Introduction

Moss animals (Bryozoa) are a common element of freshwater invertebrate assemblages, but have received relatively little attention in ecological and palaeoecological studies compared with other invertebrate taxa in lakes, e.g. insects or crustaceans. Bryozoans are sessile colonial suspension feeders that grow on submerged substrates (Wood & Okamura, 2005). Colonies are composed of asexually produced modules, called zooids, which use ciliated tentacles to create feeding currents to capture suspended food particles, including phytoplankton and bacteria (Kaminski, 1984; Wood & Okamura, 2005). Collecting bryozoan colonies can be challenging as they can be difficult to locate. Hence, bryozoans are generally not collected by standard sampling methods (e.g. kick-sampling). An alternative way to assess bryozoan presence and abundance is to collect their dormant stages, or statoblasts, which have robust, chitinous outer valves that are regularly found in flotsam, flood debris, and lake sediments (Hill et al., 2007). Statoblasts are commonly found in lake sediment records, and can therefore be analysed in palaeoecological studies to infer past dynamics of invertebrate assemblages (Frey, 1964; Francis, 2001; Okamura et al., 2013).

In modern ecosystem studies, stable carbon (δ13C) and nitrogen (δ15N) isotope analyses on aquatic invertebrates can provide information on food sources and on the length and structure of food webs of lakes (Post, 2002). For invertebrates that produce fossilizing chitinous structures, δ13C analysis of fossil remains can also provide information on past changes in the structure and carbon sources of lacustrine food webs (Wooller et al., 2008; Van Hardenbroek et al., 2014). For example, δ13C analyses on Daphnia and chironomid larvae have recently been used to reconstruct the relevance of methane-derived carbon in benthic and planktonic food webs in the past (Wooller et al., 2012; Van Hardenbroek et al., 2013a; Belle et al., 2014).

Climate strongly influences the H and O isotopic composition of lake water, which in turn determines the δD and δ18O values of lacustrine invertebrates. The stable isotopic composition of H and O in aquatic invertebrate fossils reflects the δD and δ18O value of lake water at the time when these invertebrates were alive, and δD and δ18O values of invertebrate fossils can thus provide information about past climatic change. For example, δ18O and δD values of fossil remains of aquatic insects have been identified as proxies for reconstructing past variations in lake water δ18O and δD values (e.g. Wooller et al., 2004; Verbruggen et al., 2010; Van Hardenbroek et al., 2013b). However, δ18O and δD values of aquatic invertebrates are also influenced by the δ18O and δD values of food (Wang et al., 2009; Soto et al., 2013; Schilder et al., 2015b). Reconstructions may therefore also be affected by variations in food sources available to aquatic invertebrates and in the isotopic composition of these food sources.

Despite their ubiquity and preservation in lake sediments, the potential use of statoblasts in stable isotope studies has been largely unexplored. Here we present an exploratory study of the carbon and hydrogen isotopic composition of bryozoan zooids and statoblasts collected at 23 sites in Northwest and Central Europe. We provide information on the range of bryozoan δ13C and δD values, as well as on the offsets between zooids and statoblasts under field conditions. We focused on stable carbon isotopes since invertebrate δ13C analyses are widely used in modern food web studies (Vander Zanden & Rasmussen, 1999; Grey et al., 2004a) and are increasingly analysed for palaeoecological reconstructions of carbon cycling in lakes (Frossard et al., 2014; Van Hardenbroek et al., 2014). Because our analytical set-up allowed us to simultaneously measure δD and δ13C values on relatively small bryozoan samples, we analysed hydrogen rather than oxygen isotopes. We also present the results of a laboratory study designed to characterise the offset between δ13C values of bryozoan zooids and statoblasts under controlled conditions.

Our study first investigates the relationship between zooids and statoblasts regarding their δ13C and δD values and secondly the relationship between bryozoans and their food/surrounding water. Specifically, we focus on the following questions: (1) How do δ13C values of bryozoan zooids relate to those of statoblasts under field conditions? (2) How do δD values of zooids relate to δD values of statoblasts under field conditions? (3) How are δ13C values of zooids/statoblasts related to δ13C values of their food under laboratory conditions? (4) How do δD values of zooids/statoblasts reflect lake water δD values? Our study provides a basis for a future use of δ13C and δD values of bryozoan statoblasts in palaeo food web studies and for inferring past variations in lake water δD values based on δD analyses of fossil bryozoan remains.

Methods

Field survey

Bryozoan colonies with statoblasts were collected from 23 sites in the littoral zone of lakes and ponds and from one stream. Species collected were Cristatella mucedo (Cuvier, 1798) (8 sites), Pectinatella magnifica (Leidy, 1851) (1 site), and representatives of the genus Plumatella, which were not identified to species (16 sites). Sites were visited from 2010 to 2012 and included locations in the Netherlands, Germany, and Switzerland (Table 1). Sufficient material to measure stable isotopes of ‘paired’ samples of both zooids and statoblasts was collected from most sites, but occasionally only zooids or statoblasts were available (Table 2). Simultaneously, water samples for stable isotope analysis were collected at all 23 sites in sealed containers and stored cool and dark until analysis within 2 months of collection.

Table 1 Location, date, and substrate of sampled Bryozoa
Full size table
Table 2 Number of sampled colonies and number of sites for which median stable isotope values are calculated
Full size table

Between one and six replicate, colonies were measured for each site (Tables 1 and 3). Colonies were kept cool and dissected within 24 h of collection. Gut evacuation was often incomplete when colonies died soon after detachment from their substrate. Extraneous material such as wood, algae and silt, was removed from the zooids with lancet and forceps to minimize contamination, but complete removal was not always possible due to the disintegration of fragile zooid tissues. Zooid material was freeze-dried and transferred into silver cups.

Table 3 δ13C and δD values (mean and standard deviation) of zooid and statoblasts of the three taxa studied: Cristatella mucedo (C), Pectinatella magnifica (Pe), and Plumatella (Pl)
Full size table

Mature statoblasts were identified with a dissection microscope (4-40 × magnification) and opened to remove the mass of yolk granules and germinal tissue using lancet and forceps. Statoblasts were then treated with 10% KOH for 2 h at room temperature to remove remaining attached soft tissue. This KOH treatment is commonly used in palaeolimnological studies of chitinous invertebrate remains and has been shown to have negligible effect on δ13C values of chitinous sheaths and exoskeletons (Van Hardenbroek et al., 2010; Schilder et al., 2015a). Samples were then rinsed with deionised water, freeze-dried, and transferred into silver cups for stable carbon and hydrogen isotope analysis.

Zooid tissue and matching statoblast samples from the field survey were measured on a high temperature elemental analyzer (ThermoFinnigan, Bremen, Germany) coupled to a mass spectrometer (Isoprime, Cheadle, UK). Pyrolysis temperature was set to 1,450°C. Since we attempted to measure stable isotope ratios for C and H simultaneously on small (60–160 µg) samples, the precision associated with the δ13C and δD measurements is relatively low. Replicate measurements (n = 37) on a chitin standard (Sigma Aldrich, Zwijndrecht, The Netherlands) had a standard deviation of 1.1‰ for δ13C and 3.1‰ for δD. Replicate measurements (n = 35) of a cellulose standard (Merck, Darmstadt, Germany) had a standard deviation of 1.0‰ for δ13C and 10.8‰ for δD. Stable carbon isotopes are reported relative to VPDB and stable hydrogen isotopes relative to V-SMOW. δD values of bryozoan samples were corrected for exchangeable hydrogen using the method described by Filot et al. (2006): In short, exchangeable hydrogen in the samples was equilibrated with standard water vapour of known isotopic composition. δD of bryozoan samples was calculated based on the measured δD after equilibration, the δD of the standard water vapour, and an estimated percentage of 23.9% exchangeable H in the sample, assuming that all samples have the same percentage of exchangeable H atoms.

Stable H and O isotopes of water samples from the field survey were analysed on a Finnigan MAT 250 mass spectrometer (Finnigan MAT, San Jose, CA) after equilibration of the water samples with a standard carbon dioxide using an equilibration device developed at the Physics Institute (University of Bern, Bern, Switzerland). Four small-volume samples were measured on a Picarro L1102-i analyser (Picarro Inc., Sunnyvale, CA) at the same laboratory. Standard deviations of measurements on water standards of known isotopic composition were better than 0.5‰ for δD and 0.1‰ for δ18O. For five sites, only δ18O of lake water was measured. For these five lakes, δD was estimated based on δ18O and the relationship between δ18O and δD was observed for Swiss lakes: first, the difference (Δδ18O) between measured δ18O in lake water and estimated δ18O in precipitation (Bowen & Revenaugh, 2003; Bowen, 2014) was calculated for each of our sampling locations. Similarly, ΔδD was calculated as the difference between measured lake water δD and estimated δD of precipitation for those sites where we had measured lake water δD. A linear regression was then used to estimate ΔδD as a function of Δδ18O (n = 21, r = 0.99). This relationship was used to calculate ΔδD from Δδ18O for sites without lake water δD measurements and to estimate δD based on the δD of precipitation and ΔδD. Table 1 specifies the method used to derive δD for each water sample.

Laboratory study

Colonies of Lophopus crystallinus (Pallas, 1768) and Plumatella emarginata (Allman, 1844) were grown to evaluate the δ13C values of their zooids and statoblasts and how this relates to the δ13C of their diet, particulate organic matter (POM). A microcosm for culturing bryozoans was established at constant 18°C (±1 − 2°C) as in Hartikainen & Okamura (2012). The microcosm contained deionized water spiked periodically with natural pond water. The system comprised two 16-l side tanks, housing the bryozoan colonies, connected to a 30-l main tank containing 2 goldfish. A fluorescent light tube above the main tank (Tropic Sun 5500 K, ZooMed, Ekeren, Belgium) and fish excretions promoted algal and bacterial production and hence food for bryozoans. Water was continuously circulated between the main and side tanks via airlifts and U-tubes.

Bryozoa collected from three UK sites (Barton Blow Wells in Lincolnshire, the Norfolk Broads in Norfolk, and Padworth in Berkshire) were allowed to grow for at least 30 days in the laboratory. Zooids grown de novo in the laboratory were transparent, allowing the exclusion of sediment-covered zooids and statoblasts formed in the field. This allowed us to select laboratory-grown material, which had only incorporated carbon from the POM under the laboratory conditions. After 30 days, colonies were transferred to artificial pond water for 24 h to allow gut evacuation. Zooids and mature statoblasts were separated with forceps and a lancet under a stereomicroscope (4–40× magnification). Two types of statoblasts were collected from Plumatella colonies: sessoblasts, which remain attached inside colonies, and floatoblasts, which are released into the water. All zooid materials per taxon were combined, freeze-dried, and homogenized. From this homogenized material, 3–4 replicate samples were weighed into tin capsules, and stored in a desiccator until stable isotope analysis. The same procedure was followed for statoblast material per taxon.

In addition, POM was collected at the start of the culture, at 14 days, and at 30 days to assess if δ13C values of POM changed during the study. POM was filtered onto pre-combusted filters (Whatman GF/C), freeze-dried, weighed into tin capsules, and stored in a desiccator until stable isotope analysis.

Because of their low weight (32–78 μg), zooid and statoblast samples from the culturing study could be only analysed for δ13C values. This was done on a Fisons NA 1500 NCS Elemental Analyzer coupled to a Thermo Electron Delta plus IRMS at the Geochemistry laboratory, Utrecht University, The Netherlands. Repeated measurements (n = 10) of an internal laboratory standard (NAXOS carbonate) yielded an analytical precision better than ±0.1‰.

Statistical analyses

To examine the relationship between δ13C and δD values of bryozoan soft tissue and statoblasts in the field study, we compared median values of replicate measurements per field site using nonparametric Spearman’s rank correlation coefficient (ρ) and associated significance tests (‘Hmisc’ package, R core team, 2013). Correlations were calculated for Cristatella and Plumatella separately and for all Bryozoa combined to see if similar relationships can be observed at genus level and for freshwater Bryozoa as a group. This provides useful information for palaeoenvironmental applications, where statoblasts of different genera might need to be pooled to retrieve enough material for stable isotope analyses. Nonparametric correlation coefficients and tests were selected since offsets between bryozoan soft tissues and statoblasts suggested some unusual outlier values in the δ13C measurements (see results). The same median values of replicate measurements per field site for zooids and statoblasts were used to test for significant differences in δ13C and δD values between zooids and statoblasts, using a paired samples t test. This was done for all bryozoan samples together and for Cristatella and Plumatella samples, using Past software version 2.14 (Hammer et al., 2001). We tested for significant differences between mean δ13C values of POM, zooids, and statoblast in the laboratory study using one-way ANOVA and pairwise comparisons with Tukey-HSD test (Past software version 2.14, Hammer et al., 2001).

Results

Field survey

In total 88 colonies from 21 sites were sampled (Table 1). Because of small sample quantities in some sites, reliable stable isotope measurements were only available for 80 statoblast samples and 57 zooid samples. Table 2 shows that paired samples of zooids and statoblasts from the same colony were found for Plumatella at 9 sites (23 paired samples), and for Cristatella at 7 sites (24 paired samples). Pectinatella was only found at 1 site (2 paired samples).

δ13C values of zooids and statoblasts

A remarkably large range of δ13C values of nearly 30‰ characterised bryozoans at the study sites (Fig. 1a). Plumatella zooids ranged from −48.2 to −19.4‰ compared with values of −42.3 to −22.4‰ measured on statoblasts. δ13C ranges of Cristatella were −40.0 to −26.8‰ for zooids and −39.8 to −25.2‰ for statoblasts. Offsets in δ13C between zooids and statoblasts generally ranged from −3.0 to +4.5‰. However, very large offsets were observed for four individual samples, ranging from −13.6‰ (one sample of Cristatella in Schöhsee) to +12.6‰ (one sample of Cristatella in Veenmeer and two samples of Plumatella in Aatalweiher). The overall mean of differences between median zooid δ13C values and median statoblast δ13C values per site was relatively small (1.4‰ ± 4.4‰ SD for Plumatella, 1.0‰ ± 1.9‰ SD for Cristatella, and −0.2‰ for Pectinatella, Fig. 2a). Differences between zooid and statoblast median δ13C values were not statistically significant for Plumatella, for Cristatella, or for all Bryozoa pooled (paired samples t-test).

Fig. 1
figure 1

Boxplots of δ13C values (a) and δD values (b) of the three bryozoan genera investigated in this study. Values for zooids are shown in white and for statoblasts in grey. Numbers indicate how many data points constitute each boxplot

Full size image
Fig. 2
figure 2

Stacked histograms representing the offsets between δ13C values (a) of zooids and statoblasts for Cristatella (black), Pectinatella (grey), and Plumatella (white). Average offsets calculated from the median value per site for each genus are indicated by circles of the same colour. In (b) the same is shown for δD values

Full size image

Median δ13C values of zooids and statoblasts were strongly and positively correlated (ρ = 0.70, P = 0.0019, n = 17) when all paired samples from bryozoans were examined together. Considering the taxa separately, a similar correlation was found for Cristatella (ρ = 0.85, P = 0.012, n = 7), but not for Plumatella due to the two samples in Aatalweiher with unusually low statoblast δ13C values (Fig. 3a). Without the Aatalweiher site, the correlation would also have been strongly positive for Plumatella (ρ = 0.81, P = 0.022, n = 8) and even stronger for all bryozoan samples (ρ = 0.90, P < 0.0001, n = 16).

Fig. 3
figure 3

δ13C values of statoblasts plotted against δ13C values of zooids for Cristatella, Pectinatella, and Plumatella. The dotted line indicates the 1:1 line. All data points are shown in (a); Median values for each sampling location are shown in (b) with grey lines representing the range of replicate δD values for each location

Full size image

δD values of zooids and statoblasts

The observed range of bryozoan δD values from our sites is nearly 100‰ (Fig. 1b). δD values of Plumatella zooids ranged from −213.2 to −127.4‰, compared with values of −207.2 to −125.9‰ measured on statoblasts. The ranges of Cristatella were −221.3 to −186.6‰ for zooids and −197.5 to −139.0‰ for statoblasts. Offsets in δD values between zooids and statoblasts ranged from −75 to +16 ‰ (Fig. 2b). Statoblast δD values appeared higher than δD values of zooids (Figs. 1b and 2b). The mean difference between median statoblast δD values and median zooid δD values per site was −44.8‰ ± 15.5‰ SD for Cristatella, and −12.2‰ ± 14.7‰ SD for Plumatella. These differences were statistically significant for Cristatella (two-sample t-test, t = −7.783, P < 0.001), but not for Plumatella. For Pectinatella, the median offset was −29.5‰, with too few samples for significance testing.

Only for Plumatella did we observe a positive correlation between median δD values of statoblasts and median δD values of zooids (ρ = 0.75, P = 0.0025, n = 9; Fig. 4), whereas the relationships between the median values were not significant for Cristatella or for all bryozoan samples combined. A visual examination of the scatter plots (Fig. 4) suggests that this lack of significance may be associated with more variable offsets between statoblast and zooid tissue in Cristatella than observed for the other bryozoan groups.

Fig. 4
figure 4

δD values of statoblasts plotted against δD values of zooids for Cristatella, Pectinatella, and Plumatella. The dotted line indicates the 1:1 line. All data points are shown in (a); Median values for each sampling location are shown in (b) with grey lines representing the range of replicate δD values for each location

Full size image

δD values of bryozoans and lake water

Sample pairs of zooids and lake water were available for Plumatella from 10 sites (29 paired samples), for Cristatella from 5 sites (20 paired samples), and for Pectinatella from 1 site (2 paired samples). However, when comparing median zooid δD values per site with lake water δD values based on Spearman’s correlation coefficients, no systematic relationships between lake water and zooid δD values were observed (Fig. 5a). Sample pairs of statoblasts and lake water were available for Plumatella from 16 sites (43 paired samples), for Cristatella from 7 sites (26 paired samples) and for Pectinatella from 1 site (3 paired samples). A positive correlation was observed (Fig. 5b) between median statoblast δD values from the same location and lake water δD values when considering all bryozoan samples combined (ρ = 0.56, P = 0.005, n = 24) and when considering Plumatella (ρ = 0.55, P = 0.027, n = 16), but this was not apparent for Cristatella. The lack of significant relationship for Cristatella may partly be a consequence of the lower number of localities for which samples of this species were measured.

Fig. 5
figure 5

Median δD values (with ranges indicated by grey lines) of Cristatella, Pectinatella, and Plumatella zooids (a) and statoblasts (b) plotted against δD of lake water. Grey lines represent the range of replicate δD values for each location

Full size image

Laboratory study

The laboratory study yielded three and four replicate samples of homogenized zooid tissue for Plumatella and Lophopus, respectively. For Plumatella, three replicate samples were available for both sessoblasts and floatoblasts, and for Lophopus three replicate samples of floatoblasts were collected. Measured δ13C values of zooids were in general very similar to δ13C values of sessoblasts and floatoblasts, as well as to values observed for POM (Fig. 6). ANOVA indicated statistically significant differences between POM, zooids and statoblasts for both Plumatella and Lophopus. For Lophopus, pairwise comparisons with Tukey-HSD tests indicated no significant differences between zooids and statoblasts. For Plumatella, however, Tukey-HSD tests confirmed that the observed mean differences of 1.2‰ between zooids and floatoblasts (Tukey-HSD, Q = 5.14, P = 0.023) and the 1.2‰ mean difference between zooids and sessoblasts (Q = 5.48, P = 0.014) were significant. No significant difference was observed between floatoblasts and sessoblasts of Plumatella (Tukey-HSD). Furthermore, the mean differences between zooids and POM of 1.7‰ were significant for both Lophopus (Tukey-HSD, Q = 8.22, P < 0.001) and Plumatella (Q = 8.14, P < 0.001). The mean 1.7‰ difference in δ13C values between POM and Lophopus statoblasts was significant (Tukey-HSD, Q = 8.26, P < 0.001), but no significant difference was found between POM and Plumatella sessoblasts or floatoblasts (Tukey-HSD).

Fig. 6
figure 6

Average offsets between δ13C values of food (particulate organic matter, POM, grey circles) and δ13C values of Plumatella and Lophopus zooids (open circles), floatoblasts (closed circles), and sessoblasts (closed squares) in the culturing experiment. Error bars indicate the standard deviation of three replicate measurements, unless indicated otherwise (in brackets). The standard deviation of the POM samples is shown to provide an indication of the variability of δ13C of POM in the culturing experiment

Full size image

Discussion

Large range of bryozoan δ13C values

The nearly 30‰ range of δ13C values observed for freshwater bryozoan tissues in this study is much larger than range of δ13C values previously reported by Turney (1999), Van Riel et al. (2006), and Van Hardenbroek et al. (2014). These earlier studies found δ13C values between -35 and -20‰ that largely overlap with reported ranges for phytoplankton and POM (France, 1995; Vuorio et al., 2006). At 10 sites, we found δ13C values that were lower than −35‰ and at one site, Chli Moossee, values measured for zooids were as low as −48.2 and −47.2‰. Planktonic algae can in some situations be characterised by δ13C values lower than −35‰. For example, δ13C values of −41 to −37‰ were reported by Jones et al. (1999) and Kankaala et al. (2010) for three small Finnish brown water lakes with low phytoplankton growth rates. Such low phytoplankton δ13C values, however, are very unusual for eutrophic lakes like Chli Moossee, and an additional source of 13C-depleted carbon must have been available to bryozoans. Methane-derived carbon is strongly 13C-depleted and it has been shown that different groups of freshwater invertebrates can incorporate carbon of methane-oxidizing bacteria (MOB) (Jones & Grey, 2011; Schilder et al., 2015b), leading to observed δ13C values as low as −70‰ in some invertebrate groups. The availability of MOB is especially high at the anoxic–oxic interface (Jones & Grey, 2011) and, in lakes with anoxic bottom waters, planktonic filter feeders have been observed to incorporate methanogenic carbon, leading to δ13C values lower than −50‰ in their biomass (Taipale et al., 2007; Schilder et al., 2015b). Bryozoans are sessile filter feeders, and all colonies obtained in this study originate from shallow parts of lakes down to a depth of 2 m. Richelle et al. (1994) have demonstrated that bryozoans can feed on microbial biomass. Our results suggest that, in some lakes, MOB may form a relevant part of POM in the shallow littoral zone and that bryozoans may incorporate carbon from MOB under these circumstances. Feeding partly on MOB would explain the extremely low δ13C values of Bryozoa found at Aatalweiher, Sisselenweiher, Chli Moossee, Golihübweiher, Lobsigensee, and Piepertkolk (Table 3). However, more detailed measurements of δ13C values of bryozoans and POM, and of the abundance of MOB in POM in littoral habitats would be necessary to confirm this hypothesis.

δ13C offsets between POM, zooids, and statoblasts

Freshwater consumers are usually very similar in their δ13C values compared to their diet, with consumer δ13C values on average 0–1.3‰ higher than those of their diet (DeNiro & Epstein, 1978; McCutchan et al., 2003; Peters et al., 2012). It has therefore been suggested that δ13C values of freshwater bryozoans reflect the δ13C values of phytoplankton or POM in the water column (Van Hardenbroek et al., 2014). This idea is supported by the results of our laboratory study. Although the 1.7‰ offset we observed between δ13C values of POM and cultured bryozoan zooids was statistically significant, it was small relative to the 30‰ range of δ13C values observed for zooids in the field survey.

In a study on the River Rhine, colonies of Plumatella repens and P. fungosa were characterised by δ13C values of −31.1‰ and −28.8‰, respectively (Van Riel et al., 2006). These values were substantially lower than δ13C values observed in the same study for POM (−24.27‰), which contrasts with the results of our experiments. In the same study on the River Rhine, however, Van Riel et al. also found that Plumatella δ13C was only 0.9–3.3‰ higher than δ13C values of phytoplankton (−32‰) that they estimated based on the δ13C values of dissolved inorganic carbon. The δ13C offset between Plumatella and phytoplankton reported by Van Riel et al. was therefore apparently similar to the offsets we report between POM and zooids in our laboratory study, suggesting that bryozoans were selectively feeding on phytoplankton, and that POM collected by Van Riel et al. contained organic matter not assimilated by bryozoans.

Other culturing experiments with planktonic filter feeders are in keeping with the results obtained in our laboratory study. For example, cultured specimens of Daphnia magna (Straus, 1820) were characterised by δ13C values 1.7–3.1‰ higher than their food (Power et al., 2003). In another study with Daphnia pulicaria (Forbes, 1893), this difference was 0.5 ± 0.3‰ (Schilder et al., 2015a). In our laboratory study, the 1.7‰ higher δ13C values of zooids of Plumatella and Lophopus compared with the δ13C values of their food are of similar magnitude, suggesting that zooid δ13C values provide a direct indication of the δ13C values of bryozoan diet.

In our laboratory study, we found very small offsets between δ13C values of bryozoan zooids and statoblasts, based on a diet with constant δ13C values. We observed no significant offset between δ13C values of zooids and statoblasts for Lophopus, and a small but significant 1.2‰ offset for Plumatella. This is in agreement with differences reported between whole body tissue of other aquatic invertebrates and their fossilizing, chitinous body parts. Perga (2011) showed that δ13C values of the ephippia of Daphnia from Lake Geneva were indistinguishable (±0.1‰) from δ13C values of whole body tissue. Similarly, a culturing experiment by Schilder et al. (2015a) indicated that δ13C values of Daphnia ephippia were on average 0.2 ± 0.4‰ higher than whole body tissue. Head capsules of 4th instar Chironomus riparius (Meigen 1804) larvae were on average 1.2 ± 0.9‰ and 0.9 ± 0.2‰ lower than whole body tissue in culturing experiments by Heiri et al. (2012) and Frossard et al. (2013), respectively.

Our laboratory study suggests that the δ13C offset between food and zooids (1.7‰) does not vary greatly between colonies, at least for the two taxa investigated. In contrast, the offset between body tissue and fossilizing structure can vary between 0 and 1.2‰. Variations <1.2‰ in δ13C values of statoblasts from sediment samples could therefore be the result of natural variability in the offset between zooids and statoblasts. Variations >1.2‰ can thus be interpreted as a colony-independent signal that has ecological or environmental significance. Certainly the large between-lake variability in bryozoan δ13C values we observed in the 23 sites of our field survey exceeds this 1.2‰ range. Other studies also indicated larger variability of δ13C values in ecosystem studies and in down core records. For example, Vander Zanden & Rasmussen (1999) report a range of 6‰ for δ13C values of primary consumers in modern lake ecosystems, and Van Hardenbroek et al. (2014) report a range of 5‰ for δ13C values of bryozoan statoblasts in a sediment record. The majority of this variation in δ13C values can thus be interpreted in terms of changing carbon sources, or changing δ13C values of these carbon sources.

Zooid and statoblast δ13C values under field conditions

We observed a strong correlation between the median δ13C values of zooids and associated statoblasts at the different study sites (Fig. 3), which confirms that δ13C values of statoblasts are systematically related to δ13C values of zooids. We observed a clearly greater variability in offsets between zooids and statoblasts, however, in the field survey than in the laboratory. In general, offsets in δ13C values between zooids and statoblasts ranged between −3 and +4.5‰ (Fig. 2a), with average offsets of 1.0, 1.4, and −0.2‰ for Cristatella, Plumatella, and Pectinatella, respectively. However, in three cases, statoblast δ13C was more than 10‰ lower than δ13C of zooids and the opposite was observed in one instance. These four data points clearly fall outside the regular range for offsets (Fig. 2a), and at two of these four extreme sites we also collected paired samples with offsets of only 0.9–3.6‰. This suggests that the availability of food sources that differ >10‰ can be a very localized phenomenon in time and space, possibly occurring in particular microhabitats. Differences in δ13C values of similar magnitude (10–20‰) in chironomid larvae were linked to local oxygen depletion in lakes and localized incorporation of methane-derived carbon (Grey et al., 2004b; Agasild et al., 2013).

In addition to spatial and temporal variability in the carbon sources available to bryozoans, other factors may have contributed to the large range of δ13C offsets in the field survey. Examination of colonies collected during fieldwork revealed that some of them were partly covered or interspersed by periphyton and that the guts of the bryozoans still contained variable amounts of material. Because zooids rapidly disintegrated during dissection, it is likely that non-bryozoan material was not completely removed from zooids. Variable amounts of non-bryozoan material in our samples might also explain the relatively large variability in δ13C values of replicate colonies from the same location (Table 3). Without additional isotopic analyses of POM, gut content, and periphyton at the different sites, however, we cannot draw firm conclusions about the causes for the observed variability in δ13C values of bryozoan colonies.

Taxonomic differences in δD values

Our data suggest that Plumatella statoblast δD values are clearly related to δD values of associated zooids and to δD values of lake water, whereas this is not observed for Cristatella statoblasts (Figs. 4 and 5). This might simply be explained by the low number of data points for Cristatella, but other explanations could also be considered.

One explanation for the differences between Cristatella and Plumatella may be the difference in food particles ingested by these two groups, because δD values of aquatic invertebrates are strongly influenced by δD values of their food (Solomon et al., 2009; Wang et al., 2009; Soto et al., 2013). At most of our study lakes, bryozoans can be expected to feed predominantly on planktonic algae or microorganisms feeding on them. The δD values of this food source can be expected to be closely related to lake water δD values and therefore relatively constant for a given site. However, for some organism groups, such as MOB, extremely low δD values have been reported (Whiticar, 1999; Deines et al., 2009). Furthermore, organisms feeding predominantly on terrestrial organic matter may be characterised by δD values that differ from algal organic matter produced within lakes (Karlsson et al., 2012). Kaminski (1984) demonstrated that Cristatella mucedo selects small seston (<7 µm in diameter), which can include bacteria, whereas Plumatella repens prefers slightly larger particles (ranging from 5 to 17 µm in diameter). Cristatella may therefore feed on small organisms with a more variable isotopic composition, such as chemoautotrophic or methane-oxidizing bacteria, which are less abundant in the larger particles than Plumatella feeds on.

Another explanation could be that Plumatella is firmly attached to substrates, whereas Cristatella is mobile and has been found at water depths of up to 20 m (Lacourt, 1968), both on hard and soft substrates. As lake water δD values can vary within the water column (Gat, 1995) and Cristatella colonies are capable of limited movement to different microhabitats, Cristatella could potentially incorporate different food sources and be exposed to water with different δD values than the immobile Plumatella. Even if the mechanism behind this observation is not fully understood, our results indicate that δD values of Plumatella statoblasts reflect lake water δD more closely than δD of Plumatella zooids and Cristatella tissues.

δD offsets between zooids and statoblasts

In contrast to the δ13C values, which were similar for zooids and statoblasts if median values were examined, we found that zooid δD was substantially lower than statoblast δD for most of the paired samples examined in the field survey (Fig. 2). A visual examination of Fig. 4 reveals that this is largely due to δD values for Cristatella obtained from 5 sites (i.e. Alte Aare, Piepertkolk, Schöhsee, Veenmeer in 2010, and Veenmeer in 2012), which are characterised by higher offsets between zooid and statoblast values and fall outside the scatter of other data points. These large offsets between δD of zooids and statoblasts might be linked to differences in food type and mobility as discussed above.

In addition, fractionation during the synthesis of different compounds can result in different δD values between tissues. For example, lipids are especially D-depleted compared with other tissues (Hobson et al., 1999; Soto et al., 2013). The higher atomic carbon content of zooids (mean 44%) compared to statoblasts (mean 30%) in our culture supports the idea of higher lipid content in zooids. Further experiments with controlled δD values of food and environmental water, and analysis of the chemical composition of bryozoan tissues, will be necessary to further constrain the reasons for the unexpectedly large offset in δD values between zooids and statoblasts, especially for Cristatella.

Relationship between δD of lake water and Bryozoa

Lake water δD values are more clearly related to the δD values of statoblasts than to the δD values of zooids (Fig. 5). One explanation for this may be that zooid samples are more easily affected by contamination with attached organic material and undigested particles in the guts, as discussed above. Secondly, we assumed a constant proportion of exchangeable H in our samples based on chitin and cellulose reference materials. However, the proportion of exchangeable H may differ between tissues (Wassenaar & Hobson, 2000; Schimmelmann et al., 2006). Zooid tissues are more diverse in chemical composition, leading to additional variability in δD values after correcting for exchangeable H, especially if compounded by contamination with non-bryozoan organic matter. Thirdly, differences in turnover rates between zooid and statoblast biomass might lead to incorporation of H into zooids that is different in δD from the material incorporated into statoblasts. δD values of lake water can change seasonally (Gat, 1995; Schürch et al., 2003), leading to temporal changes in δD of the water and food available to bryozoans. However, controlled experiments are required to estimate turnover rates in different bryozoan tissues and to investigate how quickly changes in δD values of water and diet are recorded in different bryozoan tissues.

Conclusions

Our results demonstrate that the C isotopic composition of freshwater bryozoan statoblasts is systematically related to the isotopic composition of the zooids over a large range of δ13C values. Offsets in δ13C values between zooids and statoblasts were considerably more variable in the field survey than in our laboratory study, with very large offsets observed for some of the sampled colonies. However, median estimates, based on statoblast δ13C values from several colonies per sampling site, were strongly related with zooid δ13C values. Similarly, median statoblast δD values based on material from several colonies per site showed a robust relationship with lake water δD and, for Plumatella, with δD values of zooids.

Statoblasts obtained in flotsam and lake sediment samples typically originate from numerous bryozoan colonies within an examined lake. Our results therefore suggest that C and H isotopic analyses on such samples can provide insights into variations of δ13C values of bryozoan zooids in lakes and in situations where bryozoans predominantly feed on algal organic matter, on variations in lake water δD. The robust nature of chitinous statoblasts makes them particularly suitable for studying the isotopic composition of lacustrine primary consumers over long time scales, using statoblasts preserved in lake sediment records. Since statoblasts also include N, O, and S, the stable isotopic composition of these elements may provide further valuable information of a distinct ecosystem component near the base of aquatic food webs.