Marine debris is a critical problem to the safety and health of global marine environments. It is defined as any persistent, manufactured or processed solid material discarded, disposed of, or abandoned, in the marine and coastal environment1. Studies have shown that marine debris can affect the income of a region2, the safety of marine vessels3, survival of marine species4,5,6, and change the physical characteristics of the surrounding local marine environment7,8. The quantity of debris in the aquatic environment is predicted to continually increase due to a combination of poor disposal practices and rapidly increasing plastic production9.

Debris has been observed in a wide range of environments including estuaries10,11 and along beaches12,13,14. Estuaries are a key ecosystem for aquatic species as they provide habitat for spawning grounds and can act as sanctuaries for a range of species15,16. Estuaries within Australia are identified as threatened ecological communities17. In August 2003, marine debris was identified as a key threatening process under the Australian Environment Protection and Biodiversity Conservation Act 1999 (EPBC Act)18. However, relatively little research has been undertaken on shoreline debris found in Australian estuaries.

Most of the major cities in Australia are situated in coastal areas, within the catchment of an estuary. In 2011 it was estimated that 82% of Australians lived within 50 km of the coast19. The shores and catchments of these cities are heavily urbanised. Urbanisation alters waterways and overland flow through the installation of drains, roads and seawalls20,21. Household pollution and stormwater drains also contribute to shoreline debris11,22,23. However, terrestrial waterways are the main transport processes by which debris is transported from the land to the coast24,25,26. To date, no comparisons between coastal and estuarine debris have been presented in the scientific literature, and the relative importance of the various sources has not been explored empirically.

We carried out a scientifically rigorous survey and corresponding analysis comparing estuary and coastal shoreline debris. Our objective was to determine what are the anthropogenic factors associated with urbanisation that influence the distribution of estuary and coastal debris. We addressed the following questions:

  • Do we find similar types and amounts of debris in coastal and estuarine sites?

  • Does an increase in local population correlate with higher debris loads?

  • Do we find higher debris loads where there is more infrastructure (e.g. stormwater drains, roads, car parks).

  • What is the relative importance of direct deposition by coastal users, urban runoff and stormwater, and onshore deposition from the marine environment in the amount of debris found on shoreline environments?


A total of 78 transects were surveyed across 26 sites in the Derwent Estuary with an average of 3.0 transects per site. There were 93 transects surveyed across 29 sites in the Tamar Estuary with an average of 3.2 transects per site. Along the east coast 53 transects were surveyed across 12 sites with an average of 4.4 transects per site. A total 751 items of debris were observed in the Derwent, 308 items observed in the Tamar and 72 items observed along the east coast. Of the total debris counts in both estuaries, plastic items were the most abundant material type. Glass items were the next most observed followed by foam, metal, paper, wood and rubber (Fig. 1). Along the east coast glass was the most observed item, followed by plastic, metal and paper (Fig. 1).

Figure 1: The percentage of different types of debris observed in the Derwent Estuary, Tamar Estuary and the east coast of Tasmania.
figure 1

Percentages are of total number of debris items observed in respective estuaries and coast.


A full model, including the terms distance downstream from uppermost sample, side of estuary, estuary basin, percentage of days wind blew onshore over the past 14 months at 1500 hours, stormwater drains within 5 km of a site, minimum distance from a site to a storm water drain outlet, distance to nearest public road, and population within 25 km of a site, captured 45.8% of the variation in the data (based on a deviance comparison, Tables 1 and 2). The model showed debris along an estuary shoreline increases significantly with an increase in the number of stormwater drains within 5 km radius of each site (Fig. 2). The east side of both estuaries had more debris than the west side and the upstream sampling sites had more debris than downstream sampling sites. Debris decreases along an estuary shoreline with an increase in distance from a site to a storm water drain outlet, i.e. the further the distance of the nearest stormwater drain to a site, the less debris will be observed. An increase in debris was found at sites closer to a public road and sites with a larger population within a 25 km radius. Overall, the Tamar Estuary had significantly less debris than the Derwent Estuary (Fig. 3).

Table 1 AIC scores of term combinations for each hypothesis model for the Derwent and Tamar Estuaries.
Table 2 Results of the GAM analysis of the Derwent and Tamar Estuaries final full model.
Figure 2: The effect of the number of stormwater drains within 5 km radius of each site has on total amount of debris observed, in the Derwent and Tamar Estuaries.
figure 2

Each point represents one transect.

Figure 3: The effect of estuary basin on the total amount of debris observed.
figure 3

The Tamar Estuary has less debris than the Derwent Estuary. Along the x-axis, D = Derwent Estuary, T = Tamar Estuary.

East Coast

The full model, including the terms beach aspect, distance to nearest public road, water catchment population, percentage of days wind blew onshore over the past 14 months at 0900 and 1500 hours and if a beach clean-up had taken place in last 12 months, captured 91% of the variation in the data (based on a deviance comparison, Table 3 and 4). The full model (AIC = 89.31) showed that debris along Tasmania’s east coast decreases significantly the further a site is from a public road (Fig. 4). An increase in population within a water catchment showed a slight increase in the amount of debris present at a site; however, this relationship was not significant (Table 4).

Table 3 AIC scores of term combinations for each hypothesis model for the east coast.
Table 4 Results of the GAM analysis of the east coast final full model.
Figure 4: The effect the distance a site is from a public road has on the total amount of debris observed along the east coast of Tasmania.
figure 4

Each point represents one transect.

Importance of sources

The support, based on AIC, for the marine transport model was stronger than the support for either the beach user or stormwater drain model, although all three models were improvements in comparison with a null model (Tables 1 and 3). However, as the final full model had the lowest AIC score of all sub models, all sub models (i.e. different debris sources) contribute to explaining the variation in debris deposition along both estuaries and the east coast.


Overall, we observed less plastic than what has been reported in other similar studies13,27; 40% of all debris detected in the Derwent and Tamar estuaries was composed of plastic. This is markedly less than the 60–80% reported in other estuary debris studies10,28,29. Unexpectedly, in our coastal surveys, only 11% of total items were plastic. This was surprising given previous beach debris surveys in Tasmania found that plastic made up the largest proportion of debris items30,31. Furthermore, an Australia-wide survey found three-quarters of coastal debris observed was plastic32 and globally plastic typically represents 60–80% of debris observed on beaches33.

Glass made up the greatest proportion of debris on the east coast of Tasmania (71% of total items) and the high glass counts reduced the proportion of plastic we observed. As glass (typically found as fragments on our surveys, KW, pers. obs.) does not float, we presumed it arrived via local deposition. We suggest that the high proportions of glass observed on our surveys is likely due to the large number of beach users and visitors in the state. More visitors provide more opportunity for littering to occur. Furthermore, the areas surveyed are comprised of rocky substrate in many instances (50% of transects in the estuaries and 20% of transects along the east coast had a rocky substrate). Glass can readily shatter on the rocks and broken glass can become trapped in crevices, as we observed (KW pers. obs.).

Of the three transport sub-models, we found that the marine transport model was the best model predicting debris deposition in both estuaries and the east coast. Our estuary marine transport model shows that debris entering the estuary upstream becomes deposited on downstream shores by wind and currents. There is a westerly prevailing wind in Tasmania and we found a significant positive influence of onshore wind on debris deposition. These two factors can help explain why we found more debris on the eastern shores of both estuaries.

The inclusion of the stormwater drain and beach user terms in the best fitting overall model indicates both are contributing sources to debris deposition. Stormwater drains have been identified as a major contributor to shoreline debris in Australia23,34. Stormwater drains are an outflow point of litter in urban run-off. We found greater amounts of debris on shores with a greater number of stormwater drain outflows within a 5 km radius, even when we controlled for local population density. The greater number of stormwater drains, the higher probability that litter will exit a drain and be deposited on the shore.

Beach users have been shown to be main contributors to debris along coastal and estuarine shores13,35. The ease of access to a beach is a main factor to the number of beach visitors36. Beaches closer to roads are generally easier to access and have higher numbers of visitors. The estuary and east coast model included a positive distance to nearest public road. This suggests the shores closer to roads experienced more visitors to potentially deposit debris on the beach. We observed nearly three times the quantity of debris along the Derwent than the Tamar survey sites (Mean = 9.6 and 3.3 items per transect, respectively). The Derwent Estuary has a larger urban centre and double the population of the Tamar Estuary, which likely accounts for the difference observed.

Our study of the debris distribution on open coasts and urbanized estuaries provides an overall conceptual model for debris input and movement in coastal regions. Heavily urbanized upper reaches of estuaries have significant inputs of debris from both stormwater and beach users. These transport sources affect local estuarine sites, but also contribute to the pool of debris in the marine system. Wind, current, and wave transport then deposit debris, resulting in concentrated areas of debris in particular portions of the coastal margin. These concentrations are generally highest near the stormwater and beach user transport sources, and steadily decrease moving down the estuary and out onto the open coastline.

Jambeck et al.9 observed that plastic waste entering the marine system was one to three orders of magnitude greater than the reported mass of floating plastic in oceans globally and there is a long standing discussion about where the ‘missing’ plastic entering the oceans resides34,37. The marine transport of debris from upper estuary sources to local downstream shores observed can explain the disparity between land-based inputs and plastic in our oceans, suggesting that the majority of the material may be deposited in coastal regions close to its sources9,38.

Our research demonstrates that local solutions, such as improvements in infrastructure, public outreach (education) and rubbish bins at popular beaches are all likely to decrease the amount of debris on our shorelines39,40,41. Similarly, litter traps on stormwater drains have shown to significantly reduce debris entering the coastal margin. For example, within the Derwent Estuary, litter traps on stormwater drains capture 136 tonnes of litter per year42. Also, stormwater traps were shown to remove 44% of litter before it reaches the San Francisco Bay43.

In summary, local marine transport of debris from upstream sources to downstream shores was found to be the main driver of debris deposition. The types and amounts of debris on estuarine shores differed to what we found in the coastal zone. The difference in debris quantity was influenced by local population size and proximity to urban infrastructure. Shores with large local populations and a high number of stormwater drains had more debris. Additionally, we showed that easily accessible shorelines have more debris, due to their high frequency of beach visitors and subsequent increased chance of littering. These results emphasise the need and importance of targeting debris at the local level.


Survey area

Tasmania is an island south of the mainland of Australia, surrounded by the Indian Ocean to the west, Southern Ocean to the south and the Tasman Sea to the east. Tasmania’s prevailing winds are from the west. Three regions of Tasmania were selected for this research: the Derwent Estuary, the Tamar Estuary and the east coast of Tasmania (Fig. 5). Sampling was conducted between August 2014 and February 2015. Survey sites were distributed along each side of the estuaries, at 12.5 km intervals from a randomly selected starting location removing any bias in site selection. Where selected sites were inaccessible and could not be surveyed, the nearest site that could be accessed was surveyed.

Figure 5: Survey sites in the Derwent Estuary, Tamar Estuary and east coast, Tasmania, Australia.
figure 5

Generated by Kathryn Willis using ArcGIS, [Desktop version 10.2.], ( LIST Coastline (MHWM) from theLIST ©State of Tasmania (Creative Commons Attribution 3.0 Australia License,

Transect surveys

Survey methodology was adapted from Hardesty et al.32. Transect surveys were conducted using stratified random sampling. A minimum of three and a maximum of six transects were conducted at each site. To avoid bias, such as the effect of higher traffic at the access point, the first transect commenced 50 m from the point of entry onto the beach. Subsequent transects were spaced a minimum of 50 m apart. When debris was found in the first three transects, the survey would terminate at the third transect. When no debris was found in the first three transects, additional transect surveys were carried out until debris was observed or a maximum of six transects had been surveyed. Each site was sampled once. To reduce any potential for inter-observer bias a single observer conducted all surveys.

Transects ran perpendicularly to the shoreline, starting at the water boundary and terminating two metres into the terrestrial coastal vegetation. The vegetation types ranged from coastal heathlands to dry eucalypt forests. Transects for swamp or marshland environments ran from the interface between the semi-submerged vegetation and open water, up to 2 m into vegetation on firm ground. All debris observed from a standing position was recorded according to material type and colour. Rubber was recorded separately from plastics, as it is frequently a natural product. To adhere with Hardesty et al.32 foam was also recorded separately from plastics. Seaweed, rocks and large drift wood items were not turned over when surveying for debris. Organic debris was not recorded unless it was identified as human modified (e.g. evidence of alteration by tools [building materials] or similar).

The following variables were recorded at each site: arrival time, number of engineered car parks, weather condition, direction of wind relative to the shore, wind speed, and number of people on the beach (Supplementary Information). The following variables were recorded at each transect: start and end time, transect length, distance from the start of the transect to the dominant debris line, beach gradient, substrate type of beach, colour of the substrate, physical structure of the backshore where the beach meets terrestrial vegetation, beach shape based on 25 m each side of the transect, and the aspect of the beach when facing the water (Supplementary Information). Only debris greater than 5.0 mm in length observed along each 2 m wide transect was recorded.

Geographical Information System analysis

All Geographical Information System (GIS) analysis was performed on ArcGIS (version 10.2). GIS was used to calculate the shortest distance from a site to the nearest public road, the distance from each site to the nearest stormwater drain outlet, population within a 5 km and 25 km radius of each site and the population within the catchment of each site.

Statistical analysis

The R statistical software44 was used for all statistical analysis. Generalised Additive Models, as implemented in the mgcv package in R45 were fitted to the estuarine and coastal data using a smooth spatial component based on a line feature to describe the location from the upper estuary shoreline outward to the open coastal shoreline. For the east coast a separate smooth spatial component using a line feature was used to describe the location from the north-eastern to the south-eastern coastline. In all cases the terms were initially included as smooth components, but then were reduced to parametric terms given their roughly linear estimates. We used a negative binomial distribution to model the count of debris items on our transects.

We tested hypotheses about the sources of debris at each site using three different sub models. The sub model for transport by surface runoff and stormwater drains included the following variables: number of cadastral parcels within a site’s catchment, number of storm water drain outlets in a 5 km radius of each site and the minimum distance from a site to a storm water drain outlet. The sub model for direct deposition by people included population within a 5 km and 25 km radius of the sites, number of people on the beach at the time of the survey, distance to the nearest public road, number of car parks at the access point. The sub model for deposition from the reservoir in the marine system included percentage of days wind blew onshore over the past 14 months at 0900 and 1500 hours, mean speed of onshore wind over the past 14 months, the aspect the shore faced, the estuary basin (Tamar or Derwent) and an interaction term for which the side of the estuary (east or west). The estuary basin and estuary side terms were excluded from the east coast models.

Environmental variables were also included that could affect deposition and re-suspension rates and detection of debris, including beach gradient, substrate type of beach, beach shape. Variables that could potentially introduce sampling error into data such as time since last clean-up, on-day wind speed and on-day weather were also included in our analyses.

Both transport models were run with every possible combination of terms. The Akaike Information Criterion (AIC) scores of each model were compared to determine the most parsimonious model. All terms from each parsimonious model were combined into a single aggregate model. The east coast data was analysed separately from the estuary data to account for the use of a different spatial component and to identify whether sources of debris differed between coastal and estuarine sites.

Additional Information

How to cite this article: Willis, K. et al. Differentiating littering, urban runoff and marine transport as sources of marine debris in coastal and estuarine environments. Sci. Rep. 7, 44479; doi: 10.1038/srep44479 (2017).

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.