The stable isotopic composition of Daphnia ephippia reﬂects changes in δ 13 C and δ 18 O values of food and water

. The stable isotopic composition of fossil resting eggs (ephippia) of Daphnia spp. is being used to reconstruct past environmental conditions in lake ecosystems. However, the underlying assumption that the stable isotopic composition of the ephippia reﬂects the stable isotopic composition of the parent Daphnia , of their diet and of the environmental water have yet to be conﬁrmed in a controlled experimental setting.

Abstract. The stable isotopic composition of fossil resting eggs (ephippia) of Daphnia spp. is being used to reconstruct past environmental conditions in lake ecosystems. However, the underlying assumption that the stable isotopic composition of the ephippia reflects the stable isotopic composition of the parent Daphnia, of their diet and of the environmental water have yet to be confirmed in a controlled experimental setting. We performed experiments with Daphnia pulicaria cultures, which included a control treatment conducted at 12 • C in filtered lake water and with a diet of fresh algae and three treatments in which we manipulated the stable carbon isotopic composition (δ 13 C value) of the algae, stable oxygen isotopic composition (δ 18 O value) of the water and the water temperature, respectively. The stable nitrogen isotopic composition (δ 15 N value) of the algae was similar for all treatments. At 12 • C, differences in algal δ 13 C values and in δ 18 O values of water were reflected in those of Daphnia. The differences between ephippia and Daphnia stable isotope ratios were similar in the different treatments (δ 13 C: +0.2 ± 0.4 ‰ (standard deviation); δ 15 N: −1.6 ± 0.4 ‰; δ 18 O: −0.9 ± 0.4 ‰), indicating that changes in dietary δ 13 C values and in δ 18 O values of water are passed on to these fossilizing structures. A higher water temperature (20 • C) resulted in lower δ 13 C values in Daphnia and ephippia than in the other treatments with the same food source and in a minor change in the difference between δ 13 C values of ephippia and Daphnia (to −1.3 ± 0.3 ‰). This may have been due to microbial processes or increased algal respiration rates in the experimental containers, which may not affect Daphnia in natural environments. There was no significant difference in the offset between δ 18 O and δ 15 N values of ephippia and Daphnia between the 12 and 20 • C treatments, but the δ 18 O values of Daphnia and ephippia were on average 1.2 ‰ lower at 20 • C than at 12 • C. We conclude that the stable isotopic composition of Daphnia ephippia provides information on that of the parent Daphnia and of the food and water they were exposed to, with small offsets between Daphnia and ephippia relative to variations in Daphnia stable isotopic composition reported from downcore studies. However, our experiments also indicate that temperature may have a minor influence on the δ 13 C, δ 15 N and δ 18 O values of Daphnia body tissue and ephippia. This aspect deserves attention in further controlled experiments. autotrophic sources of organic matter at the base of a food web (DeNiro and Epstein, 1978;Vander Zanden and Rasmussen, 1999;McCutchan et al., 2003). Likewise, stable nitrogen isotope ratios (expressed as δ 15 N values) can be used to estimate the trophic position of consumers in food webs (DeNiro and Epstein, 1981;Minagawa and Wada, 1984), and stable oxygen isotope ratios (expressed as δ 18 O values) have been found to reflect those of the water in the environment that organisms live in (Hobson, 2008;Soto et al., 2013).
Approaches are continuing to be developed that apply stable isotope ratio analysis to chitinous remains of aquatic invertebrates preserved in lake sediments (Heiri et al., 2012;Leng and Henderson, 2013). For example, the δ 13 C values of the fossil head capsules of benthic larvae of non-biting midges (Chironomidae) and δ 13 C values of the remains of water fleas of the genus Daphnia (Cladocera) have been used to investigate past changes in carbon cycling and energy pathways in lake food webs (Perga, 2011;Wooller et al., 2012;van Hardenbroek et al., 2013;Belle et al., 2014;Frossard et al., 2014). The δ 15 N values of chironomid head capsules and of Daphnia resting eggs (ephippia) have also been examined to investigate changes in nitrogen sources in an arctic lake (Griffiths et al., 2010). Past variations in lake water δ 18 O values have been reconstructed by analyzing the δ 18 O values of fossil chironomid head capsules (Wooller et al., 2004;Verbruggen et al., 2010b), and a correspondence has been found between δ 18 O values of lake water and of chironomid head capsules and Daphnia ephippia buried in surface sediments (Verbruggen et al., 2011).
Daphnia can occur in high abundances and often dominate the zooplankton community in lakes (Lampert, 2011). Being first-order consumers of algae, bacteria and detritus (Geller and Müller, 1981;Gophen and Geller, 1984;Kamjunke et al., 1999;Lampert, 2011), they form an important link between primary production and the higher orders of the pelagic food web. This makes Daphnia particularly suited for ecological investigations of freshwater ecosystems and food webs using stable isotopes. While Daphnia usually reproduce parthenogenetically, they may also reproduce sexually. Environmental cues such as food availability, photoperiod and population density (Kleiven et al., 1992;Cáceres and Tessier, 2004) may trigger sexual reproduction, upon which eggs are formed enclosed by rigid sheaths (ephippia). The chitinous ephippia are found abundantly in a wide range of lake sediment types and remain well preserved in sediments hundreds to thousands of years old (Szeroczyńska and Samarja-Korjonen, 2007). Since the chemical composition of chitinous invertebrate remains stays largely unchanged even in fossils more than 10 000 years old (Miller et al., 1993;Verbruggen et al., 2010a), they are believed to retain their isotopic composition after deposition (Heiri et al., 2012). Therefore, ephippia may provide material for reconstructing the past stable isotopic composition of Daphnia in lakes and, consequently, for investigating past conditions in aquatic food webs (e.g., Wooller et al., 2012;van Hardenbroek et al., 2013van Hardenbroek et al., , 2014Schilder et al., 2015).
The use of δ 13 C and δ 15 N values of organisms to infer likely organic carbon and nitrogen sources relies heavily on assumptions regarding the difference between δ 13 C and δ 15 N values of organisms and their diet ( 13 C, 15 N). There is a need for more controlled laboratory studies investigating 13 C and 15 N (Martínez del Rio et al., 2009) as well as the relationships between the δ 18 O values of organisms and those of environmental water (Rubenstein and Hobson, 2004). 13 C, which is generally assumed to be between 0 and +1 ‰ for a range of animals, including invertebrates (DeNiro and Epstein, 1978;McCutchan et al., 2003), has been studied for chironomids under controlled laboratory conditions (Goedkoop et al., 2006;Wang et al., 2009;Heiri et al., 2012;Frossard et al., 2013) and ranges from −0.8 to +1.2 ‰. For Daphnia magna, 13 C values range from +1.7 to +3.1 ‰ (Power et al., 2003). 15 N, which is usually assumed to be between +3 and +4 ‰ (DeNiro and Epstein, 1981;Minagawa and Wada, 1984) ranges from −1.5 to +3.4 ‰ for chironomids (Goedkoop et al., 2006;Wang et al., 2009;Heiri et al., 2012) and from +1 to +6 ‰ for Daphnia (Adams and Sterner, 2000;Power et al., 2003;Matthews and Mazumder, 2008). In terms of oxygen, the δ 18 O values of lacustrine invertebrates are strongly and positively related to the δ 18 O values of local precipitation and the water in which the invertebrates live (Wang et al., 2009;Nielson and Bowen, 2010;Verbruggen et al. 2011;van Hardenbroek et al., 2012;Soto et al., 2013), although laboratory studies have shown that the oxygen isotopic composition of the diet can also affect invertebrate δ 18 O values (Wang et al., 2009;Nielson and Bowen, 2010).
There can be distinct offsets in isotopic composition between whole-body tissue and chitinous structures of invertebrates. Culturing experiments comparing cephalopod soft tissue and their chitinous mouthparts have shown that their chitinous structures can have δ 15 N values 3 to 4 ‰ lower than soft body tissue (Hobson and Cherel, 2006). Heiri et al. (2012) reported that offsets of up to 2 ‰ between chironomid body tissue and chitinous head capsule δ 13 C and δ 15 N values are possible. For Daphnia, field studies suggest that (non-ephippial) exoskeleton parts can have 0.8 ‰ lower δ 13 C and 7.9 ‰ lower δ 15 N values than whole Daphnia (Perga, 2010), while no clear differences in δ 13 C and δ 15 N values between Daphnia and ephippia have been reported in the only available study which examined this offset for Daphnia and free ephippia collected in a vertical net trawl in Lake Geneva, Switzerland (Perga, 2011). For vertebrates, differences in stable C and N isotopic composition between tissue types have been related to differences in contents of specific compounds (e.g., relative abundance of lipids, carbohydrates and protein or of different amino acids; e.g., DeNiro and Epstein, 1978;Pinnegar and Polunin, 1999). Differences in biochemical composition also provide a potential explanation for the observed differences in δ 13 C and δ 15 N values We present results from an experiment developed to examine the relationships between the δ 13 C values of diet and the δ 18 O values of environmental water, on the one hand, and the δ 13 C and δ 18 O values of Daphnia, on the other. The experiment was specifically designed to examine whether offsets in δ 13 C, δ 15 N and δ 18 O values exist between Daphnia and their ephippia. Furthermore, we investigated whether the stable isotopic compositions of Daphnia and their ephippia are influenced by temperature by performing the experiment at two different temperatures.

Daphnia cultivation
Three ex-ephippial Daphnia pulicaria clones (LC PUL 53, 99 and 101; Möst, 2013) from Lower Lake Constance (Switzerland) that showed extensive ephippia production in culture in pre-tests were selected for the experiment. For each clone, 20 neonate Daphnia (< 48 h old) were grown in 2.5 L batch cultures prior to the experiment. From these batch cultures, seven to eight second to third clutch neonates (< 48 h old) were transferred to 180 mL jars, containing 160 mL of filtered lake water (natural abundance or labeled water, according to treatment conditions described below). The lake water was filtered with 0.45 µm glass fiber filters (Sartorius Stedim AG, Switzerland). Initially, Daphnia were fed three times per week with fresh algae, concentrated to an equivalent of 1 mg C L −1 . After day 21 of the experiment, the amount of food was doubled because the number of Daphnia in most jars exceeded 30 individuals. Experimental water was exchanged once per week and ephippia (if present) were retained in the cultures. Due to potentially higher productivity and evaporation, the water was exchanged twice per week in Treatment 4 (20 • C).

Food and water sources in the experiment
Three weeks before the experiment, two 1 L chemostats were started simultaneously to produce the algae (Acu-todesmus obliquus, Turpin) to be used as food for Daphnia in the experiment. The algae were cultivated in a "WC"-medium (Guillard, 1975). For one of the chemostats, 45 % of the sodium bicarbonate in the medium (5.67 of 12.6 mg L −1 ) was replaced by sodium bicarbonate containing 99.9 % 12 C (Sigma Aldrich, USA), lowering the δ 13 C values of the algae from this chemostat by, on average, 1.8 ± 1.2 ‰ (one standard deviation (1 SD)) (see results). Once per week, the chemostat-grown algae were harvested, centrifuged (5000 rpm) to remove residual medium, stored at 9 • C in the dark and used to feed the Daphnia during the following week. Seven days before the start of the experiment, 250 L of lake water were collected from Greifensee (Switzerland) (pH 8.0, TP 0.04 mg L −1 , TN 1.6 mg L −1 ; data provided by the Cantonal Bureau for Waste, Water, Energy and Air (AWEL, Zürich; www.awel.zh.ch)). This water was stored in the dark at 12 • C for the duration of the experiment. Of this water, 50 L were stored in a separate container, and 0.9 mL of water containing 97 % 18 O (Sigma Aldrich, USA) were added to increase the δ 18 O value of the water by 5.6 ‰ relative to the unlabeled water (see results). Before exchanging the water in Treatment 4, the water was allowed to equilibrate with the ambient laboratory air temperature (20 • C).

Experimental design
The experiment consisted of four cultivation treatments: a control treatment in which Daphnia were cultivated in untreated, filtered lake water at 12 • C on a diet of fresh chemostat-grown algae (Treatment 1), and treatments with conditions identical to Treatment 1, with the exception of the algae in Treatment 2, which had 1.8 ± 1.2 (1 SD) ‰ lower δ 13 C values. The culturing water in Treatment 3 had δ 18 O values that were 5.6 ‰ higher than in the other treatments and Treatment 4 had a temperature (20 • C) that was higher than the other treatments.
Each treatment consisted of 30 glass jars, which were sterilized using an autoclave. Prior to the experiment, each glass jar was assigned to one of three replicate groups (A, B, C). The neonate Daphnia were evenly distributed in the jars to ensure that every experimental replicate group contained 10 jars, with 3 to 4 jars per clone. All the jars for a given treatment were held in one large tray, and the jars within each treatment were evenly distributed within the trays. The trays were held in the dark in temperature-controlled incubators.
The experiment was designed to assess the following: Statistical analyses were performed with the PAST software package, version 1.97 (Hammer et al., 2001), except for tests used to compare the algae from both chemostats. To account for repeated measures, linear mixed effects models (LMEs) were applied, fitting a random intercept for each probing date with the lme function in the nlme package in the R statistical package (R Core team, 2013). Significance was analyzed using an F test. A Bonferroni correction was applied to the multiple (six) comparisons of the stable isotopic composition of Daphnia between the treatments (Tukey post hoc tests).

Sample collection
After the weekly harvest, a small portion of algae from each chemostat was rinsed with deionized water and centrifuged five times to remove the culturing medium. The concentrated algae were freeze-dried and a small aliquot (150 to 200 µg) was loaded into tin cups (6 × 4 mm, Lüdi Swiss, Switzerland) to measure the δ 13 C, δ 15 N and δ 18 O values of the algae (δ 13 C algae , δ 15 N algae and δ 18 O algae ). In each treatment, one jar was assigned to monitoring variation in δ 18 O values of the water (δ 18 O water ). Once per week, before discarding the water, 12 mL were transferred to a 12 mL glass vial with no head space (Labco, UK) and stored in the dark at 7 • C. Every second sample was analyzed for δ 18 O water values. Every third week a sample of the water in the storage barrels was collected, stored and measured for δ 18 O water values.
The experiment was terminated after 62 days. He and Wang (2006) have demonstrated that the Daphnia carbon turnover rate is 11 to 36 % per day, which suggests that after 62 days our Daphnia likely had achieved isotopic equilibrium with the experimental diet and water. Daphnia and ephippia were harvested and pooled according to treatment (1-4) and replicate group (A, B, C). Adult Daphnia were hand-picked from a Bogorov sorting tray (Gannon, 1971) with fine forceps under a binocular and freeze-dried, after which they were loaded into tin cups (6 × 4 mm, Lüdi Swiss, Switzerland; ∼ 10 to 12 individuals per measurement) for analysis of δ 13 C Daphnia , δ 15 N Daphnia and δ 18 O Daphnia values. For each treatment replicate group, three samples were prepared and measured, resulting in 36 measurements for each chemical element. Ephippia were collected and treated in 10 % KOH for 2 h to remove any algal matter and egg yolk. Replicate measurements (three each for C, N and O) of ephippia not treated with KOH were prepared to assess any influence of this treatment on the isotopic compositions of ephippia. The ephippia were loaded into pre-weighed tin cups (6 × 4 mm, Lüdi Swiss, Switzerland): ∼ 10 to 15 for δ 13 C ephippia and δ 15 N ephippia analysis and 15 to 20 for δ 18 O ephippia analysis. Three samples were prepared and measured for each treatment replicate group, except for Treatment 4, which yielded only sufficient numbers of ephippia to measure once per treatment replicate group.

Assessing the source of oxygen in Daphnia
Following Wang et al. (2009), our experimental setup was used to approximate the proportional contribution of oxygen in the Daphnia stemming from the environmental water relative to that from the diet, using the following equation: where p is the proportion of oxygen in Daphnia stemming from the water,

Stable isotope mass spectrometry
The δ 13 C and δ 15 N values of the algae, Daphnia and ephippia were measured on a Costech ESC 4010 elemental analyzer interfaced via a ThermoConflo III with a Thermo Delta V isotope ratio mass spectrometer (IRMS) at the Alaska Stable Isotope Facility (ASIF) at the University of Alaska, Fairbanks. The analytical precisions for δ 13 C and δ 15 N values are expressed as 1 SD from the mean based on the results from multiple (n = 13) analyses of a laboratory standard (peptone) and were ± 0.2 ‰ and ± 0.1 ‰, respectively. The δ 18 O values of the water samples were measured on an online pyrolysis thermochemical reactor elemental analyzer (TCEA) (Finnigan ThermoQuest) coupled to a continuous flow (Conflo III) IRMS (Finnigan MAT Delta V) at the ASIF. Analytical precision is expressed as 1 SD from the mean based on the results from multiple (n = 3) analyses of a laboratory standard (doubly labeled water; ±0.3 ‰).

Daphnia stable isotope ratios
Mean stable isotope values for Daphnia are based on 9 measurements (three measurements for each of the three replicates per treatment). The mean δ 13 C Daphnia value in Treatment 2 (where Daphnia were offered 13 C-depleted algae) was lower (−20.2 ± 0.1 ‰) than in Treatment 1 (−18.7 ± 0.1 ‰) and 3 (−17.9 ± 0.1 ‰ ) (Fig. 3). For treatments at 12 • C (1-3), the mean δ 13 C Daphnia value was 0.5 ± 0.3 ‰ higher than the mean δ 13 C algae value that Daphnia were cultured on. The mean δ 13 C Daphnia value in Treatment 4 (20 • C; −19.0 ± 0.1 ‰) was 0.2 ± 0.1 ‰ lower than the mean δ 13 C algae value. The results from all treatments in terms of δ 13 C Daphnia values were significantly different from each other (one-way ANOVA and Tukey post hoc test; Table 1) Mean δ 15 N Daphnia values at 12 • C were 5.5 ± 0.1 ‰ (Treatment 1), 5.7 ± 0.1 ‰ (Treatment 2) and 6.2 ± 0.1 ‰ (Treatment 3), and they were 3.4 ± 0.3 ‰ higher than the mean δ 15 N algae value (Fig. 3). At 20 • C (Treatment 4), the mean δ 15 N Daphnia value (6.5 ± 0.2 ‰) was 4.0 ± 0.2 ‰ higher than the mean δ 15 N algae value. All treatments, except for Treatments 1 and 2 and Treatments 3 and 4, were signif- icantly different from each other with regard to δ 15 N Daphnia values (one-way ANOVA and Tukey post hoc test; Table 1). Treatments 1 and 2 were both performed at 12 • C and with similar water in terms of δ 18 O values. The mean δ 18 O Daphnia values in these treatments were 11.7 ± 0.1 ‰ and 11.0 ± 0.2 ‰ (Fig. 3). In Treatment 3, where the mean δ 18 O water value was 5.2 ‰ higher than in the other treatments, the mean δ 18 O Daphnia value was 14.6 ± 0.3 ‰ , which was 2.9 and 3.6 ‰ higher than in Treatment 1 and 2, respectively. In Treatment 4, with δ 18 O water as in Treatment 1 and 2 but run at a higher temperature (20 • C), the mean δ 18 O Daphnia value (10.2 ± 0.2 ‰) was 1.5 and 0.8 ‰ lower than in Treatment 1 and 2, respectively. A significant difference in δ 18 O Daphnia values was found between all treatments, except for Treatments 1 and 2 and Treatments 2 and 4 (oneway ANOVA and Tukey post hoc test; Table 1).

Ephippia stable isotope ratios
In all treatments ephippia production started between day 27 and day 34 of the experiment. Until day 48 of the experiment, ephippia production was low (on average 1 to 1.5 ephippia per jar per week), after which production increased to 4.5 to 6 ephippia per jar per week in Treatments 1, 2 and 3, whereas production in Treatment 4 remained low. Across the replicate treatments (A-C), the production of ephippia was similar with, on average, 12 to 13 ephippia per jar at the end of the experiment. The majority of the ephippia were produced by clone LC PUL 99 (55 %), whereas LC PUL 101 and 53 were responsible for 23 and 22 % of the ephippia production, respectively.

Discussion
Statistically significant differences were found between nearly all treatments for all investigated Daphnia stable isotope ratios, even in cases where we expected no differences based on the manipulations. For example, Treatment 1 and 3 were identical in terms of δ 13 C values of the food source and temperature and only differed in the δ 18 O values of the water, and Treatment 1, 2 and 3 were identical in terms of δ 15 N values of the food source and temperature. However, the unexpected differences between these treatments were generally small and of the same order of magnitude as the analytical precisions associated with each element (Fig. 3). They may represent the inherent variability associated with stable isotope ratios in organisms (Schimmelmann, 2011). Alternatively, since the stable isotope ratios of the algae showed some variability over the course of the experiment (Fig. 1), a slight difference in timing in the buildup of biomass may have led to small differences in Daphnia stable isotope ratios. In previous experiments, δ 13 C Daphnia and δ 15 N Daphnia values have been found to differ as much as 1 ‰ between identical treatments (Power et al., 2003). The differences in Daphnia stable isotope ratios were much larger when comparing treatments with manipulated δ 13 C algae and δ 18 O water values to those with non-manipulated algae and water.

The food experiment: changing δ 13 C algae
Offering Daphnia algae with, on average, 1.8 ‰ lower δ 13 C algae values resulted in 1.5 to 2.1 ‰ lower δ 13 C Daphnia values. Since the δ 13 C algae values were variable over time, we cannot reconstruct the exact δ 13 C value of the carbon that Daphnia in our different treatments assimilated, and therefore we cannot calculate a precise estimate of 13 C. Based on the mean δ 13 C algae value over the duration of the experiment, however, 13 C between Daphnia and algae is estimated to be +0.5 ± 0.3 ‰ at 12 • C. This is in agreement with  commonly found 13 C values of 0 to +1 ‰ for a range of animals, including invertebrates (DeNiro and Epstein, 1978;McCutchan et al., 2003). D. magna has been reported to have a 13 C value of +1.7 ‰ at 12 • C on a diet of aquarium food (Power et al., 2003). However, in this study a lipid correction was applied to infer δ 13 C values based on C : N ratios following a model by McConnaughey and McRoy (1979). This leads to relatively higher δ 13 C values, and the procedure has been criticized, since it potentially provides biased estimates when comparing isotopic ratios of different organisms and tissues (Mintenbeck et al., 2008). Power et al. (2003) did not report the C : N of the food and Daphnia, so we cannot back-calculate the δ 13 C values they measured prior to lipid correction.
δ 13 C ephippia values also reflected the difference in δ 13 C algae values between the treatments. At 12 • C, they were not significantly different from the δ 13 C Daphnia values (although they were consistently lower at 20 • C; see below). This is in line with the findings by Perga (2011), who found that the δ 13 C value of ephippia collected in the field was slightly, but not significantly, higher than the δ 13 C value of Daphnia collected in the same net trawls. This suggests that δ 13 C ephippia values are a reliable indicator of changes in δ 13 C Daphnia values, and consequently of variations in δ 13 C values of Daphnia diet: at 12 • C, δ 13 C ephippia was 0.7 ± 0.2 ‰ higher than the mean δ 13 C algae . The absence of a clear offset in δ 13 C values between whole Daphnia and Daphnia ephippia at 12 • C is in contrast to the difference found between whole Daphnia and Daphnia exoskeletons (0.8 ‰; Perga, 2010) and between chironomid body tissue and chironomid head capsules (∼ 1 ‰; Heiri et al., 2012;Frossard et al., 2013).

δ 15 N values of Daphnia and ephippia
At 12 • C, the observed 15 N was +3.4 ± 0.3 ‰, which agrees well with 15 N values referred to in the literature (+3 to +4 ‰; DeNiro and Epstein, 1981;Minagawa and Wada, 1984). A range of 15 N values for Daphnia have been reported. D. pulicaria reared on a diet of frozen algae pellets had a 15 N of +1.4 ‰ (Matthews and Mazumder, 2008). This is lower than the 15 N we found. According to Matthews and Mazumder (2008), the low 15 N they observed may be explained by the observation that a diet consisting of detritus (dead algae) is associated with considerably (∼ 2.5 ‰) lower 15 N values than one consisting of living plant matter (Vanderklift and Ponsard, 2003). Our observed 15 N for D. pulicaria is within the range of reported D. magna 15 N values (+1 to +6 ‰; Adams and Sterner, 2000;Power et al., 2003).
δ 15 N ephippia values were lower (1.6 ± 0.4 ‰) than δ 15 N Daphnia values. In contrast, Perga (2011) found δ 15 N ephippia values to be slightly, but not significantly, lower than δ 15 N Daphnia values in the field. Together with the results of Perga (2011), our data provide an indication that δ 15 N ephippia values are indicative of δ 15 N values of Daphnia and their diet, with only relatively minor offsets between food, Daphnia and ephippia. For chironomids, differences of similar magnitude between whole-body δ 15 N values and head capsule δ 15 N values (−1 to +1 ‰) were observed over a large range of δ 15 N values (2.5 to 15 ‰; Heiri et al., 2012). Therefore, it seems likely that differences between Daphnia and ephippia δ 15 N values may also be similar across this δ 15 N range.

The water experiment: changing δ 18 O water values
δ 18 O water values were 5.2 ‰ higher in Treatment 3 than in Treatment 1 and 2, and the mean δ 18 O Daphnia values in Treatment 3 were 2.9 ‰ higher than in Treatment 1 and 3.6 ‰ higher than in Treatment 2. This implies that, as expected, differences in δ 18 O Daphnia values reflect differences in δ 18 O water , yet that, as in other invertebrates, only part of the oxygen incorporated by the Daphnia originated from the water. Wang et al. (2009) reported that 69 % of the oxygen in chironomid larvae stemmed from the water in their environment. Soto et al. (2013) estimated that 84 % of the oxygen in protein isolated from chironomids came from the water in their environment, and Nielson and Bowen (2010) reported that 69 % of the oxygen in chitin from brine shrimp came from water in their environment. Based on Eq. (1), we estimate that in our experiment 56 to 69 % of the oxygen in Daphnia came from the water, based on Treatment 1 and 2, respectively. These estimates are similar to the values reported by Wang et al. (2009) and Nielson and Bowen (2010).
δ 18 O ephippia values closely reflected differences in δ 18 O Daphnia : they were, on average, 0.9 ± 0.4 ‰ lower than δ 18 O Daphnia values. This suggests that δ 18 O ephippia may be used as an indicator of δ 18 O Daphnia , which in turn can be expected to be related to lake water δ 18 O values. This is in agreement with the correlation between surface sediment δ 18 O ephippia values and lake water δ 18 O values found in a field survey of a number of European lakes (Verbruggen et al., 2011). Power et al. (2003) reported an increase of 0.1 ‰ in 13 C values for D. magna with a temperature increase from 12 to 20 • C (and +1.4 ‰ when temperature increased from 12 to 26 • C). Therefore, we expected 13 C values for Daphnia in Treatment 4 (20 • C) to be similar to or slightly higher than in the other treatments (12 • C). 13 C values were clearly lower, however, in Treatment 4 (−0.2 ± 0.1 ‰) than in the other treatments (+0.5 ± 0.3 ‰). While we cannot exclude a negative relation between temperature and 13 C values for Daphnia, we choose to treat this result with caution due to the discrepancy with the positive 13 C values as reported in other studies (DeNiro and Epstein, 1978;McCutchan et al., 2003;Power et al., 2003). A higher lipid content of Daphnia may potentially lead to lower δ 13 C Daphnia values (Mc-Cutchan et al., 2003). However, the C : N ratios of Daphnia in Treatment 4 were slightly lower (but not significantly different; t test: t = 1.18 p > 0.05) than those of Daphnia in Treatment 1, which does not agree with a higher lipid content in Daphnia from Treatment 4 (Smyntek et al., 2007). Alternatively, 13 C depletion of algal biomass during dark respiration may have affected the δ 13 C algae in Treatment 4 disproportionately due to the higher temperature. Degens et al. (1968) found that δ 13 C values of the alga Dunaliella tertiolecta were 4 ‰ lower after 3 days in darkness. The rate of respiration by algae depends on temperature and can be 2 to 4 times higher at 20 • C than at 12 • C (e.g., Vona et al., 2004). Microbial activity in the experimental jars could have been affected by temperature and could have also influenced our results. Additionally, if Daphnia in Treatment 4 had a different timing of growth compared to Treatment 1, as might be expected, they may have been assimilating carbon from algae with different δ 13 C algae values during the main phase of their growth compared to the other treatments, since δ 13 C algae values were relatively low in the beginning and at the end of the experiment (Fig. 1). δ 13 C ephippia values were also lower in Treatment 4, and 1.3 ± 0.3 ‰ lower than δ 13 C Daphnia values. For the same reasons as outlined above, it remains unclear whether this observation is the consequence of a fundamental change in the offset between δ 13 C Daphnia and δ 13 C ephippia with temperature or whether it is affected by variations in δ 13 C algae and algal respiration rates or differences in Daphnia growth rates between our treatments. Controlled experiments over a range of temperature values analyzing not only δ 13 C Daphnia and δ 13 C ephippia values, but also δ 13 C values of respired CO 2 and microbial biomass would be desirable to further explore this issue. Although the results of Treatment 4 indicate that the difference between δ 13 C ephippia and δ 13 C Daphnia values may be more variable than suggested by the cultivations at 12 • C, the offset is still relatively small compared to the variation in δ 13 C ephippia values in lake sediment records (up to 10 ‰; e.g., Wooller et al., 2012). 15 N between Daphnia and algae was +4.0 ± 0.2 ‰ at 20 • C, 0.6 ‰ higher than at 12 • C. A small increase (0.4 ‰) in 15 N in this temperature range has also been reported for D. magna (Power et al., 2003). Power et al. (2003) found a decrease of 2.7 ‰ in 15 N values for D. magna between 20 and 26 • C, however, and Barnes et al. (2007) found a decrease of 0.6 ‰ in 15 N values for sea bass with a temperature increase from 11 to 16 • C. Previously observed 15 N values in field studies of aquatic food webs (Vander Zanden and Rasmussen, 2001), and specifically in experimental studies of Daphnia (Adams and Sterner, 2000;Matthews and Mazumder, 2008), are, in some cases, lower than +3 to +4 ‰. A potential effect of temperature on 15 N values for Daphnia which, based on presently available observations, may amount to 2.7 ‰ at temperatures above 20 • C (Power et al., 2003) therefore deserves future attention. The offset between δ 15 N Daphnia and δ 15 N ephippia in our experiment was, however, not significantly different (t test: t = 0.26 p > 0.05) between Treatment 1 (control, 12 • C) and 4 (20 • C).

The temperature experiment
The effect of temperature on oxygen isotope fractionation during the formation of chitin by aquatic organisms has not been examined previously in experimental studies. Schimmelmann and DeNiro (1986) analyzed the δ 18 O values of the chitin of marine crustaceans collected along a temperature gradient of 10 • C and van  studied the δ 18 O values of aquatic beetles in museum specimens selected to represent a temperature gradient across North America. Both studies concluded that the temperature effect on oxygen isotope fractionation during chitin formation (if any) was smaller than the variability due to minor differences in local environmental conditions. In this study we kept close control over the environmental conditions and source water δ 18 O values, and we found that δ 18 O Daphnia was slightly (0.8 to 1.5 ‰) lower with an increase of temperature by 8 • C but otherwise similar conditions. This may indicate an effect of temperature on oxygen isotope fractionation by Daphnia. We do note, however, that the potential temperature effect on oxygen isotope fractionation by Daphnia observed in our experiment was relatively small and resulted from a large difference in temperature. Therefore, δ 18 O Daphnia values most likely primarily reflect environmental water δ 18 O values. The offset between δ 18 O ephippia and δ 18 O Daphnia in Treatment 4 (20 • C) was not significantly different, however (t test: t = 0.09, p > 0.05), from that in Treatment 1 (control, 12 • C). This suggests that, in contrast to the difference between δ 18 O water and δ 18 O Daphnia , this offset is not affected by temperature in the investigated temperature range (12 to 20 • C). Verbruggen et al. (2011) measured the δ 18 O values of recently deposited ephippia from surface sediments in lakes along a geographical gradient in Europe. They found a strong correlation between δ 18 O ephippia values and lake water δ 18 O values. In their data set, the δ 18 O values of lake water increased by ∼ 4.8 ‰ with a temperature increase of 8 • C, whereas δ 18 O ephippia values increased by only ∼ 3 ‰ over this temperature gradient, a difference of ∼ 1.8 ‰. This difference is of a similar order of magnitude as the 0.8 to 1.5 ‰ lower δ 18 O Daphnia values we found with an 8 • C temperature rise. The data of Verbruggen et al. (2011) and our experimental data would therefore be in agreement with a slight temperature effect on the fractionation of 18 O between lake water and Daphnia biomass. However, other mechanisms, such as a change in timing of Daphnia ephippia production with temperature and variations in δ 18 O values of food across the examined temperature gradient could also explain varying offsets between δ 18 O water and δ 18 O Daphnia at different temperatures in the study of Verbruggen et al. (2011). Moreover, differences in air temperature at lakes, which Verbruggen et al. (2011) reported, do not necessarily lead to similar differences in lake water temperatures.

Implications for palaeoecological studies
In general, we found that the stable isotopic composition of ephippia closely reflected the stable isotopic composition of Daphnia. The offsets were consistent within treatments and between most treatments (Fig. 4), and the ephippia stable isotope ratios responded to the manipulations in δ 13 C algae and δ 18 O water that we performed. Studies investigating the δ 13 C and δ 15 N values of fossil Daphnia ephippia have recorded shifts of up to 5 to 10 ‰ in δ 13 C values (Wooller et al., 2012;Frossard et al., 2014) and 3 ‰ in δ 15 N values (Griffiths et al., 2010). Shifts of 2 to 3 ‰ in δ 18 O values have been reported for fossil chironomid head capsules (Wooller et al., 2004;Verbruggen et al., 2010b). In our experiment, the standard deviation of the offset between Daphnia and ephippia stable isotope ratios was much smaller than the reported shifts in stable isotope ratios of fossil remains: ±0.4 ‰ for δ 13 C, δ 15 N and δ 18 O (±0.8 ‰ for δ 13 C when including Treatment 4 at 20 • C). If our findings are representative of the offset in stable isotope ratios between Daphnia and their ephippia in nature, they indicate that reported shifts in stable isotope ratios of fossil ephippia can reliably be interpreted as indicating past variations in Daphnia stable isotope ratios. These in turn can be expected to reflect past changes in isotopic composition of Daphnia diet and/or the δ 18 O of the water they lived in. While experiments offer the possibility to closely control the food sources and growth conditions for Daphnia, they cannot cover the full range of environments and interactions found in nature. Further studies in the field, in the fossil record and in an experimental setting are therefore needed to confirm the findings that we present here and to improve our understanding of the relationship between the stable isotopic composition of food, ambient water and chitinous fossilizing structures produced by Daphnia and other invertebrates. Although we only cultured Daphnia at two different temper-atures, we found indications that temperature may have affected 13 C and 15 N on the one hand and the relationship between δ 18 O water and δ 18 O Daphnia values on the other in an experimental setting. Future efforts focused on constraining the effect of temperature on these offsets and relationships are therefore particularly necessary.