Lifecycle-dependent toxicity and removal of micropollutants in algal cultures of the green seaweed Ulva (Chlorophyta)

The green marine seaweed Ulva (Chlorophyta) is widely suggested as a biofilter for cost-effective bioremediation in integrated multitrophic aquaculture and wastewater management. Micropollutants (MPs), including antibiotics, endocrine disruptors, and herbicides, can severely affect humans and the environment. As these compounds may be accumulated or transformed by Ulva, its simultaneous function as an efficient biofilter and as a food and feed source might be affected. Therefore, we investigated the removal of ten MPs often found in wastewater effluents by Ulva and its associated bacteria, and characterized the effects of these MPs on the alga during two crucial lifecycle phases (germination and vegetative growth) using dose dependent tests. We monitored MP detoxification at elevated concentrations in a reductionistic tripartite Ulva mutabilis-Roseovarius-Maribacter model system to reduce interference from the fluctuating algal microbiome. Our results showed that the tripartite community was resistant to the MPs tested, although the gametes were between 2 to 140 times more susceptible based on the half-effective concentrations (EC50) than the growing vegetative alga. The herbicide atrazine and the endocrine disruptor bisphenol A proved the most toxic MPs for germinating gametes. U. mutabilis and its associated bacteria could not eliminate the tested antibiotics and herbicides but efficiently reduced the concentration of endocrine disruptors, including bisphenol A, estradiol, and ethinylestradiol, by over 98% to below the detection limit. We also confirmed that Ulva is not likely to become contaminated under the studied exposure conditions because no biotic processes are used to remove the other MPs, which emphasizes yet another benefit of its use in aquaculture. Compared to green microalgae, U. mutabilis appears to be more resistant to micropollutants.


Introduction
The use of algae to remediate nitrogen, phosphorus, heavy metals, and organic contaminants in wastewater and the environment has been suggested in several studies (Norvill et al. 2016;Mohsenpour et al. 2021;Viegas et al. 2021).Macroalgae, in particular, are regarded as an effective bioremediation technology with numerous advantages; notably, they are simple to separate from water, facilitating efficient harvesting and retention in a potential cleaning system (Qiu et al. 2017;Liu et al. 2020).In addition, seaweeds are considered a cost-effective, adaptable, and economical alternative tool for bioremediation (Arumugam et al. 2018).Among the seaweeds, the green macroalga Ulva (Chlorophyta) has various applications in integrated multitrophic aquaculture (IMTA) (Neori et al. 2003;Bolton et al. 2009;Guttman et al. 2018) as it incorporates nitrogen, phosphorus, and carbon dioxide to generate biomass for biofuel and commodity chemicals (Baghel et al. 2015;Zollmann et al. 2021;Simon et al. 2022).Ulva is often used to clean effluent from fish discharge and water for recirculation (Neori 1991;Shpigel et al. 1993;Lawton et al. 2013).
Ulva mutabilis, which is conspecific with the cosmopolitan Ulva compressa (Steinhagen et al. 2019), serves in our study as a unique model system to investigate the requirements for withstanding environmental stresses, such as micropollutants (MPs) (Ghaderiardakani et al. 2020;Vockenberg et al. 2020).MPs include antibiotics, hormones, and herbicides, and given their known negative effects on humans and the environment, exposure should be minimized wherever possible.Thus, Ulva might prove to be a powerful tool for efficient water purification and bioremediation.At the same time, herbicides and antibiotics may, however, have negative impacts on the macroalga and its microbiome.
Ulva mutabilis is excellently suited for researching bacterial-algal interactions and investigating specific bacteriainduced growth under environmental stress because it relies on the presence of bacteria for its complete development and morphogenesis (Spoerner et al. 2012).Notably, the natural microbiome of U. mutabilis can be reduced to two bacterial strains, Roseovarius sp.MS2 and Maribacter sp.MS6.These strains supply all essential algal growth-and morphogenesis-promoting factors (AGMPFs) (Ghaderiardakani et al. 2019), and the resulting tripartite community is stable and replicable in the laboratory (Wichard et al. 2015;Wichard 2023).This reductionistic microbiome minimizes bacterial interactions, thereby enabling experimental focus on algal remediation.
In addition, the ability of Ulva to act as a biofilter for various substance raise concerns about its usage as a source of food and feed.Indeed, antibiotics like chloramphenicol (CAP), erythromycin (ERY), oxytetracycline (OTC), and sulfamethoxazole (SMX) are often used in aquaculture (Chen et al. 2020).Additionally, xenoestrogens like bisphenol A (BPA), and 17α-ethinylestradiol (EE 2 ) as well as the most active natural estrogen, 17β-estradiol (E 2 ), can be found in water, thereby affecting the hormone homeostasis of vertebrates and invertebrates (Hoga et al. 2018).BPA is a highly concerning MP as it is almost ubiquitous and pseudo-persistent (Zhang et al. 2019;Zielińska et al. 2019;Hardegen et al. 2021).Herbicides such as triazine, chloracetamide, and glyphosate herbicides were selected to cover a broad range of herbicides that target enzymes identified in the U. mutabilis genome (De Clerck et al. 2018).Notably, atrazine (ATZ) blocks the quinone-binding site in photosystem II resulting in plant death through increased reactive oxygen species generation (Rose et al. 2016).Metolachlor (MTC) inhibits several enzymes involved in cell division and elongation.Fatty acid elongation and geranylgeranyl pyrophosphate cyclization enzymes are two sensitive targets.Glyphosate (PMG, N-(phosphonomethyl) glycine) inhibits enolpyruvylshikimate-3-phosphate synthase, which is required for the synthesis of aromatic amino acids via the shikimate pathway (Rose et al. 2016).
In our study, we thus investigated the removal of various MPs by U. mutabilis and its associated bacteria.Ten substances were chosen to represent antibiotics, endocrine disruptors, and herbicides (Fig. 1a) given their frequent detection in wastewater effluents, surface water, and groundwater (Hirsch et al. 1999;Kummerer 2009;Luo et al. 2014;Baldwin et al. 2016;Das et al. 2017;Sousa et al. 2018;Tran et al. 2018;Montiel-Leon et al. 2019;Nguyen et al. 2022).
We examined the effects of these substances on the survival and growth of U. mutabilis during two phases of its lifecycle using dose-response tests (Fig. 1b) and determined the ability of the alga and its associated bacteria to remove the MPs using ultra-high-performance liquid chromatography (UHPLC) coupled with electrospray ionization high-resolution mass spectrometry (ESI-HRMS).Our study highlights the resilience of the U. mutabilis tripartite community against the tested MPs as well as its suitability as an ecotoxicological model system under standardized conditions for investigating the mortality and sub-lethal effects of specific MPs, such as germination and developmental delays.

Ulva and bacteria cultivation
Haploid gametes and gametophytes of Ulva mutabilis Føyn (sl-G[mt +]; morphotype 'slender'; locus typicus: Ria Formosa, Portugal) were used in all experiments.The cultivar is referred to as Ulva.mutabilis throughout this study.Ulva mutabilis was cultivated in artificial seawater Ulva culture medium (UCM) at 18 °C ± 2 °C with a light/dark cycle of 17/7 h and at a light intensity of 40-80 µmol photons m −2 s −1 (Califano and Wichard 2018;Nahor et al. 2021).The bacterial strains, Roseovarius sp.MS2 (GenBank EU359909) and Maribacter sp.MS6 (GenBank EU359911), were propagated in a 50/50 mixture of marine broth and UCM.Gametes were purified to axenic conditions and inoculated with these two strains.To avoid sporulation, Ulva older than two weeks were only supplemented with 50% new UCM in all experiments.Therefore, the medium for experiments with Ulva consisted of 50% each of fresh and spent UCM.This procedure ensures the suppression of gametogenesis in the presence of the sporulation inhibitors (Stratmann et al. 1996;Kessler et al. 2018).

Toxicity and inhibition assay for gametes
Freshly discharged gametes were purified from bacteria using their phototactic behavior and diluted in UCM to 200 gametes mL −1 (Califano and Wichard 2018).The axenic gamete suspension was inoculated with Roseovarius sp.MS 2 and Maribacter sp.MS6 cultures (final OD 620 = 0.0001).In 96-well plates, a two-fold geometric dilution series of the MPs was prepared in triplicate at 100 µL per well.The gamete-bacterial suspension (100 µL, i.e., 20 gametes) was added to each well to reach the targeted concentration (Table S1).Additionally, controls without MP were prepared for the axenic gametes and inoculated with both bacterial strains.The plates were incubated in the dark for two days before switching to the typical diurnal light cycle.
After 14 days, the effect of the MPs on survivability and longitudinal growth after germination was evaluated using light microscopy.The no observed effect concentration (NOEC), lowest observed effect concentration (LOEC), and lethal concentration (LC) were determined by length comparison with the controls using a two-tailed t-test (see chapter data processing).Owing to the use of a two-fold geometric dilution series, the LOEC was typically twice as high as the NOEC.An observed LC refers to the death of over 98% of gametes.The average length of the individuals in the biological replicates was plotted against concentration.Dose-response curves were then calculated using the logistic function (Eq. 1) in Orig-inPro software (v.2022, OriginLab, USA) using the Levenberg-Marquardt algorithm (with A 1 = bottom asymptote;A 2 = top asymptote; p = hill slope;B = inflection point).The EC values were then calculated (with x = causes x% change in response), as follows (Eq.2) (Table S2): (1) Fig. 1 Micropollutants and experimental setup.(a) Selected micropollutants in the study, (b) parthenogenetic lifecycle of Ulva mutabilis (morphotype "slender") (Wichard 2015), and (c) experimental setup for the removal test.The effect of the micropollutants on the survivability and germination of gametes as well as on the chlorophyll-a fluorescence of growing gametophytes within 14 days was tested.
The removal of the micropollutants by the tripartite community Ulva mutabilis sp.-Roseovarius sp.-Maribacter sp. and abiotic dark and light controls was tested over 14 days.Experiments were carried out in three independent biological replicates with 20 gametes or four to seven growing specimens in each case NOEC and LOEC allow for comparison with the available literature, which is especially useful when data are unsuitable for dose-response fitting.Both LOECs/NOECs and EC x are valuable complementary values when describing the toxicity of MPs because they are derived in different ways (Green et al. 2013).However, as has been discussed in several studies, EC x values contain quantitative information and are generally more robust than NOEC and LOEC (Landis and Chapman 2011;Green et al. 2013;Tanaka et al. 2018).

Toxicity assessment based on Ulva vegetative growth
Five-week-old Ulva individuals (0.5-1.0 cm long) were used for toxicity experiments and incubated in 48-well plates with different concentrations of MPs in UCM for 14 days.As we were primarily interested in changes in chlorophyll fluorescence in response to the application of the MPs, the normalized reflectance/fluorescence was averaged across the waveband of chlorophyll fluorescence (670-760 nm) to calculate the normalized average reflectance in the fluorescence spectrum (nARFS) upon excitation of actinic light.
During the incubation experiments, the MP-induced change in the chlorophyll fluorescence intensity (F 665 nm) was used as a proxy for vitality.The effect of the MPs on Ulva was measured in biological triplicates with four to seven individuals each under the standardized conditions.EC x values were calculated based on the decrease in chlorophyll-a fluorescence, while NOEC was determined by comparison with controls under the light microscope.The range of the increasing effect of the MP was first evaluated using a twofold geometric dilution series, followed by a time-independent arithmetic dilution series measured within the area of interest.
The multimode microplate reader Varioskan Flash (Thermo Scientific, USA) was used for measurements.Fluorescence was measured in a grid of 49 points per well at 430 and 665 nm excitation and emission wavelengths, respectively, with a 12-nm bandwidth.Each well was measured three times, and the average of the top 10 values of each measurement was used for plotting and calculating the ECs (Eqs. 1 and 2) (Table S3).The atrazine concentration was log transformed for the calculations.

Determination of micropollutant removal
For the removal experiments, the effect of the MPs ATZ, BPA, CAP, ERY, PMG, MTC, and OTC on Ulva growth (2) EC x = B + log 10 100 100−x − 1 p were tested at their EC 10 concentrations in 12-well plates (Table 1).E 2 , EE 2 , and SMX were not toxic at their saturated concentrations and were tested slightly below the saturated concentrations in the UCM (Table 1).Nine wells per MP were filled with 2.2 mL of the respective solution in the UCM.Next, 200 µL were immediately sampled as the starting concentration (d0), spiked with IS, and stored at -20 °C until measurement.The remaining 2 mL were incubated in the dark ('abiotic dark', n = 3) with a standard light cycle ('abiotic light', n = 3) or with young Ulva in the tripartite community in a standard light cycle ('biotic', n = 3) (Fig. 1c).Approximately 3-5 individuals of 6-weekold (1-2 cm long) Ulva were added to each biotic well.After 14 days, 200 µL was sampled (d14), spiked with IS, and stored at -20 °C until measurement without further sample preparation upon appropriate dilution with water.For dry weight measurement, the algae were washed three times with ultra-pure water, frozen in liquid nitrogen, and freezedried until there was no further weight loss.
For all compounds that displayed no or only minor removal after 14 days (ATZ, ERY, PMG, MTC, and SMX), removal was tested over 42 days in a similar experiment.This was done to provide more time for the tripartite community to adapt to the introduced MPs.As the five MPs tested did not show any light sensitivity in the 14-day experiment, a dark control was excluded.

UHPLC-ESI-HRMS method
The MPs were separated over 10 min with a 0.3 mL min −1 flow rate and 7 min ramp from 100% A (aqueous) to 100% B (organic) in an ACQUITY UPLC BEH C 18 column (130 Å, 1.7 µm, 100 × 2.1 mm, Waters, Ireland) with column guard.The pH of both mobile phases was between 9.4 and 9.6, with A containing 90% water, 10% methanol, and 2 mmol L −1 (NH 4 ) 2 CO 3 and B containing 100% methanol and 2 mmol L −1 (NH 4 ) 2 CO 3 .The injection volume was 2 µL.Electrospray ionization (ESI) was employed with 3 kV voltage, 360 °C capillary temperature, and 400 °C auxiliary gas temperature.High-resolution mass spectrometry (HRMS) was performed in alternating negative-positive mode with a mass range between 100 and 1200 m/z and a resolution of 70,000 at 200 m/z.

Quantification of micropollutants by mass spectrometry
Isotopically labeled internal standard concentrations and appropriate analyte concentrations were used for calibration.For analysis, all samples were spiked with IS at half the molar starting concentration, diluted with water (by a dilution factor between 100 and 10,000 depending on the MPs) to reach the working range of the calibration and filtered with 0.22-µm PVDF filters.Protonated and deprotonated ions of the MPs were used for quantification with a 5 ppm mass range (Table 2).As no isotopologue was available for oxytetracycline, an external calibration was performed.
All calibration parameters are listed in Table S4.Critical signal, limit of detection (LOD), and limit of quantification (LOQ) refer to the calibration used and do not represent the full potential of the analysis method and the selected mass spectrometer.Linearity was tested using the Mandel and lack-of-fit tests (Reichenbächer and Einax 2011) using OriginPro (v.2022) software.

Data processing
Concentrations were measured before and after incubation to ascertain whether each of the three treatments eliminated MPs from the culture medium.All MPs with a significant decrease in darkness were assumed to have been eliminated through hydrolysis or sorption into the plastic well.Removal by photolysis was assumed for all MPs, with a significantly higher removal rate in light than in the dark.Elimination by sorption, uptake, or biotransformation was assumed for all MPs, with a significantly higher removal rate with the tripartite community than in light controls.For areas below the critical signal, the concentration was set as the detection limit (corrected by the dilution factor) for statistical analysis (Student´s t-test), calculations, and visualization.The relative decrease for each MP was determined and separated into the various contributing effects using the relative removal under four different conditions: (i) the removal by Ulva and its associated bacteria was calculated by subtracting the removal in the treatment of 'light abiotic controls' from the 'biotic treatment'.(ii) Removal by photolysis was calculated by subtracting the removal in the treatment of the 'dark abiotic controls' from the 'light abiotic controls'.(iii) Removal by sorption and hydrolysis was calculated directly from the treatment 'dark abiotic controls'.(iv) Removal below the LOD was regarded as possibly higher but not lower.A onetailed two-sample unpaired Student's t-test or Welch's test was used for all statistical tests on removal, depending on the homogeneity of variances.A p-value > 0.05 was deemed insignificant (NS).

Results
The toxic and inhibitory effects of ten different MPs were investigated in U. mutabilis gametes and juvenile gametophytes (1-2 cm), representing two crucial developmental phases in the parthenogenetic lifecycle of Ulva (Fig. 1b).

Survival and inhibition of the germination of gametes
Overall, there was no trend in toxicity regarding the three tested substance groups (herbicides, antibiotics, and hormonal disruptors, Table 3).The curves for eight MPs followed a clear dose-response relationship (Fig. 2 and Table S2), whereas CAP and EE 2 were not fitted owing to insufficient and variable data.All MPs inhibited the germinating and growth of the gametes of U. mutabilis; however, ERY significantly promoted longitudinal growth (up to 56% at 0.095 mg L −1 ) compared to the control (Fig. 2c).
The herbicide ATZ was most toxic for Ulva gametes, with an EC 50 value for longitudinal growth inhibition of 0.05 mg L −1 , followed by BPA with an EC 50 value of 1.28 mg L −1 (Table 3).The EC 50 values of ERY, EE 2 , CAP, and E 2 were approximately between 3-4 mg L −1 , whereas MTC, OTC, SMX, and PMG were less active, with EC 50 values of 6.11, 17.5, 50.7, and 108 mg L −1 , respectively.Similarly, ATZ and BPA were the most toxic, with LCs of 0.47 and 4.97 mg L −1 , respectively.The toxicity of the antibiotic CAP was comparable with that of the endocrine disruptor BPA, with an LC of 6.36 mg L −1 , whereas the other MPs were less toxic, with LCs ranging from 25-201 mg L −1 .PMG was non-toxic at the highest concentration tested (100.8 mg L −1 ).
EC 50 and slope can be used to determine the full range of activity for MPs and are relevant in bioremediation and aquaculture maintenance.The slopes of the dose-response curves of ATZ and CAP represent their potency at toxically relevant inhibition levels; the slopes are steep and exhibit the lowest relative distance between the no effect and maximum effect points.In contrast, ERY had the highest relative range of increasing effect over three orders of magnitude.

Toxicity and survival of growing Ulva (gametophyte)
The toxic effects of all MPs were recorded for Ulva in the tripartite community.After 14 days of incubation with varying concentrations of MPs, the chlorophyll-a activity of biological triplicates was evaluated.Eight of the ten MPs exhibited dose-dependent response relationships and were fitted (Fig. 3 and Table S3), whereas SMX and E 2 had no impact, even at saturated concentrations.NOECs were determined by visual comparison, and EC values were determined using dose-response curves (Table 4).Some concentrations had no statistical effect on fluorescence values, while damage to the algae was visible under the microscope.In these cases, the NOEC represents a more conservative estimate of toxicity compared to the EC 10 .
ATZ had a dynamic range of over two orders of magnitude and was measured in a geometrical dilution series (Fig. 3a).BPA exhibited a high distribution in the biological triplicates (Fig. 3b).Three wells in the active CAP range demonstrated high fluorescence (Fig. 3c).Presence of ERY decreased the fluorescence of Ulva by only 56% (Fig. 3d), whereas the other compounds bleached the algae.EE 2 (Fig. 3e), PMG (Fig. 3f), MTC (Fig. 3g), and OTC (Fig. 3h) indicated standard dose-response relationships.
ATZ had the highest activity with a NOEC (4.72 mg L −1 ) followed by BPA (6.2 mg L −1 ) (Table 4).The NOECs of all other drugs ranged between 50.1 and 402 mg L −1 , with SMX having the lowest activity.Similarly, ATZ and BPA were the most hazardous, with EC 50 values of 7.19 and 18.2 mg L −1 , respectively.E 2 and SMX were not toxic within the investigated concentration range, whereas the other MPs had EC 50 values ranging from 93.0 to 339.9 mg L −1 .

Quantification of micropollutants in the algal culture medium
A robust quantification methodology was required to determine the removal of MPs from the algal culture medium.For all analytes, symmetrical-shaped chromatographical peaks with minimal tailing were obtained by detecting the respective quasimolecular ions of the MPs (Fig. 4).To combine the analyses on the same reversed-phase chromatographic platform, PMG was derivatized with Fmoc reagent.Using isotopically labeled internal standards, a base calibration was performed for all compounds except OTC, where an external calibration was utilized (Fig. S4).All calibrations passed the Mandel and lack-of-fit tests for linearity.The relative standard deviations of the calibrations of ATZ, BPA, CAP, E 2 , EE 2 , PMG, MTC, and SMX were below 10%, whereas the remaining two were over 10% (Table 5).Owing to the lack of an internal standard, the calibration of the OTC exhibited large variability throughout the calibration.

Removal of micropollutants by the Ulva tripartite model system
High concentrations of MPs were preferentially evaluated to determine the suitability of U. mutabilis for bioremediation.Thus, the EC 10 limits of growing Ulva were used to identify the highest amounts possible while minimizing cell damage and toxicity.Micropollutant levels were measured before and after 14 days of incubation with a tripartite community of Ulva mutabilis-Roseovarius-Maribacter and compared with that of abiotic control treatments incubated in the dark and light (Fig. 5).Depending on the biotic and abiotic conditions, four cases were observed: (i) no changes at all, (ii) biotic-dependent changes, (iii) light-dependent changes, and (iv) treatment-independent changes.The concentrations of ATZ and SMX did not change significantly under any conditions (Fig. 5a, b).Owing to the high variability in ERY measurements, no significant difference was found (Fig. 5c).The situation was different for CAP, where the concentration decreased significantly by approximately 70% (p = 0.001) in both the biological and the light controls but not in the dark (Fig. 5d).Similarly, the OTC concentration decreased below LOD in the presence of light (in both the biological and light controls), but also decreased by 23.7% in the dark (Fig. 5e).
Notably, the BPA concentration decreased significantly (p = 0.003) under biological conditions below the LOD but did not change under abiotic conditions (Fig. 5f).Over 98% of BPA was removed in the presence of Ulva and the associated bacteria.Similarly, the hormones E 2 and EE 2 were significantly reduced (98.2% and 94.3%, respectively; p < 0.001) to below the LOD in the presence of the Ulva-Roseovarius-Maribacter model system, were not significantly reduced in the dark, and decreased only slightly in the light by 24.3% (p = 0.019) and 29.9% (p = 0.014) (Fig. 5g,  h).MTC and PMG herbicides exhibited low but significant removal rates under biological and dark conditions, respectively (Fig. 5i, j).
The concentrations of the more stable compounds ATZ, SMX, ERY, MTC, and PMG were monitored for 42 days under biological and light treatments only.The concentration of PMG was reduced by 30.4% under biotic conditions (p = 0.005) but, in contrast, increased slightly in the light control (Fig. 5k).The concentrations of ATZ, ERY, MTC, and SMX were stable for this extended period under both biotic and abiotic light conditions (data not shown).
The recovery rates for all chemicals were between 90-120%, except for PMG (78.5%),BPA (138.0%), and E 2 (148.1%)(Table 5).The relative decrease for each MP within 14 days was determined and separated into the various contributing modes of action for the removal (Fig. 6a).
Fig. 2 Inhibitory and toxic effects of different micropollutants on germinating gametes of Ulva mutabilis using three biological independent replicates with 20 gametes each.After 14 days of germination, the average length was fitted in a dose-response curve.Atrazine (a), bisphenol A (b), erythromycin (c), estradiol (d), glyphosate (e), metolachlor (f), oxytetracycline (g), and sulfamethoxazole (h) showed a dose-response relationship.Representative microscopic images of different inhibitory concentrations of erythromycin (i).Scale bar: 500 µm.The length after 14 days of uninhibited growth varied between the groups due to variability between different batches of Ulva but always matched the respective group control.The red triangles indicate the average length of the negative controls (n = 3) PMG and OTC were significantly removed by hydrolysis or sorption; photolysis removed significant amounts of CAP, E 2 , EE 2 , and OTC.Importantly, BPA, E 2 , EE 2 , and MTC were all significantly removed via biosorption, uptake, or biotransformation accounting for the strongest effect on these MPs.Indeed, the tripartite community entirely removed BPA, E 2 , and EE 2 (< LOD).The dry weight removal capacity of Ulva (0.8-12.8 mg g −1 ) suggests that it could be used as a biofilter in wastewater treatment (Fig. 6b).

Discussion
In general, the harmful effects of MPs on Ulva were minimal compared with those concentration observed in surface water, groundwater, and wastewater effluents (Das et al. 2017;Sousa et al. 2018;Tran et al. 2018).Environmental concentrations are typically many orders of magnitude below the levels that influenced Ulva in this study.Accordingly, as a holobiont, Ulva can be grown even in harsh conditions and, in some cases, can contribute to the bioremediation of

Lifecycle dependent toxicity: differences between germinating gametes and growing algae
We discovered that the tested MPs had a lifecycle stagedependent sensitivity, which has also been observed in flowering plants (Boutin et al. 2014).Gametes were more sensitive to MPs than mature Ulva specimens: Based on the EC 50 value, the gametes were 2 to 138 times more sensitive compared to growing thalli of U. mutabilis for the tested various MPs.
Although target enzymes for all three herbicides are present in the U. mutabilis genome (De Clerck et al. 2018), only ATZ demonstrated enhanced activity relative to the other MPs, whereas MTC and PMG did not.Notably, PMG had the lowest activity among all tested MPs in gametes.BPA was the second most active substance against germination.

Erythromycin displayed algistatic effects
ERY displayed the widest range of increasing effects on gametes, and longitudinal growth was inhibited over a large concentration range.The LC/LOEC ratio of ERY was 32, while the ratios of the other MPs were ≤ 8. Therefore, the slope of the ERY dose-response curve was shallowest.For the growing Ulva, ERY decreased fluorescence by only 56%, whereas the other compounds killed and even bleached the algae.This effect could be explained by the inhibition of protein synthesis in chloroplasts and mitochondria by ERY.Machado and Soares (2019a) observed an algistatic effect of ERY for the green microalga Raphidocelis subcapitata at concentrations ranging from 38-200 µg L −1 .They observed a decrease in chlorophyll-a content and mitochondrial function, which decreased growth without harming the cell membrane.This was attributed to the similarity of the chloroplast and mitochondria ribosomes to prokaryotic ribosomes, thereby inhibiting protein synthesis in these cell organelles.Ulva appears to be more resistant than Raphidocelis, with inhibitory effects in gametes at 3-49 mg L −1 (Fig. 2c, i), whereas the EC 50 value for decreasing chlorophyll-a fluorescence in growing Ulva was 340 mg L −1 .

Chloramphenicol treatment indicated decoupling of photosystem
Higher fluorescence was observed in the CAP treatment of growing Ulva with multiple repetitions at certain concentrations.CAP can serve as an electron acceptor for photosystems I and II and transfers electrons to oxygen, decoupling the photosystem and generating superoxide (Rehman et al. 2016).Therefore, CAP reduces photosynthetic efficiency by competing for electrons and damaging the cell.This effect reached a maximum in rice leaves (Oryza sativa) at 323 mg L −1 (Okada et al. 1991) and probably caused higher electron transfer rates in the remaining U. mutabilis at a CAP concentration between 254-354 mg L −1 (Fig. 3c), resulting in higher chlorophyll-a fluorescence by decoupling the photosystems in this study.

Comparison of the resilience of U. mutabilis with other Ulva species, microalgae and macrophytes
The existing literature suggested that Ulva and its associated bacteria can withstand high concentrations of MPs under harsh conditions.In the following sections, the individual MPs are discussed in the context of Ulva.Overall, U. mutabilis was highly resistant to antibiotics, whereas its gametes were more susceptible to CAP and ERY (Table 6).Endocrine disruptors displayed opposing effects on U. mutabilis; the algae were highly resilient against EE 2 , whereas the gametes were susceptible to BPA and E 2 (Table 7).Ulva gametes displayed similar susceptibility to herbicides as green microalgae, whereas growing U. mutabilis was considerably more resilient than green microalgae, macrophytes, and most other Ulva species (Table 8).
Chloramphenicol (CAP): U. mutabilis was considerably more resilient to CAP in our study (about 20 times higher EC 50 ) than has been reported during gametogenesis (Hoxmark 1975), which was most likely linked to cell damage and additional stress owing to thallus fragmentation.Similar to our study, the growth of U. lactuca was found not to be negatively affected by CAP at concentrations up to 50 mg L −1 (Leston et al. 2013).The toxic effects observed after CAP addition are caused by the photodegradation product p-nitrobenzaldehyde, which can be detoxified by algaepresumably by oxidation-to p-nitrobenzoic acid (Hoxmark and Nordby 1977).As Ulva gametes were tested at lower cell densities, they could not detoxify higher concentrations of p-nitrobenzaldehyde and were more susceptible than the growing Ulva.Furthermore, microalgae display similar susceptibility as gametes (Xiong et al. 2019b;Nguyen et al. 2022) (Table 6).Erythromycin (ERY): Growing Ulva mutabilis was extremely resilient to ERY, especially when compared with U. pertusa, even though Ulva gametes were only marginally more resistant than most green microalgae.Specifically, ERY shows a wide range of effects on different species of green microalgae (Liu et al. 2011;Machado and Soares 2019a;Ma et al. 2021;Zhang et al. 2021;Li et al. 2022) and was highly effective against U. pertusa (Chen et al. 2017).
17β-estradiol (E 2 ): Similar to previous studies on Ulva spp., U. mutabilis was not affected by E 2 , even at saturated concentrations (Yang and Hong 2012;Lu et al. 2021).No precipitation or crystallization was observed at the LCs of E 2 and EE 2 despite the tested concentrations being close to reported solubility in distilled water at 25 °C (Shareef et al. 2006).Noteworthy, Ulva gametes were more susceptible (about 5 times lower EC 50 ) than growing Ulva and the green microalgae Chlorella spp.(Huang et al. 2019) (Table 7).
17α-ethinylestradiol (EE 2 ): Gametes and growing Ulva were considerably more resilient to EE 2 than green microalgae (Balina et al. 2015;Belhaj et al. 2017;Zhang et al. 2022).The only report on the effect of EE 2 on green macroalgae identified metabolome changes in U. lactuca at concentrations as low as 1 mg L −1 (Cabral et al. 2018), which is similar to the observed effects in U. mutabilis at higher concentrations.

Atrazine (ATZ):
The inhibitory effect of ATX on Ulva gametes was comparable to the EC 50 values reported for green microalgae (Qian et al. 2008;Sun et al. 2012;Kabra et al. 2014;Camuel et al. 2017), U. pertusa, and macrophytes (Fairchild et al. 1998;Lee et al. 2019); however, the growing U. mutabilis was considerably more resilient, but the EC 50 value reached the lowest measured value (0.052 mg L −1 ) compared to the other MPs in our study (Table 8).
Glyphosate (N-(phosphonomethyl) glycine, PMG): Gametes and growing U. mutabilis displayed similar resistance to PMG as Chlorella spp.and U. intestinalis (Kittle and McDermid 2016;Kaeoboon et al. 2021).PMG inhibited photosynthesis in U. lactuca at very low concentrations; however, tolerance mechanisms were observed at higher concentrations (de Carvalho et al. 2022), which could explain the high resilience of gametes and growing U. mutabilis.In a study of four different microalgae, low EC 50 values for a PMG formulation were observed after 96 h (Hernandez-Garcia and Martinez-Jeronimo 2020).As one of the microalgae in this previous study was C. vulgaris, the formulation of PMG likely resulted in a substantial change in effectivity (Table 8).
Metolachlor (MTC): Growing U. mutabilis were highly resilient to MTC compared to green microalgae and macrophytes, whereas gametes were more resilient than all but one species (Fairchild et al. 1998;Maronic et al. 2018;Machado and Soares 2019b).This is the first study to investigate MTC toxicity in macroalgae.

Removal of micropollutants
Overall, our analytical approach was well adapted for MP quantification, with an acceptable peak shape, a low relative standard deviation of the calibration, and adequate recovery rates.Furthermore, when calibrated suitably, the developed method can also detect lower, environmentally relevant concentrations.Consideration should be given to enhance the ionization of the phenolic endocrine disruptors when improving these methods.This can be accomplished, for example, through post-column infusion or the use of atmospheric pressure chemical ionization.The observed increase in MP concentrations after 14 days in some cases is probably attributed to water evaporation, which increases the concentration at stable MP concentrations.In summary, ATZ and SMX were stable under all conditions and not removed by the tripartite Ulva-Roseovarius-Maribacter community; PMG and OTC were slightly removed by hydrolysis or sorption; CAP, E 2 , EE 2 , and OTC were partially removed by photolysis; and BPA, E 2 , and EE 2 were removed to below LOD, presumably by degradation.The tripartite community partially and reversibly removed ERY and MTC while partially removing PMG after a lag phase.
CAP and OTC undergo photodegradation (Shih 1971;Jin et al. 2017).In contrast, E 2 and EE 2 are stable under UV-A and visible light conditions but are slowly degraded in the presence of UV-B and UV-C radiation (Liu and Liu 2004;Rosenfeldt and Linden 2004;Mazellier et al. 2008;Leech et al. 2009;Matamoros et al. 2009).Owing to the medium composition and use  of 50% spent medium, organic compounds might act as photocatalysts and enhance photodegradation (Leech et al. 2009).
Although approximately 12% of ERY and MTC were removed in the presence of Ulva within 14 days, no effective removal was observed after 42 days, indicating a reversible sorption process.The tripartite community likely adapted to PMG, as its removal under biotic conditions occurred only after the longer incubation period of 42 days and not in the first two weeks.
As the tripartite community removed BPA, E 2 , and EE 2 to below the LOD, biotransformation or total metabolization is the most likely mechanism.Removal of these three endocrine disruptors by U. mutabilis have been previously examined alongside other Ulva species.For example, U. prolifera removed approximately 95% of 1 mg L −1 BPA within six days (Zhang et al. 2019), and U. pertusa 90% of 0.01 mg L −1 EE 2 and 60% of 0.01 mg L −1 E 2 within the first day of incubation (Lu et al. 2021).Notably, other species of Ulva were not found contributing to the removal of BPA, EE 2 , or E 2 (Astrahan et al. 2021).In our study, U. mutabilis displayed better or comparable removal capabilities to other Ulva species for BPA and higher removal efficiencies for E 2 and EE 2 .

Implications for the use of Ulva in aquaculture
To use U. mutabilis in IMTA, the compounds that accumulate and might be passed on via consumption of the algae must be identified when applying Ulva as a biofilter and food and feed source at the same time.Further examination of the underlying removal mechanisms of the phenolic endocrine disruptors BPA, E2, and EE2, as well as the herbicide PMG will facilitate a clearer risk assessment.However, as no biotic processes are involved in the removal of the other studied MPs, Ulva will not likely become contaminated by these under the studied exposure conditions, thus highlighting its potential for use in aquaculture.

Implications for the modes of action of algal growth factors
While U. mutabilis was resistant to the tested herbicides and antibiotics, as evidenced by high EC 50 values, all tested hormone disruptors were removed during our experiments.However, the extent to which these hormones are absorbed or biologically transformed requires further study.This is particularly important for hormone-like compounds, such as thallusin (Alsufyani et al. 2020;Dhiman et al. 2022), which are released by associated bacteria and are essential for the development of Ulva.Compounds with a potentially similar spectrum of action may likely be eliminated from Ulva to facilitate the selective action of thallusin.

Conclusions
The green macroalga U. mutabilis is resilient against many MPs at high concentrations including antibiotics, endocrine disruptors, and herbicides.Depending on the MP, Gametes were up to two order of magnitude more vulnerable than a growing tubular thalli of U. mutabilis but still more resilient than green microalgae.The bioaccumulation of the endocrine disruptors remains a critical factor for assessing Ulva in IMTA, which requires further investigation.Our developed analytical method is suitable for the direct measurement and quantification of various compounds in seawater; however, the selected detection limits could be lowered further using the analytical techniques employed here.Ulva mutabilis and its associated bacteria did not remove the tested antibiotics and herbicides from seawater.Nonetheless, the fast growth rates and broad geographical distribution of Ulva make it an ideal target for the bioremediation of endocrine disruptors, BPA, E 2 , and EE 2 , and this genus can be considered an effective biofilter for wastewater management.

Fig. 3
Fig. 3 Toxicity of different micropollutants for growing Ulva mutabilis.Chlorophyll-a fluorescence was fitted in a dose-response curve using three independent biological replicates with four to seven specimens each.In alphabetical order: (a) atrazine, (b) bisphenol A, (c) chloramphenicol, (d) erythromycin, (e) ethinylestradiol, (f) glyphosate, (g) metolachlor, (h) and oxytetracycline decreased chlorophyll-a fluorescence, while estradiol and sulfamethoxazole showed no effect, even in saturated concentrations.The red triangles indicate the average fluorescence of the negative controls (n = 3)

Fig. 5
Fig. 5 Concentrations of micropollutants in the growth medium of Ulva mutabilis before and after 14 or 42 days of cultivation.The EC 10 value of each individual micropollutant for Ulva was first adjusted in the growth medium.The tripartite community Ulva-Roseovarius-Maribacter (biotic) treatment was compared to controls with (light) and without light (dark).The values of the detection limit (corrected

Fig. 6
Fig. 6 Removal of micropollutants after 14 days.(a) Total removal was divided into effects in the dark (hydrolysis and sorption), with light (photolysis), and removal by biota.Error bars represent the mean ± standard deviation (n = 3).For removal below the limit detection, the error bars show only the positive range to 100% because removal could be higher but not lower.(b) Biological removal below LOD of endocrine disruptors normalized to Ulva dry weight

Table 1
Highest concentrations tested for micropollutant-removal experiments with the tripartite community Ulva mutabilis-Roseovarius sp.-Maribacter sp

Table 3
Micropollutants inhibit the germination and growth of Ulva mutabilis gametes.The no effect concentration (NOEC), the lowest observed effect concentration (LOEC), the half effective concentration (EC 50 ± standard deviation) for longitudinal growth, and the lethal concentration (LC) of in mg L −1 were determined (n = 3)

Table 4
Light microscopy and chlorophyll a fluorescence were used to determine the toxicity of micropollutants to growing Ulva mutabilis.Concentrations with no effect (NOEC) were observed by light microscopy.Effective concentrations (EC) were determined by the decreasing chlorophyll a fluorescence (EC standard deviation) (n = 3) (Qiu et al. 2017 ion chromatograms (EIC) of micropollutant quasimolecular ions measured by electrospray mass spectrometry in positive ( +) or negative (-) ionization modes coupled to C 18 reversedphase liquid chromatography 1 3wastewater(Qiu et al. 2017).In only one case, an environmental concentration of ATZ (40.2 µg L −1 , Baldwin et al. 2016) was reported that was higher than the NOEC and EC 10 values but lower than EC 50 determined for gametes of U. mutabilis in our study.

Table 5
Selected calibration parameters for the quantification of the micropollutants Note: * LODs are 1-3 order of magnitude below the adjusted concentrations of MPs in growth media.** Erythromycin and oxytetracycline have unusually high relative standard deviations.Further enhancements to the analytical method may be required

Table 6
Toxicity of antibiotics studied in comparison to the literature (Ulva spp.and microalgae)

Table 7
Toxicity of endocrine disruptors studied in comparison to the literature (Ulva spp.and microalgae)

Table 8
Toxicity of herbicides studied in comparison to the literature (Ulva spp., macrophytes, and microalgae)