Investigating microsized anthropogenic particles in Norwegian fjords using opportunistic nondisruptive sampling

: Norwegian fjord systems provide a host of ecosystem services and are important for recreational and industrial use. The biodiversity of Norwegian fjords has been — and still is — extensively studied since they are important for fishing and aquaculture industries. However, threats from plastic and microplastic pollution within the fjord systems are largely undocumented. Monitoring efforts of microplastic in Norway are limited to coastal biota monitoring, offshore sediments, and some investigations within Oslofjord. Here, we quantify anthropogenic microparticles in Norwegian fjord subsurface waters, including an analy- sis of distribution effects. Fifty-two samples were collected during repeated transits from Bergen to Masfjorden covering 250 km. Anthropogenic particles were identified in 89% of samples, with an average abundance within the fjord estimated to be 1.9 particles m − 3 . This report shows the ubiquitous nature of anthropogenic particles in the subsurface waters of a Norwegian Fjord system. Additionally, methods were validated for opportunistic nondisruptive sampling on-board vessels where microplastics are seldom monitored, including research vessels, commercial freight and transport, and recreational vessels. Further development and implementation of these methods in terms of sampling, chemical characterisation, and long-term monitoring will allow for microplastic quantification and can be easily adapted for worldwide implementation.


Introduction
Microplastic pollution is now regarded as a major global concern that presents threats to ecosystems and biota. These small particles of plastic, <1 mm (Hartmann et al. 2019), have various shapes, sizes, colours, and properties, which account for their patchy, yet ubiquitous presence in all environmental matrices. It has been suggested that most microplastics entering the ocean may be transported from land or directly inputed to the sea (Auta et al. 2017). Irrespective of their origins, once in the marine environment, microplastics will persist and accumulate due to the durable qualities inherent in the original user product. Microplastics and other anthropogenic particles have been identified in ecosystems around the world, which has encouraged regional and national monitoring to investigate the presence of these pollutants.
Microplastics are generally sampled from the water surface using nets, which are limited by their lower size range defined by the mesh sizes (Conkle et al. 2018;Covernton et al. 2019;Stock et al. 2019;Lindeque et al. 2020). Subsurface waters are less frequently investigated, and methods include pumping and Niskin bottles (Prata et al. 2019). As the research field has developed, so has the number of methods available for researchers to investigate microplastics (Cowger et al. 2020). With a lack of harmonised and intercalibrated methods, researchers are limited to the level at which global datasets can be compared (Zhao et al. 2018;Prata et al. 2019). Investigating the water column can be challenging due to varying weather conditions and equipment availability, and common methods can be highly weather and user dependant (Prata et al. 2019). There are currently limited methods available to enable large spatial scale coverage of microplastic abundance that are relatively inexpensive, easy to collect, and simple to use despite often unavoidable adverse weather conditions. Opportunistic nondisruptive sampling, such that samples are collected alongside other monitoring surveys, appears to be a practical way forward. Earlier progress has been made by researchers to utilise seawater intakes on research vessels or other vessels of opportunity (e.g., Lusher et al. 2014Lusher et al. , 2015Kanhai et al. 2017Kanhai et al. , 2018. These methods use vessel's water inlets, which are usually at a depth of 1-6 m, and water is passed through a series of nested sieves. An advantage of this approach is that other analytical parameters-such as nutrients, oxygen, temperature, and salinity-are collected simultaneously to the microplastic sample. Furthermore, data related to vessel speed and environmental conditions including wind speed and direction can also be collected through the same systems. This allows for an extensive dataset of variables to be compiled with a minimum error, which is especially important when methods are being optimised. Research Infrastructure have the potential to play a major role in monitoring microplastics in the future, and baseline investigations have shown promising advancements utilising FerryBox systems (Conchubhair et al. 2019). Two projects are currently focussing on integrating this technology: COMMONSENSE (an EU FP7 funded project developing cost-effective sensors, interoperable with international existing ocean observing systems, to meet EU policies requirements) and NORSOOP (a Norwegian Research Council, NFR, funded project to establish a ship of opportunity-based infrastructure to support marine and atmospheric research). While these projects are maturing, it is fundamental that researchers continue to explore the influence of sampling variables on datasets that can be fed into future investigations.
Utilising validated, simple, and cost-effective methods will allow national organisations to establish datasets that can be replicated on temporal and spatial scales. Norway is currently investigating the use of monitoring tools to investigate microplastics, including the use of bioindicators and sediments (Lusher et al. 2017;Bråte et al. 2018;Møskeland 2018). Although few in number, Norwegian studies have identified microplastics in lakes , rivers (Bottolfsen 2016;Buenaventura 2017), coastal waters (Lusher et al. 2015), coastal sediments (Sundet et al. 2015;Haave et al. 2019), offshore sediments (Møskeland 2018), and biota (Sundet et al. 2015;Bråte et al. 2016Bråte et al. , 2018. A recent report investigated the transport of microfibres into and throughout the Norwegian marine environment using Lagrangian particle tracking methods, with simulations predicting that sediment is a major accumulation zone for microfibres (Booth et al. 2018).
One marine environment understudied for the presence of microplastics are fjord systems. They are found around the world and are a prominent feature along the Norwegian coast, where nearly 1200 fjords occupy more than 90% of the estimated 29 000 km coastline (SSB 2013). Fjord systems and their associated ecosystems are recognised as unique habitats for a wide range of species (Delaval et al. 2018). Historically, the fjords have been an important factor in sustaining human populations, acting as food sources for local inhabitants. Fjords are now also utilised to support industries including aquaculture, providing jobs in local communities situated along the coastline.
Norwegian fjords are monitored for contaminants because of their prominent ecosystem role (Hordaland Fylkeskommune 2015). A few studies noted the presence of microplastics in Norwegian fjord biota and sediments (Bråte et al. 2018;Haave et al. 2019). To further expand fjord monitoring methods and identify habitats and species that may be at risk of microplastic contamination, research into the distribution and accumulation within the water column is required. In this study, we use the validated method originally presented in Lusher et al. (2014) to explore the presence and distribution of anthropogenic particles in the Norwegian fjord system on-board RV G.O. Sars. The aim of the investigation was to quantify anthropogenic particles in the subsurface waters using an opportunistic nondisruptive sampling strategy, categorise particles, and compare levels to studies in other geographical locations.

Study location
Norway's west coast is characterised by fjords and is often known as Fjord Norway. The study location in the region surrounding Bergen, in Hordaland, contains many fjords ( Fig. 1). Fensfjord to the north of Bergen feeds into the North Sea and is roughly 30 km long and 3-5 km wide. Fensfjord divides into two inner fjords: Masfjorden (north east) and Austfjorden (south east). Fensfjord has large industrial and ship traffic and includes Norway's largest port based on tonnage (Mongstad). Masfjord separates from Fensfjord by a shallow sill (75 m) at the inlet. Fjords generally have three distinct water layers, including a brackish, intermediate, and basin layer. The two top layers in fjords from this study are influenced by the Norwegian Coastal Current and the wind patterns along the coast, whereas the basin layer is limited by the sill (Aksnes et al. 2019). The sills also limit the flux of organisms inside and outside the fjords, which is known to affect population structures, e.g., supporting unique subpopulations of deep-water organisms like roundnose grenadier, Coryphaenoides rupestris (Delaval et al. 2018).

Sample collection
Samples were collected on-board RV G.O. Sars during the sampling period of 4-12 October 2014 during a research cruise between Bergen and Masfjorden. The vessel made repeated sampling transects from the outer fjord to inner Masfjorden with locations dependant on the day of the survey (Supplementary data, Table S1 1 ). Our continuous method of sampling was developed and utilised during earlier research investigating microplastics (Lusher et al. 2014(Lusher et al. , 2015. Seawater samples were collected from a depth of 6 m using the vessel's water intake over different time intervals during the research cruise. Seawater was pumped aboard using a IWAKI Magnetic Drive Pump (MDM25 160 ECFF 0221-E2) (Japan), 2.2 kW power (2 bar vacuum) through a hose to covered marine-grade stainless steel sieves (ISO3310-1, DIN-4188, 2 mm and 200 μm) on the rear deck of the vessel. The flow rate was calculated by measuring the time it took for a 1 L bottle to fill with water. Flow rate of the collected water was maintained at 1000 L h −1 . To ensure that there were no deviations between samples, a minimum of 10 flow checks were performed between samples. Sample time ranged from 1 to 12 h. Suspended particulate matter was filtered through the sieve stack for the set period. After the required filtering period, the sieve was removed and rinsed into a filtration tower and passed under vacuum onto filter papers (GF/C, 47 mm diameter, pore size 1.2 μm). This was performed in the ship's laboratory with limited airflow and reduced person traffic. Each filter paper was sealed in a Petri dish and frozen (−20°C) until returned to the laboratory for analysis under controlled conditions. During each sampling period, environmental variables including sea surface temperature, salinity, conductivity, and fluorescence and sampling variables including vessel speed were recorded.

Laboratory analysis
On return to the laboratory, individual filter papers were visually examined under a dissecting microscope (Olympus SZX10 with a mounted Q-imaging Retiga 2000R camera, 10×-20× magnification). Anthropogenic particles were counted, photographed, and recorded using commonly adopted procedures (Lusher et al. 2014(Lusher et al. , 2015. The full protocol has recently been published (Lusher et al. 2020). Particles were assigned to morphology categories (fibres, fragments, bead, and film) and their length (longest dimension in μm) and colours recorded. Categorising particles based on their polymeric structure was not conducted due to being outside the scope of this investigation. Visual identification was carried out following a strict protocol by trained and experienced observers (>3 years experience). Particles were excluded when they had cellular structures or were black and matte as these are characteristics of nonplastic particles (Lusher et al. 2020). As polymer characterisation was not used, all particles are referred to as anthropogenic microparticles (AMPs, defined as human-produced or -modified materials which can include microplastics, e.g., Misic et al. 2019;von Friesen et al. 2020). The visual identification limit of detection is reported as 200 μm reflecting the smallest size sieve used.

Contamination prevention
Cotton clothing and gloves were worn when working aboard to reduce contamination and was rolled with a lint roller prior to sampling to prevent contamination. Cotton and other nonplastic materials have a matte appearance and display ribbon-like folding that allows them to be distinguished from plastic materials. Strict laboratory controls were in 1 Supplementary data are available with the article at https://doi.org/10.1139/anc-2020-0002. place for visual identification as described in previous research including laboratory airborne monitoring using wetted filter papers exposed to air (n = 6) and replicates of filtered MilliQ-water to act as procedural blanks (n = 6). Under MARPOL regulations the grey water (treated sewage) outlet on the vessel is on the forward port side and released on average every 4 h. It should not affect the water collected, as grey water is released on the opposite side of the vessel and behind the intake. Sieves and glassware used during sampling collection were rinsed thoroughly with filtered water between samples.

Statistical analysis
All statistical analyses were performed using R (R Core Team 2019). Data were tested for normality and homology of variance. As neither were normal, nonparametric statistical tests were performed (correlations, linear regression, and comparison of means). Environmental variables were averaged for each sample duration. To describe the influence of external factors on microplastic concentration, variables believed to influence sample results were investigated. After data processing, 47 (92%) samples were used for analysis. Five samples were rejected due to oversaturation of organic matter as sampling time exceeded 6 h. Sample locations were split into different categorical (fjord location, vessel activity) and continuous variables (vessel speed, transect length, sample duration, time of day) to investigate explanatory variables affecting AMP distribution.

Results and discussion
A total number of 52 opportunistic nondisruptive samples were collected during the research cruise in the west of Norway. Initial data analysis identified that five samples should be removed due to high sample volume and collection duration (>6 h). Therefore, the final dataset for analysis included 47 samples collected over nine days during daylight hours (n = 27) and at night (n = 20). The total survey effort sampled 135 000 L of subsurface seawater and covered 250 km. Samples were collected when the vessel was underway (n = 24), stationary, and conducting CTD casts (n = 12) and carrying out including trawling and multibeam echo-sounder (n = 11). Therefore, the average transect length was 4.9 km (range 0-13.5 km).

Anthropogenic microparticle (AMP) abundance in western Nordic fjords
Analysis found AMPs in 89.4% of samples, ranging between 0 and 7 particles m −3 (mean ± SD: 1.97 ± 1.75 particles m −3 , Fig. 1). The location within the fjord system did not significantly affect the number of AMPs found (Kruskal-Wallis, H = 4.79, p = 0.09) nor did the position related to Masfjorden and the sill (Kruskal-Wallis, H = 3.52, p = 0.17). Although there is no statistical difference, samples taken outside Masfjorden appeared to contain more AMPs, which may be indicative of exposure to the North Atlantic. Further, there are high numbers of urban locations near Hjeltefjorden and a decreasing population density towards Masfjorden, thus fewer potential sources of contamination (Fig. 2). Environmental variables (temperature, salinity, conductivity, and fluorescence) and time of day did not significantly affect the number of AMPs identified (data not shown). Increasing sample size and intensity is required to further understand the seasonal variance of AMP distribution within the fjords. These results are different to previous studies showing that environmental variables such as temperature influenced samples in the northeast Atlantic (Lusher et al. 2014(Lusher et al. , 2015. This further supports the argument that sampling should consider all environmental parameters that may influence the number of AMPs found. Opportunistic nondisruptive sampling is therefore beneficial as it allows collection of environmental parameters in parallel to AMP samples.

Influence of vessel operations and sampling time
The specific activity (steaming: >5 knots; trawling: 1-5 knots; stationary: <1 knot) of the vessel did not have a significant effect on the number of AMPs recorded, although vessel speed had a slight correlation to AMPs (Fig. 3, Spearman's: 0.273, p = 0.063), and it appears that there is a clustering of samples that appear above and below 6 knots (Fig. 3). It is noted that the duration of sample collection (1 h, 2 h, 3 h, etc.) was accounted for when standardising AMPs m −3 .
These results are important as understanding the effect of vessel speed on sample collected is fundamental for valid opportunistic nondisruptive sampling. Large data sets are required to accurately map AMPs distributions, and data collected must be independent of vessel operations or accounted for accordingly. Further investigations into the influence of vessel speed and operations is required. These results are similar to previous study  showing that cruises should be assessed individually to ensure samples are not affected by vessel operations (Lusher et al. 2014(Lusher et al. , 2015. No flushing effect of sampling across the sieve stack was observed as sample time did not affect the number of AMPs found per sample (Kruskal-Wallis, H = 0.345, p = 0.55). To be consistent with previous studies, the minimum sample time in this study was 1 h. However, it is recommended that future sampling further tests for flushing effects by collecting samples over shorter or longer durations. Samples collected over durations exceeding 6 h contained large amounts of organic matter that in turn hindered the identification of particles. Thus, they were discounted from final analysis; however, further processing methods can be adapted to use an organic material removal step and should be considered for future continuous monitoring efforts especially when sampling is conducted during or in areas of peak biological activity.

Anthropogenic microparticle (AMP) characterisation
Three different types of particles (fibres, fragments, and foam) were encountered with the majority (count) being fibres (88%); no pre-production resin pellets or beads were found. The size of particles ranged from 0.2 to 16.7 mm (mean: 2.1 mm). Of the total number of particles identified, 17% were <1 mm in length (fitting the same size categories as Hartmann et al. 2019), 79% were 1-5 mm, and 4% were >5 mm. As it was not possible to use FTIR or similar to confirm polymeric identity, all particles were subject to strict visual identification procedures and referred to as AMPs (e.g., von Friesen et al. 2020). Although recent studies support the use of visual identification in the size range of AMPs in this study (e.g., Isobe et al. 2019;Lusher et al. 2020), future investigations should strive to use confirmatory techniques or in-situ quantification (e.g., Edson and Patterson 2015;Cowger et al. 2020). The limit of detection used in this study was 200 μm; further investigation should consider method adaption to target smaller particles as they are considered to be the most abundant, yet understudied component of AMPs (Haave et al. 2019).

Comparison to previous investigations
This investigation documents and characterises the composition of AMPs present in the western Norway fjords. The data collected can be directly compared to four previous investigations of the North Atlantic and Barents Sea (Lusher et al. 2014(Lusher et al. , 2015Kanhai et al. 2017Kanhai et al. , 2018 that used similar methods. The proportion of fibres and fragments are similar for all studies, interestingly this study presented a smaller fibre proportion (88%) compared to previous studies (>94%, Table 1). As the methods employed were similar, the differences in locality to sources and regional distribution could contribute to the observed variationalthough this could not be quantified in this report. These findings using very similar methods on different research vessels show opportunistic nondisruptive sampling is an appropriate method to investigate the number of AMPs in subsurface waters. It is not possible to compare data collected using this method to common surface sampling with nets. Nets are not suitable for offshore monitoring as sampling conditions-sea states and wind speed-can be extremely variable. Nevertheless, subsurface opportunistic monitoring can be used on vessels of opportunity to collect datasets on large geographical scale that can be compared. Further, they can be run alongside other international surveys that are coordinated for sampling across different vessels in parallel such as The ICES International Bottom Trawl Survey (IBTS) in the North Sea (http://ocean.ices.dk/Project/IBTS/).

Future investigations
Additional investigations could be conducted to adapt this method, such as the use of different sieve mesh sizes to account for smaller items. This is in agreement with Desforges et al. (2014) who highlight that more particles were found when using smaller-sized meshes. Automated FTIR or other analytical chemical identification procedures will greatly benefit the information gained from future studies and would require adaptation of this method to pre-treat samples to remove biological material.
It is vital to understand the influence of vessel operations and oceanography on the distribution of microplastics within the marine environment, including surface and vertical mixing of the water column. It would be beneficial to apply a model including other environmental variables that could affect the distribution, such as wave patterns, Langmuir circulation, wind speeds, and swell. Once this is understood, the model would act as a baseline for the pathways of microplastics and their fate in the marine environment. Identified pathways will highlight organisms that can be targeted to understand the effect microplastics can have on individuals and populations.

Conclusion
This study showed that opportunistic nondisruptive sampling is advantageous for collecting anthropogenic particles over large spatial scales. Particles were abundant throughout the western Norwegian fjords and no statistical differences were observed, although increased sampling intensity and targeted sampling may change this. This study documents the ubiquitous nature of anthropogenic particles in a Norwegian Fjord system and suggests that distance from urban locations and sources of anthropogenic input influences the number of AMPs found in samples. Due to the feasibility of opportunistic nondisruptive sampling to be performed alongside other research and monitoring efforts, we recommend that opportunistic nondisruptive sampling is incorporated into existing programs to further map AMPs in the environment. We suggest that this method of study is carried out in other geographical regions to look for a regional variation, and aid in modelling to characterise the spatial distribution, pathways, and fate of microplastics around the globe.