Several studies in marine waters found a correlation between mesh size used for sampling and measured concentration. For example, Lindeque et al. (2020) found 2.5-fold and 10-fold greater MP concentrations with 100 µm compared to 333 and 500 µm mesh sizes. Even more remarkable is the finding of a 100-fold greater concentration measured with 100 µm nylon net compared to 333 µm manta net by Vermaire et al. (2017) or a 10,000-fold difference of maximum concentrations for small (10 µm) and large (300 µm) MP sampled by Roscher et al. (2021). Several studies pointed out rising concentrations towards lower size fractions (Adamopoulou et al., 2021; Aigars et al., 2021; Gündoğdu and Çevik, 2017; Gündoğdu, 2017; Güven et al., 2017; Hansen et al., 2023; La Kanhai et al., 2017; Kazour et al., 2019) and therefore, the importance of similar sampling methods to obtain comparable results (Tamminga et al., 2018). It is important to mention that this trend, even though reported in some of the considered literature on freshwater samples (Kataoka et al., 2023; Dris et al., 2015; Lorenz et al., 2020), could not be confirmed in the here conducted meta-analysis. For example, Adamopoulou et al. (2021) who took samples in the open sea as well as in the inner Saronikos Gulf, did not observe rising concentrations towards lower size fractions for the latter sampling location. On the other hand, Dris et al. (2018) find that using a 80 µm net instead of a 330 µm mesh size the probability of sampling fibres increases by 250%, which is backed up by Norén (2007), who find an increase of sampled fibers by three orders of magnitude.

In order to evaluate the influence of sampling mesh size on measured concentrations and therefore enable comparison between studies, the ocean and freshwater datasets were split into six size fractions (0–25 μm, 25–100 μm, 100–200 μm, 200–300 μm, 300–400 μm, 400–600 µm). The resulting violin plots, with average concentrations in items/m³ on a logarithmic scale against mesh size in µm are shown in Figure 5 for freshwater (right violin side, green) and oceans (left violin side, blue), respectively.

Regarding freshwaters, no tendency is visible with concentrations ranging over up to six orders of magnitude (for size class 300 μm–400 µm). The smallest median concentration is observed for the largest size fraction (400 μm–600 µm) whereas the highest median concentration is observed for size fraction 25 μm–100 µm. The medians of the remaining four size fractions only differ by one order of magnitude. Contrary, the median concentrations for oceans differ over several orders of magnitude for all size fractions. A decreasing tendency of concentration with increasing mesh size is visible for all size fractions except for the largest one (400 μm–600 µm), which is in the same order of magnitude as size fraction 200 μm–300 µm and the smallest class (0 μm–25 µm), which is in the same order of magnitude as 25 μm–100 µm.

Direct comparison of both of the above described figures shows higher maximum measured concentrations in oceans and a tendency to more outliers, which is attributable to the higher absolute number of data points (blue), especially in size fraction 300 μm–400 µm.

The decreasing tendency observed in Figure 5 aligns with expectations and is probably attributable to several abiotic (caused by light, temperature, water, air and mechanical forces) and biotic (caused by organisms such as bacteria, fungi and insects) degradation mechanisms acting on the (mostly secondary) MP particles and crushing them in even smaller particles (Zhang et al., 2021; Prata et al., 2019). This effect is enhanced by longer residence times in oceans compared to rivers and thus leading to higher fragmentation in oceans (Parać et al., 2022). Therefore, an increasing number of small size fraction particles is to be expected in oceanic samples. For fibers in particular, smaller mesh sizes are necessary for adequate sampling (Prata et al., 2019), hence it should be noted that they often slip through the mesh or are collected, although they might be larger than the respective used mesh size, depending on their spatial orientation (Gewert et al., 2017). The observed increase in the average concentration of microplastics in the largest size fraction, coupled with a decrease in the smallest size fraction, might be attributable to the limited number of data points available for these specific size ranges. This tendency once again highlights the need for a more standardized and comprehensive sampling methodology to ensure accurate and representative assessments of microplastic pollution across all size ranges.

In the context of freshwater systems, where no discernible trend in microplastic concentrations depending on mesh size is observed, it is hypothesized that a higher proportion of primary microplastics may be present. This is likely due to shorter residence times of water within freshwater bodies before eventually flowing into oceans. As a result, degradation processes that would break down larger plastic particles into smaller secondary MP are less likely to occur within the limited time frame. Consequently, instead of observing an increase in the concentration of smaller secondary MP particles, we see a relatively constant measured average concentration of MP over all size fractions. This consistency is attributed to the predominance of larger primary MP that retain their size during their transit through the freshwater system. Therefore, the particle sizes remain constant during residency in the river, leading to a stable average concentration of MP in these environments.

Histograms of the respective measured concentrations in oceans and freshwaters are shown in Figure 6 with the relative frequency versus average concentration in items/m³ on a logarithmic scale. For Figure 6 the data was binned into 15 different size ranges, all equal in span width, according to the mesh sizes used in sampling. The data was then normalized such that the area under the graph equals 1. That means the number of observations in one mesh size bin is divided by the total number of observations for all bins, which gives the relative frequency.

For the oceanic samples (Figure 6b) a bimodal distribution is present with peaks at approximately 0.1 items/m³ and 1,000 items/m³ and a minimum in between them with less than 100 items/m³. A difference of 30% between the relative frequencies for the two peaks is observed. In terms of freshwater (Figure 6a) a mono-modal distribution is obtained with a peak at around 10 items/m³ with a relative frequency of approximately 40%. When combining the two distributions in one figure (Figure 6c) it is notable that the peak of the freshwater distribution is located at the trough of the oceanic distribution.

Furthermore, it is to be mentioned, that none of the presented distributions align with a normal-, log-normal or any other common distribution. This observation could indicate several underlying issues. One possibility is that the current dataset is not extensive enough, suggesting that more data points need to be sampled to achieve a more accurate and representative distribution. Another factor to consider is the influence of the sampling net’s mesh size. The mesh sizes most commonly used in microplastic sampling range from 300 µm to 400 μm, which may create significant gaps in the data. This is not merely about the number of sampling points, but rather the intervals between the mesh sizes, leading to large gaps between data points. Such gaps could result in a distorted representation of the true distribution of microplastic particles. Additionally the non-standardized sampling leads to multiple distributions layered over one another and therefore again increases the difficulty of fitting common distributions. As seen in Figure 6b concentrations between 0.01 items/m³ to 1 items/m³ are the most frequent ones, which aligns with the observation in Figure 5, where they appear as mean values for three different size fractions and are especially frequent in the most common size class (300 μm–400 µm). This underlines the importance of the mesh size used for sampling and once again shows that larger mesh sizes result in smaller measured concentrations.

It may be hypothesized that an influencing factor on the staggered peaks in the combined distribution is due to agglomeration and degradation of MP in oceanic environments. In the ocean, smaller particles tend to aggregate, leading to more frequently measured lower concentrations. This agglomeration of small particles could explain the rightward shift of the peak in the oceanic distribution compared to freshwater systems, where primarily primary MP are present. These primary MP in freshwater do not significantly agglomerate or degrade, maintaining a more stable size distribution (Munari et al., 2021; Fischer et al., 2016; Campanale et al., 2020). Additionally, the degradation of MP in oceanic environments might contribute to the observed leftward shift of one peak compared to freshwater systems. In freshwater, as previously mentioned, no significant degradation occurs due to the shorter residence time before these particles flow into the ocean. At the same time, larger particles (sometimes referred to as macro- or mesoplastics) from other sources, such as fishing nets and marine debris, could also be present in the oceanic environment (Reinold et al., 2021; Pedrotti et al., 2022; Gündoğdu and Çevik, 2017; Ruiz-Orejón et al., 2019). This combination of factors might result in the shown peak distribution and explain the left and right shift of the mono-modal particle size distribution found in rivers to the bi-modal size distribution found in oceans.

In literature, the dependency of MP concentration on the distance to the nearest WWTP is discussed and not consensus. While it is generally agreed that WWTPs, especially municipal ones, serve as sources of microplastics (fibers, fragments, and beads), some studies indicate a significant correlation (Dris et al., 2015; Napper and Thompson, 2016; Mughini-Gras et al., 2021; Ziajahromi et al., 2017; Habib et al., 1998), whereas others point out that different other effects, such as weather phenomena (Primpke et al., 2017; Stanton et al., 2020), outweigh the proximity to WWTP outlets. Additional publications report higher concentrations upstream of WWTP outlets, with values ranging from 354 items/m³ to 69 items/m³ at the nearest downstream sampling station (Schrank et al., 2022). Schmidt et al. even report three different findings in the same river at three different outlets: one showing a significant increase, one showing only a tendencial increase and the third one showing no significant difference in concentration between upstream and downstream of the outlet (Schmidt et al., 2018a). Utilizing the data collected in this study, along with spatial information about WWTP outlets, a more comprehensive analysis can be conducted, including a variety of sampling methods, weather influences and river sizes.

To find the distance of a sampling station to the nearest WWTP outlet, a python script was developed. For the oceanic sampling stations, the distance to the nearest WWTP outlet was determined using the geopy library (Esmukov, 2024) and the geodesic distance calculation method. For the river sampling stations, additionally the angle between the river flow direction and the direction to the nearest WWTP outlet was calculated and only the WWTP outlets upstream the sampling station were considered as no back mixing in rivers is a common assumption (Urgert, 2015; Wal et al., 2015). Only distances under 30 km to the nearest WWTP outlet were considered in the analysis to minimize the influence of other MP sources. Figure 7 illustrates the impact on MP concentration in water samples for both oceanic and freshwater sampling stations.

As illustrated in Figure 7, the concentration of microplastics in the water samples decreases with increasing distance from the nearest WWTP outlet. However the correlation is not significant, with a PCC of p = − 0.167 for rivers, p = − 0.011 for oceans and p = − 0.044 combined with R 2 values of 0.34 , 0.01 , 0.19 respectively.

The low correlation strength for oceanic sampling points is expected, as factors such as currents and tides have a much stronger influence on MP concentration than the distance to the nearest WWTP outlet, especially for sampling points further offshore where this distance approaches that to the nearest coastline. On the other hand, the finding of low to non-significance of the distance to the nearest WWTP outlet on MP concentration in river sampling stations contrasts with the findings of other publications as mentioned above.

Even though waste water treatment plants, of course, are a great source of MP influx into the water systems (Murphy et al., 2016; Mason et al., 2016) and may act as point sources for fibers (Ziajahromi et al., 2017; Lares et al., 2018), the influence of other sources like surface runoff (Campanale et al., 2020), atmospheric deposition (Dris et al., 2015) and industrial waste water (Horton et al., 2017) is not to be underestimated (Nizzetto et al., 2016; Liu et al., 2019; Baldwin et al., 2016; Mani and Burkardt-Holm, 2020; Schmidt et al., 2018a; Obermaier and Pistocci, 2022). Especially challenging in this regard is, for example, the filter efficiency of WWTPs regarding microplastic particles as WWTPs are expected to filter MP with varying efficiency. Also, the input into the WWTPs can differ strongly depending on the population density and industrial or agricultural usage of the area in proximity to the WWTP. The findings underline the complexity of the MP pollution problem and the need for further research and more standardized data and methods to better understand the sources and sinks of MP in the environment. The meta-analysis also shows that the correlation between higher MP concentrations in proximity to WWTP outlets shows only low significance and points to other factors outweighing and covering up the effect of MP rich WWTP effluent entering rivers and oceans.

The influence of population density on MP concentration in the water samples was analyzed using population density data from Copernicus (Joint Research Centre Data Catalogue, 2024) and the MP concentration in the water samples. The sourced data from copernicus gives the amount of inhabitants in an rectangular cell pattern over the desired region. All inhabitants in the cells inside a certain radius around the sampling sites were summed up to obtain the population density for that circular area. Plotting the concentration versus the population density in a certain radius results in Figure 8.

Because defining a meaningful radius around the sampling station is difficult and may vary for oceanic and fresh water sampling stations, the analysis was done for multiple radii. The results show a moderate increase in the MP concentration with increasing population density. However, the correlation is only significant ( | p | ≥ 0.3 ) for some of the radii as compiled in Table 2. As can be seen in Table 3; Figure 8 the correlation strength increases for the radii 5, 10 and 25 km and decreases beyond that value. The mean squared error for the straight line fits shows a similar behavior, but is only increasing until 50 km. Generally, for the river samples higher population densities were found. This finding is expected because the population density tends towards zero further offshore as there is no one living in the oceans. Also for sampling points closer to coastlines, the population density is expected to be lower compared to the inland sampling points as only the coastline can be populated whereas for sampling points in freshwater or rivers, the sampling can take place right in a city center, increasing the population density heavily. However, some oceanic sampling points also show high population densities which are supposedly near big coastal cities. As would be expected, especially these sampling points show higher MP concentrations.

Since the population density could only be calculated for the sampling stations with sufficient data within the specified radius, sampling points near the edge of the population data were not considered for larger radii. This exclusion may result in fewer extremes and, consequently, better fits in the analysis. As shown, population density in an area does seem to have an influence on the microplastic concentration, even though the correlation is relatively weak. This may be due to the fact that population density is not directly correlated with all anthropogenic activities. For example, industrial areas can exist without any population density (Gao et al., 2023a; Klein et al., 2015) and also tourism activities are not represented in this metric, which becomes particularly relevant in tourism hot spots (Sighicelli et al., 2018; Asenova et al., 2021; Fischer et al., 2016). Generally, the population density–especially for fresh water sampling–shows a significant correlation and has a much larger impact compared to the other factors investigated in this study (Mani and Burkardt-Holm, 2020; Correa-Araneda et al., 2022; Kataoka et al., 2019).

In addition to the previously mentioned factors, the distance to the next coastline was analyzed for the oceanic sampling points. Similar to the findings for the WWTP outlets, the distance to the next coastline was found to also not have a significant influence on the MP concentration in the water samples. Figure 9 shows the influence of the distance to the next coastline on the MP concentration in the water samples for all samples (a) as well as for only the samples taken with a net size of 300 μm–350 µm (b). With a PCC of p = 0.050531 and p = − 0.042682 respectively, the correlations are not significant. The R 2 values of 0.0043 and 0.053 indicate that the linear model describes the data only poorly.

In literature, often a correlation between MP concentration and distance to coast is assumed, but the results are differing heavily, ranging from no correlation (Aytan et al., 2016; Eryaşar et al., 2021; Fagiano et al., 2022; Kermenidou et al., 2023; Wakkaf et al., 2020; Zeri et al., 2018) over the expected negative correlation (Compa et al., 2020; Alomar et al., 2024; Galli et al., 2023) up to a positive correlation (Hansen et al., 2023; Lechthaler et al., 2020). Pakhomova et al. (2022) found a negative correlation for abundance of fibers with distance to coast, but no correlation for fragments. Fagiano et al. (2022) stated location to be more important for MP abundances than distance to coastlines, whereas Adamopoulou et al. (2021) highlighted the importance of ocean currents. These ambivalent results are reflected by the results obtained in this review.

In literature, wind speed is assumed to have a strong impact on measured concentrations in surface waters due to wind induced mixing (Kukulka et al., 2012; Reisser et al., 2015). Especially, Kukulka et al. (2012) found a strong effect of wind mixing on the vertical MP distribution and therefore introduced a correction factor to calculate the actual MP concentration for wind speeds > 5 m s⁻¹, which was confirmed by Reisser et al. (2015) and applied in several following publications (de Haan et al., 2019; Adamopoulou et al., 2021; Compa et al., 2020; Galli et al., 2023; Poulain et al., 2019; Ruiz-Orejón et al., 2019). In Figure 10 a scatter plot of the MP concentration versus wind speed in Beaufort is presented on a double logarithmic scale with a linear regression in gray. Although, a decreasing tendency is observable, this tendency is not statistically significant with a PCC of p = − 0.0673 and R 2 = 0.033 . Given these findings, the validity of the proposed correction factor should be critically questioned before application.

The weak correlation observed in our data suggests that the influence of wind speed on MP concentration may not be as straightforward or significant as previously reported. While earlier studies have emphasized the role of wind-induced mixing, our results indicate that this effect may be overestimated, or that other factors, not accounted for in those studies, could be influencing MP distribution. This raises important questions about the generalizability of the correction factor across different environments and conditions, emphasizing the need for further research to reassess its applicability.

Often assumed is also a dependence of measured concentration on season in which the sampling took place, but the opinions on which seasons show high or low concentrations differ (Rodrigues et al., 2019; Faruk Çullu et al., 2021; Adamopoulou et al., 2021; Aytan et al., 2016; Ben Ismail et al., 2022; Oztekin and Bat, 2017; Ruiz-Orejón et al., 2019; Tesán Onrubia et al., 2021). Exemplary, Faruk Çullu et al. (2021) found the highest MP concentration in autumn and identified seasonal differences in the occurrence of the MP species (fragments dominate in autumn and winter, fibers in spring and summer), similar to Adamopoulou et al. (2021) and Aytan et al. (2016) who found the highest concentration in autumn. They emphasized the correlation with rain events during the same day or 10 days in advance of sampling. Contrary, Ben Ismail et al. (2022) found the lowest concentration during autumn and highest concentration in spring. Similarly, Oztekin and Bat (2017) found highest MP abundances in spring and lowest values in winter. Ruiz-Orejón et al. (2019) and Tesán Onrubia et al. (2021) found the summer season to show the highest MP concentrations. Carretero et al. (2022) did not find significant differences between the MP abundances for different months, which aligns with findings of other studies (Reinold et al., 2021; Schmidt et al., 2018b). Generally, seasons represent combinations of several environmental factors, such as precipitation, temperature or wind speed, which should best be evaluated separately. Due to different climate zones in the reviewed studies (sub polar, temperate, subtropical) these environmental factors may differ heavily and increase the complexity of the evaluation. Furthermore, other factors which influence the MP concentration, such as tourism activity (Compa et al., 2020; Alomar et al., 2024; Galli et al., 2023; Ruiz-Orejón et al., 2019) or ocean currents (Adamopoulou et al., 2021; Tunçer et al., 2018; Ben Ismail et al., 2022; Pakhomova et al., 2022; Tunçer et al., 2018), might as well depend on seasons.

Although this meta-analysis includes only studies from European aquatic environments, substantial differences in rainfall regimes and touristic pressures can still be expected across sampling locations. To better resolve regional patterns in seasonal MP concentrations, sampling sites were grouped by latitude into three zones: (1) Mediterranean (35 deg N to 45 deg N), (2) Central Europe (45 deg N to 55 deg N), (3) Northern Europe (55deg N to 75deg N). A boxplot for the different seasons, comparing the three different regions is shown in Figure 11. In both Northern and Central Europe, the highest concentrations of MPs were observed in spring, followed by summer, autumn, and winter. In contrast, Mediterranean regions exhibited peak MP concentrations in summer, followed by spring, autumn, and winter. Pairwise comparison shows statistically significant seasonal variation in both Mediterranean and Northern regions, whereas in central Europe, concentrations in summer and autumn did not differ significantly. The observed seasonal patterns in microplastic concentrations appear to reflect distinct combinations of anthropogenic and climatic drivers. In Mediterranean Europe, significantly elevated MP levels are likely linked to intense seasonal tourism, increased recreational activity and limited hydrological flushing due to dry conditions. In Northern Europe peak concentrations in spring are potentially caused by snow-melt and rainfall, mobilizing accumulated plastic debris into aquatic systems. The similar tendency observed in Central Europe, combined with the lack of significance between summer and autumn, suggests a more uniform year-round input of MP, likely from continuous urban, agricultural, or industrial sources and less pronounced climatic seasonality.

These results underscore the complex interplay of multiple environmental and anthropogenic factors, indicating that MP abundance cannot be linked solely to the season. However, the influence of season remains significant and should not be underestimated.