Variability in amino acid composition of alpine crustacean zooplankton and its relationship with nitrogen-15 fractionation

Amino acids (AAs) are critical biochemical compounds for living organisms. Because of the limited capacity for their de novo synthesis in many animals, the nutritional value of food largely depends on its AA composition relative to the animal’s requirements. To improve present knowledge on AA variability in freshwater crustaceans, we studied the inter- and intraspeciﬁc variability in three contrasting species from an oligotrophic alpine lake ( Daphnia pulicaria , cladocera; Cyclops abyssorum , cyclopoid copepod and Diaptomus cyaneus , calanoid copepod). Inter-species differences were larger than intraspeciﬁc variation, conﬁrming a non-strict homeostasis in freshwater crustacean zooplankton. The intraspeciﬁc variability differed for each species: in Daphnia , it was mainly related with ontogenetic changes rather than reproduction; in Cyclops , both factors were equally important; and reproduction was the most relevant in Diaptomus . Reproduction changes were associated with serine and phenylalanine in the three species, while the AAs responsible for ontogenetic changes differed in each species. There were no gender differences in AA composition in any of the two copepod species. Free AAs formed a very low percentage of total AA pool ( , 2.7%). Taking advantage of the fact that Daphnia is the main prey for Cyclops in the lake studied, we further investigated to what extent the AA composition is related with Cyclops – Daphnia nitrogen stable isotope fractionation. Only those AAs that are both essential and are not trans -aminated during protein synthesis had a signiﬁcant correlation with nitrogen stable isotope fractionation, supporting the hypothesis that an AA imbalance can be responsible for a variable nitrogen stable isotope fractionation.

animals, the AA composition of food becomes very important in animal nutrition. This has been demonstrated in different types of terrestrial organisms such as butterflies (O'Brien et al., 2003), birds (Ramsay and Houston, 1998), farm animals (D'mello, 1994) and humans (Reeds, 2000), and also for aquatic organisms such as farmed fish (Conceição et al., 2003;Ronnestad et al., 2003), rotifers (Boëchat and Adrian, 2006), mussels (Kreeger et al., 1996) and marine copepods (Kleppel and Burkart, 1995;Guisande et al., 1999Guisande et al., , 2000. This knowledge contrasts with the few ecological studies focusing on the relative importance of AAs for animal growth and reproduction (Conklin and Provasoli, 1977;Boëchat and Adrian, 2006). Stoichiometry studies focusing on elements, mainly phosphorus, but also carbon and nitrogen (reviewed in Sterner and Elser, 2002) or fatty acids, mainly highly unsaturated fatty acids such as eicosapentanoic acid or docosahexanoic acid (e.g. Müller-Navarra et al., 2000), have received more attention than AAs. Since most of animal's nitrogen content is in the form of AAs, comparison between organisms and their food AA composition has been suggested as a step forward in using stoichiometric arguments for a better understanding of nitrogen limitation (Anderson et al., 2004). This suggestion is based on two main assumptions, namely that the consumer is unable to synthesize some AAs in significant quantities (i.e. essential AAs) and that it is homeostatic with respect to the composition in essential AAs. The former assumption is supported by numerous studies, including in crustaceans (Claybrook, 1983), whereas the latter has only been partially demonstrated and further studies are required (Cowgill et al., 1986;Anderson et al., 2004).
An imbalance between the AA composition of food and those of the consumer has been proposed as one of the main mechanisms explaining nitrogen stable isotope fractionation (i.e. change in isotope ratios, D 15 N) between diet and animal tissues (Martínez del Rio and Wolf, 2005). Nitrogen stable isotopes together with those of carbon have become a common tool for describing trophic relationships, although the physiological mechanism explaining D 15 N (crucial for estimating diet sources) is still not resolved (Martínez del Rio et al., 2009). The physiological mechanism could be caused by the deficiency in one or several essential AAs leading to greater protein intake and eventually to higher catabolism of the AAs in excess for protein synthesis favouring a preferential body elimination of lighter nitrogen (Martínez del Rio and Wolf, 2005). Few studies have provided evidence for this hypothesis with the exception of Robbins et al. (Robbins et al., 2005), in a comparative study, and Ventura and Catalan (Ventura and Catalan, 2008), in a field study, that reported a significant negative correlation between D 15 N and the protein quality of the diet (e.g. AA composition). An important unresolved question is if there are some particular AAs that are more relevant for D 15 N. The essential part of an AA (and what differentiates them biochemically) is their carbon skeleton. All the nitrogen of AAs is in the amino group, which can be trans-aminated by animals in, a priori, all AAs. However, it has been observed that some AAs appear to retain approximately the same nitrogen isotopic composition as their food source (are not trans-aminated; "source AAs"), whereas others become enriched in 15 N by the animals' metabolism (undergo trans-amination; "trophic AAs") (McClelland and Montoya, 2002;Popp et al., 2007). Different proportions of "trophic" and "source" AAs explain why animal tissues differ in 15 N (Martínez del Rio et al., 2009). Since the trans-amination is not related to the AA carbon skeleton, some trophic/source AAs are essential while others are nonessential, making a complicated physiological puzzle. What particular AAs or AA group is directly related to D 15 N has not been described and therefore needs to be explored more in depth. One may hypothesize that among the essential AAs, only those that are not trans-aminated would have a higher relevance than the others.
Zooplankton has played a central role in stoichiometry studies, particularly Daphnia, which has been extensively used as a model organism. Investigations examining the AA composition are scarce, however. Most past studies of the AA composition of zooplankton have focused on the variability in the total AA (TAA) composition of adult marine or freshwater zooplankton compared to their food, the results showing that adults have a very strict AA composition (Cowey and Corner, 1963;Cowgill et al., 1986;Guisande et al., 2000). All those studies are indicative of AA composition homeostasis in adults of crustaceans. In contrast, there are few studies focusing on the TAA intraspecific variability. Significant differences in AA composition during development have been observed in Artemia spp. and lobsters (Helland et al., 2000;Mente et al., 2001). In cladocerans and copepods, Dabrowski and Rusiecki (Dabrowski and Rusiecki, 1983) found differences in TAA between late developmental stages of different species, although they did not examine the degree of these TAA changes for each species. Changes in TAA were reported for two Calanus species during development from egg to the second naupliar stage (Laabir et al., 1999) and between nauplii of different seasons (Helland et al., 2003a). Ontogenetic variability in species of cladocerans, calanoid and cyclopoid copepods were studied in different coastal lagoons and wetlands by Brucet et al. (Brucet et al., 2005). They analysed all the lake-populations simultaneously, and found a more pronounced variability in the TAA composition related with ontogeny in copepods, while Daphnia variability was mostly related with lake type and not with ontogeny. However, the relative importance of ontogeny, reproduction and gender in AA composition changes has not been evaluated for these animal groups.
In this context, with the aim of testing the hypothesis that interspecific differences among the main freshwater groups of crustacean zooplantkton are greater than their intraspecific variation, we studied the relative proportions of the AA pools and its variation in crustacean zooplankton species of an alpine lake throughout a whole seasonal cycle. We selected a lake that had one species of each of the main freshwater crustacean groups, a cladoceran (Daphnia pulicaria Forbes), a cyclopoid copepod (Cyclops abyssorum Sars) and a calanoid copepod (Diaptomus cyaneus Gurney). In a previous study, these three species were found to have substantial changes in their elemental content of carbon and phosphorus during their life cycles associated with reproduction (Ventura and Catalan, 2005). In contrast, nitrogen was the only compound which did not change significantly. In addition, Cyclops was described as being a predatory copepod feeding primarily on Daphnia, and the D 15 N of both species was strongly correlated with their protein quality, which was interpreted as evidence of consumer AA (nitrogen) limitation (Ventura and Catalan, 2008). Therefore, these two species also present an ideal framework for investigating the specific role of AAs in D 15 N. Contrasting with Cyclops, the other two species, Diaptomus and Daphnia, are more generalistic consumers (an omnivore and herbivore, respectively), and therefore they are not suitable for this particular comparison. The specific hypotheses driving the study were (i) interspecific variability of AA composition is greater than intraspecific AA variability; (ii) reproduction, gender and ontogenetic changes are important intraspecific sources of AA variability; and (iii) essential AAs that are not trans-aminated during tissue incorporation are those most closely related with the variability in D 15 N between Cyclops and Daphnia.

Study site and sample collection
Lake Redon (formerly lake Redó) is an oligotrophic glacial cirque lake located at 2240 m a.s.l. in the central Pyrenees (42838 0 N, 0846 0 E). It is relatively large (24 ha) and it is one of the deepest lakes (73 m) in the Pyrenees. It is dimictic, covered by ice during half of the year, usually from mid-December until late May or the beginning of June. The lake has average microbial biomass ratios of 10:2:2:1 for phytoplankton:bacteria:heterotrophic nanoflagelates:ciliates during the ice-free period . Rotifers are usually less abundant than crustaceans (the ratio between Daphnia and rotifer biomass is approximately 10:1 during the ice-free period; Camarero et al., 1999).
The lake was sampled on 14 occasions from December 1998, just after the lake was completely ice covered, to December 1999, after it was ice-covered again. Samples were collected at the deepest point of the lake either after drilling the ice-cover or from a platform anchored throughout the summer at the same point. Zooplankton samples were collected by vertical hauls from 65 m to the surface with a 200 mm net. Individuals were kept alive and transported cold (48C) until they were frozen (2208C) in the laboratory a few hours after collection.
The three crustacean zooplankton species studied had contrasting life histories, although all produced a single cohort per year. A complete description of their life cycles during the study period can be found in Ventura and Catalan (Ventura and Catalan, 2005) and their trophic relationships in Ventura and Catalan (Ventura and Catalan, 2008). Cyclops is a predatory copepod feeding primarily on Daphnia in this lake (Ventura and Catalan, 2008). Its life cycle was synchronized with that of Daphnia being the only two planktonic crustacean species inhabiting the lake during the whole year. Adult females of Daphnia from the previous summer survived below the ice-cover, reproduced during the end of the ice-covered period and the first 2 months of the ice-free period, and disappeared from the lake afterwards, leaving the juveniles growing up during the summer ( Fig. 1) (Ventura and Catalan, 2005). Adult females (termed "first cohort females" in our study) were therefore clearly distinguishable from the juveniles born during the ice-free period (termed "second cohort females"). We therefore compared the AA composition of the two cohorts, of the reproducing and non-reproducing first cohort females, and their eggs. Similar to Daphnia, adult females of Cyclops survived below the ice-cover in order to reproduce during the end of the ice-covered period. Adult males were present in the lake from December to February below the ice-cover. Nauplii appeared the following summer, just after the ice-cover melted (copepods have six naupliar, and five copepodite stages before adults develop) and coincided with the spring production maximum. Copepodites CI -CIII dominated Cyclops biomass during July and August, and copepodites CIV and CV were present in the lake from September until November ( Fig. 1) (Ventura and Catalan, 2005). Therefore, we sampled two generations and the stages analysed consisted of adult males, adult reproductive and non-reproductive females, and four copepodites (CII -CV), which were amalgamated in two groups, one with CII and CIII and the other with CIV and CV. The other copepod, Diaptomus, dominated the plankton in July and August, spending the rest of the year as resting eggs ( Fig. 1) (Ventura and Catalan, 2005). Due to its fast development, only adult Diaptomus were considered in this study. Unlike Cyclops, adult males and females were present in the lake during the same period. The July and August samples coincided with the beginning and the end of the reproduction period and were therefore expected to cover the maximum range of the AA variability for the adults of the species.

Sample preparation and chemical analysis
After thawing the samples, from 10 to several hundreds of individuals, depending on the weight of each species and stage, were quickly sorted under a dissecting microscope and were either placed into pre-weighed tin capsules for stable isotope analysis or in Teflon capsules for the analysis of AAs. From two to six combined sample, replicates were analysed for each species and stage for each sampling date, with the exception of a few cases where sample material did not suffice. Individuals were kept cold (,48C) during the sorting process, which was always carried out within a few hours of collection. Dry weights were determined for all samples after drying at 608C for 24 h and weighing on a high precision microbalance (Ohaus Analytical Plus, AP250D-0).
For AA analysis, between 0.6 and 0.9 mg of dried sample was vacuum-sealed and hydrolysed with HCl 6N at 1128C for 16 h (Fountoulakis and Lahm, 1998). An internal norleucine standard was included in every sample prior to hydrolysis in order to minimize among sample variability. Samples were subsequently dried under vacuum at room temperature. Dried samples were re-dissolved with 0.5 to 1 mL (depending on the concentration of the sample) of buffer solution ( pH 2.2) prior to analysis and purified by filtration (0.65-mm Durapore filter; Millipore, Bedford, MA, USA). The AA analysis was performed on a Biochrom20 (Amersham-Pharmacia) ion-exchange AA auto analyser following the ninhidrine method of Spackman et al. (Spackman et al., 1958). A standard solution of 20 AAs was run for every 10 samples. The detection limit was 10 pmol for all AAs and 50 pmol for proline. The samples injected were always greater than the limit of detection for all AAs considered. Analytical precision had a CV of 3.9%. Tryptophan, due to its degradation under acid hydrolysis (Fountoulakis and Lahm, 1998), has not been considered in this study. Cysteine and methionine are other AAs that usually degrade easily during the hydrolysis process. Together with taurine, they are the three AAs containing sulphur. Therefore, their possible degradation was checked by analysing for elemental sulphur (EA 1108 CHNS-O Carlo Erba Instruments elemental analyser). This analysis showed that the AA sulphur accounted for most of the elemental sulphur indicating a good recovery of most sulphurcontaining AAs. Also due to the acid hydrolysis, Fig. 1. Seasonal changes in water temperature, chlorophyll a, the biomass of the three crustacean planktonic species of Lake Redon (Cyclops abyssorum, Daphnia pulicaria and Diaptomus cyaneus), the relative contribution of Cyclops abyssorum small copepodites (I -III), large copepodites (IV and V) and adults of the total C. abyssorum biomass, and the relative contribution of Daphnia pulicaria ovigerous and non-ovigerous females (first cohort females) and juveniles (second cohort females) to the total D. pulcaria biomass. Daphnia juveniles were those not reaching sexual maturity, which in this year corresponded to a size lower than 2 mm. For temperature and chlorophyll a, filled circles are the water column average values; dotted lines follow maximum and minimum values recorded on each sampling occasion. asparagine and glutamine were measured together with aspartic acid and glutamic acid, respectively. AAs are expressed as relative mass percentage of TAA throughout this paper. These units were chosen rather than absolute concentrations (e.g. mg AA/mg C or DW) since a description of the biochemical variability of the present species and study period showed that there was an important seasonal variability in the lipid and carbohydrate content (up to 50% of their DW) (Ventura and Catalan, 2005), which were much greater than the changes in the AA composition. Using the relative units, we focus on the variability in the AA composition rather than the factors affecting the gross biochemical composition. However, we provide the TAA concentration (in % dry weight) which allows for calculation of the absolute concentration also used in the literature.
In order to determine the relative importance of the free AA (FAA) pool with respect to proteins, we used the concentration of taurine as an indicator of the FAA pool. Taurine is the only AA which does not form proteins, and one of the main FAAs in crustaceans (Claybrook, 1983). A bibliography search of joint measurements of taurine and the concentration of FAAs was obtained from Ventura (Ventura, 2006), and these were found to relate linearly (Fig. 2). The good fit (r 2 ¼ 0.90, P , 0.001) suggested that taurine concentration could be used as a predictor of the FAA pool. Therefore, we estimated the relative proportion of the FAA by applying the regression formula of Fig. 2 to the measured taurine concentration.
Dried samples for d 13 C and d 15 N stable isotope analysis were analysed in a Delta C Finningan MAT mass spectrometer. Results are reported relative to atmospheric nitrogen and PeeDee belemnite carbonate as references. Reproducibility was better than 0.1‰ and 0.3‰ for d 13 C and d 15 N, respectively. Further methodological details are provided in Ventura and Catalan (Ventura and Catalan, 2008).

Statistical analysis
To test whether the different species or the stages within each species could be distinguished based on their AA composition, we performed multiple discriminant analyses (MDA). The MDAs enabled us to determine which AAs were the most significant in discriminating between two or more groups (species and stages). We performed a first MDA (MDA1) with the various stages and other sources of intraspecific variation of the three different species as separate factors in one single analysis in order to find out if the interspecific variability was greater than the intraspecific variability. The sources of intraspecific variation of the AA composition evaluated consisted of gender, development stages (ontogenetic changes), differences between reproductive and nonreproductive females in the three species and also males in Diaptomus. Gender differences were only considered in the two copepod species, since Daphnia mainly reproduced asexually. Ontogenetic changes could only be tested in Daphnia and Cyclops due to the short life cycle of Diaptomus. Each replicate was entered separately in the analysis. The variable number of samples for each species or stage was due to the natural variability of their different seasonal occurrences. A different MDA was performed for each species (MDA2 for Daphnia, MDA3 for Cyclops and MDA4 for Diaptomus) with the above stages as discriminating groups in the MDA in order to test if they could also be distinguished in terms of their AA composition and to find out which particular AAs were involved with the different developmental stages of each species. In the specific case of Diaptomus due to the lower number of samples (eight cases for four variables) present, we ran a forward selection procedure to select only the combination of AAs best discriminating among the different groups. MDA analysis is a method previously applied in AA comparisons of zooplankton (Guisande et al., 1999(Guisande et al., , 2000(Guisande et al., , 2003Boëchat and Adrian, 2005;Brucet et al., 2005) and therefore its application facilitates comparison with previous studies.
Average Euclidean distance (D jk ) was used to test if the interspecific variability of AA composition was greater than the intraspecific variability and to estimate the dissimilarity between the AA compositions of Cyclops and Daphnia for comparison with the nitrogen stable isotope composition. We compared Cyclops with Daphnia  (Cowey and Corner, 1963;Suyama et al., 1965;Jeffries, 1969;Jeffries and Alzara, 1970;Srinivasagam et al., 1971). DW stands for dry weight. since the former was described as being a predatory copepod feeding primarily on Daphnia in the same lake and study period (Ventura and Catalan, 2008). The Euclidean distance is a metric distance given by the Pythagorean formula (Legendre and Legendre, 1998): where X ij and X ik are the percentages of the AA i of species 1 ( j) or species 2 (k), and n is the number of AAs considered. The distance was calculated with all the AAs when evaluating interspecific versus intraspecific AA variability. For comparison of the distance between Cyclops and Daphnia, we calculated the distance of all AAs or separately for the essential AAs (lysine, phenylalanine þ tyrosine, leucine, valine, threonine, isoleucine, histidine, cysteine þ methionine), non-essential AAs (glutamic acid, aspartic acid, alanine, arginine, glycine, proline and serine), trophic AAs (alanine, asparagine þ asparctic acid, glutamine þ glutamic acid, isoleucine, leucine, proline and valine) and source AAs (arginine, glycine, lysine, serine, threonine and phenilalanine þ tyrosine). The average Euclidean distance between Cyclops and Daphnia using the different AA groups was compared with the D 15 N of both species with linear regression analysis.
In order to determine which particular AAs could be most imbalanced in Daphnia in comparison with the predator, Cyclops, we calculated the AA imbalance as follows (Conceição et al., 2003): where diet AA and consumer AA are the contributions of a given AA (i) to the AA profiles of the consumer.
Differences in individual AAs were tested with one-way ANOVA analysis (Underwood, 1997) with species or stage as independent variable and the percentage of the AAs as the dependent variable. AA percentage data were previously normalized with arcsine root square transformation when required, and a test of homogeneity of variances was performed before ANOVA. To test if Euclidean distances of interspecies versus intraspecies comparison were significantly different, we used one-way ANOVA with the Euclidean distance between all the pairs of samples analysed as dependent variable and if the comparison was from among species comparison (three cases) or within species (three more cases) as independent variables. We used Tukey's post hoc test for detecting significant differences among or within species. All statistical analyses were performed with SPSS V17.
The average Euclidean interspecific distance was two times greater than the average intraspecific distance (3.7 + 1.1 and 1.7 + 0.6, respectively; F 5,1080 ¼ 706, P , 0.001; Tukey's P , 0.001; Fig. 3). The two copepods had a more similar composition to each other than to the cladoceran, while the three species had a similar intraspecific distance (Fig. 3).
The AA composition of the three species studied was species specific (Fig. 4A). The most abundant AAs for the two copepods (in order of importance and not taking into account asparagine/aspartic acid and glutamine/glutamic acid which were analysed together) were alanine, leucine, tyrosine, lysine and arginine, which altogether accounted for 37% of the TAA pool. For the cladoceran, the most abundant AAs were leucine, arginine, alanine, valine, threonine and lysine, accounting for 40% of the TAA pool (Table II). The MDA1, including all the stages of the three species, was able to classify the stages of the three species based on their AA composition (100% of the samples being correctly classified with the exception of half of the Diaptomus non-reproductive females that were identified as males; Fig. 4A and Table III). The first axis of the MDA1 explained 47% of the variance and mainly segregated the two copepods from the cladoceran. The AAs significantly correlated with this first axis were proline, serine and threonine, their relative concentrations being higher in Daphnia, and glycine and taurine, with higher relative concentrations in the two copepods (Table III). The second axis explained 39% of the variance and segregated Diaptomus from the other two species. Diaptomus had a higher alanine and serine relative concentration, while Cyclops and Daphnia had higher glycine concentration.

Intraspecific variation
Within each species, there were no differences in the relative proportion of TAA, protein AA (PAA), FAA, essential AA and nonessential AA pools between most of the different stages, with the exception of Cyclops males that had a lower relative concentration than the females in the lake during the same period (2.9 + 0.1 and 1.6 + 0.5% dry weight of TAA for females and males, respectively; F 1,3 ¼ 15.8, P ¼ 0.028) and Daphnia eggs that had higher TAA relative concentration than the first cohort females (F 1,8 ¼ 2.8, P ¼ 0.1 Tukey's post hoc test P ¼ 0.013) ( Table I). In contrast, there was a decrease from non-reproducing females to reproducing females of FAA and essential AA pools of Cyclops (F 1,9 ¼ 20.5, P ¼ 0.001 and F 1,9 ¼ 7.2, P ¼ 0.025 respectively), a slight decrease in the females of Daphnia (F 1,8 ¼ 2.8, P ¼ 0.1 and F 1,8 ¼ 3.3, P ¼ 0.1 respectively) and in both males and females of Diaptomus (F 1,6 ¼ 3.8, P ¼ 0.1 and F 1,6 ¼ 2.8, P ¼ 0.1, respectively).
The MDA2 correctly classified 100% of the samples of Daphnia (Fig. 4B and Table III). The first axes summarized 74.9% of the variance and separated the three stages analysed: first cohort females, second cohort females and eggs. The AAs most strongly associated with the axes were proline and arginine, more abundant in the eggs ( positive) side, and glycine and leucine, more abundant in second generation females (negative) side. The second axis explained the remaining 22.4% of the variation and separated the second generation females and eggs (negative) side from the reproductive and non-reproductive first generation females ( positive) side. The AAs most strongly associated with the axes were positively, threonine and glutamine/glutamic acid and negatively, arginine. Further comparison of individual AAs between reproducing and non-reproducing first cohort females with one-way ANOVAs indicated that reproducing females had higher histidine, serine and cystine (P ¼ 0.003, P ¼ 0.003 and P ¼ 0.007, respectively) and lower phenylalanine (P ¼ 0.021).
For Cyclops, results of the MDA3 analysis showed that 100% of the cases were correctly classified (Table III) (Fig. 4C). The first axes explained 78.2% of the variance and mainly segregated the CIV-V copepodites, the Fig. 3. Average Euclidean distance (with 10th and 25th percentiles) of the interspecific (shaded box plots) and intraspecific (white box plots) distance in the amino acid composition. Distances were computed among all samples analysed and therefore the intraspecific distances include both the distances between stages and within stages. Abbreviations stand for Cyclops abyssorum (C.a.), Daphnia pulicaria (D.p.) and Diaptomus cyaneus (D.c.). Letters above each box-plot denote significantly different distances from the one-way ANOVA Tukey's post hoc test (P , 0.001). Total AA (TAA) concentration is expressed as mass percentage of dry weight (DW) and is subdivided in the protein AA pool (PAA) and the free amino pool (FAA) that are expressed as the mass percentage of TAA. The relative proportions of essential and nonessential AA pools (EAA and NEAA, respectively) are also calculated. Letters after SDs denote ANOVA statistically significant stages (a, d) or reproducing versus non-reproducing females (b, c) at P , 0.05 or *P , 0.1.
reproductive females and males from the nonreproductive females and CII -III copepodites. The AAs most significantly associated with the first axis were glutamine/glutamic acid and phenylalanine in the nonreproductive females, males and copepodites CII -III ( positive) side and serine at the negative side. The second axis explained 12.9% of the variance and mainly separated copepodites CII -III from the rest, leucine and serine being correlated with copepodites CII -III and valine and glutamine/glutamic acid with the rest (Table III). Further comparison of individual AAs between males and females during the months they coexisted in the lake (January and February) with one-way ANOVAs confirmed that there were no statistical differences for any of the 18 AAs analysed and thus no gender differences in Cyclops in terms of their AA composition.
The MDA4 correctly classified all the samples of Diaptomus in one single significant axis ( Fig. 4D and Table III), showing that almost all the variability was associated with reproduction, while only negligible differences were related to gender. The AAs that were selected by the forward selection procedure and that significantly explained this variation were asparagine and phenylalanine that were associated with the nonreproducing adults and threonine and serine with the reproducing adults (Table III).
In order to find out if a particular essential AA might be related to D 15 N, we calculated the relative imbalance between Cyclops and Daphnia. Those AAs with a higher weight in explaining D 15 N Cyclops -Daphnia variance were expected to have (i) a negative imbalance in the consumer with respect to the diet and (ii) D 15 N Cyclops -Daphnia should increase with decreasing relative concentration of this individual AA in the diet. Among the different essential AAs, only lysine had an average negative imbalance between both species (Fig. 6). A decrease in D 15 N Cyclops -Daphnia with the relative concentration of each individual essential AA in Daphnia was analysed with Pearson correlation (Table IV). Among the eight essential AAs analysed, only the essential and source AA lysine had a significant negative correlation.

Interspecific variability
Our results confirm the hypothesis that interspecific variation in TAA composition was more pronounced than intraspecific variation. The overall AA interspecific distance was twice that of intraspecific distance (Fig. 3). In addition, although almost all samples were successfully classified to their corresponding stages for each species by the MDA1, the two first axes of the MDA1 basically distinguished among the three species (Fig. 4A). These results complement a previous extensive study focusing only on interspecific variation of alpine lake zooplankton in which most (85%) of the adult females of 13 species were successfully classified to their respective species (Guisande et al., 2003). Therefore, it can be concluded that alpine freshwater zooplankton have a non-strict homeostasis (Sterner and Elser, 2002) in their AA composition. This also implies that if there are predatory interactions among them, it is likely that predators may encounter dietary imbalances in their requirements. While phosphorus limitation in freshwater primary consumers has been documented frequently (Hessen, 1992), predators have been reported to face frequently nitrogen limitation (White, 1993;Anderson et al., 2004). We suggest that in Missing values for a given AA are those excluded from the analysis of the tolerance test. A graphical representation can be found in Fig. 4.
JOURNAL OF PLANKTON RESEARCH j VOLUME 32 j NUMBER 11 j PAGES 1583-1597 j 2010 these organisms, AAs might be a dietary limiting biochemical more often than is commonly assumed.

Intraspecific variability
The degree of intraspecific variability was comparable in the three species (average Euclidean distance within each species of Fig. 3 and dispersal of the points in the MDA1 of Fig. 4A), although the relative importance of ontogeny, reproduction and gender was not the same for each species. Cyclops had a similar AA composition variability related with ontogeny and reproduction, while Daphnia had a higher variability related with ontogeny than reproduction. In the only previous study of ontogenetic intraspecific AA variability of both Fig. 5. Comparison between (A) the Cyclops abyssorum-Daphnia pulicaria nitrogen isotopic fractionation (D 15 N) and the trophic amino acids average Euclidean distance, (B) the source AAs average Euclidean distance, (C) the essential AAs average Euclidean distance, (D) the nonessential AAs average Euclidean distance, (E) the trophic and essential AAs average Euclidean distance, (F) the source and essential AAs average Euclidean distance, (G) the trophic and non-essential AAs average Euclidean distance and (H) the source and nonessential AAs average Euclidean distance. The different AAs belonging to each group are described in Table II. copepods and cladocerans, Brucet et al. (Brucet et al., 2005) also found an important effect of ontogeny in copepods, although they found little ontogenetic variability in Daphnia. In these latter species, all the variability was associated with differences among populations belonging to different lakes. Our results show that when studying the intraspecific variability in one particular lake, the ontogenetic AA variability of copepods and cladocerans can be of comparable magnitude. Furthermore, we found different AAs responsible for the AA variation in each species ( proline, arginine, glycine and leucine in Cyclops; glutamine/glutamic acid, phenylalanine and serine in Daphnia). In other species of copepods, Brucet et al. (Brucet et al., 2005) found valine, tyrosine, proline and alanine as the AA with a higher weight in the ontogenetic variability of other copepod species. These results altogether suggest that the AAs involved in the ontogenetic variability of crustaceans are species specific. Although in this study we have not analysed the AA composition of nauplii, previous studies have shown they are the stage most similar to adults in terms of their composition (Brucet et al., 2005), and that they are very restrictive in their AA composition, not developing further if maternal diet is deficient (Laabir et al., 1999). Therefore, our results of lower intra-versus interspecific AA variability would most likely have been the same if the naupliar stage of copepods had been included.
Various studies have described the negative effects of an AA imbalanced diet for reproductive success in copepods (Cowgill et al., 1986;Guisande et al., 1999Guisande et al., , 2000Laabir et al., 1999;Lacoste et al., 2001;Helland et al., 2003b). In this study we show that, in addition, adult female copepods and cladocerans, and males of Diaptomus change their AA composition with reproduction. Contrary to the ontogenetic effects, two of the AAs involved with the changes related with reproduction were the same in the three species (the essential AA phenylalanine and the nonessential serine). Phenylalanine is an aromatic AA which acts as a precursor of tyrosine and other compounds involved in Fig. 6. Average relative essential amino acid (AA) imbalance between Cyclops abyssorum and Daphnia pulicaria during a seasonal cycle. See text for calculation procedure. Results are expressed in dry weight percentage of the total AA composition. Error bars are standard deviations. White and black bars are source and trophic AAs, respectively. Grey bars are those amino acids that have not been described to belong to any of the two categories. Asterisks denote significant negative correlation between the C. abyssorum-D. pulicaria nitrogen isotopic fractionation and the AA composition of D. pulicaria as described in Table IV  important physiological functions, while serine is involved in lipid biosynthesis and also a precursor of melanine. The results of this study suggest that they are relevant during the reproduction of crustaceans. In addition to the changes in individual AAs, we also found a tendency for the FAA content to decrease from non-reproductive to reproductive females. This agrees with other studies showing that eggs contain a larger proportion of FAAs and that the AA composition of eggs is decisive for the successful development of the embryos (Guisande et al., 1999;Laabir et al., 1999;Helland et al., 2003a, b). Contrasting with the ontogenetic and reproduction effects, we found no changes in the AA composition associated with gender in the two copepod species. The only difference attributed to gender was a higher proportion of the FAA pool in Cyclops females compared with males. This contrasts with the equal proportion between both sexes in Diaptomus, which might reflect the different reproductive strategies described for freshwater calanoids and cyclopoids. This result also agrees with the findings of a parallel study (Ventura and Catalan, 2005), in which Diaptomus males were found to follow similar elemental composition changes with reproduction as females, while the elemental content of Cyclops males did not change with time. This was attributed to the requirement of Diaptomus to mate every egg clutch, while Cyclops females can produce multiple clutches of eggs with only one impregnation. Therefore, the present results on the changes in the FAA pool also support the hypothesis that the different reproductive strategies of copepods have a relevant effect on the elemental and biochemical homeostasis of males.

AAs involved in D 15 N
There are different physiological factors that might influence D 15 N, including those related with fasting, growth and diet quality in terms of protein quantity or quality (reviewed by Martínez del Rio and Wolf, 2005). While most of these factors have been demonstrated either experimentally or through field studies, the effects of varying AA composition have scarcely been studied (Martínez del Rio et al., 2009). In a previous study, we showed that the carbon isotopic variation in Cyclops juveniles could be explained by fitting an isotopic growth model based on feeding entirely on Daphnia. However, this was not the case for nitrogen isotopic variability. Cyclops nitrogen isotopic composition variation and the Cyclops to Daphnia D 15 N were closely correlated to the dissimilarity in the AA composition between the two species (Ventura and Catalan, 2008).
In this study, we show that among the different AAs, only those that are both essential and source AAs had a high and significant correlation with D 15 N. Among these, only lysine had a negative imbalance in the consumer as it increased its relative concentration in the diet when the D 15 N decreased. This suggests that the imbalance of only one AA can be responsible for a high metabolic turnover, which can ultimately be responsible for a higher D 15 N. Our results therefore are in agreement with the hypothesis that an AA deficiency can be responsible for a variable D 15 N (Martínez del Rio and Wolf, 2005).
Although the recognition of a variable D 15 N might seem a difficulty for a widespread use of 15 N as a trophic position estimator, it opens new perspectives for its interpretation, since it could be used as an indicator of the relative AA imbalance of an animal and as a consequence, an indicator of the relative degree of AA limitation and as a result of the overall nitrogen limitation. Although this might seem difficult to apply to general field studies, future studies using individual AA stable isotope composition of carbon or nitrogen (e.g. Larsen et al., 2009) might substantially help interpretations in stable isotope trophic ecology.