Herbivorous cladoceran essential fatty acid and cholesterol content across a phosphorous and DOC gradients of boreal lakes : Importance of diet selection

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY 4.0 https://creativecommons.org/licenses/by/4.0/ Herbivorous cladoceran essential fatty acid and cholesterol content across a phosphorous and DOC gradients of boreal lakes : Importance of diet selection © 2023 The Authors. Freshwater Biology published by John Wiley & Sons Ltd. Published version Keva, Ossi; Litmanen, Jaakko J.; Kahilainen, Kimmo K.; Strandberg, Ursula; Kiljunen, Mikko; Hämäläinen, Heikki; Taipale, Sami J. Keva, O., Litmanen, J. J., Kahilainen, K. K., Strandberg, U., Kiljunen, M., Hämäläinen, H., & Taipale, S. J. (2023). Herbivorous cladoceran essential fatty acid and cholesterol content across a phosphorous and DOC gradients of boreal lakes : Importance of diet selection. Freshwater Biology, 68(5), 752-766. https://doi.org/10.1111/fwb.14061 2023 752 | wileyonlinelibrary.com/journal/fwb Freshwater Biology. 2023;68:752–766. Received: 8 December 2021 | Revised: 14 December 2022 | Accepted: 23 January 2023 DOI: 10.1111/fwb.14061 ORIGINAL ARTICLE Herbivorous cladoceran essential fatty acid and cholesterol content across a phosphorous and DOC gradients of boreal lakes— Importance of diet selection Ossi Keva1 | Jaakko J. Litmanen1 | Kimmo K. Kahilainen2 | Ursula Strandberg3 | Mikko Kiljunen1 | Heikki Hämäläinen1 | Sami J. Taipale1 1Department of Biological and Environmental Science, University of Jyväskylä, Jyväskylä, Finland 2Lammi Biological Station, University of Helsinki, Lammi, Finland 3Department of Environmental and Biological Sciences, University of Eastern Finland, Joensuu, Finland Correspondence Ossi Keva, Department of Biological and Environmental Science, University of Jyväskylä, P.O. Box 35 (YA), Jyväskylä FI40014, Finland. Email: [email protected] Funding information Academy of Finland; Suomen Kulttuurirahasto Abstract 1. Eutrophication has been shown to increase production of nutritionally lowquality cyanobacteria and decrease the longchain polyunsaturated fatty acid (PUFA) content of seston. Contrarily, lake browning inhibits cyanobacteria contribution in seston and favours poorly grazable mixotrophic algal species. These environmental changes have probable impacts on the diet and longchain PUFA content of primary consumers. However, herbivorous zooplankton may preferentially retain PUFAs through diet selection for optimal growth and reproduction, but such selective feeding is challenging to document in nature owing to the difficulties in quantifying zooplankton diet. 2. Here, we sampled seston and herbivorous cladocerans (Daphnia sp. and Bosmina sp.) from lakes (n = 23) in Finland along eutrophication (total phosphorous) and browning gradients (dissolved organic carbon [DOC]). We analysed the fatty acid content of seston (mg FA/g POC [particular organic content]) and cladocerans (mg FA/g C), and estimated available and consumed diet biomass percentages with quantitative fatty acid signature analysis. Cladoceran diet preference was evaluated as the difference between consumed and available food sources, to understand if they preferentially retain high nutritional quality diet. 3. Generally, lake chemistry and morphometry poorly explained seston and cladoceran longchain PUFA contents. However, multiple linear models for shorter chain PUFAs (linoleic acid [LA] and alphalinolenic acid [ALA]) performed better in explaining variation in the LA and ALA content of seston (20% and 11%) and cladocerans (36% and 46%, respectively). The factors most strongly and positively associated with the LA and ALA content of seston and cladocerans were phosphorus and DOC concentrations, respectively. 4. Seston and cladoceran PUFA contents were clearly uncorrelated. In most of the sampled lakes, highquality diet (i.e., diatoms and cryptomonads) was preferred by cladocerans and lowquality diet (cyanobacteria) was avoided. Lake chemistry This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2023 The Authors. Freshwater Biology published by John Wiley & Sons Ltd. [Correction added on 29 September 2023, after first online publication: The copyright line was changed.] | 753 KEVA et al. 1 | INTRODUCTION Lake water chemistry is largely dependent on catchment geology and land use. For instance, the use of fertilisers in agriculture and intensive forestry with drainage ditching networks increase leaching of nutrients and organic matter from catchments to lakes (e.g., Finér et al., 2021; Kritzberg et al., 2020). Lake water chemistry and light climate have myriad impacts on phytoplankton community structure and biomass. For example, cyanobacteria thrive in eutrophic lakes, whereas chrysophytes and diatoms have high biomass proportions in oligotrophic phytoplankton communities (e.g. Lepistö & Rosenström, 1998; Taipale, Vuorio, et al., 2016). Largesized raphidophytes, that are poorly grazed by zooplankton, have been shown to thrive in brown water lakes (Lebert et al., 2018; Strandberg et al., 2020), but mixotrophic chrysophytes, cryptomonads and dinoflagellates also are common in such lowlight conditions (Lepistö & Rosenström, 1998; Senar et al., 2021). Phytoplankton taxa differ in their size and polyunsaturated fatty acid (PUFA) and sterol contents. Therefore, differences in phytoplankton community structure, for example resulting from eutrophication or browning, may alter the availability of these important biomolecules to consumers (MüllerNavarra et al., 2000, 2004; Strandberg et al., 2020; Taipale, Vuorio, et al., 2016, 2019; Trommer et al., 2019). Cladocerans, in particular, are a key component in transferring energy and important biomolecules from primary producers and basal sources to secondary consumers, and thus are of major interest in trophic ecology (e.g., de Bernardi et al., 1987). Some previous studies have shown a clear relationship between seston and herbivorous cladoceran PUFA contents (Brett et al., 2006; Francine et al., 2022; Ravet et al., 2010), whereas some have reported no relationship (Persson & Vrede, 2006; Smyntek et al., 2008). As the carbon renewal time in Daphnia generally is 6 days (Taipale et al., 2011), the strongest correlations have been found between Daphnia fatty acid (FA) profiles and those of potential food sources determined a week before (Taipale et al., 2009). Not all available phytoplankton is suitable for herbivorous cladocerans: for example, largesized raphidophytes or dinoflagellates are inedible for the smallest cladocerans. This weakens the PUFA transfer from primary producers to primary consumers (Johansson et al., 2016; Peltomaa et al., 2017; Strandberg et al., 2020). Copepods are known to actively select their diet (Fryer, 1957) and it has been suggested that this selection is driven by the biomolecular composition of the diet items (DeMott, 1986; DeMott & Moxter, 1991). A mismatch between algalproduced and copepodassimilated FAs supports a selective retention of FAs by copepods (Schneider et al., 2017). By contrast, it has been previously considered that herbivorous cladocerans (Daphnia) as filter feeders do not actively select food items (Butler et al., 1989), apart from size selection based on the size limitations of the feeding apparatus (Brendelberger, 1991; Irvine, 1986). However, Hartman and Kunkel (1991) showed that Daphnia is able to feed selectively on large particles, although the role of smaller particle selection in lakes remains unknown. Juvenile and adult Daphnia can locate habitats with highquality food (Schatz & McCauley, 2007), possibly attracted through perception of increased ingestion rate and odour (Jensen et al., 2001). Furthermore, it has been suggested that Daphnia are able to reduce their feeding rate when disadvantageous food is present (Lampert, 1981). An experimental study has shown that during cyanobacteria blooming, Daphnia is able to prefer other food sources (Gladyshev et al., 2000). Thus, rapid detection of highquality food patches could support rapid population growth as a consequence of the short life cycle and parthenogenetic reproduction strategy in herbivorous Daphnia (Ebert, 2005). A recent field study, using a FAbased model to estimate cladoceran diet composition, suggested that herbivorous cladocerans preferentially fed on cryptomonads, chrysophytes and diatoms in a eutrophic and a brown water lake (Taipale, Aalto, et al., 2019). It is not known if cladoceran dietary selection is similar in lakes with different trophic status. The uniqueness of FA composition of different phytoplankton groups is the rationale for FAbased modelling in assessing seston composition and cladoceran diet composition (Galloway et al., 2014; Litmanen et al., 2020; Strandberg et al., 2015). Although there are still knowledge gaps in the retention and conversion pathways of some FAs, these caveats could be bypassed by using cultured algae and feeding experimentderived cladoceran FA libraries to estimate composition (Galloway et al., 2014; Strandberg et al., 2015). In this study we investigated the effects of various environmental predictors, mainly nutrient and DOC concentrations as well poorly explained cladoceran diet preference, but highquality preference was positively associated with lake average depth. 5. In summary, our spacefortime study approach did not reveal that eutrophication or browning downgraded the seston nor cladoceran PUFA quality. We found no correlation with seston and cladoceran PUFA content, but a clear mismatch between available and consumed diet. Our results suggest a selective feeding strategy of cladocerans, possibly through foraging in highquality algae patches or selective assimilation of PUFAs. KEYWORDS browning, eutrophication, nutritional quality, seston composition, zooplankton diet 13652427, 2023, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/fwb.14061 by University Of Jyväskylä Library, Wiley Online Library on [30/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 754 | KEVA et al. as lake morphometry parameters, using a spacefortime study approach, on (a) the content of PUFAs and sterols in seston and cladocerans, (b) the dietary preferences of cladocerans and (c) the trophic transfer of PUFA from seston to cladocerans. Previous studies have shown that eutrophication has a negative effect on the sestonic longchain PUFA content, but a weak positive effect on the shortchain PUFA content (e.g., MüllerNavarra et al., 2004). We were particularly interested in linoleic acid (LA, c18:2n6), arachidonic acid (ARA, c20:4n6), alphalinolenic acid (ALA, c18:3n3), eicosapentaenoic acid (EPA, c20:5n3) and docosahexaenoic acid (DHA, c22:6n3), which are considered beneficial for the growth and reproduction success of primary consumers (e.g., Arts et al., 2001; MüllerNavarra et al., 2004; Peltomaa et al., 2017). EPA might be the most physiologically important PUFA promoting cladoceran growth and reproduction (MartinCreuzburg et al., 2010; Ravet et al., 2010). There may not be any physiological demands for DHA in cladocerans as dietary DHA has been shown to be retroconverted to EPA and thus support cladoceran growth (MartinCreuzburg et al., 2010). In addition to FAs, sterols are linked with cladoceran metabolism, and some phytosterols such as cholesterol support growth more than others (e.g., MartinCreuzburg et al., 2010, 2014; Peltomaa et al., 2017). Our predictions are: (P1) Even though herbivorous cladocerans (Daphnia, Bosmina) can locate nutritionally highquality spots, eutrophication and browning will have a negative impact on the longchain PUFA content of herbivorous cladoceran as a result of the decreased nutritional quality of grazable phytoplankton. (P2) A strong positive relationship between seston and cladoceran PUFA indicates no diet selection, whereas the lack of relationship suggests the opposite. As the proportion of EPA and DHA synthesising grazable phytoplankton taxa is supposedly low in more eutrophic and brown lakes, we expect that the cladoceran diet preference of these taxa would increase with lake productivity and browning to maintain high cladoceran biomass production. 2 | MATERIALS AND METHODS 2.1 | Study area and field work We collected lake water, seston and zooplankton from 23 Finnish lakes (Figure 1; Table 1). The study lakes are in southern and central Finland with varying water chemistry characteristics (dissolved organic matter [DOC]: 5– 22 mg/L; total phosphorous [TotP]: 4– 149 μg/L; total nitrogen [TotN]: 300– 2,638 μg/L; Table 1) and morphometries (lake surface area: 0.1– 10,751.0 ha, catchment area: 0.1– 931.5 km2, mean depth: 1.2– 21.1 m, maximum depth: 1.9– 68 m). Lake water chemistry, seston and zooplankton samples were collected from epilimnion in most cases once in late summer (July– August) between 2016 and 2020 (Table 1). Lake water chemistry data (TotP, TotN, DOC, total organic carbon [TOC] and chlorophylla [chla]) were based on duplicate analyses of samples taken for the present study, or were derived from an opensource database HERTTA (Finnish Environment Institute) or from previous studies (Taipale, Galloway, et al., 2016). TotP and TotN concentrations were analysed with an automated discrete photometric system (Gallery™ Plus Automated Photometric Analyser; Thermo Fisher Scientific). TOC and DOC were analysed with a total organic carbon analyser (TOC500 and TOCL; Shimadzu). We had TOC data from only 16 lakes and predicted TOC concentrations for the remaining lakes from DOC concentration data using a linear regression model (r2 = 0.93; Figure S1). Lake particular organic matter (POC) concentration was derived for all of the lakes using the equation POC = TOC − DOC. Lake morphometric data (lake area, mean and maximum depth) were derived from HERTTA and open terrain maps (National Land Survey of Finland). Catchment area size and percentage coverage of five major land cover categories (urban, agriculture, forested, wetland and water [lakes and rivers]) were derived using the open VALUEtool (Finnish Environment Institute, https://paikkatieto.ymparisto.fi/value), which uses open terrain maps and the CORINE land cover database 2012 (Table 1). FIGURE 1 Study lakes located in southern and central Finland. For lake names, morphometry and water chemistry, see Table 1. 20 10 14 21 19 25.0°E 5km N S WE 13 11 5 60°N 65°N 25°E 30°E 100km N S WE 22 16 12 92,7 23 3 17 aissuR dnalniF nedewS 8 15 61.2°N 61.0°N 1,4 6 18 13652427, 2023, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/fwb.14061 by University Of Jyväskylä Library, Wiley Online Library on [30/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License | 755 KEVA et al. For the seston samples, lake water was collected from the top of the epilimnion (1 m) in the middle of lake with a Limnos sampler (3.5 L) and filtered through 50μm mesh. A subsample of the sieved water (range 80– 1,150 ml, mean ± SD 240 ± 228 ml) was filtered through a cellulose filter (Whatman cellulose nitrate membrane filter: pore size: 0.3 μm, diameter 47 mm) and freezedried (−50°C, 48 h). Zooplankton samples were collected from the same sites using a plankton net (50μm mesh- ) with vertical hauls. Live zooplankton samples were washed to a beaker with tap water and herbivorous cladocerans (Daphnia spp. and Bosmina spp., hereafter cladocerans) were separated from the bulk sample using a glass plate technique whereby cladocerans were trapped to the surface layer and copepods to the bottom of a beaker. A glass plate was dipped into the beaker to pick the trapped cladocerans from the surface and concentrated to a 1.5 ml polypropylene tube. The cladoceran samples were freezedried (−50°C, 48 h) in the tubes and stored in a deep freezer (−80°C, maximum duration 3 months) before lipid extraction. Cladocerans were freezedried before they emptied their stomachs; however, the contribution of gut mass on cladoceran total mass is low (c. 5%; Feuchtmayr & Grey, 2003) and thus also the confounding impacts on cladoceran PUFA content are considered minimal in our samples as well. Daphnia and Bosmina FA profiles have been shown to be very similar (Persson & Vrede, 2006; Ravet et al., 2010), and thus they were pooled in all subsequent FA analyses. 2.2 | Laboratory methods Lipids of freezedried filters and cladocerans (target mass 1.5 ± 0.25 mg) were extracted using chloroformmethanol solution (2:1 vol) in an ultrasonic bath and spiked with internal standard (Free c23:0). Duplicates of each sample were analysed when sample material was not limiting. Distilled ultrapure water (0.75 ml) was added to Kimax tubes to separate watersoluble compounds from lipids. Methylation reagent (1% H2SO4 [sulfuric acid] in methanol) was added to samples and the methylation reaction was catalysed with water bath incubation (50°C for 20 h). The produced FA methyl esters (FAMEs) were extracted to hexane (run volume 300 μl in most cases) and run with a gas chromatograph (GC) attached to a mass spectrometer (MS) (GC2010 Plus and QP2010 Ultra; Shimadzu). In most runs, the GC column was Zebron ZBFAME (30 m + 5 m guardian × 0.25 mm × 0.2 μm). Each GC– MS run started with the oven temperature of 50°C for 1 min, which was raised by 10°C/min to 130°C, 7°C/min to 180°C, and 2°C/min to 200°C at which it was held for 3 min before the oven temperature was raised by 10°C/min to 260°C. The injection temperature was 270°C, the interface temperature 250°C and the linear velocity in the column was 36.3 cm/s. Samples were run with splitless injection method. In few latest runs, a Zebron DB23 GC column (60 m × 0.25 mm × 0.2 μm) was used with a similar run temperature program but adjusted to the longer column. The FAs were identified based on ion spectra and quantified based on fourpoint standard mixture calibration curves (15 ng/μl, 50 ng/μl, 100 ng/μl, 250 ng/μl; GLC 566c, NuChek Prep) with GCMS solution software version 4.42 (Shimadzu, Japan). The calibration curves were analysed before each run and coefficients of determination between peak area and standard FA concentrations were always higher than 0.995. The recovery percentage of the internal standard (c23:0) was 94.0 ± 10.3% (n = 78) and this was used to adjust FA concentrations of the samples. The average sum of FA profile error between duplicate samples for seston and cladoceran samples were 6.3 ± 4.1% (n = 14 pairs) and 10.8 ± 8.5% (n = 18 pairs), respectively. The seston FA content unit was calibrated to mg FA/g POC and for cladoceran FA the unit was mg FA/g C. The cladoceran carbon content (%) was derived from an elemental analyser (FlashEA 1112; Thermo Fisher Scientific). The whole FA profiles of seston and cladocerans were used for reconstructing seston composition and cladoceran diet. LA, ARA, ALA, EPA and DHA are the only FAs for which content data are presented in this study, and abbreviated in the text as PUFA. Lakespecific PUFA content data can be found in Table S1. Sterols were analysed according to Taipale, Hiltunen, et al. (2016) from the lipid extraction of the samples where sufficient sample was available. Briefly, trimethylsilyl derivatives of sterols were analysed with GC– MS with a Phenomenex ZB5 Guardian column (30 m × 0.25 mm × 0.25 μm). Sterols were identified using characteristic ions (Taipale, Hiltunen, et al., 2016) using GCMS solution software. Alpha cholestane was used as internal standard and for recovery correction. The seston sterol content was corrected with lake POC concentration (mg/g POC) and cladoceran sterol content was calculated for C mass (mg/g C). Only cholesterol is reported in this study as it is the main sterol for cladocerans (e.g., Peltomaa et al., 2017). 2.3 | Seston and cladoceran diet composition estimation The measured FA profiles of seston and cladocerans were used to estimate seston and cladoceran diet compositions. We used Quantitative FA Signature Analysis in R (QFASAR) (Bromaghin, 2017; Iverson et al., 2004) with χ2 distance measure (Stewart et al., 2014). The method had previously been validated for Daphnia diet estimation (Litmanen et al., 2020). Seston composition estimation had been validated for FASTAR (Galloway et al., 2015; Strandberg et al., 2015) but Litmanen et al. (2020) found that QFASAR with χ2 distance measure provides more accurate results than FASTAR on Daphnia diet estimation and thus was used for seston composition estimation as well. The method was applied with previously determined laboratory culture and monoculture feeding experimentderived FA profile libraries (Galloway et al., 2014; Litmanen et al., 2020; Strandberg et al., 2015). This allowed us to estimate the cladoceran diet composition (biomass %) of diatoms, cryptomonads, chlorophytes, chrysophytes, euglenoids, dinoflagellates, cyanobacteria, raphidophytes, actinobacteria, methane oxidising bacteria (MOB), terrestrial organic matter (tPOM) and terrestrial organic matter 13652427, 2023, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/fwb.14061 by University Of Jyväskylä Library, Wiley Online Library on [30/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 756 | KEVA et al. TABLE 1 Lake, sampling date, location, size and water chemistry, and catchment area data. M/Y, month and year of the sampling, Alt, altitude (m a.s.l.), LA, Lake area (ha) CA, lake catchment area (km2), log10(CA:LA), log10transformed catchment area to lake area - ratio, AD, lake average depth (m), MD, lake maximum depth (m), chla, chlorophyll a concentration (μg/L), TotN, total nitrogen concentration (μg/L), TotP, total phosphorus concentration (μg/L), DOC, dissolved organic matter concentrations (mg/L). Urban area (Urb), agriculture area (Agr), forested area (For), wetland area (Wetl), waterbody area (Wat) indicates the landuse coverage (%) from catchment area. Last row represents an average of all lakes separated with dashed line. no. Lake M/Y Coordinates (°N; °E) Alt (m) LA (ha) CA (km2) Log10 CA:LA AD (m) MD (m) Chla (μg/L) TotN (μg/L) TotP (μg/L) DOC (mg/L) Urb (%) Agr (%) For (%) Wetl (%) Wat (%) 1Enäjärvi 8/20 60.35; 24.38 49 492.3 34.1 0.84 3.2 9.1 145.9 2,638 149 7.6 21.1 19.0 45.5 0.6 13.8 2Villikkalanjärvi 7/16 60.77; 26.03 40 717.4 411.4 1.76 2.9 8.9 22.6 1,470 120 8.5 4.0 31.0 61.8 0.3 2.9 3Sääskjärvi 7/16 60.83; 26.21 53 40.4 65.1 2.21 2.4 5.0 28.8 938 103 6.1 3.7 33.2 55.3 0.3 7.6 4Tervalampi 8/20 60.32; 24.42 48 41.1 80.4 2.29 1.2 3.5 106.9 1816 81 11.6 13.4 16.3 59.3 1.6 9.4 5Kyynäröjärvi 8/18 61.12; 24.99 95 24.4 28.3 2.06 1.3 3.0 44.9 967 77 16.9 2.6 25.5 70.3 0.3 1.4 6Tuusulanjärvi 7/18 60.44; 25.05 38 592.0 88.7 1.18 3.2 9.8 23.0 660 59 9.0 24.7 27.0 39.8 0.9 7.5 7Pyhäjärvi 7/16 60.72; 25.99 40 1,298.3 457.7 1.55 21.1 68.0 20.8 1,350 58 7.6 4.2 30.4 59.1 0.4 5.9 8Pusulanjärvi 7/16 60.64; 23.40 38 207.0 223.4 2.03 4.9 10.6 25.7 720 47 6.5 5.4 15.4 69.9 0.4 8.8 9Hiidenvesi 7/16 60.38; 24.19 32 2,909.9 931.7 1.51 6.7 29.4 15.3 1,018 38 8.6 6.7 16.2 67.4 0.8 8.8 10 Vesijärvi 7/16 61.02; 25.61 81 10,751.0 510.1 0.68 6.1 40.0 5.0 440 37 5.0 8.3 18.3 51.1 1.9 20.5 11 Kataloistenjärvi 8/18 61.02; 24.94 127 106.8 11.3 1.02 1.2 1.9 11.1 620 35 8.9 4.2 24.0 60.4 3.5 7.9 12 Alajärvi 8/16 63.01; 23.85 99 1,107.7 130.5 1.07 1.4 7.0 24.7 907 63 19.5 3.3 6.3 78.9 0.7 10.8 13 Lovojärvi 8/18 61.08; 25.03 108 4.6 6.7 2.16 7.7 17.5 24.4 785 49 13.5 22.0 31.5 40.4 4.2 1.8 14 Nimetön 8/18 61.23; 25.19 152 0.4 0.5 2.10 8.6 11.6 9.0 828 31 22.0 0.0 0.0 99.1 0.0 0.9 15 Eräjärvi 8/16 61.57; 24.61 87 835.2 58.2 0.84 2.1 10.0 14.2 610 26 7.7 3.9 15.4 64.8 3.0 12.8 16 Jyväsjärvi 7/16 62.24; 25.77 78 330.0 353.5 2.03 7.0 25.0 9.5 561 25 9.0 12.1 6.4 73.5 0.6 7.5 17 Älänne 8/16 63.48; 28.13 144 10.0 14.4 2.16 3.2 15.4 11.0 480 21 10.2 0.2 0.5 66.2 1.8 31.3 18 VähäValkjärvi 8/18 61.19; 25.09 126 2.2 0.1 0.66 1.3 4.0 20.0 581 17 6.0 0.0 0.0 75.5 0.0 24.5 19 Haarajärvi 8/18 61.23; 25.18 141 14.0 2.1 1.18 6.1 12.0 6.0 390 13 12.0 0.0 0.0 89.6 1.3 9.2 20 Haukijärvi 8/18 61.22; 25.14 131 2.1 5.6 2.43 3.8 8.0 7.0 370 11 13.0 0.6 0.0 98.5 0.5 0.4 21 Hokajärvi 8/18 61.24; 25.10 141 8.4 6.3 1.88 2.2 6.0 3.0 330 77.0 1.1 0.0 95.2 0.1 3.6 22 Hämeenjärvi 7/16 61.32; 27.27 84 130.4 14.4 1.04 4.5 11.7 6.0 300 76.2 1.9 0.8 83.6 0.4 14.4 23 SuuriVahvanen 8/16 61.69; 27.54 89 131.8 7.0 0.73 4.3 15.0 2.1 305 45.9 2.4 0.3 73.0 0.3 23.9 Average lake – – 88 859 149.6 1.54 4.6 14.5 25.5 830 47 9.9 6.3 13.8 68.6 1.0 10.2 13652427, 2023, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/fwb.14061 by University Of Jyväskylä Library, Wiley Online Library on [30/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License | 757 KEVA et al. consumed by microbes (mtPOM). The same resource groups were estimated for the seston composition (biomass %). Phytoplankton quality for cladoceran diet was categorised to three quality groups (high, moderate, low) based on longchain PUFAs and cholesterol content and on a previous laboratory feeding experiment showing differences in Daphnia growth and reproduction fed with different diets (Table S2; Peltomaa et al., 2017): Diatoms and cryptomonads, which have high EPA and DHA content and hightomoderate sterol content, were categorised as highquality diet (Peltomaa et al., 2017; Taipale, Hiltunen, et al., 2016). Moderatequality diet consists of chlorophytes and chrysophytes as they typically have moderatetolow EPA and DHA content and high sterol content (excluding Chlamydomonas), as well as euglenoids with high EPA and DHA but low sterol content (Peltomaa et al., 2017). Lowquality diet includes dinoflagellates, which have high EPA and DHA content but low sterol content, and cyanobacteria which are generally low in PUFAs and sterols, and high in saturated FAs. However, there is taxonspecific variation in the biomolecule contents within the taxonomic groups listed previously (Ahlgren et al., 1992; Peltomaa et al., 2017; Taipale et al., 2013; Taipale, Hiltunen, et al., 2016). Cladoceran diet preference was calculated as the percentage point difference between consumed (cladoceran diet composition) and available (seston composition) diet (Strauss, 1979; Taipale, Aalto, et al., 2019), where positive values indicate preference and negative values avoidance; a 20% difference between available and consumed diet was considered to be a crucial difference. 2.4 | Data analysis Most of the Finnish lakes are considered as phosphorous limited (e.g. Kauppi et al., 1993). As TotN and TotP had very high positive correlation (Figure S2), we selected only TotP to use as an indicator of eutrophication in the statistical analyses. We used lake DOC concentration as an indicator of browning. Lake TotP and DOC concentrations were clearly uncorrelated in the sampled lakes (Figure S2). We used linear regression analysis to test the relationship between lake chemistry (TotP and DOC) and seston and PUFA content (P1). Linear regression analysis also was conducted to test the relationship between lake chemistry and seston composition, cladoceran diet and cladoceran diet preference (P2). Moreover, general linear models (GLMs) were used to test if and how much lake chemistry (TotP, DOC and their interaction) and morphometry (average depth, lake area, catchment:lake area ratio) could enhance the explanatory power for seston and cladoceran PUFA and cholesterol content (P1). Selection of the variables used in GLMs was based on visual observation of the correlation matrix, where we decided to discard catchment characteristic parameters as many of them correlated heavily with lake chemistry (Figure S2). The same explanatory variables were used to estimate seston and cladoceran diet composition and cladoceran diet preference with GLMs (P2). Stepwise forward and backward selection with the Akaike information criterion (AIC) were used to select the most parsimonious GLM models. All statistical analysis were conducted in the R environment (version 3.5.3; R Core Team, 2019). Both direction stepwise GLM model selection with AIC were conducted using the MASS package (Venables & Ripley, 2002). We used an αlevel of 0.05 to test specific null hypotheses. Graphical illustrations were done with the base and ggplot2 packages (Wickham, 2016). 3 | RESULTS 3.1 | Seston and cladoceran FA content along lake chemistry and morphometrics Variation in seston PUFA content was weakly associated with lake water chemistry (Figure 2). Only lake TotP and seston LA had a positive relationship (F1,21 = 5.1, adj.r2 = 0.16, p = 0.04). The explanatory power of GLMs were slightly better compared to simple linear regressions (Table 2); however, none of the lake morphometric variables were statistically significantly related to seston biomolecule content: LA (adj.r2 = 0.20), ARA (no variables selected), ALA (adj.r2 = 0.11), EPA (no variables selected), seston DHA (no variables selected) and cholesterol (adj.r2 = 0.18; Table 2). From these, the only statistically significant model was for seston LA (p = 0.04) where TotP was positively connected (t = 2.4, p = 0.03), and DOC negatively connected (t = −1.5, p = 0.16) with seston LA (Table 2). In contrast to content data, seston PUFA concentrations (mg/L) showed a clear positive relationships with lake TotP: LA (adj.r2 = 0.51, p < 0.01), LIN (adj.r2 = 0.38, p < 0.01) and EPA (adj.r2 = 0.19, p = 0.03; Figure S7). Lake TotP was positively connected with cladoceran DHA content (F1,21 = 7.6, adj.r2 = 0.16, p = 0.04; Figure 2) and lake DOC was positively connected with cladoceran LA (F1,21 = 13.3, adj.r2 = 0.36, p < 0.01) and ALA (F1,21 = 16.5, adj.r2 = 0.41, p < 0.01; Figure 2) content. The GLM models enhanced the explanatory power of cladoceran ARA (adj.r2 = 0.23), ALA (adj.r2 = 0.46) and cholesterol (adj.r2 = 0.54) content compared to simple linear regression models. Lake area was selected to the most parsimonous cladoceran ARA and ALA models with a negative relationship. The cladoceran cholesterol content model included lake average depth and lake TotP which both had a significant positive relationship with cladoceran cholesterol content (Table 2). 3.2 | Seston algae composition and cladoceran diet along environmental variables In our study lakes, lowquality algae groups contributed most to the seston composition (pooled lakes mean ± SD: 52.4 ± 26.8%) followed by moderatequality (23.6 ± 21.6) and highquality algae groups (4.3 ± 7.7). However, this was not reflected in cladoceran diet where the estimated high- , moderateand lowquality algae diet contributions were 22.2 ± 20.3, 10.6 ± 11.9 and 33.4 ± 19.5%, respectively. Lake TotP did not have a relationship with the availability of 13652427, 2023, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/fwb.14061 by University Of Jyväskylä Library, Wiley Online Library on [30/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 758 | KEVA et al. FIGURE 2 Linear regressions between lake chemistry (a, c: TotP; b, d: DOC) and biomolecule contents of seston (a, b: mg/g POC) and cladoceran (c, d: mg/g C). Biomolecules: LA, ARA, ALA, EPA, DHA and cholesterol, presented in different rows (1– 6, respectively) of the figure. Diamonds represent values from different lakes. Linear regression lines with 95% confidence intervals (grey shaded areas) are drawn if the relationship was statistically significant. Regression equation and model statistics (F, p and adj.r2) also are presented if the model was statistically significant. In case of nonsignificant regression, only adj.r2 values are shown. adj.r2=0.02 5101520 adj.r2=0.00 5101520 adj.r2=0.01 5101520 adj.r2=0.02 5101520 adj.r2=0.01 5101520 adj.r2=0.11 5101520 y=0.53x−0.36 F(1; 21)=13.3 adj.r2=0.3; p0.06< 1 5101520 adj.r2=0.05 5101520 y=1.06x−1.37 F(1; 21)=16.5 adj.r2=0.413; p0.0<1 5101520 adj.r2=−0.02 5101520 adj.r2=−0.05 5101520 adj.r2=−0.10 5101520 )C g / gm( tnetnoc elucelomoib narecodalC )COP g / gm( tnetnoc elucelomoib notseS y=0.11x+1.76 F(1; 21)=5.1 630.0=p ;1.0=2r.jda 0 10 20 30 40 050100 150 AL adj.r2=−0.03 0 5 10 15 050100 150 ARA adj.r2=0.11 0 25 50 75 050100 150 ALA adj.r2=0.04 0 20 40 60 050100 150 APE adj.r2=−0.10 0 1 2 050100 150 loretselohC adj.r2=−0.04 0 10 20 30 050100 150 AHD μ)L/g(P-toTDOC (mg/L) μ)L/g(P-toTDOC (mg/L) (a1) (b1) (a2) (b2) (a3) (b3) (a4) (b4) (a5) (b5) (a6) (b6) adj.r2=−0.05 0 5 10 15 20 050100 150 AL adj.r2=0.01 0 5 10 15 050100 150 ARA adj.r2=−0.05 0 10 20 30 050100 150 ALA adj.r2=−0.05 0 10 20 30 050100 150 APE adj.r2=0.16 0 5 10 15 050100 150 loretselohC y=0.03x+0.37 F(1; 21)=7.6 adj.r2=0.23; p=0.01 0.0 2.5 5.0 7.5 050100 150 AHD (c1) (d1) (c2) (d2) (c3) (d3) (c4) (d4) (c5) (d5) (c6) (d6) 13652427, 2023, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/fwb.14061 by University Of Jyväskylä Library, Wiley Online Library on [30/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License | 759 KEVA et al. different seston quality groups, cladoceran diet group consumption or cladoceran diet preference (Figure 3). Lake DOC had a positive relationship with the available moderatequality seston biomass % (F1,21 = 5.5, adj.r2 = 0.17, p = 0.03; Figure 3) and a negative relationship with available lowquality seston biomass % (F1,21 = 5.4, adj.r2 = 0.17, p = 0.02; Figure 3). Similar trends were observed for the consumed cladoceran diet quality groups: a positive relationship with lake DOC and consumed moderatequality diet biomass % (F1,21 = 9.8, adj.r2 = 0.29, p < 0.01; Figure 3) and a negative relationship with lowquality diet biomass % (F1,21 = 9.9, adj.r2 = 0.29, p < 0.01; Figure 3). A positive trend with lake TotP and available tPOM in seston (F1,21 = 11.4, adj.r2 = 0.32, p < 0.01) was observed (Figure S5). Available chlorophyte contribution was positively correlated (F1,21 = 4.4, adj.r2 = 0.13, p < 0.05) and cyanobacteria negatively correlated (F1,21 = 5.1, adj.r2 = 0.16, p < 0.05) with lake DOC (Figure S6). A positive correlation between lake DOC and cladoceran chlorophyte consumption (F1,21 = 11.7, adj.r2 = 0.32, p < 0.01) was observed (Figure S6). Cladoceran diatom (F1,10 = 9.3, adj.r2 = 0.43, p < 0.01) and chlorophyte consumption (F1,10 = 14.9, adj.r2 = 0.56, p < 0.01) was positively correlated with cladoceran cholesterol content (Figure S9). The predictive power for available and consumed seston quality was enhanced with GLM models (Table S2). GLM models explained 22.2% and 26.3% of the variation in available moderateand lowquality seston biomass, respectively. DOC had a positive (t = 1.6, p = 0.13) and lake area a negative relationship (t = −1.5, p = 0.14) with moderatequality seston biomass %. DOC had a negative relationship (t = −2.5, p = 0.02) and lake average depth a positive relationship (t = 1.9, p = 0.07) with lowquality seston biomass %. Model name and statistics Variable Coefficient SE t p LA seston (mg FA/g POC) (Intercept) 7.619 4.967 1.534 0.141 F(2,20) = 3.76, p = 0.041 TotP 0.116 0.049 2.376 0.028 adj.R2 = 0.200, RSE = 8.707 DOC −0.604 0.414 −1.458 0.160 ARA seston (mg FA/g POC) (Intercept) 1.684 0.688 2.449 0.023 F(0,22), RSE = 3.298 ALA seston (mg FA/g POC) (Intercept) 5.641 6.684 0.844 0.408 F(1,21) = 3.67, p = 0.069 T o t - P 0.214 0.112 1.915 0.069 adj.R2 = 0.108, RSE = 19.968 EPA seston (mg FA/g POC) (Intercept) 9.100 3.555 2.560 0.018 F(0,22), RSE = 17.049 DHA seston (mg FA/g POC) (Intercept) 4.138 1.478 2.800 0.010 F(0,22), RSE = 7.086 Chol. seston (mg FA/g POC) (Intercept) 1.695 0.550 3.084 0.022 F(2,6) = 1.85, p = 0.237 DOC 0.053 0.041 1.298 0.242 adj.R2 = 0.175, RSE = 0.496 Log(CA/LA) −0.324 0.258 −1.257 0.255 LA cladocera (mg FA/g C) (Intercept) −0.364 1.570 - 0.232 0.819 F(1,21) = 13.30, p = 0.002 DOC 0.528 0.145 3.647 0.002 adj.R2 = 0.359, RSE = 3.044 ARA cladocera (mg FA/g C) (Intercept) 10.442 1.756 5.948 <0.001 F(1,21) = 7.58, p = 0.012 Log(LA) −0.909 0.330 −2.754 0.012 adj.R2 = 0.230, RSE = 4.044 ALA cladocera (mg FA/g C) (Intercept) 4.120 4.314 0.955 0.351 F(2,20) = 10.24, p = 0.001 DOC 0.868 0.276 3.148 0.005 adj.R2 = 0.456, RSE = 5.268 Log(LA) −0.775 0.473 −1.637 0.117 EPA cladocera (mg FA/g C) (Intercept) 12.891 1.846 6.982 <0.001 F(0,22), RSE = 8.855 DHA cladocera (mg FA/g C) (Intercept) 0.372 0.596 0.624 0.539 F(1,21) = 7.58, p = 0.012 TotP 0.027 0.010 2.753 0.012 adj.R2 = 0.230, RSE = 1.779 Chol. cladocera (mg FA/g C) (Intercept) −2.483 2.116 −1.174 0.271 F(2,9) = 7.49, p = 0.012 AD 0.633 0.208 3.042 0.014 adj.R2 = 0.541, RSE = 3.600 TotP 0.076 0.030 2.515 0.033 TABLE 2 General linear models explaining seston and cladoceran PUFA and cholesterol content. The most parsimonious models are selected with forward and backward selection applying Akaike information criterion. The most parsimonious models for each biomolecule are separated with solid horizontal lines. Fstatistics, pvalue, adjusted R2 and residual standard error (RSE) are presented for each biomolecule model under Modelcolumn. For the selected best models, variable coefficient, standard error, tvalue, and pvalue are presented. Variables with p < 0.05 are bolded. Abbreviations in Variablecolumn are Log(LA), Log(Lake area [ha]); Log(CA/ LA), Log(Catchment area to lake area ratio [km2/km2]); AD, average depth; TotP, total phosphorus (μg/L); DOC, dissolved organic matter (mg/L). 13652427, 2023, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/fwb.14061 by University Of Jyväskylä Library, Wiley Online Library on [30/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 766 | KEVA et al. Taipale, S. J., Vuorio, K., Strandberg, U., Kahilainen, K. K., Järvinen, M., Hiltunen, M., Peltomaa, E., & Kankaala, P. (2016). Lake eutrophication and brownification downgrade availability and transfer of essential fatty acids for human consumption. Environment International, 96, 156– 166. Tang, Y., Yang, X., Xu, R., Zhang, X., Liu, Z., Zhang, Y., & Dumont, H. J. (2019). Heterotrophic microbes upgrade food value of a terrestrial carbon resource for Daphnia magna. Limnology and Oceanography, 64, 474– 482. Trommer, G., Lorenz, P., Lentz, A., Fink, P., & Stibor, H. (2019). Nitrogen enrichment leads to changing fatty acid composition of phytoplankton and negatively affects zooplankton in natural lake community. Scientific Reports, 9, 16805. Venables, W. N., & Ripley, B. D. (2002). Modern applied statistics with S (fourth ed.). Springer. Vuorio, K., Järvinen, M., & Kotamäki, N. (2020). Phosphorus thresholds for bloomforming cyanobacterial taxa in boreal lakes. Hydrobiologia, 847, 4389– 4400. Watson, S. B., McCauley, E., & Downing, J. A. (1997). Patterns in phytoplankton taxonomic composition across temperate lakes of differing nutrient status. Limnology and Oceanography, 42, 487– 495. Wauthy, M., & Rautio, M. (2020). Permafrost thaw stimulates primary producers but has a moderate effect on primary consumers in subarctic ponds. Ecosphere, 11, e03099. Wenzel, A., Vrede, T., Jansson, M., & Bergström, A. K. (2020). Daphnia performance on diets containing different combinations of highquality algae, heterotrophic bacteria, and allochthonous particulate organic matter. Freshwater Biology, 66, 157– 168. Wickham, H. (2016). ggplot2: Elegant graphics for data analysis. SpringerVerlag. Xie, P., Iwakuma, T., & Fujii, K. (1998). Changes in the structure of a zooplankton community during a Ceratium (dinoflagellate) bloom in a eutrophic fishless pond. Journal of Plankton Research, 20, 1663– 1678. Yates, M. C., Derry, A. M., & Cristescu, M. E. (2021). Environmental RNA: A revolution in ecological resolution? Trends in Ecology and Evolution, 36, 501– 609. SUPPORTING INFORMATION Additional supporting information can be found online in the Supporting Information section at the end of this article. How to cite this article: Keva, O., Litmanen, J J., Kahilainen, K K., Strandberg, U., Kiljunen, M., Hämäläinen, H., & Taipale, S J. (2023). Herbivorous cladoceran essential fatty acid and cholesterol content across a phosphorous and DOC gradient of boreal lakes— Importance of diet selection. Freshwater Biology, 68, 752– 766. https://doi.org/10.1111/fwb.14061 13652427, 2023, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/fwb.14061 by University Of Jyväskylä Library, Wiley Online Library on [30/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License