Study site description and sampling.
Sediment cores (one core per each lake from the deepest part of the lake) were taken from three lakes in Latvia (Supplementary Fig. S1). Lake Pinku is oligotrophic/mesotrophic lake located in a glacier depression. It is of high water quality and since 2004 the lake and its surrounding area has been a part of a protected nature park. The nearest regional motor road is approximately 600 m away. The lake's surface area is 29 ha, length 1.3 km, greatest width 0.3 km, average depth 4.3 m, greatest depth 20 m56. There is one inflow ditch and another outflow ditch. The core (25 cm length) was taken in August 2019 (Lat: 56° 59' 58.20", Long: 21° 41' 14.72") by a 5.2cm inner diameter Kajak corer.
Lake Usmas is among Latvia's largest lakes, with a surface area of 3469.2 ha and a water volume of 0.19km3. It is mesotrophic/eutrophic lake, located in a glacier depression and has several islands. Part of the lake (outside sampling area) belongs to a nature reserve. Apart from this area, the lake is a famous destination for recreational activities. The major motor road is approximately in 680 m distance. The lake is also surrounded by several local roads. Lake’s length is 13,5 km, greatest width 6.2 km, average depth 5.4 m, greatest depth 27 m. There are more than 10 inflow rivers and ditches and one outflow river57. The core (25 cm length) was taken in August 2019 (Lat: 57° 13’ 45.32’’, Long: 22° 10’ 26.97’’) by a 5.2cm inner diameter Kajak corer.
Lake Seksu is eutrophic, small (surface area 7.9 ha, length 550 m, greatest width 350 m) and shallow (average depth 2.5 m, greatest depth 6 m) lake located in the vicinity of Latvia's capital city, Riga. Major motor roads are located on both sides of the lake (the nearest - one of the most extensively used major motor roads in the country is in approximately 850 m distance). Inland dune forests mainly surround the adjacent catchment. The lake is part of the drinking water supply system enriching the groundwater level near the drinking water pump station. Due to that, access to the lake is limited, and a fence surrounds the lake. Water in the lake was artificially replenished between 1953 and 1965 from close to the eutrophic lake in order to increase the water level. The lake has no outflow, and there is an inflowing ditch58. The core (45 cm length) was taken in February 2019 (Lat: 57° 2’ 10.35’’, Long: 24° 21’ 6.73’’) by an 8 cm inner diameter Kayak/HTH gravity corer.
The sediment cores were divided in the field into 1 cm sections, placed in specially prepared (washed and muffled) glass jars, covered by aluminium foil and metal lids. Samples were stored in a cold room. Part of the material (4 to 5 cm3) of each section was dried and used for 137Cs/210Pb and spheroidal carbonaceous particles (SCPs) dating as well as for dry weight estimation, chemical and physical analyses.
Microplastic samples preparation and analysis
MP samples were prepared as a consecutive section with 5 cm intervals of mixed subsamples of the core. However, due to the very high water content of the topmost samples of Lake Seksu, the upper sample of the core consisted of 10 cm. Samples purification was done by applying a multi-step treatment method (Supplementary Fig. S7) adapted from ref.59–64.
Samples were analysed using Fourier Transform Infrared micro-Spectroscopy (µFTIR (Perkin Elmer Spotlight 400). A sub-sample of the total 5 mL sample (at least 0.5 mL) was taken using a capillary glass pipette (micro-classic, Brand GmbH, Germany) and filtered through 11×11 mm Si-filter (Fraunhofer Institute for Reliability and Microintegration, Germany). In case sample contained low number of particles and for the blank samples the total volume of the sample (5 mL) was analysed. Filters were left to dry for 12 h at room temperature. The analysis was done applying µFTIR-Imaging technique in transmission mode in a spectral range of 4000 − 750 cm− 1 at 8 cm− 1 resolution. The whole surface of the filter was scanned, and IR spectra of particles were obtained. The polymer assignments of the analysed particles were based on comparison with a FTIR spectral library developed at Tallinn University of Technology and in Leibniz Institute for Polymer Research Dresden. Spectral libraries comprise spectra of artificial polymers and natural organic and inorganic materials. The threshold for accepting the match was set to 70%, but all matches were verified by the operator as well. Cross-validation of measured spectra between Tallinn University of Technology and Aalborg University, Department of the Built Environment was made. At the same time a light microscope image of the inspected particles was produced for visual inspection and size determination65. Plastic particulates were also identified and categorised by size class for greatest length and width dimensions (major dimension is the longest side of the particle and minor dimension is the shortest side of the particle). Principles of measuring particles were taken from Sun & Liu 66protocol for measuring cell bio volume and surface area. We followed only measurement instructions from this work. To present measurements, we used 50µm intervals from 51 to 550 µm. One size step (50 µm) below and above fractioning sieves size was chosen to recover fibres as much as possible.
Quality control and blank tests
In order to avoid airborne and cross-contamination or unintentional loss of MP particles, several precautionary measures were taken. All equipment that was used for samples storage and treatment or came in contact with the samples in any other way, was made from either glass, PTFE or metal when possible and was thoroughly rinsed with filtered Milli-Q water before use. The polymer spectrum of all plastic materials, which were in contact with samples and not possible to replace with glass or other alternatives (e.g. sediment corer, gloves, bottle corks etc.) was recorded and excluded from further polymer analysis. Cotton laboratory coats of a specific colour (green) and blue or green nitrile gloves were worn while working with samples. The samples treatment was performed in the laminar flow cabinet. The same beaker was used when possible throughout the treatment process for each sample, rinsing it with filtered Milli-Q water between each treatment step. Samples were covered with aluminium foil when not processed or when placed in the shaking-heating bath located in the fume hood. All reagents were filtered through a glass fibre filter (pore size 1.2 µm).
Laboratory and field blanks were run parallel to real samples, applying the same processing steps to collect information on sample contamination degree during sample treatment. The air background sample showed contamination of 3.4 fibers h− 1 (only, no fragments) on average. Procedural and field blanks contained mostly viscose (65.7–100% of all polymers found in the blank samples), most likely from the clothing. Viscose was consequently removed from the further data analysis. Other polymer particles were detected only in very low numbers (few particles per blank sample). Recovery tests were performed with triplicate lake sediment samples spiked with standardised 100 red ⌀100 µm PS beads, density 1.05 g/cm3 (Sigma-Aldrich product no. 56969-10ML-F). The spiked samples were processed as described in the protocol for sediment samples. Extracted beads were easily identified due to their distinct appearance and were counted under a light microscope Leica DM400 B LED.
Core chronology samples preparation and analysis
The activity of total lead isotope 210Pb was determined indirectly by measuring polonium isotope 210Po using alpha spectrometry. Freeze-dried sediment samples of 0.2 g were spiked with a 209Po yield tracer and digested with concentrated nitric acid HNO3, perchloric acid HClO4, and hydrofluoric acid HF at a temperature of 100°C (CEM Mars 6 microwave digestion system, USA). Next, the solution was evaporated with 6 M hydrochloric acid HCl to dryness, and then dissolved in 0.5 M HCl. Polonium isotopes were spontaneously deposited within 4 h on silver discs67. After deposition, the discs were washed with methanol and analysed for 210Po and 209Po using a 7200-04 APEX Alpha Analyst spectrometer (Canberra, USA) equipped with PIPS A450-18AM detectors. The samples were counted for 24 h. A certified mixed alpha source (234U, 238U, 239Pu, and 241Am; SRS 73833-121, Analytics, Atlanta, Georgia, USA) was used to check the detector counting efficiencies, which varied from 30.9–33.9% for the applied geometry. Two blank samples were analysed with each sample batch to additionally verify the quality of the chemical procedure.
According to the black carbon combustion continuum model of Hedges, et al. 68 and Masiello 69, SCP only form during industrial fuel combustion at high temperature (> 1000°C). A load of SCP along the sediment sequences were estimated and followed the methodology of Rose70. The sediments were subjected to sequential chemical treatment using H2O2, potassium hydroxide (KOH) and HCl to remove organic material, silicates and carbonates, respectively. Lycopodium tablets71 with a known amount of spores were added as markers allowing estimation of SCP per sample. Slides for the microscope were prepared afterwards and all SCP within the whole slide was counted under a light microscope at 400 times magnification. Identification criteria for SCP counting followed Rose72. The concentrations of SCP were calculated as a number of particles per 0.2 gram dry mass of sediment. Across all sites, the record of SCP starts at the beginning of the 20th century and rapidly increases since the 1950s. The peak in SCP emissions in Latvia occurred in 1982+/-1073. After rapid increase and peak in SCP concentration followed a decline which coincided with the collapse of the Soviet Union when numerous manufactures and air pollutants halted their production in the Baltic region. The SCP occurrence pattern in our study mirrors worldwide SCP pattern change74 following the fuel combustion pattern: 1950 - the rise of SCP; 1982 - the peak of SCP; 1991 - the decrease of SCP.
Results of the 210Pb dating with the CFCS (Constant Flux Constant Sedimentation) model and the SCP analyses were used to build an age-depth model using the Clam deposition model75 package with a 95.4% confidence level in the R environment66. The mean weighted value of the modelled age was selected.
LOI samples preparation and analysis
The dry weight of 1-cm-thick subsamples with a 1-cm3 volume was determined after oven-drying at 105°C until constant weight. The organic matter content of the sediment was determined by loss-on-ignition (LOI) at 550°C for 4 h. The carbonate matter was calculated as the difference between the LOI at 950°C and the LOI at 550°C. Because the weight loss after 950°C is the amount of CO2 evolved from carbonate minerals, to get the actual percent of CO3, the weight after 950°C combustion was multiplied by 1.3676,77. Dry bulk density (BD, g‧cm− 3) was estimated on the base of the LOI for all samples. Non-carbonate siliciclastic matter, here referred to as minerogenic matter content, was obtained by subtracting organic and carbonate matter from the total sample weight after final combustion. All values for organic, carbonate and mineral matter are expressed as percentages.
Data analysis and statistical assessment
Particles were categorised into size ranges (with intervals of 50 µm) up to size class 51 to 550 µm considering two diametrical dimensions (major and minor dimension). Plastic densities data were obtained from existing databases. If density range was rather wide, e.g. for PA and PUR what can be made in a variety of densities and hardnesses, particular polymers were not included in the density analysis78,79. Principal Component Analysis (PCA) combined with factor analysis was applied to the data to understand which variables and factors are driving the transport of particles into deeper sediment layers. In order to include information about particle elongation (or shape) on a continuous scale, the Aspect ratio80 was calculated for each analysed particle. Aspect ratio is a shape descriptor defined by the ratio of the minimum to maximum Feret diameter80:
$$AR= \frac{{X}_{Feret min}}{{X}_{Feret max}}$$
,
(for easier interpretation of the results, the AR is expressed in decimal fractions; in the reviewed literature, this form of expression is used less often than the length-to-width ratio). In addition, a variable such as relative bulk density has been added to the dataset to normalize substantial variations in bulk density between sediments of different lakes. The most representative dimensions of PCA were determined according to the commonly used methodology, implementing Kaiser–Guttman criterion81together with the scree plot of eigenvalues and biplot visualization output, therefore PC1 (29.7%) and PC3 (18.8%) were selected as the best explanations of variances (PC2 was equal to 19.3%, but visualization of individuals’ groups was weaker). PCA output, variable coordinates, quality of the factor map COS2 and variable contribution can be found in Supplementary Table 3.
Spearman’s rank correlation test between age (date) of sediments, Aspect ratio and Equivalent Spherical Diameter was performed on the entire dataset of the pooled data of the three lakes in order to increase data length and include data on Lake Usmas, which was not sufficient (only 4 data lines) for a separate analysis. Thus, it was not possible to use layer depth as a correlating variable for the pooled data due to the different bulk density and of the sediments. Instead, the age of the sediments was used as a measure of depth, normalized to bulk density. The verified relationships were considered statistically significant at p < 0.05. For the analysis, the aspect ratio was conventionally divided into 4 groups: group 1 is AR < 0.25; group 2 is 0.25 ≤ AR < 0.5; group 3 is 0.5 ≤ AR < 0.75; group 4 is AR ≥ 0.75, to show differences in proportional composition of that groups in each sediment layer. The Equivalent Spherical Diameter (ESD)82, as a standardized particle fraction index, was calculated according to the formula:
where A is a particle's major dimension (or XFeret max) and C is a minor dimension (or XFeret min). Similarly to AR, to see how the proportion of particle fractions varies in different sediment layers, ESD was divided into four groups according to the lower quartile, median, and upper quartile of the ESD distribution: group 1 is ESD < 62 µm; group 2 is 62 µm ≤ ESD < 120 µm; group 3 is 120 µm ≤ ESD < 190 µm, group 4 is ESD ≥ 190. Spearman's rank correlation test was chosen because the AR proportional composition variables were not normally distributed, moreover Date is an ordered variable.
Elbow points of MP concentrations in sediment cores were calculated by means of akmedoids package (version 1.3.0) in R environment83. Analysis was done for lakes Pinku and Seksu only. The number of data from Lake Usmas was not sufficient for this analysis.
Data exploration, artworks, and statistical analyses were performed using R software for Windows, release 4.0.383 and GNU Image Manipulation Program (GIMP), release 21.10.3084.