Modelled and observed plastic pollution on remote Scottish beaches

Beach-cleans conducted on the west coast of Scotland investigated the distribution of land-and marine-sourced litter and compared these with a particle tracking model representing the presumed principal land-based source. Modelled particles dispersed widely, even reaching the remote northwest coast, with ‘hotspots ’ and ‘coldspots ’ on windward and leeward coasts respectively. In beach sampling, however, land-sourced litter represented only 19% of items by count and 8% by weight, while marine-sourced litter represented 46% by count and 62% by weight. The source of the remainder could not be identified. Windward coasts had an average count of 1859 litter items per 100 m, and weight of 14,862 g per 100 m. Leeward coasts had an average count of 32 litter items per 100 m and weight of 738 g per 100 m. Field observations and model predictions were consistent in many respects for land-sourced litter, however marine-sourced litter is dominant on many coastlines.


Introduction
Plastic litter accumulation in the ocean and on coastlines is an increasing problem worldwide. Previous studies have shown that the most abundant type of litter is plastic, representing around 60-80% of the total mass identified (e.g., Derraik, 2002;Topcu et al., 2013;Nelms et al., 2017;Turrell, 2019;Morales-Caselles et al., 2021;Smith and Turrell, 2021). Jambeck et al. (2015) estimated that 4.8-12.7 million tonnes (Mt) of plastic waste enter the ocean globally per year, which contributes to at least 5.25 trillion floating particles. Plastics are an ecological hazard due to ingestion by or entanglement with marine wildlife (e.g., Gregory, 2009;Wilcox et al., 2015;Nelms et al., 2016), by modification of habitats, or through their role as a transport vector for invasive species (e.g., Kiessling et al., 2015). Furthermore, plastic litter can alter the aesthetics of marine and coastal environments, and thus can have a negative impact on tourism industries (Nelms et al., 2017). Our understanding of the extent of plastic pollution can be improved by providing evidence on the abundance, distribution, and composition of litter in the environment (Nelms et al., 2022). This can also help to inform and implement strategies targeted at mitigating the problem.
It has been documented that the abundance, distribution, and composition of plastic litter exhibits considerable spatial variability (e. g., Galgani et al., 2021;Buhl-Mortensen et al., 2022), due to the transport of positively buoyant plastics by the combination of physical processes such as currents, winds, and waves . It has been suggested that this variability is also influenced by proximity to sources of pollution e.g., centres of population, riverine systems, industrial areas, fishing grounds, aquacultural and agricultural farms, and touristic hotspots (Kaandorp et al., 2021). Thus, the dominant composition may differ by location. Sources can be grouped into land-based (e. g., populated areas, riverine systems) and marine-based (e.g., fishing grounds, shipping lanes) (Thompson et al., 2009). Meijer et al. (2021) estimated that between 0.8 and 2.7 Mt of plastic litter currently enters the ocean per year from riverine systems (i.e., a land-based source). It has been suggested that plastic litter entering the world's oceans predominantly accumulates on coastlines (Onink et al., 2021) or on the seafloor (Maes et al., 2018). On the west coast of Scotland, Turrell (2020a) estimated that >90% of macro-plastics (items >5 mm) came from one major riverine source in the south-west of the region, the River Clyde (Fig. 1).
Among marine-based sources, a specific type of waste is Abandoned, Lost or Discarded Fishing Gears (ALDFG), e.g., fishing rope, fishing nets, creel pots, and fishing line. As the material of fishing gear has transitioned to more durable synthetic polymers, the quantity and distribution of ALDFG in the ocean and on coastlines has likely increased due to the long lifespan of these materials (Gilman, 2015). ALDFG are problematic as they are designed to catch marine animals, intentionally and unintentionally, pose navigational hazards, and damage in-use fishing gear which leads to more gear loss (Deshpande et al., 2020). Some recent studies have attempted to quantify the input of ALDFG to the ocean. For example, Deshpande et al. (2020) reported that commercial fishing off Norway contributes around 380 t of ALDFG per year. Kuczenski et al. (2021) calculated the median estimate for gear lost globally by fishing in 2018 was 48,400 t, but this did not include abandoned or discarded gear. Even though attempts have been made to calculate the loss of ALDFG, there is still a critical knowledge gap with respect to the magnitude of the problem (Gilman, 2015;Haarr et al., 2020;Lebreton et al., 2022).
While it is widely assumed that marine plastic litter mostly originates from land-based sources (Morales-Caselles et al., 2021), recent studies have shown that ALDFG is a significant source of plastic litter in the ocean (e.g., Lebreton et al., 2022) and has been found to accumulate on remote islands (e.g., Lavers and Bond, 2017;Kaviarasan et al., 2020). Schmuck et al. (2017) suggested that the density of beached litter, in particular macro-plastic (items >5 mm) was higher on uninhabited beaches, and similarly Galgani et al. (2021) state that increasing quantities of plastic litter are being found in remote areas. A 'remote area' is defined as any location distant from dense populations and urban centres, and may have limited accessibility.
The west coast of Scotland, the study region of this paper, has a complex coastline of many islands, sea-lochs (fjords), and embayments. The region experiences prevailing south-westerly winds (Aleynik et al., 2016;Turrell, 2019;Allison et al., 2022), a variety of tidal ranges, and intense weather conditions (Sabatino et al., 2016). The Clyde Sea is bordered to the north by the most industrialised region of the west coast, with considerable development, high population density, recreational activities, and sewage disposal (Baxter et al., 1979). It is therefore a potential source of ocean pollution. Immediately off the west coast of mainland Scotland and to the north of the Clyde Sea lies a chain of islands collectively known as the Inner Hebrides, including Islay, Mull, Tiree, and Skye. A second chain of islands, located on the shelf further out towards the eastern Atlantic, are collectively known as the Outer Hebrides (e.g, Uist, Barra, and Lewis and Harris). In contrast to the Clyde Sea, these remote regions have very low population densities and rural communities, with little industrial activity.
The Scottish Coastal Current (SCC) (Fig. 1) originates in the Irish Sea (MacKay and Baxter, 1986), exiting via the North Channel and flowing northwards at ~5 km/d up the west coast (Simpson and Hill, 1986) while receiving input from rivers and sea-lochs, including the Clyde Sea.
In the Little Minch, the SCC splits, with part continuing northwards into the Upper Minch, and part travelling around the southern tip of the Outer Hebrides archipelago, before heading northwards up the western coast of the Outer Hebrides. The Islay Front, situated west of the island of Islay and north of Ireland marks the transition from the coastal region of relatively shallow water and strong tides, to more stratified, Atlanticinfluenced waters to the west (MacKay and Baxter, 1986). Beach-clean monitoring campaigns often focus on coastal clean-ups which apply citizen-science (e.g., Nelms et al., 2017;Watts et al., 2017;Turrell, 2019;Smith and Turrell, 2021) and these provide a useful tool to understand the composition, distribution, and abundance of plastic litter. However, more standardisation and coordination is needed to compare data (Nelms et al., 2022), and estimate influxes of landsourced (Turrell, 2020a) and marine-sourced (Smith and Turrell, 2021) litter. It is common to quantify beach litter by count of litter items (e.g., Nelms et al., 2017). However, Smith and Turrell (2021) highlight the importance of recording both count and weight of litter items as the method chosen can alter the perception of the dominant source type (e. g., Haarr et al., 2020;Smith and Turrell, 2021). For example, from 80 surveys of North Sea beached litter, contributions from marine-based sources measured only 6% by count but 41% by weight (Smith and Turrell, 2021).
Although the effectiveness of citizen-science beach-clean activities is well known, for example in the removal of litter from the environment whilst raising public awareness, beach-clean surveys are often not optimized. To effectively minimise beached litter, it is important to use clean-up resources more efficiently by targeting the most impacted areas. It is also important to identify principal litter sources, both in terms of nature of the source and geographical location, to tackle the problem before entry into the marine environment. To understand the spatial abundance, distribution, and composition of marine litter, longterm datasets with broad geographical coverage are required (Nelms et al., 2022). However, the collection of such large datasets can be costly, time-consuming, and labour intensive (Haarr et al., 2020;Nelms et al., 2022). Various field technologies are rapidly being developed to survey beached litter, for example, unmanned aerial vehicles (UAVs), more commonly known as drones. Such methods can help to overcome the aforementioned survey limitations by expanding the monitored area, the frequency of surveys, and accessing remote areas (Andriolo et al., 2021a;Andriolo et al., 2021b;Escobar-Sánchez et al., 2021;Cocking et al., 2022;Goncalves et al., 2022). Hydrodynamic models combined with particle tracking models, are an effective tool to predict high impact beaching sites of plastic litter, and many are being developed for specific regions (e.g., Gutow et al., 2018;Jalon-Rojas et al., 2019;Lebreton and Andrady, 2019;van der Mheen et al., 2019;Onink and Laufkötter, 2020;van der Mheen et al., 2020;Onink et al., 2021;Allison et al., 2022). Previous modelling studies have demonstrated that such coastal spatial variability in beaching is wind-driven and have addressed the importance of onshore winds in areas of high deposition (e.g., Critchell and Lambrechts, 2016;Gutow et al., 2018;Rios et al., 2018;Allison et al., 2022). Gutow et al. (2018) report that particles are pushed ashore by prevailing onshore westerly-winds and, similarly, Allison et al. (2022) found that windward (i.e., the prevailing wind direction is directed onshore) coasts had higher beached loadings. While particle tracking is an effective method, it is important to consider model predictions with field observations from any study region. Furthermore, previous studies have found drones to be successful in mapping beached litter, with image detection identifying characteristics of macro-litter e.g., type, material and size, and distribution (e.g., Andriolo et al., 2021a;Andriolo et al., 2021b;Goncalves et al., 2022). Thus, drones could provide a further tool to validate existing particle tracking models and assess beached litter in remote regions.
Here, we build on the robustness of an existing particle tracking model by using a beach-clean dataset, collected within the study region, to validate model predictions. The specific objectives of the paper are to (1) support model predictions with field observations from the region examining the spatial variation, (2) investigate the composition of beached-plastic litter to understand its sources, and (3) identify areas of high ('hotspots') and low ('coldspots') beached loadings on the west coast of Scotland.

Survey protocol
The survey protocol followed that of the Marine Conservation Society (MCS) and OSPAR guidelines (OSPAR Agreement 2020) to remain consistent with previous monitoring studies. Nelms et al. (2022) suggested that in order to collect scientifically meaningful data it is essential that all citizen-science or beach clean-up studies follow a standard methodology, which is often not the case. OSPAR states that it is standard practice to survey a 100 m (minimum 10 m) random horizontal transect marked along the most recent high-tide line. This can be identified by a line of seaweed or marine debris. Volunteer beach-cleaners walk along the transect to search and collect litter items from the space between the high-tide line to the back edge of the beach, which could be marked by natural structures e.g., vegetation, sand-dunes, or man-made structures e.g., walls or pavements. To account for any instance where a 100 m transect is not possible all data should be standardised to 100 m (e.g., Nelms et al., 2017;Nelms et al., 2020;Turrell, 2020b). This occurs, for example, in circumstances where the length of the beach is <100 m (i.e., pocket beaches) due to rocks or cliffs, or removal and transport of litter items from 100 m is not possible due to high quantities, or limited accessibility to the beach.
For this study, only items that were visible and physically able to be removed from the foreshore were included in the analysis. Mega items (e.g., large ropes, creel pots, tyres; Supplementary Fig. S1) were left insitu due to burial, or extraction and transport limitations. All samples that were collected were labelled with the date, location, and survey team names, and stored back in the laboratory. Care was taken to minimise any breakage of brittle items before post-processing.

Survey locations
In total there were 49 beach-clean surveys across 27 beaches ( Fig. 1) on the northwest coast of Scotland. A beach here is defined as any stretch of coast where beach-clean surveys occurred, with samples collected from four main foreshore substrate types: sandy, coarse sand, pebble small, and pebble large (Fig. 2). The study region is subject to semidiurnal tides (twice daily), that are macro-tidal with a range of between 4 and 5 m. Of the 27 beaches, 22 were surveyed twice, and five were surveyed once. All surveys were completed between August 2022 and March 2023. Each beach-clean location was selected following the results of Allison et al. (2022), in order to validate model predictions of beached plastic litter spatial distribution and composition. Model 'hotspots' were predicted on windward (i.e., the prevailing wind direction is directed onshore) coasts and 'coldspots' on leeward (i.e., the prevailing wind direction is directed offshore) coasts, of the northwest islands. This pattern being consistent with the south-westerly prevailing winds that occur along this coastline (Allison et al., 2022). Therefore, at least two beaches, one from a 'hotspot' and one from a 'coldspot' were selected on the Western Isles. The daily averaged wind speed (m/s) and direction are presented in a wind rose for the year 2022 ( Fig. 1). Wind data were derived at a defined location in the Atlantic Ocean (7.8 • W, 56.46 • N) from a high-resolution weather forecast model of the west coast of Scotland (Skamarock et al., 2008;Aleynik et al., 2016).

Post-processing of samples
To prepare the collected litter for analysis, all items collected were first washed to remove any adhering material (e.g., sand, seaweed). The next stage was to air-dry the litter. Once completely dried, all items were divided into litter categories (n = 131) defined by OSPAR (OSPAR Agreement 2020-01). For the purposes of this study, only plastic categories were used (n = 77). A full list of litter item categories and corresponding OSPAR ID values is provided in Supplementary Table S1.
For each plastic litter category (n = 77), a count of litter items (from all 49 beach-clean surveys) and dry weight (g) (from 43 beach-clean surveys) were recorded. These two methods of measuring quantity were selected based on previous studies (e.g., Hong et al., 2014;Jang et al., 2014;Ali and Shams, 2015;Lee et al., 2017;Haarr et al., 2020;Smith and Turrell, 2021). Dry weight was measured on Brecknell C3236 weighing scales (3 kg capacity and 0.1 g resolution).

Spatial distribution 2.4.1. Mainland vs. island coasts
In order to explore the spatial distribution of beached plastic litter across the study region, the complete dataset (n = 49) was pooled into eight defined locations (Fig. 1); River Clyde, Clyde Sea, Kintyre, Oban -'Mainland' survey sites, and Islay, Mull, Tiree, and Skye -'Island' survey sites (Table 1). Island survey sites were classified as remote i.e., distant from dense populations and urban centres, and mainland survey sites were classified as urban i.e., industrialised areas consisting of dense populations.
The average beached loading of plastic litter items per 100 m was calculated, both by count (NP/100 m) and weight (g/100 m), for assessment of the variability of beached loadings between urban, mainland survey sites, and remote, island survey sites. The average weight (g) per litter item was also calculated for each plastic litter category, and for each identified sub-source (see below), for a better understanding of the weight class of each item. All data analysis used the computer programming language R (R Core Team, 2021).

Windward vs. leeward coasts
As the west coast of Scotland is subject to prevailing south-westerly winds (Fig. 1), which has been previously documented to influence the deposition of beached plastic litter onshore (Critchell and Lambrechts, 2016;Gutow et al., 2018;Turrell, 2018Turrell, , 2019Allison et al., 2022), all survey sites were categorised by their coastal orientation (Table 1).
Of particular interest was the spatial variability of beached plastic quantity between windward and leeward coasts of island survey sites. To consider the effect of coastal orientation, and validate the predictions of Allison et al. (2022) an average beach loading per 100 m of surveyed coastline was calculated by count (NP/100 m) and weight (g/100 m) across each island by coastal orientation.

Input source characterisation
To understand the origins of beached plastic litter, the composition of each beach-clean sample was analysed by dividing all individual plastic litter item categories (n = 77) into three groups based on their proposed source: i.e., land-sourced, marine-sourced, or of unknown source (Supplementary Table S1), and these are referred to as the source type. Examples of the grouping of items into source types during postprocessing are provided in Fig. 3.
Within each of the three source types, plastic litter items were also assigned to a sub-source as follows. Land-sourced: public sub-source (e. g., drink bottles or crisps packets); medical sub-source (e.g., single-use face masks or syringes); and sewage sub-source (e.g., wet wipes or tampons). Marine-sourced: fishing sub-source (e.g., fishing nets or rope); and shipping sub-source (e.g., industrial packaging or strapping bands). Of unknown source: fragment sub-source (no longer identifiable) and other items that did not fit into a litter category (e.g., paint brush handle; Supplementary Table S1).

Spatial regressions
The spatial patterns of the composition of beached plastic litter has been previously reported to vary relative to the proximity to the source (Turrell, 2019(Turrell, , 2020aKaandorp et al., 2021). A linear regression model was fitted to the data to test whether an increase in water-distance from a principal land-based litter source in an urban setting, to areas of remoteness, had a significant effect on the composition of beached plastic litter. The composition was evaluated based on the percentage of each source type found; i.e., land-sourced, marine-sourced, or of unknown-source. A fixed point within the Clyde Sea was chosen (4.86 • W, 55.97 • N) to represent the principal land-based litter source as it was estimated by Turrell (2020a) that 93% of land-sourced beached litter originated from within the River Clyde (location shown in Fig. 1).

Basic litter statistics
A total of 3.5 km of coastline was surveyed across the study region from 27 beach-clean locations yielding a total count of 10,901 plastic litter items, and weight of 86,013 g. This gives an average weight per plastic litter item of 7.9 g.
The average number of plastic litter items per 100 m (NP/100 m) for the full study region was 813 ± 2132 NP/100 m, with a median value of 85. The average weight was 6464 ± 14,674 g/100 m, with a median value of 865 g. Although, both average and median values are presented here, following the recommendation of Smith and Turrell (2021) it was concluded the average value provides a better representation of the overall dataset.
Across all plastic litter item categories (n = 77), a minimum weight per item of 0.1 g and a maximum weight per item of 1936 g was measured (see Supplementary Table S1 for detailed information on individual item weight). Any item removed that weighed >1000 g was considered as 'mega plastic'. One fishing buoy weighing 1832 g, one large fishing net categorised as 'tangled rope' weighing 1936 g, and one piece of sponge at 1894 g, fell into this category, and alone accounted for 6.6% of the total items weighed.

Mainland vs. island coasts
To quantify the spatial distribution of beached plastic litter relative to the remoteness of each survey site, the data were separated by island (remote) or mainland (urban) survey sites. On the mainland, the average count of plastic items was 211 ± 239 NP/100 m and an average weight of 2457 ± 3589 g/100 m. Across the islands, there was an average count of 1229 ± 2705 NP/100 m, and an average weight of 9949 ± 19,305 g/ 100 m (Supplementary Table S2).
Two islands, Skye and Mull, were of particular interest due to their reported high quantities of beached litter. The average count on Skye was 3426 ± 3901 NP/100 m, and similarly on Mull, with an average count of 2942 ± 4496 NP/100 m (Table 2; Fig. 4A). In terms of weight, plastic items removed from Skye measured 31,405 ± 36,991 g/100 m, while from Mull a weight of 11,015 ± 17,050 g/100 m was recorded (Fig. 4B).
These observed high quantities primarily stem from windward coastal survey sites on each island; Talisker Bay located on the west coast of Skye, and Black Beach located on the northwest coast of the Mull (see Supplementary Fig. S2 for a visualisation of their geographical location). Talisker Bay (Skye) had a total count of 13,531 litter items, yielding an average count of 6766 NP/100 m. The total weight for this survey site was 124,601 g, and on average the loading by weight was 62,301 g/100 m. Similarly, Black Beach (Mull) had a total count of 17,480 litter items yielding an average count of 8740 NP/ 100 m. The total weight was 63,864 g, and on average the loading by weight was 31,932 g/100 m. Overall these survey sites alone represented 77% of the count data, and 68% of weight data and were determined as beached plastic litter 'hotspots' within the dataset.

Windward vs. leeward coasts
To quantify the spatial distribution of beached plastic litter relative to coastal orientation, the island data were analysed separately for windward coasts (i.e., the prevailing wind direction is directed onshore) or leeward coasts (i.e., the prevailing wind direction is directed offshore).

Table 2
An overview of the average loading of plastic litter items per 100 m of coastline processed by count (NP/100 m) and weight (g/100 m) of all items recovered from all surveys within the defined 8 geographical locations (n = number of beach-clean surveys and SD = standard deviation).   Across windward island survey sites, the average plastic count was 1859 ± 3189 NP/100 m, with an average weight of 14,862 ± 22,601 g/ 100 m. In contrast, leeward island survey sites were found to have quantities two orders of magnitude less, with an average count of 32 ± 35 NP/100 m, and an average weight of 738 ± 988 g/100 m (Fig. 5A-B; Supplementary Table S3). Hence, beached loadings on windward coasts were in general 58 times greater by count and 20 times greater by weight than beached loadings on leeward coasts.
Supplementary Table S4 describes in detail the observed beached litter quantity and composition for each survey site and in particular Talisker Bay located on Skye and Black Beach located on Mull. See Supplementary Fig. S3 for images of Talisker Bay and Supplementary  Fig. S4 for images of Black Beach, taken from the field.

Input source characterisation
In terms of plastic litter composition across the full dataset, by source type, marine-sourced litter had the highest count and weight at 46% and 62%, respectively (Fig. 6, 'All'). The same was found across the island dataset, with 48% of count data and 64% of the weight data represented by marine-sourced litter (Fig. 6, 'Island'). However, for mainland data land-sourced litter was dominant by count at 47%, compared with marine-sourced litter only representing 27% (Fig. 6A, 'Mainland'). Conversely, in terms of weight, marine-sourced litter accounted for 55% of the data, whereas land-sourced litter only accounted for 14% (Fig. 6B, 'Mainland').
The composition of plastic litter was found to alter spatially relative to the remoteness of the selected survey sites (i.e., between remote island sites and urban mainland sites) (Fig. 7). Across the island survey sites (Islay, Tiree, Mull and Skye. See Fig. 1 for their geographical locations), a greater percentage of marine-sourced litter was observed, by both count and by weight. For example, on Tiree, 51% of the total plastic litter count was marine-sourced, and only 17% land-sourced. The same was evident by weight, with 67% marine-sourced, and only 9% landsourced. Whereas, across the mainland survey sites, the opposite was observed, for example, within the Clyde Sea, 63% of the total plastic litter count was land-sourced compared with only 16% marine-sourced. In terms of weight, the same was found, with 32% of the data landsourced, and 18% marine-sourced (Supplementary Table S5). A descriptive account of source type for each location discussed here is provided in Table 3. However, the composition of plastic litter did not alter spatially relative to coastal orientation (i.e., between windward and leeward coasts). In the case of the island survey data marine-sourced litter represented the greatest percentage of the dataset for both windward and leeward coasts (Fig. 5C-D; Supplementary Table S6).
The largest sub-source of beached plastic litter across the full dataset was 'fishing' with 41% by count and 58% by weight (Fig. 6, 'All'). For island data, 'fishing' was also the highest sub-source accounting for 43% and 60% by count and weight, respectively (Fig. 6, 'Islands'). The composition differed for the mainland data with 'public' litter found as the highest sub-source by count (Fig. 6A, 'Mainland'). However, 'fishing' remained the largest sub-source by weight at 48% (Fig. 6B, 'Mainland'). The percentages of each sub-source are presented in Supplementary  Table S7.
Looking at each individual plastic litter category present in the dataset, by count, the most common items were fishing nets (0-50 cm) at 19%, fragments (0-2.5 cm) at 16%, followed by fragments (2.5-50 cm) at 14% (Fig. 8A). By weight, rope (>1 cm) at 21%, fragments (2.5-50 cm) at 16%, and rope (<1 cm) at 12%, were three commonest items (Fig. 8B). The difference in the most common items between count and weight, shows the discrepancy possible between these measures. A list of the representation of all categories is provided in Supplementary Table  S1.

Spatial regressions
The percentage that each source type (i.e., land-sourced, marinesourced, or of an unknown-source) represented at each survey site was calculated relative to water distance (km) from the principal land-based source, the Clyde Sea. Across the beach-clean surveys, based on the absolute number of items recovered, by count and by weight, there was no statistically significant difference in plastic litter beached-loadings (Table 4). In terms of composition percentage (i.e., the percentage of each sample that was land-sourced, marine-sourced or from an unknown source), trends were observed as water distance increased from the Clyde Sea to remote and less populated areas of little industrialised activity (Table 4). The percentage of land-sourced plastic litter recovered decreased, whereas the percentage of marine-sourced plastic litter recovered increased by both count and weight (Fig. 9). For items of unknown source, there was no statistically significant difference between sites relative to water-distance from the Clyde Sea ( Fig. 9; Table 4).

Discussion
The purpose of this study was to investigate the spatial abundance, distribution, and composition of plastic litter on the west coast of Scotland. Of particular focus was to identify 'hotspots' and 'coldspots' of beached litter to validate predictions from a particle tracking model developed for the region. The model predicted a higher deposition of beached litter on windward coasts, rather than on leeward coasts. In the current model, this dynamic is driven by a combination of the northerly coastal current and the prevailing south-westerly winds in the region. Future models will also incorporate the effects of tide, wave and storm surge height as decribed by Turrell (2018).
Overall, the dataset presented in this study suggested that on average the Scottish west coast had 813 NP/100 m of surveyed coastline. In comparison, a previous analysis of available data for the region, collected as part of the Marine Conservation Society's citizen-science beach-clean monitoring program, found there were on average 380 NP/100 m of surveyed coastline (Allison et al., 2022). The difference Fig. 6. Plastic beach litter composition for all survey sites, island (remote) survey sites and mainland (urban) survey sites. Inner plots: source type (i.e., land-sourced, marine-sourced, unknown-sourced). Outer plots: sub-source (i.e., public, sewage, medical, fishing, shipping, non-sourced and fragments) by (A) count data and (B) weight data. observed between these results suggests previous evaluations could have underestimated the beached loadings, and thus, demonstrates the importance of the optimisation, consistency, and standardisation of beach-clean monitoring efforts as beached loadings exhibit large spatial variability. This is supported by Nelms et al. (2022), in which they state that data sharing is essential to avoid beach-cleaning of the same area, and consequently skewing data. Galgani et al. (2021) also report that more coordination is required before reliable estimates can be made on trends of plastic litter.
To our knowledge, currently, there is no documented dry-weight data for beached plastic litter on the west coast of Scotland. For all the survey sites documented here, there was an overall average of 6434 g/ 100 m of surveyed coastline, with an average dry weight per item of 7.9 g. It is common practice to report input estimates by mass (e.g., Jambeck et al., 2015), therefore a record of dry weight is important to record for any region, in order to convert from number of litter items to mass of litter items for future input estimates.
Island coastlines have been found to have a higher deposition of plastic litter, which is primarily of non-local origin, and is mainly due to oceanic transport (e.g., Thiel et al., 2013;Perez-Venegas et al., 2017). As there were apparent differences observed between the mainland and island datasets (Supplementary Table S2), separate monitoring and data Fig. 7. Percentage contribution of plastic beach litter from each of the three input source types (i.e., land-sourced, marine-sourced, and unknown source) represented graphically by (A) count data and (B) weight data, in the 8 defined geographical areas. See Fig. 1 for geographical area names. The source types are represented by greenland-sourced; bluemarine-sourced; and greyunknown source. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Table 3 An overview of the average loading of plastic litter items per 100 m of coastline divided into three source types (land-sourced, marine-sourced, unknown source) processed by count (NP/100 m) and weight (g/100 m) of all items recovered from all surveys within the 8 defined geographical locations (n = number of beach-clean surveys and SD = standard deviation).
Here we report that island coastlines on the west coast of Scotland had an average beach loading of at least an order of magnitude greater than mainland coastlines. The higher values observed demonstrate the variability in beached loadings between remote coastlines that are exposed and less populated, with urban coastlines that are densely populated and industrialised areas. Greater beached loadings on remote coastlines could be a result of less beach-clean activities because of harder to reach survey sites, therefore developing an accumulation over time. Alternatively, these areas are more exposed to local oceanic sources of litter, e.g., from local marine industries.
In addition, Asensio-Montesinos et al. (2021) found that on remote beaches storms were one of the main drivers of floating plastics onshore, including large meso-plastic items, and that the burial, trapping, and degradation of litter items occurred more on pebble beaches. Hengstmann et al. (2017) also observed that beached litter items were trapped by vegetation or pebbles. They report at all survey sites more litter items were found in coarse sediments and vegetation at the back of the beach, rather than at the high tide line. Coarse sediments can act as a depression that traps litter items preventing them being dispersed by wind, and therefore accumulation occurs in these regions. Furthermore, results from a mark-recapture study of beached litter items found that there was no correlation in deposition and retention events between surveyed beaches (Solbakken et al., 2022). This further suggests that these processes are primarily determined by local site-specific physical factors e. g., wind direction and speed, wave, tide and storm surge height, and currents in combination with individual beach characteristics e.g., beach profile, sediment type, and exposed or sheltered coasts (Andriolo et al., 2020). Schmuck et al. (2017) also found higher plastic quantities on less accessible beaches, and suggested this could be due to less frequent, or non-existent beach-clean monitoring. Lavers and Bond (2017) reported that on the remote, uninhabited Henderson Island in the South Pacific, the density of plastic litter was the highest reported of any other study, with up to 671 items/m 2 . It was therefore suggested that this island, and many others that may be unreported, could be acting as a sink for the increasing inputs of plastic litter into the environment.
Examples of such sites observed within our dataset would be Talisker Bay on Skye and Black Beach on Mull ( Supplementary Fig. S4). Both are south-west facing (windward) coasts, of pebble substrate, and received beach loadings of at least one to two orders of magnitude more than all other sites by count (range: 3-8740), and at least two to three orders of magnitude greater than all other sites by weight (range: 60-62,000 g; Supplementary Table S4).
On the mainland (windward) coasts of the west coast of Scotland, this study found on average there was 211 NP/100 m and 2457 g/100 m of surveyed coastline. However, in comparison, Smith and Turrell (2021) calculated that on the east coast of mainland Scotland (leeward) there was an average of 18 NP/100 m and 279 g/100 m of surveyed coastline. Similar results were found for island coastlines, with windward coasts found to have beach loadings, by count and weight, of at least two orders of magnitude greater than leeward coasts ( Fig. 5A-B; Supplementary Table S3). These results coincide with previous studies that suggest that exposed coastlines, subject to onshore winds, received higher beached loadings (Critchell and Lambrechts, 2016;Kaandorp et al., 2021). Allison et al. (2022) reported that for the same study region a particle tracking model predicted high deposition of plastics on windward shores. This information could enable a selection of sites by beach-clean monitoring programs or governmental organisations, that could be targeted to achieve maximum impact of litter removal.
The importance of analysing the data by the characteristics of their geographical location (e.g., population density, industrialisation and accessibility) is further emphasised through source type. Fig. 6 demonstrated the variation between urban mainland survey sites and remote island survey sites, with mainland found to be dominated by landsourced litter, and islands found to be dominated by marine-sourced litter. However, on the mainland, by count, the data was dominated by land-sourced litter at 47%, but by weight the data was dominated by marine-sourced litter at 55%. A similar observation was reported by Smith and Turrell (2021), further emphasising that both metrics are important when interpreting results.
When the dataset was divided into sub-sources, the greatest contributor of land-sourced litter was found to be public litter. Turrell

Table 4
The linear model results for both count data and weight data, describing the goodness of fit, and the significance of explanatory variables (i.e., distance from the principal land-based source), against the response variable (i.e., quantification of plastic items by source type). Statistically significant values are highlighted in bold. (2020a) reported that public litter was the largest input into the River Clyde. In the case of marine-sourced litter, the greatest contributor was fishing waste. Other studies have also documented that beached plastic litter on remote island coastlines are likely to be from fisheries activities (e.g., Unger and Harrison, 2016;Lavers and Bond, 2017;Nelms et al., 2017;Perez-Venegas et al., 2017). The composition of plastic litter was previously reported to vary relative to the proximity to the source (Kaandorp et al., 2021). While the quantity of beached loadings does not change relative to the principal land-based source, in this case, the River Clyde, the composition of litter does differ the further away the survey site is from the principal landbased source. This also further clarifies the differences in composition observed between urban mainland survey sites and remote island survey sites to which mainland data (close to the principal land-based source) were primarily land-sourced, and island data (further from the principal land-based source) were primarily marine-sourced.

Conclusions
This study looked at a region (the west coast of Scotland) with a complex coastline and many islands, but a single major urban area (Glasgow and its surroundings) that was presumed to be the major landbased source of litter to the marine environment (via the River Clyde). Beach surveys, however, showed that this single land-based source does not solely account for the observed loading and composition of plastic litter. In particular, marine-sourced litter was often dominant on beaches that were remote from the land-based source.
The data demonstrates that, to fully understand the extent of plastic pollution, an intensive assessment of remote coastlines is required. However, in areas such as the west coast of Scotland, where site accessibility is often limited, data collection can be restricted. In addition, although the efforts of citizen-scientists are documented to be successful (e.g., Nelms et al., 2017), volunteers are typically concentrated near urban areas, due to accessibility, costs, and transport etc. Therefore, data on the quantity and composition of beached plastic litter can potentially be biased.
By gathering information from accessible coastlines, through various field methods and using this information to verify particle tracking models, the models have the ability to identify remote litter 'hotspots'. In doing-so, it is hoped that key resources (i.e., personnel, transport) can be used in an efficient way to maximise our understanding of marine plastic pollution as well as optimising its removal and prevention.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Data will be made available on request.