Introduction

Freshwater habitats are suffering a dramatic biodiversity crisis. Nevertheless, the consequences of biodiversity loss remain largely unknown and insufficient information is available to make predictions on future status and trends of freshwater biodiversity, ecosystem functioning and services. It is hence crucial to investigate the impact of diversity loss of aquatic organisms and all the potential cascading and feedback effects on aquatic food webs.

Phytoplankton communities are composed of numerous species bearing a variety of functional traits and represent a treasure trove of biodiversity. They form the base of aquatic food webs and through photosynthesis, are responsible for roughly half of the global primary production (Falkowski, 1994). They also contributes to nutrient cycling and regulation and maintenance of all higher trophic levels in aquatic ecosystems. Phytoplankton provides important biochemical constituents to consumers such as carbohydrates, fatty acids, amino acids, sterols and vitamins (Peltomaa et al., 2017). Among these, a particular focus has been placed on essential fatty acids (EFAs), which play an important role in the development, health and reproduction of zooplankton.

EFAs must be obtained from the zooplankton’s diet because many animals cannot synthesize them de novo (Pond et al., 1996; Bell et al., 2007), although new findings are questioning this long-standing belief (Kabeya et al., 2018a; Boyen et al., 2022). The recent discovery of enzymes called "methyl end-desaturases" in many aquatic invertebrates, including cnidarians, molluscs, annelids and crustaceans, which enable them to produce PUFAs de novo endogenously, have suggested that the biosynthesis of these compounds in the aquatic environment is much more widespread than previously thought (Monroig et al., 2013; Kabeya et al., 2018a, 2021; Monroig & Kabeya, 2018; Boyen et al., 2022).

Phytoplankton biochemical traits are highly diverse and depend on both the respective algal species and their physiological condition (Fink et al., 2011; Lang et al., 2011). As a consequence, a loss or change in nutritional value of phytoplankton will impact their nutritional quality for zooplankton and consequently the production and transfer of essential biomolecules through the food web in general (Müller-Navarra et al., 2004; Lau et al., 2021). Phytoplankton functional trait diversity is therefore the key determinant of the fitness of planktonic consumers, influencing the overall functioning of pelagic ecosystems (Irwin & Finkel, 2017). In the light of biodiversity loss, it should be of primary concern to understand the relationships between the variability in phytoplankton functional diversity and their effects on zooplankton communities, food web dynamics, and ecosystem processes.

To address this question, experimental research needs to directly manipulate the diversity of taxa and traits in order to identify and interpret the effects of changing biodiversity on planktonic community functioning and aquatic ecosystems (Lacoste et al., 2001; Engel et al., 2017; Irwin & Finkel, 2017; Zhao et al., 2019; García-Girón et al., 2020; Krztoń et al., 2022). For example, phytoplankton density, biomass, composition and diversity can be directly manipulated by addition or removal of species or assembling communities with different taxonomic or functional composition and diversities. However, as phytoplankton is composed of microscopic organisms, these controlled changes become particularly challenging in laboratory or field experiments dealing with natural phytoplankton assemblages.

According to Hammerstein et al. (2017), gradients of hydrodynamic disturbance can be used as an easily manageable tool to alter the diversity of natural algal communities. There, mechanical disturbance such as mixing or shaking of natural phytoplankton communities particularly affects stress-sensitive species (Elmqvist et al., 2003; Gallagher et al., 2015), generating communities with different composition, richness, taxonomic and trait diversity.

We here used this disturbance method to simulate the loss of phytoplankton functional trait diversity by applying different intensities of mechanical disturbance to a natural lake phytoplankton community. Subsequently, we provided the altered phytoplankton communities as a food source to Daphnia longispina (O. F. Müller, 1776) and Eudiaptomus graciloides (Lilljeborg, 1888), here selected for their distinct feeding modes and diet selectivity and as representatives of the Cladocera and Copepoda, the two main taxonomical and functional zooplankton groups in freshwater environments.

The aim of the experiment was to investigate how changes in phytoplankton community composition and trait diversity may affect the development, fitness and reproductive success of herbivorous zooplankton feeding on them. Moreover, we investigated how D. longispina and E. graciloides responded to changes of phytoplankton nutritional quality, analyzing and comparing their fatty acid profiles with the lipid profiles of the altered phytoplankton communities.

Although the two grazers are quite similar in size and occupy same freshwater habitats, they exhibit distinct and specific differences in diet preferences, selectivity, reproductive strategies and life cycles. Daphnia longispina are opportunistic filter feeders that cannot select food particles individually. They only retain food items that have appropriate size to pass the filtering appendages (Geller & Müller, 1981; Gophen & Geller, 1984). Eudiaptomus graciloides on the other hand, feed much more selectively than cladocerans. They possess mechanical and chemical sensors on their mouthparts, antennae and body surface. These sensors allow them to perceive the motility and chemical composition of their dietary items (DeMott, 1986; Heuschele & Selander, 2014).

Considering this, we thus expect that, in general, the disturbance method will modify phytoplankton trait composition and diversity and this will strongly modulate and influence their palatability and accessibility for zooplankton grazing. Specifically, altered phytoplankton communities.

  1. (i)

    will affect D. longispina and E. graciloides differently, due to their different feeding strategies and selectivities

  2. (ii)

    will result in a variety of life-history trait effects/responses in both grazers, which may include variation in developmental time, body size and fecundity

  3. (iii)

    will primarily penalize daphnids, which, being unable to perceive phytoplankton nutritional quality, will require extra energy investment and efforts in food uptake through filtration to satisfy their nutritional and physiological demands

  4. (iv)

    will mainly favor copepods, which, being able to select their food particles individually, will actively choose the food items which most closely match their dietary needs.

This experiment thus aids our understanding of how loss of functional traits in the resource community could impact consumer traits and clarifies to what extent the dietary dependency occurs in D. longispina and E. graciloides. Furthermore, it tests whether consumers may have evolved adaptation strategies to fulfil their physiological requirements in response to variation or reduction of resource quality due to biodiversity loss, thereby modifying the energy transfers to higher trophic levels.

Materials and methods

Phase I: plankton sampling and disturbance method

Natural seston samples were collected in June 2020 at the surface of lake Fühlinger See, a recreational lake and former gravel pit located north of the city of Cologne in Western Germany (51° 01′ 21.0″ N 6° 55′ 48.1″ E). Plankton was filtered in situ through a 150 µm mesh to remove large zooplankton. Upon transportation to the laboratory, the samples were placed in a climate chamber at 20 ± 0.5 °C (similar to the June surface water temperature of the lake) and under 100 µE s−1 m−2 PAR for 1 day for acclimation, before evenly distributing them into fifteen 1 L borosilicate glass Erlenmeyer flasks.

The first phase of the experiment consisted of the manipulation of the phytoplankton community by applying three different disturbance intensity levels: no disturbance, intermediate disturbance and high disturbance (replicated five times each) to create phytoplankton trait diversity loss and community changes. Experimental flasks were continuously subjected to mixing using laboratory stirrers with different intensities. The “undisturbed” treatment was not mixed, phytoplankton subjected to the “intermediate disturbance” was gently mixed on a horizontal magnetic laboratory shaker with a speed set to 70 revolutions min−1, while the shaker for the “high disturbance” treatment was set to 135 revolutions min−1. Every 4 days, 25 vol. % of each sample was removed under a sterile atmosphere and replaced by fresh growth medium (1/5 of WC medium, Guillard & Lorenzen, 1972, diluted with ultrapure water). Those subsamples were subsequently analyzed for phytoplankton biomass, pigments, fatty acids and community composition (see below).

Phase II: feeding experiment and zooplankton life-history trait analysis

In the second phase of the experiment, the phytoplankton communities altered via different disturbance treatments were subsequently used as the food source for the calanoid copepod Eudiaptomus graciloides (25 newly hatched nauplii per replicate) and the cladoceran Daphnia longispina (15 neonates per replicate, in order to equalize grazer biomass between species). The animals were placed into fifteen (3 treatments, n = 5) replicate glass jars with 200 mL aged, membrane-filtered (pore size: 0.45 µm) tap water each and fed one of the three differently disturbed algal communities every second day equivalent to approx. 50,000 phytoplankton cells ml−1. All the disturbed algal communities were filtered through a 30 µm gauze before being supplied to zooplankton. During the experiment, individual body mass (µg ind−1), body size at first reproduction (SFR, µm), mass-specific somatic growth rates (MSGR, mg day−1), size-specific somatic growth rates (SSGR, mm day−1), time at first reproduction (days) and clutch size (CS, as total number of eggs produced in every food treatment and individual number of eggs per female per food treatment) were recorded as life-history traits for both zooplankton species (for details see Titocci & Fink, 2022). The carapace length for D. longispina and the prosome length of all E. graciloides egg-carrying females were measured under a stereomicroscope (Axioskop 40, Carl Zeiss Microscopy, Germany) and analyzed through the image-processing application Image J for the calculation of the SFR. The SSGR was determined as the increase in body length from the beginning of the experiment till maturity. The MSGR was calculated according to the formula:

$$ \left[ {\left( {\ln \, \left( {W_{t} } \right) - \,\ln \, \left( {W_{0} } \right)} \right)} \right]\, \times \, \, t^{ - 1}, $$

where W0 is the initial dry weight of neonates/nauplii, Wt is the weight of the individuals after reaching maturity, divided by the time to first reproduction (Lampert & Trubetskova, 1996).

Analytical methods

Biomass determination

Algal samples were filtered and collected on pre-combusted GF/F filters, dried at 60 °C for 24 h and then weighed on a microbalance (Sartorius CP2 P, accuracy 1 μg) for biomass determination. All surviving adult individuals of D. longispina and E. graciloides at the end of phase II were placed in aluminum boats and then freeze-dried, before determining their weight with the microbalance. Subsequently, the dried animals were placed in glass tubes with 5 ml dichloromethane/methanol (2:1, v/v) and stored at − 20 °C for subsequent fatty acids extraction.

Fatty acid analysis

Fatty acids were extracted from phytoplankton and zooplankton and converted to fatty acid methyl esters (FAMEs) according to Windisch and Fink (2018). The FAME composition of the samples was analyzed using a gas chromatograph (6890N GC System, Agilent Technologies, Waldbronn, Germany) equipped with a DB-225 capillary column (30 m, 0.25 mm i.d., 0.25 μm filmthickness; J&W Scientific, Folsom, CA, USA, 1 µl splitless injection), He as carrier gas and a flame ionization detector. The GC oven temperature program was as follows: initial 60 °C for 1 min, then heated at 120 °C/min to 180 °C, held for 2 min, then heated at 50 °C/min to 220 °C, constant for 13 min and finally heated at 120 °C/min to 220 °C, where temperature was held for 10 min. Individual sample fatty acids were identified based on the retention times of FAME standards (Sigma-Aldrich) and their quantification was performed with internal standards (C19:0, C23:0) and previously established calibration functions for each FAME (Couturier et al., 2020).

Pigment analysis

Samples for pigment analysis were filtered and collected on pre-combusted GF/F filters and stored dark in aluminum foil at -20 °C until further analysis. Pigments were extracted using acetone at 4 °C overnight in darkness. 50–100 µL of internal standard (trans-β-apo-8’-carotenal, Sigma-Aldrich). After extraction, each sample was centrifuged, and 1 mL of the supernatant was stored at -20 °C prior to injection into the HPLC column within 72 h after the extraction. Pigments were analyzed using the HPLC System (Prominence HPLC system; Shimadzu Co., Kyoto, Japan) with a binary pump LC-20AB, auto-sampler SIL-A20C, column oven CTO-10AC set at 40 °C, diode array detector (PDA) SPD-M20A and a reverse phase Spherisorb ODS2 column (stationary octadecyl-phase C18) with the dimensions 250 mm × 4.6 mm, particle size: 5 µm. The pigments were separated with a method modified after (Garrido & Zapata, 1993). The solvents used were methanol: 1 M ammonium acetate: acetonitrile (50:20:30, v/v, Solvent A) and acetonitrile: ethyl acetate (50:50, v/v, Solvent B). The gradient system used was as follows: 0 min: A: 90%, B: 10%; 2 min: A: 90%, B: 10%; 26 min: A: 40%, B: 60%; 28 min: A: 10%, B: 90%; 30 min: A: 10%, B: 90%. The composition of the solvents was returned to initial conditions over a 1 min gradient, followed by 2 min of system re-equilibration before the next sample was injected. The flow rate was 1 ml min−1. Absorbance was recorded in the PDA from 350 to 700 nm. Pigments were identified by the retention times and the absorption spectra, which were obtained from previous measurements of the pure pigment standards. Peak areas were integrated at 436 nm and corrected for internal standard.

Phytoplankton identification

Phytoplankton samples were fixed in Lugol's iodine solution and counted according to Utermöhl (1958). For taxonomic identification, the whole bottom area of the chamber was checked first at low magnification (100x-200x) for large forms. Small species were next counted on pairs of two perpendicular diametrical transects at higher magnification (630x). At least 400 cells were identified and counted to species level (where possible) and 20–50 filaments or colonies per species were measured and converted to single cell counts. Biovolumes were estimated using geometric shapes and equations (Hillebrand et al., 1999) and expressed in µm3 L−1 as absolute and relative biomass. The phytoplankton species were classified in major taxonomical groups and according to their morpho-functional characteristics were arranged in 7 morphology-based functional groups (MBFG) as described by Kruk et al. (2010). In the MBFG system, Group I includes all small organisms with a high surface to volume ratio, group II small flagellated organisms with a siliceous exoskeletal structure, group III is represented by large filaments (with aerotopes), group IV by organisms of medium size lacking specialized traits, group V is formed by unicellular flagellates of medium to large size, group VI consist of non-flagellated organisms with siliceous exoskeletons, and group VII is represented by organisms that form large mucilaginous colonies. Shannon diversity (H’) was calculated from phytoplankton MBFGs, pigments and fatty acids with n = 3 each.

Data analysis

All data were analyzed using R (version 3.3.3). Biomass data, Shannon indices and the zooplankton life-history traits were checked for normal distribution with a Shapiro–Wilk’s test and for homogeneity of variances with a Levene’s test. One-way and two-way ANOVA and Tukey's post-hoc tests were used to analyze and describe their changes in the different disturbance treatments and time. Permutational multivariate analysis of variance (PERMANOVA, (Anderson, 2006) was used to test whether differences in phytoplankton community composition, fatty acids and pigments composition were statistically significant among the disturbed treatments and over time. Moreover, permutational multivariate analysis of variance was used also to investigate variations in the fatty acid profile of D. longispina and E. graciloides raised by feeding the three disturbed phytoplankton communities. Non-metric multidimensional scaling (NMDS; package vegan) was used to ordinate zooplankton fatty acids based on Bray–Curtis dissimilarities. Similarity percentages (SIMPER; package vegan) were used to identify and quantify the contribution of individual and classes of fatty acids to the overall Bray–Curtis dissimilarities within treatments and grazers. Finally, correlation analysis between phytoplankton and zooplankton fatty acids was computed using R stats’s cor.test function. In calculating correlation coefficients, dietary fatty acids and consumer’s fatty acids were used in absolute values, expressed on a dry weight basis. Pearson correlation coefficients were used when data were normally distributed, and Spearman's correlation coefficient were used when variables were non-normally distributed. For all statistical analyses, P values below 0.05 were used to indicate the statistical significance. Graphics were generated using ggplot2 v3.3.3 (https://ggplot2.tidyverse.org) and dplyr v1.0.3 (https://CRAN.R-project.org/package=dplyr) in R.

Results

Phase I: phytoplankton trait alteration by disturbance

Morpho-functional groups

The natural phytoplankton community was initially dominated by Dinobryon sp., which had the highest biomass (73%) followed by unicellular flagellates of medium to large size belonging to morpho-functional group V and mainly represented by dinophytes and chlorophytes, 20% (Fig. 1a, Supplementary Table 1a). After one week of disturbance, a significant change in biomass composition was observed between the starting community and all the experimental treatments, (Permanova Time, df = 1, F = 41.001, P < 0.001) while no significant differences between disturbance treatments were observed (Permanova Treatments, df = 2, F = 2.364, P = 0.057). This was seen in the replacement of Dinobryon sp. by several taxa of diatoms (MBFG VI) and organisms of medium size lacking specialized traits (MBFG IV) in all treatments and probably caused by the transfer of a field community to laboratory conditions (different medium, light, mixing). After three and five weeks of differential disturbance, a clear shift in phytoplankton morpho-functional groups and diversity emerged between treatments (Fig. 1a). Reduced or absent turbulent mixing appeared to facilitate the potential of cyanobacteria to dominate the phytoplankton community in the undisturbed and intermediately disturbed treatments, shifting the species composition of phytoplankton communities in favor of buoyant cyanobacteria (MBFG III, 61% and 47%, respectively, Supplementary Table 1a) such as Anabaenopsis sp., Limnothrix sp., Oscillatoria sp., Planktothrix agardhii (Gomont) Anagnostidis & Komárek and organisms that form large mucilaginous colonies (MBFG VII, 12% and 3%, respectively, Supplementary Table 1a). In contrast, the intense turbulence in the highly disturbed treatment promoted the rapid decline of filamentous cyanobacteria and facilitated the increase of organisms of medium size lacking of specialized traits, mainly represented by different taxa of green algae (MBFG IV) and the spread of pennate diatoms (MBFG VI) which alone, contribute in this treatment to 75% of the total biomass in week 5 (Supplementary Table 1a). At the end of the experiment, the highest phytoplankton density was found in the undisturbed community (1.21 × 109 cells L−1, Supplementary Table 2), while the highly disturbed treatment registered the highest biomass (1.48 × 1011 µm3 L−1, Supplementary Table 2). In terms of diversity, a significant reduction in species richness and functional diversity was observed over time and between treatments (Supplementary Table 2, 3). In particular, the phytoplankton community that received no and intermediate levels of disturbance exhibited a general decrease of 6–10% of the taxa richness and the lowest functional diversity, which was registered in week 3 and 5 (Fig. 1a, Supplementary Table 2, 3).

Fig. 1
figure 1

Distribution of phytoplankton morpho-functional and biochemical traits before the manipulation (start) and throughout the course of the disturbance experiment after 1, 3 and 5 weeks of disturbance application. All the results are mean values (n = 3 for biovolume, n = 5 for fatty acids and pigments). a Phytoplankton community structure was expressed as relative biomass (% of total biovolume) of each morphologically based functional group (MBFG from I to VII, Kruk et al., 2010, top row) followed by b relative abundances of fatty acids and c pigments in the undisturbed, intermediate and highly disturbed phytoplankton communities. Shannon diversity was calculated from MBFG groups, fatty acids and pigment at the beginning of the experiment (filled diamonds, start) and for the phytoplankton community subjected to no (filled circles), intermediate (filled squares) and high (filled triangles) disturbance during weeks 1, 3 and 5 of the experiment, respectively. H’ values are mean ± standard deviation (n = 3 for functional diversity and n = 5 for fatty acid and pigment diversity)

Full size image

Biochemical traits

The changes in phytoplankton community composition and diversity mentioned above were also reflected in changes in their biochemical traits depending on disturbance (Fig. 1b, c). Indeed, the highest fatty acid diversity was registered after the first week of manipulation, both in the highly disturbed and undisturbed treatment (H′ = 2.37, 2.33 respectively, Supplementary Table 2, 3). In all treatments, a significant reduction of fatty acid diversity was registered over time (PermANOVA Time, df = 2, F = 34.063, P < 0.001, Supplementary Table 3). The highest total content of fatty acids was found in the highly disturbed community (95.7 µg per mg DW in week 3 and 80.47 µg per mg DW in week 5 Supplementary Table 2, 3) followed by the intermediate treatment which had a peak of 73.59 µg per mg DW in week 3. The undisturbed phytoplankton community showed a lower total FA content compared to the other treatments along the whole experiment (Supplementary Table 2, 3). Even though significant differences in fatty acid class composition and content were registered between the starting community and the altered communities (permANOVA Time, df = 2, F = 8.339, P < 0.001), no significant differences in fatty acid class composition were found between treatments after the first week of manipulation (permANOVA Treatment, df = 2, F = 2.08, P = 0.106) (Fig. 1b, Supplementary Table 1b). Nevertheless, changes in fatty acids composition were evident through the later phases of the experiment. Specifically, a significant decrease of polyunsaturated n-3 fatty acids was recorded in the undisturbed treatment (from week 1 to week 5, SIMPER analysis, P < 0.001), caused mainly by a dramatic drop in C20:5 n-3 (EPA), C18:4 n-3, C22:6 n-3 (DHA) and C18 fatty acids. A constant increase of monounsaturated fatty acids (MUFA) was found in the highly disturbed treatment (SIMPER analysis, P < 0.01), mainly due to contribution of single fatty acids such as C16:1 n-9 and C18:1 n-9.

After excluding Chl a, pheophytin and beta-carotene, which were the most abundant pigments and most widely occurring in all classes of algae, our pigments analyses demonstrated how the composition and content of accessory and marker pigments strongly reflected the phytoplankton taxonomic composition and subsequent variation during the disturbance experiment (Fig. 1c, Supplementary Table 1c). Significant differences in the concentrations of pigments were found between treatments and time (permANOVA treatment, df = 3, F = 7.17, P < 0.001; permANOVA time, df = 3, F = 11.36, P < 0.001). Specifically, the starting phytoplankton community showed the highest pigment diversity (H′ = 2.26) with 15 of 17 pigments detected in quite similar proportions, reflecting a more diverse and heterogeneous phytoplankton taxonomical composition. After one week of disturbance, no significant changes were detected in pigment composition between treatments. However, pigment diversity showed a gradual decrease with the lowest values reached in week 3 in the intermediate and undisturbed treatments (H′ = 0.79, 0.97, respectively, see Supplementary Table 2). Alloxanthin (3.2%), a typical cryptomonad pigment, was initially present, but almost disappeared in the later weeks. Moreover, the relative abundance of Chlorophyll c, a marker pigment for chrysophytes, was quite high in the starting phytoplankton community (5.6%), but gradually decreased over time in the other treatments corresponding to the replacement chrysophyte taxa detected through microscopic observations. Chlorophyte marker pigments (Violaxanthin, Neoxanthin, Lutein, Chlorophyll b) showed an increase in their relative abundances in all the treatments over time, matching the increasing occurrence of green algae in the later phase of the experiment. Moreover, Zeaxanthin and Echinenone, marker pigments of cyanobacteria, reached relatively high concentrations in the undisturbed and intermediate disturbed treatments (4.8% and 4.7%), but extremely low abundances in the highly disturbed treatment (0.3%) at the end of the disturbance experiment. The opposite trend was observed for Fucoxanthin, the characteristic pigment of diatoms, which contributed almost 10.2% in the highly disturbed treatment in week 5. Other diagnostic pigments for phytoplankton groups such as Peridinin, Diadinoxanthin and Diatoxanthin had relatively low concentrations, with mean contributions < 1% (Supplementary Table 1c).

Phase II: dietary impact on zooplankton

Life-history traits

The two herbivorous grazers D. longispina and E. graciloides showed a different response in growth, development and reproduction once fed the disturbance-altered phytoplankton communities (Table 1). In D. longispina, only the trait clutch size showed significant variation between food treatments. The total number of eggs produced per individual at the first brood, and consequently the potential number of offspring available for the future generations was significantly reduced when D. longispina was fed with the intermediate and undisturbed phytoplankton diets compared to those fed with the highly disturbed phytoplankton diet (one-way ANOVA; df = 2, F = 4.554, P < 0.05). E. graciloides on the other hand was strongly affected by the disturbance-altered phytoplankton communities. Several life-history traits such as clutch size (total number of eggs), body mass and age at first reproduction (one-way ANOVA, P < 0.05 in all previously mentioned traits). E. graciloides grown on the highly disturbed phytoplankton community were bigger in terms of body mass at maturation, showed higher size-specific growth rate and needed less time to became mature and reproduce for the first time, respect to the others. Moreover, E. graciloides fed the highly disturbed phytoplankton community showed the highest total egg production (one-way ANOVA, df = 2, F = 6.24, P < 0.05) while in the animals fed the intermediate disturbed treatment exhibited high variability in the reproduction phase, with two out of five replicates without egg-carrying females. At the end of the experiment, the sex ratio (♀/♂) was on average more than 1, indicating a predominance of females in favor of males in all treatments.

Table 1 Reproductive traits and life-history parameters of individuals of D. longispina and E. graciloides (mean ± SD, n = 5) fed with undisturbed, intermediately and highly disturbed phytoplankton diets
Full size table

Fatty acid composition

Twenty-seven fatty acids, from C16:0 to C24:1 n-9 were identified and compared among the two grazers fed the different food treatments. A peak, always present of uncertain identity found between C22:6 n-3 and the internal standard C23:0 was omitted from the calculations (Tables 2, 3). Two-dimensional non-metric multidimensional scaling (NMDS) ordination plots revealed a significant and clear difference in fatty acid profiles between the two grazer taxa and the dietary treatments (Fig. 2). Three distinct and separate clusters could be identified for E. graciloides fed the different algal diets, showing that the copepods’ fatty acid profiles were different in each diet treatments, while they overlapped in D. longispina. The permANOVA did not indicate significant differences in D. longispina fatty acid body content and class composition (permANOVA, df = 2, F = 0.91, P < 0.46, Table 2). For E. graciloides however, significant variations in the fatty acid content were observed (permANOVA, df = 2, F = 6.72, P < 0.001, Fig. 2). For example, the total FA content (per dry weight) was significantly higher in the calanoids fed the highly disturbed community (51.40 ± 12.83 ng per µg DW) compared to calanoids from the other treatments, which showed approx. half of the total lipid content (29.78 ± 7.04 ng per µg DW intermediate and 27.27 ± 15.43 ng per µg DW in undisturbed treatments, respectively). Moreover, except for saturated fatty acids (SAFA) where no differences were observed between treatments, the calanoids fed the highly disturbed phytoplankton community were significantly richer in n-3 and n-6 PUFA, as well as in monounsaturated fatty acids (Table 3). Copepods fed intermediately and undisturbed phytoplankton communities showed high content of SAFA (Fig. 2). When we compared the fatty acid contents between the two grazer taxa, we observed significant differences between them (permANOVA, P < 0.001). In general, the total lipid contents was higher in daphnids than in copepods and a different selective accumulation of certain fatty acids was observed (Table 2, 3). Specifically, respect to E. graciloides, D. longispina tended to accumulate the double content of EPA and its precursor stearidonic acid (SDA) once fed with the no and intermediate disturbed phytoplankton communities, while no significant differences in those fatty acids content were observed in both grazers fed the high disturbed phytoplankton community. ARA was significantly more present in daphnids than copepods (all treatments) while an opposite trend was registered for DHA, which was accumulated in almost four times higher amount in the body content of E. graciloides fed the three disturbed phytoplankton communities compared to D. longispina.

Table 2 Fatty acid composition and concentration (ng per µg dry weight) of D. longispina fed the experimental diets: no, intermediate and highly disturbed phytoplankton communities. Values expressed as means and standard deviations, with n = 5
Full size table
Table 3 Fatty acid composition and concentration (ng per µg dry weight) of E. graciloides fed the experimental diets: no, intermediate and highly disturbed phytoplankton communities. Values expressed as means and standard deviations, with n = 5
Full size table
Fig. 2
figure 2

Non-metric multidimensional scaling (NMDS, top panels) based on Bray–Curtis dissimilarities for the fatty acid (FA) profile for D. longispina (left) and E. graciloides (right) fed the experimental diets. Below, absolute abundances (ng FA/µg dry weight, mean + SD in each FA class, n = 5 per treatment) of SAFA, MUFA, n-3 and n-6 PUFA from adult D. longispina and E. graciloides fed the undisturbed (light blue) and intermediate (green) and highly (light red) disturbed phytoplankton communities. Asterisks denote significant differences (P < 0.05) between different diets

Full size image

In order to identify the FAs that have the most potential to serve as dietary trophic markers for cladocera and copepods, we compared their fatty acid composition to that of their respective diets (Fig. 3). Fatty acid composition of the highly disturbed phytoplankton community was strongly correlated with that of both grazers. The correlation between the fatty acid composition of E. graciloides and their diet was strongest for SDA (C18:4 n-3, P < 0.05, r2 = 0.83), EPA (C20:5 n-3, P < 0.01, r2 = 0.93) and C22:1 n-9 (erucic acid, P < 0.05, r2 = 0.80). We found a significantly positive correlations between D. longispina and their diet for ALA (α-linolenic acid, C18:3 n-3, P < 0.05, r2 = 0.76) and a significantly negative correlation with erucic acid (P < 0.05, r2 = 0.80). The fatty acid composition of E. graciloides matched that of their diets much more closely than for D. longispina. This was particularly evident in the highly disturbed treatment, where except for C20:0, all fatty acids from C18:2 n-6 to C22:6 n-3 in E. graciloides showed strong positive correlations (r2 > 0.5) with the lipid composition of the respective diet, while for D. longispina, correlation coefficients were on average lower in all treatments. In the undisturbed treatment, D. longispina showed high positive correlation with two monounsaturated fatty acids: C22:1 n-9 and C16:1 n-7, and negative correlation with ALA and DHA, while E. graciloides showed a significant positive correlation in C18:1 n-9c (P < 0.05, r2 = 0.80), followed by C20:3 n-3 and DHA even though the slope of the relationship were not significant. Moreover, in some cases a mismatch between the nutritional content of the diets and the biochemical composition of the consumers were found. In fact, although in low concentration, a general presence of C18:1 n-9t, C20:1 n-7, C21:0 and C20:4 n-6 has been observed in D. longispina body content despite the absence of those fatty acids in their diets. For E. graciloides this result was even more pronounced. Indeed, nine further fatty acids (C18:1 n-9t, C20:0, C21:0, C20:1 n-7, C20:2 n-6, C20:3 n-5, C20:3 n-6, C20:4 n-6 and C24:1) were recorded in E. graciloides despite their absence in their respective phytoplankton diets.

Fig. 3
figure 3

Correlations between phytoplankton and zooplankton fatty acid (FA) content and composition. Correlogram between fatty acid content of D. longispina and E. graciloides with their respective dietary FA from the undisturbed, intermediately and highly disturbed phytoplankton communities. The color intensity and size of the point is proportional to the absolute value of the Pearson/Spearman correlation coefficient, the stronger the correlation (i.e., the closer to − 1 or 1), the darker and larger the points. The color legend shows a negative correlation (red color) when the two variables varied in opposite directions, and a positive correlation (blue color) when the two variables varied in the same direction. Significant correlations (P < 0.05) are indicated by a white asterisk

Full size image

Discussion

Phase I: phytoplankton trait alteration by disturbance

In our experiment, alterations in the physical environment, induced by the application of different disturbance intensities, resulted in profound shifts in composition and structure of a natural phytoplankton community. These responses depended on the responses of the single taxa and the extent to which their tolerances to mixing conditions contributed to subsequent variations in dominance in their respective populations. Despite receiving the same environmental conditions in terms of light, temperature and dissolved nutrients, the capacity to accumulate biomass varied significantly across taxa. Phytoplankton trait dynamics differed depending on the hydrodynamic mixing regime. In particular, diatoms and motile algae of medium size lacking specialized traits consistently dominated under sustained mixing, while buoyant species and large mucilaginous colonies built the majority of communities at low disturbance levels. These responses resembled those that occur in natural freshwater environments, where seasonal phytoplankton succession is basically controlled by the physical environment and water circulation (movement) represents one of the main factor in conditioning the availability of light and nutrient resources for planktonic organisms (Margalef, 1978). Hydrodynamic plays a crucial role in phytoplankton succession, where stronger hydrodynamic forces typically shift phytoplankton communities towards large species (cells, colonies or long filaments) and low surface area:volume ratio (Reynolds, 1984). In our experiment, high turbulence strongly influenced and re-structured phytoplankton composition favoring taxa with increased size and cell volume, which resulted in a high community biomass, without a numerical increase. In the highly disturbed treatment, almost 75% of the algal biomass were represented by medium to large sized pennate diatoms. This corresponds to previous findings that turbulent environments typically favor larger phytoplankton species (Lepistö & Rosenström, 1998; Arin et al., 2002; Rokkan Iversen et al., 2010). Under low disturbance conditions, the strategies of cell division and chain elongation as well as cell adhesion and mucilage formation appeared to dominate. In the undisturbed and intermediately disturbed communities, we observed the highest densities and a predominance of colonial and filamentous cyanobacteria. Indeed, mucilaginous sheaths coupled with the presence of gas vacuoles appear to be beneficial traits for organisms living in calm or intermediate disturbed water, allowing them to enhancing buoyancy and floating movements for a better utilization of light and nutrients (Fogg & Walsby, 1971; Reynolds & Walsby, 1975). Moreover, the high occurrence of nitrogen fixing cyanobacteria like Pseudoanabaena, Chroococcus, Oscillatoria, more frequently found in the undisturbed treatment as layers of thin mats attached to flask substrates, confirmed their ecological adaptation to slow moving waters (Marcarelli et al., 2008).

On the contrary, other species like Dinobryon sp. declined rapidly within the first week of the experiment. This was probably not due to mechanical stress induced by the hydrodynamic disturbance, as this occurred in all treatments irrespective of the disturbance treatment. In this case, unfavorable laboratory conditions such as light and cultivation medium may have not met the nutritional requirements of this species. Dinobryon, like many chrysophytes, are mixotrophs, a characteristic particularly favorable under nutrient-poor conditions. Thus, the enriched nutrient conditions supplied in our experiment may have put Dinobryon at a competitive disadvantage in relation to other taxa, disfavoring their growth and survival (Sandgren, 1988).

Overall, in our experiment, hydrodynamic stress strongly influenced changes in species dominance, composition and traits diversity, with the high disturbed community showing in general the highest values of richness and functional diversity. A similar pattern emerged for the biochemical trait of the phytoplankton’s pigments. Specifically, a clear match between the pigment and the taxonomic composition was particularly evident in the latest phase of the experiment. Zeaxanthin and Echinenone, marker pigments of cyanobacteria, reached relatively high concentrations in the undisturbed and intermediate disturbed treatments and Fucoxanthin, the characteristic pigment of diatoms contributed greatly in the highly disturbed treatment. The similar changing dynamics observed clearly demonstrated how the classical taxonomic identification together with the use of pigments as biomarkers are both valid and robust methods for phytoplankton functional determination and community dynamics investigation (Mackey et al., 1996; Schlüter et al., 2000; Irigoien et al., 2004; Ilić et al., 2023).

The highest total content of FA was registered in the highly (week 3 and week 5) and intermediately (week 3) disturbed communities. Those exhibited more than double the FA content compared to the starting community and the undisturbed treatment. This suggest that, intermediate and high levels of mixing contributed to enhance the chemical composition and nutritional efficient in terms of fatty acids content in these treatments. Although, in our understanding, there are no evidences that hydrodynamic disturbance may enhance lipid productivity, from our result we can suppose that stress conditions, and specifically, mixing may have favored more suitable conditions of circulation and exchange of the nutrients in the cultures respect to the unmixed treatment, implying a possible acceleration in phytoplankton metabolism with consequent increase of activation of biochemical pathways for fatty acids synthesis and storage. This find some correspondence with several studies which reported the synergistic effect of several stress conditions on the improvement of lipid productivity in selected algal species (Ho et al., 2014; Singh et al., 2015; Kwak et al., 2016).

Even though no significant differences in fatty acid class composition were found between treatments after the first week of manipulation, significant changes in FA classes’ proportion were registered over time. In particular, a constant relative increase of MUFA was found during the late phase of the experiment, especially in the highly disturbed treatment. This was mainly due for the contribution of few single peaks of monounsaturated fatty acids such as C16:1 n-7, C16:1 n-9 and C18:1 n-9, commonly recognized as diatom and flagellate lipid biomarkers (Reuss & Poulsen, 2002; Taipale et al., 2013). This corresponds to the observed dominance of diatoms (75% of total biomass at the end of the experiment) and contributed to the lowest fatty acid diversity in this treatment. In contrast to this, n-3 PUFA decreased significantly in the undisturbed treatment. This can be explained by the gradual increase of cyanobacteria and chlorophytes in this treatment. Indeed, these two phytoplankton groups are known to be poor producers of PUFAs and are commonly classified as non-EPA and non-DHA-synthesizers (Ahlgren et al., 1990; Taipale et al., 2013; Jónasdóttir, 2019).

In general, alteration of species distribution and dominance due to the disturbance levels have been strongly reflected in variation in phytoplankton traits composition and dynamics along the experiment highlighting even more the importance of investigating not only taxonomic but also functional diversity of competing species in order to understand the effects of biodiversity and even more of biodiversity loss across trophic levels (Duffy et al., 2007).

Phase II: dietary impact on zooplankton

In the second phase of the experiment, the disturbance-altered phytoplankton communities had distinct impacts on functional traits and fitness of the two grazers D. longispina and E. graciloides, mainly as consequence of their feeding strategies.

As expected, phytoplankton community changes strongly modulated the algal palatability and accessibility for E. graciloides. Thanks to mechanical and chemical sensors, E. graciloides is able to perceive the chemical composition and movements of its prey (DeMott, 1986). They actively select and choose food items which most closely match their metabolic requirements in the altered phytoplankton communities. In particular, due to their selective feeding mode, copepods have resulted exceptionally well adapted to optimize their nutrition in the highly disturbed community, which provided them the highest fitness. In fact, under a mainly diatom-dominated diet, the copepods exhibited higher body mass, and a higher occurrence of ovigerous females in a shorter period of time that were subsequently able to produce a higher number of eggs compared to copepods in the other food treatments. Moreover, after analyzing the biochemical composition of E. graciloides, almost the double total fatty acid content was found in individuals fed the highly disturbed community than in individuals fed and grown on the undisturbed and intermediate disturbed communities as their diet.

In particular, the dominance of diatoms accounted for the high amount of PUFA n-3, PUFA n-6 and the elevated amounts of C16:1 n-7, a commonly recognized diatom biomarker, found in the body of calanoids fed with the highly disturbed treatment. This confirmed the ability of E. graciloides to select the most nutritional diet item from a mixed phytoplankton community. Furthermore, this supports the general view that diatom-dominated communities promote high biomass, growth and reproduction in calanoid copepods (Legendre, 1990; Kleppel, 1993) and that PUFAs and long-chain fatty acids are fundamental constituents for oogenesis in crustaceans in general (Payne & Rip**ale, 2000; Broglio et al., 2003) and for copepod reproduction in particular (Støttrup & Jensen, 1990; Jónasdóttir, 1994). Moreover, in our experiment, the highly disturbed community was also the one with the highest functional diversity compared to the other diets with lower taxonomic richness and biodiversity, supporting the idea that dietary diversity increases the probability of obtaining a nutritionally complete dietary mixture for herbivores (Groendahl & Fink, 2016).

In contrast, calanoids fed the undisturbed and intermediate disturbed community showed lower body weight, longer time needed for maturation and reproduction and approximately half of the total lipid content and higher accumulation of saturated fatty acids in their body. Since most animals cannot (efficiently) synthesize PUFAs de novo, the availability of PUFA is largely dependent by the phytoplankton community composition (Strandberg et al., 2015). The undisturbed and intermediate disturbed diets were mainly composed by green algae and cyanobacteria. These phytoplankton groups are typically characterized by their lack of long-chain PUFAs (Ahlgren et al., 1990; Brown et al., 1997; Lacoste et al., 2001; Von Elert & Wolffrom, 2001). This nutritional deficiency may thus have caused a strongly negative impact on copepod reproduction (Von Elert & Stampfl, 2000). In these treatments, due to a poorer quality and availability of palatable resources, E. graciloides could not maximize all components of fitness (growth, development and reproduction) simultaneously. In this trade-off, they apparently invested more into survival and growth, rather than into reproduction. This was further confirmed by the lowest total egg production registered in these treatments and with the finding of egg-carrying females in only two out of five replicates in the intermediate disturbed treatment. Moreover, our results corroborate those of previous experiments where Eudiaptomus sp. exhibited low egg production, hatching success and survival rates when fed monoalgal diets of Chlamydomonas klinobasis (Skuja) Fott, Tetradesmus obliquus (Turpin) M.J. Wynne, and Synechococcus elongatus (Nägeli) Nägeli as representatives of green algae and cyanobacteria (Titocci & Fink, 2022).

In contrast to E. graciloides, the disturbance-dependent alteration of the phytoplankton community did neither significantly affect life-history traits, nor FA composition of D. longispina. This can probably be explained by the different feeding strategies of the two grazers. Although the two crustacean zooplankton grazers are quite similar in size and therefore they can be quite equivalent in terms of grazing pressure and biomass, they exhibit completely distinct feeding modes, selectivity and diet preferences. D. longispina are unselective filter feeders, and thus unable to exercise a specific impact based on the phytoplankton organisms’ nutritional quality, but only as function of numeric cell density (Barnett et al., 2007; Kiørboe, 2011). As a consequence, unselective filter-feeding may be a disadvantageous feeding strategy in a scenario of phytoplankton diversity loss.

Consumers’ fatty acid composition generally reflect the FA composition of their diets (Iverson, 2009). However, the fatty acid composition of E. graciloides matched that of their diets much more closely than the composition of D. longispina. This was particularly evident in the highly disturbed treatment, suggesting once more how the selective feeding favors a preferential direct assimilation on more nutritious cells within higher food quality diets. Moreover, our FA analyses showed that in general, zooplankton had a significantly higher proportion of essential fatty acids than phytoplankton, suggesting a preferential retention of EFAs in both grazers (Kainz et al., 2004; Brett et al., 2009b), despite some differences in PUFA accumulations between the two taxa.

D. longispina accumulated twice as much EPA and its precursor SDA when fed with low PUFA communities. This corroborates the generally acknowledged key role of EPA for cladoceran fitness (e.g., von Elert, 2002; Abrusán et al., 2007; Windisch & Fink, 2018) and that FA internal transformation, mobilization and bioconversion mechanisms when essential fatty acids availability was limiting (Bell & Tocher, 2009; Twining et al., 2021) have been occurred. Moreover, as previously reported (Persson & Vrede, 2006; Smyntek et al., 2008; Brett et al., 2009a), ARA was significantly more present in D. longispina than E. graciloides (all treatments) while an opposite trend was registered for DHA, which was accumulated in almost four times higher amount in the body content of E. graciloides fed the three disturbed phytoplankton communities compared to D. longispina. This is in agreement with previous findings showing that copepods have the ability to synthesize or regulate DHA (Ravet et al., 2010; Kabeya et al., 2021) and confirm the eco-physiological role of docosahexaenoic acid for copepod growth and reproduction (Kainz et al., 2004; Arts et al., 2009a; Chen et al., 2012; Deschutter et al., 2019). The consumer-specific PUFA demands may have prompted transformations in consumers’ fatty acids metabolism to actively regulate and/or convert some missing dietary fatty acids and to compensate for the low availability and abundance of PUFAs in order to face the altered biochemical composition of the diets and maximize consumer’s fitness. Even though it is generally believed that the ability of crustacean zooplankton to bioconvert and modify fatty acids is scarce (Sargent & Falk-Petersen, 1988; Bell et al., 2007) the presence in our experiment, although in low concentration, of several fatty acids in both grazers’ tissues despite the absence in their respective phytoplankton diets, support even more the hypothesis that D. longispina and E. graciloides have modified fatty acids via internal transformation, such as selective FA retention and bioconversion, by elongating and/or desaturating shorter chained fatty acid precursors into PUFAs (Bell & Tocher, 2009; Boyen et al., 2022) during the experiment. Findings in support to this hypothesis are still very controversial, and our understanding of selective PUFAs accumulation, allocation and retention in freshwater zooplankton is still limited. In general, several studies revealed a mismatch between FA composition of zooplankton and its food (Desvillettes et al., 1997; Hessen & Leu, 2006; Persson & Vrede, 2006; Smyntek et al., 2008; Arts et al., 2009b; Taipale et al., 2011) and recent advances in genomic analyses have proved the occurrence of some desaturases enzyme in a plethora of invertebrates, which would enable them to biosynthesize PUFAs de novo (Monroig & Kabeya, 2018), corroborating the idea that PUFAs synthesis and conversion may be more widespread in the animal kingdom than previously assumed (Kabeya et al., 2018b, 2021; Nielsen et al., 2019; Boyen et al., 2020, 2022; Twining et al., 2021). Thus, phytoplankton fatty acid composition plays overall an important role in regulating the life-history traits of herbivorous zooplankton, however, biochemical deficiencies of the food may imply the potential endogenous PUFA synthesis and bioconversion independently of the diet’s composition, with consumers facing higher metabolic costs, and potentially associated reduced fitness. Thus, it becomes of primary importance to understand the ability and the extent of consumers to modify the dietary fatty acids (Galloway & Budge, 2020; Jardine et al., 2020).

In our experiment, the key nutritional contrasts in the altered phytoplankton communities generated by the disturbance method have influenced and driven some behavioral and/or metabolic adaptation in consumers, demonstrating that a very plastic response in lipid transfer from phytoplankton to zooplankton may occur in freshwater environments. In the prevalent patterns of biodiversity loss and environmental changes, this may have large-scale implications for food web dynamics in aquatic environments.

Conclusions

Our disturbance method was able to alter for the taxonomic and trait composition of a natural phytoplankton community. It is thus a useful tool for the simulation of biodiversity loss and in particular the loss of stress-sensitive species. The profound shifts in phytoplankton composition and structure subsequently modulated the fitness and lipid composition of higher trophic levels represented by the two grazers D. longispina and E. graciloides via their distinct feeding modes and diet selectivities. The fatty acid composition of the phytoplankton diets and zooplankton consumers matched only partially. This suggests that consumers actively transform dietary fatty acids to adjust the dietary PUFA composition to their physiological needs under unfavorable conditions. Overall, this study highlights how a loss of functional traits in the resource community could impact consumer traits and eventually may lead to community re-organization and ecological adaptation in trophic dynamics under environmental change and biodiversity loss.