Introduction

Despite nearly two decades of intensive research, the effect of diatoms on copepod secondary production is still an unresolved question. It seems, however, clear that (1) all or most diatom species induce reduced hatching success after some days of incubation if offered as a monospecific diet (Turner et al. 2001 and references therein; Dutz et al. 2008), (2) diatom effects on other life-history parameters such as egg production, female mortality and juvenile development are more diverse (e.g., Ban et al. 1997; Carotenuto et al. 2002) and (3) egg production and hatching success are often uncoupled and regulated by different factors (Jónasdóttir and Kiørboe 1996). Still, the questions about the mechanisms and relevance of the diatom effects remain unsolved.

Until now, several mechanisms have been suggested to be the reason behind decreased hatching success (and sometimes nauplii development) on diatom diets. Firstly, α,β, γ, δ-polyunsaturated aldehydes (PUA) have been observed to block the embryogenesis of copepod eggs and therefore to reduce the hatching success (Miralto et al. 1999). Secondly, nutritional (either biochemical or mineral) deficiencies of diatoms have been shown to induce poor food quality of pure diatom diets, which, similarly to PUA effects, reduce hatching success (Jónasdóttir and Kiørboe 1996; Jónasdóttir et al. 1998; Jones and Flynn 2005). Recent studies add a physiological explanation concerning low assimilation efficiency of essential compounds in copepod guts (Dutz et al. 2008), a depletion of polyunsaturated fatty acids (especially eicosapentaenoic acid EPA) in certain diatoms caused by the PUA production (Wichard et al. 2007) and compounds inducing oxidative stress (Fontana et al. 2007) to the list of suggested mechanisms behind the negative effects of diatoms. Since it seems, at present, likely that neither PUA production nor direct nutritional limitations can fully explain the diatom effects (Poulet et al. 2006, 2007; Dutz et al. 2008; Wichard et al. 2008), the actual mechanism(s) remain, however, unknown.

Similarly, field studies on copepod–diatom interactions present controversial results, and the field relevance of diatom effects is under debate. While some studies observed decreased hatching success during diatom blooms (Miralto et al. 2003), other studies found no evidence of negative effects of high diatom concentrations on hatching success in nature (Irigoien et al. 2002). Studies including grazing experiments give similarly contrasting results: while ingestion of PUA-producing diatoms had a negative effect on hatching success and early nauplii survival of Calanus pacificus during diatom blooms (Leising et al. 2005; Pierson et al. 2005), high ingestion of diatoms coincided with high hatching success and early nauplii development of Calanus finmarchicus (Koski 2007). Although selective omnivorous feeding should lead to diverse (not exclusively diatoms) diets (see Kleppel 1993), some copepod species seem, however, to select for diatoms (Meyer-Harms et al. 1999; Koski 2007), which may or may not lead to harmful effects for reproduction (Pierson et al. 2005; Meyer-Harms et al. 1999, respectively). From an evolutionary perspective, for species, which have annual peak abundance during the spring diatom bloom, it should be essential to be able to feed on diatoms and to produce viable eggs. For recruitment to be successful, the hatched nauplii should in addition be able to develop on diatom-dominated diets. The evolutionary aspects of copepod–diatom interactions, such as shown for copepods developing during dinoflagellates blooms (Colin and Dam 2002a), have, however, not yet been investigated.

Until now, several laboratory studies have shown low development rates and high nauplii mortality of several copepod species on monospecific diatom diets (Poulet et al. 1995; Carotenuto et al. 2002; Ianora et al. 2004), suggesting, similarly to hatching success, that PUA production (Carotenuto et al. 2002; Ianora et al. 2004) or algae mineral (Jones and Flynn 2005) and biochemical (Klein Breteler et al. 2005) composition are the factors determining the quality of diatoms for somatic growth and development of copepods. However, several copepod species have also been successfully grown from early nauplius stages to adults with monospecific diatom diets, without any obvious negative effects (Paffenhöfer 1976; Harris and Paffenhöfer 1976; Vidal 1980a, b). The effect of diatoms on copepod nauplii seems thus to be more diverse than the effect on hatching success: sometimes contrasting results have been obtained with the same diatom species or even between different replicate experiments (Carotenuto et al. 2002).

In diatom studies, maternal diet seems to be equally important in determining the development of the next generation than the actual diet of the extant generation (Poulet et al. 1995, 2003; Ianora et al. 2004), although a pure diatom diet can also be inadequate for development irrespective of the maternal diet (Carotenuto et al. 2002). Potentially both PUA production and nutritional deficiencies could also induce maternal effects. Poulet et al. (1995, 2003) showed that ≥70% of the nauplii produced on maternal diatom diets had serious malformations leading to high mortality, and attributed the effects to PUA production. However, poor egg and nauplii quality can also be induced with low food concentration (Guisande and Harris 1995) or starvation (Poulet et al. 2003), suggesting that the maternal effects of poor nutrition and PUA production on copepod development may be similar. The reports of low juvenile growth (Carotenuto et al. 2002), arrested embryonic development (Ban et al. 1997) and malformations of nauplii (Poulet et al. 1995) on diatoms that are not listed as PUA-producers (Wichard et al. 2005a), as well as the observations of the gradually decreasing development rates with increasing nutrient limitation (Klein Breteler et al. 2005), further stress the role of other factors in determining the food quality of diatoms. Simultaneous measurements of juvenile growth rate and mortality with the algae nutritional quality and production of deleterious PUA are therefore necessary.

Studies that directly link copepod growth or development on diatom diets to either mineral or biochemical limitations are, however, rare (Jones and Flynn 2005; Klein Breteler et al. 2005), as are the studies including actual quantification of PUAs (Carotenuto et al. 2005 for cladocerans). Further, most of the studies concerning copepod development and diatoms have been done with a limited number of diatom species (namely Thalassiosira rotula, Phaeodactylum tricornutum and Skeletonema costatum), excluding, e.g., the genus Chaetoceros which is among the dominant spring bloom diatoms in, e.g., North Sea (Riebesell 1991). The present study focuses on these aspects by estimating (1) how common the negative effects of diatoms are on copepod development and nauplii mortality and (2) whether the arrested development is connected to deleterious PUAs or food nutritional quality. For this purpose we measured development, growth and juvenile mortality of a common boreal spring copepod Temora longicornis on 11 different monospecific diatom diets and related the growth and development rates to the ingestion of PUA, particulate organic carbon (POC), nitrogen (PON), polyunsaturated fatty acids (PUFA) and sterols. In addition, the suspect algae were tested for toxicity using a bio-assay approach of Jónasdóttir et al. (1998). We show that, although some diatoms are of inferior food quality, this is unlikely to be connected to the PUA content of the species or due to a direct limitation by a single dietary nutritional compound.

Materials and methods

Algae and copepods

Experiments were conducted using a calanoid copepod T. longicornis, originating from the central North Sea, but cultured in laboratory for over ten generations. Copepod cultures were kept at 14°C in dark, and fed in excess (>400 μg C l−1) a mixture of Rhodomonas sp., Thalassiosira weissflogii and Heterocapsa sp. Nauplii for the experiments were collected from the stock culture directly at the start of the experiments, using a 140 μm net which separated early naupliar stages (NI–IV) from late naupliar stages, copepodites and adults. To additionally ensure an even quality of nauplii, a treatment with a standard diet (Rhodomonas sp.) was included in every series of experiments (see below).

Algae used in experiments were the cryptophyte Rhodomonas sp. and the diatoms T. weissflogii (grown at the National Institute of Aquatic Resources, Technical University of Denmark; strain unknown), T. rotula strains CCMP1647 and 1018, Thalassiosira pseudonana strains CCMP1010 and 1335, S. costatum CCMP1281, Leptocylindricus danicus CCMP469, P. tricornutum CCMP630, Chaetoceros affinis CCMP158, Chaetoceros decipiens CCMP173 and Chaetoceros socialis (originating from the Royal Netherlands Institute for Sea Research; strain unknown; Table 1). Algae were cultured in 1–2 l batch cultures with F/2 + Si medium (Guillard 1975), at 18°C, an irradiance of 100 μmol photons m−2 s−1 and 14:10 h light–dark cycle. The algae were kept in exponential growth phase by diluting 30–50% of the culture 3–4 times a week. The cell concentration in cultures was checked 3–4 times a week using a Coulter Counter (Multisizer 3; Beckman Coulter): a low algae biomass (<20 μg C ml−1) was taken as an indication of an exponential growth phase. All analyses were performed on exponentially growing algae.

Table 1 Strain number, volume (μm3), carbon and nitrogen content (pg cell−1), C:N ratio (weight), content of polyunsaturated fatty acids EPA, ARA and DHA, total PUFAs (pg cell−1) and Δ5 sterols (pg cell−1) and the fatty acid ω3:ω6 ratio in the algae used in experiments (mean ± SD)
Full size table

The volumes of the algal cells were measured 3–4 times a week by using the Coulter Counter except for species which formed chains (Chaetoceros spp.). For these latter, the volume was evaluated by using an inverted microscope. Carbon and nitrogen content of algae were determined by combustion in a Carlo Erba Analyser, after filtering 5–10 ml of the culture on combusted GF/F filters (2–11 replicates), and kept frozen until analysis. The replicate samples were taken at different time points before and during the experimental period in order to account for the potential variability in the cultures. For T. pseudonana CCMP1335, S. costatum, C. decipiens, and C. socialis carbon and nitrogen contents were not measured, but estimated from the average volume:carbon and volume:nitrogen regressions obtained for the other diatom species (r = 0.90, n = 7, P < 0.01). Despite the cell nitrogen content being a rather linear function of the cell volume, a high C:N ratio of some species (most notably P. tricornutum; Table 1) indicated nitrogen stress. It should thus be noted that, although the algae were kept in an exponential growth state, this did not necessarily exclude a possibility of a nutrient limitation.

Fatty acid and sterol analyses were conducted from 1–4 replicate samples, with a slight adjustment of standard methods (Klein Breteler et al. 1999) as described in Dutz et al. (2008). Here we considered only samples taken at different time points to be replicates; in fact each replicate measurement presented in Table 1 consisted of 2–3 replicate analysis (samples taken at the same day). Extracted FAMEs were analyzed on a GC–MS (Agilent 6890 with PTV inlet and Agilent 5973 mass selective detector) on an Agilent DB23 (60 m × 0.25 mm) column using helium as a carrier gas and subsequently, after silylation on a GC–MS (as above) on a Sil-5 (25 m × 0.32 mm) column using helium as a carrier gas. Retention times were compared to those of known FAME and sterol mixtures (from Matreya and Larodan, respectively). Sterols are presented here as the total concentration of three identified Δ5 sterols, cholesterol (Cholesta-5-en-3β-ol), brassicasterol (24-β-methylcholesta-5,22-dien-3β-ol) and campesterol (24α-methylcholesta-5-en-3β-ol). Total concentration of PUFAs represents a sum of 11 C16-C22 PUFAs. The algal volumes, carbon and nitrogen contents, C:N ratios, and concentrations of PUFAs eicosapentanoic acid (EPA; 20:5 ω3), arachidonic acid (ARA; 20:4ω6) and docosahexaenoic acid (DHA; 22:6ω3), total PUFAs and Δ5 sterols, as well as the fatty acid ω3 to ω6 ratios, are listed in Table 1.

For quantification of PUA, samples were taken once during the experiments and prepared as described in Wichard et al. (2005b), derivatized with O-(2,3,4,5,6-pentafluorobenzyl) hydroxylamine hydrochloride, and extracted with hexane. The potential PUA production in the algal diets was quantified by GC–MS in triplicates. Except for T. rotula CCMP1018, the measured PUA concentrations (or lack of PUA) were consistent with previous measurements from the same species/strains (Wichard et al. 2005a). However, T. rotula CCMP1018, which a few months earlier was producing only trace amounts of PUA (Dutz et al. 2008), produced elevated amounts of PUA at the time of the experiments and formed a distinguishable pattern of PUA compared to the strain T. rotula CCMP1647. The total PUA production of CCMP1018 seems thus to be very variable, ranging from zero to ca 6 fmol cell−1 (Pohnert et al. 2002; Wichard et al. 2005a). The produced amounts per cell and the distribution of various PUAs are presented in Table 2.

Table 2 Concentration (fmol cell−1 ± SD) and profile (%) of polyunsaturated aldehydes (PUA) in the species with a detectable amount of PUA
Full size table

Experiments

Temora longicornis was grown from early naupliar stages on 11 different monospecific diatom diets (Table 1), a control diet of Rhodomonas sp., and 1:1, 1:3 and 3:1 mixtures (in carbon) of Rhodomonas sp. with P. tricornutum and Rhodomonas sp. with T. rotula CCMP1018. The monospecific diets were offered at a volume concentration of 2.2 (±0.9) ppm, which corresponded to ca 240 (±100) μg C l−1. This concentration corresponds to an average spring bloom concentration in, e.g., the North Sea (Riebesell 1991; Kiørboe and Nielsen 1994). There were, however, two exceptions for this concentration: Rhodomonas sp. had a 2–3 times lower volume to carbon ratio than diatoms, resulting in a high carbon concentration of 760 (±22) μg C l−1 and S. costatum was offered to copepods in a lower concentration of 0.8 ppm (ca 50 μg C l−1). The mixtures had a total food concentration of 1.2 (±0.4) ppm or 250 (±80) μg C l−1.

The set up of the experiments followed the protocol described in Koski et al. (1998): a high number of nauplii (ca 1,000 l−1) were placed in 1.2 l bottles filled with the aimed food suspension, the bottles were sampled three times weekly, and the sample volume was adjusted to keep the copepod biomass in the bottles constant (the removed biomass represented approximately the biomass increase due to growth). Simultaneously with the sampling, >80% of the food suspension was renewed with reverse filtration. The food concentration never decreased by more than 20% during the incubations. The number, length (total length of nauplii, prosome length of copepodites and adults ± 20 μm) and development stage of copepods in samples were analysed using a binocular microscope (at least 15, but typically >30 individuals per sample). The experiments were terminated either when all copepods had died, or when the first adults appeared. The experiments were conducted in four sessions during a period of ca 6 months, with each diatom experiment being repeated 1–3 times. Simultaneously with every set of a diatom experiment, a Rhodomonas sp. control was run (see Fig. 1a), to ensure that the quality of nauplii remained stable: if nauplii development in Rhodomonas sp. control was not optimal, the whole set of experiments were omitted.

Fig. 1
figure 1

Temora longicornis. Development on a monospecific diatom diets and on the control diet Rhodomonas sp. and b in 1:1, 1:3 and 3:1 mixtures of Rhodomonas sp. + Phaeodactylum tricornutum and Rhodomonas sp. + Thalassiosira rotula CCMP1018 (mean stage as a function of time). Different symbols indicate different experiments; in a the symbols in diatom experiments correspond to the simultaneously run Rhodomonas sp. controls, in b open symbols and dashed or dotted lines indicate mixtures, closed symbols and solid line Rhodomonas sp. The line indicates a three parameter power function. For abbreviations, see Table 1

Full size image

Three to six replicate grazing experiments were conducted with nauplii together with each of the diatom species, Rhodomonas sp., and mixtures, to check that any failure in development was not due to low ingestion. At the start of the experiments, 40–60 nauplii of stages IV–VI were collected from the stock culture and placed in 330-ml bottles containing the aimed food suspension. After 24 h of adaptation to food, copepods were carefully transferred into a new food suspension and three replicate control bottles without animals were set up. The concentration of cells in the start suspension was counted from replicate samples using a coulter counter. After 24-h incubation in dark, cell concentration was recounted with the coulter counter, the contents of the bottles were carefully filtered onto a 30 μm net and flushed into a petri-dish, and the number, condition (dead/alive) and development stage of copepods were analysed. All experiments were conducted at 14°C, in a walk-in temperature controlled room, and the bottles were rotated at a speed of ca one round-per-minute.

Development was expressed as an increase in mean stage (calculated from the stage frequency distribution in the samples) against time. Copepod carbon content was estimated based on length measurements and length:dry weight regression of Klein Breteler et al. (1982), assuming carbon content to be 40% of the dry weight (Mullin 1969). Instantaneous growth rates were estimated assuming an exponential increase in carbon content over the course of time (e.g., Huntley and Lopez 1992). Instantaneous rates of mortality during development were calculated after correcting for sampling mortality according to Klein Breteler et al. (2004). To avoid bias due to a low number of individuals in the samples with species inducing high mortality (mean stage based only on few individuals), only samples with ≥15 individuals were used to estimate development and growth rate, whereas all samples were used to calculate mortality. This resulted in somewhat different numbers of observations in development/growth rate versus mortality rate (cf. Tables 3, 4; Fig. 1). Filtration and ingestion rates were calculated according to Frost (1972). To calculate weight-specific ingestion, the mean development stage measured in the end of the grazing experiments was converted to carbon, by using a mean stage:carbon (based on length) regression obtained for copepods fed Rhodomonas sp. (r = 0.98, n = 25, P < 0.0001). Gross growth efficiency was calculated by dividing the weight-specific ingestion with the weight-specific growth rate.

Table 3 Temora longicornis: parameters of linear model of growth, relating ln mean weight (μg C) to time (days)
Full size table
Table 4 Temora longicornis: parameters of linear model of mortality, relating ln no. of individuals to time (days)
Full size table

Growth and mortality between treatments were tested for differences using an analysis of covariance (ANCOVA), while filtration and weight-specific ingestion rates were tested using a one-way analysis of variance (ANOVA). In addition, filtration and ingestion rates were formally tested against zero (insignificant ingestion/filtration) using a one-sample t test. Tukey HSD post hoc test was used for pairwise comparisons. To test whether growth or mortality were related to the quantity or quality of ingested food, Spearman rank order correlation analysis was run for weight-specific growth, daily mortality, carbon and nitrogen ingestion, C:N ratio, fatty acid ω3 to ω6 ratio, and ingestion of PUA, total PUFA, linolenic acid (LIN; C18:3ω3), EPA (20:5ω3), ARA (20:4ω6), DHA (22:6ω3) and Δ5 sterols.

Results

Temora longicornis was able to complete its development on the control algae Rhodomonas sp. and on the diatoms T. weissflogii, T. rotula strain CCMP1647, L. danicus, and S. costatum. In contrast, development was arrested in late copepodite stages when copepods were fed T. rotula CCMP1018, either of the T. pseudonana strains, P. tricornutum or any of the Chaetoceros species (Fig. 1a). However, with T. rotula CCMP1647 there was a large variation between replicate experiments, with complete development in one experiment, but arrested development in the other, although the growth rate of nauplii on Rhodomonas sp. in simultaneous incubations was invariably high indicating a stable quality of nauplii (Fig. 1a). In food mixtures, development was faster than on pure Rhodomonas sp. diet when Rhodomonas sp. was supplied with T. rotula CCMP1018, but slower (1:1 and 1:3) or similar (3:1) than on pure Rhodomonas sp. when Rhodomonas sp. was supplied with P. tricornutum (Fig. 1b).

Growth and mortality reflected the development, with highest growth rates (14–16% body carbon day−1) on T. weissflogii, S. costatum, T. rotula CCMP1647, Rhodomonas sp., all Rhodomonas sp. + T. rotula CCMP1018 mixtures and Rhodomonas sp. + P. tricornutum 3:1 mixture (slope in Table 3), and insignificant or low mortality (≤7% day−1) with the same species/mixtures and L. danicus (slope in Table 4). Based on the analysis of covariance, the rest of the species were divided into two groups: with T. rotula CCMP1018, T. pseudonana CCMP1010, P. tricornutum, C. affinis, C. decipiens and Rhodomonas sp. + P. tricornutum 1:3 the growth was 7–13% body carbon day−1 and the mortality 17–29% day−1, while with T. pseudonana CCMP1335 and C. socialis growth was insignificant and mortality 44–47% day−1 (Tables 3, 4). There were thus significant differences in both growth and mortality between the different species (ANCOVA; P < 0.0001). Further, there was a significant negative correlation between growth and mortality (Spearman rank correlation, P < 0.01), suggesting that with the species inducing high growth rate mortality was low (Fig. 2).

Fig. 2
figure 2

Temora longicornis. Weight-specific growth [μg C (μg C)−1 day−1] as a function of specific mortality (day−1) in development experiments (mean ± SE). Closed symbols indicate monospecific diets, open symbols mixtures. Parameters from the Spearman rank correlation are indicated in the figure; only species which were ingested by T. longicornis are included in the figure

Full size image

Of the species, which did not support full development, neither T. pseudonana CCMP1010 nor any of the Chaetoceros species were ingested by T. longicornis nauplii (Fig. 3). Most of the remaining species were filtered with a rate ranging from 0.01 to 0.04 ml ind.−1 h−1 (Fig. 3a), which resulted in a weight-specific ingestion of 0.1–0.6 μg C (μg C)−1 day−1 for the nauplii (Fig. 3b). Due to the significantly higher filtration rate on S. costatum than on any of the other food species (Tukey HSD; P < 0.001), there were significant differences in filtration rate between different food species (1-way ANOVA; F 7 = 19, P < 0.001), although the differences in ingestion rate were only weakly significant (ANOVA; F 7 = 3.5, P < 0.05). The high filtration rate on S. costatum was likely due to an accidentally low food concentration with this species (see “Materials and methods”). When P. tricornutum was mixed with Rhodomonas sp., it was never ingested, irrespective of the food ratio. In contrast, in Rhodomonas sp. + T. rotula CCMP1018 mixture, both species contributed to ingestion of nauplii (Fig. 4). There was no correlation between the C:N or ω3: ω6 ratio of the diet, ingestion of carbon, nitrogen, PUFA (either total or LIN, ARA, EPA and DHA separately), Δ5 sterols or PUAs and either nauplii growth or mortality (Spearman; P > 0.05; Fig. 5).

Fig. 3
figure 3

Temora longicornis nauplii. a Filtration rate (ml ind.−1 h−1) and b weight-specific ingestion rate [μg C (μg C)−1 day−1] on monospecific diets (mean ± SE). Ns Not significantly different from zero. For other abbreviations, see Table 1

Full size image
Fig. 4
figure 4

Temora longicornis nauplii. a Filtration (ml ind.−1 h−1) and b weight-specific ingestion rates [μg C (μg C)−1 day−1] in food mixtures (mean ± SE). Grey columns: Rhodomonas sp., open columns: Phaeodactylum tricornutum, striped columns: Thalassiosira rotula CCMP1018. Ns Not significantly different from zero. Other abbreviations as in Table 1

Full size image
Fig. 5
figure 5

Temora longicornis. Weight-specific growth [μg C (μg C)−1 day−1] as a function of weight-specific ingestion of carbon and nitrogen [μg (μg C)−1 day−1], PUFAs LIN (18:3ω3), EPA (20:5ω3) and DHA (22:6ω3), total PUFA and Δ5 sterols [ng (μg C)−1 day−1] or PUA [fmol (μg C)−1 day−1], as well as the C:N ratio (μg C:μg N) of the diet (mean ± SE). Only species which induced significant ingestion rates were included. Closed symbols indicate monospecific diets, open symbols mixtures. The dotted line indicates a significant linear regression between growth and C:N ratio of the diet, obtained when the aldehyde containing diets were removed (y = 0.17 − 0.005x; r = 0.83, P < 0.05; see “Summary”)

Full size image

The nauplii gross growth efficiency with species where both growth and ingestion were significant varied between 0.11 (T. rotula CCMP1018) and 0.49 (T. rotula CCMP1647). The average gross growth efficiency on diatoms was 0.30 ± 0.15, which was similar to the gross growth efficiency on Rhodomonas sp. (0.23). However, the carbon concentration in Rhodomonas sp. experiments was 2–3 times higher than with diatoms (see “Materials and methods”), which likely resulted in over-saturated feeding conditions and thus a lower gross growth efficiency. A generally higher gross growth efficiency of Rhodomonas sp. was also suggested by mixture experiments: in mixtures where only Rhodomonas sp. was consumed (Rhodomonas sp. + P. tricornutum) the nauplii gross growth efficiency was high (>0.5), whereas when both Rhodomonas sp. and T. rotula CCMP1647 contributed to the total ingestion, the gross growth efficiency was substantially lower (0.17; Table 5).

Table 5 Temora longicornis: carbon gross growth efficiency of nauplii on Rhodomonas sp., on diatom diets and in mixtures
Full size table

Discussion

The diatom community in the North Sea during the peak development of T. longicornis nauplii includes species such as Chaetoceros spp., Thalassiosira spp., S. costatum, Rhizosolenia spp., Coscinodiscus spp. and Leptocylindricus spp., and tends to consist of several species (Riebesell 1991; Maar et al. 2002), with bloom concentrations ranging from ca 140 μg C l−1 (Kiørboe and Nielsen 1994) up to 800 μg C l−1 (Halsband and Hirche 2001). Our choice of species and food concentrations thus reflected the actual spring bloom conditions.

Temora longicornis was able to complete its development on 4 out of 11 diatom species tested (T. weissflogii, T. rotula CCMP1647, L. danicus and S. costatum), whereas the other 7 species proved to be inadequate food for nauplii development (T. rotula CCMP1018, two strains of T. pseudonana, all three Chaetoceros sp. and P. tricornutum). Four of the diets, which resulted in low development and high mortality, were not ingested by T. longicornis nauplii (all Chaetoceros species, T. pseudonana CCMP1010). Since the Chaetoceros species were both spiny and chain-forming, and the remaining species T. pseudonana was among the largest species offered, morphology and/or size were sufficient to explain the lacking ingestion. There was no indication that diatom derived PUAs such as 2,4,7-decatrienal would function as feeding deterrents (Jüttner 2005; Leising et al. 2005), but it rather seemed that copepod nauplii were feeding on diatoms if they were morphologically capable of feeding on them.

Of the ingested species, three induced low development and high mortality (T. rotula CCMP1018, T. pseudonana CCMP1335 and P. tricornutum), whereas the development was completed with four species (see above). The diatom species which are reported to have deleterious effects or to induce complete development are often the same throughout the literature: e.g., T. rotula has been reported to have both positive (Paffenhöfer and Harris 1976; Harris and Paffenhöfer 1976) and negative (Carotenuto et al. 2002; Poulet et al. 2003) effects on juvenile development, as has S. costatum [respectively, Verity and Smayda (1989) and Ianora et al. (2004) for positive and negative effects; Table 6]. It seems that the only intensively studied diatom species which consistently fails to support complete development is P. tricornutum, while with the other more frequently used diatoms the results are not consistent (Table 6), and variation even between replicate experiments of the same series is high (Carotenuto et al. 2002, this study). However, due to, e.g., its lack of absolute silica requirements and unusual Si uptake kinetics (Del Amo and Brzezinski 1999), P. tricornutum neither represents an average diatom, nor is a relevant species during the spring bloom (see, e.g., Riebesell 1991). This makes it of somewhat limited interest when considering copepod–diatom interactions.

Table 6 Summary of the literature on development, juvenile mortality and/or malformations of early naupliar stages of copepods fed either different monospecific diatom diets or diets dominated by diatoms (in situ and mesocosm experiments)
Full size table

We suggest that diatom effects on juvenile development and growth can be divided into four groups similarly to Ban et al. (1997): (a) species which induce fast development (high growth) and low mortality, (b) species which induce arrested development and high mortality, (c) species with which development is completed but mortality high, and (d) species which do not support complete development but promote low mortality. Since juvenile growth and mortality generally seem to be coupled (nauplii die if they can not moult into the next stage; Lopez 1991), a response following the first two groups is most common. A few exceptions have, however, been recorded: Calanus helgolandicus and T. longicornis were able to complete their development on S. costatum and T. rotula, respectively, although the mortality was high (Ianora et al. 2004; Carotenuto et al. 2002, respectively), while the development of T. longicornis and Pseudocalanus elongatus on nutrient-limited T. weissflogii was significantly slowed down, without any evident effect on mortality (Klein Breteler et al. 2005; Table 6). The diatoms in the present study fall in the first two categories, with the observed negative correlation between growth and mortality suggesting a coupling between these two processes. Our results therefore show that some diatoms are inadequate food for nauplii development, whereas with some species the development can be completed. With low ingestion excluded, we considered (1) diatom PUA production, (2) presence of other diatom toxins, (3) nutritional quality and (4) nutritional limitation following PUA production as possible reasons behind the observed response of copepod nauplii on diatoms diets.

Diatom toxicity

The potential adverse effects of diatoms result from certain polyunsaturated aldehydes (PUA), which are generated from PUFAs upon a cell disruption during grazing (Pohnert et al. 2002). Adding these PUAs to the maternal diet of Calanus helgolandicus induced arrested development and increased rates of nauplii mortality (Ianora et al. 2004). The poor development, high mortality and occurrence of birth defects of copepods associated with deleterious PUAs have also been observed with monospecific diets of T. rotula and S. costatum (Miralto et al. 1999; Pohnert et al. 2002; Ianora et al. 2004). Although the present study adds T. pseudonana (strain CCMP1335) to the list of “bad” diatom species, this strain did not produce any PUAs. Moreover, we did not find any evidence of an adverse effect of PUAs on growth or mortality of T. longicornis nauplii (Fig. 5). The growth and development with three diatoms which have a potential for PUA production was comparable to that on Rhodomonas sp. In contrast, two species, which were ingested but did not support a complete development, did not have a potential to produce any PUAs.

To check if the negative effect could be due to unknown toxins, we used the toxicity bioassay approach as suggested by Jónasdóttir et al. (1998) and further developed by Colin and Dam (2002b). In this toxicity bioassay the suspect species is mixed in different ratios with the control diet, and if the species has a toxic effect, the copepod response should fall below the reference line connecting growth rate (or egg production or hatching) at 0 and 100% of the control diet (Fig. 6a) or gross growth efficiency on different concentrations of the control diet (Fig. 6b). The two species chosen for the bioassay were P. tricornutum and T. rotula CCMP1018: P. tricornutum because it is supposed to be a potential producer of the unsaturated 12-oxo-dodeca-5,8,10-trienoic acid, which is shown to have adverse effects in sea urchin bioassays (Pohnert et al. 2002), and T. rotula CCMP1018 because of its elevated and potentially variable production of PUA at the onset of the development experiments. With neither of the approaches had P. tricornutum or T. rotula CCMP1018 a toxic effect. In Rhodomonas sp. + P. tricornutum mixture, P. tricornutum did not have any effect on growth, obviously due to the strong selection of nauplii for Rhodomonas sp. when the two species were mixed (Fig. 4). In contrast, T. longicornis nauplii benefited from T. rotula 1018 if it was mixed with Rhodomonas sp. (Fig. 6), suggesting either a beneficial effect of pure food quantity (increased ingestion) or possibly supplement of some nutritional component by the diatom. Thus, the results of the toxicity bioassay, together with the lacking correlation between ingestion of deleterious PUAs and growth or mortality, strongly indicated that other factors than the parameters connected to diatom toxicity are behind the failed development on certain diatom diets.

Fig. 6
figure 6

Temora longicornis. Weight-specific growth [μg C (μg C)−1 day−1] as a function of a percentage of control food Rhodomonas sp. (%) in the diet and b weight-specific ingestion [μg C (μg C)−1 day−1] of Rhodomonas sp. as a single species diet (mean ± SE). Closed symbols: Rhodomonas sp. + Thalassiosira rotula CCMP1018, open symbols: Rhodomonas sp. + Phaeodactylum tricornutum, grey symbols: Rhodomonas sp. as a monospecific diet. The lines connect the weight-specific growth in a 0 and 100% of Rhodomonas sp. or b in the Rhodomonas sp. ingestion at a concentration of ca 50 and 200 μg C l−1, thus indicating a growth rate which is only dependent on the Rhodomonas sp. concentration (a) or ingestion (b). The symbols on the line indicate that food species is neither toxic nor beneficial, below the line that the food species is toxic and above the line that the food species is beneficial (Jónasdóttir et al. 1998). In a the dotted line indicates Rh + Ph mixture, the solid line Rh + Tr mixture. The ingestion and weight-specific growth of Rhodomonas sp. at 50 μg C l−1 from M. Koski (unpublished data) and Koski et al. (2006), respectively

Full size image

Nutritional deficiencies

Alternatively, arrested development may be due to nutritional deficiencies. Jones and Flynn (2005) argued that single diatom diets are insufficient in terms of mineral nutrients. This is supported by the observations of the detrimental effects of N- or P-limited T. weissflogii on juvenile development, because in similar incubations, nutrient replete T. weissflogii induced high growth rates and low mortality (Koski et al. 1998; Klein Breteler et al. 2005). Klein Breteler et al. (2005) further showed changes in PUFAs and sterols of nutrient limited algae and suggested that these biochemical limitations, induced by mineral limitations, could be the reason behind the failed development on nutrient limited diatoms. According to a recent study of Wichard et al. (2007), also the production of PUAs can change the fatty acid content of diatoms during ingestion: a rapid depletion in PUFAs (especially EPA, the precursor of decatrienal and heptadienal) occurs during the production of PUAs, which suggests a PUFA limitation in diatoms which have a potential for PUA production. Evidence from the freshwater literature further indicates the role of diverse PUFAs in crustacean growth (Müller-Navarra et al. 2000), showing increased growth rates of Daphnia with supplementation of algal diet with single PUFAs (von Elert 2002).

Besides PUFAs, also Δ5 sterols are suggested to be essential for the growth of juvenile copepods (Klein Breteler et al. 1999, 2004, 2005). Similarly to PUFAs, sterols can not be synthesised de novo by crustaceans, and must thus be derived from the algal food (Klein Breteler et al. 2005 and references therein). A role of sterols in moulting of crustaceans (D’Abramo et al. 1984) further suggests sterols as limiting factors for juvenile growth, although direct evidence from supplement experiments is still missing.

However, our results did not show any straightforward relationship between food nutritional components and growth or mortality of T. longicornis nauplii, neither for C:N ratio nor for the ingestion of any of the essential PUFAs or the Δ5 sterols tested (Fig. 5). Similarly, since the development was arrested also in species which do not have a potential for PUA production, a PUFA limitation following from PUA production is unlikely. The development was typically arrested in the last naupliar–first copepodite stages. It appears that if the food species is not ingested or is directly toxic for copepods, the development ceases in the first feeding stages (see Huntley et al. 1987), while if the food species is nutritionally poor, the development tends to be arrested at metamorphosis (see, e.g., Klein Breteler et al. 1999; Koski et al. 2006). After passing the first feeding stage, the transition from nauplii to copepodites seems therefore to be a critical phase in nauplii growth (Lopez 1996). Carrillo et al. (2001) suggested that metamorphosis might need substantial amounts of phosphorus, while Epp and Lewis (1980) measured increased metabolic rate at the nauplii–copepodite transition. In contrast to freshwater cladocerans, copepod life-cycle may be too complicated to obtain a correlation between a single limiting nutritional factor and growth or mortality, but different life-stages may have different nutritional needs and limitations. Although growth in general might be limited by PUFAs, sterols or nitrogen, these correlations will disappear if specifically metamorphosis fails in the lack of, e.g., phosphorus, or is more food quantity limited due to increased metabolic demands.

Intra-experimental variability

Similar to Carotenuto et al. (2002), there was a large variability between replicate experiments when copepods were feeding on T. rotula strain CCMP1647 (Fig. 1a). Similar intra-experimental variability is also presented in Ianora et al. (2004) and results obtained in more frequently used algae (Table 6) seem to confirm the large variability in response of even the same copepod species on diverse diatom diets. Copepod–diatom interactions are both species and strain-specific (Pohnert et al. 2002), and changes in algae growth conditions can have large consequences for nutritional quality (Klein Breteler et al. 2005) and also change the PUA production (Ribalet et al. 2007). In addition, maternal diets have potentially a large effect on the development of the new generation, which has been shown thoroughly with diatom diets (Ianora et al. 2004; Table 6) and starvation (Poulet et al. 2003, 2007).

The possible explanations for the large variability between experiments could therefore be (1) use of different algae strains with different properties, (2) changes in algae condition affecting PUA production, nutritional quality or both or (3) maternal effects complicating the nauplii response to the diet. However, most studies have used exponentially growing algae, and conflicting effects occur even between replicate experiments with the same copepod species and algae strain. In the present study we used 11 different species and strains, always during an exponential growth state, and paid extra attention to maintain constant growth conditions of the algae. In addition, when possible, replicate samples were taken for algae mineral and biochemical composition during different time points of the experimental period (see “Materials and methods”; Table 1), to account for a potential variability in the algae condition. The effect of maternal condition was minimised by using copepods originating from a standard culture and by monitoring the development on a standard diet. Thus, besides potential changes in algae condition and maternal effects as complicating factors, additional factors seem to be involved in determining juvenile’s response on diatom diets.

Summary

Our results show that whereas some diatom species are inadequate food for T. longicornis nauplii, the development can be completed with several other diatom species, irrespective of their production of deleterious aldehydes. For some of the common diatom species, lacking ingestion was sufficient to explain the low development and growth. However, we failed to identify the reason behind copepod’s response on the ingested diatom diets, but neither toxicity nor nutritional quality (represented by C:N, PUFA and sterols) could directly explain the observed growth. It thus appears that either the nauplii–diatom interactions are affected by something else, such as, e.g., assimilation of nutritional compounds during the gut passage (Dutz et al. 2008), or that the nutritional needs of development stages are too complex to be described by a single nutritional compound.

Additionally, since different algal species can lack different nutritional elements (or contain different harmful substances), it is difficult to assess limitations by single compounds without direct supplement experiments. For instance, in the present study, removing the aldehyde containing diets (both single species and mixtures) would result in a significant negative correlation between growth and C:N ratio of diet (Spearman −0.899, n = 6, P < 0.05), suggesting nitrogen limitation of growth, although the relationship is driven by a single data point, P. tricornutum (Fig. 5). To estimate limitation by, for instance, PUFAs, one should thus remove both the aldehyde producers and the nitrogen-deplete diets, while PUA effects should ideally be studied after excluding the species deficient in PUFAs and/or nitrogen. However, removing PUFA- and N-deficient species would involve subjective estimates of the threshold levels of limitation, and would not remove the PUFA-deficiencies resulting from N-limitation (Klein Breteler et al. 2005) or PUA production (Wichard et al. 2007). Besides taking into account that specific limitations can occur during different stages of growth, the future studies aiming to identify a single limiting and/or harmful substance behind the diatom effect on growth should thus consider using single algal species, either manipulated by, e.g., nutrient stress, or supplemented by the missing nutritional element or harmful substance.

Whatever the reason behind the observed responses, our results suggest that a diatom spring bloom consisting mainly of species which nauplii are morphologically incapable of ingesting (Chaetoceros spp.), with only a low concentration of alternative species, will not support a complete development of T. longicornis, whereas a bloom dominated by, e.g., Thalassiosira spp. and Leptocylindricus spp. will. Although the egg production of T. longicornis seems to be dependent on the spring bloom (Kiørboe and Nielsen 1994; Peterson and Kimmerer 1994), its in situ mortality and growth/development rates are, however, relatively unknown, as are their effects on population dynamics. In general, similar negative effects of diatoms for cohort development, as reported for Calanus helgolandicus in the Adriatic Sea (Ianora et al. 2004), have not been described for T. longicornis. Instead, the few available measurements on the development and/or growth rate in situ seem to suggest fast development during the spring bloom (Bakker and van Rijswijk 1987; Peterson and Kimmerer 1994; M. Koski unpublished data).

The generally high in situ food diversity would suggest that the food quality of monospecific diatom diets would be of limited importance for the cohort development in nature. However, T. longicornis nauplii seem to have a limited capacity to utilise food mixtures (Koski et al. 2006) and the juvenile development is regularly limited by food quality and/or quantity on an annual basis (M. Koski unpublished data). Further, the few studies dealing with nauplii selectivity show a preference for diatoms and mainly size-selective grazing (Irigoien et al. 2003), suggesting that nauplii diet would mainly consist of the most available and largest cells (thus diatoms). It can therefore not be excluded that the nauplii–diatom interactions could influence population dynamics and, e.g., result in inter-annual variations in the copepod abundance, such as shown by Durbin et al. (2003) or Halsband-Lenk et al. (2004). Until further field studies have been conducted, the relevance of diatom effects for the annual dynamics of T. longicornis in nature, however, remains speculative.