All plant treatments were created in quadruplicate and solution treatments were created in triplicate. Laboratory blanks were included with each sample extraction and pure methanol was analyzed in each UPLC/MS/MS run to check potential contamination. Surrogates were used in all sample analyses to account for losses during extraction and matrix effects during instrumental analysis. Recovery of the surrogates was used to calculate the actual concentration of each target analyte. Recoveries of surrogates in plant tissue and nutrient solution samples are listed in Table S5.1 of the Supporting Information. Statistical analysis of data including ANOVA with Tukey’s Honestly Significant Difference, linear regression, and t-test was performed using R . Significance was assigned at p ≤ 0.05. Carrot, lettuce, and tomato plants grown in both environments were found to be generally healthy, and no difference in biomass was detected between plants grown in solution with or without PPCP/EDCs. For the same plant species, those from the warm-dry environment generally had greater bio-masses. One tomato plant from the cool-humid treatment had yellow, stunted leaves and was excluded from the analysis. The nutrient solution pH was measured at each solution exchange, and was found to average pH 5.2 for carrot, pH 5.3 for lettuce, and pH 6.0 for tomato during the study. The average pH values were used to calculate the neutral fraction and the pH-adjusted octanol-water partition coefficient for the different PPCP/EDCs, as described in Wu et al. . Only small differences in neutral fractions and log Dow values were seen between treatments for the same compound,vertical vegetable tower mostly for compounds with pKa values near the solution pH. Based on the primary ionic state in the nutrient solution, the selected PPCP/EDCs were placed in either the anionic, cationic, or neutral group .
The transpired mass for each plant was measured at every solution exchange and the data were used to calculate the cumulative transpiration . For lettuce and tomato, the different temperatures and air humidity resulted in significantly different transpired masses . The differences were smaller for carrot seedlings , likely due to the considerably smaller leaf masses of the carrot plants. The mean transpired masses in the cool-humid and warm-dry treatments during the 21 d of growth were, respectively, 65.50 ± 19.36 and 194.33 ± 30.72 g/d for lettuce, 127.04 ± 15.52 and 503.38 ± 59.76 g/d for tomato, and 16.82 ± 8.05 and 55.31 ± 26.41 g/d for carrots. For the same plant type, the warm-dry environment induced a 3-4-fold increase in plant transpiration as compared to the cool-humid environment. The dissipation of PPCP/EDCs from nutrient solution during the hydroponic growth of plants may be attributed to plant uptake and microbial degradation in the solution. The change in PPCP/EDC concentrations was measured on day 10, after 2 d incubation. In the spiked nutrient solutions without plants, most PPCP/EDCs showed limited dissipation from the solution , suggesting that these compounds were mostly stable in the nutrient solution . The only exception was atorvastatin, where 49.0% and 61.7% were not recovered for the cool-humid and warm-dry treatments, respectively . In the presence of plants, levels of PPCP/EDCs in the solution significantly decreased compared to the plant-free control. For example, after exposure to a tomato plant, about 38.8% of the initially spiked diclofenac was not recovered from the solution for the cool humid treatment and about 75.6% for the warm-dry treatment, while there was essentially no chemical loss in the plant-free container . When all compounds were pooled, removal from the solution was found to be consistently greater in the warm-dry treatment than in the cool-humid treatment. This difference was statistically significant for lettuce and tomato , but not for carrot , likely due to its very small biomass. For example, in the cool-humid and warm-dry treatments, the respective losses of gemfibrozil were 18.2% and 28.6% for carrot, 64.5% and 89.2% for lettuce, and 55.6% and 91.8% for tomato .
These trends clearly suggested that the warm-dry environment and the corresponding larger plant biomass in the warm-dry treatments, contributed to enhanced PPCP/EDC dissipation in the nutrient solution . The transpired mass over the 2 d period was compared to the measured removal of the anionic, cationic, or neutral PPCP/EDCs over the same period to assess the effect of plant transpiration on the removal of PPCP/EDCs from the nutrient solution. A significant positive relationship was found for each group of compounds , suggesting that the removal of PPCP/EDCs in the nutrient solution increased with transpiration for both ionic and neutral compounds, and across different plant species. The separation of PPCP/EDCs by ionic state in the regression analysis decreased the model residuals for both the cationic and neutral groups, as compared to a linear regression with all compounds grouped together , showing that consideration of ionic states better describes the interaction of PPCP/EDC and transpiration. Transpiration had the greatest impact on removal of neutral compounds, as shown by a slope of 0.048 for the linear regression , followed by anionic compounds , while removal of cationic compounds was least affected by transpiration . Since neutral compounds are expected to move through root membranes according to diffusion, and ionic compounds are subject to electrical effects, it is reasonable to expect that transpiration exerts the most effect on neutral compounds. Other compound characteristics besides ionic state, such as hydrophobicity and stability, may also influence PPCP/EDC dissipation in the nutrient solution and may help explain the remaining regression residuals. To evaluate the contribution of adsorption to root structures, log Dow values for each group of compounds were compared to their removals in the nutrient solution. However, no significant relationship was found for any of the treatments , suggesting that log Dow alone was not a good predictor for PPCP/EDC removal from the nutrient solution. It must also be noted that due to the small number of compounds in the cationic group, the analysis may not be sufficiently strong to be generalized for other cationic PPCP/EDCs.
To facilitate comparisons of PPCP/EDC accumulation among different compounds and between different treatments, a bio-concentration factor was calculated by dividing the concentration of a compound in a plant tissue after the 21 d cultivation to the concentration in fresh solution . In this study, atorvastatin, diclofenac, and clofibric acid were the least accumulated , while perfluorooctanoic acid, diazepam, and diuron were the most accumulated compounds . After averaging across all compounds and plant types, BCF values for root tissues were found to be significantly higher than those for leaves , with the respective mean BCF values of 51.3 and 21.0. These BCF values suggest that many PPCP/EDCs have the ability to bio-accumulate in plant tissues, and the overall accumulation into roots likely exceeds that into leaves. In addition, some PPCP/EDCs may be accumulated to relatively high levels. In general, BCFleaf values followed the order cationic ≥ neutral > anionic and BCFroot values were in the order anionic > neutral ≥ cationic,vertical farming equipments suggesting that accumulation of cationic and neutral compounds was somewhat similar. However, anionic compounds were accumulated significantly less than cationic or neutral compounds in leaves, but significantly more in the roots. The reversed trends of accumulation between leaf and root tissues were mainly caused by the behavior of anionic compounds. For anionic PPCP/EDCs, accumulation in root was significantly more than in leaf , with the mean BCF root at 72.8 while the mean BCFleaf at 3.3. In comparison, accumulation into leaf and root tissues was similar for cationic or neutral compounds . Overall, these results suggest that root tissues may accumulate high levels of anionic compounds, while in leaf tissues, cationic and neutral compounds may be more prevalent. A few other studies have considered some of these same PPCP/EDCs under hydroponic conditions, but often used higher spiking concentrations. Herkltoz et al. investigated the growth of cabbage in solution spiked with carbamazepine, sulfamethoxazole, and trimethoprim at 232.5 µg/L and found BCF values of 0.045 – 0.081 in leaf tissues and 7.04 – 10.92 in root tissues, values similar to this study for sulfamethoxazole and carbamazepine in root , but lower than carbamazepine accumulation in leaves or trimethoprim accumulation . In another study, Zhang et al. measured the uptake of clofibric acid by Scirpus validus from a culture spiked at 0.5 – 2 mg/L, and observed wetweight BCFs of 9.5 – 32.1 in leaf tissues and 6.6 – 23.2 in root tissues. These values were similar to the uptake of clofibric acid in this study . Wu et al. examined many of the same compounds at similar concentrations in nutrient solution growing cucumber, lettuce, pepper, or spinach under greenhouse conditions and observed similar BCF values in leaf and root tissues. The different environment conditions influenced bio-concentration of the PPCP/EDCs in the test plants. The mean overall BCF in the warm-dry treatment was 33.7, which was greater than that in the cool-humid treatment , although the difference was not statistically significant , likely due to the large differences in plant biomass and the wide range of chemicals used in this study. However, when BCFleaf was correlated to the transpired mass during the 21 d of plant growth, a positive correlation was observed for anionic, cationic, and neutral compounds . This result suggests that the mass flow of water caused by plant transpiration influenced the accumulation of PPCP/EDCs in leaves. Transpiration had the greatest impact on the leaf bio-concentration of cationic PPCP/EDCs, as shown by a model slope of 0.0067, while the effect was less for neutral PPCP/EDCs and much less for anionic PPCP/EDCs, suggesting that increased transpiration will have the greatest effect on leaf uptake of cationic compounds and little effect on leaf uptake of anionic compounds. This result is somewhat different than that seen for the removal of PPCP/EDCs from the nutrient solution. The difference may be attributed to other factors in addition to plant uptake, including microbial degradation in the nutrient solution.
In contrast, a relationship between BCFroot and transpired mass was only observed for the neutral group . High residuals in the linear model analysis further suggested that other factors, such as plant species, metabolism after uptake, and likely other compound properties, may also be important in describing PPCP/EDC accumulation into plant tissues. For anionic compounds, it is known that the negative charged molecules may experience repulsion from negatively charged root cell membranes, and that plant accumulation of anions may be mainly due to diffusion of the neutral fraction through the membrane and ion trap effects . A comparison of BCF values of anionic compounds in all plants with their respective log Dow showed a negative correlation for BCF leaf or BCF root , suggesting that anionic compounds with lower effective hydrophobicity had higher accumulation in the leaf or root tissues . This effect was greatest for root tissues, and the slope of the linearized regression was -54, while for leaf tissues the slope was only -0.63, suggesting other factors besides hydrophobicity may have a larger impact on the aerial uptake of anionic compounds. The cationic fraction of a compound may slowly diffuse through plant membranes due to electrical attraction between the positively charged molecules and the negatively charged cell membrane, while the neutral fraction may diffuse with preference to compounds of moderate hydrophobicity . In this study, a positive correlation was observed between BCF leaf and log Dow for cationic PPCP/EDCs in all plants , suggesting that more hydrophobic cationic PPCP/EDCs have a higher accumulation potential in leaf tissues. Further, this effect was relatively strong, with a slope of 10 for the linear regression, as compared to a slope of 6.6 for neutral compounds or -0.63 for anionic compounds. It has been shown that the accumulation of cationic organic compounds in aerial tissues was the greatest for compounds with log Kow between 2.5 –5.5 . In this study, for example, uptake of dilantin into the leaves was greater than that of trimethoprim . In comparison, no significant correlation was observed between BCFroot and log Dow for cationic compounds , suggesting that other factors also contributed to the accumulation of cationic compounds in roots. However, it must be stated again that the limited pool of cationic compounds in this study hampered a more conclusive examination of cationic PPCP/EDCs and that the assumption merits further validation.