Copper-contaminated soil compromises thermal performance in the springtail Folsomia candida (Collembola)

(cid:129) Extensive use of copper-based products in agriculture and industry calls for in-depth risk assessment. (cid:129) Global warming will exacerbate toxicity of heavy metals on soil invertebrates. (cid:129) Cu pollution shifted the thermal performance curve and reduced thermal tolerance of springtails by damaging cell membranes.

The widespread agricultural and industrial emissions of copper-based chemicals have increased copper levels in soils worldwide.Copper contamination can cause a range of toxic effects on soil animals and influence thermal tolerance.However, toxic effects are commonly investigated using simple endpoints (e.g., mortality) and acute tests.Thus, how organisms respond to ecological realistic sub-lethal and chronic exposures across the entire thermal scope of an organism is not known.In this study, we investigated the effects of copper exposure on the thermal performance of a springtail (Folsomia candida), regarding its survival, individual growth, population growth, and the composition of membrane phospholipid fatty acids.Folsomia candida (Collembola) is a typical representative of soil arthropods and a model organism that has been widely used for ecotoxicological studies.In a full-factorial soil microcosm experiment, springtails were exposed to three levels of copper (ca.17 (control), 436, and 1629 mg/kg dry soil) and ten temperatures from 0 to 30 °C.Results showed that three-week copper exposure at temperatures below 15 °C and above 26 °C negatively influenced the springtail survival.The body growth was significantly lower for the springtails in high-dose copper soils at temperatures above 24 °C.A high copper level reduced the number of juveniles by 50 %, thereby impairing population growth.Both temperature and copper exposure significantly impacted membrane properties.Our results indicated that high-dose copper exposure compromised the tolerance to suboptimal temperatures and decreased maximal performance, whereas medium copper exposure partially reduced the performance at suboptimal temperatures.Overall, copper contamination reduced the thermal tolerance of springtails at suboptimal temperatures, probably by interfering with membrane homeoviscous adaptation.Our results show that soil organisms living in copper-contaminated areas might be more sensitive to thermally stressful periods.
Science of the Total Environment 897 (2023) 165334

Introduction
Soils are the sinks of metallic pollutants from anthropogenic activities, for instance, the massive use of copper additives in the current industry and agriculture (Durães et al., 2018).The primary source of copper pollution originates from air or water emissions from mining, smelting and foundry industries, and crop protection against fungal pests in agriculture (Kozlov and Zvereva, 2007;Lamichhane et al., 2018;Masindi and Muedi, 2018).In addition, copper-containing pig slurry (since copper is used as a food additive for piglets) has been applied to agricultural soil for decades and is now of growing concern (Formentini et al., 2015;Jensen et al., 2016).In consequence, copper contamination constitutes an increasing threat to soil organisms.One such copper-polluted hotspot is located in Hygum, Denmark, where copper sulphate was used for wood preservation, representing an example of industrial pollution (Scott-Fordsmand et al., 2000).Copper contamination in Hygum reached a concentration as high as 2900 mg Cu/kg dry soil and has existed for >90 years (Scott-Fordsmand et al., 2000).
The increasing frequency and magnitude of heat waves, and metal pollution, are predicted to continue due to global climatic change and anthropogenic activity (Järup, 2003;Rahmstorf and Coumou, 2011;Buckley and Huey, 2016;IPCC, 2022;Jørgensen et al., 2022).The interaction between the effects of pollutants and the thermal environment will be a key factor shaping soil animal population responses to further environmental change.Long-term copper contamination may harm biodiversity and species richness, especially in soil invertebrates such as earthworms and springtails (Naveed et al., 2014).Studies that assess pollutants across the entire thermal scope of an organism in combination with chronic exposure are, however, lacking.Most studies on the interaction between metal pollution and natural stressors have been done in short-term experiments using spiked soils (Holmstrup et al., 2008;Buch et al., 2016;Callahan et al., 2019).However, the toxicity and bioavailability of metals in spiked soil is unrealistically high and has little relevance for field-contaminated soils where copper is much less available for living organisms (Smolders et al., 2009;Neaman et al., 2020).Hence, there is an urgent need for toxicological assessments based on field-collected contaminated soils and realistic temperature scenarios.The classic endpoint in toxicological studies is survival, but the combined effect of chronic copper contamination and climatic stressors might profoundly impact sub-lethal endpoints like body growth and population growth.
Excess loss of soil biodiversity will have considerable implications for soil organic matter dynamics, the biogeochemical cycle of ecosystems, and soil fertility (Rath and Rousk, 2015;Lakshmi et al., 2020).Springtails are one of the most abundant arthropods as they are widespread and are found in soils on every continent.Springtails are able to efficiently regulate their internal copper concentration (Ardestani and van Gestel, 2013;Ardestani et al., 2014).However, continuous exposure to elevated concentrations may potentially influence the springtails' growth, reproduction, and behaviour, likely by inducing oxidative stress (Maria et al., 2014).Metal-mediated formation of excess free radicals in the organism can lead to peroxidation of phospholipids, the main component of the cell membranes, resulting in damage to double bounds of the phospholipid fatty acid (PLFA) chains (Valko et al., 2005).This will interfere with homeoviscous adaptation, which can change the membrane's response to temperature change and consequently impede normal biochemical reactions (Hazel, 1995;Hill et al., 2012).Alterations in membrane phospholipid composition can provide a mechanistic link to the compromised thermal performance (Ernst et al., 2016).A previous study has shown that copper exposure had negative effects on membrane phospholipids and hence impaired the cold tolerance of the earthworm Dendrobaena octaedra (Bindesbøl et al., 2009).Therefore, extreme weather events such as bare soil freezing or heat waves could result in a dramatic decline in population size or even lead to extinction (Holmstrup et al., 2010;Laskowski et al., 2010).
In this study, we aimed to investigate the combined effect of copper exposure and different temperatures on survival, body growth, and reproduction of the springtail F. candida in a chronic exposure scenario using a microcosm approach.Besides life-history traits, we measured the PLFA composition of the springtails as an indicator of lipid peroxidation and membrane fluidity.We hypothesised that 1) survival, growth, and reproduction of F. candida would show a unimodal distribution with temperature; 2) exposure to copper has deleterious effects on the thermal performance curve (TPC) by lowering the performance at optimal temperature and reducing the tolerance at the extreme ends of the tested temperatures; 3) copper exposure decreases the degree of unsaturation of membrane phospholipids.

Springtails
Folsomia candida was cultured in Petri dishes floored with plaster of Paris (moist gypsum/charcoal, 8:1, w/w), kept at 20 ± 0.2 °C with a photoperiod of 12 L:12D and fed dried baker's yeast.Several batches of twenty springtails of similar size were transferred to newly made Petri dishes and left for two days to obtain age-synchronized eggs (OECD, 2016).Upon hatching, the juveniles were fed dried baker's yeast.After 12-14 days, age-synchronized springtails were ready for the experiment.We randomly measured the initial dry weight of springtails in 18 subgroups containing 10 animals, and the average dry weight was 0.013 ± 0.003 mg.

Soil
Copper-contaminated soils were collected from a former timber preservation yard at Hygum (Jutland, Denmark), where the copper contamination gradient across a 50 × 100 m field has been well characterized (Scott-Fordsmand et al., 2000).The upper soil layer (approximately 20 cm) was collected from three areas representing 'low', 'medium', and 'high' copper levels (see later for copper concentrations).Before use, soils were thoroughly homogenized, dried at 80 °C for 48 h and sieved through a 2 mm mesh.Further description of the test soils can be found in Pedersen et al. (1999) and Maraldo et al. (2006).

Microcosm experiment
Plastic beakers (7.8 cm depth, 4.6 cm bottom diameter, 6.1 cm upper diameter) with 50 g moist soil (39.5 g dry soil and 10.5 mL demineralized water) were used as an experimental unit.The experiments were performed at 10 constant temperatures (0, 2.5, 5, 10, 15, 20, 22, 24, 26, and 30 °C) in climate chambers (ICP110, Memmert, Schwabach, Germany) for 21 days.Each beaker contained 10 springtails and was sealed with a plastic lid to prevent water loss.Six replicates were assigned to each copper level for each temperature.Springtails destined for temperatures at or below 10 °C were pre-acclimated to the cold environment by gradually decreasing culture temperature from 20 to 10 °C over 24 h (Waagner et al., 2013).Springtails destined for temperatures above 10 °C were kept at 20 °C.During acclimation, the springtail groups were starved, ensuring that the average body dry mass remained the same.Beakers were then placed in the temperature chambers for 21 days.The actual temperature was recorded every 5 min using TinyTag temperature loggers with an accuracy of ±0.1 °C (Gemini Data loggers, Chichester, UK).The actual average temperatures were calculated at the end of the experiment (0.2, 2. 5, 5.3, 9.9, 14.8, 20.2, 22.1, 24.0, 25.7, 30.0 °C, Fig. S1) and used for further analysis.During the experiment, animals were fed with dried baker yeast, and the beakers were aerated every second day by briefly releasing the lids.At the end of the experiment, the soil in each beaker was carefully poured into a tray and spread out using a needle.The survivors of the ten added adult springtails were collected using an aspirator, transferred to a 1.5 mL microcentrifuge tube, and stored at −80 °C until further analysis.The remaining soil was transferred back to the beaker, and deionized water was added.After gentle stirring with a spoon, the beaker was left for 1-2 min until the juveniles floated on the surface.A photo was then taken to record the number of juveniles on the water surface.
Samples of adults were dried for 48 h using a freeze dryer (LyoQuest-55, Telstar, Terrassa, Spain) and counted to obtain survival.Using a Sartorius Micro SC 2 balance with an accuracy of ±1 μg (Sartorius AG, Göttingen, Germany), we weighed the samples of adults and calculated the average dry weight of the adults of each replicate.The number of juveniles was counted manually using ImageJ (Version 1.53 k, National Institute of Health, Maryland, USA) as a counter (Schneider et al., 2012).

Copper concentration in soil and animals
Soil samples from different concentrations were prepared in technical triplicate for copper concentration measurement.Fifty springtails in each replicate (3 for each copper exposure level) were exposed under the same copper exposure setup for the same period at 20 °C.For the copper quantification, 2 mL of 68 % nitric acid was added to glass vials containing ca. 0.3 g subsamples.The vials were put into the cells of a heating block in which temperature gradually increased from 80 to 120 °C and remained constant until the content in vials became transparent.Digested samples were diluted to 25 mL using 5 % nitric acid and placed at room temperature overnight.The supernatant was transferred to 5 mL plastic centrifuge tubes, and the copper concentration was measured using inductively coupled plasma-optical emission spectrometry with an accuracy of ±0.1 mg/kg (5800 ICP-OES, Agilent, Germany).Calibration was performed using copper calibration standards (Cu Pure standard, 1000 μg/mL, PerkinElmer, USA).The efficiency of the digestion was verified by analysing certified reference material (oyster tissue material from the National Institute of Standards and Technology, U.S. Department of Commerce and lobster hepatopancreas from National Research Council Canada).The measured concentration was approximately 95 % of the certified values.

Phospholipid fatty acid composition
To obtain enough material for PLFA analysis, we pooled every two replicates of animals, creating biological triplicates for extraction and measurement of PLFA composition.Crude extraction of total lipids was performed following the procedures described by Waagner et al. (2013).Springtails were transferred to 12 mL glass centrifuge tubes containing 3 mL methanol, 1.5 mL chloroform and 1.25 mL PO 4 buffer.Samples were vortexed for 1 min and placed at room temperature (approximately 20 °C) for 2 h.Another 1.5 mL PO 4 buffer and 1.5 mL chloroform were added, and samples were whirly mixed again and left overnight at room temperature.Samples were centrifuged at 3000 ×g for 10 min to extract the organic compounds dissolved in chloroform from the watery phase.The watery phase was removed by suction, and the chloroform phase was transferred to a new glass centrifuge tube and evaporated under nitrogen flow.To separate the polar phospholipid from other lipids, the sample was redissolved in 300 μL chloroform and then slowly vacuum-filtered through solid-phase silica columns (100 mg, Bond Elute, Agilent, USA), preconditioned with 1.5 mL chloroform.The neutral and medium polarity lipids were eluted with 1.5 mL chloroform and 6 mL acetone and discarded.Polar lipids (mainly phospholipids) were eluted with 1.5 mL methanol and dried under nitrogen flow.Polar lipids were trans-esterified by mild alkaline methanolysis (Dowling et al., 1986).After that, 500 μL methanol, 500 μL toluene, and 1 mL 0.2 mol•L −1 KOH were added to the samples, which were briefly whirly mixed and incubated at 37 °C for 15 min in a water bath.After trans-methylation, 2 mL heptane, 300 μL 1 mol•L −1 acetic acid and 2 mL ELGA water were added to samples, whirly mixed for 1 min, and centrifuged at 1500 ×g for 5 min.The upper organic phase containing fatty acid methyl esters (FAMEs) was transferred to a new tube, and the extraction was repeated by adding 2 mL heptane to the acetic acid/water phase.The heptane-FAMEs mixture was evaporated under nitrogen flow, and the FAMEs were redissolved in 500 μL heptane.Samples were then run on a gas chromatograph coupled with mass spectrometry (GCMS-QP2010 Plus, Shimadzu, Japan) equipped with an auto-sampler (Columbia, USA).The FAMEs were separated on a Supelco Omegawax 320 column (length 30 m, inner diameter 0.32 mm; Sigma-Aldrich, Stockholm, Sweden) using helium as the carrier.Samples of 1 μL were injected in split mode (split ratio 10:1) at 220 °C, and the oven was programmed to remain the temperature at 60 °C for 2 min, increase it to 280 °C over 16 min and hold it at the final temperature for 2 min.Column flow was 1.79 mL•min −1 at a pressure of 90.7 kPa.The GCMS inner-phase temperature was set at 200 °C, and the ion source temperature was at 220 °C.Identifying individual fatty acids was based on the mass spectra of known FAME standards (Supeleco 37 component FAME mix, Sigma Aldrich, USA).
Fatty acids were designated as X:YnZ, where X indicates the number of C atoms, Y indicates the number of double bonds, and Z indicates the position of the first double bond counting from the methyl end of the molecule.Areas of identified peaks were quantified using external standards, and mol % distributions were calculated.PLFAs accounting for >0.5 mol% were reported.The degree of unsaturation was calculated as Ʃ(% monoenes + 2 × % diens +3 × % trienes…)/100.The average chain length was calculated by the weighted average number of C atoms in the fatty acid chain.

Statistical analysis and modelling
The distributions of survival, average weight and numbers of juveniles were examined using the 'fitdistrplus' package in R (Delignette-Muller and Dutang, 2015) and found to be beta-binomial distribution, gamma distribution and gamma-Poisson distribution, respectively.To obtain thermal-related parameters from TPCs of survival, individual growth, and population growth, we used the R package 'rTPC' to select candidate models (Padfield et al., 2021) in R (version 4.1.2)(R Core Team, 2021).First, we selected the models based on probability, Q 10 effect or enzyme kinetics for survival, individual and population growth as candidates (Table S1).Each model was fitted separately to the control group datasets on survival, average weight, and number of juveniles.The prediction error and relative quality of the models were then evaluated based on the Akaike information criterion (AIC) (Akaike, 1998).The scores of sigma value, AIC, AICc, and Bayesian information criterion (BIC) are shown in Table S2.Second, we fitted the models with the least AIC values based on the three copper levels across ten temperatures.Accordingly, survival, individual growth and population growth, respectively, were fitted with modified gaussian, Weibull, and O'Neill models using the 'gnlm' R package and the corresponding 95 % confidence intervals were calculated (Swihart et al., 2017).An F test was conducted to determine the statistical difference between TPCs under different levels of copper exposure.Then, we extracted maximum performance (R max ), optimal temperature (T opt ) and temperature breadth of 80 % maximum performance (T br80 ) and calculated their 95 % confidence interval.
Since survival, body growth, and population growth were not normally distributed, Aligned Rank Transform (ART) one-way analysis of variance (ANOVA) followed by Bonferroni's post hoc test was performed on survival, body growth, and population growth of F. candida at each temperature.One-way ANOVA followed by Bonferroni's post hoc test was performed on the internal concentration of copper in animals.For PLFA data, we performed principal component analysis (PCA) on each component and multivariate analysis of variance (MANOVA) on the PLFA indices to present the overall effects of copper and temperature.Following the MANOVA test, the PLFA data was analysed using two-way ANOVA followed by post-hoc multiple comparison Bonferroni's tests in order to evaluate the effects of copper and temperature, respectively.Then to illustrate the relation between the unsaturation degree of phospholipids and temperature, as well as the effect of copper on this correlation, simple linear regression and analysis of covariance (ANCOVA) was performed at temperatures 0-5 °C and 5-26 °C separately.The principal components in PCA were selected based on parallel analysis.Simple linear regression and PCA were executed in GraphPad Prism 9.4.1 (GraphPad Software, San Diego, California, USA), while ANOVA, MANOVA and ANCOVA were performed using the packages 'ARTool' (Elkin et al., n.d.; Wobbrock et al., n.d.), 'lme4' (Bates et al., 2015) and 'car' (Fox and Weisberg, 2019) in R (version 4.1.2)(R Core Team, 2021).All data were expressed as mean ± standard error of mean (SEM).Statistical significance was set at P < 0.05.

Results and discussion
Springtails (Collembola) are among the most abundant species distributed in all continents and hence most likely to experience high temperature variation across day and season in their life cycles.The present study showed that the thermal performance of F. candida was negatively influenced by copper exposure by measuring survival, body growth, population growth, and PLFA composition.It reveals the relationship between fitness and temperatures across the entire range of tolerated temperatures and how copper contamination influences this relationship.The study shows the importance of including the temperature effect when assessing metal toxicity in soil organisms.

Bioavailability of copper in Hygum soil
The copper concentrations in the experimental soils were determined to be 17.3 ± 0.5, 435.7 ± 6.9 and 1628.9 ± 15.6 (n = 3, measurement replicates) mg/kg dry soil, corresponding to low (L), medium (M), and high (H) copper exposure levels.After 3 weeks of copper exposure, the internal copper concentrations of springtails at 20 °C were 42.3 ± 1.5, 47.1 ± 1.5, and 81.6 ± 1.9 (n = 3) mg/kg dry weight at low, medium, and high copper exposure, respectively.The much lower internal concentration of copper of springtails exposed in polluted soil is likely due to the ability of springtails to regulate internal copper concentrations, but also because heavy metals in field-collected soil are significantly less bio-available than in spiked soil as a result of ageing processes, higher organic matter content and adsorption capacity for metals in field soils as reported by previous studies (Spurgeon and Hopkin, 1995;McBride et al., 2009;Neaman et al., 2020).These results indicate the importance of using field-contaminated soils in ecotoxicological studies.The internal copper concentration of springtails in the high exposure group was significantly higher than those in low and medium exposure groups (P < 0.0001, Bonferroni's test).There was no significant difference of internal copper concentration between low and medium exposure groups, indicating this species was capable of regulating its internal concentration of copper in medium polluted soil (up to 400 mg copper/kg dry soil), but not in soil with a high level of copper.A previous study on uptake and elimination kinetics of copper has been conducted on F. candida in LUFA 2.2 soil spiked with 100 μg Cu(NO 3 ) 2 /g dry soil and reported the internal copper concentration to be around 90 mg/kg dry tissue after 14 days (Ardestani and van Gestel, 2013), presenting a high bioavailability of copper in newly contaminated soil.

Survival
After three weeks of temperature exposure, the springtail survival showed a bell shape relation with temperature with no mortality across a temperature plateau from around 10 to 22 °C (Fig. 1).Jegede et al. (2017) reported that the average survival of F. candida was over 80 % and showed no statistical difference at 20 and 26 °C, indicating a potential plateau of thermal optimum.When temperatures were below 5 °C or above 26 °C, the mortality increased significantly (Fig. 1), suggesting that adverse effects start to accumulate when temperature exceeds these limits.A study on cold acclimation in F. candida showed that acclimation to 0 °C for 24-72 h significantly reduced the cold tolerance compared with those acclimated to 5 °C (Waagner et al., 2013).Relatively higher survival at 0 °C in this study might be due to the preacclimation to 10 °C, which enhanced the cold tolerance of F. candida.
Under the pressure of copper contamination, no significant reduction in survival was found at temperatures between 5 and 26 °C (Fig. 1).However, survival of springtails in the low copper group was significantly higher than that in the medium and high copper group at 0 °C (P = 0.0001, Bonferroni's test) and 2.5 °C (P < 0.0001, Bonferroni's test), which confirmed our hypothesis that copper exposure reduced the cold tolerance of springtails.From the fitted models and the 95 % confidence intervals, we also observed lower survival of springtails in the high copper exposure group between 26 and 30 °C (Fig. 1).Hence, copper exposure caused high mortality at both extreme ends of the temperature range and decreased T br80 (Table 1).A previous study showed synergistic interaction between copper and freezing temperatures in another springtail (Protaphorura armata) and in the earthworm D. octaedra (Holmstrup et al., 2010).The synergistic interaction between effects of copper and low temperatures has also been reported in a study of Enchytraeus albidus, where low survival of copper exposed animals under constant freezing and freeze-thaw cycles was observed (Boas et al., 2016).The decreased survival at high temperatures in our study is in accordance with another study in which the heat tolerance of F. candida and Choreutinula americana was significantly reduced by copper exposure (Callahan et al., 2019).
Like other ectotherms, springtails adapt to temperature changes by modifying the membrane PLFA composition and adjusting the fluidity of biological membranes (Hazel and Williams, 1990).More polyunsaturated fatty acids (PUFAs) are required at low temperatures to maintain the cell membrane in a functional liquid-crystalline state, whereas more saturated Fig. 1.Thermal performance of survival in Folsomia candida after 3-week exposure to different level of copper contamination with the fitted thermal performance curves (TPC).Points with error bars were the mean ± SEM.Ribbons were 95 % confidence interval of the fitted model.Green, yellow and red indicated low (control), medium and high soil copper concentrations, respectively.fatty acids are needed at high temperatures to prevent membrane fusion (Hazel, 1995;Ernst et al., 2016).This process is termed homeoviscous adaptation (Sinensky, 1974).Appropriate membrane properties are fundamental for cellular biochemistry reactions.However, membrane phospholipids are susceptible to abiotic stressors, which in our case would be copper exposure and extreme temperature.Excessive internal copper could have caused peroxidation in membrane phospholipids of F. candida and hence contributed to higher mortality at low temperature.Since the PUFAs have more double bonds on the fatty acyl chain and are thus more susceptible to reactive oxygen species, the effect of copper exposure on survival would be higher at the more extreme temperatures.
Copper's influence on survival at ecologically relevant low and high temperatures, implies a potential risk that soil organisms living in contaminated areas can potentially lead to higher mortality during summer or winter, or other extreme weather events.

Body growth
Unlike the bell-shaped TPC for survival, the average weight of springtails showed a unimodal relation with temperature, peaking at the optimal temperature (21 °C) and decreasing rapidly in the suboptimal high temperature range (Fig. 2).This unimodal relation confirmed our hypothesis and is in line with other TPC studies on body growth (Niehaus et al., 2012;Schmalensee et al., 2021;Villeneuve et al., 2021).Difference of TPC models between the low and high copper exposure groups (F 53,54 = 0.4822, P < 0.01, F test) indicated the average weight of springtails at high copper levels was reduced at temperatures above 20 °C (Fig. 2).This finding suggests increased toxicity of copper at high temperature and may explain a previous observation that no significant copper effect was found on the growth of F. fimetaria when exposed at near-optimal temperature (20 °C) using Hygum soil (Scott-Fordsmand et al., 2000).However, Scott-Fordsmand et al. (2000) reported a significant reduction in size or growth caused by copper using newly spiked soil.Other studies on metal toxicity using spiked soils showed a likewise significant impact on body growth in animals at lower metal concentrations than using field collected soils (Scott-Fordsmand et al., 1997;Nursita et al., 2005;Xu et al., 2009a).In a comparative study on field collected soil and spiked soil with copper, the LC50/EC50 of survival, reproduction and juvenile size was significantly lower in spiked soil than in field collected soil (Scott-Fordsmand et al., 2000).This discrepancy in results emphasises the importance of using field-collected soil for ecotoxicological experiments as follow up studies after laboratory experiments, as the toxicity and bioavailability of metal in spiked soils are much higher than those in field-collected soils (McBride et al., 2009;Neaman et al., 2020).
The reduced growth of springtails at high copper levels may be because metal toxicity resulted in less food consumption and oxidative stress and consequently tilted the balance of the energy budget away from growth into detoxification or antioxidation to guarantee survival, which explains that the T br80 of body growth of springtails was around 10 °C lower than that of survival (Table 1).Several studies have found that under stressful Note: population growth was indicated by the number of juveniles at the end of trial.Brackets indicate the 95 % confidence interval of the mean.R max : highest performance; T opt : optimal temperature; T br80 : temperature margin at 80 % max performance.temperatures and metal exposure, springtails were less active in food search and ingestion (Sørensen et al., 1995;Boiteau and Mackinley, 2012;Kwak and An, 2016).On the other hand, heavy metals will cause oxidative stress in organisms and thus consume energy reserves (Valko et al., 2005;Ferreira et al., 2015).Our study showed no detectable copper effect on body growth at temperatures below 15 °C, suggesting that low temperature significantly slowed the overall biochemical reactions, including metal uptake rate (Janssen and Bergema, 1991), and therefore masked the toxic effect of copper.Combined with the survival results, a moderate decrease in temperature (10-15 °C) might postpone the toxic effect of copper contamination.

Population growth
The thermal performance of population growth, indicated by the number of juveniles at the end of the trial, followed a unimodal curve along with temperature, peaking at around 25 °C (control) and dropping at temperature above T opt .(Fig. 3).Mallard et al. (2020) reported a similar unimodal relation between temperature and the number of eggs produced and showed that F. candida was able to produce eggs from 6 to 26 °C.In our study, the TPCs under medium and high copper exposure differed significantly from the TPC at low copper exposure (F 56,56 = 0.4393, P < 0.01; F 56,56 = 0.3637, P < 0.01, F test).The high copper concentration had deleterious effects on population growth, cutting R max of population growth by almost 50 %.(See Fig. 3.) Population growth includes other important but variable factorsthe production of eggs, hatching rate, and juvenile survival.Tranvik et al. (1993) reported that the egg production in two springtail species (Onychiurus armatus and Isotoma notabilis) were delayed and reduced after being exposed to the mixture of Cu and Zn for three weeks.Besides, since the eggs were constantly exposed to copper contamination in the soil, the hatching rate was further reduced (Tranvik et al., 1993).Xu et al. (2009b) reported a negative correlation between soil copper concentration and hatchability of springtail eggs.Moreover, they even found colour change of the eggs in high copper exposure, suggesting that copper ions can penetrate through eggshells and accumulate in embryos.Our findings suggest that low and high temperatures significantly restrict population growth.Various external factors make recruitment of springtails more difficult under chemical exposure, which was also indicated by the much lower T br80 for population growth compared to survival and body growth (Table 1).Even if the individuals can survive the sub-optimal temperatures, the population may still risk extinction, especially under the pressure of prolonged extreme weather events or chemical contamination.Note: ANOVA: analysis of variance; Temp: temperature; SFA: saturated fatty acid; MUFA: monounsaturated fatty acid; PUFA: polyunsaturated fatty acid; CL: average chain length; UI: unsaturation index, degree of unsaturation; d.f.: degree of freedom.The 'overall' effect was indicated by multivariate analysis of variance (MANOVA) using all fatty acid components as dependent variables.

Phospholipid fatty acid composition
Biological membranes, consisting of a phospholipid bilayer and embedded proteins, are essential for organisms to sense and adapt to environmental changes.Environmental changes (temperature, osmolarity, humidity, etc.) induce transient fluidity change in membranes and then trigger the molecular pathways to restructure PLFA composition and subsequently adjust membrane fluidity, a process known as homeoviscous adaption (Hazel and Williams, 1990;Hazel, 1995;Los and Murata, 2004).This adaptation is mainly achieved by regulating the proportion of PUFA in the membrane (Mikami and Murata, 2003;Los and Murata, 2004;Ernst et al., 2016).Hence, the PLFA composition was measured, and the degree of unsaturation in membrane was calculated to illustrate copper effects on the thermal response of F. candida.
In the present study, a total of 11 different PLFA's were identified, including 3 saturated fatty acids (SFAs), 4 monounsaturated fatty acids (MUFAs), and 4 PUFAs (see Fig. S2 & Table S1).The most abundant PLFAs were 20:5n3 (ca. 25 mol%), 18:0 (ca.21 mol%), 18:1n9c (ca.16 mol%), 20:4n6 (ca.16 mol%), and 16:0 (ca. 10 mol%).The PLFA composition of F. candida in our study is similar to what has been found in previous studies on springtails (Holmstrup et al., 2002;Haubert et al., 2008), specifically with the main components of 20:4n6 and 20:5n3 (Chamberlain and Black, 2005;Sechi et al., 2014;Holmstrup and Slotsbo, 2018).All the PLFAs and the related indices were significantly influenced by temperature, which was shown in both the Two-way ANOVA (P < 0.0001, Tables 2& S3, Figs. 4 & S3) and the PCA (Fig. 5A).At temperatures between 5 and 26 °C, we observed a strong negative correlation between temperature and degree of unsaturation (P < 0.05, F test, Table S4 & Fig. S4), which is an indicator positively correlated with membrane fluidity.The increase in the degree of unsaturation with decreasing temperature was caused by the upregulation of PUFAs (18:2n6 and 20:5n3, Fig. 5B), indicating that the springtails increased the membrane fluidity to maintain normal cellular function.By prolonging the cold acclimation at 5 °C by 24 h, Waagner et al. (2013) observed an increased ratio of unsaturated to saturated fatty acid in phospholipids of F. candida.The significant effects of temperature on PLFA composition have been found in not only other species of Collembola (Haubert et al., 2008;Holmstrup and Slotsbo, 2018) but also in bacteria (Cronan, 1975), fungi (Martin et al., 2007) and fish (Liu Fig. 4. Degree of unsaturation (UI) of membrane phospholipids in Folsomia candida after 3-week exposure to different level of copper contamination under temperatures from 0 to 30 °C.Data were shown as mean ± SEM with asterisk indicating statistical significance (***: P < 0.001 and ****: P < 0.0001).Green, yellow and red indicated low (control), medium and high soil copper concentrations, respectively.et al., 2020).However, the correlation between the degree of unsaturation and temperature in the present study had not been reported in Collembola (Haubert et al., 2008;Holmstrup and Slotsbo, 2018).One of the major explanations might be the selection of temperature, which is wider than the previous research.
At temperatures below 5 °C, a positive correlation was, surprisingly, found between membrane fluidity and temperature (P < 0.05, F test, Table S4 & Fig. S4), indicating the ability of homeoviscous adaptation was hampered.In the copper-exposed groups, the significant decreases in membrane fluidity indicated synergistically interacting effects of low temperature and copper toxicity (F 2, 21 = 10.234,P < 0.001, ANCOVA, Table S4 & Fig. S4).Bindesbøl et al. (2009) demonstrated that high concentrations of copper reduced freeze tolerance in the earthworm Dendrobaena octaedra by inducing lipid peroxidation and consequently decreasing 18:2n6 in phospholipids.High-dose copper could catalyse the formation of hydroxyl radicals and cause oxidative stress (Valko et al., 2005;Calap-Quintana et al., 2017), which would attack the double bonds on the longchain PUFA, resulting in lipid peroxidation, loss of membrane integrity, and consequently, high mortality (Aucoin et al., 1995;Halliwell and Gutteridge, 2015).A fast reduction in the degree of unsaturation was recorded between 26 and 30 °C (Fig. 4), suggesting substantial stress generated by high temperature.However, a reduction in membrane fluidity at a high temperature might be less accountable for the high mortality in F. candida.Instead, the failure of protective transcription and synthesis of heat shock proteins, which are responsible for coping with heat stress, will be considered the major lethal factors (Lindquist, 1986;Sørensen et al., 2003).

Conclusion
The present study demonstrated the effects of copper exposure on the thermal performance of F. candida at temperatures from 0 to 30 °C concerning survival, individual growth, reproduction, and PLFA composition.Our results suggested that exposure to high dose copper had significant negative effect on F. candida, specifically reducing thermal scope of survival, impairing body growth at high temperature and lowering the R max of population growth.During the adaptation to low temperatures, copper interfered with the adjustment of membrane fluidity by inducing oxidative stress and lipid peroxidation.

Fig. 2 .
Fig. 2. Thermal performance of body growth (average weight of each replicate) in Folsomia candida after 3-week exposure to different levels of copper contamination with the fitted thermal performance curves (TPC).Points with error bars were the mean ± SEM.Ribbons were 95 % confidence interval of the fitted model.Green, yellow and red indicated low (control), medium and high soil copper concentrations, respectively.

Fig. 3 .
Fig. 3. Thermal performance of population growth (number of juveniles) in Folsomia candida after 3-week exposure to different levels of copper contamination with the fitted thermal performance curves (TPC).Points with error bars were the mean ± SEM.Ribbons were 95 % confidence interval of the fitted model.Green, yellow and red indicated low (control), medium and high soil copper concentrations, respectively.

Fig. 5 .
Fig. 5. Principal component analysis of phospholipid fatty acid composition in Folsomia candida after 3-week exposure to different level of copper contamination under temperatures from 0 to 30 °C.(A) Principal component scores with dots colored based on temperature and labeled with low (L), medium (M) and high (H) copper exposure level.(B) Loadings of fatty acid components of phospholipids from principal component analysis, indicating the contributions of variables to the components.

Table 1
Extracted thermal parameters from the selected models.

Table 2
Results of Two-way ANOVA and MANOVA analysis on indices related to phospholipid fatty acid composition of Folsomia candida.