Tolerance of whitefish (Coregonus lavaretus) early life stages to manganese sulfate is affected by the parents

European whitefish (Coregonus lavaretus) embryos and larvae were exposed to 6 different manganese sulfate (MnSO4) concentrations from fertilization to the 3‐d‐old larvae. The fertilization success, offspring survival, larval growth, yolk consumption, embryonic and larval Mn tissue concentrations, and transcript levels of detoxification‐related genes were measured in the long‐term incubation. A full factorial breeding design (4 females × 2 males) allowed examination of the significance of both female and male effects, as well as female–male interactions in conjunction with the MnSO4 exposure in terms of the observed endpoints. The MnSO4 exposure reduced the survival of the whitefish early life stages. The offspring MnSO4 tolerance also was affected by the female parent, and the female‐specific mean lethal concentrations (LC50s) varied from 42.0 mg MnSO4/L to 84.6 mg MnSO4/L. The larval yolk consumption seemed slightly inhibited at the exposure concentration of 41.8 mg MnSO4/L. The MnSO4 exposure caused a significant induction of metallothionein‐A (mt‐a) and metallothionein‐B (mt‐b) in the 3‐d‐old larvae, and at the exposure concentration of 41.8 mg MnSO4/L the mean larval mt‐a and mt‐b expressions were 47.5% and 56.6% higher, respectively, than at the control treatment. These results illustrate that whitefish reproduction can be impaired in waterbodies that receive Mn and SO4 in concentrations substantially above the typical levels in boreal freshwaters, but the offspring tolerance can be significantly affected by the parents and in particular the female parent. Environ Toxicol Chem 2017;36:1343–1353. © 2016 SETAC


INTRODUCTION
Manganese (Mn) and sulfate (SO 4 ) occur naturally in the aquatic environment [1,2]. Median Mn and SO 4 concentrations in Nordic surface waters range from 3.2 mg/L to 65 mg/L and 1.3 mg/L to 3.8 mg/L, respectively [3][4][5]. Although Mn and sulfur (S) are essential nutrients [1,2], excessive concentrations of Mn and SO 4 can be toxic to aquatic organisms [6,7]. Mining and mineral processing are 2 of the major anthropogenic sources of Mn [1]. Similarly, SO 4 is often a prevalent contaminant in mine water, and it can contribute substantially to salinization of the waterbodies receiving the mine waters [8]. The metal mining industry has adopted and developed biomining processes, in which microorganisms are utilized in metal recovery, and biomining is considered to have economic and environmental advantages compared with conventional recovery methods [9]. In Europe, the first commercial application of biomining utilizing bioheap leaching technology was established in 2008 in northeastern Finland [10]. Since the mine started to operate, concentrations of Mn and other metals, as well as SO 4 , in the waterbodies receiving the mine effluents have been elevated, and an accidental gypsum pond leakage at the mine in late 2012 caused deterioration of nearby water quality [5,[11][12][13]. This has raised concern about the effects of Mn and SO 4 especially on the commercially important boreal freshwater fish.
The early life stages of fish, larvae in particular, are generally more sensitive to chemical toxicants than adults [14]. Offspring stress tolerance can depend on their genetic background and especially on the female parent [15,16]. Metal exposure during early development is known to disturb developmental processes, reduce hatching rate and larval body size, and cause both embryonic and larval malformation and mortality [17]. Compared with other metals, such as cadmium (Cd), copper (Cu), and zinc (Zn), the toxicity of Mn to aquatic organisms is suggested to be low [6,18]. The 25% inhibition concentration (IC25) of Mn on survival and growth of brown trout (Salmo trutta) early life stages has been reported to be 4.67 mg Mn/L to 8.68 mg Mn/L, Mn being more toxic in soft water than in hard water [6]. In soft water, Mn concentrations of 0.32 mg/L to 0.35 mg/L have disturbed the mineral uptake and skeletal calcification of brown trout larvae [19]. For SO 4 , previously reported IC25 values affecting embryo development of coho salmon (Oncorhynchus kisutch), embryo-to-alevin development of rainbow trout (Oncorhynchus mykiss), and larval mortality of fathead minnow (Pimephales promelas) were 1264 mg/L, 501 mg/L, and 933 mg/L, respectively [7]. Mixture toxicity studies on aquatic organisms, such as salmonid embryos, tropical duckweed (Lemna aequinoctialis), green hydra (Hydra viridissima), and pulmonate snail (Amerianna cumingi), focusing on both SO 4 and a cationic metal, such as calcium (Ca 2þ ) or magnesium (Mg 2þ ), suggest that the cation is the toxic cause rather than the SO 4 [20,21].
External stressors can also activate defense mechanisms in aquatic organisms, and oxidative stress is often associated with a strong stress [22]. Metals and salinity changes are known to modulate oxidative stress responses in fish [23,24], and the main antioxidant enzymes protecting organisms from oxidative damage are catalase (CAT), glutathione peroxidase (GPx), glutathione-S-transferase (GST), and superoxide dismutase (SOD) [25,26]. Metal-binding proteins, metallothioneins (MTs) [27], have been considered as suitable biomarkers for metal exposure [25]. Although the exact role of MTs is still unclear [27], they are known to regulate the availability of essential and nonessential metals [28]. Also, the induction of MT gene transcription can correlate with metal tolerance, as observed with Cd-exposed turbot (Scophthalmus maximus) larvae [29].
The present study was designed to specifically assess the critical mixture concentration of Mn and SO 4 that reduces survival of a native boreal fish species, European whitefish (Coregonus lavaretus) embryos and larvae and disturbs their yolk utilization for growth. We conducted a continuous laboratory-scale manganese sulfate (MnSO 4 ) exposure with whitefish embryos and hatched larvae to investigate the effect of the parental combination on the sensitivity of whitefish early life stages to MnSO 4 ; the Mn body residues of the whitefish eggs and the 3-d-old larvae; and the transcript abundance of cat, gstt, mt-a, and mt-b in the embryos and larvae under the MnSO 4 exposure. These results bring new information for assessing the effects of SO 4 -induced salting and Mn on the reproduction of European whitefish stock, and they allow comparison of the species sensitivity with the effects of salting and Mn in freshwaters worldwide.

Test species and chemicals
Newly stripped whitefish eggs of 4 females and milt of 2 males (Rautalampi stock from the Finnish Game and Fisheries Research Institute, Laukaa, Finland) were transported from the hatchery to the laboratory for fertilization. The eggs were kept in plastic boxes and on ice until fertilization. Milt was transported (40 min) in oxygen-filled Minigrip plastic bags and kept on ice. Before fertilization, the milt was pipetted into microcentrifuge tubes and placed into a cool block (Echtotherm Chilling/Heating dry bath, Torrey Pines Scientific) at 5 8C.
Manganese sulfate monohydrate, MnSO 4 Á H 2 O (Emsure, ACS, Reag. PhEur; Merck; purity 98.8%), was weighed into 4 MnSO 4 Á H 2 O stock solutions of 6.4 mg/L, 160 mg/L, 4000 mg/L, and 100 000 mg/L into ultrapure water (Ultra Clear UV UF TM; Evoqua), and the solutions were stored at 4 8C in the dark prior to use. New stock solutions were made 2 times (6.4À4000 mg MnSO 4 Á H 2 O/L) to 3 times (100 000 mg MnSO 4 Á H 2 O/L) during the course of the experiment. Suprapur HNO 3 (65%, Merck) was used in the water sample acidification, and acid washes were done with analytical grade HNO 3 (Merck). The reagents (HNO 3 and HCl) used in tissue sample digestion were of analytical grade (Sigma-Aldrich), and only high-purity water of 18.2 MV cm resistivity produced by a PURELAB Ultra water purification system supplied by Elga was used throughout with the tissue samples.

Test setup and fertilization
Prefiltered (1 mm; 155383-03, model BP-410-1; Pentek) Lake Konnevesi (Konnevesi, Central Finland) water was used as the control water and spiked with MnSO 4 to 6 nominal MnSO 4 exposure concentrations of 0.06 mg/L, 0.29 mg/L, 7.2 mg/L, 35.7 mg/L, 179 mg/L, and 893 mg/L, respectively (element-specific concentrations are presented in Table 1). The continuous laboratory-scale MnSO 4 exposures were started from fertilization and ended 3 d after the hatching of the larvae. The experiment lasted for 160 d, starting on 7 November 2013 and ending on 16 April 2014, when nearly all the embryos had either died or hatched (with 2 alive but unhatched embryos in the concentration of 41.8 mg MnSO 4 /L). During the winter period (experiment days 1À108), the water temperature development followed natural Lake Konnevesi water temperature. The water temperature elevation was started on experiment day 109, resulting in an approximately 0.2 8C daily mean water temperature increase. The mean, minimum, and maximum water temperatures of all the pools during the whole experiment period were 3.6 8C, 1.0 8C, and 10.3 8C, respectively. The light rhythm followed the local natural light rhythm (Konnevesi, Central Finland), resulting in approximately 7:17-h, 5:19-h, 6:18-h, 9:15-h, 12:12-h, and 14:10-h mean monthly light:dark cycles during the experiment time. The embryos were sampled at the end of the winter period to represent the embryonic development during winter period, and, correspondingly, the 3-d-old larvae represent development during the spring period.
A full-factorial breeding design was applied by fertilizing the eggs of each female (F1, F2, F3, and F4) with the milt of both males (M1 and M2) separately to produce all 8 different female-male combinations in 3 replicates for each MnSO 4 exposure concentration and the control (Figure 1). The sperm motility of the males was inspected with an Integrated Semen Analysis System (ISASv1 Casa; Proiser) before fertilization. Approximately 50 eggs to 200 eggs per replicate were fertilized on plastic Petri dishes with 5 mL to 10 mL of milt using the corresponding exposure or control water as the sperm activation water. The activation water temperature was 5 8C. A few minutes after fertilization, each replicate egg batch was placed into a plastic box (350 mL, polypropylene; Greiner) containing 100 mL of the corresponding exposure or control water and taken immediately into the experiment room at 5 8C temperature. Eggs that were left over from the fertilization were stored for size analysis at -20 8C. Fertilization success was estimated under a light microscope from 10 eggs/replicate 4 d to 5 d after fertilization; after investigation, those eggs were not returned for further incubation. The egg batches were moved into plastic (LE-marked, Robusto; OKT) incubation pools (inside size: 565 mm Â 365 mm Â 220 mm) containing 12.5 L of the corresponding test water 6 d after fertilization. There was 1 pool for each MnSO 4 exposure and 1 control pool; the eggs were placed into compartment grids, 1 replicate egg batch per 1 randomly selected compartment. The compartment grids (350 mm Â 350 mm Â 70 mm) containing 36 compartments (50 mm Â 50 mm Â 40 mm) were made of plexiglass, with a 750-mm mesh (PETP, 07-750/53; Sefar Petex) glued (Acrifix 192) to the bottom of the compartments. Water depth both underneath and above the eggs was 30 mm (total water depth 60 mm), allowing sufficient water circulation for the eggs. Before the onset of hatching, the compartments were divided into 4 sections with thin plexiglass slides allowing 3-d incubation of the larvae with 24-h accuracy of the individual hatching time. The grids and plexiglass slides were acid washed (10% HNO 3 ), and the grids and incubation pools were soaked in the corresponding exposure or control water before the eggs were placed into them. Pool waters were aerated with glass Pasteur pipets (10% HNO 3 acid washed) from 2 opposite sides of the pool into opposite directions to enhance adequate water circulation. Pools were protected from contamination with loosely placed clear plastic film covers on top, and polystyrene covers were placed on the pools for 3 mo in early December to mimic the ice cover typical of boreal regions at that time of the year.

Quality control of the exposure
The water renewal intervals were 3 d to 4 d, and 4 L from each pool was changed at a time. At every water renewal time, both new exposure and control waters for the next water renewal were prepared and left to aerate and stabilize to the incubation temperature.
Incubation water temperature, pH (744 pH meter, Metrohm, Professional Plus, YSI; and SevenGo pH meter SG2, Mettler Toledo), conductivity (Professional Plus, YSI), and oxygen concentration (ProOdo, YSI) were monitored at both the beginning and the end of the experiment and before and after water renewals. Mean water oxygen concentration (AE standard error [SE]) of all the pools during the experiment was 12.6 AE 0.1 mg/L (min 11.0 mg/L and max 15.2 mg/L), and pH was 6.66 AE 0.01 (min 5.46 and max 7.49) (Supplemental Data, Table S1). Degree days (cumulative sum of mean daily temperatures during the whole incubation period) were calculated for each pool with linear interpolation using the water temperature before every water renewal. Dissolved organic carbon of prefiltered Lake Konnevesi water (i.e., newly made control water) was determined at the beginning of the experiment and during embryo and larval sampling (TOC-L, Total Organic Carbon Analyzer; Shimadzu), and the result was a mean dissolved organic carbon concentration of 7.7 AE 0.2 mg/L. Ammonium concentrations of the pool waters were analyzed in an accredited laboratory (FINAS T142; EN ISO/IEC 17025) according to a standard method [30] once after 2 mo of incubation. The mean ammonium concentrations (AE SE) of all the pools before and after the water renewal were 361 AE 35.1 mg/L (min 260 mg/L and max 500 mg/L) and 253 AE 23.6 mg/L (min 180 mg/L and max 340 mg/L), respectively.
Manganese (Mn) and sulfur (S) concentrations in the incubation waters were monitored, and other common elements such as aluminum (Al), arsenic (As), Cd, calcium (Ca), chrome (Cr), cobalt (Co), Cu, iron (Fe), lead (Pb), magnesium (Mg), Figure 1. A schematic illustration of the experiment setup with a timeline and procedures. The exposures were started at fertilization and continued until all larvae were hatched. Fertilization success was estimated 4 d to 5 d after the fertilization from 10 eggs per replicate from each exposure concentration and the control. In the spring period, water temperature was elevated gradually and hatching started at experiment day 119. On average, 50% of all the embryos had hatched on experiment day 136 and larval samples were collected after 3-d incubation of the larvae. In the highest exposure concentration, none of the embryos hatched. See Supplemental Data, Table S4 for more detailed information about the number of samples of each sample type in each concentration and parent pair.

Effects of manganese sulfate on whitefish early life stages
Environ Toxicol Chem 36, 2017 nickel (Ni), phosphorus (P), potassium (K), sodium (Na), strontium (Sr), uranium (U), and Zn were analyzed as well (Supplemental Data, Table S2). Sulfate concentrations were estimated based on S concentrations, assuming that all the S was present as SO 4 in the well-aerated exposure and control waters. Filtered (50-mL sterile syringe, BD Plastipak, 25-mm GD/XP syringe filters, 0.45-mm polyvinylidene fluoride [PVDF] with polypropylene, Whatman) and unfiltered water samples from the newly made control and MnSO 4 -spiked waters were collected twice at the beginning of the experiment, and unfiltered pool water samples were collected twice, 1 d and 3 d after the eggs were placed into the pools. Afterward, unfiltered pool water samples were collected just before and after every water renewal and at the end of the experiment. Water samples were collected into metal-free plastic tubes (polypropylene, 50 mL or 15 mL; VWR), acidified immediately after sampling by adding 6 (50-mL samples) or 2 (15-mL samples) drops of HNO 3 , and stored at 4 8C in the dark until the analyses. The total numbers of analyzed samples are shown in Supplemental Data, Table S2. The chemical element concentrations of the waters (Supplemental Data, Table S2) were analyzed with inductively coupled plasma-optical emission spectrometry (ICP-OES; Optima 8300, PerkinElmer); in case of low Mn concentrations, the Mn analysis was performed with electrothermal atomic absorption spectrometry (AAnalyst 800; PerkinElmer). Water concentration results above the limits of quantification (LOQs) with a relative standard deviation (RSD) below 10% were accepted. The LOQs were defined according to US Environmental Protection Agency method 200.7 [31].
The control (i.e., background) water mean total concentrations (AE SE) of MnSO 4 , Mn, and SO 4 for the whole experiment period were 5.5 (AE 0.1) mg/L, 1.5 Â 10 À2 (AE 0.3 Â 10 À2 ) mg/L, and 5.5 (AE 0.1) mg/L, respectively ( Table 1). The measured total MnSO 4 , Mn, and SO 4 exposure concentrations varied between 100.4% and 107.7%, 27.9% and 107.6%, and 102.1% and 108.6% of the nominal concentrations that included the corresponding background levels (Table 1; Supplemental Data,  Table S3). During the first month of the experiment, the Mn concentrations in 2 of the lowest exposure concentrations fell below the nominal levels and remained there until the end of the experiment. It is probable that Mn could have been oxidized in the well-aerated pool water and thus adsorbed onto the pool and compartment grid surfaces and/or onto the developing eggs. In this case, if both the Mn was oxidized and the amount of Mn adsorbed was uniform in each pool, the difference between the measured and nominal Mn concentrations would thus have been the most extreme in the low-concentration pools compared with the higher-concentration pools.
Most of the control water Mn was not in dissolved form (mean Mn dissolved 7.5%), whereas the MnSO 4 concentration increase in exposure waters gradually increased the proportion of dissolved Mn (range of the mean Mn dissolved 78À99%). Sulfur was in dissolved form in the MnSO 4 exposure (range of the S dissolved 99À100%) and control (mean S dissolved 99%) waters.

Mortality and overview of embryo and larval sampling
Dead embryos were counted and removed 3 times per week during the first month of the experiment and twice per week afterwards. Hatching started at 119 d after fertilization, and the mean hatching peak of all parent pairs was reached 136 d after fertilization. During the hatching period, hatched and dead larvae were counted daily.
Detailed information on the embryo and larval samples of every sample type is given in Supplemental Data, Table S4 according to MnSO 4 exposure concentration and parent pair. Both embryos and larvae were sampled for tissue element concentration and gene expression analyses, whereas only larvae were sampled for growth and yolk consumption analyses. Embryo samples were collected before the beginning of the water temperature elevation at the end of the winter period, on experiment days 102 (for tissue concentration analyses) and 105 (for gene expression analyses). The hatched larvae were incubated for 3 d under exposure conditions before larval sampling. For the 3-d-old larvae, samples were collected from experiment day 131 to experiment day 146.
Growth, yolk consumption, and egg size One growth and yolk consumption sample from each replicate contained 1 to 10 larvae from the same hatching day and a maximum of 4 larvae per replicate was measured. The samples were collected into 1.5-mL microcentrifuge tubes (StarLab), excess water was removed, and approximately 1 mL of 10% neutralized formalin (1:9 v/v of 37% formalin and Na 2 HPO 4 3.55 g/L, NaH 2 PO 4 Á H 2 O 7.3 g/L dissolved in ultrapure H 2 O) was added. The samples were stored at À20 8C in the dark until analysis. Thereafter, the samples were thawed on ice and rinsed with ultrapure water (Ultra Clear UV UF TM); then, yolk and carcass were separated and placed into preweighed aluminum cups. The samples were dried at 40 8C for 24 h and weighed. The initial egg size of each female was measured by analyzing the dry weight from 16 to 20 eggs per female. The initial egg size samples were stored at À20 8C in the dark without formalin fixation.

Tissue concentration analysis
The embryo tissue concentration samples were analyzed as whole eggs (no dechorionation), and thus those samples are referred to as eggs. A maximum of 4 eggs ($48.4 mg wet wt resulting in 8.7 mg dry wt) from each replicate was collected into each tissue concentration sample. To get enough material for the analysis, the egg tissue concentration samples of the 3 replicates for each parent pair in each concentration were pooled (see Supplemental Data, Table S4). With the 3-d-old larvae, the larval samples were replenished until 10 3-d-old larvae/sample/replicate were obtained. The sample replicates of the larval tissue concentration were not pooled. The egg and larval tissue concentration samples were collected into 1.5-mL microcentrifuge tubes (StarLab), washed 3 times with filtered (50-mL sterile syringe, BD Plastipak, 25-mm GD/XP syringe filters, 0.45-mm PVDF with polypropylene, Whatman) Lake Konnevesi water, blotted dry, placed into preweighed 1.5-mL microcentrifuge tubes (StarLab), and stored at À20 8C in the dark prior to analysis. The sampled larvae were anesthesthetized with sodium hydrogen carbonate containing fruit salt (Samarin, Cederroth), and some of the larvae were photographed under a microscope (SteREO, Discovery V8, AxioCam ERc 5s, Zen lite 2011, Zeiss) for malformation investigation before washing and storing. In the course of photography, the larvae at 197.8 mg MnSO 4 /L were observed to be opaque, but because no proper malformations were observed from any of the larvae at any exposure concentration, the malformation data were not analyzed further.
Freeze-dried and weighed egg and larval samples were digested in aqua regia (HNO 3  . Eggs were digested in 1.5-mL microcentrifuge (StarLab) tubes by moisturizing the sample with ultrapure water and adding 8 drops of aqua regia, followed by sonication in 2-min to 3-min cycles at 30 8C to 40 8C. Larval samples were digested in 13-mL tubes (polypropylene, Sarstetd) by moisturizing the sample with ultrapure water and adding 10 drops of aqua regia, followed by sonication in 3-min cycles at 55 8C. The sonication was repeated 2 times to 12 times until the samples were fully digested. Samples were shaken, and sample tube pressure was released by opening the caps between each sonication cycle. The digested samples were filtered (Whatman no. 41, GE Healthcare Life Sciences) and filled to a final volume of 5 mL with ultrapure water.
The element concentrations of the eggs (Supplemental Data, Table S5) and the larvae (Supplemental Data, Table S6) were analyzed with ICP-OES (Optima 8300, PerkinElmer). The same LOQ and RSD limit requirements were used for both the water and the tissue concentrations, and the tissue concentration results are presented as upper bound concentrations (values below LOQ and/or RSD > 10% are replaced with LOQ), unless noted otherwise.

Gene expression
Gene expression samples were collected at the incubation temperature to avoid sudden temperature changes that may affect embryo and larval gene expressions. A maximum of 3 embryos per sample from each replicate was collected into 1.5-mL microcentrifuge tubes (StarLab), washed 3 times with filtered (50-mL sterile syringe BD Plastipak, 25-mm GD/XP syringe filters, 0.45-mm PVDF with polypropylene; Whatman) Lake Konnevesi water, blotted dry, and placed into 1.5-mL microcentrifuge tubes (StarLab). With the 3-d-old larvae, 1 sample from each replicate contained 5 larvae from the same hatching day. The larvae were sampled into 1.5-mL microcentrifuge tubes (StarLab), and the excess water was removed with a needle and a syringe. Both the embryonic and larval samples were immediately frozen in liquid nitrogen and stored at À80 8C until analysis.
The embryonic and larval expression of the target genes mt-a, mt-b, gstt, and cat was analyzed with quantitative reverse transcriptionÀpolymerase chain reaction (qRTÀPCR) using ribosomal protein L2 (rl2) and beta actin as reference genes (Supplemental Data, Table S7). The selected reference genes had the most stable expression among treatments from all the reference genes tested. The mt-a gene was obtained from Hansen et al. [32]. Other target gene primers were designed with Primer3, Ver 4.0.0 [33,34], and the reference gene primers were designed with AmplifX Ver 1.5.4 [35]. The specificity of the genes was checked with Primer-BLAST (http://www.ncbi.nlm. nih.gov/tools/primer-blast/) [36]. The RNA extraction and integrity analysis, deoxyribonuclease treatment, complementary DNA synthesis, and amplification reactions were done as described in Vehni€ ainen and Kukkonen [37]. The qPCR run was done using CFX96 Real-Time PCR cycler (Bio-Rad), and the protocol was 95 8C for 3 min; 40 cycles of 95 8C (10 s), 58 8C (10 s), and 72 8C (30 s); 95 8C for 10 s; and melt curves from 65 8C to 95 8C with 0.5 8C intervals. Samples were run in duplicate using clear 96-well PCR plates (Bio-Rad). A no-template control was always run for each gene.  Equation 4), excluding the fertilization loss of the cross-fertilizations, were estimated according to the following equations

Mortality calculations and statistics
where N 0 is the number of live eggs before fertilization, N F is the number of successfully fertilized eggs, N W is the number of embryos that had survived by the end of the winter period just before embryo sampling, N S is the number of embryos in the beginning of the spring period just after embryo sampling , N L is number of live larvae 3 d after hatching, and 0.01 is a constant added to avoid 0 values when taking the natural logarithm (ln). Instantaneous mortality (Z) was assumed to be a normally distributed variable. A full-factorial general linear model (GLM), including the main effects of male, female, and MnSO 4 exposure and their interactions was used to test for the effect of MnSO 4 exposure on Z F and Z T . The highest MnSO 4 concentration at which the femalespecific total survival did not differ significantly from the control values (no-observed-effect concentration [NOEC]) were tested with analysis of variance (ANOVA) and, depending on the homogeneity of variances, the pairwise comparisons were done with a one-sided Dunnett's test or Dunnett's T3 test. The lethal MnSO 4 concentrations resulting in 50% mortality of the offspring (LC50) with 95% confidence limits were calculated for each female separately by pooling the data of both males per female. The female-specific LC50 values for MnSO 4 were calculated according to total offspring mortality and the mean measured total MnSO 4 concentrations of each pool during the whole experiment period using each female's mean total mortality in the control pools as natural response rates. Data analyses were made with the Probit model using an ln-transformed covariate. A heterogeneity factor was used in the confidence limit calculations because the significance level of the Pearson goodness-of-fit chi-square test was less than 0.15.
A full-factorial GLM, including the main effects of male, female, and MnSO 4 exposure, with their interactions and degree days as a covariate, was used to test for the effect of MnSO 4 exposure on growth and yolk consumption of the 3-d-old larvae. The mean carcass and yolk dry mass of the larvae in each replicate were used in the analyses.
A full-factorial GLM, including the main effects of female and MnSO 4 exposure and their interactions, was used to test for the effect of MnSO 4 exposure on the egg and larval Mn and S tissue concentrations. Because of mortality, the number of samples was small, and thus the male effect was not included in the data analysis. The concentrations below the LOQ or with RSD above 10% were replaced with upper bound concentrations.
The critical body residues of Mn taken up by the eggs and larvae causing 50% mortality of the observed individuals (CBR50) with 95% confidence limits were calculated separately for eggs and egg-to-larvae stages. Because of mortality, the number of samples was small, and thus the CBR50 values could not be distinguished according to the parent pairs or females. In all the egg samples, the Mn concentrations were above LOQ, with an RSD less than 10%, whereas for the larvae, an upper bound (concentrations below the LOQ or with RSD above 10% Effects of manganese sulfate on whitefish early life stages Environ Toxicol Chem 36, 2017 replaced with LOQ) estimate for the CBR50 value was calculated. Data analyses were made with the Probit model using an ln-transformed covariate. A heterogeneity factor was used in the confidence limit calculations because the significance level of the Pearson goodness-of-fit chi-square test was less than 0.15. Because the egg tissue concentration replicates of each exposure concentration and control were pooled according to the parent pair, the corresponding mean winter embryonic mortality values were used in the egg Mn CBR50 analysis, and the mean mortality of the control samples was used as a natural response rate. The total mortality values were used in the egg-to-larval analysis, and thus the egg-to-larval CBR50 values represent the entire exposure period from fertilization until the larvae were 3 d old. The natural response rate in the egg-to-larvae analysis was the mean total mortality of the control samples. Embryonic and larval gene expression differences of 3 and 2 parent pairs, respectively, were analyzed, and thus the female and male effect could not be tested. The pair selection was based on the offspring total mortality: a pair with low (F4 Â M2), intermediate (F2 Â M1), and high (F3 Â M2) offspring total mortality. The degree days of the analyzed replicates were taken into account, and thus no effects of developmental differences were expected. The gene expression differences between the parent pairs and treatments were analyzed with log10-transformed normalized expressions using the GLM full-factorial effect model structure. In addition, to distinguish the differences in embryonic gene expression among the 3 parent pairs, the exposure concentration was excluded from the factors and only the parent pairs were compared with one-way ANOVA including Tukey's honest siginficant difference (HSD) or Tamhane's T2 post hoc tests depending on the homogeneity of the variances. Also, the differences in gene expression of the offspring of each parent pair separately were analyzed for both embryos and larvae with one-way ANOVA, including Tukey's HSD or Tamhane's T2 post hoc test depending on the homogeneity of the variances.
All statistical analyses were done with SPSS (IBM SPSS Statistics 22).

Fertilization, mortality, NOEC, and LC50 values
The Z F of the whitefish eggs was significantly affected by both the MnSO 4 exposure concentration and female parent but male effect or interactions between the variables were not found ( Table 2 and Figure 2A). Similarly, MnSO 4 exposure and female, but not male, parent had a significant effect on the Z T ( Figure 2B). The significant interaction between MnSO 4 exposure and female effects indicates that some of the females had produced more MnSO 4 -tolerant embryos than others, and this was also evident from the female-specific NOEC and LC50 values (Table 3). Although there was a significant interaction between MnSO 4 exposure and male effects as well, the differences between the males were not as clear as those between the females at different concentrations.

Growth and yolk consumption
The carcass dry weight of the 3-d-old whitefish larvae was significantly affected by MnSO 4 exposure and female parent ( Table 4). The male effect was not significant, but the interaction between exposure and male effects were. Degree days, MnSO 4 exposure, and female parent had a significant effect on the yolk dry weight of the 3-d-old larvae ( Table 4). The interactions between female and MnSO 4 exposure effects, and between parent pair and MnSO 4 exposure, were also significant. On average, the larvae of each female had more yolk left at the 41.8 mg MnSO 4 /L concentration compared with the control larvae ( Figure 3).

Mn and S concentrations in the eggs and the larvae
The Mn concentrations of the eggs were significantly affected by MnSO 4 exposure (GLM, F ¼ 207.137, df ¼ 6,   Table 5; Supplemental Data, Table S8). The MnSO 4 exposure also had a significant effect on the larval Mn concentrations (GLM, F ¼ 58.810, df ¼ 5, p < 0.001), but the female effect was not significant (GLM, F ¼ 1.069, df ¼ 3, p ¼ 0.368). Significant interactions between MnSO 4 exposure and female effects were not found with either the eggs or the larvae (GLM, p > 0.05; Supplemental Data, Table S8). Compared with the eggs, the 3-d-old larvae had roughly 8 to even 60 times lower Mn body burdens. The egg CBR50 value for Mn was 9.08 mmol/g dry weight, with 95% confidence limits of 7.13 mmol/g to 12.81 mmol/g dry weight and a natural response rate of 30.5%. For the egg-to-larvae period, the CBR50 value was 0.88 mmol/g dry weight, with 95% confidence limits of 0.56 mmol/g to 2.05 mmol/g dry weight and a natural response rate of 29.1%. Sulfur was not concentrated in either eggs or larvae (GLM, p > 0.05; Supplemental Data, Table S8).

Gene expression
The MnSO 4 exposure did not have a significant effect on the transcript abundance of any of the target genes in the embryos (Figure 4 and Table 6). However, cat, mt-a, and mt-b expression differed significantly between the embryos of the 3 different parent pairs ( Table 6). As the exposure concentration was excluded from the analysis and only the embryonic gene expression differences between the parent pairs were analyzed, the significant differences in cat, mt-a, and mt-b expression between the parent pairs were still observed (ANOVA, cat: . Pairwise comparison showed that cat expression of the embryos of the parent pair with high offspring total mortality (F3 Â M2) was significantly higher than that of the embryos of the other 2 pairs (Tukey's HSD, F2 Â M1: p ¼ 0.015; F4 Â M2: p ¼ 0.001). The mt-a expression of the embryos of the parent pair with the lowest offspring total mortality (F4 Â M2) was significantly lower than that of the embryos of the 2 other parent pairs (Tukey's HSD, F2 Â M1: p ¼ 0.028; F3 Â M2: p ¼ 0.018), whereas mt-b expression of the embryos of the parent pair with intermediate offspring total mortality (F2 Â M1) was significantly lower than that of the embryos of the parent pair with high Table 3. Female-specific NOEC and LC50 values of MnSO 4 (mg/L) according to total mortality of the offspring of each female, and the natural response rate according to the control mean total mortality of the female's offspring   (Figure 4 and Table 6). According to the GLM results, mt-b was differently induced among the 2 pairs, F2 Â M1 and F4 Â M2, and the joint effect between MnSO 4 exposure and the parent pair was significant as well (Table 6). Even though the GLM results revealed significant differences only in mt-b induction between the parent pairs, the individual parent pair analyses showed that both mt-a and mt-b were significantly induced only in the larvae of the pair with the lowest offspring total mortality (F4 Â M2; ANOVA, mt-a: F ¼ 10.569, df ¼ 4, p ¼ 0.003; mt-b: F ¼ 7.176, df ¼ 4, p ¼ 0.009), whereas the larvae of the pair with an intermediate offspring total mortality (F2 Â M1) had no mt induction (ANOVA, mt-a: In pairwise comparisons, larval mt-a and mt-b expression of the pair F4 Â M2 was significantly higher in MnSO 4 exposures of 5.9 mg/L (mt-a: Tamhane's T2, p ¼ 0.035; mt-b: Tukey's HSD, p ¼ 0.038) and 41.8 mg/L (mt-a: Tamhane's T2, p ¼ 0.043; mt-b: Tukey's HSD, p ¼ 0.020) than in the control larvae. However, mt-b expression of the larvae reared at the 12.8 mg MnSO 4 /L concentration also differed significantly from expression in the larvae reared at 41.8 mg MnSO 4 /L (Tukey's HSD, p ¼ 0.032). According to the GLM results, MnSO 4 exposure or parent pair did not have a significant effect on gstt or cat expression of the larvae (Table 6), and the individual parent pair analyses did not reveal any differences either. The numbers of valid replicates for both embryos and larvae for each target gene and parent pair are given in the Supplemental Data, Table S9.

DISCUSSION
Under experimental conditions, early life stages of whitefish were sensitive to MnSO 4 , and the variation in their tolerance   Table S9.
was significantly affected by the parent fish, the female in particular. The female had a significant effect on fertilization success, offspring total mortality, and growth and yolk consumption of the larvae, whereas the male alone did not have a significant effect on those endpoints. The MnSO 4 exposure caused a significant induction of mt-a and mt-b, but only in the larvae of the pair with the lowest total offspring mortality.
The female-dependent differences in offspring tolerance to toxic chemicals or unfavorable environmental conditions could be because of genetic or environmentally induced variability. In methylmercury-exposed mummichog (Fundulus heteroclitus) embryos, some females living in an unpolluted environment produced more tolerant offspring than other females of the same population, and it was suggested that the variability in offspring methylmercury tolerance was linked to genetic differences between the females [15]. Also, survival of freshwater-adapted European whitefish embryos under chronic osmotic stress has been shown to depend significantly on their female parents [16]. However, there is evidence that if the parent fish is exposed to metals before spawning, the metal tolerance can be maternally transferred to the offspring as well. For example, female fathead minnows that had been exposed to Cu produced larvae with a higher Cu tolerance [38]. In the present study, however, the parent fish were hatchery-reared and thus likely had a uniform life history and had been reared in an unpolluted environment. Thus, the female effect observed in the present study was more likely the result of natural individual variation. The male effect was not observed or remained vague because of a low number of males.
The higher MnSO 4 tolerance of the offspring of F4 could also be related to the larval mt-a and mt-b induction observed in parent pair F4 Â M2. The ability of Cd-exposed turbot larvae to induce MT gene transcription has previously been correlated with their Cd tolerance [29], and thus the activation of detoxification processes could have been the reason for the higher MnSO 4 tolerance of the offspring of F4 as well. However, the offspring gene expression of the parent pair F4 Â M1 was not investigated, and only the larvae had induced expression of mt-a and mt-b, whereas in the embryos, the observed differences in gene expression between the parent pairs did not seem to be connected to offspring survival. Also, induction of mt-a and mt-b in the whitefish larvae did not show a consistent concentration-related pattern. Previously, a concentration-related pattern of MT induction has been observed from several tissue types of Cd-exposed juvenile river pufferfish (Takifugu obscurus) [39]. On the other hand, the expression patterns of MT have also been shown to depend on exposure duration [39], and different metals can cause very different induction patterns as well [40].
The MnSO 4 exposure slightly inhibited yolk consumption of all the larvae in the 41.8 mg MnSO 4 /L exposure concentration and with nearly all females, exposure also reduced the dry weight of the larval carcass. Metal exposure can cause an inability to utilize yolk reserves, as has been demonstrated with Cu-exposed common carp (Cyprinus carpio) larvae [41], and the larval growth of brown trout also has been shown to be reduced after Mn exposure [6]. In addition, increased osmoregulatory cost can reduce the larval length of whitefish [16].
Because Mn accumulation in the eggs was substantially higher than in the larvae, it is most likely that Mn was blocked by the chorion and/or perivitelline fluid, as previously demonstrated with fish embryos exposed to Cd [42] and Cr [43]. Such a conclusion is also supported by the finding that, in the embryos, none of the target genes were significantly affected by MnSO 4 exposure. The observed difference between Mn concentrations in eggs of the different females is most likely because of the differences in the size of the offspring, as F3 had the largest eggs and larvae but nearly always had accumulated the least Mn. This was most likely because the surface area to body mass ratio was smaller for the eggs and larvae of F3 than in the other females. Thus, such female-related differences also support the view that Mn had accumulated in the chorion and/or pervitelline fluid.
Also, the egg and egg-to-larval Mn CBR50 values (95% confidence limits) of 9.08 (7.13À12.81) mmol/g dry weight and 0.88 (0.56À2.05) mmol/g dry weight, respectively, suggest that the chorion and/or perivitelline fluid had protected the developing embryo from Mn. According to the CBR50 values, the eggs seemed to be far more tolerant to Mn compared with the whole development period from egg to 3-d-old larva. However, we used total offspring mortality instead of larval mortality alone when estimating the egg-to-larval CBR50 value, because the actual larval mortality during the 3-d rearing was rather negligible. The egg-to-larval CBR50 value thus represents the entire exposure period from fertilization until the larvae were 3 d old, which may better represent real-life conditions in the water bodies affected by the mining effluents.
In the present experiment, we were unable to determine whether the observed responses were because of Mn or SO 4 alone or because of their interaction. However, if it is assumed that the mixture toxicity effect of MnSO 4 in the exposures would have been solely additive for the 2 substances, and so back-calculating the MnSO 4 NOEC and LC50 values to the respective Mn and SO 4 exposure concentrations, we can make some comparisons with the previously reported toxicity values of Mn and SO 4 . The female-specific NOEC range for Mn and SO 4 would thus be 0.04 mg/L to 12.5 mg/L and 5.8 mg/L to 29.3 mg/L, respectively. The female-specific MnSO 4 LC50 values varied from 42.0 mg/L to 84.6 mg/L, with a 95% confidence limit of 33.9 mg/L to 147.1 mg/L. Such values are within the MnSO 4 exposure concentration range of 12.8 mg  [7] levels, the toxicity values of the early life stages of other fish species, whitefish embryos and larvae seemed to be slightly more or equally sensitive to Mn than those of other fish species, whereas the whitefish early life stages were far more sensitive for SO 4 . However, the difference in the water hardness between the different experiments makes comparison of the toxicity values uncertain. The test chemical was MnSO 4 with 98.8% purity, and especially at the highest exposure concentration, elevated levels of Cu, Ni, and Pb were measured, with the Cr concentration being already elevated at 41.8 mg MnSO 4 /L (Supplemental Data, Table S2). However, compared with a previous experiment with rainbow trout embryos and larvae [44], the effect of the Cu impurity in the present experiment was likely negligible, as was the effect of Ni when estimated according to the annual average environmental quality standard concentration for Ni (20.0 mg/L) [45]. Also, the egg and larval tissue concentrations of Cu did not increase as the exposure concentration increased, and the tissue concentrations of Ni were below the LOQs. Exposure water Pb concentrations above the annual average environmental quality standard (7.2 mg/L) [45] were observed once at 41.8 mg MnSO 4 /L, and increasingly from 197.8 mg MnSO 4 /L and 965.0 mg MnSO 4 /L concentrations, but the egg and larval Pb concentrations were below the LOQ. For Cr, a 499.2 mg/L concentration has been reported to increase the larval mortality of common carp [43], and the mean water Cr concentration (AE SE) at the highest MnSO 4 exposure concentration (422.2 AE 45.0 mg Cr/L) was close to that value. According to other studies, only a Cr concentration of several mg/L is toxic to fish embryos and larvae [46,47]. In the present study, however, the egg Cr concentrations were elevated at 4 of the highest exposure concentrations. Thus, a possible interference of Pb and Cr especially at the highest exposure concentration cannot be completely excluded. In addition, after approximately 1 mo of incubation, brown precipitates and/or bacterial growth was observed on the walls of some of the pools and compartment grids (visually observed in pools containing 5.9À41.8 mg MnSO 4 /L), which may have affected the Mn balance of those pools.
Finally, although the present study does not allow us to distinguish the individual roles of Mn or SO 4 , the results give relevant insights into the field conditions of freshwaters under the impact of metal mining, as the concentrations of both Mn and SO 4 are often elevated in water bodies receiving mining effluents [5,48]. In the case of the bioheap leaching mine in northeastern Finland, from approximately 2010 onward, the annual mean Mn and SO 4 concentrations have increased in the waterbodies impacted by the mine, with the highest SO 4 concentrations being even several thousand mg/L and the highest Mn concentrations several hundred mg/L [5,12]. Whitefish is still caught from some of the less impacted lakes, and in those lakes the highest annual mean values of Mn and SO 4 have been found to be 2022 mg/L and 257 mg/L, respectively [12,49]. When these concentrations are compared with the findings from the present study, it can be seen that the observed concentrations of the impacted lakes may have adverse effects on whitefish early life stages. However, as has been demonstrated in earlier experiments, fish populations can also adapt to chemical stressors [50,51].

CONCLUSIONS
Continuous exposure to MnSO 4 decreased whitefish embryonic survival in relation to the MnSO 4 exposure concentration. Offspring tolerance to MnSO 4 exposure depended on the female parent in particular, resulting in substantial differences in offspring instantaneous total mortality already at moderate MnSO 4 concentrations. Also, larval expression of metalregulating genes indicated that better offspring MnSO 4 tolerance is linked to the induction of metal-regulating genes. The present study has shown that whitefish reproduction success can be impaired in populations under waterborne MnSO 4 concentrations of approximately 40 mg/L in boreal soft waters. As a relevant continuation of the present experiment, we will perform a corresponding field study to assess the implications of mining effluents on whitefish reproductive success in field conditions.