Introduction

Negative effects of harmful phytoplankton on grazers, ranging from feeding inhibition and physiological incapacitation to reduction in egg production and egg hatching rates, are common (see Turner et al. 2001 and references therein). Traditionally, diatoms have been considered as a basis for a short and effective food chain leading to a high fish production (Steele 1974); a view which is confirmed in a number of more recent field studies (Irigoien et al. 2002 and references therein). During the last 15 years, however, this view has been challenged by studies showing reduced copepod hatching success and/or egg production on diatom-dominated diets (e.g., Miralto et al. 1999; Dutz et al. 2008). Several mechanisms have been suggested to cause these negative effects. Firstly, inhibitory chemical compounds, identified as polyunsaturated aldehydes (PUA), have been shown to block embryogenesis in copepods and sea urchins and have antiproliferative and apoptotic effects on human carcinoma cells (Miralto et al. 1999). These compounds have been detected in many diatoms (Miralto et al. 1999; Wichard et al. 2007), including several strains of the Skeletonema marinoi (recently separated from Skeletonema costatum, Sarno et al. 2005), which is a common spring bloom diatom in the Baltic Sea, particularly in eutrophic areas (Ask et al. 2006). Secondly, nutritional inadequacy of diatoms, such as fatty acid or sterol, have also been observed to induce low production rate and hatching success of copepod eggs (Jónasdóttir 1994; Jónasdóttir et al. 2009). Recent findings have added possible explanations for the negative effects of diatoms, such as a rapid depletion of polyunsaturated fatty acids (especially eicosapentaenois acid EPA) caused by the PUA production upon mechanical wounding (Wichard et al. 2007), a compound inducing oxidative stress (Fontana et al. 2007a, b), and a low assimilation efficiency of essential compound with a shorter residence time in copepod guts (Dutz et al. 2008). Still, the exact mechanisms remain unknown with numerous inconsistencies in the results from different investigations.

Typically, the effect of diatoms on the feeding or reproduction of copepods can vary greatly among species and is situation specific, due to differences in the species and genetic composition of the bloom (see Turner and Tester 1989; Rynearson et al. 2006) or differences in nutritional requirements of the copepod (Hasset 2004). Variation in inhibitory effects of diatoms does also occur at the intraspecific strain level (Ask et al. 2006; Koski et al. 2008; Dutz et al. 2008). For instance, different strains of Thalassiosira rotula have been observed to have both positive and negative effects on copepods development (Koski et al. 2008) and different strains of Skeletonema costatum induced variable hatching success of Eurytemora affinis (Ask et al. 2006). Further, S. marinoi in culture has also been observed to release pronounced bursts of PUAs (heptadienal and octadienal) that are strongly dependent on the growth phase (Vidoudez and Pohnert 2008). Recently, Taylor et al. (2009) isolated nine different strains of S. marinoi in the Gullmar Fjord, Skagerrak and reported large differences in the PUA production between seasons. The clones isolated showed different physiological responses and degree of genetic heterogeneity indicating that different populations succeed each other in the fjord (Saravanan and Godhe 2010). These observations indicate that there can be differences in the toxic and biochemical properties of diatoms depending on strains, genetic variability, growth stage of the algae, and season.

In nature, phytoplankton cells are patchily distributed (Daro 1988) over a distance of a few centimeters to meters (Leising and Franks 2000). To maximize exposure to limiting resources, copepods need to be able to find and remain within the patches of high food concentration (Dagg 1977; Tiselius 1992). There is increasing evidence that copepods have some ability to response quickly and locate the high resource patch (Buskey 1984; Tiselius and Jonsson 1990; Tiselius 1992). Some copepod species are believed to decrease their swimming speed or jump frequency when encountering a patch of high food concentration (Acarti tonsa, Tiselius 1992; Temora longicornis, and Pseudocalanus elongatus, Tiselius and Jonsson 1990) suggesting that they may be using some type of “area-restricted search” behavior (Leising 2002). Studies on behavioral adaptation of zooplankton aggregation and swimming when the food patches consist of a potentially toxic or harmful algae are rare. However, by analyzing the behavioral response of copepods to a high concentration of toxic Karenia brevis, Cohen et al. (2007) showed that Temora turbinata actively avoided dense patches of this species. Pierson et al. (2005) found a similar pattern with Calanus pacificus females, which strongly avoided the layer with a high diatom concentration, which was suggested to result in a high reproductive success.

Motivation of this study was to investigate the strain- and species-specific effect of S. marinoi on copepod physiology and behavior, in relation to both nutritional quality and toxicity of the algae. Specifically, we investigated the strain-specific effect of S. marinoi on feeding, egg production, and hatching of three common planktonic calanoid copepods, A. tonsa, P. elongatus, and T. longicornis. We analyzed copepod behavior in order to see whether copepods alter their distribution and swimming behavior as a response to different biochemical properties and PUA content of the alga. Our results bring new detailed information on copepod–diatom interactions in the level of species and strains, as well as provide one of the few existing data sets where both algae nutritional and toxic compounds, as well as copepods ingestion of these, are measured.

Materials and methods

Algae and copepods

We used three copepod species and three Skeletonema marinoi strains, as well as Rhodomonas sp., which served as a good quality control food. All algae were offered to the copepods as a single food source, at a concentration of ca. 200 μg C l−1. All three copepod species, Acartia tonsa, Temora longicornis, and Pseudocalanus elongatus, were tested with a chain forming S. marinoi strain GF04-9B (9B; 264 μm3 cell volume) and Rhodomonas sp. (RHO) in all experiments. A. tonsa was additionally tested with two other S. marinoi strains, GF04-1G (1G; 622 μm3) and GF04-7J (7J; 141 μm3; Table 1).

Table 1 Strain number, volume (μm3), carbon and nitrogen content (pg cell−1), C/N ratio (weight), and content of 20:5(n-3) fatty acid (EPA), 22:6 (n-3) fatty acid (DHA), total (tFA) and polyunsaturated fatty acid (16–22 PUFA) (fg μm−3), sterol ∆5 (fg μm−3), cholesterol (fg μm−3) and total PUA (fmol cell−1) in Rhodomonas sp., and the three Skeletonema marinoi strains
Full size table

All strains originated from the culture collection of the University of Gothenburg, Sweden and were isolated at different times of the year 2004 (1G; February, 7J; June and 9B, September; Saravanan and Godhe 2010). All strains have been verified to belong to the same species although the genetic heterogeneity was high and physiological responses different (e.g., growth rate, biovolumes) depending on season, temperature, and salinity. All strains were also observed to produce different ranges of PUA after several months of cultivation under identical conditions, with the highest PUA production in 1G (ca. 0.7 fmol cell−1), second highest in 9B (ca. 0.3 fmol cell−1), and lowest in 7J (ca. 0.2 fmol cell−1, Taylor et al. 2009) suggesting that the differences are not due to the experimental set of conditions. At the time of the present study, the PUA production was, however, changed (Table 1). It is thus unknown to which degree the strains reflected their original properties or genetic composition after several years of cultivation, but the differences observed in PUA production, biochemical composition, and morphology still suggested a high variability in a number of important properties.

Copepods were cultured in the laboratory at 15°C, in a salinity of ~35‰ in the dark, with a mixture of food consisting of Rhodomonas sp., Thalassiosira weissflogii, and Heterocapsa sp., offered at a saturating concentration of >400 μg C l−1 (Koski et al. 2006). When adults occurred, both females and males were sorted out, and ca. 60 actively swimming individuals (depending on species) were isolated under dissecting microscope. The individual carbon content of copepods was estimated from the measured length and the length–dry weight relationship by Klein Breteler et al. (1982) for T. longicornis and P. elongatus and Kiørboe et al. (1985) for A. tonsa. The estimated individual carbon content was 3.30 μg C for A. tonsa, 5.67 μg C for T. longicornis, and 8.15 μg C for P. elongatus.

All algae were grown under identical conditions in batches of 1–2 l with F/2 + Si medium at 18°C with a light/dark cycle of 14:10 h. Algae were kept in exponential growth phase by diluting the culture every ca. 3 days. The algae volume was determined either by using a Coulter counter (RHO and 1G) or an inverted microscope (9B and 7J), by measuring ca. 30 cells (Zeiss Axiovert S100, 400× magnification) and using an appropriate geometric formula for the volume determination. As 1G strain always consisted of single cells (Table 1), it was appropriate to use an electronic particle counter to determine its abundance. During the experiments, mean carbon content per cell was estimated from the volume, assuming an average carbon content of 0.077 pg μm−3 as measured earlier for Skeletonema costatum (Dutz et al. 2008; Koski et al. 2008). Later, ca. 20 ml of the algae cultures were filtered on duplicate precombusted (6-h at 450°C) Whatman GF/F filters for the analysis of particulate organic carbon and nitrogen with a Carlo Erba elemental analyzer, using standard procedures and a combustion temperature of 1,030°C (Ask et al. 2006).

Similarly, a defined volume (80–100 mL) of each algae culture was filtered for the quantification of PUA and lipid. PUA were analyzed according to Vidoudez et al. (2011). In short, cells were transferred in a buffer containing 25 mM O-(2,3,4,5,6-pentafluorobenzyl) hydroxylamine hydrochloride (PFBHA, derivatization grade >99%, Sigma–Aldrich) prior disruption by sonication to trigger the production of the PUA. The derivatized PUA were then extracted with hexane and analyzed by using a gas chromatography–mass spectrometer (GC–MS). Fatty acids and sterols were analyzed according (Jónasdóttir et al. 2011). In short, cell fatty acids were transesterified in methanol/acetyl chloride 9:1 (>99.0%, Sigma–Aldrich) before extraction with hexane. The same samples were then extracted twice with chloroform to recover the sterols. Chloroform and hexane phases were combined and, after silylation with N-nethyl-N-(trimethylsilyl) trifluoroacetamide (Macherey–Nagel, Düren, Germany), were analyzed by GC–MS. Sterols are presented as the total concentration of three identified ∆5 sterols, cholesterol, brassicasterol, and campesterol. Total concentration of polyunsaturated fatty acids (PUFA) represents the sum of the identified 16–22 fatty acids. The mineral and biochemical properties of alga were measured once in the course of the 8-day experiments.

Grazing and egg production

The experiments were carried out for 8 or 10 days, with daily measurements of egg production and hatching, and two separate 24-h experiments to determine pellet production and ingestion rates (days 3–4 and 7–9). For A. tonsa feeding on 7J, the incubation was ended at day 5 due to high mortality rate (see result). Experiments were conducted with 4–5 experimental replicates, and 2–3 controls without copepods were added for grazing experiments. Food suspension was renewed daily. Depending on the species, 8–15 copepods, including 1–2 males, were incubated in 625 ml bottles containing the appropriate food suspension. The bottles were carefully closed to prevent any air bubbles and kept in suspension by rotating at 1 rpm at 15°C.

Every 24-h (A. tonsa and T. longicornis) or 48-h (P. elongatus) interval, contents of the bottles were gently poured onto a 180 μm (copepods) and 50 μm (eggs) mesh sieve, the number of copepods and eggs were counted, the condition of copepods checked, and dead copepods were removed. The difference in incubation time resulted from the absence of egg sacs for all P. elongatus female at the beginning of the experiment and production of a new egg sac was assumed to take >24 h (Koski et al. 2006). However, to be sure of not including any bias due to incubation time, we also calculated the egg production of P. elongatus based on the egg ratio method (Runge and Roff 2000), which gave similar rates. Empty egg shells were counted and included in the calculations of egg production to account for cannibalized eggs. However, they rarely contributed for more than 2% of the total egg number. The produced eggs were flushed into 330 ml bottles in filtered sea water, and the hatching success was determined after another 3 days of incubation, by counting the remaining eggs and nauplii. At the end of the experiment, copepods were preserved for length measurements, which were later conducted using a stereo microscope (Leica MZ6, 20× magnifications).

For grazing experiments, cell concentrations at the start and end of the incubations were counted from subsamples preserved with Lugol’s iodine solution, using an inverted microscope (>400 cell sample−1; 7J and 9B) or a Coulter counter (Multisizer™ 3; RHO and Skeletonema marinoi 1G). From our observation, 1G strain always consisted of single cells with a few 2 cell chains. To ensure that the chain length did not change due to the presence of copepods, additional control experiment were performed to compare the microscopic cell counts with the Coulter counter counts, in the absence and presence of copepods. The cell numbers counted with both methods were comparable, and the differences were not influenced by the number of copepods present in the incubations (data not shown). Ingestion and clearance rates were estimated using the method of Frost (1972). For pellet volume, ca. 30 pellets per treatment were measured using a stereo microscope (Leica MZ6, 40× magnifications), assuming a cylinder shape.

Carbon-specific egg production was calculated by multiplying egg production by the egg carbon mass (μg C) and dividing by the female body mass (μg C). Carbon content of T. longicornis and P. elongatus eggs were estimated to be ca. 0.09 and 0.19 μg C egg−1, respectively (Koski et al. 2006), while A. tonsa eggs were estimated to be 0.046 μg C egg−1 (Kiørboe et al. 1985). The efficiency of egg production was calculated from the ratio of egg produced to the ingested food carbon.

When Temora longicornis was feeding on RHO, we observed a clear discrepancy with previous results (Dutz et al. 2008), with negligible egg and pellet production and hatching success (see results). As the simultaneously measured maximum egg production with the 9B strain of Skeletonema (days 7–8) was relatively high, we interpreted this low reproduction to be due to some unexplained problem with the RHO culture at the time of this incubation, and not related to the condition of T. longicornis (such as age, past-feeding history, or fertilization limitation). When a comparison with RHO was necessary (e.g., egg production, hatching, and egg production efficiency), we used previous measurements done in the same culture of T. longicornis and RHO, under similar conditions (Dutz et al. 2008). The egg production and hatching rates of T. longicornis on 9B should, however, be considered as conservative minimum rates.

Behavioral changes

Copepods behavioral response to different Skeletonema marinoi strains was studied as described in Tiselius (1992), by creating a patchy food distribution. In these experiments, we compared behavioral responses (percentage distribution, swimming speed, and turning angle in the food patch) of all three copepod species with food (9B) and without food (filtered seawater; FSW), with A. tonsa additionally tested with 1G and 7J. We used both fed and 24-h starved individuals to account for the possible effects of a hunger response. The patchy food distribution was created by using a salinity gradient, where the upper (32‰) and bottom (36‰) part of the aquarium contained filtered seawater, and food at a concentration of ca. 200 μg C l−1 was added in the middle (34‰) part of the 0.5 l aquarium (20 × 5 × 5 cm). The patch in the middle of the aquarium was made by gently siphoning through a modified tube positioned at the bottom of the aquarium by starting with the lightest water (Fig. 1). To the food patch was added a few drops of food color, to recognize the food layer. To check for the effect of salinity gradient and food dye, control aquariums were created similarly, but excluding the algae from the middle layer. As all copepods were coastal species and have a wide salinity tolerance (see e.g., Holste et al. 2009; Renz and Hirche 2006), the steep gradient used in this study was considered acceptable.

Fig. 1
figure 1

Schematic figure of aquarium used in the behavior experiment

Full size image

Prior to the experiment, adult female copepods were kept in 2 l containers with RHO at a concentration of ca. 300 μg C l−1 (fed experiments) or without food (starved experiments) at 15°C for 24 h. At the start of the experiments, 20 copepods of each species were carefully introduced to the top of the aquarium with a pipette and filmed as they sank to the food layer. Observations were made over a 2-h period approximately the same time of the day, between 14:00 and 18:00. Both the experiment (middle layer with food) and control (middle layer without food) were run during the same day.

For filming, we used a 2D setup with Panasonic CCD-F10 camera equipped with a 35 mm lens, connected to a synchronizer, a mixer, a time–date generator (Panasonic WJ180), and a video cassette recorder, VCR (Panasonic NV-FS200 HQ). The aquarium was illuminated from the back by colliminated light, the only light source for filming. Two-dimensional projection of swimming tracks were digitized using LABTRACK (Bioras, Kvistgard, Denmark) (Kiørboe 2007). The program allows setting of thresholds for size, minimum track length, minimum velocity, and contrast, ensuring that only the target particles are tracked (Titelman and Kiørboe 2003). Between 7 and 25 swim tracks, which varied between 1 s to 6 min in duration, were observed for each individual. The time step was always 0.04 s. Swimming speed was measured by counting from position differences between consecutive frames as described in Titelman and Kiørboe (2003) and turning angle as the differences over time steps of three frames.

Percentage of copepods distribution in the food region was determined every 30 min by tracing the movements on the monitor screen. Further details on filming technique and analysis are given in Titelman and Kiørboe (2003).

Statistics

To test for differences between food sources for each copepod species, a one-way analysis of variance (ANOVA) followed by Tukey honestly significant different (HSD) was used. If the assumption for the ANOVA were not met (normality and equality of variances), we used a nonparametric Kruskal–Wallis (KW) test. For P. elongatus and T. longicornis, where only two food types were tested, we used a simple t test. We used the mean values of 2 days (day 3–4 and 7–9) for clearance, ingestion, daily ration and pellet production, and the average over 8 or 10 days for egg production. In the behavioral studies (vertical distribution, swimming speed, and turning angle), we used a 2-way ANOVA to test for differences between fed and starved copepods and diet. Spearman rank correlation analysis was used to evaluate any correlations between copepod responses and food nutritional composition or diatom aldehydes. Only the days, when grazing experiments were conducted, were included in the analysis.

Results

Ingestion and clearance rate

The average clearance rate ranged from about 11 (1G) to 17 (RHO) ml ind−1 d−1 in Acartia tonsa, 16 (9B) to 19 (RHO) ml ind−1 d−1 in Pseudocalanus elongatus, and 13 (RHO) to 32 (9B) ml ind−1 d−1 in Temora longicornis (Fig. 2). A. tonsa ingested RHO, 9B, and 1G at a rate ranging from 1.7 (1G) to 3.1 (9B) μg C ind−1 d−1 with significantly higher ingestion of 9B than other food items (1-way ANOVA, F3 = 9, p < 0.001, Tukey, p < 0.05), while neither clearance nor ingestion of 7J were significantly different from zero (t test, p > 0.05). For P. elongatus and T. longicornis, average ingestion rate was about 3 μg C ind−1 d−1 (for both algae), and 2 (RHO) or 6 (9B) μg C ind−1 d−1, respectively. No significant differences were observed in clearance and ingestion rate of P. elongatus between diets, but 9B was cleared and ingested more by T. longicornis than RHO diet (t test, p < 0.05).

Fig. 2
figure 2

Copepod clearance (ml ind −1 d−1; columns) and ingestion (μg C ind−1 d−1; symbols) rates on different diets (mean ± SD). (RHO) Rhodomonas sp., (9B), Skeletonema marinoi GF04-9B (7J), Skeletonema marinoi GF04-7J, and (1G) Skeletonema marinoi GF04-1G. Different letters denote ingestion rates that are significantly different from each other (Tukey HSD, p < 0.05). Ns not significantly different from zero (t test, p > 0.05)

Full size image

Ingestion of A. tonsa in terms of 20:5(n-3) fatty acid (EPA), PUFA and tFA was also significantly higher for 9B than for other algae (KW, H = 9.57, H2 = 11.58, H2 = 6.62, respectively, p < 0.01; Fig. 3b, d, e). In contrast, A. tonsa feeding on RHO showed a high daily ration of both ∆5 sterol and 22:6(n-3) fatty acid (DHA) of >12 ng ind−1 d−1, while significantly lower ingestion of sterols and DHA (<2 and 5 ng ind−1 d−1, respectively) were obtained when feeding on 9B and 1G (1-way ANOVA, F2 = 140, KW, H = 12.5, respectively, p < 0.01; Fig. 3c, f). For P. elongatus, no significant differences were observed in ingestion rates in terms of nitrogen, PUFA, total fatty acids (tFA), and EPA between the diets (t test, p > 0.05), but the ingestion of ∆5 sterol and DHA was significantly higher, >16 ng ind−1 d−1, when feeding on RHO than when feeding on 9B (t test, p < 0.05). Similarly, ingestion rates of T. longicornis in terms of nitrogen, EPA, PUFA, and tFA on 9B were significantly higher than RHO, but significantly lower in terms of ∆5 sterol and DHA.

Fig. 3
figure 3

a Ingestion of nitrogen (μg ind−1 d−1), b EPA (ng ind−1 d−1), c DHA (ng ind−1 d−1), d PUFA (ng ind−1 d−1), e tFA (ng ind−1 d−1) f sterol (ng ind−1 d−1), and g PUA (nmol ind−1 d−1), of copepods (Aca: Acartia tonsa, Pseu: Pseudocalanus elongatus, Tem: Temora longicornis) feeding on diatoms and RHO (means ± SD). Striped columns: RHO, black columns: 9B, open columns: 1G. Only species/strains which induced significant ingestion rates were included. Different letters denote treatments that are significantly different from each other (Tukey HSD, p < 0.05)

Full size image

Pellet production rate

Pellet production reflected ingestion, with copepods on 9B diet generally producing most pellets (Fig. 4). The pellet production ranged from 2.8 (7J) to 18 (9B) pellets ind−1 d−1 in A. tonsa, 5 (RHO) to 9 (9B) pellets ind−1 d−1 in P. elongatus, and 2 (RHO) to 54 (9B) pellets ind−1 d−1 in T. longicornis. Differences in average pellet production by A. tonsa between diets were significant (1-way ANOVA, F3 = 17, p < 0.001) with the highest rate observed in copepods feeding on 9B (Tukey, p < 0.001). Further, pellets produced by A. tonsa feeding on 9B had generally larger volume than pellets produced from other food items (Table 2), with the smallest pellets observed from 7J diet. Thus, the total feces volume of A. tonsa was significantly higher on 9B than on other diets (1-way ANOVA, F3 = 21, p < 0.001, Tukey, p < 0.001; Fig. 4). No significant differences were observed in pellet production of P. elongatus between different diets, but production in T. longicornis was significantly higher feeding on 9B than RHO (t test, p < 0.05).

Fig. 4
figure 4

Total feces volume, (×106 μm3 d−1; columns), and pellet production (pellet ind−1 d−1; symbols) of copepods feeding on diatoms and RHO (mean ± SD). Abbreviations as in Fig. 2; different letters denote treatments that are significantly different from each other (Tukey HSD, p < 0.05.)

Full size image
Table 2 Pellets volume (×105 μm3 d−1; average of 30 pellet per treatment) and the efficiency of egg production (E-EP) in terms of carbon (C) feeding on Rhodomonas sp. (RHO) and three Skeletonema marinoi strains
Full size table

Egg production and hatching success

The daily egg production rate and hatching success of A. tonsa, P. elongatus, and T. longicornis are shown in Fig. 5a. Egg production rates ranged from about 0.8 to 23 eggs female−1 d−1 in A. tonsa, 0.5 to 2.4 eggs female−1 d−1 in P. elongatus, and 0.6 to 7.9 eggs female−1 d−1 in T. longicornis, which corresponded to a weight-specific egg production of 0.01–0.32, 0.01–0.06, and 0.02–0.12 μg C (μg C)−1 d−1, respectively. There were strain-specific differences in the mean A. tonsa egg production, with significantly highest production on 1G, second highest on RHO and 9B, and the lowest on 7J (1-way ANOVA, F3 = 16, p < 0.001, Tukey, p < 0.001). No differences were found for P. elongatus between diets (t test, p > 0.05). The egg production of T. longicornis fed on 9B was initially low, but increased in the course of the experiment, reaching a similar level in days 7–8 as measured previously on RHO (Dutz et al. 2008). The egg production on RHO at the present experiments was unexpectedly low and did not increase similarly to the egg production on 9B. As T. longicornis in both treatments (RHO and 9B) originated from the same batch and thus had a similar history, we concluded that the unexplained result on RHO was due to the quality of RHO rather than T. longicornis (see “Materials and methods”).

Fig. 5
figure 5

a Egg production rate (egg female−1 d−1 ± SD) and b egg hatching success (%) of copepods feeding on diatom and RHO during the 8 days incubation (Acartia tonsa, Temora longicornis) and 10 days incubation (Pseudocalanus elongatus). Abbreviations as in Fig. 2. For Temora longicornis egg production, previous measurements on RHO (EPR) are given for comparison (Dutz et al. 2008; symbol on dotted line). Open symbols without line show the hatching on diatoms as a percentage of hatching in RHO. Hatching of T. longicornis in RHO was estimated based on previous result in Dutz et al. (2008). Note different scales of the x-axis

Full size image

Hatching success at the start of the experiment ranged from 56 to 66% in A. tonsa, 64 to 72% in P. elongatus and was 33 to 70% in T. longicornis (9B) (Fig. 5b). For T. longicornis with RHO, hatching success was terminated at day 3 as too few eggs were produced for accurate incubations. Hatching success of A. tonsa was similar with all diets until day 6, when the hatching success on 9B decreased to less than 30% in last day of the incubation. The hatching of P. elongatus and T. longicornis continuously diminished after onset of feeding on 9B diet, although the hatching success with RHO was variable: with P. elongatus, the hatching success with 9B corresponded on average to 63% of the hatching success with RHO. If T. longicornis hatching success on RHO was assumed as in Dutz et al. (2008); the hatching success with 9B corresponded to ca. 35% of the hatching with RHO (Fig. 5b). For 7J strain, hatching success could not be measured, since too few eggs were produced.

Egg production of A. tonsa was positively correlated with the ingestion of DHA and sterol (Spearman rank correlation: p < 0.05; Table 3) and negatively correlated with the C/N and EPA/DHA ratios of the diet. However, there was no significant correlation between hatching and ingestion of any of the nutritional components or PUAs. No significant correlation was observed between the egg production of P. elongatus and the diverse food components, but hatching was found to significantly correlate with the DHA and sterol content of the diet (Spearman rank correlation: p < 0.05). Hatching success of T. longicornis with 9B was not correlated with any of the nutritional components or PUAs (Spearman rank correlation: p > 0.05), whereas egg production correlated negatively with ingestion in terms of carbon, nitrogen, lipid (EPA, PUFA, and tFA), and PUA (Spearman rank correlation: p < 0.05).

Table 3 Spearman correlation coefficients between reproductive response parameters (EPR; eggs fem−1 d−1 and HS; %) and ingestion of mineral and biochemical components and polyunsaturated aldehydes (PUAs) in Acartia tonsa (n = 25), Pseudocalanus elongatus (n = 19), and Temora longicornis (n = 11) (using Rhodomonas sp. (T-RHO, Dutz et al. 2008) as comparison) females
Full size table

Calculation of the efficiency of egg production in A. tonsa showed that with RHO and 1G, the ingested carbon was utilized with a higher efficiency (0.21 and 0.20) than with 9B (0.10; Table 2). For T. longicornis, the carbon egg efficiency with 9B was ca. 0.06, ca. 3 times lower than observed previously on RHO (Dutz et al. 2008). In P. elongatus, the efficiency did not exceed 0.09 with either RHO or 9B.

Behavioral effects

A general assessment of copepod distribution was made, based on the percentage of copepods in the thin layer with food. Both starved and fed A. tonsa typically resided in the middle layer, T. longicornis in the upper layer, and P. elongatus in the bottom layer (Fig. 6). For fed and starved A. tonsa, approximately half of the copepods were present in the thin middle layer, with significantly higher abundance in the food layer with Skeletonema (strains 9B and 1G) than in FSW control (2-way ANOVA, F3 = 7, p < 0.001). For both T. longicornis and P. elongatus, <20% of the copepods remained in the food layer. No significant differences between the experiment (food) and FSW control were observed in both fed and starved P. elongatus and starved T. longicornis. In contrast, copepod concentration in the food layer was elevated in fed T. longicornis (2-way ANOVA, F1 = 8, p < 0.001, Tukey, p < 0.001). We thus did not observe any avoidance of the food patch containing Skeletonema, irrespective of the PUA production of the strain.

Fig. 6
figure 6

Distribution (%) of a fed and b starved copepods in different treatment. Upper water column without algae; Mid water column with algae; and Lower water column without algae. Different letters denote treatment significantly different from each other (Tukey HSD, p < 0.05). Abbreviations as in Fig. 2

Full size image

Comparison of swimming speed showed that copepods moved an average of ~5 mm s−1 in all treatment conditions, with the highest swimming in fed A. tonsa with 9B strain (6.2 mm s−1). The turning angle ranged from 0.001 to 0.04 with the largest angle observed in starved T. longicornis with 9B strain. Only starved A. tonsa (2-way ANOVA, F3 = 6, p < 0.05; Fig. 7) and fed P. elongatus (2-way ANOVA, F1 = 8, p < 0.05) significantly altered their swimming speed in response to food treatments, while both fed and starved A. tonsa, fed P. elongatus, and starved T. longicornis altered their turning angle (A. tonsa, 2-way ANOVA, F3 = 23, for A. tonsa, P. elongatus, F1 = 11 T. longicornis, F1 = 5). Typically, these copepods moved significantly more in food treatment and turned significantly less when placed in FSW but no reduction in swimming activity was observed in the presence of S. marinoi.

Fig. 7
figure 7

Swimming speed (mm s−1) and turn angle (θ) of fed and starved Acartia tonsa, Pseudocalanus elongatus, Temora longicornis in different food treatment (mean ± SD) observed in a mid column with algae. Different letters denote significantly different treatment from each other according to Tukey HSD, p < 0.05. Abbreviations as in Fig. 2

Full size image

Discussion

Strain-specific effects of Skeletonema marinoi

Our study revealed strong strain-specific effects of Skeletonema marinoi on ingestion, egg production, and hatching success of Acartia tonsa, summarized in Table 4. All three Skeletonema strains differed in size, chain length, and carbon content (Table 1) and induced very different feeding. Typically, 7J was not eaten, while 1G was ingested at a similar rate to the control RHO, and 9B was ingested at even higher quantities. 7J had a low PUA content, suggesting that the low feeding was not influenced by a potential effect of PUA as a feeding deterrent. Instead, low ingestion could have been due to the small cell size, combined with a relatively long chain length (Table 1), which could have made this chain difficult to ingest. Of the remaining strains, egg production was high on 1G and moderate to low on RHO and 9B, while the hatching success was only reduced on 9B (last day of the incubation). Later, we consider diatom aldehydes, nutritional compounds, and alternative explanations as potential reasons explaining the observed patterns.

Table 4 Overview of effects on the physiology and behavior of three copepod species when fed three different strains of Skeletonema marinoi
Full size table

Diatom aldehydes

Acartia tonsa has been reported to have a higher reproductive success when feeding on Skeletonema than when feeding on Phaeocystis pouchetii (Verity and Smayda 1989) and ciliates (Ederington et al. 1995). However, low fecundity and hatching of Acartia spp. have also been demonstrated with the same diatom species in several laboratory studies (Ban et al. 1997; Ianora et al. 2003) and in the field, with only 12% of the eggs hatching during a Skeletonema bloom (Miralto et al. 1999). These observations of a low reproductive success have been attributed to the production of PUA (Ianora et al. 2009).

In the present study, egg production and hatching of A. tonsa had no direct connection to the ingestion of PUA (Table 3). This finding is in agreement with recent laboratory (Temora longicornis, Dutz et al. 2008) and field (from the English Channel off Roscoff, Calanus helgolandicus, Poulet et al. 2006; Wichard et al. 2008) studies showing no evidence that PUAs were involved in the overall observed reproduction failure. Further, the daily ingested amount of PUA by A. tonsa in this study was estimated to be 0.004–0.02 μg ind−1 d−1 (29–86 pmol ind−1 d−1), which is higher than the 100 pg (=0.0001 μg) daily ingestion of decadienal, which was reported by Ianora et al. (2004) to have strong adverse effects on copepod reproduction. Our estimated ingestion rate was also close to the highest ingestion reported by Dutz et al. (2008) of 100 pmol ind−1 d−1(0.1 μg ind−1 d−1), (note that there is a calculation error in Fig. 2f of Dutz et al. 2008; with a one magnitude difference), which did not generate apparent effects on reproduction. The results indicate that either there is no constant critical level of PUA ingestion for harmful effects to occur or other factors than PUAs explain the observed strain-specific effects for reproduction.

Nutritional deficiency

The diatom strains also differed in their composition of lipids, especially with respect to the contents of specific PUFAs and sterols, as well as in their C/N ratio. Of the two ingested strains, 9B and 1G had a relatively similar fatty acid contents (per volume unit, Table 1). However, there was a fourfold difference in the volume-specific sterol content of these strains, with 9B containing ca. 3 times as much sterols as 1G. In addition, the C/N ratio was lower in 9B and this strain thus appeared more beneficial than 1G (for sterol and nitrogen content). Nevertheless, 1G induced higher egg production, hatching success and gave higher egg production efficiency.

If the egg production of A. tonsa was plotted against ingestion of different compounds, the egg production appeared positively correlated with the ingestion of DHA and sterols and negatively correlated with the EPA/DHA and C/N ratios. Various studies have reported the effect of these compounds on copepods growth and recruitment. For instance, a low C/N ratio (Jones and Flynn 2005) or a high concentration of unsaturated fatty acid (Kleppel et al. 1998; Broglio et al. 2003), especially EPA and DHA (Jónasdóttir 1994; Jónasdóttir and Kiørboe 1996; Shin et al. 2003) and sterols (Klein Breteler et al. 1999; Hasset 2004) in the diets have been shown to have beneficial effects. The C/N ratio of the algae varied from 4.8 to 6.3, indicating that the differences in C/N ratios were small (Table 1). Although the ratios in our study were lower than the C/N previously reported to have significant effect on fecundity (Kiørboe 1989), it appears that particulate nitrogen content of food did have an effect on A. tonsa egg production, as also shown in previous studies (Koski et al. 2006; Peters et al. 2007). However, even though stoichiometric analysis can be indicative for food quality limitation, there is evidence that C/N ratios are not adequate predictors to assure reproductive success in crustaceans (reviewed by Broglio et al. 2003) and the effect of mineral nutrients can be connected to changes in biochemical compounds of the algae (Klein Breteler et al. 2005).

The importance of the balance in EPA/DHA ratios has been observed in several previous studies (Støttrup and Jensen 1990; Jónasdóttir 1994; Jónasdóttir and Kiørboe 1996), with a smaller ratio tending to promote a high reproductive success in copepods. Although it is not really clear why low EPA/DHA ratio profits copepod reproduction, it has been demonstrated regularly that inappropriate composition of essential fatty acids lead to a lower reproductive success (e.g., Evjemo et al. 2008).

The significant relationship of egg production with sterol supports other studies indicating that the sterol composition obtained by females is of some importance in controlling reproduction rates (Klein Breteler et al. 1999; Hasset 2004). Recent studies reported that cholesterol is the dominant sterol in crustaceans and is required for growth, egg production, and hatching (Hasset 2004; Crockett and Hassett 2005). However, in our study, no cholesterol was detected in the 1G diet (Table 1), suggesting that high egg production rate could be maintained without significant cholesterol content in the diet.

Alternative explanations

Hatching success of A. tonsa was not correlated with the ingestion of carbon, nitrogen, PUFAs, sterols, or PUAs. It is possible that A. tonsa hatching success suffered from a deficiency of other biochemical compounds, not included in our measurements, or was influenced by specific metabolites. These could include, e.g., amino acids, proteins, various vitamins (e.g., Kleppel and Burkart 1995), dimethyl sulfide, and other unidentified metabolites (see Barofsky et al. 2009). Besides that, the production of fatty acid hydroperoxides (FAHs) and oxylipin such as hydroxyacids and epoxyalcohols have been observed to depress the viability of copepods (Fontana et al. 2007b). Recently, Ianora et al. (2010) observed negative effect on Temora stylifera feeding on Pseudo-nitzschia delicatissma, which does not produce PUA but oxylipins, suggesting that these metabolites may be another important factor behind the negative effects of some diatoms diet in copepods. However, it is uncertain at present how widespread this oxylipin production is compared to PUAs.

Behavioral effects

In the present study, higher aggregation of A. tonsa in diatom food patch clearly showed that this copepod is able to locate high food concentration. This result is consistent with previous study using a similar setup (Tiselius 1992) where a high aggregation of A. tonsa remained in the thin layer of the diatom Thalassiosira weissflogii. As A. tonsa does not store energy reserves, it should have developed strong behavior to remain in the food patches. Distinct effects on A. tonsa behavior were also observed between FSW control and food treatment in our study. The pattern of increased retention in a food layer can result from area-restricted search strategy in which swimming and turning angles increase when entering a food patch (Tiselius 1992).

In contrast, Bochdasky and Bollen (2004) reported no aggregation of Acartia hudsonica in the food layer of Skeletonema costatum suggesting that this diatom is not a preferred food source. However, although no aggregation was evident in their 12-h observation, copepods were found crowding in the food patch in the first 2 to 4 h of the experiment. We assumed that the behavioral changes in relation to different food properties and PUA production would be fast, and 2-h observations would thus be adequate to assess the potential changes in copepods swimming activity. Because the average time that it takes for the copepods to fill their guts in high food concentrations is relatively short (e.g., 2-h; Leising and Franks 2000), we argue that the behavioral changes during the first hours following changed food conditions would be the most relevant. Although we did not observe any differences in copepod behavior related to the different strains, we cannot, however, exclude that copepod swimming and aggregation behavior could have changed during a longer experimental duration.

Thus, our result indicates that PUA-producing algae do not exert any deterrent response on copepods. This result supports previous study showing that A. tonsa exhibit minimal behavioral effects when exposed to Karenia brevis or brevetoxins with higher toxic accumulation in their body (Cohen et al. 2007). In contrast, Temora turbinata in the Gulf of Mexico actively avoided dense concentration of Karenia brevis, suggesting that the behavior of this species provide some refuge from Karenia brevis toxicity (Lester et al. 2008). There might thus be species-specific differences in copepod behavior as a response to toxic algae, although this was not visible with S. marinoi.

Copepod-specific effects of the 9B strain

The 9B strain of Skeletonema induced a reduction in egg production in A. tonsa and P. elongatus and a reduction in hatching success in A. tonsa and T. longicornis, while maximum egg production of T. longicornis (day 7–8) and P. elongatus hatching success remained relatively high. It has been long known that similar diatom species produce inconsistent effects for different copepod species (Ianora et al. 2009). For instance, Ban et al. (1997) showed a strong reduction in fecundity and egg viability in Acartia clausi but not in hatching of Calanus helgolandicus and neither of the two in Calanus finmarchicus, when feeding on S. costatum. We can exclude the effects of growth-stage specific PUA production in the food algae, since the 9B strain was used at the same time and with the same concentration for all three copepod species. However, Ianora et al. (2009) also suggest that diatom effects on copepod recruitment can vary greatly between copepod species, probably due to differing copepod sensitivity or detoxification capacity. Although A. tonsa did not show any significant correlation between egg production/egg hatching and PUA ingestion with the three S. marinoi strains (see earlier), the other two copepods species were only tested with 9B strain. Thus, our results cannot exclude differences between the copepod species in sensitivity or detoxification capacity.

It also appears that copepods have species-specific differences in fatty acid requirements (Jónasdóttir et al. 2009 and references therein). In the present study, we only observed a positive correlation between egg production of A. tonsa and hatching of P. elongatus and DHA, and sterol ingestion. The reason behind these species-specific responses remains unclear, but could, for instance, originate from different internal sources of lipid in females (Lee et al. 2006), which may result in different dependency on fatty acid ingestion in our study.

It appeared that only A. tonsa was remaining in the food patch, while T. longicornis and P. elongatus were mostly located either above or below it. Similar to A. tonsa, T. longicornis has also been observed to increase their time spent in a region of high food concentration (Woodson et al. 2005). As T. longicornis is a species with high metabolic requirements and low energy reserves, it can also be expected to respond quickly to changes in food quantity and quality (Mayzaud et al. 1992). However, although 9B was ingested at higher rate than the other food items in the grazing experiments, T. longicornis aggregated mostly in the upper part of the aquarium with FSW. The cause of this response is not clear, but one possible explanation is that density gradients may act as barriers to vertical migration for this copepod, thus resulting in aggregations at these boundaries due to a physical or a behavioral barrier (Woodson et al. 2005).

Pseudocalanus elongatus was observed to be distributed mostly in the bottom of the aquarium. P. elongatus has been found to store energy in the form of lipid droplets, available as long-term energy reserve and can thus survive starvation longer than T. longicornis (Koski and Breteler 2003). It has also been observed that Pseudocalanus minitus and Calanus finmarchicus are capable of withstanding long period of starvation, suggesting that they can metabolically remove themselves from changes in food environment (Dagg 1977). Fast behavioral response to food patches may, therefore, not be that essential for P. elongatus survival. A starving response by T. longicornis was demonstrated by a higher turning frequency in 9B than in the FSW control. In contrast, starved P. elongatus showed no significant response in their behavior, suggesting that species-specific life strategies give species-specific behavioral characteristic in feeding and swimming.

In conclusion, the observed negative effects of S. marinoi on feeding, egg production, and hatching success indicate that one diatom species can range from being a high quality food to directly inducing inhibitory effects. Different strains of S. marinoi thus induce widely different responses on copepod physiology, ranging from strongly reduced feeding (7J) and moderately reduced egg production (9B) to high ingestion, egg production and hatching (1G). Although inconclusive, our results show clearly that (1) PUA production of diatoms cannot explain the reduction in copepod reproduction, (2) all S. marinoi strains are not harmful, and (3) specific lipids typically appear profitable for reproduction, although the different nutritional demands of different species and life stages, as well as the potential co-limitation of different compounds, make it unlikely that one element could explain the reproduction.

Although the species diversity in phytoplankton blooms is typically lower than at the times of intermediate phytoplankton biomass (Irigoien et al. 2004), diatoms, and especially S. marinoi, have a high strain-specific genetic variability (Saravanan and Godhe 2010). Our results show that this strain-specific variability is an important factor controlling the interactions between diatoms and their copepod grazers. It is therefore important to identify the specific characteristics of local diatoms in order to evaluate their potential impact on copepods (Pohnert et al. 2002). Future studies should consider these strain-specific interactions, together with the interacting effects of different mineral, biochemical, and toxic compounds, and their relevance for copepod secondary production. In our opinion, looking into details of copepod-diatom interactions might prove useful when predicting the copepod population dynamics during diatom blooms.