Comparative Assessment of Pharmaceutical Removal from Wastewater by the Microalgae Chlorella sorokiniana , Chlorella vulgaris and Scenedesmus obliquus Comparative Assessment of Pharmaceutical Removal from Wastewater by the Microalgae Chlorella sorokiniana , Chlorella vulgaris and Scenedesmus obliquus

In view of risks associated with the discharge of pharmaceuticals in the aquatic envi‐ ronment, the objective of this work was to assess the removal of paracetamol, salicylic acid and diclofenac from water by a microalgae‐based treatment. For a comparison purpose, the growth and kinetic parameters for the removal of drugs were determined for three different microalgae strains, namely Chlorella sorokiniana , Chlorella vulgaris and Scenedesmus obliquus . It was found that the drugs removal efficiency by these strains was related to their growth. Comparing the three pharmaceuticals, the salicylic acid was the most efficiently removed, especially by S. obliquus (>93% batch culture, >99% semicontinuous culture) and C. sorokiniana (>73% batch culture, >93% semicontinuous culture). Contrarily, paracetamol was the most poorly removed, the maximum efficien ‐ cies being those attained by C. sorokiniana (>67% batch culture, >41% semicontinuous culture). On the other hand, diclofenac was efficiently removed only by S. obliquus (>98% batch culture, >79% semicontinuous culture). For the three considered drugs, C. vulgaris was the strain showing the lowest removal capacity. The large differences here revealed between microalgae strains regarding their removal capacity of pharma‐ ceuticals, pointed to the strain selection as a key issue for a successful application in wastewater treatment.


Introduction
Emerging contaminants (ECs) include a wide range of compounds and may be defined as naturally occurring, manufactured or man-made chemicals or materials that have been found or are suspected to be present in various environmental compartments and whose toxicity or persistence are likely to significantly alter the metabolism of a living being [1]. Among them, pharmaceuticals have received considerable attention with respect to their environmental fate and toxicological properties over the last decade [2]. Pharmaceuticals represent an especially worrying class since they were designed to cause a physiological response and their presence in the environment may affect non-target individuals and species [3]. This concern on pharmaceuticals presence in the aquatic environment has led to the recent consideration by European regulations within the Water Framework Directive (2000/60/EC) (WFD). The Commission proposal of 31 January 2012 foresaw the inclusion of three pharmaceuticals, namely diclofenac, 17-beta-estradiol (E2) and 17-alpha-ethinylestradiol (EE2) in the list of priority substances. Instead, by the EU Decision 2015/495, these compounds together with another estrogen (E1) and three antibiotics (azithromycin, clarithromycin and erythromycin) were finally included in the first watch list of substances to be monitored in all member states to support future reviews of the priority substances list [4].
Pharmaceuticals in domestic sewage or from hospital or industrial discharges end in municipal sewage treatment plants (STPs), but conventional wastewater treatments have been reported to be ineffective in the removal of such pollutants, with efficiency values of <5 to 40% [5]. In fact, STPs were not originally designed for the removal of pharmaceuticals due to the non-existence of limiting regulations on their discharge [6,7]. Consequently, STPs are important sources of such pollutants in the aquatic environment [8,9]. In this regard, Verlicchi et al. [10], who reviewed the occurrence of 118 pharmaceuticals in the influent and effluent of 244 STPs, found that the occurrence of some of them in the effluent discharged into surface water bodies may pose a mediumhigh (acute) risk to aquatic life. Among the studied pharmaceuticals, diclofenac was shown to have the highest average mass load (240 mg/1000 inhabitant) in the effluents of municipal STPs [10]. The removal efficiencies of diclofenac in conventional STPs have been reported to be about 17% [11], which translates into relative high concentrations in the corresponding effluents [12].
In the recent years, phytoremediation of waters by using photoautotrophic aquatic organisms such as algae has gained attention for the removal of both organic and inorganic pollutants [13][14][15]. Microalgae are characterized by high photosynthetic efficiency, high growth rates, wide adaptability and high potential to remove inorganic nutrients from the wastewater. The principal mechanism of algal nutrient removal is their uptake into the cell biomass [16]. The main advantages of using microalgae for nutrients removal during the tertiary treatment of wastewaters are the possibility of recycling the assimilated nitrogen and phosphorus into algal biomass as a fertilizer, as a source of products (e.g. paraffin, olefin, glycerol, protein, anti-oxidant, pigment, plastic, etc.), or as biofuel, and also the generation of an oxygenated high-quality effluent [17]. However, although the capability of microalgae wastewater treatments systems to remove organic matter and nutrients has been deeply studied, little is known about the removal of ECs, such as pharmaceuticals, by algae. In fact, it has already been claimed the necessity of further studies on the removal of this sort of pollutants by algal systems [18].
In this context, the aim of this study was to determine and compare the potential of green microalgae Chlorella sorokiniana, Chlorella vulgaris and Scenedesmus obliquus to remove paracetamol, salicylic acid and diclofenac from water. The strains used in this work were selected since they are known to have fast growth rates and potential for wastewater treatment due to their tolerance to the severe environmental conditions found in municipal wastewater and some industrial wastewaters [19].

Microorganisms and culture conditions
The microalgae strains used in this study were C. sorokiniana CCAP 211/8 K from UTEX Culture Collection of Algae, C. vulgaris SAG 221-12 from SAG Culture Collection of Algae and S. obliquus SAG 276-1 from SAG Culture Collection of Algae. These microalgae strains are among the most commonly used for wastewater treatment have high growth rates and are able to grow under a wide range of conditions [19], which motivated their choice for this study.
Bubbling column photobioreactors (PBRs) with spherical bases (40 mm diameter and 300 mm height with 300 ml capacity) were used for the experimental setup, keeping an operating volume of 250 ml. In each PBR, the Mann and Myers culture medium was inoculated with the required volume of the corresponding pre-cultured microalgae in order to have an initial concentration of about 3 × 10 6 cells ml −1 .
During the experimental phase, the culture was aerated with filtered air (0.22-µm sterile airventing filter, MillexFG50-Millipore), at a rate of 0.3 v/v/min, enriched with CO 2 at 7% v/v, which was injected on demand to keep a constant pH (pH = 7.5 ± 0.5), as controlled by a pH sensor. The irradiance supplied during this phase was 370 µE m −2 s −1 , which was provided by eight fluorescent lamps (58 W, 2150 lumen, Philips, France). The photoperiod was maintained in 12:12 h light/dark and the temperature in 25 ± 1°C.

Experimental setup
PBRs were operated in batch mode until the end of the exponential growth phase and then under semicontinuous mode till the growth parameters remained constant at the steady state.
During the batch culture, an aliquot of 5 ml was daily taken from each PBR for the analytical determinations, this volume being replaced with distilled water to keep the operation volume. During the semicontinuous culture, 30% of the culture volume was daily harvested and used for analysis, this volume being replaced with fresh medium. Throughout the experiments, the growth of the culture was daily monitored by the determination of biomass concentration and cell density. The removal of pharmaceuticals was daily determined by the analysis of the remaining concentration of this drug in the culture medium. All analyses were conducted in triplicate.

Analytical methods
Biomass concentration (Cb) was determined by optical density at 680 nm (OD 680 ) by spectrophotometric (UV/visible spectrophotometer BECKMAN DU640) and verified by dry weight. Preliminary studies were conducted to determinate the relationship between dry weight and OD 680 for each strain; as shown in Eq. (1) for C. sorokiniana, in Eq. (2) for C. vulgaris and in Eq. (3) for S. obliquus: Dry weight measurements were performed by filtering 10 ml of culture through a 0.45 µm Whatman filter, which was then washed with 20 ml HCl (0.5 M) to dissolve precipitated salts. Then, the filtrate was dried in an oven at 80°C for 24 h. Additionally, the growth of the culture was measured as cell density (Nc) by cell counting with a Neubauer chamber.
The initial and remaining pharmaceuticals concentration in the culture medium was quantified by a Waters HPLC 600 equipped with a 2487 Dual λ Absorbance Detector. A Phenomenex Gemini-NX C18 column (5 µm, 250 mm × 4.6 mm) was used for the separation. The wavelengths of detection were 246 nm for paracetamol, 236 nm for salicylic acid and 276 nm for diclofenac. The mobile phase consisted of a mixture of acetonitrile to water (30:70, v/v) for the analysis of paracetamol and a mixture of acetonitrile to water to orthophosphoric acid (70:30:0.1, v/v/v) for salicylic acid and diclofenac. HPLC quality acetonitrile (CH 3 CN) and orthophosphoric acid (H 3 PO 4 ) from Prolabo Chemicals and ultrapure water obtained by a Millipore System were used for the preparation of the mobile phase. Before use, each mixture was passed through a Millipore 0.45-µm pore-size filter and degasified in an ultrasound bath for 30 min. Before analysis, all the samples were centrifuged twice at 7500 rpm for 10 min (SIGMA 2-16P centrifuge). For the chromatographic analysis, the mobile phase flow rate was 1 ml min -1 and the injection volume was 100 µl.

Data analysis
Growth kinetics were resolved in OriginPro 8 using the classic model originally described by Verhulst [21] called logistic model, which has been proved to fit the growth of microalgae [22]. The logistic model fits to a sigmoidal curve that describes the relationship between microorganisms' growth and density in limited environmental conditions (Eq. (4)).
Where N (g l −1 ) is the algal density at time t (h), K (g l −1 ) is the carrying capacity (the maximum algal density reached in the culture), a is a constant in the logistic model that refers to the relative position from the origin and indicates the duration of the lag phase and r (d −1 ) is the specific growth rate.
Furthermore, the kinetic curves for the removal of pharmaceuticals were fitted to the logistic model. In each case, the parameter K (g l −1 ) is the maximum removal capacity by the microalgae in the culture. The parameter a is a constant in the logistic model that indicates the delay in the beginning of the target compounds removal and the parameter r (d −1 ) is the specific removal rate.
Finally, differences among the strains with respect to the kinetic parameters of growth and removal of pharmaceuticals were compared by a non-parametric test using IBM SPPS Statistics 21. The comparison of means was performed by means of the U Mann-Whitney test.
Significance was defined at p ≤ 0.05.
For the removal of pharmaceuticals, the volumetric efficiency for each target compound was calculated as the difference between its average concentration in the influent (C inf ) and in the effluent (C efflu ) at every sampling day, considering the daily dilution rate of the corresponding operation stage (D) (Eq. (5)). During the batch culture these efficiencies were cumulatively expressed as milligram per litre and as milligram per litre per day during the steady state of the semicontinuous culture: The specific efficiency of the removed pharmaceuticals was calculated as the ratio between the volumetric efficiency and the biomass concentration (Cb) (Eq. (6)). Likewise, during the batch culture these efficiencies were cumulatively expressed as milligram per gram per biomass and as milligram per gram day during the steady state of the semicontinuous culture:

Growth of the culture
The growth curves of C. sorokiniana, C. vulgaris and S. obliquus during the batch culture, of either the treatments or the controls, showed a typical sigmoidal growth of 8-10 days until reaching the steady state. On the other hand, during the semicontinuous mode, daily dilution rates produced instability and the growth rate declined throughout several days until the growth parameters remained constant during the steady state. This instability is a typical behaviour in the microalgae culture when the growth conditions change and it is related with an adaptation phase (Figures 1-3).   differences among the treatments were analysed according to growth kinetic parameters, as shown in Table 1.
The addition of paracetamol increased the lag phase of the strains of the genus Chlorella compared with the positive controls (p ≤ 0.05), as it can be seen for the values of the parameter a in Table 1. However, in the case of S. obliquus, the addition of the drug did not modify the beginning of the exponential growth phase compared with the positive control (p > 0.05). Furthermore, there were significant differences among the treatments with paracetamol, C. vulgaris showed a quite longer lag phase than S. obliquus and this one than C. sorokiniana.
At the end of the batch culture, the biomass concentration was increased above 49% by the presence of paracetamol in the C. sorokiniana culture (CCS+, 1.40 ± 0.29 g l −1 ; PCS, 2.09 ± 0.02 g l −1 ) and was increased above 31% in the C. vulgaris culture (CCV+, 2.60 ± 0.15 g l −1 ; PCV, 3.42 ± 0.15 g l −-1 ) compared with their positive control, as shown in Figure 1 and confirmed by K values in Table 1. However, S. obliquus culture was not significantly modified by the addition of the drug and the maximum algal density reached in the treatment (PSO, 3.27 ± 0.15 g l −1 ) was similar to the positive control (CSO+, 3.46 ± 0.08 g l −1 ). In spite of the different response of the strains to the presence of paracetamol, there were not significant differences between PCV and PSO, even though the value reached for the parameter K in the case of CSO+ was significantly larger than for CCV+. Still, the carrying capacity of the PCS treatment was significantly lower than for PCV and PSO.
Respect to microalgae growth rate (r), there was significant differences between the paracetamol treatment for C. vulgaris (PCV, 1.08 ± 0.04 d −1 ) and the corresponding positive control (CCV+, 0.84 ± 0.06 d −1 addition, when comparing the treatments of the three strains, there were not significant differences between the paracetamol treatments of C. vulgaris and S. obliquus strains (PCV, PSO) despite there were significant differences between their respective positive controls (CCV+, CSO+).

Growth of the culture under salicylic acid addition
The microalgae growth curves of C. sorokiniana, C. vulgaris and S. obliquus under the presence of salicylic acid, the corresponding positive controls and their fittings to the logistic kinetic model, are represented as values of biomass concentration versus time in Figure 2. The differences among the treatments were analysed according to growth kinetic parameters, as shown in Table 2.
Regarding the parameter a, the addition of salicylic acid increased significantly the lag phase of the strains C. vulgaris and S. obliquus compared with the positive controls. Also, C. sorokiniana treatment showed a higher a value than the positive control, in spite of the difference being not significant ( Table 2). Comparing the treatments with salicylic acid, C. vulgaris showed a quite longer lag phase than C. sorokiniana and S. obliquus.
As it can be seen in Figure 2, the maximum algal density reached at the end of the batch culture was significantly higher in the treatments with salicylic acid for all strains here considered as compared with the positive controls. The C. sorokiniana treatment increased their biomass concentration above 52% (CCS+, 1.40 ± 0.29 g l −1 ; SCS, 2.14 ± 0.13 g l −1 ), C. vulgaris above 18% (CCV+, 2.60 ± 0.15 g l −1 ; SCS, 3.09 ± 0.23 g l −1 ) and S. obliquus above 36% (CSO+, 3.46 ± 0.08 g l −1 ; SCS, 4.71 ± 0.30 g l −1 ) over their respective positive controls at the end of the batch culture. However, under salicylic acid, the carrying capacity of S. obliquus was significantly larger than those of C. sorokiniana and C. vulgaris.  . Moreover, the growth rate of SSO was significantly lower than that of SCS and SCV.

Growth of the culture under diclofenac addition
The microalgae growth curves of C. sorokiniana, C. vulgaris and S. obliquus under the presence of diclofenac, the corresponding positive controls and their respective fittings to the logistic kinetic model are represented as values of biomass concentration versus time in Figure 3. The differences among the treatments were analysed according to growth kinetic parameters, as shown in Table 3.
There were significant differences respect the parameter a (p ≤ 0.05) between the positive control and the corresponding treatment of each strain of microalgae, reaching higher values in the case of the treatments with diclofenac. Therefore, the presence of diclofenac produced a delayed response in the beginning of the exponential growth phase compared with the positive control. Comparing the treatments with diclofenac, C. sorokiniana showed a longer lag phase than C. vulgaris and S. obliquus As it can be seen in Figure 2, the treatments with diclofenac achieved significantly higher biomass concentration than their respective positive controls. At the end of the batch culture, the C. sorokiniana treatment showed an increase of biomass concentration above 45% (CCS+, 1.58 ± 0.11 g l −1 ; DCS, 2.30 ± 0.03 g l −1 ), C. vulgaris above 35% (CCV+, 1.96 ± 0.13 g l −1 ; SCV, 2.65 ± 0.10 g l −1 ) and S. obliquus above 11% (CSO+, 1.34 ± 0.03 g l −1 ; SCS, 1.49 ± 0.05 g l −1 ) over their respective positive controls.   K value, which was significantly higher than those determined for the C. sorokiniana and the S. obliquus treatments.
With respect to microalgae growth rate, there were significant differences between the positive control and the corresponding treatment for the two strains of the genus Chlorella here used. The C. sorokiniana growth rate was significantly increased under the presence of this drug (CCV+, 0.72 ± 0.04 d −1 ; DCS, 0.96 ± 0.01 d −1 ). This significant increase was also confirmed for C. vulgaris (CCV+, 0.56 ± 0.00 d −1 , DCV, 0.74 ± 0.01 d −1 ). However, no significant differences were determined in the case of S. obliquus (CSO+, 0.79 ± 0.03 d −1 , DSO, 0.81 ± 0.09 d −1 ).

Removal of pharmaceuticals
The pharmaceutical concentration in each reactor was daily monitored and compared with the concentration of each pharmaceutical in the corresponding negative control. The concentration of the pharmaceuticals here studied decreased over the time in the treatments with microalgae, either with C. sorokiniana (PCS, SCS, DCS), C. vulgaris (PCV, SCV, DCS) or S. obliquus (PSO, SSO, DSO). Meanwhile, no concentration reduction was observed in the negative controls (CP−, CS−, CD−). Therefore, it may be assumed that the pharmaceuticals concentration decrease in the microalgae treatments was due to the removal by the microalgae.

Removal of paracetamol
The removal curves of paracetamol by each strain of microalgae and the corresponding fittings to the logistic kinetic model during the batch mode are displayed in Figure 4(a). In addition, differences among the treatments were analysed according to removal kinetic parameters, as shown in Table 4.
Regarding the parameter a, there were no significant differences between C. vulgaris and S. obliquus in the lag phase for the removal of paracetamol. However, C. sorokiniana showed a significantly longer response at the beginning of the removal of this drug than the other two strains.  The parameter K values in Table 4 revealed that C. sorokiniana (PCS, 17.62 ± 0.91 mg l −1 ) reached a carrying capacity 2.8 times higher than C. vulgaris (PCV, 6.23 ± 0.02 mg l −1 ) and 1.7 times higher than S. obliquus (PSO, 10.41 ± 1.58 mg l −1 ). In the same way, the removal rates revealed significant differences among the treatments, with C. sorokiniana showing the quickest removal (PCS, 1.01 ± 0.06 d −1 ) and C. vulgaris the slowest one (PCV, 0.77 ± 0.01 d −1 ), which is in agreement with the determined K values.
As a consequence of the different responses obtained for the removal parameters between the strains, at the end of the batch culture, efficiencies in the removal of paracetamol above 67% for C. sorokiniana, 21% for C. vulgaris and 40% for S. obliquus were achieved. These results evidenced a larger removal capacity of paracetamol by C. sorokiniana, followed by S. obliquus, and C. vulgaris.  The average volumetric efficiencies on the paracetamol removal by each strain at the steady stage of the semicontinuous culture are depicted as percentages in Figure 4(b).
The paracetamol volumetric efficiency reached values above 41% for C. sorokiniana, 12% for C. vulgaris and 9% for S. obliquus. Moreover, the ratios between the volumetric efficiency and the microalgae biomass are shown in Table 4 as specific efficiencies. These results revealed that C. sorokiniana cells removed above 7.2 times more paracetamol than C. vulgaris and 8.4 times more than S. obliquus per gram of biomass. On the other hand, the paracetamol removal per gram of biomass was similar between S. obliquus and C. vulgaris.

Removal of salicylic acid
The removal curves of salicylic acid by each strain of microalgae and the corresponding fittings to the logistic kinetic model during the batch mode are displayed in Figure 5(a).
In addition, differences among the treatments were analysed according to removal kinetic parameters, as shown in Table 4.
In the case of C. sorokiniana there were significant differences respect the parameter a, which indicated that the beginning of the removal of salicylic acid had a delayed response as compared with the lag phase of C. vulgaris and S. obliquus.
The average salicylic acid volumetric efficiencies by each strain at the steady stage of the semicontinuous culture are depicted as percentages in Figure 5(b). The paracetamol volu-  metric efficiency did not showed significant differences between the strains C. sorokiniana and S. obliquus, reaching values above 93% for SCS and 99% for SSO. However, the salicylic acid volumetric efficiency of C. vulgaris (above 22%) was more than four times lower than by the other strains. Moreover, the obtained specific efficiencies revealed that C. sorokiniana removed above 12.4 times more salicylic acid than C. vulgaris and 4.5 times more than S. obliquus per gram of biomass ( Table 4).

Removal of diclofenac
The removal curves of diclofenac by each strain of microalgae and the corresponding fittings to the logistic kinetic model during the batch mode are displayed in Figure 6(a). In addition, differences among the treatments were analysed according to removal kinetic parameters, as shown in Table 4.
The a values were similar (p > 0.05) for all the treatments, which indicated that the three strains showed the same delayed response in the removal of diclofenac. However, regarding the maximum removal capacity, there were significant differences between the treatment with S. obliquus (DSO, 22.43 ± 0.20 mg l −1 ), which removed 1.5 times more diclofenac than by C. sorokiniana (DCS, 14.55 ± 0.73 mg l −1 ) and 1.4 times more than C. vulgaris (DSO, 15.52 ± 0.26 mg l −1 ).
Concerning the removal rate, the obtained results revealed significant differences among the treatments. The quickest removal rate was attained by C. sorokiniana (DCS, 2.03 ± 0.33 d −1 ), with removal values 1.6 times higher than S. obliquus (DSO, 1.25 ± 0.19 d −1 ) and 1.4 times higher than C. vulgaris (DCS, 1.44 ± 0.05 d −1 ). Despite the differences between strains regarding the removal parameters, at the end of the batch culture, efficiencies above 65% for C. sorokiniana, 69% for C. vulgaris and 98% for S. obliquus were achieved.
The average volumetric efficiencies for the diclofenac removal in the steady stage of the semicontinuous culture are showed in Figure 6(b). The volumetric efficiency for S. obliquus (above 79%) was 2.6 times higher than for C. sorokiniana and 3.7 times higher than for C. vulgaris. Moreover, the ratios between the volumetric efficiency and the microalgae biomass are shown in Table 4 as specific efficiencies. The determined values revealed that S. obliquus removed above 3.0 times more diclofenac than C. sorokiniana and above 5.4 times more than C. vulgaris per gram of biomass.

Discussion
In view of the obtained results, it may be inferred that the presence of paracetamol, salicylic acid and diclofenac modified the growth parameters of the strains here studied. In most of the treatments, the addition of the pharmaceutical increased the biomass concentration, which may be explained by the fact that these pharmaceuticals were an additional source of organic carbon. It is well known that the genus Chlorella and Scenedesmus can have a mixotrophic growth. However, S. obliquus did not show a significant increase of microalgae biomass under the addition of paracetamol or diclofenac. These results suggest that the other removal mechanisms, apart from metabolism, may be involved.
The fact that removal curves displayed a similar trend than growth curves points to the association between the microalgae growth and the removal efficiency of pharmaceuticals.
In view of the obtained results, it may be concluded that paracetamol was more efficiently removed by C. sorokiniana, either per litre or per gram of biomass (>67% batch culture, >41% semicontinuous culture), in spite of the biomass concentration reached in the culture being the lowest one among the three strains. Also, the removal rate by C. sorokiniana was the fastest one, in spite of showing the lowest growth rate among the paracetamol treatments. However, the addition of paracetamol in the C. sorokiniana culture produced the largest increase in the biomass concentration compared with the corresponding positive control (>49%).
On the other hand, S. obliquus showed the highest salicylic acid removal capacity at the end of the batch culture (>93%) and also at the steady state of the semicontinuous culture (>99%). However, the removal rate by S. obliquus was the lowest one among the salicylic acid treatments. The highest removal rate was reached by C. sorokiniana, which showed a removal per gram of biomass 4.5 times larger than S. obliquus. Furthermore, the increase of biomass under the addition of salicylic acid was above 52% in the C. sorokiniana treatment, while for S. obliquus was above 36%.
Regarding diclofenac, despite C. sorokiniana cells attained a higher removal rate and the higher growth rate, it may be stated that S. obliquus was the strain that reached the highest removal efficiency (>98% batch culture, >79% semicontinuous culture) with more diclofenac removed either per litre or per gram of biomass.
Comparing the three pharmaceuticals, the salicylic acid was more efficiently removed, with C. sorokiniana and S. obliquus showing the highest efficiencies. Contrarily, the paracetamol was the less efficiently removed. In all cases, C. vulgaris showed the lowest efficiencies for the three pharmaceuticals. These results may be related with the specific strain characteristics, the mechanisms involved in the removal and the particular properties of each pharmaceutical.
As in this work, published results on the removal of ECs by microalgae have revealed different efficiencies depending on the pollutant and on the microalgae strain. For example, Gattullo et al. [23] demonstrated that Monoraphidium braunii was able to remove up to 48% of bisphenol A with an initial concentration of 4 mg l −1 . de Wilt et al. [14] reported removal efficiencies by C. sorokiniana, grown in wastewater streams, up to 60-100% for diclofenac, ibuprofen, paracetamol and metoprolol. However, under identical conditions, the removal of carbamazepine and trimethoprim was incomplete and did not exceed 30% and 60%, respectively [14]. Wang et al. [24] studied the removal of phenol by Chorella sp. culture, obtaining removal efficiencies up to 100% from an initial concentration of 500 mg l −1 in 7 days. Peng et al. [25] reported removals above 95% of progesterone by S. obliquus and Chlorella pyrenoidosa, nearly complete removal of norgestrel by S. obliquus and almost 40% of norgestrel by C. pyrenoidosa. Likewise, Hom-Díaz et al. [15] studied the elimination of the hormones E2 and EE2 from anaerobic digestate centrate by the microalgae Selenastrum capricornutum and Chlamydomonas reinhardtii. After 7 days of culture, these authors [15] determined removals above 88% for E2 and above 60% for EE2. Furthermore, Matamoros et al. [26] studied the capability of microalgae-based wastewater treatment systems to remove diclofenac, among other 25 emerging organic contaminants. These authors [26] determined diclofenac removal efficiencies above 82% under HRT of 4 days and above 92% under HRT of 8 days during the warm season (11-26°C, on a daily average). These efficiencies are higher than the here obtained under an HRT of 80 h and temperature of 25±1°C. Differences must be related, at least to some extent, to the fact that microalgae monocultures were used in this work while Matamoros et al. [26] worked with mixed microalgae strains present in the wastewater, mostly identified as Stigeoclonium sp., diatoms, Chlorella sp. and Monoraphidium sp.

Conclusions
Among the here considered strains, S. obliquus displayed the highest removal efficiency for salicylic acid and diclofenac, while C. sorokiniana did it for paracetamol. On the other hand, C. vulgaris showed the lowest efficiencies for the three pharmaceuticals. Comparing the three pharmaceuticals, the salicylic acid was more efficiently removed while paracetamol removal was the less efficient. These differences may be related with the specific strain characteristics, the mechanisms involved in the removal and the particular properties of each pharmaceutical. The obtained results pointed to the feasibility of using the microalgae here considered in bioremediation systems and revealed that this sort of studies are key for the selection of the strain, which depends on the application. Still, further research is needed to assess the mechanisms involved in the removal of pharmaceuticals by these strains.
respectively). Marta Otero acknowledges University of León for the extension of her RYC-2010-05634 contract.