3.1. Concentrations of Organic Contaminants
shows the concentrations of each contaminant group in the different lakes. The results are expressed as sum of 12 congeners of PFAS (ng g−1
ww), sum of the two congeners and four respective metabolites of DDT (ng g−1
dw) and sum of 14 congeners of PCB (ng g−1
dw) in zooplankton samples. Detailed data regarding contaminants concentrations are reported in Tables S4–S6
Differences between lakes were statistically significant (Anova one-way, p-value < 0.001 for all compounds; N = 51 for PFAS, N = 72 for OC). In detail, PFAS concentrations in zooplankton from Lake Iseo were lower than those in Lake Como and Maggiore (Tukey test, p-value < 0.001), which showed no differences between them; DDT concentrations in Lake Maggiore were significantly higher than in the other lakes (Tukey test p-value < 0.001) and PCB levels in zooplankton in Lake Como were lower than in the samples that were collected in Lake Maggiore and Iseo (Tukey test p-value < 0.001).
In general, Lake Maggiore showed the highest concentrations for each group of contaminants, with mean values of 7.6 ng g−1 ww for PFAS, 65.0 ng g−1 dw for DDT, and 65.5 ng g−1 dw for PCB.
High levels of DDT and its metabolites in this lake are due to the presence of a point source from a chemical plant located on the River Toce, an important tributary of Lake Maggiore, as already explained in the section “Study Area”. The factory produced technical DDT from 1948 to 1996, but the contamination is still present, because these compounds accumulated into the soils around the industrial area [21
There are not factories that produce PFAS in this area, but lakes are subjected to the effluents of both industrial and urban wastewater treatment plants (WWTPs) and to diffuse pollution from atmospheric deposition. PFAS are not removed in standard treatments of wastewater and enter in water bodies [31
]. The basin of Lake Maggiore is characterized by the most extended area, the highest number of inhabitants, and the highest percentage of densely populated area (2%) among the basins of the studied lakes in accord to the European report on degree of urbanization [26
] (Table 1
). The same ranking of PFAS contamination has been highlighted in fish that were sampled in the same areas [32
], and in that work the source of PFAS for Lake Maggiore was hypothesised to be Lake Lugano, which belongs to the Lake Maggiore basin. Lake Iseo, which collects the waters of the smallest basin with the lowest population and number of WWTPs, showed the lowest PFAS concentrations among the studied lakes (mean value: 3.2 ± 5.7 ng g−1
It is more difficult to address the differences in PCB zooplankton concentrations, because the contamination is very old, and no point sources can be identified in the lake basins. In fact, the differences in concentrations cannot be directly related to the basin areas or the inhabitant number. Nonetheless, Lake Iseo has a significantly higher mean concentration of total PCB than Lake Como (40.6 ± 40.1 and 20.9 ± 21.1 ng g−1 dw, respectively), and this result could be linked to the great exploitation of hydroelectric power plants in Valcamonica, during the economic development after the second World War, which largely used PCB as dielectric fluids in transformers.
Regarding Italian subalpine lakes, this is the first study of PFAS contamination in zooplankton, while data regarding DDT and PCB are abundant for Lake Maggiore [33
], but sporadic for the other lakes. The last determination of DDT and PCB concentrations in zooplankton in Lake Iseo, dating back to 2010, showed that current DDT concentrations are lower, while PCB concentrations are stable [34
]. In Lake Como, the comparison with older data showed a decrease in the concentrations of both organo-halogenated compounds [11
], but there are not enough data to claim a significant decreasing trend.
Concentrations of PFAS in pelagic invertebrates in this study are higher than those that are reported in Baltic Sea [36
], where the sums of PFSA and PFCA in zooplankton were only 0.11 ± 0.02 and 0.12 ± 0.01 ng g−1
ww, respectively. On the contrary, they are comparable with the concentrations that were measured in the Gironde estuary (France) [37
] and in the Arctic Canadian Lakes that are not contaminated by local airport [38
3.2. Pattern of Contamination
When considering the composition pattern of PFAS accumulated in zooplankton (Figure 3
), PFOS was detected in 96% of the samples and it was the predominant compound in all lakes, reaching the maximum concentration of 18.9 ng g−1
ww in Lake Maggiore. It represented 32% of total PFAS concentrations in zooplankton in Lake Iseo, 52% in Lake Como, and 67% in Lake Maggiore. The other two perfluoroalkyl sulfonic acids detected (PFBS, PFHxS) were only determined in significant concentrations in Lake Iseo. Regarding perfluoroalkyl carboxylic acids (PFCA), long-chain compounds (C > 9) predominated in zooplankton, while short-chain compounds (with 6–7 carbon atoms) were only detected in few samples (about 10%) and at lower concentrations (maximum value: 0.9 ng g−1
ww). PFOA was detected in about 65% of samples, but it only represented 11.5% of the total PFAS concentration in Lake Iseo, and about 6% in the other two lakes.
Concentrations of PFTrDA (C13) and PFTeDA (C14) were lower than those of other long-chain PFCA, probably because these compounds have higher affinity for particles and sediment is their main sink [39
We analysed the whole dataset of individual congeners of PFAS by a Principal Component Analysis (PCA) (Figure 4
). Loading plot on the first two components, which globally explain 54% of the total variance, helps to identify common behaviour among the individual PFAS congeners (Figure 3
). Three different groups are gathered in the loading plot: PFOA, PFBS, and PFHxS compose the first, which is maximum on the second component and orthogonal to the first one. PFOA shows Kow
similar to PFHxS [40
], and this group of compounds was higher in the samples of Lake Iseo than in the other lakes, which suggests a specific contamination source for this lake. The second group is formed by PFTrDA (C13) and PFTeDA (C14) and it is orthogonal to the first component and parallel to the second one, but in the negative direction. The third group showed PFOS (C8) laying in the same direction of the other long-chain PFCA (8 < C < 13); it is rather orthogonal to the other two groups and it includes the most bioaccumulable and biomagnificable PFAS congeners. In fact, PFBS and PFHxS are the most soluble congeners and PFTrDA (C13) and PFTeDA (C14) are not readily bioavailable because of their molecular size [41
]. The coefficients in the second eigenvector are correlated with Kow
of PFAS substances [40
], except for PFOA, which is uncorrelated (Figure S2
), suggesting that lipophilicity cannot be used to model bioaccumulation of PFOA in zooplankton.
Looking at the bidimensional score plot, lakes Como and Maggiore samples cannot be distinguished, while the Lake Iseo data are better described by the second component where PFOA, PFBS, and PFHxS loadings predominate.
In the group of DDT compounds, metabolites of DDT and their isomers were predominant over the parental compounds (op’ DDT and pp’ DDT). Technical DDT products generally contained about 75% of pp’ DDT, 15% op’ DDT, and other compounds in very small amounts. DDT isomers are known to degrade into DDE and DDD under aerobic and anaerobic conditions. Therefore, the increase of the percentage of DDE and/or DDD and a > 1 ratio DDE/DDT indicated that there are no recent inputs to the environment [43
]. DDE represented more than 40% of the total concentrations in all lakes and its ratios with DDT were 2.4, 5.3, and 2.3 for Lakes Maggiore, Iseo, and Como, respectively, suggesting that the contamination is old and no recent inputs of parental compound occurred (Figure 3
). pp’ DDE was the main compound detected in zooplankton and it was measured in all samples with concentrations that ranged from 0.3 ng g−1
dw in Lake Como to 38.3 ng g−1
dw in Lake Maggiore (Table S5
PCB 153 was the congener with the highest frequency of detection (>94%), followed by PCB 101 (91.5%), PCB 44, PCB 180, and PCB 138 (all up to 70% of total samples). In Lake Maggiore, PCB 153 was the congener with the highest concentrations (11.0 ± 8 ng g−1
dw), while PCB 149 prevailed in Lake Como (6.6 ± 12 ng g−1
dw) and PCB 52 in Lake Iseo (11.0 ±18 ng g−1
dw). If we grouped PCB congeners in seven classes based on their number of chlorine atoms, concentrations raised with the increase of number of chlorine atoms until the hexachlorobiphenyl (hexa-CB) group, and then tended to decrease (Figure 3
). Accordingly, the prevalent group was hexa-CB, which constituted 35.4% of total PCB concentration, reached a maximum of 60.5 ng g−1
dw in Lake Maggiore. The pattern of PCB congeners probably reflected the Aroclor mixtures (Aroclor 1256 e 1260) most used in the past in Italy [44
While examining loading plot in the PCA of PCB and DDT compounds, gathered in isomer groups (Figure 5
), we can see that the coefficients of PCB and DDT isomer groups in the second component are significantly correlated with Kow
), except for octachlorobiphenyl (octa-CB). The peculiar octa-CB behaviour cannot be easily explained, but it could be related to the fact that octa-CBs were only determined in Lake Maggiore. As in the case of PFAS, the second component is related to the contaminant lipophilicity and explains 12% of the total variance. Nevertheless, it should be noted that the slope of the correlation between coefficients in the second eigenvector and Kow
of PCB and DDT is five-times higher than that interpolated for PFAS (Figure S2
The score plot shows that data from different lakes cannot be distinguished, but Lake Maggiore has the highest variability, while the lowest one is shown by Lake Como, as also evident in Figure 2
3.3. Role of Zooplankton Size and Seasonality on Contaminant Levels
Data that were collected in this study allowed for in-depth insight of the role of zooplankton ecology in the contaminant accumulation. Zooplankton has been sampled in different size fractions in order to separate species that are characterized by different trophic levels. Details on size fractions collected in the different lakes can be found in Tables S2 and S7
reports data on biomass and taxa composition.
The smallest and the intermediate fractions (≥200 and ≥450 µm) included all crustacean species living in the lakes but had different total biomass, because, in the former, we could collect also the smallest and youngest specimens, having a more complete picture of the zooplankton community. The greatest size fraction (≥850 µm) mainly contained the biggest individuals of Cladocera (generally Daphnia for primary consumers and predators).
We only analysed zooplankton data from Lakes Como and Maggiore, because, for Lake Iseo, there were enough data for the lowest size fraction (200 µm), but the total sampled biomass for the other two fractions was insufficient to complete all the chemical analyses. PFAS data have been analysed as a whole dataset. Since we have shown that there are no statistically significant differences between Lake Como and Lake Maggiore for PFAS data (Figure 2
), while for DDT and PCB, the datasets have been separately analysed for each lake (Figure S1
No significant differences were observed between zooplankton size fractions for all contaminants (Figure S1
). According to a biomagnification hypothesis, the biggest fraction, which contains more predators than filter-feeder or herbivores crustaceans, should be the preferred fraction for contaminant accumulation. On the contrary, our results showed that the biggest fraction had no statistically significant differences with the others, and the 850 µm-fraction was clearly less contaminated than the 450 µm-one for DDT and PCB in Lake Maggiore. Piscia et al. [45
] suggested that in the smaller fractions there were more copepods, richer in lipids than cladoceran species, and therefore more available to bioaccumulate organic contaminants. Principal Component Analysis of taxonomic compositions and contaminant concentrations, expressed as total concentrations of each chemical family, (Figure 6
) showed that chemical concentrations were orthogonal to (i.e., independent from) the taxa of planktonic organisms, but inversely correlated with the total zooplankton biomass and temperature. The score plot showed that the colder seasons (autumn and winter) positively correlated with all of the contaminant concentrations in zooplankton.
This result is partially confirmed by comparing concentrations in the different seasons (Figure 7
), which shows a similar qualitative trend for all compounds: the concentrations were higher in colder months than in spring and summer, with a characteristic U-shape from winter to autumn. For PFAS, these differences were not statistically significant, while winter DDT concentrations in Lake Como were significantly higher than spring ones (p-value
< 0.05, Anova and Tukey tests), and, in Lake Maggiore, there were significant differences between winter and both warmer seasons and between autumn and summer. PCB followed the same trend as DDT and both lakes showed significant differences between seasons: in Lake Como (p-value
< 0.01), there were significant differences between winter and both warmer seasons and between autumn and spring; in Lake Maggiore (p-value
< 0.05) there were significant differences between summer and colder seasons. No interaction between the considered variables (size and seasons) was evidenced by two-way-Anova test.
The characteristic U-shaped trend of DDT and PCB concentrations in Lake Maggiore was observed since the beginning of the monitoring activities and it did not vary between years [33
]. The inverse relationship between concentrations in zooplankton and zooplankton biomass might be associated with the shift in diet of zooplanktonic specimens because of the different availability of nutrient along the year. Changes in resource availability and environmental conditions (the decrease of food availability or the increase of metabolic costs) can lead to changes in trophic interactions [46
]. For example, δ15N‰ of all zooplanktonic species changed along the years, increasing in the cold seasons, as shown in [35
]. During spring and summer, phytoplankton is easily available and filter feeders rely on this food source, while, during autumn and winter, they need to eat also bacteria, protozoa, or organic particles to obtain enough energy to live. Additionally, Campbell et al. [47
] observed that organisms, which live in cold water from glaciers in an unproductive environment and low nutrients, often become richer in lipid and OC content, indicating that nutrient limitation at the base of the food web can affect the uptake of contaminants at higher trophic levels.
The differences of concentrations throughout the year might be also explained by “the biomass dilution effect”, as proposed by Taylor et al. [6
], who observed that DDT and PCB concentrations varied across lakes according to an inverse relationship with their planktonic biomass. The same effect, as observed for polycyclic aromatic hydrocarbons in plankton of the Mediterranean and Black Seas, was explained by a reduction of water concentrations by adsorption on dissolved organic matter and suspended sediments that peak during summer algal bloom [48
In Italian lakes, which were studied in the present work, the seasonal trend was much stronger for the chlorinated compounds than for PFAS. Variations in PFAS were quite limited, as in the Gironde estuary, where PFAS only varied up to a factor of 2.5× for zooplankton and 2.3× for shrimps in different seasons [49