Copper pollution exacerbates the effects of ocean acidification and warming on kelp microscopic early life stages

Ocean warming (OW), ocean acidification (OA) and their interaction with local drivers, e.g., copper pollution, may negatively affect macroalgae and their microscopic life stages. We evaluated meiospore development of the kelps Macrocystis pyrifera and Undaria pinnatifida exposed to a factorial combination of current and 2100-predicted temperature (12 and 16 °C, respectively), pH (8.16 and 7.65, respectively), and two copper levels (no-added-copper and species-specific germination Cu-EC50). Meiospore germination for both species declined by 5–18% under OA and ambient temperature/OA conditions, irrespective of copper exposure. Germling growth rate declined by >40%·day−1, and gametophyte development was inhibited under Cu-EC50 exposure, compared to the no-added-copper treatment, irrespective of pH and temperature. Following the removal of copper and 9-day recovery under respective pH and temperature treatments, germling growth rates increased by 8–18%·day−1. The exception was U. pinnatifida under OW/OA, where growth rate remained at 10%·day−1 before and after copper exposure. Copper-binding ligand concentrations were higher in copper-exposed cultures of both species, suggesting that ligands may act as a defence mechanism of kelp early life stages against copper toxicity. Our study demonstrated that copper pollution is more important than global climate drivers in controlling meiospore development in kelps as it disrupts the completion of their life cycle.


Results
Meiospore germination. After 6 days, the percentage of germinated meiospores (i.e., with visible germ tube, Fig. 1) was calculated for both species. OA and OW had no significant effect on the germination of meiospores of both M. pyrifera and U. pinnatifida (germination >85%) (Fig. 2). A significant (P < 0.001) detrimental effect of copper (i.e., 5-18% reduction) on germination of meiospores of both species was observed in all treatment combinations, except under ambient temperature and current pH (Fig. 2). The greatest (P < 0.05, Tukey test) additional effect of copper in the reduction of germination was observed at current pH and OW in U. pinnatifida (18%) (Fig. 2b).
Gametophyte size. Sexual differentiation of gametophytes ( Fig. 1) occurred at the 15 th day of culture for both species under No-Cu conditions. Gametophyte development and sexual differentiation were significantly (P < 0.001) inhibited by copper exposure (Fig. 4). In the No-Cu treatment, gametophytes of M. pyrifera grew significantly bigger (18% increase for males and 46% increase for females; P < 0.05, Tukey test) under OA at 16 °C, and males were significantly (P < 0.001) bigger (29-54%) than females under all pH and temperature combinations (Fig. 4a). In contrast, only the size of female gametophytes of U. pinnatifida was significantly (P < 0.05, Tukey test) reduced (24%) by OA at 12 °C compared to those at pH T 8.16 at 12 °C (Fig. 4b). Germling growth rate during the recovery period. The growth rate of sexually ambiguous germlings ( Fig. 1) during the recovery period (after stopping copper addition to the media of the Cu-EC 50 treatments on day 9) was calculated from day 12 to 18 for both species. In M. pyrifera, recovery of germling growth rate was significantly (P = 0.026) 7 and 25% greater in OA conditions compared to the current pH T treatment at 12 or 16 °C, respectively (Fig. 5a). In contrast, OW, OA and their interactions did not significantly affect germling growth rate of U. pinnatifida during recovery (Fig. 5b). Moreover, recovering germlings of both kelps did not differentiate into male or female gametophytes by the end of the experimental period (day 18). When comparing the growth rate during the recovery period ( Fig. 5) with that during the Cu-EC 50 exposure (Fig. 3), the germling growth rate of M. pyrifera significantly (P < 0.001) increased. That increase in growth rate of M. pyrifera was 29-33% greater under pH T 7.65 and 12 °C compared to pH 8.16 and 16 °C (P < 0.05, Tukey test). There were no statistical differences between growth rate during copper exposure and recovery for U. pinnatifida germlings under all pH and temperature combinations.
Total dissolved copper (Cu T ) concentrations. At the nominal copper concentrations corresponding to the Cu-EC 50 for M. pyrifera (2.36 µM Cu) and U. pinnatifida (3.62 µM Cu), Cu T concentrations measured in the fresh media were 2.51 and to 3.84 µM Cu T , respectively (Table 1). During the first 9 days of culture, when the media corresponding to the different treatments was renewed every third day, the dissolved concentration of Cu T was reduced by 64-71% for M. pyrifera and by 66-72% for U. pinnatifida, under all temperature and pH treatment combinations . Although copper was not added on the 12 th and 15 th day of culture, Cu T was still detectable in the culture media, but at much lower concentrations compared to the period when copper-treatment media was renewed every 3 days (Table 1).

Figure 2.
Percentage meiospore germination of (a) M. pyrifera and (b) U. pinnatifida after 6 days of culture in a factorial combination of two temperatures (12 and 16 °C), two pH (pH T 7.65 and 8.16) and two copper (No-Cu, and Cu-EC 50 = 2.36 and 3.62 µM Cu for M. pyrifera and U. pinnatifida, respectively) treatments. Bars represent mean ± SD (n = 4). Significant subgroups are grouped by the lowercase groups as a > b > c (Tukey, P < 0.05). Note that the y-axis has a break from 10 to 70% in both graphs. pinnatifida, under all temperature and pH treatment combinations . Although copper was not added on the 12 th and 15 th day of culture, Cu' was detectable but at much lower concentrations compared to the copper-added period ( Table 1).

Copper-binding ligand (L) concentration.
Due to the high Cu T in the media during days 0-9, L concentrations present in the cultures were undetectable. After the removal of Cu from the media (day 9), L was detected on day 12 and 15 (Table 1). At 12 °C and under both pH treatments, L release was higher (by >40%) on day 12 than day 15 in the culture media of both M. pyrifera and U. pinnatifida. At 16 °C, L release in the culture media of both kelps under pH 8.16 was higher (by >30%) on day 12 than day 15 while under pH 7.65 L production was lower (>50%) on day 12 than day 15 (Table 1).

Discussion
Our results reveal that local anthropogenic drivers such as copper pollution have a greater impact on kelp meiospore survival and ontogenic development than global climate drivers such as OW and OA. While the independent effects of OW and OA on different early life history stage processes (e.g., meiospore germination and gametophyte growth) are mostly insignificant, the effect of copper is negative and magnified through the different developmental stages. For example, in our experiments, copper exposure (Cu-EC 50 ) had only a moderate negative effect (5-18% reduction) on meiospore germination for all OW treatment and the ambient temperature/OA treatment and no effect for the ambient pH/temperature treatment. However, the subsequent growth of germlings was reduced by 43-68%, and sexual differentiation was inhibited regardless of seawater temperature and pH. The different sensitivities of early life history stage processes, not only to copper exposure, but also to other environmental drivers, are related to the fact that meiospore germination is an autogenous process supported by cellular  [38][39][40] while gametophyte growth and subsequent life-history processes are dependent on the photosynthesis and factors affecting this process 40,41 .
Despite the initial germination process being autogenous, copper exposure can interfere with germ tube initiation in brown seaweeds. In the Fucales, Ca 2+ movement across the cell membrane of zygotes generates an electrical gradient that initiates the movement of negatively charged vesicles into the basal pole, leading to adhesion and rhizoid formation and germ tube formation 42,43 . An excess of copper may alter Ca 2+ membrane permeability inhibiting cellular polarization and delay germination in brown macroalgae 42,43 . For example, germination and rhizoid elongation in the brown macroalgae Fucus serratus (Order Fucales) was inhibited by copper at 0-2.11 µM Cu 44 while Cu-EC 50 for spore germination in M. pyrifera and U. pinnatifida are 2.36 and 3.62 µM Cu T , respectively. Following the transport of copper into the cytosol, copper disrupts enzyme-active sites and cell division 45,46 . In various organelles, copper interferes with mitochondrial electron transport, respiration, ATP production, and photosynthesis in the chloroplast 47 . The multiple effects of copper in subcellular organelles is likely responsible for the more pronounced effects on germling and gametophyte growth compared to meiospore germination in our experiment.
In general, the production of L seems to be the first line of defence by macroalgae against copper toxicity conferring some degree of tolerance by neutralizing the negative effect of copper 21,24 . In our study, when copper was removed, M. pyrifera germling growth rate was significantly enhanced under OA regardless of temperature whereas U. pinnatifida germling growth did not significantly increase under OA, OW or ambient conditions, indicating a more serious disruption of the development of M. pyrifera germlings under copper stress. In addition, since L cannot to be detected when Cu' is in excess of CuL 27 , after removing copper from the medium at day 9 onwards, L was detected in cultures of both species. L in the No-Cu treatments was significantly lower than that in Cu-EC 50 treatment, especially for U. pinnatifida, suggesting that L release is an active response to Cu stress that enabled the cells to resume growth during the recovery phase. The production of L in response to copper exposure has been reported in adult macroalgae 21,24 , but, to our knowledge, this is the first study showing L production by microscopic early life history stages of the Order Laminariales. treatments under control (no-copper addition) treatment. No data is available for the Cu-EC 50 treatment as sexual differentiation was inhibited by copper exposure. Bars represent mean ± SD (n = 4). Significant subgroups are grouped by the lowercase groups as a > b > c > d (Tukey, P < 0.05). Note that the y-axis has a break from 100 to 500 µm 2 .
Scientific RePoRts | (2018) 8:14763 | DOI:10.1038/s41598-018-32899-w L were not detectable during copper exposure (day 1-9) using our method, but the observed difference between Cu T and Cu' under No-Cu and Cu-EC 50 treatment indicates the presence of L, and we suggest that L was already actively produced during that period. The apparent L production may have helped to protect and detoxify kelp cells, thereby promoting germling growth, albeit at very low rates.
Considering that M. pyrifera and U. pinnatifida were exposed to different germination Cu-EC 50 values (2.36 µM for M. pyrifera and 3.62 µM for U. pinnatifida 5 ), it is noteworthy that growth rate under these Cu-EC 50 treatments (Fig. 3) was comparable between the two species. However, during recovery (day 12-18), M. pyrifera germling growth rates were significantly enhanced under OA, regardless of temperature while the growth of U. pinnatifida germling remained at the same rate at that observed during copper exposure, regardless of pH and temperature treatments. As the Cu-EC 50 was 35% greater for U. pinnatifida (3.84 µM Cu T ) compared to M. pyrifera (2.51 µM Cu T ), it is likely that the higher levels of remaining Cu in the U. pinnatifida cultures adsorbed into the container and/or the cell wall and negatively affected growth rate recovery germlings. In contrast, there was less Cu remaining in the M. pyrifera cultures and the higher pCO 2 ( Table 2) enhanced the growth rate of germlings.
The species-specific response to copper toxicity observed for M. pyrifera and U. pinnatifida may also be attributed to bacteria that inhabit macroalgal surface. The surface of macroalgae is a nutrient-rich habitat that is optimal for colonization by bacteria 48 . Bacteria associated with macroalgae can also exude L for binding and transport metals required for several physiological processes of bacteria such as nitrogen fixation 49 . Thus, L released by bacteria can play an additional protective role for macroalgae against metal pollution 24,49 . For example, research on M. pyrifera from California indicated that populations from highly copper-polluted coasts have epibiotic bacteria with greater copper tolerance compared with those from less copper-impacted coasts 50 . Therefore, although bacteria were not observed by light microscopy during our experiment, it is possible that bacteria increased the metal tolerance of different life stages of kelps under copper exposure, but this needs further investigation.
The effects of copper on M. pyrifera and U. pinnatifida became apparent at the germling and gametogenesis stages of both species, with the growth rate of germlings being significantly lower and gametophyte development (growth and sexual differentiation) arrested in all temperature and pH treatments when copper was added. Trace  51,52 . At high concentrations, ROS are toxic to all organisms, oxidizing proteins, lipids and nucleic acids that often lead to structural aberrations, mutagenesis, and cell death 51 . Consequently, the presence of ROS likely resulted in the observed low germination and growth rates under copper exposure. In addition, this response might be related to copper inhibiting the utilization (but not the production) of vesicle-stored reserves, e.g., alginic acid 38,40,53 , as the growth of germlings depends on the formation of the new cell wall that contains alginic acid 54,55 . This suggests that copper was preventing cell expansion and growth of M. pyrifera and U. pinnatifida germlings during the Cu-EC 50  The initial Cu T concentrations were reduced during the 9-day Cu-EC 50 exposure likely due to adsorption into exopolysaccharides. This suggestion is supported by the observations that macro-and microalgae and bacteria, over-produce cell wall polysaccharides in response to trace metal pollution to avoid absorption of metals into the cell 56,57 . The negatively-charged active groups (i.e., hydroxyl, sulphate, and carboxyl) of polysaccharides are strong ion-exchangers, and so have a high capacity to bind (i.e., bioadsorb) trace metal ions such as Cu 2+ 56 . The extracellular polysaccharide, alginate (i.e., insoluble salt of alginic acid), occurs naturally in brown macroalgae (Ochrophyta, Phaeophyceae) as a major structural component of the matrix of cell wall 56 . It is possible that developing meiospores (in both No-Cu and Cu-EC 50 treatments) produced enough alginate to block cellular entry of copper to the cytosol, thereby limiting subcellular toxic effects of copper. This protective mechanism might have been operating during the development of M. pyrifera and U. pinnatifida meiospores in our experiment, but to assess this, further studies on the production of cell wall polysaccharides by kelp meiospores during copper exposure are required.
Millions of meiospores (e.g., >5 × 10 3 cell mL −1 cm of sorus area −2 ) are produced by one fertile sporophyte 29 so the individual and interactive effects of OW and OA on meiospore germination would be small (15%) and cause little concern. However, upon release, the meiospores are exposed to different abiotic drivers that can already significantly reduce their number before any effect of OW and OA (Fig. 6). These factors include: (1) large-scale hydrodynamics, such as currents affecting the density and the physical transport of the larval pool; (2) micro-hydrodynamics, such as small-scale currents and spatial variability that may determine settlement; and (3) substrate availability and quality, substrate preference and spore settlement behaviour, e.g. phototaxis, chemotaxis 58 . Furthermore, grazing on gametophytes and juvenile sporophytes can further contribute to the decimation and the collapse of the local kelp population 59,60 . The surviving individuals (gametophytes) give rise to the next life history stage (sporophytes) which may be able to withstand exposure to OW and OA due to better acclimation and subsequent adaptation 61 . However, the effect of copper, a relevant local stressor, is more concerning as sexual differentiation and subsequently, sexual reproduction will be arrested further compromising the development of the next generation of sporophytes.

Methods
Preparation of trace metal clean, laboratory-ware. All laboratory-ware used for stock solution preparation, seawater sampling and meiospore cultures were acid-cleaned and ultrapure water-rinsed to reduce contamination by trace metals contamination 36 , microalgae and bacteria 24 . Manipulation (e.g., culture media renewal and sampling) during the experiment was performed inside a laminar flow cabinet to minimize contamination 36 . Copper stock solution and nominal concentrations. The copper stock solution was prepared by dissolving CuCl 2 in ultrapure water (2 g L −1 CuCl 2 , i.e., 14.9 mM Cu) in a 100-mL polycarbonate bottle (Nalgene TM , Nalge Nunc International Corporation, NY, U.S.A.) and stabilized by adding 2 M HCl until reaching pH 2.4 36 . This stock solution was prepared at the beginning of the experiment. The nominal copper concentrations used in this experiment were the theoretical Cu concentration received when diluting the stock solution for each treatment. We aimed for the species-specific Cu T concentrations that inhibits 50% of germinations (Cu-EC 50 treatment) of 2.36 µM for M. pyrifera and 3.62 µM for U. pinnatifida 5 . No-Cu treatment corresponded to the media without any Cu stock addition.
Total dissolved copper (Cu T ) analysis. Cu T in the 0.2 µm-filtered solution was measured in the fresh culture media before exposure to the meiospores and at day 0-15. Cu T in the culture media was determined from Figure 6. Schematic representation of the bottleneck effect produced by drivers that may influence meiospore settlement and subsequent development into an adult sporophyte population of kelps. Results of the current experiment indicate that early life stages of kelps are susceptible to the interaction between OW, OA and Cu but other drivers may affect the same and different developmental stages. For example, swimming meiospores are affected by large-scale hydrodynamics (e.g., currents) impacting their density and physical transport to the substratum. Micro-hydrodynamics (e.g., small-scale currents) may determine meiospore settlement and subsequent development. Simultaneously, early life history stages of kelps are constantly stressed by the interactions of abiotic (e.g., OW, OA, Cu) and biotic (e.g., grazing) drivers. All these interactions control the dynamic and structure of the adult populations. Adapted from Pineda 66 . Diagram is not drawn to scale.
Scientific RePoRts | (2018) 8:14763 | DOI:10.1038/s41598-018-32899-w two replicates of each copper treatment and the analytical blanks. An amount of 0.15 mL of culture medium was diluted in 4.25 mL of ultrapure water and acidified with 0.10 mL of HNO 3 (acidified sample, 4.5 mL 2% HNO 3 ) and stored until analysis. Total copper concentrations were quantified by inductively coupled plasma mass spectrometry (ICP-MS) 5 .

Labile copper (Cu') and copper-binding ligand (L) analyses. Cu' and L in solution were measured
from the meiospore culture media on day 0-15. Due to a large number of samples and time requirements for each analysis, Cu' and natural L concentrations were determined from one sample of each copper treatment by cathodic stripping voltammetry (CSV) of freshly thawed samples at the ambient pH. Good reproducibility was demonstrated for one sample (i.e., U. pinnatifida, 12 °C and pH T 7.65) on day 6. Due to the limited seawater volumes available, the large number of samples and the fact that most of the samples contained high amounts of copper (i.e., in excess of natural L), a kinetic approach as described for Fe with 1-nitroso-2-naphthol in Witter and Luther III 27 was used here for Cu with salicylaldoxime (SA) as the competing L 62 . Briefly, to determine Cu' , [CuL] and therefore [L], SA was added to a final concentration of 100 µmol L −1 . The current was monitored immediately after the addition of SA and until the system reached equilibrium, which was reached at a maximum of 2 h. The current measured right after the addition of SA represents the Cu' and the difference after 2 h additionally include [CuL]. This assumes that due to the large excess of SA over L, any Cu' formed during the dissociation of CuL will react faster with SA than with L, and the product CuSAx will not revert to Cu' during the timescale of the analysis 27 . At the end of the equilibration time a two-fold standard addition was performed to derive the Cu' concentration. The standard addition curves were in all cases linear, indicating that no free L was present at that time. For samples at Cu-EC 50 concentrations, no CuL could be detected, i.e. there was no significant difference between the current measured right after the addition of SA and after equilibration of 2 h. However, we could still derive Cu' from our analysis. For No-Cu treatments and samples taken during the recovery period, i.e. 9-15 days, the current increased during the equilibration time and [CuL] could be approximated. The errors associated with the analysis, however, were too large to derive meaningful stability constants from these measurements and we therefore report only the L concentrations. Seawater pH measurements. The seawater pH during the experiment was measured on the total scale (pH T ) at 12 and 16 °C using a pH electrode (Orion ROSS Sure-Flow semi-micro, ORI8175BNWPW) connected to a pH meter (Thermo Scientific Orion 720 A pH/ION Meter). The electrode slope was determined using temperature equilibrated pH 7 and pH 9 buffers (colour coded, NIST traceable). pH was measured on the total scale using TRIS and 2-aminopyridine buffers in synthetic seawater to calibrate the electrode 66 . Seawater samples representing the two pH T treatments were collected and fixed with mercuric chloride for determining seawater carbonate chemistry. Total alkalinity (AT) was measured using the closed-cell titration method and dissolved inorganic carbon (DIC) was measured directly by acidifying the sample 66 . The seawater carbonate chemistry of each pH treatment at both temperatures was calculated using the measured AT, DIC, pH, salinity, and temperature (Table 2) with the SWCO2 software 67 .
Seawater pH treatments. The seawater used in the experiment was collected at the same time as the sporophylls and had a salinity of 36‰. The seawater was filtered (0.2 µm) to reduce microalgal and bacterial contamination and kept overnight in previously sterilized 2 L-polycarbonate bottles at the respective temperature treatment before use. After filtration, nutrients (10 µM NaNO 3 and 1 µM NaH 2 PO 4 ) were added to the seawater to avoid nutrient limitation. The present pH treatment corresponded to the non-manipulated seawater (pH T 8.16, defined as ambient treatment). To obtain the lowest seawater pH treatment (pH T 7.65, defined as OA treatment), equal volumes of 0.5 M HCl and 0.5 M NaHCO 3 were added to the seawater 68,69 until pH T reached 7.65 at 16 °C. Seawater with the corresponding pH T was freshly prepared every three days to renew the culture medium.
Effects of seawater temperature, pH, and copper on meiospore development. Meiospore release and cultivation were performed as described in Leal et al. 32 . Briefly, from each species, discs (2 cm 2 ) of mature sorus were cut from the sporophylls using a cork borer. Pools of excised sori (total of ca. 50 g of 2 cm 2 discs each) of both kelp species were separately immersed in seawater of the different pH T and temperature treatments for 15 min. After release, meiospores were dispensed (final density of 25,000 cell·mL −1 ) into culture flasks (Corning ® 75 cm 2 , polystyrene cell culture flask with phenolic-style cap) containing seawater with the corresponding pH T and temperature but without copper addition. To avoid meiospore mortality during swimming and settlement that could change the initial density, meiospores were allowed to settle for 3 h before exposure to the respective nutrient-amended seawater and copper concentration treatments. Exposure to copper lasted 9 days, which is the time observed to inhibit gametogenesis in both kelp species 5 . Meiospore cultures of both species under the respective pH T and copper treatments were prepared in two sets and each one was exposed to the respective temperature treatment ( Chamber, Contherm Scientific Co. Ltd., New Zealand). PAR (metal halide lamps Philips HPI-T 400 W quartz), with a photoperiod of 12 h light: 12 h dark, was measured with a spherical quantum sensor (LI-193, LI-COR, Lincoln, Nebraska) connected to a light meter (LI-250A, LI-COR, Lincoln, Nebraska) and adjusted to 55 ± 2 and 54 ± 1 µmol photons·m −2 s −1 in the 12 °C-and 16 °C-culture room, respectively. Thereafter, Cu-treated samples were allowed to recover using the culture medium with nutrients but without copper addition, under the respective scenario. Analytical blanks (i.e., seawater with each copper concentration under the respective pH and temperature conditions but without biological material) corresponding to each copper treatment were also prepared. The media of the meiospore cultures with the appropriate pH T , copper, and nutrients (to avoid nutrient depletion), were renewed every 3 days. Cu T concentrations in the treatments and blanks were measured as described above.
Meiospore development. M. pyrifera and U. pinnatifida meiospore germination (%), germling growth rate (%·day −1 ), gametophyte size (µm 2 ) and sex ratio, during the experiment, were obtained from photograph (5.1 M CMOS camera, UCMOS0510KPA) taken every three days from at least five haphazardly chosen visual fields, using an inverted microscope (200×, Olympus CK2; Olympus Optical Co. Ltd., Tokyo, Japan). Photographs were analysed using the ToupView 3.5 digital camera software (ToupTek Photonics, Zhejiang, China). Meiospores with visible germ tubes were considered germinated and the germination percentage was calculated from 350 individuals per replicate after 6 d of culture. The size of sexually ambiguous growing meiospores (germlings) and sexually-differentiated male and female gametophytes was obtained from an average of 30 individuals per replicate after 12 and 15 d of culture, respectively. Germling growth rate under No-Cu and Cu-EC 50 treatments were separately calculated during exposure and recovery. For germlings under No-Cu treatment, growth rate was calculated from 0 to 12 d, before sexual differentiation was observed in both kelps. For germlings under Cu-EC 50 treatment, growth rate was calculated during copper exposure (0-9 d) and during recovery period (12-18 d). Growth rate (%·day −1 ) was calculated as [(W t /W 0 ) 1/t −1] × 100, where W 0 is the initial size, W t is the final size, and t is days of culture 70 . At day 15, when sexual ambiguity was resolved, male and female gametophytes were counted and the sex ratio, expressed as the frequency of males per progeny, was calculated as no. ♂/(no. ♂ + no. ♀) 69 .

Statistical analyses.
We did not statistically compare the two species in the present work because M. pyrifera and U. pinnatifida showed species-specific responses to copper 5 , OA and/or OW 32,33 in previous studies. Percentage germination and germling growth rate (%·day −1 ) were logit transformed 71 . All the data satisfied Normality (Kolgomorov-Smirnov test) and homogeneity of variances (Levene's test). Three-way ANOVA (P < 0.05) was used to test the statistical significance of differences in meiospore germination, germling growth rate during copper exposure, gametophyte size, germling growth rate (copper exposure vs. recovery period) between temperature, pH, copper treatments, and sex. Two-way ANOVA (P < 0.05) was used to test the statistical significance of differences in gametophyte sex ratio and germling growth rate (during recovery) between temperature and pH. When significant interactive effects were observed in the ANOVAs (at α = 0.05), the significant main effects of the factors (i.e., temperature, pH, copper treatments and sex) were subordinated, and the interaction(s) becomes the focus of the analysis 72 . A post hoc Tukey test (P < 0.05) was applied when a significant effect (single, two-and/or three-way interactions) of independent variables was observed. The ANOVA analyses were run using the software SigmaPlot version 12.0 (Systat Software, Inc., San Jose, CA). ANOVA statistical results for M. pyrifera and U. pinnatifida are listed in Supplementary Tables S1 and S2, respectively.