Long-term bioconcentration kinetics of hydrophobic chemicals in Selenastrum capricornutum and Microcystis aeruginosa

Algae

Cultures of S. capricornutum (NIVA-CHL1, Rhaphidocelis subcapitata) and M. aeruginosa (NIVA-CYA043) were supplied by the Norwegian Institute for Water Research (NIVA). The algae were cultured in a 25% Z8 mineral medium [22], prepared axenically in 0.2 μm filtrated Nanopure water (Sybron-Barnstead, Dubuque, IA, USA). The light regime was 16: 8 h light: dark at 100 μE/m². Temperature was 20 ± 2°C. Algal subsamples were used for bioconcentration experiments after the stationary growth stage had been reached.

Monitoring of algal density during culturing, CB and PCB bioconcentration, and elimination

Dry weight (DW; mg/L) was measured by drying at 105°C until constant weight was reached. Cell numbers and particle size distribution were measured with a Coulter Multisizer II (Coulter Electronics, Mijdrecht, The Netherlands). Total organic carbon (TOC) was measured in unfiltered samples (mg/L) and in 0.45 μm filtrated samples (DOC; mg/L) with an OIC (College Station, TX, USA) Model 700 TOC analyzer. Particulate carbon concentrations (POC; mg/L) were calculated from TOC and DOC by differencing. Organic carbon fractions of the algal cells were calculated as POC/DW. Comparability of subsamples was assured by using a Retsch (Dassel, Germany) suspensions divider.

Chemicals

1,2,3,4-Tetrachlorobenzene (TeCB; 98%) and pentachlorobenzene (QCB; 98%) were obtained from Aldrich (Zwijndrecht, The Netherlands); hexachlorobenzene (HCB; >98%) was obtained from BDH Chemicals (Poole, UK), and 2,4,4′-trichlorobiphenyl (PCB28), 2,2′,5,5′-tetrachlorobiphenyl (PCB52), 2,2′,4,5,5′-pentachlorobiphenyl (PCB101), 2,3′,4,4′,5′-pentachlorobiphenyl (PCB118), 2,2′,3,4,4′,5′-hexachlorobiphenyl (PCB138), 2,2′,4,4′,5,5′-hexachlorobiphenyl (PCB153), and 2,2′,3,4,4′,5,5′-heptachlorobiphenyl (PCB180) were obtained from Dr. Ehrenstorfer (Augsburg, Germany). Nanograde organic solvents (acetone, diethyl ether, 2,2,4-trimethylpentane) were obtained from Promochem (Wesel, Germany) and from Mallinckrodt (Paris, KY, USA).

Bioconcentration experiments

Quadruplicate BCFs were determined for TeCB, HCB, PCB28, PCB52, PCB101, PCB118, PCB138, PCB153, and PCB180 at the following conditions: (1) M. aeruginosa, 10°C; (2) S. capricornutum, 10°C; and (3) S. capricornutum, 20°C. The algal cultures (10 L) were spiked with 200 μL of a solution of each test compound in acetone. To avoid saturation with respect to the HOC aqueous solubilities, the spike solution was added to the continuously stirred cultures in several equal portions over a 5-h period. The spiked suspensions were incubated for 5 or 6 weeks at 10 ± 1°C or at 20 ± 2°C. At incremental time intervals, triplicate 0.6-L subsamples were taken for monitoring algal density, size, growth, and concentrations of the test compounds in/on algal cells. After equilibration, aqueousphase HOC analysis in 0.45-μm filtrates (glass fiber, Whattmann GF/C, Maidstone, UK) were performed. After the bioconcentration experiments, gas purge–induced desorption experiments (n = 3) were performed immediately.

Gas purge experiments

General explanation. The gas purge method involves bubble generation in water with HOCs (determination of Henry's law constant) or in the algal suspension incubated with HOCs (elimination studies). If the bubbles remain in solution for a sufficient period of time, HOC transfer from the water to the bubbles will occur until equilibrium is attained. With algae, the depletion of HOCs in the aqueous phase causes HOC elimination from the cells. The resultant HOC concentration in the gas leaving the headspace then depends on Henry's law constant, as measured in controls without algae, and on the HOC elimination rate (from the cells). The HOC concentration in the gas can be measured with a trap.

Gas purge procedure. The purge procedure was described earlier [10], but it is briefly explained here. Three 0.8-L subsamples from equilibrated cultures were placed in a thermostatted (10 ± 0.1°C), glass-stoppered, 0.9-L bottle. The HOCs were purged onto 11 to 14 Tenax traps (40–60 mesh, Chrompack, Bergen op Zoom, The Netherlands) in 5 or 6 weeks at a flow of 0.5 L/min under continuous stirring. The gas flow (constant within 2%) was adjusted with flow controllers and was passed through vessels containing water to prevent volume reduction in the algal suspensions. Tenax traps (trap efficiency, 100%) were replaced at incremental intervals, and these consisted of glass tubes containing glass wool and approximately 0.2 g Tenax. After purging, HOCs in the algae were analyzed so that we could account for remaining HOC fractions. Mass balances were calculated from the amounts of test compound that were found in the Tenax traps, in the algae, and on the container walls after purging based on the amount initially added to the systems. Before and after purging, a full characterization, including DW, DOC, and TOC, was performed.

For the determination of Henry's law constants, the HOCs were purged from Nanopure water at 10 ± 0.1°C. Spike volume was 5 μL. Equilibration and purge time were 1.5 h and 5 d, respectively.

HOC cleanup and analysis

HOCs were isolated from the filtered algae by multiple extractions with acetone. Combined acetone extracts were shaken with diethyl ether. Acetone was removed by extractions with water. The organic fraction was separated, and the aqueous phase was extracted again with diethyl ether. Both diethyl ether fractions were combined and were shaken again with water. The organic fraction was separated, dried with potassium sulphate (p.a.), concentrated to 5 ml on a Kuderna-Danish apparatus, (Wheaton Science Products, Millville, NJ, USA) and finally concentrated to 1 ml under nitrogen. The resultant solution was cleaned by eluting with diethyl ether over deactivated alumina. The extract was concentrated to 2 ml after addition of 2 ml 2,2,4-trimethylpentane with QCB as an internal standard.

Hydrophobic organic compounds were isolated from water samples by extraction with diethyl ether. After drying with p.a., the samples were concentrated and cleaned, as described above. Hydrophobic organic compounds were desorbed from container walls by thorough rinsing with acetone. Acetone fractions were extracted, concentrated, and cleaned as described above. Hydrophobic organic compounds were extracted from Tenax traps with 15 ml diethyl ether (100% recovery). The extracts were concentrated to 2 ml under nitrogen after addition of 2 ml 2,2,4-trimethylpentane with QCB as an internal standard.

A Hewlett-Packard (Avondale, PA, USA) 5890 double-column gas chromatography system equipped with two ⁶³Ni electron capture detectors was used for CB and PCB quantification. Further details of the analytical procedure were described in previous studies [6, 10].

Data analysis

Recoveries were calculated for the S. capricornutum and M. aeruginosa accumulation experiments. After incubation, bioconcentration factors on a DW basis (BCF^batch_DW), were calculated from solid phase and aqueous concentrations. The purge data were analyzed using first-order compartment models, which are briefly explained below. For a full discussion of these models, we refer to our earlier publications [6, 10]. This analysis also yields DW-based BCF (BCF^GP_DW) and kinetic data.

The gas purge–induced elimination of HOCs from the aqueous phase is described as an irreversible first-order volatilization process [13]

(1)

in which t is time (d), C_W (μg/L) is the aqueous phase chemical concentration referred to aqueous phase volume V_w (L), k_v is a first-order volatilization rate constant (per day), F is the gas flow (L/d), R is the gas constant (atm·m³·mole¹·K⁻¹), T is the absolute temperature (K), and H is Henry's law constant (atm·m³·mole⁻¹). When the volatilizing chemical is trapped on Tenax columns, C_W can be calculated from the cumulative trapped amounts, Q^TENAX, using

(2)

in which Q_t=0 (μg) is the amount of chemical initially present in the system. Values for k_v were calculated by fitting the cumulative mass purged from solution to the integration of Equations 1 and 2 [15] and were then converted to H values (Eqn. 1).

The transport to and the desorption from algae at low growth rates can be modeled as a one-compartment, first-order process

(3)

in which C_A (μg/L) is the algal phase chemical concentration referenced to suspension volume and k₁₂ and k₂₁ are uptake and elimination rate constants (per day). The course of the aqueous concentration during gas purging is given by

(4)

The dimensionless BCF* can be calculated from Equation 3 at steady state

(5)

BCF* quantifies the bioconcentration with respect to the true aqueous concentration and may be normalized to DW or organic carbon concentration of the algal suspension. The analytical solution to the Equations 3 and 4 was reported by Hasset and Millicic [17]. If the rate of gas purging is sufficiently slow such that equilibrium between algae and water is maintained, then the above three-parameter model (k₁₂, k₂₁, k_v) simplifies to a two-parameter model k₁₂/k₂₁ = BCF^GP_DW, k_v [17]

(6)

in which C_T (μg/L) is the sum of aqueous and sorbed chlorobenzene concentrations (C_T = C_A + C_W), and BCF^GP_DW, k₁₂/k₂₁) is the bioconcentration factor on a DWbasis, as determined by gas purging. If k_v (or H) and DW are known, Equation 6 allows the calculation of BCF^GP_DW. Parameters were estimated DW by nonlinear least-squares regression using the analytical solutions to the models. In some cases, the amounts of PCB153 and PCB180 purged from suspension were too low to enable subsequent parameter estimation. The amount of test compound in the system at t = 0 (Q_t=0) was calculated by adding the trapped amounts of the test compound to the amount found in the adsorbent after the purge experiment.

Algal density

Because of the low temperatures, average growth rates in doublings per day, based on cell numbers, were low during batch incubation: 0.002 ± 0.003 per day for M. aeruginosa at 10°C, 0.0036 0.009 for S. capricornutum at 10°C, and 0.005 ± 0.004 for S. capricornutum at 20°C. Characteristics of the cultures are summarized in Table 1. During gas purging, no losses of cells were observed for S. capricornutum at both temperatures. However, the density of the M. aeruginosa culture gradually decreased during gas purging. The DOC concentrations indicate significant exudation for S. capricornutum at 10°C. Particle size distributions remained similar in all cases except for that of S. capricornutum at the end of the incubation, during which a second peak occurred at 3.5 μm.

Henry's law constant determinations

Henry's law constant values at 108C were calculated from the cumulative mass purged from solution, except in the cases of PCB118 and PCB180. For PCB118 and PCB180, the mass purged was too small to allow direct fitting to Equations 1 and 2. Following Murray and Andren [23], the H values for PCB118 and PCB180 were calculated directly as the ratio of gas-phase concentration at time t = (Q^TENAX/[F.t]) and at initial aqueous concentration, assuming that the aqueous concentration did not change significantly during the first hours of purging. Estimated H values are presented in Table 2.

Uptake of chlorobenzenes and PCBs

The total recovery of individual homologues varied from 50 to 100%. Balances were less complete for the least and for the most hydrophobic chemicals. Uptake was monitored to check whether equilibrium had been reached, and it was not the subject of detailed study. The graphs of the uptake data are available on request. For all three experiments, accumulation was rapid in the first week. For S. capricornutum (20°C) and M. aeruginosa, no further uptake occurred. Selenastrum capricornutum cells at 10°C still accumulated some PCBs in the last week of the experiment.

Bioconcentration factors

Apparent BCFs were calculated from algal and aqueous phase measurements after 33, 34, or 36 d (see Table 1). The log BCF_DW−log K_ow curves (Fig. 1A), as determined by batchwise equilibration, were nonlinear and had a maximum at log K_ow = 6.5 to 7. No clear difference was found for the log BCF values measured at 10 and 20°C (Fig. 1A). Note that no decisive conclusions regarding a temperature effect can be drawn. The differences in apparent BCF values were also influenced by differences in DOC concentration and probably by the degree of equilibration.

Gas purge–induced elimination

The removal curves were fitted to the first-order one-compartment model and are presented in Figures 2 and 3 (upper curves). For the experiments at 10°C, Equation 6 fitted the experimental data very well (Fig. 2A and B and Fig. 3, upper curves). At 20°C (Fig. 2C), the fit for the more hydrophobic PCBs is slightly worse.

The BCF^GP_DW values determined by gas purging (Eq. 6) are presented in Figure 1B. Because of the low volatility of the higher chlorinated PCBs, data for compounds with log K_ow > 7 were not available for this graph, so the curvature seen in Figure 1A is not observed. Figure 1B shows that the difference in bioconcentration between the two species is reduced. If binding to algal exudates is significant, the reduction may be explained by the fact that theBCF^GP_DW values relate to the free aqueous HOC concentrations. The relationship of the pooled log BCF values to the log K_ow values up to log K_ow = 7 is fairly linear (r² = 0.743; p = 4 × 10⁻⁶; n = 17).

Henry's law constant

Estimated H values were compared to literature data in Table 2 [14-16, 20, 24]. The results are similar to those presented by Ten Hulscher et al. [16]. For the PCBs, the general decrease with increasing molecular weight and the decrease of H with the decreasing degree of ortho-chlorine substitution (e.g., PCB118 vs PCB101) agree with previous reports [15].

Uptake of chlorobenzenes and PCBs

The mass balances for the uptake experiment were less complete for the least and the most hydrophobic chemicals. The same pattern and ranges of recoveries were reported by Stange and Swackhamer [4], who presented a discussion on possible sinks in this type of experiments. Following their arguments, we explain that the low recoveries for the more volatile and less adsorbing chlorobenzenes and PCBs are likely the result of evaporation losses during sampling and gas leakage. Poor recoveries for the more hydrophobic PCBs may relate to the stronger associations with the container walls. Incomplete recoveries did not interfere with the data interpretation of BCF, as they were based on measured concentrations in the media and the algae rather than on a procedure that required us to measure one and calculate the other by difference.

The observed rate of uptake (data not shown) is similar to that of earlier reports. Chlorobenzenes and PCBs have been found to reach equilibrium over a time period of days [6, 10] or weeks [4, 5], respectively. The net PCB uptake by S. capricornutum cells at 10°C may have been retarded by an increase in the elimination of PCBs due to exudation of DOC. For the S. capricornutum experiments, the PCB180 concentration decreased after the first week. Among the test compounds, PCB180 had the highest affinity to DOC or algal exudates [9, 25]. For this very hydrophobic compound, the enhanced elimination may have overcompensated the uptake.

Bioconcentration factors

The log BCF_DW−log K_ow curves (Fig. 1A), as determined by batch equilibration at 10°C, agree closely with those previously reported by Stange and Swackhamer [4]. First, they found a maximum log BCF at log K_ow 6.5 for S. capricornutum, which is comparable to our values for the same species. Second, they found a slight increase of log BCF with log K_ow at log K_ow > 6.5 for the cyanobacterium Anabaena, as observed for the cyanobacterium M. aeruginosa in the present study. Third, they found relatively low log BCF values for the cyanobacterium at log K_ow < 6.5, which is also observed here. The main difference is that our values are one-half a log unit higher. Two explanations for this may be offered. First, in the current study, incubation times were longer (33–36 d) as compared with 12 d in Stange and Swackhamer [4]. Second, in our experiments DOC levels were relatively low (Table 1) as compared with those in Stange and Swackhamer [4] (70–120 mg/L). This may have resulted in a larger bioavailability of PCBs in our experiments.

Gas purge–induced elimination

The good fit of the first-order one-compartment model (Fig. 2A and B and Fig. 3, upper curves) suggests that for both species, uptake and elimination kinetics at 10°C are sufficiently fast, with respect to the volatilization rate k_vC_w, to maintain equilibrium between HOCs in the algae and the water. At 10°C, this volatilization rate is very low because of low Henry's law constants. At 20°C (Fig. 2C) the fit is worse but is not severe enough to make a two-compartment model statistically preferable. Although no alterations were detected visually, it is possible that the algal suspensions may have changed during gas purging. Besides physiological changes related to aging, such as changes in lipid content, exudates may have been formed (e.g., S. capricornutum) because of metabolic activity. For M. aeruginosa, gas purging removed cells from the bulk of the suspension. The excellent fit of the one-compartment model throughout the entire 5- or 6-week elimination period suggests that the changes, observed or not, had no significant effect on the overall sorptive properties of the population. During the M. aeruginosa experiments, cells appeared to accumulate in a liquid film on the inner surface of the flasks and near the outlet to the Tenax traps. No cells or cell fragments were detected in the Tenax traps themselves. We assume that the transport of the test compounds from this film to the Tenax traps occurred at the same rate as did transport from the bulk suspension. Hence, the amounts trapped in the Tenax were not significantly affected by the loss of cells.

The effect of slow accumulation kinetics on the apparent BCF

The aim of this study was to assess the possible influence of slow kinetics on the magnitude of the steady-state BCF for PCBs. First, minimum values for the uptake and elimination rate constants, which were still consistent with the results of the purge experiments, were calculated. Second, the maximum values of the rate constants at which the slow kinetics hypothesis was still consistent with the apparent BCF values reported in this study and in the literature were calculated. Finally, the calculated constants were compared. The calculations were performed on the data for PCB138 and PCB153 in the S. capricornutum experiment at 10°C. This experiment offered the greatest degree of comparability with literature data [4, 5] and showed no loss of cells.

Calculation of lower limits for rate constants from purge data. The fitting of Equation 6 yielded estimates of BCF^GP_DW that were equal to the ratio k₂₁/k₁₂. When the magnitudes of these rate constants were larger than certain minimum values, then the elimination from the algae was not rate limiting with respect to the known volatilization rate, and equilibrium was maintained. However, had the constants been lower than these minimum values, the elimination would have been rate limiting, and the continuous equilibrium state would not have been observed. Using Equations 3 and 4, the elimination of PCB138 and PCB153 from S. capricornutum cells was simulated for various magnitudes of k₂₁, keeping the ratio k₂₁/k₁₂ (=BCF^GP_DW) constant at the value obtained with Equation 6. The DW volatilization rate constants for PCB138 and PCB153 were approximately 1 per day. The numerical integration of the differential equations was carried out using the simulation software package TUTSIM (Meerman Automation, The Netherlands). The results (Fig. 3) indicate that the equilibrium binding situation found in our experiments implies k₂₁ values that are larger than 1.5 per day (PCB138 and PCB153) and k₁₂ values that are larger than 250 per day (PCB138) and 30 per day (PCB153). At lower k values, the observed equilibrium would not have occurred, and the model would no longer fit the data (Fig. 3). Equilibrium binding was also observed for the other PCBs. In our experiments, the k_v values had a constant ratio to the Henry's law constant and were all of the same order of magnitude. Therefore, a value of 1 per day is a safe minimum value for k₂₁ for all PCBs and species considered in this study.

Recently, Axelman et al. [26] found evidence from field data that algal cell size has an effect on bioconcentration rate constants. They estimated the effect of algal cell size on elimination rate constants using radial diffusion model concepts. Following their method (eqns. 1–7 in [26]), we obtained values for k₂₁ in the range of 10 to 50 per day for k₂₁ for the pentachloro- and hexachlorobiphenyls under our experimental conditions (10°C; cell diameter 4–5 μm; log K_oc = 7−7.5). These high values are consistent with the current results from our purge experiments.

Calculation of upper limits for rate constants assuming growth dilution. When algal growth is significant and when DOC effects play no role, Equation 3 should be extended to [1, 2, 6]

(7)

with k_G (per day) as the first-order rate constant for algal growth. This model has the following steady state solution [6]

(8)

in which BCF^G is the bioconcentration factor for growing phytoplankton, BCF^N is the bioconcentration factor at nongrowth conditions, equal to k₁₂/k₂₁, and the superscripted “e” refers to equilibrium concentrations. If k_G cannot be neglected when compared with k₂₁, then k_G affects the bioconcentration factor. For PCB138 and PCB153 (log K_ow = 6.8-7), the magnitude of the BCF^N underestimation for S. capricornutum can be estimated from Figure. 1A and appears to be at least one log unit. For this culture, a first-order growth rate of k_G 5 0.02 per day was measured. It follows from Equation 8 that a factor 10 reduction of BCF as a result of growth at a rate of 0.02 per day requires an elimination rate constant of approximately 0.002 per day.

For PCBs with hydrophobicity similar to that of PCB138 and PCB153 (log K_ow of 6.8-7.0), Swackhamer and Skoglund reported lipid-normalized log BCF_G values of approximately 10⁷ for Scenedesmus sp. at low growth rate (fig. 1 in ref [5]). For the same species and log K_ow range, they found log BCF^G values of approximately 106 at a growth rate of 0.13 per day (fig. 2 in ref [5]), which represents a factor 10 decrease of BCF. The same authors also report a factor 10 reduction of BCF in the 6.8 to 7.0 log K_ow range for a summer field collection of mixed algae (higher growth rate), as compared with a winter field collection (low growth rate). It follows from Equation 8 that a factor 10 reduction of BCF as a result of growth at a rate of 0.13 per day requires an elimination rate constant of 0.014 per day.

Recently, Skoglund et al. [27] evaluated laboratory- and field-derived BCF values for PCBs using a two-compartment kinetics model for bioconcentration in growing phytoplankton. In their experiments, DOC concentrations were considerably higher (S. capricornutum: 45–93 mg/L) and were acknowledged to affect the BCF values, but they were not taken into account in the data analysis because the quality of the DOC measurements was in question [27]. Elimination (depuration) rate constants for the slow accumulating fraction were less than 0.06 per day for all species studied and approximately 0.001 per day for S. capricornutum. These were calculated from equation 5 in Skoglund et al. [27], using uptake rate constants at log K_ow = 6.8-7 from figure 1 in Skoglund et al. [27], and using surface adsorption partitioning coefficients from table 2 in Skoglund et al. [27].

The above analysis shows that the hypothesis of slow bioconcentration kinetics requires very low rate constants and therefore cannot explain the curvature of log BCF−log K_ow plots for S. capricornutum and M. aeruginosa unicells. If elimination from S. capricornutum cells had been as slow as the rate constants suggested, no equilibrium binding would have occurred in our purge experiments. The k₂₁ value measured for PCB138 or for PCB153 by gas purging is at least factor 100 to 500 higher than the value derived from steadystate BCFs reported in Swackhamer and Skoglund [5]. The recent analysis by Skoglund et al. [27], which used a kinetics model, requires elimination rate constants that are at least three orders of magnitude lower than the minimum value of 1.5 per day. We therefore suggest that both the effect of growth on BCF and the curvature of log BCF−log K_ow plots may need another explanation.

The effect on the apparent BCF of binding to DOC

Several authors have proposed that the binding to DOC is a mechanism of reduced bioavailability of HOCs for aquatic organisms (e.g., [4, 27-29]). Analogous to the effect of DOC on sediment/water partition coefficients [30], the effect of DOC on the apparent BCF^batch_DW can be quantified by

(9)

Because DOC (kg/L) and apparent BCF^batch_DW values were measured in this study, the equation has two unknowns: the partition coefficient describing the compounds' equilibrium distribution between water and DOC (K_DOC; kg/L) and the bioconcentration factor for phytoplankton in the absence of DOC (BCF^true_DW; kg/L). Assuming that bioconcentration in algae primarily concerns partitioning into the lipid fraction [4, 5], BCF^true_DW can be assumed to be proportional to K_ow

(10)

The proportionality factor, a, is dependent on the lipid content and class and therefore may be different for different species and physiological states. Lara et al. [9] reported PCB K_DOC values for exudates of brown algae in seawater. The values were of the same order of magnitude as those for marine humic substances. For freshwater algal exudates, no K_DOC literature data are available. Recently, we reported PCB K_DOC values that were of the same order of magnitude as K_ow for extracellular carbon isolated from S. capricornutum cultures, using gas purge experiments [29]. Exudates may consist of fractions of low molecular weight DOC as well as colloidal DOC and surface active DOC [31]. Further, bacteria may contribute to the DOC pool in batch experiments. The magnitude of these fractions may vary considerably in time and among species and temperatures, so that the choice of any fixed value for K_DOC would be arbitrary. We therefore assume that K_DOC for a series of chemically similar hydrophobic compounds is proportional to K_ow

(11)

Combining Equations 9 through 11 yields

(12)

Equation 12 was fitted to the BCF^batch_DW values found for M. aeruginosa (DOC = 4.8 mg/L), S. capricornutum (DOC = 65.6 mg/L) at 10°C, and S. capricornutum (DOC = 3.9 mg/L) at 20°C. The nonlinear least-squares parameter estimation was performed with the optimization routine available in Quattro Pro (R5.0, Borland International). The model fits are shown in Figure 4. An overview of model parameters and the ANOVA of the nonlinear regression is presented in Table 3. For M. aeruginosa, a satisfactory fit is found for α = 0.05 and for β = 0.02. The value for α can be interpreted as a lipid content of 5%, which is a plausible value for this species [32]. Karickhoff et al. [33] reported a value of β = 0.63 for sedimentary organic carbon (K_OC = 0.63 K_ow). The value of 0.02 for exudates is 30 times lower and therefore confirms their rather hydrophillic nature. The value of 0.02 is in agreement with data on the association constants of PCBs for humic materials [25]. For S. capricornutum, Equation 12 also satisfactorily follows the trend, but it overestimates the BCF for PCB180. Note that Equation 12 only accounts for the general trend in hydrophobicity. The lack of fit for individual congeners may be explained by properties affecting the partition constants, which are not accounted for by the model. Besides hydrophobicity, the factors that may influence bioconcentration factors or DOC association constants are the degree of orthosubstitution, stereochemistry, planarity, or total surface area [25, 34, 35]. Further, for the more hydrophobic PCB congeners, the traditional K_ow may be an inappropriate parameter to describe the actual hydrophobic interactions with algal lipids [26]. As the species and dissolved macromolecules are different, different values for α and β are also found for S. capricornutum (Table 3). The α values are rather high and suggest that lipid normalized BCFs may be an order of magnitude higher than K_ow. This has also been noted in earlier reports [4, 6, 27]. At 10°C, the value for β that quantifies the curvature of the log BCF−log K_ow plot is similar to the β value for M. aeruginosa, suggesting that the exudates of the two species at 108C have similar affinities for HOCs. The higher value for β at 20°C may relate to differences in DOC properties, (e.g., a different contribution of bacteria and exudates to the total DOC pool). From the results in Table 3 we conclude that the hypothesis that there is significant binding of PCBs to DOC is consistent with our current results. No conclusions on the effects of temperature or species and DOC type on BCF^batch_DW values can be drawn.