Strain-specific metabarcoding reveals rapid evolution of copper tolerance in populations of the coastal diatom Skeletonema marinoi.

Phytoplankton have short generation times, flexible reproduction strategies, large population sizes and high standing genetic diversity, traits that should facilitate rapid evolution under directional selection. We quantified local adaptation of copper tolerance in a population of the diatom Skeletonema marinoi from a mining-exposed inlet in the Baltic Sea and in a non-exposed population 100 km away. We hypothesized that mining pollution has driven evolution of elevated copper tolerance in the impacted population of S. marinoi. Assays of 58 strains originating from sediment resting stages revealed no difference in the average tolerance to copper between the two populations. However, variation within populations was greater at the mining site, with three strains displaying hyper-tolerant phenotypes. In an artificial evolution experiment, we used a novel intraspecific metabarcoding locus to track selection and quantify fitness of all 58 strains during co-cultivation in one control and one toxic copper treatment. As expected, the hyper-tolerant strains enabled rapid evolution of copper tolerance in the mining-exposed population through selection on available strain diversity. Within 42 days, in each experimental replicate a single strain dominated (30%-99% abundance) but different strains dominated the different treatments. The reference population developed tolerance beyond expectations primarily due to slowly developing plastic response in one strain, suggesting that different modes of copper tolerance are present in the two populations. Our findings provide novel empirical evidence that standing genetic diversity of phytoplankton resting stage allows populations to evolve rapidly (20-50 generations) and flexibly on timescales relevant for seasonal bloom progressions.

Behaviour of freshwater snails (Radix balthica) exposed to the pharmaceutical sertraline under simulated predation risk.

Due to their potential for affecting the modulation of behaviour, effects of selective serotonin reuptake inhibitors (SSRIs) in the environment are particularly interesting regarding interspecies interactions and non-consumptive effects (NCEs) induced by predator cues in prey organisms. We evaluated the effects of sertraline (0.4, 40 ng/L, 40 µg/L) over 8 days on activity and habitat choice in the freshwater snail Radix balthica, on snails' boldness in response to mechanical stimulation (simulating predator attack), and their activity/habitat choice in response to chemical cues from predatory fish. We hypothesised that sertraline exposure would detrimentally impact NCEs elicited by predator cues, increasing predation risk. Although there were no effects of sertraline on NCEs, there were observed effects of chemical cue from predatory fish on snail behaviour independent of sertraline exposure. Snails reduced their activity in which the percentage of active snails decreased by almost 50% after exposure to fish cue. Additionally, snails changed their habitat use by moving away from open (exposed) areas. The general lack of effects of sertraline on snails' activity and other behaviours in this study is interesting considering that other SSRIs have been shown to induce changes in gastropod behaviour. This raises questions on the modes of action of various SSRIs in gastropods, as well as the potential for a trophic "mismatch" of effects between fish predators and snail prey in aquatic systems.

Transcriptional Responses as Biomarkers of General Toxicity: A Systematic Review and Meta-analysis on Metal-Exposed Bivalves.

Through a systematic review and a series of meta-analyses, we evaluated the general responsiveness of putative transcriptional biomarkers of general toxicity and chemical stress. We targeted metal exposures performed on bivalves under controlled laboratory conditions and selected six transcripts associated with general toxicity for evaluation: catalase, glutathione-S-transferase, heat shock proteins 70 and 90, metallothionein, and superoxide dismutase. Transcriptional responses (n = 396) were extracted from published scientific articles (k = 22) and converted to log response ratios (lnRRs). By estimating toxic units, we normalized different metal exposures to a common scale, as a proxy of concentration. Using Bayesian hierarchical random effect models, we then tested the effects of metal exposure on lnRR, both for metal exposure in general and in meta-regressions using toxic unit and exposure time as independent variables. Corresponding analyses were also repeated with transcript and tissue as additional moderators. Observed patterns were similar for general and for transcript- and tissue-specific responses. The expected overall response to arbitrary metal exposure was an lnRR of 0.50, corresponding to a 65% increase relative to a nonexposed control. However, when accounting for publication bias, the estimated "true" response showed no such effect. Furthermore, expected response magnitude increased slightly with exposure time, but there was little support for general monotonic concentration dependence with regard to toxic unit. Altogether, the present study reveals potential limitations that need consideration prior to applying the selected transcripts as biomarkers in environmental risk assessment. Environ Toxicol Chem 2023;42:628-641. © 2022 The Authors. Environmental Toxicology and Chemistry published by Wiley Periodicals LLC on behalf of SETAC.

Effects of 17α-ethinylestradiol on individual life-history parameters and estimated population growth rates of the freshwater gastropods Radix balthica and Bithynia tentaculata.

Studies of aquatic environments exposed to 17α-ethinylestradiol (EE2) have demonstrated detrimental effects on fish communities. However, much less is known about effects on macro-invertebrates and especially how long-term exposure may affect critical life stages and ultimately population dynamics. We studied the effects of EE2 on relevant endpoints for population growth in two common freshwater gastropods, Radix balthica and Bithynia tentaculata, that differ in reproductive, foraging and anti-predator strategies and endocrine systems. We quantified critical life-history parameters (mortality, somatic growth rate, days to and size at first reproduction, egg production and hatching success) in a concentration-response, life cycle experiment. The two species responded to EE2 exposure in different ways, B. tentaculata showing a significantly lower somatic growth rate and R. balthica a higher somatic growth rate. However, the magnitudes of the effects were small and EE2 exposure did not have any significant effect on estimated population growth rates for the two snail species. The significant effects of EE2 on individual endpoints, but not on population growth rate for both species, emphasise the importance of evaluating higher level effects from long-term exposure studies.

Molecular biomarker responses in the freshwater mussel Anodonta anatina exposed to an industrial wastewater effluent.

Using a selection of molecular biomarkers, we evaluated responses in freshwater mussels (Anodonta anatina) exposed to effluent from an industrial wastewater treatment facility. The aims of this work were to (1) assess biomarkers of general toxicity under sublethal exposure to an anthropogenic mixture of chemicals, represented by an arbitrary effluent, and (2) evaluate the potential of A. anatina as a bioindicator of pollution. Adult mussels (n = in total 32; 24 males and 8 females) were exposed (96 h) in the laboratory to a fixed dilution of effluent or to a control treatment of standardized freshwater. Metal concentrations were in general higher in the effluent, by an order of magnitude or more, compared to the control. Toxic unit estimates were used as proxies of chemical stress, and Cu, Ni, and Zn were identified as potential major contributors (Cu> Ni > Zn). Six transcriptional (cat, gst, hsp70, hsp90, mt, sod) and two biochemical (AChE, GST) biomarkers were analyzed in two tissues, gills, and digestive glands. Out of the 16 responses (eight biomarkers × two tissues), 14 effect sizes were small (within ± 28 % of control) and differences non-significant (p > 0.05). Results did however show that (1) AChE activity increased by 40% in gills of exposed mussels compared to control, (2) hsp90 expression was 100% higher in exposed female gills compared to control, and (3) three marker signals (AChE in both tissues, and hsp70 in gills) differed between sexes, independent of treatment. Results highlight a need for further investigation of molecular biomarker variability and robustness in A. anatina.

Evaluation of transcriptional biomarkers using a high-resolution regression approach: Concentration-dependence of selected transcripts in copper-exposed freshwater mussels (Anodonta anatina).

We tested concentration-dependence of selected gene transcripts (cat, gst, hsp70, hsp90, mt and sod) for evaluation as biomarkers of chemical stress. Contrary to the common approach of factorial designs and few exposure concentrations, we used regression across a high-resolution concentration series. Specifically, freshwater mussels (Anodonta anatina) were acutely (96 h) exposed to Cu (13 nominal concentrations, measuring 0.13-1 600 µg/L), and transcripts were measured by RT-qPCR. In digestive glands, cat, hsp90 and mt decreased with water Cu (p < 0.05), but response magnitudes saturated at < 2-fold decreases. In gills, gst, hsp70, hsp90 and mt increased with water Cu (p < 0.05). While hsp70, hsp90 and mt exceeded 2-fold increases within the exposure range, high Cu concentrations were required (38-160 µg/L). Although gill responses were generally more robust compared to digestive glands, overall small response magnitudes and moderate sensitivity may set limit for potential application as general biomarkers of chemical stress.

Intraspecific variation in metal tolerance modulate competition between two marine diatoms.

Despite widespread metal pollution of coastal ecosystems, little is known of its effect on marine phytoplankton. We designed a co-cultivation experiment to test if toxic dose-response relationships can be used to predict the competitive outcome of two species under metal stress. Specifically, we took into account intraspecific strain variation and selection. We used 72 h dose-response relationships to model how silver (Ag), cadmium (Cd), and copper (Cu) affect both intraspecific strain selection and competition between taxa in two marine diatoms (Skeletonema marinoi and Thalassiosira baltica). The models were validated against 10-day co-culture experiments, using four strains per species. In the control treatment, we could predict the outcome using strain-specific growth rates, suggesting low levels of competitive interactions between the species. Our models correctly predicted which species would gain a competitive advantage under toxic stress. However, the absolute inhibition levels were confounded by the development of chronic toxic stress, resulting in a higher long-term inhibition by Cd and Cu. We failed to detect species differences in average Cu tolerance, but the model accounting for strain selection accurately predicted a competitive advantage for T. baltica. Our findings demonstrate the importance of incorporating multiple strains when determining traits and when performing microbial competition experiments.

Antipredator phenotype in crucian carp altered by a psychoactive drug.

Predator-inducible defenses constitute a widespread form of adaptive phenotypic plasticity, and such defenses have recently been suggested linked with the neuroendocrine system. The neuroendocrine system is a target of endocrine disruptors, such as psychoactive pharmaceuticals, which are common aquatic contaminants. We hypothesized that exposure to an antidepressant pollutant, fluoxetine, influences the physiological stress response in our model species, crucian carp, affecting its behavioral and morphological responses to predation threat. We examined short- and long-term effects of fluoxetine and predator exposure on behavior and morphology in crucian carp. Seventeen days of exposure to a high dose of fluoxetine (100 µg/L) resulted in a shyer phenotype, regardless of the presence/absence of a pike predator, but this effect disappeared after long-term exposure. Fluoxetine effects on morphological plasticity were context-dependent as a low dose (1 µg/L) only influenced crucian carp body shape in pike presence. A high dose of fluoxetine strongly influenced body shape regardless of predator treatment. Our results highlight that environmental pollution by pharmaceuticals could disrupt physiological regulation of ecologically important inducible defenses.

Toxicity of fungicides to natural bacterial communities in wetland water and sediment measured using leucine incorporation and potential denitrification.

We assessed potential toxicity of fungicides to natural bacterial communities from a constructed wetland, located in southern Sweden, and compared the sensitivity of two endpoints indicating bacterial activity, leucine incorporation, and potential denitrification, in detecting toxicity. The effects of eight fungicides (benomyl, carbendazim, carboxin, captan, cycloheximide, fenpropimorph, propiconazole, and thiram), two bactericides (bronopol and chlortetracycline) as controls, and one reference compound (3,5-dichlorophenol), were tested in a water-sediment microcosm set-up. Leucine incorporation was measured in both the water and sediment column, while potential denitrification was measured for the entire microcosm. The bactericides and the reference compound gave sigmoid concentration-response curves for both endpoints in all but one case. The fungicides thiram, captan, and benomyl, and to a lesser extent fenpropimorph and propiconazole had quantifiable toxic effects on leucine incorporation, with EC(50) values ranging from 3 to 70 mg l(-1), while carbendazim, carboxin, and cycloheximide had little effect at the investigated concentrations. Only thiram and captan inhibited potential denitrification; the other fungicides showed no quantifiable effect. A greater toxic effect on leucine incorporation was recorded for bacterial communities associated with the water column, compared to the sediment column, for all tested compounds. Leucine incorporation was the more sensitive method for toxicity assessment of bacterial communities, and also allowed for a rapid and simple way of comparing exposure in the sediment and water column, making it an attractive standard method for community based toxicological assays in aquatic environments.

Transcriptional and biochemical biomarker responses in a freshwater mussel (Anodonta anatina) under environmentally relevant Cu exposure.

Molecular biomarkers, like gene transcripts or enzyme activities, are potentially powerful tools for early warning assessment of pollution. However, a thorough understanding of response and baseline variation is required to distinguish actual effects from pollution. Here, we assess the freshwater mussel Anodonta anatina as a biomarker model species for freshwater ecosystems, by testing responses of six transcriptional (cat, gst, hsp70, hsp90, mt, and sod) and two biochemical (AChE and GST) biomarkers to environmentally relevant Cu water concentrations. Mussels (n = 20), collected from a stream free from point source pollution, were exposed in the laboratory, for 96 h, to Cu treatments (< 0.2 µg/L, 0.77 ± 0.87 µg/L, and 6.3 ± 5.4 µg/L). Gills and digestive glands were extracted and analyzed for transcriptional and biochemical responses. Biological and statistical effect sizes from Cu treatments were in general small (mean log2 fold-change ≤ 0.80 and Cohen's f ≤ 0.69, respectively), and no significant treatment effects were observed. In contrast, four out of eight biomarkers (cat, gst, hsp70, and GST) showed a significant sex:tissue interaction, and additionally one (sod) showed significant overall effects from sex. Specifically, three markers in gills (cat, mt, GST) and one in digestive gland (AChE) displayed significant sex differences, independent of treatment. Results suggest that sex or tissue effects might obscure low-magnitude biomarker responses and potential early warnings. Thus, variation in biomarker baselines and response patterns needs to be further addressed for the future use of A. anatina as a biomarker model species.

Differences in metal tolerance among strains, populations, and species of marine diatoms - Importance of exponential growth for quantification.

Strains of microalgae vary in traits between species and populations due to adaptation or stochastic processes. Traits of individual strains may also vary depending on the acclimatization state and external forces, such as abiotic stress. In this study we tested how metal tolerance differs among marine diatoms at three organizational levels: species, populations, and strains. At the species level we compared two pelagic Baltic Sea diatoms (Skeletonema marinoi and Thalassiosira baltica). We found that the between-species differences in tolerance (EC50) to the biologically active metals (Cu, Co, Ni, and Zn) was similar to that within-species. In contrast, the two species differed significantly in tolerance towards the non-essential metals, Ag (three-fold higher in T. baltica), Pb and Cd (two and three-fold higher in S. marinoi). At the population level, we found evidence that increased tolerance against Cu and Co (17 and 41 % higher EC50 on average, respectively) had evolved in a S. marinoi population subjected to historical mining activity. On a strain level we demonstrate how the growth phase of cultures (i.e., cellular densities above exponential growth) modulated dose-response relationships to Ag, Cd, Co, Cu, and Zn. Specifically, the EC50's were reduced by 10-60 % in non-exponentially growing S. marinoi (strain RO5AC), depending on metal. For the essential metals these differences were often larger than the average differences between the two species and populations. Consequently, without careful experimental design, interactions between nutrient limitation and metal stress may interfere with detection of small, but evolutionary and ecologically important, differences in tolerance between microalgae. To avoid such artifacts, we outline a semi-continuous cultivation approach that maintains, and empirically tests, that exponential growth is achieved. We argue that such an approach is essential to enable comparison of population or strain differences in tolerance using dose-response tests on cultures of microalgae.

Persistent organic pollutants (POP) in a benthic omnivore--a comparison between lake and stream crayfish populations.

We investigated the accumulation of PCB and DDT in crayfish populations in 10 streams in southern Sweden. The results were compared with an earlier study on crayfish in lakes from the same area. We found that the concentration of pollutants in crayfish did not differ between the two types of systems. Variation in body burden was higher in stream living crayfish probably because of the higher influence from pollutants deposited in the catchment area and the more dynamic transport in streams. In streams, p,p'-DDE concentrations were positively correlated to trophic status (total phosphorous) while PCB did not show any correlation with the nutrient regime. Further, mean SigmaPCB and p,p'-DDE concentrations in crayfish did not correlate in streams. We suggest that the sources of the two pollutants differ for stream living crayfish. The results indicate that crayfish in streams are affected to a higher degree to pollutants in the catchment area and the precipitation regime. In lakes, internal processes govern uptake of pollutants in crayfish.

Bioaccumulation and trophodynamics of the antidepressants sertraline and fluoxetine in laboratory-constructed, 3-level aquatic food chains.

Although reports of pharmaceutical bioconcentration in aquatic organisms are increasing, less is known about trophic transfer in aquatic food webs. The bioaccumulation and trophodynamics of sertraline and fluoxetine, 2 selective serotonin reuptake inhibitors (SSRIs) frequently detected in aquatic environments, were tested by exposing constructed aquatic food chains to SSRIs under controlled laboratory conditions. Both of these ionizable, weak base pharmaceuticals showed lower bioaccumulation factors (BAFs) with increasing trophic level (i.e., no biomagnifications) in 2 3-level food chains (Acer platanoides, fed to Asellus aquaticus, in turn fed to Notonecta glauca or Pungitius pungitius). Mean sertraline BAFs in A. platanoides, A. aquaticus, N. glauca, and P. pungitus were 2200 L/kg, 360 L/kg, 26 L/kg, and 49 L/kg, respectively, and mean fluoxetine BAFs 1300 L/kg, 110 L/kg, 11 L/kg, and 41 L/kg, respectively. The weak influence of diet was further demonstrated by measured BAFs being equal to or lower than measured bioconcentration factors (BCFs). Organism lipid content was not positively correlated with BAFs, suggesting that other processes are driving interspecific differences in SSRI bioaccumulation. The empirically derived parameter values were introduced into a proposed bioaccumulation model, and a poor correlation was found between modeled and empirical BAFs (predicted r2 = -0.63). In conclusion, the apparent lack of biomagnification of these ionizable pharmaceuticals suggests that environmental concern should not necessarily focus only on higher trophic levels, but also on species showing high BCFs at any trophic level. Environ Toxicol Chem 2017;36:1029-1037. © 2016 SETAC.

Low levels of persistent organic pollutants (POPs) in New Zealand eels reflect isolation from atmospheric sources.

Polychlorinated biphenyls (PCBs) and organic pesticides (i.e., DDTs) were measured in long finned eels (Anguilla dieffenbachii) in 17 streams on the west coast of South Island, New Zealand. Very low levels of PCBs and low levels of ppDDE were found. The concentrations of PCBs and ppDDE were not correlated within sites indicating that different processes determined the levels of the two pollutants in New Zealand eels. The PCBs probably originate from atmospheric transport, ppDDE levels are determined by land use and are higher in agriculture areas. The low contamination level of these aquatic systems seems to be a function of a low input from both long and short-range transport as well as few local point sources. No correlation could be found between lipid content and persistent organic pollutants (POPs) concentration (as shown in previous studies) in the eels which could be explained by low and irregular intake of the pollutants.

Assessing Potential Vulnerability and Response of Fish to Simulated Avian Predation after Exposure to Psychotropic Pharmaceuticals.

Psychotropic pharmaceuticals present in the environment may impact organisms both directly and via interaction strengths with other organisms, including predators; therefore, this study examined the potential effects of pharmaceuticals on behavioral responses of fish to avian predators. Wild-caught juvenile perch (Perca fluviatilis) were assayed using a striking bird model after a seven-day exposure to psychotropic pharmaceuticals (the antidepressants fluoxetine or sertraline, or the ß-blocker propranolol) under the hypotheses that exposure would increase vulnerability to avian predation via increasing the probability of predator encounter as well as degrading evasive behaviors upon encounter. None of the substances significantly affected swimming activity of the fish, nor did they increase vulnerability by affecting encounter probability or evasive endpoints compared to control treatments. Counter to our expectations, fish exposed to 100 µg/L fluoxetine (but no other concentrations or pharmaceuticals) were less likely to enter the open area of the arena, i.e., less likely to engage in risky behavior that could lead to predator encounters. Additionally, all fish exposed to environmentally relevant, low concentrations of sertraline (0.12 µg/L) and propranolol (0.1 µg/L) sought refuge after the simulated attack. Our unexpected results warrant further research as they have interesting implications on how these psychotropic pharmaceuticals may affect predator-prey interactions spanning the terrestrial-aquatic interface.

Influence of pH-dependent aquatic toxicity of ionizable pharmaceuticals on risk assessments over environmental pH ranges.

Due to variation in pH, ionizable pharmaceuticals entering aquatic environments experience different degrees of ionization, which may affect toxicity. Using data from toxicity testing at only neutral pH may potentially under- or overestimate actual toxicity at pH ranges found in natural aquatic environments. Here we show relative pH-dependent acute toxicity to Daphnia magna for the pharmaceutical weak bases fluoxetine and sertraline, as well as the weak acids naproxen, diclofenac, ibuprofen and ketoprofen. A probabilistic modelling approach using the pH-dependent toxicity data for D. magna and an environmental pH distribution based on over 4000 European running waters from 21 countries predicted that environmental toxicity for the weak bases may be underestimated if pH 7 was assumed. The model predicted median underestimation by a factor of 3, with 90% of the model results ranging from 1 to 6. Consequently, due to the slightly basic nature of the European running waters, weak acid toxicity may be overestimated by a factor of 2. Predicted median toxicity was 0.5 of that assuming pH 7 with 90% of the results ranging from 0.03 to 5. Because aquatic pH exhibits large variation both within and between countries, we advise the use of site-specific risk assessments for ionizable pharmaceuticals in making informed water management decisions.

Ecological implications of altered fish foraging after exposure to an antidepressant pharmaceutical.

Pharmaceutical residues are increasingly detected in environmental and biological samples, some at levels known to adversely affect non-target organisms; however, less is known of how these organism-level effects relate to the ecology of aquatic systems. Foraging processes may be used as behavioral endpoints that link effects on individuals to the population and community levels, enabling risk assessment of environmental contaminants at larger ecological scales. In this study, we performed feeding trials using juvenile Eurasian perch (Perca fluviatilis) exposed to the selective serotonin reuptake inhibitor (SSRI) sertraline to test the hypothesis that sertraline alters foraging ecology of the fish in terms of their functional response. We found an exposure-dependent decrease in feeding with increasing sertraline concentrations. Further experiments revealed that feeding rates decrease at both low and high prey densities, indicating effects on both attack rate and handling time, respectively. Because the functional response can shape consumer-resource dynamics, such effects may alter the stability of predator-prey systems and consequently, community structure.

Trophic magnification factors: considerations of ecology, ecosystems, and study design.

Recent reviews by researchers from academia, industry, and government have revealed that the criteria used by the Stockholm Convention on persistent organic pollutants under the United Nations Environment Programme are not always able to identify the actual bioaccumulative capacity of some substances, by use of chemical properties such as the octanol-water partitioning coefficient. Trophic magnification factors (TMFs) were suggested as a more reliable tool for bioaccumulation assessment of chemicals that have been in commerce long enough to be quantitatively measured in environmental samples. TMFs are increasingly used to quantify biomagnification and represent the average diet-to-consumer transfer of a chemical through food webs. They differ from biomagnification factors, which apply to individual species and can be highly variable between predator-prey combinations. The TMF is calculated from the slope of a regression between the chemical concentration and trophic level of organisms in the food web. The trophic level can be determined from stable N isotope ratios (δ(15) N). In this article, we give the background for the development of TMFs, identify and discuss impacts of ecosystem and ecological variables on their values, and discuss challenges and uncertainties associated with contaminant measurements and the use of δ(15) N for trophic level estimations. Recommendations are provided for experimental design, data treatment, and statistical analyses, including advice for users on reporting and interpreting TMF data. Interspecies intrinsic ecological and organismal properties such as thermoregulation, reproductive status, migration, and age, particularly among species at higher trophic levels with high contaminant concentrations, can influence the TMF (i.e., regression slope). Following recommendations herein for study design, empirical TMFs are likely to be useful for understanding the food web biomagnification potential of chemicals, where the target is to definitively identify if chemicals biomagnify (i.e., TMF > or < 1). TMFs may be less useful in species- and site-specific risk assessments, where the goal is to predict absolute contaminant concentrations in organisms in relation to threshold levels.

Temperature and resource availability may interactively affect over-wintering success of juvenile fish in a changing climate.

The predicted global warming may affect freshwater systems at several organizational levels, from organism to ecosystem. Specifically, in temperate regions, the projected increase of winter temperatures may have important effects on the over-winter biology of a range of organisms and especially for fish and other ectothermic animals. However, temperature effects on organisms may be directed strongly by resource availability. Here, we investigated whether over-winter loss of biomass and lipid content of juvenile roach (Rutilus rutilus) was affected by the physiologically relatively small (2-5 °C) changes of winter temperatures predicted by the Intergovernmental Panel on Climate Change (IPCC), under both natural and experimental conditions. This was investigated in combination with the effects of food availability. Finally, we explored the potential for a correlation between lake temperature and resource levels for planktivorous fish, i.e., zooplankton biomass, during five consecutive winters in a south Swedish lake. We show that small increases in temperature (+2 °C) affected fish biomass loss in both presence and absence of food, but negatively and positively respectively. Temperature alone explained only a minor part of the variation when food availability was not taken into account. In contrast to other studies, lipid analyses of experimental fish suggest that critical somatic condition rather than critical lipid content determined starvation induced mortality. Our results illustrate the importance of considering not only changes in temperature when predicting organism response to climate change but also food-web interactions, such as resource availability and predation. However, as exemplified by our finding that zooplankton over-winter biomass in the lake was not related to over-winter temperature, this may not be a straightforward task.