6 AIMS OF THE STUDY
8.2 Plastic types and particle sizes
In all studied environments, the proportions of plastic types of MPs roughly followed the production and use rates of plastic types (PlasticsEurope, 2019). PE, PP and PET are the most produced plastic types, and they were the most common plastic types in the studied samples independently of the sampled environment or matrix (Figure 9). However, lake water samples were analysed with different methods than other samples, which can affect the results. The most prominent difference was high number of PAN fibers in the lake water samples, compared to the other samples.
Moreover, lake waters did not contain any PA, but the sample pre-treatment method was suspected to decompose it.
Figure 9. Percentage distributions of plastic types in different environments and matrixes.
The lake water samples in publication I were analysed with the method, which included visual selection of particles, instead than the sea water and fish samples in
0 % 10 % 20 % 30 % 40 % 50 % 60 % 70 % 80 % 90 % 100 %
Lake water Sea water Fish
PE PP PET PA PAN PS PMMA PVC ABS
have found in the Central Baltic Sea that MPs accumulate to certain depths of the water column, according to the stratification.
To compare the results to the surface waters of the studied area, concentrations of
>100 µm MPs were similar or lower in the vertical transects than in the surface waters, where the concentration was 0–6.8 MPs/m3 (Setälä et al., 2016). However, the concentrations of >50 µm MPs were remarkably higher in haloclines and thermoclines than concentrations in the surface water or water column close to the seashore, where it was on average 0–22 MPs/m3 (Railo et al., 2018). All the evidence indicates that in stratified waters, MPs do accumulate to the layers where density changes rapidly.
In addition to the water samples, MPs ingestion in fish was studied in Lake Kallavesi by quantifying >20 µm MPs in fish GITs. Vendace contained on average 25 ± 50 MPs/fish, and perch 11 ± 16 MPs/fish (Table 7). The deviation between the samples was rather high, arising from the finding that some fish individuals had ingested numerous MPs, while the most had ingested very few MPs. The estimated mass fractions of MPs per fish mass were 0.37 ± 1.0 µg/g for vendace and 0.38 ± 0.67 µg/g perch. The count of ingested MPs was not statistically different between species.
Similarly, the share of fish individuals who had ingested MPs did not differ significantly between species: 25% of vendace and 17% of perch had ingested MPs.
The difference between sampling sites was non-significant for vendace, which probably move around the lake and do not represent the sampling site, but the lake generally. Contradictory, the difference between sampling sites was significant for perch, and it indicated that the perch from one sampling site out of five had ingested MPs, but perch from other sites have not. The reason for the difference could not be resolved, but the site where ingestion mainly happened was close to a construction site, where houses were built close to the lakeshore. That could be one source of MPs to the lake.
McNeish at el. (2018) have suggested that both the concentration of MPs in the environment and feeding behaviour of fish affect the ingestion of MPs, and Roch et al. (2019) have suggested the uptake to be passive or accidental. In Lake Kallavesi, Perch and vendace did not ingest significantly different amounts of MPs, although their feeding behaviours differ. Moreover, although perch have similar feeding behaviour, they ingested different numbers of MPs in different sites. The results support the suggestion that multiple factors affect the ingestion of MPs by fish, or it is accidental or coincidental. Comparable studies about northern freshwater fish does not exist, but in Germany 16.5% of lake fish had ingested >40 µm MPs (Roch et al., 2019). Considering the higher particle size limit, the percentage of ingestion per individuals is on the similar range in this study.
The water samples from Lake Kallavesi contained on average 169 MPs/m3 with particle size >50 µm. From fish, all >20 µm particles were analysed. Because of different analysis methods and particle sizes, the values cannot be unambiguously compared, but fish contained approximately the same numbers of MPs than 0,1 m3 = 100 L water.
8.2 PLASTIC TYPES AND PARTICLE SIZES
In all studied environments, the proportions of plastic types of MPs roughly followed the production and use rates of plastic types (PlasticsEurope, 2019). PE, PP and PET are the most produced plastic types, and they were the most common plastic types in the studied samples independently of the sampled environment or matrix (Figure 9). However, lake water samples were analysed with different methods than other samples, which can affect the results. The most prominent difference was high number of PAN fibers in the lake water samples, compared to the other samples.
Moreover, lake waters did not contain any PA, but the sample pre-treatment method was suspected to decompose it.
Figure 9. Percentage distributions of plastic types in different environments and matrixes.
The lake water samples in publication I were analysed with the method, which included visual selection of particles, instead than the sea water and fish samples in
0 % 10 % 20 % 30 % 40 % 50 % 60 % 70 % 80 % 90 % 100 %
Lake water Sea water Fish
PE PP PET PA PAN PS PMMA PVC ABS
publications II and III. Therefore, the results may not be unambiguously comparable.
Because publication I did not include as wide quality control measures as the others, it is probably the most biased study in this thesis. It did not include blank or recovery samples for sampling, pre-treatment, or microscopy and FTIR analysis. Visual selection of particles can be highly subjective and biased towards MPs of for example certain shape, size, material, or colour. Recovery rate tests could have revealed biases.
Moreover, because blank samples were prepared only for microscopy and FTIR, the potential contamination during sampling and pre-treatment was not assessed.
Therefore, results in publication I can be over- or underestimated for certain MP types. If all publications in this thesis had similar QC measures, the accuracy of both methods could be assessed, and the results could be more comparable.
The high number of PAN fibers in lake water can be caused by textile-based contamination, because the probability for contamination was not measured during sampling or pre-treatments. Similarly, half of the lake water samples were filtered through a tube made from PVC, which was not tested for contamination. Possible contamination can explain why PVC is more common than in other sample types.
However, publication I included the most inclusive spectral libraries, while the libraries in publication III were composed based on the preliminary knowledge from the lake water samples. Especially in publications II and III, the reduced number of library spectra could have biased the distribution of polymers, if material that was not included in the libraries was common in samples.
Particle sizes fractions were determined in the lake water samples according to the pore sizes of the filters. Therefore, they are not exact measured values. Because particles can aggregate and stick to the filters, fraction filtration does not necessarily capture particles with corresponding size. Moreover, because the sampled volume was different for each filter, distribution of particle sizes cannot be presented as percentages, but is presented as MPs/L in Table 8.
For the sea water and the fish samples, particle sizes were determined from the FPA-FTIR data as the longest dimensions of particles. Sea water samples were sampled with 100 µm net, or as bulk samples, which were filtered through 50 µm filters right after sampling. Therefore, the smallest (<50 or 100 µm) particles were not representative values. However, Table 8 shows the approximate distribution of particle sizes in different environments and sample types. It allows a rough estimation for abundance of MPs with different sizes in the aquatic environments.
For the fish samples, all particles larger than 20 µm were analysed. However, particle size distribution of MPs in fish does not necessarily represent the distribution in the water environment they live. Fish favour to prey zooplankton of specific sizes during their different life stages (Jacobson et al., 2019; Northcote and Hammar, 2006).
Because the studied fish were small, they were more likely to ingest small particles than very large ones. Moreover, fish can indirectly uptake MPs with food or water (Batel et al., 2016; Nelms et al., 2018). The indirect uptake could hypothetically favour small particles.
Table 8. Proportion of particle sizes of MPs in different environments and matrixes. Note the different units. *Not representative because of the methods.
Lake water
MPs/L Sea water
% Fish
%
20–50 µm - 31* 48
50–100 µm 0,15 33* 33
100–300 µm 0,012 31 18
>300 µm 0,0014 5 1
In the lake water samples (publication I), concentration of 50–100 µm MPs was hundred times higher than concentration of >300 µm MPs. In fish samples from the same lake (publication III), the difference was only 30 times. While small particles were notably common in the lake, the sea water samples contained higher proportion of larger >300 µm particles. The number of samples from both environments is limited, but based on this data, larger MP particles seem to be more common in marine environment than in lake. However, lake water was sampled from the surface, whereas sea water was from the haloclines and thermoclines and between them. From that point of view, small MPs may be more common in the surface, while larger ones in the water column. One possibly biasing factor is that different samples were collected, pre-treated, and measured with different methods. Further, because the method in publication I was validated less extensively compared to the methods in publications II and III, the results of publication I may be more biased compared to the results of other publications.
Generally, many studies have reported that small MPs are systematically more abundant than larger ones in various environments. For example Roch et al. (2019) have mathematically estimated that >95% of MPs in water and fish would be smaller than 40 µm. Similarly, Setälä et al. (2016) have found that 100 µm net collects more MPs than 300 µm in the surface waters of the Baltic Sea. Besides the similar
publications II and III. Therefore, the results may not be unambiguously comparable.
Because publication I did not include as wide quality control measures as the others, it is probably the most biased study in this thesis. It did not include blank or recovery samples for sampling, pre-treatment, or microscopy and FTIR analysis. Visual selection of particles can be highly subjective and biased towards MPs of for example certain shape, size, material, or colour. Recovery rate tests could have revealed biases.
Moreover, because blank samples were prepared only for microscopy and FTIR, the potential contamination during sampling and pre-treatment was not assessed.
Therefore, results in publication I can be over- or underestimated for certain MP types. If all publications in this thesis had similar QC measures, the accuracy of both methods could be assessed, and the results could be more comparable.
The high number of PAN fibers in lake water can be caused by textile-based contamination, because the probability for contamination was not measured during sampling or pre-treatments. Similarly, half of the lake water samples were filtered through a tube made from PVC, which was not tested for contamination. Possible contamination can explain why PVC is more common than in other sample types.
However, publication I included the most inclusive spectral libraries, while the libraries in publication III were composed based on the preliminary knowledge from the lake water samples. Especially in publications II and III, the reduced number of library spectra could have biased the distribution of polymers, if material that was not included in the libraries was common in samples.
Particle sizes fractions were determined in the lake water samples according to the pore sizes of the filters. Therefore, they are not exact measured values. Because particles can aggregate and stick to the filters, fraction filtration does not necessarily capture particles with corresponding size. Moreover, because the sampled volume was different for each filter, distribution of particle sizes cannot be presented as percentages, but is presented as MPs/L in Table 8.
For the sea water and the fish samples, particle sizes were determined from the FPA-FTIR data as the longest dimensions of particles. Sea water samples were sampled with 100 µm net, or as bulk samples, which were filtered through 50 µm filters right after sampling. Therefore, the smallest (<50 or 100 µm) particles were not representative values. However, Table 8 shows the approximate distribution of particle sizes in different environments and sample types. It allows a rough estimation for abundance of MPs with different sizes in the aquatic environments.
For the fish samples, all particles larger than 20 µm were analysed. However, particle size distribution of MPs in fish does not necessarily represent the distribution in the water environment they live. Fish favour to prey zooplankton of specific sizes during their different life stages (Jacobson et al., 2019; Northcote and Hammar, 2006).
Because the studied fish were small, they were more likely to ingest small particles than very large ones. Moreover, fish can indirectly uptake MPs with food or water (Batel et al., 2016; Nelms et al., 2018). The indirect uptake could hypothetically favour small particles.
Table 8. Proportion of particle sizes of MPs in different environments and matrixes. Note the different units. *Not representative because of the methods.
Lake water
MPs/L Sea water
% Fish
%
20–50 µm - 31* 48
50–100 µm 0,15 33* 33
100–300 µm 0,012 31 18
>300 µm 0,0014 5 1
In the lake water samples (publication I), concentration of 50–100 µm MPs was hundred times higher than concentration of >300 µm MPs. In fish samples from the same lake (publication III), the difference was only 30 times. While small particles were notably common in the lake, the sea water samples contained higher proportion of larger >300 µm particles. The number of samples from both environments is limited, but based on this data, larger MP particles seem to be more common in marine environment than in lake. However, lake water was sampled from the surface, whereas sea water was from the haloclines and thermoclines and between them. From that point of view, small MPs may be more common in the surface, while larger ones in the water column. One possibly biasing factor is that different samples were collected, pre-treated, and measured with different methods. Further, because the method in publication I was validated less extensively compared to the methods in publications II and III, the results of publication I may be more biased compared to the results of other publications.
Generally, many studies have reported that small MPs are systematically more abundant than larger ones in various environments. For example Roch et al. (2019) have mathematically estimated that >95% of MPs in water and fish would be smaller than 40 µm. Similarly, Setälä et al. (2016) have found that 100 µm net collects more MPs than 300 µm in the surface waters of the Baltic Sea. Besides the similar
environments than studied in this thesis, in the marine sediments of a Norwegian fjord >95% of MPs were smaller than 100 µm (Haave et al., 2019). Moreover, in the sub-surface waters of the Arctic Sea, close to Greenland, >11 µm MPs were analysed and 93% of the MPs were smaller than 300 µm and 69% smaller than 100 µm (Rist et al., 2020). In conclusion, smaller MPs seem to be everywhere more abundant than larger ones, which denotes that the small sizes should be monitored as well.