species also undergo diel vertical migrations, swimming to depth during the day, and returning to the surface during the night, to avoid even larger piscivorous visual predators. Thus, they may be adapted to feed at night under low light levels (Nielsen and Perry, 1990; Macy et al., 1998; Schabetsberger et al., 2000; Sørnes and Aksnes, 2004). Since the feeding rates of these larvae are typically correlated with light intensity, however, predation rates on copepods will increase nearer the surface, particularly during the night when light may preclude the ability of these fish to feed at intermediate depths. Yet copepods may still need to risk feeding within this higher risk upper surface layer in order to obtain an adequate daily ration. We hypothesize a relatively generic foraging behavior that should allow a copepod to achieve a dynamic balance between the time it needs to feed within the surface layer, and risk aversion through avoidance of this layer during non-feeding periods. We propose that copepods might make short intermittent ‘forays’ into the surface, high- 27 j NUMBER 10 j PAGES 987–1001 j 2005 concentration high-mortality-risk layer to feed, returning to intermediate depths between these forays, during which they reduce their feeding activity while they digest (Fig. 1). Such behavior of plankton has been suggested before (Gauld, 1953; Pearre, 1973; Mackas and Bohrer, 1976; Pearre, 1979) and has been termed the ‘hunger/satiation hypothesis’, as used to explain ‘midnight sinking.’ Evidence supports the contention that copepods do not continuously feed, even during the night (Mackas and Bohrer, 1976; Mackas and Burns, 1986; Rodriguez and Durbin, 1992; Kleppel et al., 1988). In these cases, a copepod may fill its gut rather rapidly, on the scale of minutes to an hour, and then require a similar period of time, or longer, to digest before beginning to feed again. This ‘refractory’ period, during which the copepod may also need time to rebuild its enzyme pool (Head et al., 1984; Mackas and Burns, 1986; Hassett and BladesEckelbarger, 1995), could serve as a natural time during which a copepod might leave the more dangerous or harmful surface region. Beyond the reasons listed above for avoiding the surface layer, there may also be other physiological or energetic reasons that make it advantageous to sink out of the surface layer between feeding events (as reviewed in Pearre, 2003). Here, we examine whether such a proposed foraging behavior could confer an advantage of increased feeding and/or reduced mortality for copepods. We used an individual-based model (IBM) of adult female Calanus pacificus, feeding within a water column characterized similarly to natural conditions found during the spring bloom within Dabob Bay, WA. Because many of the parameters controlling the foray behavior (FB) are poorly constrained, we further conducted a detailed sensitivity analysis of the effects of our parameter choices on both copepod fitness and the ability to detect FB through a simulated field-sampling scheme. Lastly, we also investigated how foray versus non- FB s could 0 High chlorophyll Feeding 15 Depth (m) JOURNAL OF PLANKTON RESEARCH Low chlorophyll 50 Digesting 100 Sunset Time Sunrise Fig. 1. Conceptual diagram of the foray behavior (FB). Copepods swim into the upper, high-food, high-mortality layer to rapidly feed, then sink to the lower, low-food, low-mortality layer during a refractory digestive period, returning to the surface multiple times during the night to feed. Shading of copepod body represents relative gut fullness. 988 A. W. LEISING ETAL. j COPEPOD FEEDING FORAYS differentially affect the prey field and feeding success of copepod predators, which includes the larvae of several important managed fisheries species, such as hake, herring, northern anchovy and salmon (Bollens and Frost, 1989). This particular ecosystem was chosen to test this hypothesized behavior for several reasons: C. pacificus is known to undertake diel vertical migrations to avoid surface visual predators (Osgood and Frost, 1994), some of which are still active at night (Bollens and Frost, 1989); the surface mixed layer within Dabob Bay typically flows out of the bay and is where the highest concentration of phytoplankton is found during the spring (Osgood and Frost, 1996; Dagg et al., 1997); the surface spring bloom often contains diatoms known to have harmful effects on the reproduction of C. pacificus (Pierson et al., 2005a). Thus all three of our hypothetical criteria that should favor FB are present within this system. Lastly, C. pacificus has been observed to have relatively long (1–3 h) low activity, non-feeding periods between active feeding bouts (Mackas and Burns, 1986), which would facilitate sinking out of the upper surface layer, as in our proposed foraging mechanism. METHODS Overview One thousand copepods were placed within a simulated 200 m deep water column, with 20 mg Chl L1 within the surface 0–15 m layer, and 1 mg Chl L1 within a 15–50 m intermediate layer, similar to situations previously observed within Dabob Bay (Frost, 1988; Dagg et al., 1997; Pierson et al., 2005b). Copepods were given swimming and feeding behaviors based on our foray hypothesis (described in detail below). Mortality probability was arbitrarily set to 10% and 1% per night within the upper and lower layers, respectively. We consider mortality to be higher within the upper layer due to (i) visual predation, which is light dependent and greater near the surface even during the night, and (ii) advective loss of the surface layer, due to estuarine flow. Although we do not have direct estimates of the absolute value of copepod mortality from Dabob Bay, the critical point here is to examine the effects of the relative mortality between layers; our sensitivity analysis below investigates changes in the ratio of mortality between the layers. A simulated zooplankton trap was placed at 17 m depth, just below the high concentration surface layer. So as not to alter the overall concentration of copepods within the simulation, the trap does not stop the copepods as they swim between layers, but simply notes their direction of passing. The total number of copepods that swam up or down past the trap over 45-min intervals was thus recorded. The gut contents (measured in units of ng Chl) of all copepods which passed by the ‘trap’ were also recorded at the end of the 45-min interval. To more realistically simulate this trapping process for gut contents, at the exact moment when a copepod swam past the ‘trap’, an identical duplicate of that copepod was created and held at the trap depth for the remainder of the 45-min period, during which it was not allowed to feed but continued to digest its stomach contents. At the end of the 45-min interval, the gut contents of these duplicate copepods were then recorded, and these duplicate copepods were then discarded from the simulation. This directional-trapping scheme was chosen as it reflects a fairly simple and practical method that could theoretically be used in the field to measure the flux of copepods between surface layers, even though it is not a regularly used method. Also at 45-min intervals, we simulated sampling the water column as would be done using opening-and-closing vertical net tows, sampling the surface, intermediate, and deepest layer, in order to obtain the relative abundance of copepods within each layer over the course of the night. The simulation starts (t = 0 h) at the point when copepods would first migrate up from their daytime depth (some point near local sunset), and ends12 h later (t = 12 h), just after local sunrise. The model time step was 10 s. Model equations and parameters The model can be broken down into five major behavioral components: (i) Initial vertical migration to shallow layers from depth (ii) Spatial foraging control within the upper layers (iii) Ingestion and gut filling (iv) Refractory period and sinking (v) Termination of foraging behavior and migration to depth Initial vertical migration All copepods started the simulation at a depth of 195 m. The timing of surface ascent for each copepod was chosen randomly from a normal probability distribution function, with a mean value of 1 h after the start of the simulation, and a standard deviation of 30 min for the standard runs. This formulation led to 68% of the population starting its migration toward the surface within a window from 30 to 90 min after the start of the simulation (68% is the integrated area within 1 stdev of the mean of a Gaussian curve). Migration speed of the copepods was set to 100 m h1 and was kept constant for all runs. Each copepod migrated up at this 989 JOURNAL OF PLANKTON RESEARCH j VOLUME speed until it reached a depth somewhere between 0 and 50 m that was randomly predetermined for each copepod, at which point the copepod began to forage and feed. Spatial foraging Copepods foraged within the simulated water column using a vertical random walk. For an unbiased random walk, the copepod has a 50/50 chance of swimming either up or down at each time step (10 s for this model), with a step size set by, Sstep. However, to simulate the FB, which calls for the copepods to swim upward into the high-concentration layers while feeding, the distance of upward motions was set to 20% greater than Sstep, and the distance of downward motions was 20% less than Sstep. For the standard run, Sstep was set such that the average swimming speed was 2.5 cm s1, similar to the maximum swimming speeds seen for C. pacificus (Landry and Fagerness, 1988). Ingestion and gut filling Ingestion rate as a function of food concentration was modeled following a Michaelis-Menton type-II curve, where the ingestion rate, I was I ¼Imax C ðks þC Þ ð1Þ where Imax is the maximum ingestion rate (ng Chl h1), C is the concentration of phytoplankton at the location of the copepod (mg Chl L1) and ks is the half-saturation constant (mg Chl L1). Imax was set to 100 ng Chl h1 and ks was set to 5 mg Chl L1 (Frost, 1972, 1985). Imax was chosen so that, when combined with the digestion and refractory periods (described below), average ingestion over longer periods would be similar to observations (referenced below). Gut content, G, at any instant in time, was modeled as the combination of the ingestion rate, minus the gut clearance rate such that dG ¼I GR dt ð2Þ where R is the instantaneous gut clearance rate coefficient (h1), and I and C are as above. Each copepod was given a unique value of R for the duration of each model run, drawn from a uniform random distribution of 25% of the standard value of R, which was set at 2 h1, based on data from laboratory experiments (Mackas and Burns, 1986; Dagg et al., 1989). This variability was added to reflect the possibility that within a population, copepods will have slightly different physiological states, based on their individual history, and may also have different body weights (copepod weight was not explicitly modeled). 27 j NUMBER 10 j PAGES 987–1001 j 2005 Copepod gut content was allowed to increase to a maximum of Gmax, which was set to the equivalent of 30 ng Chl, based on levels of average individual gut content measured in both field and laboratory experiments (Mackas and Burns, 1986; Dagg et al., 1989, 1997). Refractory period and sinking Once a copepod’s gut content, G, reached Gmax, the copepod was considered to enter a ‘refractory’ period, during which feeding ceased, yet digestion continued. During this period, gut contents decreased according to dG ¼ GR dt ð3Þ until G reaches H, the hunger threshold (ng Chl) at which feeding recommences. H was set to 5 ng Chl, based on field and laboratory measurements of minimum gut contents (Mackas and Burns, 1986; Dagg et al., 1989, 1997). During this refractory period, the FB copepods ‘sink’ at a fixed speed of 1 cm s1. This speed is 4 higher than measured sinking speeds for C. pacificus (2.5 mm s1; Landry and Fagerness, 1988) but is meant to be consistent with our view that downward movement may consist of both sinking and periods of active swimming. Using a novel plankton wheel apparatus, Hardy and Bainbridge (Hardy and Bainbridge, 1954) observed sustained periods of downward swimming for Calanus finmarchicus, with average speeds ranging from 2.9 cm s1 over a 2-min interval to 1.3 cm s1 over a 60-min interval. As there are no published data like this for C. pacificus, our analysis below includes an examination of the sensitivity of the model to decreasing the sinking speed from this value. Upon G reaching H, the copepods ‘awaken’ and continue to feed and swim again with the bias random walk described above. Given the above parameters for ingestion and gut clearance, average time to maximum gut fullness (assumed to be 30 ng Chl), within the high-concentration layer (20 mg Chl L1) was 36 min. Average ingestion rate over a 6 h period, including the non-feeding digestive periods, was 33 ng Chl h1, and average time for the gut to empty from a full stomach down to the hunger threshold was 45 min [these rates and parameters are similar to those previously found for C. pacificus from Dabob Bay (Frost, 1985; Dagg et al., 1989, 1997)]. Migration to depth At a fixed time after the start of the simulation (t = 9.5 h), all copepods were forced to migrate back to 195 m at the previously used migration speed of 100 m h1, regardless of their condition or location within the surface layers. Copepods did not feed during this downward migration. 990 A. W. LEISING ETAL. j COPEPOD FEEDING FORAYS Standard run Given the above parameterizations, we ran a standard scenario where we compared the hypothesized FB to a random ‘control’ behavior. For the random control (RC) behavior, all parameters were the same as the FB; however, the copepods have no upward bias in their random walk (equal upward and downward step sizes). These control copepods cease feeding during the refractory period, just as in the FB; however, the control copepods do not sink during this time and continue to swim randomly up and down. Sensitivity analysis As many of the parameters controlling the various aspects of the copepods’ physiology and foraging behavior are not well constrained, a sensitivity analysis of the effect of altering these parameters was conducted. For each parameter set to be tested (summarized in Table I), the model was run six times: twice (once with FB and once with the RC behavior) at the standard parameter set, twice (foray and control) at a higher value of a single parameter and twice (foray and control) at a lower value of the parameter. Both biological and environmental parameters were considered. The biological parameters investigated were the gut clearance rate, R, the maximum ingestion rate, Imax, the hunger threshold, H, the swimming speed (step size), Sstep, and the ‘sinking’ speed. All biological parameters were varied 25% from their standard value, except the hunger threshold, H, which was varied by 50%. The ‘environmental’ parameters investigated were the relative height of the layers, the relative concentration of the layers and the relative mortality rates between the layers. Lastly, we examined the effect of altering the length of the trap interval on our ability to detect FB, as a more practical consideration. For each simulation with a changed parameter value, we calculated the percentage change in average gut contents of both upward- and downward-caught copepods versus their paired run using the standard parameter value. Average gut content was calculated as the average of all copepods trapped in samples taken from t = 3 to t = 9 h (n = 9). The initial and final samples were not used, as they contained copepods which had either just migrated up from depth at the start of the simulation or were forced to descend at the end of the simulation, regardless of their physiological state. Therefore, the motion of these copepods past the traps was not foraging-behavior dependent and not relevant to the comparison. Within each run, to test for differences in gut contents between upward- and downward-traveling copepods, we calculated an analysis of variance (ANOVA), comparing the four treatments: FB copepods caught swimming up, FB copepods caught swimming down, control behavior copepods caught swimming up and control behavior copepods caught swimming down. In all cases except for altering the concentration of food within the lower layer, the ANOVA revealed a significant difference between at least two of the treatments. We then conducted post-hoc paired t-tests, pairing the samples by time, comparing the upwardversus downward-caught copepods for either FB or the control behavior, using only those samples from t = 3 to t = 9 h (thus n = 9 for these t-tests). Table I: Standard parameter values Parameter Value Description Units Imax 100 (75,125) Maximum ingestion rate ng Chl h1 ks 5 Ingestion half-saturation constant mg Chl L1 H 5 (2.5, 7.5) Hunger threshold ng Chl R 2 (1.5, 2.5) Gut clearance rate h1 Sstep 0.025 (.018, .031) Average step size m Gmax 30 Maximum stomach content ng Chl Sink 0.01 (.0075, .0125) ‘Sink’ speed m s1 Cu 20 Upper layer Chl concentration mg Chl L1 Cl 1 (5, 10) Lower layer Chl concentration mg Chl L1 Mu 10 Mortality in the upper layer Individual probability (%) Ml 1 (5, 7.5) Mortality in the lower layer Individual probability (%) Lu:Ll 15:35 (20:30, 25:25) Relative layer vertical size, upper : lower m Deltat 10 Time step of model s Numbers in parentheses indicate the altered value of the parameter used for the sensitivity analysis. Chl, chlorophyll. 991 To compare differences in copepod success among the different parameter sets, two metrics were calculated: copepod fitness (the ratio of food ingested to probability of mortality) and feeding success (total amount of food ingested per night). Only those copepods that ate a certain minimum ration were used for the calculation of average copepod fitness, with the minimum ration defined as the mean amount of food eaten per night by the individuals from a particular run. The comparison was limited in this fashion, because many of the copepods using the RC behavior never fed within the highfood, high-mortality layer. Thus, they had low mortality probabilities but also very low nightly ingestion. Although we did not model it explicitly, copepods require a certain minimum daily ration in order to meet their daily metabolic requirements, and in the case of the current model, they also need to meet the energetic demands associated with their diel vertical migration. Without knowing exactly what this minimum daily ration was for our simulated copepods, we therefore used the average minimum ration defined above as a conservative means for eliminating copepods that may not have met their metabolic requirements from the comparison. The fitness and feeding success metrics for the FB copepods were then compared with the values for their paired RC. This was done as a percent anomaly, calculated as the difference between the average value of the metric for the foray copepods and the paired RC, all divided by the average value of the metric for the control copepods. We also compared the fitness metric between the foray and control behaviors for each paired run set using a two-sample t-test with unequal variance. To assess effects of this behavior on the feeding success of potential copepod predators, in a similar manner we also calculated the anomaly of copepod abundance within the surface and intermediate layer, comparing each foray-behavior run with its paired random-behavior control. 27 NUMBER 10 j PAGES 987–1001 j 2005 pycnocline (maximum was 36 mg L1 at 6 m depth) and a sharp decrease to <1 mg L1 below 12 m. To mimic this situation in the model, the upper layer depth was set to 0–10 m, with the lower layer set to 10–50 m, with Chl values within these layers set to 16 and 1 mg L1, respectively. The standard parameter values were used to control the simulated copepods, with no additional ‘tuning.’ To compare the model data with the fieldcollected data, simulated vertical net tows were taken at similar intervals during the model run, sampling the same depth bins as the field data. Because there was some advection in the field, and some variability in total numbers from sample to sample, all data was scaled within a particular cast by the maximum number of copepods collected so that the samples could be more readily compared over time. RESULTS Standard scenario On the basis of the simulated data for our zooplankton trapping approach, it is clearly possible to differentiate between the ‘foray’ and ‘RC’ foraging strategies (Fig. 2). In both cases, there was a large upward flux of copepods in the early evening, and a large downward flux in the early morning, simply due to the overall vertical migration behavior. However, for the FB, there are several pulses of upward and then downward exchanges of copepods between the high-food layer and the low-food layer, as can be seen in the upward- versus Caught up Comparison with field data 600 500 400 300 200 100 0 Foray behavior 0 Average gut content Field data were collected from Dabob Bay, WA (47 46.0770 N, 122 50.2320 W), 9 and 10 April, 2003. Zooplankton samples were collected with a vertically hauled Puget Sound net (1 m diameter, 209-mm mesh) deployed at 1–2 h intervals throughout the night from 9:00 pm to 1:00 pm the next day. During each interval, samples were collected from four depth strata: 100–50 m, 50–25 m, 25–10 m and 10 m-surface. Upon retrieval, contents of the cod ends were placed in 500-mL glass jars and preserved in 10% formalin seawater solution. Calanus pacificus females were enumerated under a dissecting microscope. On these dates, a relatively strong pycnocline was located between 10 and 12 m, with an average Chl concentration of about 16 mg L1 above the j (ng Chl copepod ) VOLUME Number caught per sample interval j –1 JOURNAL OF PLANKTON RESEARCH 14 12 10 8 6 4 2 0 2 4 6 8 10 12 0 Foray behavior 0 2 4 6 8 10 12 0 Time (h) Caught down Random behavior 2 4 6 8 10 12 Random behavior 2 4 6 8 10 12 Time (h) Fig. 2. Results of the standard foray behavior (FB) case versus the controls. Upper panels show number of copepods trapped per 45-min interval, going either up or down. Lower panels show the average and standard deviation of the gut contents of all individuals in the traps. 992 A. W. LEISING ETAL. j COPEPOD FEEDING FORAYS Individual probability of mortality downward-caught trap data over time. From these data, it is possible to conclude that copepods probably made at least 2 if not 3 forays into the upper layer during the time of the simulation. The actual average number of times that the copepods filled their guts within the surface layer (a measure of the number of forays conducted) was 2.48 (a value that could not be measured with traps but was tracked within the model). For the RC behavior, there was exchange between the two layers during the night, but it was of much lower magnitude than the FB, and there were no detectable coordinated pulses, as there were within the FB case. There was also a larger difference between the average gut contents of upward- versus downward-caught copepods for the FB than for the RC (Fig. 2, lower panels). In the field, if such samples were taken rather than determining the individual gut fluorescence for each copepod (as has essentially been done to produce the data for Fig. 2, lower panels), the average gut content would be determined from replicates of batches of copepods, e.g. five batches of 10 copepods each from within each net might be analysed for their average gut fluorescence and then the mean and variance of these five samples from each net could be compared for statistical differences. Simulating this more realistic procedure, t-tests were performed on the 10 pairs of samples taken during hours 3–9 of the simulation, comparing upward- and downwardcaught copepods from the FB runs, and separately comparing upward- and downward-caught copepods from the RC runs (alpha was set to 0.005 for each t-test, under the Bonferroni correction criteria; the total alpha for the ten pairs of samples = 0.05). Based on this analysis, mean copepod gut content of FB was significantly greater for downward- than upward-caught copepods for all 4.0% 10:1% 3.5% 3.0% sample times except for t = 5.25 h. For the RC, there was no significant difference between the mean gut content of upward versus downward copepods for five out of these 10 sample times. Thus, based on a procedure more similar to what might be conducted with real samples, the FB would result in samples where differences in gut contents between upward- and downward-caught copepods could be detected. Although it could not be estimated using trap samples similar to those that might be deployed in the field, the IBM model allowed tracking of the total ingestion and probability of mortality of each individual copepod throughout the simulation. For the standard run, mortality rates were set to 10% per night within the upper layer and 1% per night within the deeper layer. Thus, the total nightly probability of any individual copepod dying was a function of the time it spent within each layer, given the mortality probability within that layer. Under the standard setting, copepods with FB had a much lower individual probability of mortality than the RC copepods, for a given amount of food ingested (Fig. 3). The calculated fitness metric—food ingested divided by probability of mortality—was also significantly higher for the foray copepods than the RCs under the standard parameter set (two sample t-test, unequal variance; P < 0.001). As the mortality rate of the lower layer increased, the relative mortality difference between the foray and random behavior copepods decreased; the fitness metric was significantly greater for the foray copepods than the controls when the lower layer mortality was raised to 5% night1 (two sample t-test, unequal variance; P < 0.01) but not significantly different than the controls at 7.5% night1 (two sample t-test, unequal variance; P > 0.05). However, no matter 10:7.5% 10:5% 2.5% 2.0% 1.5% 1.0% 0.5% Random Foray 0.0% 0 100 200 300 400 0 100 200 300 400 0 100 200 300 400 Total ingestion (ng copepod–1) Fig. 3. Comparison of predicted individual morality probability versus total food ingested per copepod per night, for the foray behavior (FB) versus the random swimming control, versus three different levels of relative mortality pressure within the lower layer. Mortality rate within the upper layer was held constant at 10% per night. Mortality within the lower layer was set to 1% (left), 5% (middle) and 7.5% (right) per night. Upper layer food concentration was 20 the lower layer. 993 JOURNAL OF PLANKTON RESEARCH j VOLUME what the mortality rates were, each copepod from within the foray simulations always obtained at least a certain minimum nightly ration (in these cases, >100 ng copepod1), whereas there were a large number of individuals from within the random behavior simulation that had low relative nightly ingestion (Fig. 3). Sensitivity analysis Decreasing the maximum ingestion rate, Imax, led to a lower flux of individuals between the two surface layers (Fig. 4, top left panel), as indicated by the lower number of individuals caught within the virtual traps, as compared to the standard case (Fig. 4, top center panel). Increasing the value of Imax did not greatly increase the flux of individuals between the layers (Fig. 4, top right panel). Evidence of forays was detectable via the trap data regardless of the value of Imax, as shown by the relative timing of peaks for upward- versus downwardcaught copepods. There was little change in the average gut content of copepods caught in the traps, for either the FB or RC simulations, with a change in Imax (Fig. 4, middle panels), although gut contents of downward traveling copepods increased slightly for both as Imax Fig. 4. The effects of altering the maximum ingestion rate, Imax, versus number of copepods caught in the virtual traps (top panels), gut levels (middle panel) and ingestion versus mortality (bottom panels). Top panels show number caught per trap, going up or down, from the foray-based simulations only, with ingestion rate increasing (gut filling time decreasing) from left to right. Middle panel shows gut contents of all individuals averaged from all traps taken from t = 3 to t = 9 h, for both the foray-based behaviors and for the random swimming controls. Error bars show 1SD of all samples taken from t = 3 to t = 9 h. Bottom panels show predicted individual mortality probability versus total food ingested. 27 j NUMBER 10 j PAGES 987–1001 j 2005 increased. Regardless of the value of Imax used, there was always a larger difference between the downwardand upward-caught copepods for the FB case than for the RC; for both cases, gut contents of downwardtrapped copepods was significantly greater than upward trapped copepods, although the significance level was an order of magnitude greater for the foray case (Table II). The relative difference in individual probability of mortality for a given amount of food ingested for FB versus RC increased with an increase in Imax, while counter intuitively, the maximum amount of food ingested per night decreased as Imax increased (Fig. 4, bottom panels). This is due to the rules used to control the ‘refractory’ period between feeding bouts. At a high value of Imax, the gut is filled more quickly to Gmax, at which point the copepods stop feeding and enter the refractory phase (and sink if they are using the FB). Thus a copepod with a high value of Imax would spend more time within the refractory, non-feeding phase, sink out of the upper layer more often and subsequently achieve a lower total nightly ingestion. This may seem suboptimal, although these same copepods had a much lower mortality risk, than those that remained within the surface layer longer. Changing the gut clearance rate, R (which also has a direct effect on the time spent within the refractory, nonfeeding phase), had a large effect on the amplitude of the variability of copepods caught going up versus down over time (Fig. 5, top panels). Decreasing R led to larger differences between the gut contents of trapped copepods caught up versus down for FB, whereas increasing the gut clearance rate constant decreased the difference in gut contents of the copepods between the FB and RC (Fig. 5, middle panel), making gut contents alone a less useful criteria for detecting forays, although differences in gut contents were significant for all cases (Table II). Relative difference between foray and control copepod mortality for a given amount of food ingested, decreased with an increase in the gut clearance rate constant; foray copepods with a higher gut clearance rate spend more time within the surface layer (less time within the refractory phase where they sink into the lower layer) and thus have a mortality more similar to the random behavior controls. For the highest gut clearance rate constant used, there were a number of individuals for both the control and FB copepods that reached very high nightly ingestion levels (Fig. 5, bottom right panel). This was because the value of 2.25 h1 was the average gut clearance rate used, whereas, as described above, individuals were randomly given a range of clearance rates up to 25% greater than this average value. Given the standard value of Imax, and the maximum food concentration within the upper layer, copepods with gut clearance rates >2.7 h1 could never completely fill their guts 994 A. W. LEISING ETAL. j COPEPOD FEEDING FORAYS Table II: Results of the sensitivity analysis on average gut content of trapped copepods Parameter Change in % change average gut content Paired t-test P value, parameter Imax H R Sstep Sink Control up Control down Foray up 25% 24 12 17 +25% 19 6 28 50% 4 4 2 Lu : Ll Trap interval Foray down Control Foray 2 .004 .0005 21 .01 .0001 0.8 .008 .0003 .0005 +50% 6 3 5 3 .006 25% 39 34 45 46 .014 .00004 +25% 28 16 24 27 .0074 .0067 25% 1 12 14 .014 .0001 +25% 5 16 0.2 1.6 .002 .0002 25% 2.2 4.6 4.4 .005 .0035 +25% Cl up versus down 7.9 5 175 10 7 1.5 0.3 22 .003 .00007 104 184 22 .36 .58 226 22 .38 .19 154 89 +55% 1 5 +133% 4 2 3 25% 38 17 +25% 28 27 0.7 2.9 7.2 .003 .0001 3 .005 .0009 36 28 2.6 106 1.5 1011 20 24 .009 .007 For all t-tests, n = 9. Fig. 5. The effects of altering the gut clearance rate, R, on the trap catch (top), gut content (middle) and ingestion versus mortality (bottom). From left to right, the gut clearance rate constant increases from 1.5 h1 to 2 and to 2.5 h1, which changes the time between forays from 1 to 0.9 and to 0.8 h. to Gmax and therefore never left the upper layer. In all three cases, however, copepods using the FB had significantly higher values of the fitness metric than the controls (two-sample t-test, unequal variance; P < 0.01). Changing the hunger threshold, H, had little or no effect on the gut contents of copepods caught within the traps, or the mortality of the copepods for a given amount of food ingested (Fig. 6, middle and bottom panels), although it had a large effect on the flux of copepods between layers (Fig. 6, top panel). Increasing the threshold and thereby decreasing the time spent within the refractory, non-feeding (and sinking for the FB) period, resulted in an increase in the number of individuals swimming between the two layers. Conversely, decreasing the hunger threshold (meaning that copepods must digest more of their stomach contents before ‘reviving’) led to lower exchange between layers and therefore fewer forays per individual per night (Fig. 6, left top panel). At higher threshold levels, copepods spent less time in the refractory phase, and more time feeding, and thus nightly ingestion reached higher maximum values for some copepods (Fig. 6, bottom right panel). As for ingestion rate, in all three cases, copepods using the FB had significantly higher values of the fitness metric than the controls (two-sample t-test, unequal variance; P < 0.01). 995 JOURNAL OF PLANKTON RESEARCH j VOLUME Fig. 6. Effect of altering the hunger threshold, H, on the trap catch (top), gut content (middle) and ingestion versus mortality (bottom). Hunger threshold increases from left to right from 2.5 to 5 and to 7.5 ng per copepod. Table II summarizes the results of the remaining sensitivity analyses on the gut contents of the trapped copepods. Compared to the physiological parameters, swimming speed, Sstep, had less of an effect on the gut content of the trapped copepods. The difference in gut contents between upward- and downward-caught copepods was highly significant for the foray copepods (P < 0.005) regardless of whether swimming speed was increased or decreased. Sinking rate also had little effect on the model outcome for average gut content, with the greatest change for downward-caught FB copepods with a 25% reduction in sinking rate (a decrease of 22%). Even with this decrease in gut content, the difference in gut content between upward- and downwardcaught copepods was still highly significant (P < 0.005). Changing the vertical extent of the layers led to little or no change in the model results for either gut content (Table II) or numbers caught within the traps (data not shown). Increasing the concentration of food within the lower layer to 5 mg Chl L1, led to large changes in the gut contents of foray and control copepods (Table II), with the difference between upward- and downward-caught copepods being indistinguishable (P > 0.05), although forays were still detectable by examining the number of copepods caught swimming up or down versus time (data not shown). At 10 mg Chl L1, not only were gut contents no different between controls and FB but also 27 j NUMBER 10 j PAGES 987–1001 j 2005 were forays not detected from the vertical flux data. This is because many of the copepods filled their guts completely within the lower layer, and initiated sinking from within that layer, since this food concentration was well above the half-saturation constant of their feeding response (which was set to 5 mg Chl L1). As a result, there was no significant difference between upward- and downward-trapped copepods for either the foray or control cases under these higher lower-layer food levels (Table II). The last parameter examined was the time interval between trapping the copepods. As would be expected, numbers of copepods caught per trap increased as trap interval increased (data not shown), which would make detection of FB easier. Differences between the gut contents of upward- versus downward-caught FB copepods increased as trap interval decreased, (Table II), as seen by the highly significant P values for a decreased trap interval, although the difference was still significant for the longer trap interval. To summarize the effects of varying these different parameters on the copepods and their potential predators, as described above we also calculated an anomaly of copepod fitness measured in two different ways: the ratio of ingestion to mortality (relative fitness) and total nightly ingestion (feeding success; Fig. 7), along with the anomaly in copepod abundance within the surface (upper) and intermediate (lower) layers (Fig. 8). For the best-guess parameter set, copepod concentration within the surface layer was nearly 40% higher for the FB simulation versus the RC behavior simulation (Fig. 8). Relative fitness of the FB copepods was nearly 45% higher than RCs (Fig. 7). Growth potential (total amount eaten) was also higher for the FB, although the individual variability was very high (as denoted by the large error bars). The largest increases in copepod abundance within the upper surface layer occurred when any parameter was changed such that the time to fill the gut was increased, e.g. lowering Imax or raising the gut clearance rate. Under these parameter choices, copepod fitness did not increase relative to the controls, although copepod growth potential was on average higher but with higher individual variability. Conversely, highest copepod fitness was achieved when parameters were set to minimize the time taken to fill the gut, e.g. increasing Imax or decreasing the gut clearance rate. These cases also led to no increase in growth potential versus the controls. Comparison with field data We also ran our model in such a way as to simulate the conditions present on 9 April, 2003, from within Dabob Bay, WA, and compared it to field samples of copepod vertical distribution from that same time period (Fig. 9). 996 j COPEPOD FEEDING FORAYS 225% Ingestion/mortality 175% Total ingestion 125% 75% 25% –25% –75% te ra te th r H es ig ho Eq h t ld ua hr es lho si ld H zed ig la co h y nc low er Lo en er s w tra la sw tio ye n r Lo im sp w e si nk ed sp ee d Lo w ra ea cl t h gu gu H ig w Lo nc e ce an ig ea r H cl t St ra ax ax h w Lo d an da r Im Im ra y –125% fo Average individual fitness anomaly (relative to control) A. W. LEISING ETAL. Parameter set Fig. 7. Average relative fitness (ingestion/mortality) and feeding success (total ingestion) of all copepods from each simulation. Error bars show 1SD of the fitness metric for all individuals (n = 1000) from within each simulation. From left to right, categories represent (i) the standard ‘bestguess’ foray behavior (FB) case, (ii) FB with the value of Imax lowered by 25% from the standard case, (iii) FB with Imax raised by 25%, (iv) FB with gut clearance rate, R, decreased by 25%, (v) FB with R increased by 25%, (vi) FB with the hunger threshold, H, decreased by 50%, (vii) FB with H increased by 50%, (viii) FB with the vertical extent of the layers changed from upper = 0–15 m, lower = 15–50 m, to upper = 0–25 m, lower = 25–50 m, (ix) FB with the concentration of food in the lower layer raised from 1 m chlorophyll (Chl) L1, to 10 mg Chl L1, (x) FB with a lower swim speed of 1.875 cm s1 and (xi) FB with a lower sink speed of 0.75 cm s1. Upper layer 150% Lower layer 100% 50% 0% –50% Im Lo w gu t ig w Lo H fo rd St an da H ig ax h cl Im e h ar gu an ax t ce cl ea ra ra te nc Lo e ra w te th r H es ig ho Eq h t ld hr ua es lho si H ze ld ig d h co la l o nc w ye e e r Lo ntr r l s a a w t y sw ion er i m Lo w s si pe nk ed sp ee d –100% ra y Copepod abundance anomaly (relative to control) 200% Parameter set Fig. 8. Average abundance anomaly of foray copepods versus their paired random behavior control, within either the surface (upper) layer or the intermediate (lower) layer, from simulated vertical net tows taken every 45 min over a 6-h period (from t = 3 to t = 9 of the simulation). Categories (horizontal axis) as described for Fig. 7. Error bars show 1SD of the average abundance from the 9 ‘net tows’. To compare between profiles over time, for each profile (model or field) the data has been scaled by the maximum copepod concentration from within that profile. Whereas there is not a clear match at all time points between the model and field data, the general trends of the field data do not preclude the model results; albeit, 997 JOURNAL OF PLANKTON RESEARCH j VOLUME 27 j NUMBER 10 j PAGES 987–1001 j 2005 Fig. 9. Comparison of vertical distribution data taken from the model standard run (black bars, right hand side) versus field-collected data for Calanus pacificus females (grey bars, left hand side), taken on 9 and 10 April 2003 from Dabob Bay. Numbers across the top indicate the time of the tow in decimal hours. For comparison between the model and field data, the abundance values (originally as number of copepods m3) at each time point are scaled by the maximum abundance for that vertical series. Time points with no bars indicate no field data was taken during that time point. there may be other behaviors which could have led to such distributions. A random behavior, however, would not have led to the distribution shown, as random swimming results in a nearly exact equal abundance of copepods within the two surface layers during most of the night; this was not observed. Instead, the field data seems to be more consistent with the idea that the copepods could have undergone at least two forays during the early morning, one near 2:30 am (t = 2.5) and one near 4:45 am (t = 4.75). Our model also has the copepods leaving the surface waters too soon to begin the normal downward leg of their daily vertical migration. Unfortunately, the temporal resolution of the field sampling was not detailed enough to evaluate the success of the model, and it is unclear whether the copepods from the field may have undergone an initial migration into the surface waters upon their initial ascent from depth, as there are no samples from this time. DISCUSSION If FB is used by C. pacificus, then our model results indicate that in most cases this behavior confers a large advantage to feeding success and a considerable reduction in mortality, relative to copepods that just swim and feed at random throughout the upper water column. As might be expected, the mortality benefit decreases as the mortality rate within the lower layer approaches the mortality rate of the surface layer. However, no matter what the relative mortality rates are, if grazing rates are food limited within the lower layer, then the FB always confers a feeding advantage to the copepods, since it enables them to spend most of their feeding time within the high-food region, and most of their digestive/refractory time within the low-food region. Under the random behavior scenario, a few individuals had higher nightly total ingestion than any of the FB copepods. However, all of the FB copepods typically received a minimum daily ration much higher than a large proportion of the randomly behaving copepods. Thus random behavior may confer an advantage to a few individuals, whereas the FB ensures a certain level of success for all individuals, although these few random foragers also had the highest probability of mortality. Field evidence supports the existence of foray-like behaviors in certain cases. Simard et al. (Simard et al., 1985) found that C. finmarchicus within the St. Lawrence estuary, where Chl was mainly found only above 30 m within the fresh, warm surface layer, were rapidly filling their guts upon their initial migration into the surface and then quickly descending below this layer. The copepods would then ascend again before dawn to feed a second time. Simard et al. (Simard et al., 1985) point out that the copepods gained an additional benefit from ascending into the surface layer to feed, since these waters were significantly warmer than deeper layers, thus enabling a much more rapid filling of the gut. Durbin et al. (Durbin et al., 1995) also observed similar 998 A. W. LEISING ETAL. j COPEPOD FEEDING FORAYS behavior for C. finmarchicus within the Gulf of Maine. Tarling et al. (Tarling et al., 2002) observed that C. finmarchicus within the Clyde Sea abandoned the high-food surface layer fairly rapidly upon the arrival of high numbers of krill within this layer. However, Tarling et al. (Tarling et al., 2003) also concluded that such synchronized departure from the surface occurred regardless of the hunger state of the copepods, yet acknowledged that there still could be satiation responses leading to exchange of individual between the surface and deeper layers. Our modeling results suggest that even with relatively asynchronous initial upward migration and with a relatively high level of variability in individual physiology (approximately, our gut clearance rates varied by 25% between individuals), synchronous sinking and re-ascent can occur through most of the night. Within Dabob Bay, numerous studies have shown that concentrations of C. pacificus are typically higher below the surface, high-food layer, than within (Frost, 1988, references therein; Dagg et al., 1989, 1997, 1998; Bollens et al., 1993; Pierson et al., 2005b). Yet from measurements of gut contents (Dagg et al., 1989, 1997) and egg production (Pierson et al., 2005a), it is clear that individuals must be feeding within the surface layer, at least for some portion of the night. There is also evidence that predation on C. pacificus is high within this surface layer, and that there are seasonal correlations between the extent of vertical migration behavior and potential predator abundance (Bollens and Frost, 1989). Further, although the data are limited, the correspondence between our model results and the only set of field data amenable to comparison seems to rule out random foraging behavior, at the very least, and certainly does not preclude the existence of foray-like behavior. From another point of view, it is equally important to consider what effect the existence of FB might have on the feeding success of copepod predators. Figures 7 and 8 show a summary of the results of the sensitivity analyses in terms of both copepod success and changes in copepod abundance within the two upper layers. If a predator is limited only to the surface-most layer, then changes in the abundance of copepods within that layer caused by different copepod behaviors should affect the feeding of the predator when the prey is limiting. Under the standard case scenario of FB, the average nightly abundance of copepods within the upper surface layer increased by nearly 35% (Fig. 8) compared to copepods that swam randomly. As might be expected, the greatest increases in average night-time surface copepod abundance occurred whenever the physiological parameters led to longer gut-filling times (and thus higher surfacelayer residency times), such as when the maximum ingestion rate was lowered, or when the gut clearance rate was raised (Fig. 8). However, for these cases the fitness of the foray copepods was no different than for randomly behaving copepods. For both the standard foray parameter set, and the ‘high threshold’ sensitivity run, the abundance of copepods within the surface increased, and the fitness of the copepods increased. As counterintuitive as this may seem, it therefore appears that there are conditions that increase the success of both the predator and the prey. It is possible that such an increase in surface abundance of copepods may not always increase the success of the predators. For instance, the FB may lead to somewhat bimodal surface copepod densities over time, due to the semi-synchronous nature of the forays set up by the initial upward migration after sunset. If these surface copepod density vary between very low and very high values, then the predators may go from a situation of severe prey limitation, to a situation where their ingestion response is saturated. In this case, although the nightly average surface abundance of copepods might be higher when the copepods use FB, the end result may actually be a poorer feeding environment for the predators. Under what conditions were forays detectable by our trapping scheme? This is a critical question, for although our model results suggest advantages of the FB, its existence in natural settings may remain unknown unless it is possible to detect it via practical sampling schemes. Here, we proposed a scheme that could be carried out using vertically towed, opening and closing nets to sample distributions, along with upward and downward facing traps (the technical details of which seem surmountable), which could collect upward and downward swimming copepods over time. These samples could then be assayed for gut fluorescence to detect differences in feeding state. Other markers besides gut Chl concentrations could be used equally as well [such as high performance liquid chromatography (HPLC) or highmagnification digital imagery of gut contents, or more advanced assays; see Pearre, 2003] or might complement the gut Chl measurements. Given these conditions, and under our ‘best-guess’ scenario for the parameters controlling the swimming and feeding of C. pacificus, it seems probable that our simulated field-sampling scheme would be able to detect the use of FBs by the copepods. However, there were also cases where our simulated sampling scheme would not be able to detect forays, or rather, certain elements of the sampling would reveal little about the copepods behavior. First, copepod gut fluorescence alone became less useful in detecting forays when the gut clearance rate was increased (Fig. 5). Gut clearance rate may vary with temperature (Dam and Peterson, 1988), feeding history (Kiørboe and Tiselius, 1987), and possibly food concentration (Mullin et al., 999 JOURNAL OF PLANKTON RESEARCH j VOLUME 1975; Pasternak, 1994), and could therefore increase to levels making gut fluorescence a poor indicator of FB. Potentially, measurements of gut clearance rates under ambient conditions might be necessary to augment the field sampling described above. Decreased sinking or rather downward displacement speeds also decreased the utility of gut fluorescence as a marker of FB. In this case, there may be other markers, such as lipid content, or Chl to phaeopigment ratios, which might be better markers of recent feeding history. Even if physiological markers fail to differentiate FB, directional trap sampling may still be used in most cases to infer the use of FB. Of course, trapping may be susceptible to problems of high variability typical of such net sampling, or worse due to the nature of the proposed upward- and downward-facing traps. Our analysis showed that closer spaced temporal sampling could increase the ability to resolve the peaks and troughs of the semi-synchronous forays, yet decreasing the sample interval decreases the number of copepods caught per trap, and may also push the limits inherent in physically deploying such traps. Although not shown here, the addition of both multiple upward– downward facing traps at the same depth, yet at a different location, along with more traps spaced at different depths, would increase the ability of such a system to detect FB, and reduce interpretive problems due to sample variability. Although there may be other, possibly more complex behaviors which could lead to the distributions and observations supporting the hunger/satiation foraging hypothesis, our modeling results suggest that a relatively simple behavior like that proposed here could lead to such observations. Whether the behavior is as simple as we suggest, or more complicated, the implications of such behaviors are significant for both the population dynamics of the copepods and the feeding environment for their predators (Pearre, 2003). Specifically, our results suggest not only that a foray-like behavior could greatly enhance the feeding success of the copepods while reducing their individual probability of mortality, but that there could be periods of increased copepod abundance within particular strata which would in turn enhance the prey field for planktivores. Further, our proposed sampling scheme seems technically feasible and should be able to detect the existence of foray-like behaviors under most cases. In conclusion, given the broad global distribution and trophic importance of Calanus species, both as grazers on phytoplankton and food for commercially important fish species, conducting further field and laboratory studies to determine the nature and existence of such behaviors is warranted. 27 j NUMBER 10 j PAGES 987–1001 j 2005 ACKNOWLEDGEMENTS A.W.L. thanks the staff at the Pacific Fisheries Environmental Laboratory for many comments and helpful discussions during the course of the development of this model. This article is dedicated to G.A. Paffenhöfer, whose work on copepods has pioneered aspects of copepod feeding behavior on multiple scales and stimulated many of the questions leading up to this study. The comments from two anonymous reviewers significantly helped this work. As always, the crew of the C.A. Barnes were invaluable in obtaining the copepod distribution data. J. Pierson and S. Cary were supported by the NSF through contract OCE-0118044 to B. Frost. REFERENCES Alcaraz, M., Paffenhöfer, G.-A. and Strickler, J. R. (1980) Catching the algae: a first account of visual observations on filter-feeding calanoids. In W. C. Kerfoot (ed.), Evolution and Ecology of Zooplankton Communities. University Press of New England, London, pp. 241–248. Ban, S., Burns, C., Castel, J. et al. (1997) The paradox of diatomcopepod interactions. Mar. Ecol. Prog. Ser., 157, 287–293. Bjørnsen, P. K. and Nielsen, T. G. (1991) Decimeter scale heterogeneity in the plankton during a pycnocline bloom of Gyrodinium aureolum. Mar. Ecol. Prog. Ser., 73, 263–267. Bollens, S. M. and Frost, B. W. (1989) Zooplanktivorous fish and variable diel vertical migration in the marine planktonic copepod Calanus pacificus. Limnol. Oceanogr., 34, 1072–1083. Bollens, S. M., Osgood, K. E., Frost, B. W. et al. (1993) Vertical distributions and susceptibilities to vertebrate predation of the marine copepods Metridia lucens and Calanus pacificus. Limnol. Oceanogr., 38, 1827–1837. Castro, L. R., Bernal, P. A. and Gonzalez, H. E. (1991) Vertical distribution of copepods and the utilization of the cholrophyll a-rich layer within Concepcion Bay, Chile. Est. Coast. Shelf Sci., 32, 243–256. Dagg, M. J., Frost, B. W. and Newton, J. (1997) Vertical migration and feeding behavior of Calanus pacificus females during a phytoplankton bloom in Dabob Bay, USA. Limnol. Oceanogr., 42, 974–980. Dagg, M. J., Frost, B. W. and Newton, J. (1998) Diel vertical migration and feeding in adult female Calanus pacificus, Metridia lucens and Pseudocalanus newmani during a spring bloom in Dabob Bay, a fjord in Washington USA. J. Mar. Syst., 15, 503–509. Dagg, M. J., Frost, B. W. and Walser, W. E. J. (1989) Copepod diel migration, feeding and the vertical flux of phaeopigments. Limnol. Oceanogr., 34, 1062–1071. Dam, H. G. and Peterson, W. T. (1988) The effect of temperature on the gut clearance rate constant of planktonic copepods. J. Exp. Mar. Biol. Ecol., 123, 1–14. Durbin, E. G., Campbell, R. G., Gilman, S. L. et al. (1995) Diel feeding behavior and ingestion rate in the copepod Calanus finmarchicus in the southern Gulf of Maine during late spring. Cont. Shelf Res., 15, 539–570. Frost, B. W. (1972) Effects of size and concentration of food particles on the feeding behavior of the marine planktonic copepod Calanus pacificus. Limnol. Oceanogr., 6, 805–815. Frost, B. W. (1985) Food limitation of the planktonic marine copepods Calanus pacificus and Pseudocalanus sp. in a temperate fjord. Arch. Hydrobiol. Beih. Ergebn. Limnol., 21, 1–13. 1000 A. W. LEISING ETAL. j COPEPOD FEEDING FORAYS Frost, B. W. (1988) Variability and possible adaptive significance of diel vertical migration in Calanus pacificus, a planktonic marine copepod. Bull. Mar. Sci., 43, 675–694. Osgood, K. E. and B. W. Frost (1996) Effects of advection on the seasonal abundance patterns of three species of planktonic calanoid copepods in Dabob Bay, Washington. Cont. Shelf Res., 16, 1225–1243. Gauld, D. T. (1953) Diurnal variations in the grazing of planktonic copepods. J. Mar. Biol. Ass. U. K., 31, 461–474. Osgood, K. E. and Frost, B. W. (1994) Ontogenetic diel vertical migration behaviors of the marine planktonic copepods Calanus pacificus and Metridia lucens. Mar. Ecol. Prog. Ser., 104, 13–25. Hardy, A. C. and Bainbridge, R. (1954) Experimental observations on the vertical migrations of plankton animals. J. Mar. Biol. Ass. U. K., 33, 409–448. Harris, R. P. (1988) Interactions between diel vertical migratory behavior of marine zooplankton and the subsurface chlorophyll maximum. Bull. Mar. Sci., 43, 663–674. Hassett, R. P. and Blades-Eckelbarger, P. (1995) Diel changes in gutcell morphology and digestive activity of the marine copepod Acartia tonsa. Mar. Biol., 124, 59–69. Head, E. J. H., Wang, R. and Conover, R. J. (1984) Comparison of diurnal feeding rhythms in Temora longicornis and Centropages hamatus with digestive enzyme activity. J. Plankton Res., 6, 543–551. Ianora, A., Miralto, A., Poulet, S. A. et al. (2004) Aldehyde suppression of copepod recruitment in blooms of a ubiquitous planktonic diatom. Nature, 429, 403–407. Kiørboe, T. and Tiselius, P. T. (1987) Gut clearance and pigment destruction in a herbivorous copepod, Acartia tonsa, and the determination of in situ grazing rates. J.Plankton Res., 9, 525–534. Paffenhöfer, G.-A., Strickler, J. R. and Alcaraz, M. (1982) Suspensionfeeding by herbivorous calanoid copepods: a cinematographic study. Mar. Biol., 67, 193–199. Pasternak, A. F. (1994) Gut fluorescence in herbivorous copepods: an attempt to justify the method. Hydrobiologia, 292/293, 241–248. Pearre, S. J. (1973) Vertical migration and feeding in Sagitta elegans verrill. Ecology, 54, 300–314. Pearre, S. J. (1979) Problems of detection and interpretation of vertical migration. J.Plankton Res., 1, 29–44. Pearre, S. J. (2003) Eat and run? The hunger/satiation hypothesis in vertical migration: history, evidence and consequences. Biol. Rev., 78, 1–79. Peterson, W. (1998) Life cycle strategies of copepods in coastal upwelling zones. J. Mar. Sys., 15, 313–326. Pierson, J. J., Halsband-Lenk, C. and Leising, A. W. (2005a) Reproductive success of Calanus pacificus during diatom blooms in Dabob Bay, WA. Progress in Oceanography. In press. Kleppel, G. S., Peiper, R. E. and Trager, G. (1988) Variability in the gut contents of individual Acartia tonsa from waters off Southern California. Mar. Biol., 97, 185–190. Pierson, J. J., Leising, A. W., Halsband-Lenk, C. et al. (2005b) Vertical distribution and abundance of Calanus pacificus and Pseudocalanus newmani in relation to chlorophyll a concentrations in Dabob Bay, WA. Progress in Oceanography. In press. Lampert, W. (1993) Ultimate causes of diel vertical migration of zooplankton: new evidence for the predator avoidance hypothesis. Arch. Hydrobiol. Beih. Ergebn. Limnol., 39, 79–88. Peterson, W. T., Miller, C. B. and Hutchinson, A. (1979) Zonation and maintenance of copepod populations in the Oregon upwelling zone. Deep-Sea Res., 26A, 467–494. Landry, M. R. and Fagerness, V. L. (1988) Behavioral and morphological influences on predatory interactions among marine copepods. Bull. Mar. Sci., 43, 509–529. Rodriguez, V. and Durbin, E. G. (1992) Evaluation of synchrony of feeding behavior in individual Acartia hudsonica (Copepoa, Calanoida). Mar. Ecol. Prog. Ser., 87, 7–13. Mackas, D. and Bohrer, R. (1976) Fluorescence analysis of zooplankton gut contents and an investigation of diel feeding patterns. J. Exp. Mar. Biol. Ecol., 25, 77–85. Sameoto, D. D. (1984) Environmental factors influencing diurnal distribution of zooplankton and icthyoplankton. J. Plankton Res., 6, 767–792. Mackas, D. L. and Burns, K. E. (1986) Poststarvation feeding and swimming activity in Calanus pacificus and Metridia pacifica. Limnol. Oceanogr., 31, 383–392. Schabetsberger, R., Brodeur, R. D., Cianelli, L. et al. (2000) Diel vertical migration and interaction of zooplankton and juvenile walleye pollock (Theragra chalcogramma) at a frontal region near the Pribilof Islands, Bering Sea. ICES J. Mar. Sci., 57, 1283–1295. Macy, W. K., Sutherland, S. J. and Durbin, E. G. (1998) Effects of zooplankton size and concentration and light intensity on the feeding behavior of Atlantic mackerel Scomber scombrus. Mar. Ecol. Prog. Ser., 172, 89–100. Simard, Y., Lacroix, G. and Legendre, L. (1985) In situ twilight grazing rhythm during diel vertical migrations of a scattering layer of Calanus finmarchicus. Limnol. Oceanogr., 30, 598–606. Miralto, A., Barone, G., Romano, G. et al. (1999) The insidious effects of diatoms on copepod reproduction. Nature, 402, 173–176. Mullin, M. M. and Brooks, E. R. (1976) Some consequences of distributional heterogeneity of phytoplankton and zooplankton. Limnol. Oceanogr., 21, 784–796. Mullin, M. M., Stewart, E. F. and Fuglister, F. J. (1975) Ingestion by planktonic grazers as a function of concentration of food. Limnol. Oceanogr., 20, 259–262. Napp, J. M., Brooks, E. R., Matrai, P. et al. (1988) Vertical distribution of marine particles and grazers. II. Relation of grazer distribution to food quality and quantity. Mar. Ecol. Prog. Ser., 50, 59–72. Sørnes, T. A. and Aksnes, D. L. (2004) Predation efficiency in visual and tactile zooplanktivores. Limnol. Oceanogr., 49, 69–75. Tarling, G. A., Jarvis, T., Emsley, S. M. et al. (2002) Midnight sinking behavior in Calanus finmarchicus: a response to satiation or krill predation? Mar. Ecol. Prog. Ser., 240, 183–194. Tarling, G. A., Jarvis, T. and Matthews, J. B. L. (2003) Calanus finmarchicus descends in response to the arrival of krill – better unfed than dead. Mar. Ecol. Prog. Ser., 252, 307–310. Tiselius, P. and Jonsson, P. R. (1990) Foraging behavior of six calanoid copepods: observations and hydrodynamic analysis. Mar. Ecol. Prog. Ser., 66, 23–33. Neilson, J. D. and Perry, R. I. (1990) Diel vertical migration of marine fishes: an obligate or facultative process? Adv. Mar. Biol., 26, 115–168. Townsend, D. W., Cucci, T. L. and Berman, T. (1984) Subsurface chlorophyll maxima and vertical distribution of zooplankton in the Gulf of Maine. J. Plankton Res., 6, 793–802. Ohman, M. D. (1990) The demographic benefits of diel vertical migration by zooplankton. Ecol. Monogr., 60, 257–281. Zaret, T. M. and Suffern, J. S. (1976) Vertical migration in zooplankton as a predator avoidance mechanism. Limnol. Oceanogr., 21, 804–813. 1001
© Copyright 2026 Paperzz