Toxicity and removal of pharmaceutical and personal care products: a laboratory scale study with tropical plants for treatment wetlands

Pharmaceutical and personal care products (PPCPs) are chemically and biologically active substances that can be persistent in the environment (Santos et al. 2010). They are a part of group of substances known as emerging contaminants. The United States Environmental Protection Agency (USEPA) defines them as new chemical products without regulation that generate an unknown impact in their surroundings (Deblonde et al. 2011). Concerns about the presence of PPCPs and their possible effects in aquatic ecosystems have arisen in the past years, since continuous discharge of these compounds may cause accumulation and irreversible damage to wildlife and human health (Liu & Wong 2013; Ortiz et al. 2013); however, the effects on human beings are not well identified to date (Santos et al. 2010).

PPCPs identified in natural environments belong to a diverse group of organic substances. Sources of these are pharmaceutical products such as analgesics, antibiotics, anticonvulsants, lipid regulators, and hormones, and daily use personal care products such as soaps, lotions, sun blocking lotions, and toothpaste, among others (Liu & Wong 2013). The main source of these compounds to nature is through inadequate disposal of unused or expired medicines from homes and hospitals, by excretion, indiscriminate use of fertilizers, fungicides and pesticides in agriculture, and direct discharge of veterinary products (Santos et al. 2010). In those ways, they reach the aquatic environments through point source discharges such as effluents from sewer system and wastewater treatment plants (Zhang et al. 2014), as well as from non-point sources such as rainfall runoff and leachates from agricultural and livestock production (Mompelat et al. 2009).

Conventional technologies for wastewater treatment and drinking water treatment for human consumption cannot efficiently remove this type of compound (Dordio et al. 2010), so these substances remain in the water, even after wastewaters have been treated, or surface waters processed to drinking water (USEPA 2002). More advanced treatment technologies such as ozonation, advanced oxidation, activated carbon, nanofiltration, and reverse osmosis can remove most of the PPCPs present in wastewaters reaching elimination rates above 90% (Sui et al. 2010; Yang et al. 2011; WHO 2012). However, these technologies are not widely used due to their high cost and operational and maintenance demands (Bolong et al. 2009).

Treatment wetlands are systems widely used, of relative low cost for secondary and tertiary treatment of wastewaters (Matamoros et al. 2008; Dordio et al. 2010; Li et al. 2014), and having shown successful results in the elimination of PPCPs (Matamoros et al. 2007, 2009; Ávila et al. 2013; Verlicchi et al. 2013). In this type of systems, the interactions of plants, microorganisms, and water, allow the removal of contaminants from wastewaters, ranging from conventional pollutants to persistent compounds (Zhang et al. 2015). Macrophytes carry out chemical, physical and biological processes such as volatilization, absorption, accumulation, adsorption, translocation, and degradation for the elimination of the pharmaceutical compounds (Zhang et al. 2014).

Dordio et al. (2010) reported that plant species such as Thypa sp. in hydroponic environments reach removal efficiencies of 96, 97, and 75% for ibuprofen, carbamazepine, and clofibric acid, respectively. Zhang et al. (2013a) reported absorption as the principal elimination mechanism for caffeine in hydroponic environments planted with Schoenoplectus tabernaemontani. On the other hand, the results from Matamoros et al. (2008) indicate that ibuprofen degrades 51% in subsurface flow treatment wetlands, highlighting that low aerobic processes in the systems evaluated were the major contributors to ibuprofen degradation. Studies performed by Hijosa-Valsero et al. (2010) ascribe the elimination of naproxen, ibuprofen, diclofenac, carbamazepine, methyl dihydrojasmonate, galaxolide, and tonalide compounds to the presence of plants in treatment wetlands because they have the capacity to retain particles in suspension, release O₂, serve as support for biofilm, and insulate during low temperatures. For Zhang et al. (2014) the principal removal mechanisms for PPCPs are photolytic degradation, sorption over sludge or particles, filtration, chemical oxidation, photocatalysis, and microbial degradation. In treatment systems based on plants, the absorption and the translocation are the primary mechanism for removal.

The diffusion processes of PPCPs in plant tissues depend on their physical–chemical characteristics, which include the compounds' hydrophobicity (K_WO) and water solubility. Molecules with K_WO values between 0.5 and 3.5 present moderate hydrophobicity allowing the compound to move freely in the plant between the lipid layer and the aqueous phase. In contrast, highly hydrophobic compounds (K_WO>3.5) are more prone to be absorbed in the roots of plants, while highly polar drugs with values of K_WO<0.5 are incorporated by the plant and degraded by different metabolic processes (phytodegradation) (Dordio & Carvalho 2013; Li et al. 2014; Zhang et al. 2014). Most recently, Brunhoferova et al. (2021) tested the removal of a mixture of 27 micropollutants (pharmaceuticals, pesticides, herbicides, fungicides, and others) from aqueous solution by phytoremediation, with comparable and even higher removal efficiency than conventional wastewater treatment plants. The main objective of this study was to assess the tolerance and elimination capacity of PPCPs in planted microcosm reactors with Heliconia psittacorum, Ciperus haspan, and Hedychium coronarium.

The PPCPs selected for the study were: acetylsalicylic acid (aspirin), ibuprofen, naproxen, diclofenac, ketoprofen, caffeine, diclofenac, methyl-hydrojasmonate (cis – MDJM), galaxolide, and tonalide, which are products frequently consumed and freely available in the market. The products were commercially acquired in local pharmacies in Colombia, which distribute medicines manufactured by Genfar, MK, La Santé, and Bayer. Table 1 presents the characteristics of each of the evaluated compounds.

The study was carried out in two stages: Stage I – Tolerance, and Stage II – Removal. For Stage I, the species selected were H. psittacorum, C. haspan and H. coronarium, common wetlands plants in tropical areas. The plants were selected and collected from natural wetlands, and transported to laboratory, where their roots were thoroughly washed with distilled water to remove soil and humic compounds. Eighteen individuals of similar size of the studied species were randomly selected to be hydroponically planted in 1 liter glass reactors lined with aluminium foil (to inhibit algae development).

The reactors were filled with an adapted Hoagland nutrient solution (Alkhatib et al. 2013) to guarantee the nutritional requirements of the plants (202 g L⁻¹ KNO₃, 147 g L⁻¹ CaCl₂, 15 g L⁻¹ Fe-EDTA, 493 g L⁻¹ MgSO₄.7H₂O, 159.98 g L⁻¹ NH₄NO₃, 2.86 g L⁻¹ H₃BO_3, 1.81 g L⁻¹ MnCl.4H₂O, 0.22 g L⁻¹ ZnSO₄.7H₂O, 0.051 g L⁻¹ CuSO₄, 0.12 g L⁻¹ Na₂MoO_4, and 136 g L⁻¹ KH₂PO₄). After an adaptation period of 3 weeks in hydroponic conditions, six groups (triplicates) of reactors per each plant species were stablished. Five groups of reactors were spiked with a PPCP cocktail of the selected compounds at different concentrations (10, 100, 1,000, 5,000, and 10,000 by PPCP); likewise, one group of blanks (triplicate) filled only with nutrient solution (0 μg/L by PPCP) were placed as controls. The reactors were exposed to natural environmental conditions (under translucid shelter to protect against rainfall and allow the passage of natural light), with day/night cycles of ±12 h, and temperature ranging from 24 °C (day) to 17 °C (night). A total of 54 reactors were monitored (Table 2).

The reactors corresponding to Stage II were set up with the two species showing the best tolerance in Stage I. From each species two similar-sized individuals were taken and placed in 8-liter reactors, lined with aluminum foil and fed with Hoagland nutrient solution (Alkhatib et al. 2013). The reactors were spiked with the same PPCP cocktail used in Stage I, but with a concentration of 25 μg/L by compound. The photodegradation effect was determined by setting up two blank reactors without plants and exposed to the same environmental conditions of the rest of the mesocosms. Two additional reactors without plants and covered were set up to determine the microbial degradation effect; 8 reactors in total were implemented to monitor the PPCPs' elimination (Table 2). The reactors were submitted to the same environmental conditions as in Stage I.

Stage II also had an experimental period of 21 days, where grab water samples were taken on day 1, 3, 12 and 21. For the sampling, each unit was homogenized, and 1,000 mL of sample were taken and bottled in amber color glass bottles to be preserved at 4 °C (Standard Methods – APHA 2012). Thereafter, the samples were analyzed by the Laboratory of Chromatography and Mass Spectrometry – CROMMAS of Universidad Industrial de Santander – UIS (Colombia).

The water samples were filtered through fiberglass filters with a pore diameter size of 0.45 μm and 47 mm, and subsequently acidified with concentrated chloride acid up to pH<2. Afterward, 500 mL were percolated with polymeric PE cartridges packed with 100 mg of Strata X, previously prepared with 5 mL of n-hexane, 5 mL of ethyl acetate, 10 mL of MeOH and 10 mL of Milli Q water. The filtration rate was adjusted to approximately 10 mL min⁻¹. Then, the cartridges were dried and finally the analytes were eluted with 5 mL of ethyl acetate. The extract was evaporated until dried completely under a soft flow of reconstituted nitrogen in 175 μL of methanol (Matamoros & Bayona 2006).

The sample extracted in the SPE phase was injected into an AT 6890 series plus chromatographer (Agilent Technologies, Palo alto California, USA) coupled with a selective mass detector (Agilent Technologies, MSD 5797 Inert XL) operated in SIM mode (selected ion monitoring). The column employed in the analysis was DB-5MS ((5% phenyl)-methylpolysiloxane, 60 m×0.25 mm×0.25 μm). The injection was carried out in the splitless mode (Vol inj.=2 μL). The column used was capillary HP-5 (5% phenyl, 95% dimethyl-polysiloxane) (30 m×0.25 mm ID, 0.25 μm thickness phase), the He gas travel was 1.0 mL min⁻¹, and the injection volume was 2 μL in splitless mode. The methylation of the carboxylate acid group was carried out in line in a hot gas injector with an added 10 μL solution of trimethylsulfonium hydroxide TMSH (0.25 mol L⁻¹ in methanol) to 50 μL of sample before the injection. In the ion selective monitoring mode the following ions were selected:

Methyl ester salicylic acid (92/120/121/152), methyl ester ibuprofen (161/162/177/220), methyl ester naproxen (170/185/186/244), methyl ester diclofenac (214/216/242/309), methyl ester ketoprofen (191/209/210/268), caffeine (82/109/165/194), dihydrojasmonate (83/153/156/226), galaxolide (213/243/244/258), tonalide (201/243/244/258).

To evaluate the results of the impact of PPCP concentrations over biomass growth of the tested plant species, a non-parametric analysis was carried out using the Mann Whitney test. (P<0.05). An ANOVA and Tukey HSD analysis was implemented (P<0.05) to identify significant differences in the removal of PPCP among phytodegradation, biodegradation, and photodegradation reactors. The results were analyzed with the statistical package XLSTAT 2021.5.1.1204 (Addinsoft 2021).

Figure 1 presents results of RGR of tested plants exposed to different PPCP concentrations. After 21 days of test, the results showed a decreasing trend in biomass gain for all the plants as the spiked PPCP concentration increases. At 1,000 μg L⁻¹H. psittacorum and H. coronarium plants showed a strong growth inhibition in comparison to control plants (at 0 μg L⁻¹), with a depletion in RGR of 101 and 97% respectively (without significant differences among them), whereas Ciperus haspan was the least affected plant with a decrease in RGR of ±40%.

When reactors were spiked with PPCPs from 5,000 μg L⁻¹, all the tested plants showed a decrease in RGR above 100%, denoting a loss of vegetal biomass. Likewise, the tolerance index (TI%) showed that from 1,000 μg L⁻¹, H. psittacorum and H. coronarium plants presented marked symptoms of toxicity, while for C. haspan plants, the same marked toxicity signs were expressed from PPCP concentration of 5,000 μg L⁻¹ (Figure 1).

The impact of different PPCP concentrations on the growth of stems and roots (growth effect – GE) can be seen in Figure 2. Similarly to the impact of different PPCPs concentration on RGR, the GE of H. psittacorum and H. coronarium plants was affected from concentrations of 1,000 μg L⁻¹, a fact evident after the second monitoring week, when growth inhibition was observed and represented in biomass losses. After 21 days of experiment, the average GE of stems for H. psittacorum and H. coronarium spiked with PPCPs at 1,000 μg L⁻¹ were ±85% and ±99% lower than in control plants (0 μg L⁻¹), whereas the average GE of stems C. haspan was ±25% higher than in control plants. From 5,000 μg L⁻¹ all tested species showed stem growth inhibition and loss of height due to withering. Regarding the GE of roots, in C. haspan it was ±91% lower that in control plants (0 μg L⁻¹ PPCPs), whereas the GE of roots for H. psittacorum and H. coronarium showed growth inhibition and loose in length due to withering when spike with PPCPs at 1,000 μg L⁻¹.

Table 3 presents a comparison of the appearance of chlorosis among tested plant species. The results indicated that after 21 days of exposure, in reactors spiked with PPCPs concentrations below 100 μg L⁻¹, the plants did not show evidence of chlorosis. At 100 μg L⁻¹ concentration, C. haspan and H. coronarium plants showed ‘low’ and ‘middle’ signs of chlorosis respectively, signs that start to be evident with ‘low’ chlorosis levels since day 7; after 21 days of experiment H. psittacorum did not show chlorosis signs when spiked with PPCPs at 100 μg L⁻¹. At the end of the test the H. coronarium and H. psittacorum plants displayed ‘high’ chlorosis symptoms when spiked with PPCP concentrations above 1,000 μg L⁻¹, whereas C. haspan plants just reach a ‘high’ chlorosis level. For PPCP concentration of 10,000 μg L⁻¹, H. coronarium and H. psittacorum plants reached signs of ‘necrosis’. Control plants fed only with Hoagland nutrient solution did not show any symptom of chlorosis.

Regarding growth expressed as biomass, there are statistically significant effects (p<0.05) in C. haspan and H. coronarium (Table 4), while the effect of the concentration over biomass growth presents significant differences on a p<0.05 between concentrations of 0 and 1,000 μg/L, and the corresponding concentrations to 5,000 and 10,000 μg/L (Table 5). These differences are corroborated through a hierarchical conglomerate analysis that allowed identifying and classifying two main groups in agreement to their concentrations (Figure 3).

The loss of biomass in the plants exposed to high PPCP concentrations may be explained by the decrease of the redox capacity on the cellular membranes that are affected by the exposition to xenobiotic agents and cause the increase of reactive oxygen species that drive the cellular death of the plant (Iori et al. 2012). Another physiological response to the stress caused by the added pollutants is reflected by the inhibition of radicular growth that disturbs the plant processes making it prone to diseases and inducing changes in the chlorophyll content, thus decreasing its photosynthetic capacity (Nadeem et al. 2014).

The results of Stage I allow the conclusion that the species C. haspan and H. psittacorum and are the most tolerant to exposure to high PPCP concentration, and are therefore fit to be planted in treatment wetlands for the elimination of PPCPs.

Figure 4 shows the removal curves of the compounds acetylsalicylic acid, ketoprofen, and tonalide from the nutrient solution. The removal of acetylsalicylic acid in the first 24 h reaches 75% for H. psittacorum and 84% for C. haspan, biodegradation, and photodegradation reactors; given that all reactors with final concentrations below 3.9 μg L⁻¹ (quantification limit of the analytic method employed), it is hard to determine the main removal mechanism, however, based on the evidence reported in previous studies, its removal it may be specially attributed to the degradation by biological processes (Hijosa-Valsero et al. 2010; Zhang et al. 2012a; Li et al. 2014). The compounds ketoprofen and tonalide were 100% removed during the first 24 hours, and as in the case of acetylsalicylic acid, it is not possible to distinguish the main elimination pathway given the quick removal of the mentioned compounds. However, several studies consider that the photolytic degradation is the main mechanism for the decrease in the recalcitrant compounds such as ketoprofen (Hijosa-Valsero et al. 2010; Matamoros & Salvadó 2012; Reyes et al. 2012; Zhang et al. 2012a; Zhang et al. 2014), whereas tonalide is a lipophilic compound that tends to adhere to solids on different surfaces (Ávila et al. 2010; Hijosa-Valsero et al. 2010), with sorption the main elimination path due to its hydrophobic character and high K_ow of 3.933 (Zhang et al. 2014).

Removal curves of the compounds caffeine, diclofenac, galaxolide, ibuprofen, naproxen, and methyl dihydrojasmonate (MDJM) are presented in Figure 5. Caffeine is removed 96% in the planted reactors, 90% in the covered reactors, and 80% in the reactors exposed to light, suggesting that the main removal mechanism is given by the action of plants; the obtained results are comparable with those reported by Zhang et al. (2013a) and Matamoros et al. (2012), who in planted hydroponic reactors reported caffeine removals of 99%, being the absorption by plants the main removal mechanism, and in a lesser proportion by photocatalytic action and microbial degradation. The elimination of caffeine by the action of plants is mainly associated to its ionizable character, polarity and high solubility in water, characteristics that allow it to be absorbed by the plant and translocated to the plant's organs (Li et al. 2014).

Diclofenac is removed in 93% by photolytic action and in a lesser proportion by phytodegradation (H. psittacorum – 66%; C. haspan – 79%), and biodegradation (59%); Zhang et al. (2012b) found that Diclofenac may be eliminated in 80% by photodegradation, however the plant absorption also represents a passive elimination mechanism of this compound due to its high hydrophobicity; Matamoros et al. (2012) also reported removals superior to 99% on reactors exposed to light, classifying Diclofenac as a compound highly photodegradable.

Galaxolide is removed 76% in H. psittacorum reactors, 83% in C. haspan reactors, 77% by photodegradation, and 65% by biodegradation; these results are similar to those found in other studies (Hijosa-Valsero et al. 2010; Carranza et al. 2014) that report removals between 0 and 80%, however it is hard to establish which is the predominant removal mechanism in the assessed reactors. The removal efficiency of ibuprofen at the end of the experiment was 99% for planted and photodegradation reactors, whereas biodegradation reactors reached removal efficiencies of 92%. The degradation curve of this compound indicates that the main removal mechanisms were phytodegradation and photodegradation; similar results were found by Hijosa-Valsero et al. (2010), who established that the removal of ibuprofen is given under a metabolic aerobic mechanism, and the elimination is favored by the presence of plants, attributing this fact to the modification on the redox values in the roots. Zhang et al. (2011) also reported high removal efficiencies of ibuprofen on planted systems, attributing its elimination to low K_ow and the rhizosphere action that favors an oxidant environment, favoring microbial activity; Dordio et al. (2010) stated that biodegradation is the main elimination mechanism due to lipophile moderated by its low K_wo that benefits movement through cellular membranes and inclusion into the transpiration plant's root.

Naproxen is removed more than 90% in all the reactors for the entire period of the test, and the relationship between exposition and elimination times, indicate that photodegradation is the main elimination mechanism; these results are coherent with the ones reported by Zhang et al. (2013b), who obtained removals of more than 90% in the reactors exposed to the sunlight, thus indicating the importance of the photocatalytic action in the elimination of naproxen. At the same time, Cardinal et al. (2014) indicated that direct photolysis represents 100% elimination of naproxen.

MDJM is 99% removed in reactors planted with con H. psittacorum, 74% in reactors planted with C. haspan, 84% in reactors exposed to light, and 66% in the covered reactors without plants. These results are coherent with previous studies where MDJM is exposed as a hydrophilic substance easily degradable with removal efficiencies of relatively high masses (>80%) mainly as a result of its high biodegradation rate (Matamoros & Salvadó 2012), and where the presence of plants highly contributes in its elimination (Hijosa-Valsero et al. 2010).

The ANOVA and Tukey HSD analysis of removal rates at the end of the experiment (Table 6), confirm the results of Stage II, indicating that for certain compounds (MDJM, diclofenac, and ibuprofen) there are significant differences by type of reactor (p<0.05).

In Figure 6 is summarized the removal rate of the PPCP compounds, which are classified in two big blocks in agreement with the tested elimination mechanism in each reactor: block 1, where easily absorbed compound are found and eliminated by the action of plants, and a second block formed by compounds eliminated by photocatalytic action. Plants play a significant role in the direct absorption of many contaminants, even when not having the support of specific drivers for absorption of xenobiotic organic compounds, making the PPCPs inside plants being driven by diffusion according with the hydrophobicity of each compound (Li et al. 2014).

The results of the experiment indicated that the compounds ibuprofen, caffeine, MDJM and galaxolide are removed in a more efficient way in planted reactors. The compound MDJM is principally removed by H. psittacorum, while the galaxolide is more efficiently removed by C. haspan. Caffeine and ibuprofen presented the same elimination rate in both species. Even though there is not a clear differentiation in the removal mechanism for acetylsalicylic acid and tonalide, several authors conclude that their removal is mainly due to the action of the plants (Ávila et al. 2010; Hijosa-Valsero et al. 2010; Zhang et al. 2012a; Li et al. 2014).

The photolysis occurs when the compounds absorb photons to form oxygen reactive species (Cardinal et al. 2014). In this experiment, around 90% of diclofenac is removed mainly in reactors exposed to the sunlight. Similarly, naproxen is removed mainly by the photolytic action and in lesser way by the action of the H. psittacorum. On the other hand, the elimination rates of ketoprofen did not allow fairly distinguishing between the main elimination paths, but the literature reviewed highlights photodegradation as the main removal mechanism (Hijosa-Valsero et al. 2010; Matamoros & Salvadó 2012; Reyes et al. 2012; Zhang et al. 2012a; Zhang et al. 2014).

The toxicity response to PPCPs in the tested plants is reflected in the reactors with concentrations up to 1,000 μg L⁻¹; above that concentration evidence was found of growth inhibition, loss of biomass, and necrosis, forming two main groups according to concentration and impact on plants. A first group corresponding to the absence of toxic responses for concentration inferior to 1,000 μg L⁻¹, and a second group that reports toxicity from 5,000 μg L⁻¹. In this form the species H. psittacorum and C. haspan are distinguished as the most tolerant and suitable for the removal of PPCPs.

The highest removal of the assessed compounds occurred during the first 24 hours of experimentation, thus identifying two blocks of PPCP elimination: one block formed by the compounds acetylsalicylic acid, caffeine, galaxolide, ibuprofen, MDJM, and tonalide that could be removed in more than 90% by the action of the plants or their exudates, and second block formed by the substances diclofenac, ketoprofen, and naproxen that are mainly eliminated by photolytic action. The results allow the conclusion that the presence of macrophytes in the assessed reactors favor the removal of studied PPCPs and show an interesting potential to be implemented in real-scale wetland systems to treat that kind of pollutant in tropical areas.

Further studies should be carried out using lower concentrations of the tested compounds, such as the ones found in wastewaters, as well as in natural waters. Also, to complement the results of this study, it is recommended to carry out further studies to identify intermediate metabolites derived of the degradation of the studied compounds, as well as the analysis of to corroborate the bioaccumulation and/or transformation potential of tested PPCPs in different plant tissues.

Funding for this work was provided by Colciencias (Science, technology, and innovation department in Colombia), the World Bank, and the Universidad Tecnológica de Pereira. We also thank staff of Water and Sanitation Research Group for assistance with the experimental setup and sampling.

All relevant data are included in the paper or its Supplementary Information.