Ocean acidification challenges copepod phenotypic plasticity

Ocean acidification is challenging phenotypic plasticity of individuals and populations. Calanoid copepods (zooplankton) are shown to be fairly plastic against altered pH conditions, and laboratory studies indicate that transgenerational effects are one mechanism behind this plasticity. We studied phenotypic plasticity of the copepod Acartia sp. in the course of a pelagic, large-volume mesocosm study that was conducted to investigate ecosystem and biogeochemical responses to ocean acidification. We measured copepod egg production rate, egg-hatching success, adult female size and adult female antioxidant capacity (ORAC) as a function of acidification (fCO2 ∼ 365–1231 μatm) and as a function of quantity and quality of their diet. We used an egg transplant experiment to reveal whether transgenerational effects can alleviate the possible negative effects of ocean acidification on offspring development. We found significant negative effects of ocean acidification on adult female size. In addition, we found signs of a possible threshold at high fCO2, above which adaptive maternal effects cannot alleviate the negative effects of acidification on egg-hatching and nauplii development. We did not find support for the hypothesis that insufficient food quantity (total particulate carbon < 55 μm) or quality (C : N) weakens the transgenerational effects. However, females with high-ORAC-produced eggs with high hatching success. Overall, these results indicate that Acartia sp. could be affected by projected near-future CO2 levels.

Abstract. Ocean acidification is challenging phenotypic plasticity of individuals and populations. Calanoid copepods (zooplankton) are shown to be fairly plastic against altered pH conditions, and laboratory studies indicate that transgenerational effects are one mechanism behind this plasticity. We studied phenotypic plasticity of the copepod Acartia sp. in the course of a pelagic, large-volume mesocosm study that was conducted to investigate ecosystem and biogeochemical responses to ocean acidification. We measured copepod egg production rate, egg-hatching success, adult female size and adult female antioxidant capacity (ORAC) as a function of acidification (fCO 2 ∼ 365-1231 µatm) and as a function of quantity and quality of their diet. We used an egg transplant experiment to reveal whether transgenerational effects can alleviate the possible negative effects of ocean acidification on offspring development. We found significant negative effects of ocean acidification on adult female size. In addition, we found signs of a possible threshold at high f CO 2 , above which adaptive maternal effects cannot alleviate the negative effects of acidification on egg-hatching and nauplii development. We did not find support for the hypothesis that insufficient food quantity (total particulate carbon < 55 µm) or quality (C : N) weakens the transgenerational effects. However, females with high-ORAC-produced eggs with high hatching success. Overall, these results indicate that Acartia sp. could be affected by projected near-future CO 2 levels.

Introduction
Increased concentrations of carbon dioxide (CO 2 ) in the atmosphere is changing the carbon chemistry of the world's oceans. CO 2 dissolves in seawater, thereby decreasing ocean pH. Ocean acidification is increasing fast and pH is expected to decrease by a further 0.14-0.43 pH units during the coming century (IPCC, 2007). Acidification can cause various problems to biochemical and physiological processes in aquatic organisms. In addition to affecting calcification of calcareous organisms, maintenance of acid-base equilibrium of body fluids may become more difficult and have consequences, for example, on protein synthesis, metabolism and volume control (Whiteley, 2011).
In a changing environment, populations can respond in three main ways: through plastic responses of individuals, through genetic changes across generations or through escaping in space or time by modification of phenology. During rapid change, phenotypic plasticity, i.e. the ability of an individual or a population to alter its physiological state, appearance or behaviour in response to the environment, is of major importance (West-Eberhard, 2003). Theory predicts that higher plasticity evolves in extreme environments and that spatial heterogeneity and dispersal select for higher plasticity (Chevin et al., 2013). One could, therefore, hypothesise that organisms inhabiting a variable environment, such as the study area, could be fairly plastic in their response to ocean acidification because they have to cope with both seasonal and sudden changes in pH (Almén et al., 2014;Lewis et al., 2013).
Proteomic studies suggest that oxidative stress is a common co-stress of temperature and acidification (Tomanek, 2014). Increased production of reactive oxygen species (ROS) may result in increased antioxidant and/or repair costs as well as in reduced investment in reproduction or other functions, such as immune defence. In addition, increased production of ROS may lead to accumulation of oxidative damage and to acceleration of senescence (Monaghan et al., 2009). There can also be a connection between maternal oxidative balance and offspring quality. In birds, for example, females allocate diverse antioxidants to the eggs that protect the embryo from oxidative stress. This maternal effect has a positive effect on offspring development and growth (Rubolini et al., 2006).
Copepods (zooplankton) are indispensable to the functioning of the whole pelagic ecosystem and contribute significantly to many ecosystem services (Bron et al., 2011). They provide, for example, food for early life stages as well as some adults of many economically important fish species (Steele, 1974;Cushing, 1990).
Previous results suggest that calanoid copepods have high buffering capacity against projected ocean acidification for the year 2100 and beyond (Kurihara and Ishimatsu, 2008;Weydmann et al., 2012;McConville et al., 2013;Vehmaa et al., 2013), meaning that they are able to survive, grow, develop and reproduce in lower pH (Reusch, 2014). However, there are also studies showing negative impacts on moderate CO 2 levels (Fitzer et al., 2012), whereas most of the negative impacts have been discovered for extreme carbon storage scenarios (Kurihara et al., 2004;Mayor et al., 2007;Weydmann et al., 2012). Many studies have tested only one life stage, adult females, and have, therefore, possibly underestimated the effects of ocean acidification on copepods (Cripps et al., 2014a). There are indications that transgenerational effects are the mechanism responsible for the high plasticity of copepod reproduction against altered pH conditions (Vehmaa et al., 2012). This maternal effect is most likely dependent on the condition of the mother, the availability of food and the quality of her diet (Vehmaa et al., 2012;Pedersen et al., 2014a). Paternal effects can also influence offspring traits. The exposure of both parents to CO 2 leads to fewer adverse effects on egg production and hatching than exposure of only gravid copepod females (Cripps et al., 2014b). Thor and Dupont (2015) also highlight the importance of testing transgenerational effects. They found significantly lower copepod egg production after two generations when exposed to 900 and 1500 compared to 400 µatm, but transgenerational effects alleviated the negative CO 2 response in 1500 µatm (Thor and Dupont, 2015).
We tested direct and indirect effects of ocean acidification (i.e. via food quantity and quality) on the copepod Acartia sp. egg production rate (EPR), egg-hatching success (EH), female body size measured as prosome length (PL), as well as antioxidant capacity (ORAC). The study was conducted in association with the KOSMOS (Kiel Off-Shore Mesocosms for Ocean Simulations) project in the Baltic Sea (Paul et al., 2015). The study was intended to cover the low-productivity late spring and early summer period, i.e. the post-spring bloom period when pCO 2 concentrations are at the annual minimum. Over the annual cycle, pCO 2 and pH vary substantially at the study site as a result of biological activity and mixing/upwelling of CO 2 -enriched deep water (Niemi, 1975;Omstedt et al., 2014). There are also strong spatial gradients in seawater pCO 2 /pH, most prominently between the surface layer and the CO 2 -rich deeper waters (Almén et al., 2014). Thus, the copepods in the study area are likely to experience strong changes in seawater carbonate chemistry, both seasonally and during their diurnal migration. Total particulate carbon (TPC < 55 µm) was used as the measure of food quantity. Food quality was indicated by the carbon to nitrogen ratio of the same sized fraction of seston (C : N < 55 µm) (Elser and Hasset, 1994;Sterner and Hessen, 1994). In addition, in order to separate transgenerational plasticity (i.e. maternal and paternal effects) and the effect of environment on copepod egg-hatching and development, we performed an egg-transplant experiment. Half of the produced eggs were allowed to develop in respective mesocosm water and the other half in water collected outside the mesocosm bags.
Due to the high buffering capacity of Acartia sp., we hypothesised that there are no f CO 2 -related differences in egg production rate, egg-hatching success and prosome length between the mesocosms. In addition, we hypothesised that copepod eggs hatch and develop better in the same environment in which they are produced, because transgenerational effects can alleviate the negative effects of environmental change. Our third hypothesis stated that low food quantity (TPC) and poor quality (high C : N) will weaken the maternal effect by deteriorating the condition of the mother. Finally, we tested whether mothers with higher ORAC produce better quality offspring (EH) by calculating correlation coefficients between the two variables.

Materials and methods
The study was performed in summer 2012 in the vicinity of Tvärminne Zoological Station on the south-western coast of Finland. Six large mesocosms were moored on site at the beginning of June. To enclose the natural plankton community, the mesocosms were left open with only a 3 mm-sized mesh net covering the top and the bottom during filling. After 4 days, the net was removed and the top was pulled up 1.5 m above the water surface and closed at the bottom (Riebesell et al., 2013;Paul et al., 2015). pH was ∼ 8 and f CO 2 concentrations in the mesocosms prior to adjustment were 237 ± 9 µatm (average ± SD of daily measurements from all bags). Four mesocosm were manipulated with CO 2 -enriched seawater, during three consecutive days to reach f CO 2 concentrations of 600-1650 µatm (Paul et al., 2015). Two untreated mesocosms were used as controls. The water column was mixed at the beginning of the experiment to avoid salinity stratification. Due to outgassing, CO 2 was also added on day 15 to the upper 7 m of the high-CO 2 mesocosms to maintain the treatment levels. No nutrients were added.

Sampling
Sampling took place once a week during the first 4 weeks of the experiment, and once more at the end of the whole experiment (days 3, 10, 17, 24 and 45). Mesozooplankton were sampled from all mesocosms by taking two hauls with a 300 µm net (17 cm diameter) from 17 m depth. The samples were rinsed into containers with 4 L of seawater from respective mesocosm, taken from 9 m depth with a water sampler (Limnos, Hydrobios). On the same day, integrated water samples (0-17 m) were collected from all mesocosms and the Baltic Sea, directly into 1.2 L Duran bottles that were closed without head space. Water samples were kept in cool bags and zooplankton samples were protected from light until transported to a temperature-and light-controlled room at Tvärminne Zoological Station within 4 h. The light-dark cycle in the room was 16 : 8 h and light intensity was 7 µmol photons m −2 s −1 (LI-COR LI-1000). Temperature followed the in situ temperature [9 • C (day 3), 11 • C (day 10), 15 • C (day 17), 16 • C (days 24 and 45)].

Measurements of egg production, egg-hatching success and prosome length
Twenty adult Acartia sp. (17 females and 3 males) were picked with pipettes from each sample using stereo microscopes and gently placed in prefilled glass bottles with respective mesocosm water. The bottles were closed without head space, to minimise CO 2 -outgassing during the incubation. pH in the bottles was measured before closing and right after opening them at the end of the incubation using an Ecosense pH 10 pH/temperature pen (Table S1 in the Supplement). The pen was calibrated with standard buffer solutions (Certipur, Titripac pH 4.00, 7.00 and 10.00) every second day. The bottles were incubated in temperature-and lightcontrolled room in the conditions described above (Sect. 2.1) and mixed 3 times a day, and their place on the shelf was changed randomly. After the incubation (24.3 ± 2.3 h, average ± SD), the copepods and produced eggs were filtered using 250 and 30 µm sieves respectively. The copepods were counted and their viability checked before preserving them in RNAlater (Sigma). RNAlater can affect size (Foley et al., 2010), and the effect depends on the number of segments in the animal, i.e. the more segments, the larger effect. Shrinkage is ∼ 15 % for copepods (E. Gorokhova, Stockholm University, personal communication, 2015). Prosome length of the preserved female copepods was measured using a stereo microscope (Leica MZ12) and ocular micrometre (total magnification 100×). As all the measured copepods were adult females, we assume the shrinkage to be of a similar proportion for all individuals, which means that our results are quite conservative and comparable between mesocosms.
In the egg transplant experiment, the collected eggs were divided for hatching into two 50 mL petri dishes with different conditions; one dish was filled with respective mesocosm water and the other with Baltic water. The pH of the water was measured as above before the incubations and right after the petri dishes were opened after the incubation (Table S1). The eggs were counted before the petri dishes were completely filled and sealed without head space using Parafilm. Egg hatching was followed by counting the number of remaining eggs on the dish through the lid using a stereo microscope twice a day. When the number of eggs had remained the same on two consecutive counting times, the dishes were opened and the water containing the remaining eggs and hatched nauplii was preserved with acid Lugol's solution. The hatching incubation time varied between 63.9 and 137.6 h, depending on incubation temperature. Acartia sp. nauplii stages were determined and the number of nauplii and remaining copepod eggs counted using a stereo microscope.
Some adults, copepodites, nauplii or eggs could have ended up in the incubation bottles or petri dishes with the unfiltered incubation water. The possible additional adults and their contribution to the EPR, eggs copepod −1 d −1 ) were taken into account as EPR was calculated using the number of eggs and adult Acartia sp. females found in the incubation bottles after the 24 h incubation. When estimating the egghatching success (EH, %), the total number of hatched Acartia sp. nauplii and remaining eggs at the end of the hatching incubation were compared with the number of eggs were counted before the hatching incubation. If the total number exceeded the egg number prior to hatching, the most developed nauplii (> N 4 ) were considered to be carry-over individuals and were, therefore, not considered in the estimation of EH. For an estimation of the nauplii development rate, the development index (DI) was calculated (Knuckey et al., 2005) accordingly.
where N i is the assigned stage value (0 for eggs, 1 for N 1 , 2 for N 2 and 3 for N 3 and N 4 ) and n i the number of individuals at that stage. We assume all the Acartia sp. adults and nauplii to be species A. bifilosa. However, because another Acartia species, A. tonsa occurs in the area in late summer too (Katajisto et al., 1998), we cannot be completely sure that we only had one species in the experiments.

Antioxidant capacity
For ORAC samples ∼ 25 live female Acartia sp. were picked from every zooplankton sample onto a piece of plankton net in the temperature-and light-controlled room on days 3, 10, 17 and 31. The net containing the copepods was folded and stored in Eppendorf tubes at −80 • C. The samples were homogenised in 150 µL Tris-EDTA buffer containing 1 % sarcosyl. The antioxidative capacity was assayed as ORAC (Ou et al., 2001). As a source of peroxyl radicals, 2, 2-azobis (2-amidinopropane) dihydrochloride (AAPH) (152.66 mM) was used and fluorescein was used as a fluorescent probe (106 nM). We used trolox (218 µM, Sigma-Aldrich) as a standard and the assay was performed on a 96-well microplate and to each well, 20 µL sample, 30 µL AAPH and 150 µL fluorescein were added. ORAC values were normalised to protein and expressed as mg Trolox eq. mg protein −1 . Protein concentration was measured with NanoOrange ® (Life Technologies).

C : N and TPC
Samples for TPC and C : N were collected onto GF/F filters (Whatman, nominal pore size 0.7 µm) using gentle vacuum filtration (< 200 mbar), then stored in glass petri dishes at −20 • C. GF/F filters and petri dishes were combusted at 450 • C for 6 h before use. Gauze prefilters were used to separate the size fraction < 55 µm. Filters were not acidified to remove inorganic carbon; therefore total particulate carbon was used. C and N concentrations were determined on an elemental analyser (EuroEA) following Sharp (1974), coupled by a Conflo II to a Finnigan Delta Plus mass spectrometer and were used to calculate C : N ratios in mol : mol. For further details on sampling and analyses, please refer to Paul et al. (2015).

Statistics
The effect of acidification and food quantity and quality on Acartia sp. EPR, PL, ORAC and nauplii development index (DI) was tested using linear mixed effect models (LMM) with restricted likelihood (REML) approximation from the nlme-package (Pinheiro et al., 2014), where EPR, PL or ORAC were used as response variables, f CO 2 , TPC (< 55 µm) and C : N as fixed explanatory variables and repeated measure of the mesocosms over time as a random factor (Table 1). Due to the binomial nature of the data, the effect of f CO 2 , TPC (< 55 µm) and C : N on EH was tested with a generalized linear mixed model (GLMM) with Laplace likelihood approximation, binomial error structure and logit link function from the lme4-package ) (Table 1). The average of f CO 2 , TPC (< 55 µm) and C : N measurements from each mesocosm within 3 days before the zooplankton sampling were used as explanatory variables for EPR, ORAC and EH, because 2-3 days are considered to be an appropriate acclimatisation period for A. bifilosa (Yoon et al., 1998;Koski and Kuosa, 1999). For PL, the average of all f CO 2 , TPC (< 55 µm) and C : N measurements from the start of the mesocosm experiment were used since PL reflects the environmental conditions of the whole lifespan of the animal. In addition, day 3 was excluded in the LMM testing the PL (Table 1), since 3 days is too short a period for detecting differences in copepod size. Egg to adult generation time for A. bifilosa at 17 • C is approximately 16 days of which ∼ 7.5 d taken by nauplii stages and ∼ 8.5 days by copepodite stages (Yoon et al., 1998). Collinearity between all explanatory variables was checked. Temperature was not considered in the models, because it changed similarly in all the bags (Paul et al., 2015). The model simplifications were done manually in a backward stepwise manner by removing the non-significant effects and by using Akaike's information criterion (AIC). We report t or z statistics (EH) of the retained fixed effects.
To separate the effect of the hatching environment from the maternal environment, EH and DI were divided by the corresponding values measured in the Baltic Sea water. The ratio of Mesocosm EH (or DI) / Baltic EH (or DI) > 1 indicates that eggs hatch or develop better in the maternal conditions (Mesocosm water), whereas the ratio < 1 indicates that eggs hatch or develop better in the Baltic Sea water. The effect of maternal environment (f CO 2 , TPC (< 55 µm) and C : N) on the ratio was tested with LMM, where the ratio of Mesocosm EH / Baltic EH and Mesocosm DI / Baltic DI were used as response variables; f CO 2 , TPC (< 55 µm) and C : N as fixed explanatory variables; and repeated measure of the mesocosms over time as a random factor. The model simplifications were made as above.
To test whether maternal ORAC correlates with egghatching success, Spearman rank correlation tests were used. Data from days 3, 10 and 17 were included in the test (n = 17, EH result for MC (mesocosm) 6 in day 3 is missing) because those are the days when both ORAC and EH were measured.
All the statistical analyses were performed using software R 3.0.2 (R Core Team, 2013), and the significance level was 0.05.
Neither food quantity (TPC, < 55 µm), food quality (C : N, < 55 µm), nor ocean acidification (f CO 2 ) had a statistically significant effect on copepod egg production (Table 2), even though there seemed to be variations in those parameters between the mesocoms (Table 3). Table 1. The structure of the full LMM or GLMM models that were used to test the effects of ocean acidification, food quantity and food quality on copepod EPR, EH, PL, ORAC, the ratio of EH mesocosm / EH Baltic and the ratio of nauplii DI mesocosm / DI Baltic. The sampling days that were included in each of the models are listed. Repeated measures of the same mesocosm bags were used as a random effect in all the models, because copepods that come from the same bags are more alike than copepods from different bags.
Response Ocean acidification X X X X TPC (< 55 µm) Food quantity C : N (< 55 µm) Food quality DI MC/Baltic (LMM) f CO 2 Ocean acidification X X X X TPC (< 55 µm) Food quantity C : N (< 55 µm) Food quality Table 2. T-statistics of the retained fixed effects in the linear mixed effect models testing the effects of TPC (< 55 µm), C : N and f CO 2 on EPR, female PL and female ORAC. Repeated measures of same mesocosm bags were used as a random effect in all the models, because copepods that come from the same bags are more alike than copepods from different bags. PL of Acartia sp. females increased during the first week of the study; however there seemed to be some differences between the mesocosms already on day 3, which was not included in the analysis (Fig. 1b). From day 10 onwards, the smallest A. bifilosa adults were found in the mesocosm with the highest f CO 2 concentration (Fig. 1b). f CO 2 , but also TPC (< 55 µm) had a statistically significant negative impact on copepod body size ( Table 2).
ORAC of the female copepods increased from day 3 to day 10 in all mesocosms (Fig. 1c). Interestingly, on day 3 ORAC was highest in the three mesocosms and had the highest f CO 2 treatment, whereas on day 31 the situation was reversed and ORAC was lowest in the three mesocosms with the highest f CO 2 (Fig. 1c). Despite this, only TPC (< 55 µm) had a statistically significant effect on ORAC, which decreases with increasing TPC ( Table 2).
The overall EH was high throughout the study: over 80 % of the Acartia sp. eggs hatched. As seen for EPR, PL and ORAC, EH also increased from day 3 to day 10 in all mesocosms (Fig. 1d). Variance in the EH between the four samplings was highest in the mesocosms with highest f CO 2 , whereas EH varied the least and remained > 90 % in both control mesocosms (MC 1, MC 5). In spite of this, only TPC (< 55 µm) had a statistically significant negative effect on EH (Table 4). Eggs that were produced in MCs 3, 5, 6 and 7 had fairly similar hatching success in Baltic water, whereas the hatching success of eggs that were produced in MCs 1 (control) and 8 (the highest f CO 2 ) was alternately either lower or higher than in the other MCs (Fig. 1e).  (Paul et al., 2015). Table 3. Ranges of f CO 2 , TPC < 55 µm and C : N < 55 µm that were used as explanatory variables in the full LMM and GLMM models. Three-day averages (measured within the latest three days of the sampling day) were used for testing the effects of the explanatory variables on copepod EPR, ORAC and EH, whereas average of all measurements since the start of the experiments until the sampling day were used when testing the effects of the explanatory variables on copepod size (PL). Variations in f CO 2 , TPC < 55 µm, and C : N < 55 µm in the course of the study are presented in Paul et al. (2015).

Egg-hatching and nauplii development in mesocosm vs. Baltic Sea conditions
Neither the maternal food quantity (TPC) nor the quality (C : N) had a statistically significant effect on offspring quality (EH and DI) in the egg transplant experiment ( Table 5). The f CO 2 was the only detected variable in the maternal environment that influenced the ratio of EH and DI between mesocosm and Baltic conditions. Egg-hatching success for eggs hatching in the mesocosm water differed from eggs hatching in the Baltic water. On days 3 and 10, hatching success was higher in the mesocosm water for the control (MC 1, MC 5) and for low-f CO 2treatment bags (MC 7, MC 6), whereas eggs produced in high-f CO 2 -treatment bags (MC 3, MC 8) showed higher hatching in the Baltic water (Fig. 2a). Thus, there may be a threshold f CO 2 for hatching success at high f CO 2 . However, on days 17 and 24 the f CO 2 treatment did not have a clear effect on hatching success. Nevertheless, f CO 2 had a statistically significant negative effect on the ratio of EH mesocosm / Baltic, meaning that egg hatching was higher in the Baltic water than in the maternal environment when the maternal environment had a high f CO 2 (Table 5). When the maternal environment had low f CO 2 , the situation was reversed. The level of f CO 2 also had a significant negative effect on the DI mesocosm / Baltic ratio ( Fig. 2b; Table 5).

Correlations between antioxidant capacity and offspring quality
Copepod ORAC was correlated significantly with copepod egg-hatching success. The relationship between the two variables is positive and stronger for eggs developing in the mesocosm water (ρ = 0.75, p < 0.001) than for eggs developing in the Baltic water (ρ = 0.62, p = 0.007) (Fig. 3).

Discussion
In this study, conducted in semi-natural mesocosm environments, reproduction of the Acartia sp. copepod showed high phenotypic buffering against acidification, i.e. the species was able to maintain similar egg production rates and high egg-hatching success in all f CO 2 conditions. Nevertheless,  Figure 2. Development of the ratio of (a) EH mesocosm / EH Baltic and (b) nauplii development index (DI) mesocosm / DI Baltic during the study. Ratio > 1: higher EH or DI in the mesocosm water (maternal environment) than in the Baltic Sea water, ratio < 1: lower EH or DI in the mesocosm water (maternal environment) than in the Baltic Sea water. Note that because of different development times, the DI values are not comparable between the days. The f CO 2 (µatm) values represent the average in days 1-43 (Paul et al., 2015). we found a significant negative effect of ocean acidification on adult female size. Even more interestingly, we found signs of a possible threshold at high f CO 2 for offspring develop- Table 5. T statistics of the retained fixed effects in the LMMs testing the effect of f CO 2 , TPC (< 55 µm) and C : N on ratio of EH mesocosm / EH Baltic and nauplii development index (DI) mesocosm / DI Baltic. Ratio > 1: higher EH or DI in the mesocosm water (maternal environment) than in the Baltic Sea water, ratio < 1: lower EH or DI in the mesocosm water (maternal environment) than in the Baltic Sea water. Repeated measures of same mesocosm bags were used as a random effect in both models, because copepods that come from the same bags are more alike than copepods from different bags.  (Fig. 2). However, we did not find support for the third hypothesis that lower TPC and higher C : N would weaken the maternal effect by deteriorating the condition of the mother. Conversely, higher TPC < 55 µm correlated negatively with egg-hatching success, adult female size and antioxidant capacity, whereas C : N ratio did not correlate with any of the measured variables significantly. Copepods were possibly food limited in all the mesocosms, especially after day 17 due to a sharp decline in Chl a concentrations and in phytoplankton community size structure (Paul et al., 2015). Dominance of picophytoplankton that are too small to be consumed by copepods could be the reason for the observed negative effects of food quantity and may also have masked the food quality effect. Also, after day 17 egg production rate was so low that it was practically impossible to find differences in egg production between the mesocosms. Finally, we found a positive correlation between maternal antioxidant capacity and egg-hatching success, suggesting that the female antioxidant defence might also protect the embryo from oxidative stress. The fact that Acartia sp. egg production and egg hatching were unaffected by high f CO 2 but the egg transplant experiment revealed that development was slower for nauplii at high CO 2 supports the importance of looking beyond egg production and egg hatching, which is also pointed out by Pedersen et al. (2014b). They concluded that the first endogenously feeding nauplii stages of Calanus finmarchicus are more sensitive to CO 2 -induced acidification than eggs or later nauplii stages (Pedersen et al., 2014b). Longer developmental times in high CO 2 / low pH have been observed in crustaceans, echinoderms and molluscs (Cripps et al., 2014a and references therein). Weydmann et al. (2012) also reported a significant developmental delay for Calanus glacialis eggs when exposed to highly acidified conditions. Pedersen et al. (2014a) observed that development of C4 copepodites of C. finmarchicus was delayed by 8.9 days in high-CO 2 treatments in comparison to the control condition, when also the previous generation had been exposed to the same conditions.

Response variable
We expected maternal effects to be most obvious in a highstress situation (high-f CO 2 treatments), as seen for three-spined sticklebacks in a study testing the effects of global warming (Shama et al., 2014). Instead, egg hatching was higher and nauplii development faster in the maternal environment than in the Baltic water, when the maternal environment had low f CO 2 (low stress). In the high-f CO 2 maternal environment the opposite response was observed, thus indicating that maternal effects are in fact weak and cannot compensate for the higher f CO 2 levels that correspond to nearfuture levels or that the eggs are damaged by the high-f CO 2 . This suggests that Acartia sp. and its reproduction are after all somewhat sensitive to ocean acidification. However, the effects were not as clear over the following weeks as at the beginning of the study, which may be due to an overall low egg number and large variation in hatching after day 17 or to acclimation of the copepods to the treatment conditions. In addition, the maternal effects seemed to weaken over time. This could be due to the weakening condition of the mothers. In the absence of fish predators, zooplankton density, especially Bosmina sp. (cladocerans), increased strongly in the mesocosms (Lischka et al., 2015). Senescence and food limitation were thus plausible problems for copepods and a likely cause of weakening maternal provisioning. In addition, conditions in the Baltic Sea changed after day 17 due to an upwelling event, which caused an increase in f CO 2 and decrease in pH (Paul et al., 2015). This might have made the Baltic conditions less favourable for copepod egg development and evened out the differences between high-f CO 2 mesocosms and the Baltic conditions.
A few studies have highlighted the importance of testing for transgenerational effects to avoid over-or underestimation of the effects of ocean acidification on copepods. Thor and Dupont (2015) found decreasing egg hatching of Pseudocalanus acuspes with increasing pCO 2 . In addition, transgenerational effects alleviated the negative effects on egg production and hatching of the second generation when the mothers had been acclimatised to the same treatment. Also, a reciprocal transplant experiment showed that the effect was reversible and an expression of phenotypic plasticity (Thor and Dupont, 2015). Contrary to the current study, Pedersen et al. (2014a) found no effect of the CO 2 environment on egg hatching or development of prefeeding nauplii stages N 1 and N 2 in their multigenerational study using C. finmarchicus. However, the development time of larger nau-plii and copepodite stages was increased by pCO 2 , although the development delay was not detected in the following generation (Pedersen et al., 2014a). Vehmaa et al. (2012) studied combined effects of ocean acidification and warming and found indications that negative effects on Acartia sp. reproductive success can partly be combated with maternal effects. The used pH treatments (−0.4 from ambient) were at the same level as the low-f CO 2 treatments in this study (MC 6, MC 7), which makes the results of the two studies consistent.
The measurements of female copepod antioxidant capacity were done in order to provide possible additional information of the maternal provisioning of the offspring. A preferable practice in oxidative stress studies is to measure several of the four components consisting of free radical production, antioxidant defences, oxidative damage and repair mechanisms (Monaghan et al., 2009). In the current study we only have the estimate for the defences, ORAC measurements, which makes our conclusions slightly more uncertain. However, an earlier study with the same species has indicated that, at intermediate stress levels, an upregulation of the antioxidant system enhances protection against oxidative damage, but at higher stress, the pro-oxidants may exceed the capacity of the antioxidant system and lead to oxidative damage (Vehmaa et al., 2013). In this study, upregulated antioxidant defence seemed to have a positive effect on offspring quality, as indicated by the positive correlation between female ORAC and egg-hatching success. Higher ORAC in the two highest f CO 2 mesocosms at the beginning of the study could be a sign of an upregulated antioxidant system in a sudden stressful situation, whereas the lowest ORAC in the highf CO 2 treatments at day 31 (Fig. 1c) could be caused by prolonged stress and exhausted antioxidant defence. The change from positive to negative effect in the course of the study could explain why f CO 2 did not show a significant correlation with ORAC, whereas food quantity (TPC < 55 µm) did. Ismar et al. (2008) showed that Acartia spp. development can be either slow or altered by certain algal groups causing death before the first copepodite or reproductive stage. A non-optimal diet could explain why higher food quantity would cause smaller adult female size, lower egg-hatching success or lower antioxidant capacity. Skeletonema-diatoms had fairly high abundance in the mesocosms during the first days of the experiment when egg-hatching success was lowest in every mesocosm, but then declined rapidly. Diatomdominated phytoplankton composition has been shown to cause low copepod egg-hatching success in the field (Miralto et al., 1999). Another quality aspect is the size and shape of the food, which may make it difficult to ingest or assimilate. From day 16 onwards, over 50 % of chlorophyll a was in picophytoplankton (< 2 µm) (Paul et al., 2015), which is too small for Acartia consumption (Rollwagen Bollens and Penry, 2003). Since we did not study what the copepods preyed upon, we can only speculate on diet quantity and quality. Satiated food conditions can strengthen the maternal or transgenerational effects. The transgenerational ef-fects were of minor importance for hatching success in C. finmarchicus when exposed to long-term high-CO 2 and foodlimited conditions (Pedersen et al., 2014a). Long-term stress and food limitation could thus also be a reason for weakening maternal effects in the current study.
We found body size (prosome length) to be negatively affected by high CO 2 . The result seems to be mostly driven by the mesocosm with the highest f CO 2 (MC 8), where the adult Acartia sp. copepods were smallest on all the four sampling times that were included in the analysis (days 10, 17, 24 and 45) (Fig. 1b). It takes ∼ 8.5 days for a sixthstage nauplius of A. bifilosa to develop through the five copepodite stages and reach adulthood at 17 • C (Yoon et al., 1998). According to the Bělehrádek's temperature function it takes 12-15 days for VI nauplii to reach adulthood at 9-11 • C (Bělehrádek, 1935;McLaren, 1966). The constants used in the equation (α = 1008, a = −8.701) were the same as used in Dzierzbicka-Glowacka et al. (2009) for the Baltic Sea Acartia spp. It is thus possible that the copepods could have developed through several stages causing the differences in prosome length between the treatments on day 10. Lowered pH may have increased copepods' energy requirements and if energy is reallocated towards maintaining homeostasis, their somatic growth can be reduced. Pedersen et al. (2014a) found C. finmarchicus body size to be inversely related to pCO 2 . They also found a higher respiration rate under more acidified conditions and claimed that increased energy expenditure via rising respiration and consecutive decreasing growth and reproduction could lower the energy transfer to higher trophic levels, thus hampering the productivity of the whole ecosystem (Pedersen et al., 2014a). This is especially alarming when considering the projected climate warming, since copepod size is negatively correlated with temperature (Foster et al., 2011). In addition to temperature, food quantity and quality can affect the copepod body size (Hart and Bychek, 2011) and create surprising combined effects with acidification. Garzke et al. (2016) reported an indirect positive effect of pCO 2 on copepod body size, which was explained by higher food availability when acidification acted as a fertiliser for phytoplankton. Temperature and food also interact because temperature affects the respiration and metabolism, thus the satisfying diet depends on temperature (Boersma et al., 2016). If high-CO 2 treatment (MC 8) caused a developmental delay in maturation, as could be interpreted from the prosome length results (Fig. 1b), the maturation would have occurred at a different temperature than in other mesocosms and possibly in non-optimal food conditions. Anyway, higher food quantity and quality would be expected to increase copepod size, contrary to our results. It is, therefore, possible that the used food quantity (TPC < 55 µm) and quality estimates (C : N < 55 µm) do not fully describe the diet that Acartia sp. was consuming in the mesocosms.
Adult copepods have in general shown robustness against acidification (Mayor et al., 2012;McConville et al., 2013), whereas eggs and nauplii appear to be more sensitive (Cripps et al., 2014b;Fitzer et al., 2012). In addition, there seem to be notable differences in sensitivity between species. Nauplii production, adult female fatty acid content and ORAC of Eurytemora affinis were not affected by f CO 2 in the current mesocosm campaign . Similarly, Lewis et al. (2013) found differences in ocean acidification sensitivity between the species Oithona similis and Calanus spp. (C. glacialis and C. hyperboreus). They argued that O. similis is more sensitive to future ocean acidification than Calanus spp., because O. similis remains in the surface waters whereas Calanus spp. migrates vertically and encounters wider pCO 2 ranges daily than O. similis (Lewis et al., 2013). The same applies to Acartia sp. and E. affinis in our study area. Although Acartia spp. is exposed to natural variability in pH environment due to daily variations as well as due to staying at greater depths during the day (low pH in deep water), it does not reside as deep as E. affinis (Almén et al., 2014) and may, therefore, show higher sensitivity than E. affinis during the current mesocosm campaign .
The results obtained for Acartia sp. reproduction in the current study seem to contradict the results obtained for the Acartia sp. abundance determined in the mesocosms. Although our results indicate that Acartia sp. reproduction is in fact sensitive to ocean acidification, no f CO 2 effect was found for the abundance of this species (Lischka et al., 2015). It is possible that 45 days was not long enough to detect small negative effects of CO 2 on copepod size, egg hatching and nauplii development, to be reflected in copepod abundance. In addition, especially at the beginning of the study, Acartia eggs in the mesocosms might have ended up in the sediment trap before hatching due to slow development at low temperatures, which might have made it difficult to detect differences in Acartia abundance between the mesocosms. On a longer timescale, small acidification-induced delays in offspring development could translate into negative effects for the copepod population and on energy transfer within the pelagic food web. In addition, warming will probably enhance the sensitivity of the species towards ocean acidification (Vehmaa et al., 2012(Vehmaa et al., , 2013.

Conclusions
Our results support the idea that it is important to look beyond egg production as hatching and development can be more sensitive to ocean acidification. Parental effects will likely be important in mediating some of the negative effects of ocean acidification. For Acartia sp., the transgenerational (maternal) effects may alleviate negative impacts of ocean acidification but potentially only under exposure to medium levels of CO 2 . We did not find support for the hypothesis suggesting that poorer food quantity and quality would weaken the maternal effect by deteriorating the condition of the mother. This could be due to the overall food lim-itation, especially during the latter half of the study, or the fact that our estimates of food quantity and quality did not describe the diet in a satisfactory manner. Nevertheless, maternal antioxidant defence seems to correlate positively with offspring egg-hatching success. Overall, these results indicate that Acartia sp. could in fact be affected by CO 2 levels predicted for the year 2100 (IPCC, 2007). However, it is important to remember that this study shows how today's copepods would react to tomorrow's world; thus these results do not take into account the possible effects of evolutionary adaptation. Transgenerational effects can buffer short-term detrimental effects of ocean acidification, thus giving time for genetic adaptation and assisting persistence of populations under climate change.
The Supplement related to this article is available online at doi:10.5194/bg-13-6171-2016-supplement.