Modelling plastic fluxes with INCA-macroplastics in the Imus catchment: impacts of long-term accumulation and extreme events

Plastic environmental pollution is threatening water resources, aquatic ecosystems, and human wellbeing but is still highly uncertain with global fluxes to sea of 0.4–13 Mt\yr, and up to 517 Mt of mismanaged plastics on land. Catchment modelling tools are required to challenge current knowledge, simulate impacts of management initiatives, and complement global and observation-based studies. Here we present the first spatiotemporally explicit model for mismanaged plastic mobilization and transport from land to sea from the INtegrated CAtchment (INCA) family. INCA-Macroplastics encompasses all components of the catchment, is driven by available data (weather, population, solid waste) and enables calibration and validation against diverse observations (river monitoring, household surveys). INCA-Macroplastics was applied to the Imus River, Philippines, one of World’s most polluted rivers. Given large uncertainties on catchment plastic retention, two calibrations and two emission scenarios were developed to describe catchment plastic fluxes, residence time and stocks over 1990–2020. Plastic fluxes to the sea are highly variable over years and seasons (55–75% exported during the wet season) and have increased exponentially over 1990–2020 from 5–100 to 2000–15000 tons\yr. INCA-Macroplastics is the first model handling plastic accumulation on land and highlights the importance of extreme flooding events in mobilizing and transporting legacy plastics. Model outputs explicitly show that current land plastic pollution can impact fluxes to the ocean for up to 30 years into the future. INCA-Macroplastics is useful to provide tailored recommendations for local monitoring, testing waste management scenarios and pointing towards future research avenues.


Introduction
Release of mismanaged plastic waste (MPW) in the environment is expected to follow a dramatic increase, if mitigation efforts are not rapidly scaled up (Borrelle et al 2020), threatening aquatic ecosystems and human livelihoods (Landrigan et al 2020).Following the rapid economic growth of middle income countries, large booming populations experience increased living standards with increase demand for plastic (Mihai et al 2022).Waste collection infrastructures, when present in these countries, are often not scaled to the increasing plastic waste amounts (Godfrey 2019).Mitigation measures have been introduced to reduce mismanagement and stimulate circularity (Geng et al 2019) but their impacts have been limited thus far (Dauvergne 2018), highlighting the need for more articulated solutions (UNEP 2021).
Huge amounts of plastic, between 0.4 and 13 Mt, reach the ocean every year but extremely large discrepancies still exist between various estimates (Jambeck et al 2015, Boucher and Friot 2017, Lebreton et al 2017, Schmidt et al 2017).At the catchment scale, the uncertainty typically spans two orders of magnitude and is attributed to limited quantitative estimates on plastic waste production and mismanagement, especially in middle income countries (Meijer et al 2021), as well as the lack of knowledge on MPW mobilization and transport pathways from land to-and within-aquatic ecosystems.The stock of MPW accumulated on land and potentially available for transfer to waters is another major knowledge gap.Presumably only 4.7% of the total amount of MPW since the 1960s is currently in the ocean, while the remaining 95.3% is likely stuck on land (Isobe and Iwasaki 2022).An uncertain fraction of this huge stock can be transported further during extreme floods or other catastrophic events (Roebroek et al 2021).
Several studies point at river discharge as a key source of MPW to coastal area (Castro-Jiménez et al 2019, Schirinzi et al 2020), but only one study has provided quantitative plastic flux data during flood events (van Emmerik et al 2023).In fact, measuring plastic during floods presents many logistical challenges.An observation-based study demonstrated (re)mobilization of about 70% of riverbed microplastic during the 2015/2016 winter flooding in the UK (Hurley et al 2018).Hence, extreme floods are likely responsible for a large proportion of total plastic flux to the sea given that large stocks of mismanaged plastic litter are still standing on land.Modeling both land plastic stocks and transport during extreme events can provide useful insights into plastic mobilization and transport from land.
South-East Asia is one of World's most vulnerable regions to natural disasters.Located in the Pacific Ring of Fire and the typhoon belt, many countries rank in the top ten on the World Risk Index, Jha et al (2018).Extreme events, such as typhoons and floods, cause severe material damage, economic losses and are likely responsible for a large proportion of annual plastic emissions.The combination of a high release of plastic litter and high mobilization potential through wind, rain and extreme hydroclimatic events makes this region particularly pertinent to develop models for MPW transport from land to sea.
Most existing riverine macroplastic models focus on estimating plastic emissions from land to sea at the global scale at monthly to yearly resolution with a probabilistic approach (e.g.Lebreton et al 2017, Schmidt et al 2017, Meijer et al 2021).Spatiotemporally explicit models for riverine transport have mainly been developed for micro-and nanoplastics (Nizzetto et al 2016, Besseling et al 2017, Whitehead et al 2021, Domercq et al 2022) pulling knowledge from modeling natural particles that are relatively uniform.In contrast, macroplastics are highly variable in shapes and plasticity implying contrasted behaviors in rivers that are difficult to predict (Kooi et al 2018).Recent modeling efforts provide a better representation of plastic traps within a catchment, i.e. vegetation, riverbank roughness (e.g.Newbould et al 2021, Mellink et al 2022), but these can still not resolve hydrological processes at the catchment scale with a temporal resolution capable of capturing extreme events and do not keep track of all MPW stocks on land.
Most modeling rely on MPW generation rates estimated from national financial and waste statistics (Lebreton and Andrady 2019, Meijer et al 2021, Roebroek et al 2021) which entails high uncertainty and lack of representation of informal practices.To get more precise estimates of MPW stocks and export to sea, and help adapt and update local policies, modeling tools are required to use MPW data at scales of waste collection systems, e.g.regions.Furthermore, MPW is typically described as production rates, neglecting the build-up of temporary MPW reservoirs on land.Models should rely on local waste data and consider build-up on land over time to increase their accuracy.
Here we present INCA-Macroplastics, the first integrated and spatiotemporally explicit model for MPW mobilization and transport throughout a catchment.We applied the model to the Imus River catchment (Philippines; figure 1), likely one of the top 50 most polluted rivers on Earth (Meijer et al 2021).The model is driven by locally gathered MPW data from municipal to regional sources and constrained with a field survey on household plastic waste generation.First, diversity indexes were computed on river macroplastic observations gathered following standard methods to yield information on spatial homogeneity and source (Clayer et al 2021).Modeling was then performed following two calibrations to illustrate two contrasted plastic retention regimes in river systems, i.e. preventing or promoting plastic accumulation in the river.Model outputs are compared to macroplastic observations for ground-truthing.Finally, a sensitivity analysis was performed to identify the main sources of uncertainty and quantify the impact of initially planned mitigation, i.e. stricter implementation of the Ecological Solid Waste Management Act of 2000(RA 9003) on plastic fluxes to the ocean.

Model formulation
INCA-Macroplastics is a semi-distributed catchment model developed within the Mobius framework (Norling et al 2021, Mobius 2019/2022) simulating burial, transport, and (im)mobilization of macroplastics throughout a river catchment (figure 2).Model formulation is described in detail in text S1.In brief, the PERSiST hydrological submodule (Futter et al 2014), estimating hydrological variables such as overland flow, and river shear stress is coupled to a probabilistic box model determining the fate of macroplastics on land and to an in-river plastic transport model simulating plastic retention and export to sea.Lists of model input data, static properties and parameters are given in tables S2-S4.Macroplastics are classified into four types: bottles, bags, containers, and other items, based on their observed relative abundance in river (text S2).Plastic types, and their drag coefficient (C D|i ) and drag area (A i ; table S3), are defined based on their shape which is a determining factor for mobilization and transport, governing the intensity of water dragging (Carvill 1993).
Inputs of MPW of type i (MPW i ; item day −1 ) to soils is expressed as a function of time t (Lebreton and Andrady 2019, Meijer et al 2021, Roebroek et al 2021): where m i (kg item −1 ) is the average mass of a plastic item of type i, f i (unitless) is the proportion of a plastic type i in plastic waste, P is population (cap), SW is the solid waste production rate (kg cap −1 d −1 ), β (unitless) is the plastic content in solid waste and M (unitless) is the fraction of solid that is mismanaged.Population density was assumed to be the only driver of spatial heterogeneity in releases of MPW into the environment.
Processes on land include MPW burial, transport to riverbanks by wind or overland flow with uneven efficiency depending on the land use type (Meijer et al 2021).Buried items can only be remobilized during extreme rain events when surface runoff exceeds a given threshold.For example, the amount of free plastic items in each sub-catchment is given by: where M S→R (items day −1 ) is the mobilization rate, through wind and surface runoff (Meijer et 1(a)) was converted to water discharge using a rating curve established at this station (Sedigo et al 2022).The average discharge at Daang Hari over 2017-2020 was 1.49 m 3 s −1 .

Case-study and model input data
Daily precipitation and air temperature over 1997−2020 are from the CvSU-PAGASA Agrometeorological Station (14.1980N, 120.8835E) and was complemented (1990−1996 and wind speed over 1990−2020) with bias corrected ERA5 data (Hersbach et al 2020).Note that wind ERA5 data did not allow to resolve extreme wind events.Yearly time series of population counts were from the 2015 census for each Barangay (Philippine Statistics Authority) adjusted using World Bank population growth index (World Bank 2022) while time series for solid waste production, plastic content in waste and waste mismanagement were aggregated from various local, regional and national authorities reports (table S2, figure S1 and text S4).Plastic waste production in 2020 was also corroborated with a household plastic waste generation survey performed in three different cities within the catchment (figure S2).In addition, to this baseline emissions scenario, referred to as 'historical' , we modeled the impact of a stricter enforcement of RA 9003, by lowering waste mismanagement by 30% compared to baseline, in the 'RA 9003 enforcement' scenario (figure S1(b) and text S4).
A 5 m digital elevation map, the 2015 land use map and the administrative boundary map (National Mapping and Resource Information Authority-NAMRIA) were combined in QGIS 3.30 to delineate the catchment and sub-catchment boundaries and determine surface area, slope, average distance to river, population density and MPW release within each sub-catchment.

Model calibration and validation
The PERSiST module was calibrated to best predict observed discharge at Daang Hari Bridge (figure 1(a)) over 2017-2020 with Nash-Sutcliffe (NSE) and Kling-Gupta efficiencies (KGE) as objective functions.
Current monitoring technologies yield datasets of river plastic litter that is insufficient to rigorously constrain model uncertainties (van Emmerik and Schwarz 2020).Hence, conventional calibration against observations was not possible.To counter this hindrance, the modeled weight concentration of river plastic litter was manually checked to fall within the range of observations and to follow similar trends along the downstream transect.Monitoring of macroplastic litter in the river was performed during the wet and dry season following standard methods with visual counting and trawl nets (text S2).In addition, model parameters were constrained with literature data (table S4) and then calibrated to allow for seasonal variations following two setups: 'LandAcc' and 'RiverAcc' .These calibrations were used, respectively, to avoid and promote long-term accumulation of plastics in the river to simulate two contrasted plastic retention regimes of river systems.The detailed calibration procedure is described in text S5.

Sensitivity analysis (Monte Carlo simulation)
To evaluate to which parameters plastic releases to sea were most sensitive to, a Monte Carlo sensitivity analysis was performed, sampling each parameter set 200 times, giving a distribution of possible outcomes from the model.One Monte Carlo simulation was run for each of the calibration setups 'LandAcc' and 'RiverAcc' and both MPW scenarios, i.e. 'historical' and 'RA 9003 enforcement' .The parameter included in the sensitivity analysis, their default values, and uncertainty ranges are described in text S6 and table S4.Parameters selected were uncertain or had significant impact on plastic releases.The total effect indexes (Saltelli et al 2010) were determined for each parameter (equation (S24)).

Observed macroplastics
Macroplastic abundances and weight concentrations in the Imus River all show a significant increase downstream (figures 3(a), (b) and (d)) except macroplastic abundance from manual collection during the dry season.Macroplastic abundance was significantly higher during the wet season compared to the dry season for visual counting only (table S7).In contrast, total macroplastic weight picked up in the river was significantly higher during the dry compared to the wet season.None of the other diversity indexes, showed significant differences indicating widespread and homogenous sources of pollution which supports our model parametrization (see equation ( 1)).

Modeled hydrology, plastic fluxes, and seasonal patterns
PERSiST satisfactorily simulated daily and monthly water discharge (table 1; figure 4(a)).In addition, PERSiST captured the extreme surface runoff expected on July 22nd and 23rd 2018 in urban areas (figure 4(b)) when several barangays were flooded up to waist high water level (NDRRMC 2018).
Plastic weight concentrations simulated with the LandAcc calibration along the Imus River are within the range reported for observations showing typical increase downstream (figure 3(e)).In addition, modeled weight concentrations are slightly higher during the dry season than during the wet season for the LandAcc calibration consistent with observations.In contrast, the RiverAcc calibration yielded weight concentrations outside the range of observations and the highest weight concentrations during the wet season.Over 2015-2020, plastic exported to sea was 13.8 ± 50.6 and 24.7 ± 39.2 tons d −1 for the LandAcc and RiverAcc calibration, respectively which is within the range estimated for the Imus River by Meijer et al (2021); 0.8-82 tons d −1 ).Daily fluxes showed strong seasonal variations spanning more several orders of magnitude (figure 5(a)).
MPW can be transported by surface runoff or wind from land to river.Note that extreme winds were not represented in ERA5 data preventing wind to show strong patterns which is a limitation that should be addressed in future model applications.Surface runoff (figures 4(b) and (c)) and discharge (figure 5(c)) showed the strongest seasonality peaking during the wet season.In consequence, the modeled plastic exports to sea were highest in August (figure 5(a)).Hence, as for river discharge, a larger fraction of the annual plastic export (30%) occurred during peak wet season (July-August) consistent with observations from the Mekong River in Vietnam showing similar seasonality (Haberstroh et al 2021a).

Modeled long-term accumulation and sensitivity analyses
In response to increasing plastic usage, waste production and catchment population, the macroplastic litter on land, in river and exported to the sea showed a strong increase spanning several orders of magnitude (figure 6).While most plastic litter ends up buried on land for the LandAcc calibration, with the RiverAcc calibration most of the plastics were exported (table 2).
The total effect indexes from the Monte Carlo sensitivity analysis for the historical scenario show that over-land transport processes (table 3) are the Panel f shows the location map of the monitoring sites (as in figure 1).Observed series showing significant decreasing trends with distance to river outlet are highlighted with a star (Pearson coefficient; P < 0.05).most influential for the LandAcc calibration.In contrast, for the RiverAcc calibration, the detachment from riverbanks is by far the most influential process.These contrasted behaviors show that for the LandAcc calibration, plastics on land represent a large source of litter (figure 5(c)) while for the RiverAcc calibration, plastics on riverbanks represent this primary source.However, the RiverAcc calibration was also sensitive to over-land transport processes where plastics on land represent a secondary source of litter.When plastic emissions are reduced from 2006 onwards, in the RA 9003 enforcement scenario, macroplastic residence times on land for the LandAcc calibration showed a much larger decrease than for the RiverAcc calibration (table 4) and waste mismanagement becomes a key sensitive process (table 3).
In fact, the stocks of free and buried litter on land in the RA 9003 enforcement scenario are, respectively, 43% and 27% lower than in the historical scenario (table 2), making the emissions to the river, and exports to sea more dependable on contemporary inputs of mismanaged waste easily exposed to transport.In contrast to LandAcc, the RiverAcc calibration shows larger decrease in plastic export compared to the LandAcc calibration (table 2).In summary, the RiverAcc calibration represents a catchment where river plastic fluxes are responding fast to changes on land, while the LandAcc represents a catchment with longer retention times on land and little accumulation in the river.

Export during extreme events
The July 2018 flood was well reproduced by PERSiST and the largest export of plastic litter was simulated during that event.An estimated 251 and 124 tons of plastic were released on July 22nd and 23rd, respectively, representing 16% (or 2 month-equivalent) of the average annual plastic flux to the sea over 2015-2020.In addition, the annual plastic flux was 2.6 times larger in 2018 than the 2015-2020 average flux for the LandAcc calibration (figure 6(k)).In contrast, the 2018 annual plastic flux under the RiverAcc calibration was smaller (figure 6(l)) given the smaller plastic stock on land (figures 6(c) and (d)).Analogously, under consecutive extreme events, most of the available plastic stock on land would be mobilized during the first event, leaving less plastics available for mobilization during subsequent events.analyses to identify most influential processes governing plastic transport throughout catchments.Here, the sensitivity analysis points to two processes governing plastic loads variability: overland transport and the detachment from riverbanks (table 3).This is also in line with current research priorities proposed by van Emmerik et al (2023) to focus on quantifying mobilization thresholds, plastic transport velocities in response to varying driving forces, and deposition thresholds.Note that INCA-Macroplastics is designed to take these thresholds as input parameters.Another complementary solution is to apply the Deep Dive approach to plastic litter found in various catchment compartments (Falk-Andersson 2021).This approach has been successfully applied before on beach clean-ups and consists in gathering information directly from collected plastic items, such as expiry date, brands, batch number, product design, helping to determine their origin and age.If applied more systematically, this approach would enable the estimation of plastic residence times and better understand their origin.However, precise estimations of MPW are a complementary and essential source of quantitative information for the successful implementation of the Deep Dive approach into modeling efforts.

On-land versus in-river plastic retention: vegetation as a key
Macroplastic observations in the Imus River show marked but non uniform changes in the composition and abundance of plastic litter found in the Imus River between the dry and wet season (figure 3).Although at the catchment scale there is no significant and clear difference between the two seasons, differences were locally large (tables S5-S7; figure 3).Categories of plastic found were different, and abundance usually show a decrease from the dry season to the wet season (figure 3), i.e. after the first wet season flush.In addition, the diversity and evenness indexes were relatively high and uniform across the sites and sampling seasons (tables S5-S7).These observations support high plastic (re)mobilization across the river system because of random-like dispersion processes and suggest the presence of multiple sources with diverse plastic types (Clayer et al 2021).These observations are also at odds with long-term accumulation in the river and suggest that the LandAcc 26.8 ± 6.4 [9.7 ± 0.5] 97.8 ± 14.4 [36.9 ± 1.9] 15.0 ± 6.5 [6.4 ± 0.3] 0.9 ± 0.5 [0.5 ± 0.1] RA 9003 enforcement 97.5 ± 15.0 [50.1 ± 5.9] 15.1 ± 6.3 [7.9 ± 1.0] 70.4 ± 13.0 [36.6 ± 4.8] 11.5 ± 5.2 [6.0 ± 0.8] 0.9 ± 0.8 [0.5 ± 0.1] calibration is the most appropriate (figure 6).Others have proposed that river environments, in particular floodplains, can act as permanent plastic traps (Calcar andEmmerik 2019, van Emmerik andSchwarz 2020).In the Philippines, the steep landscape and dense shrubby vegetation likely implies frequent trapping, remobilization, and non-permanent trapping in the river.This suggests that river environments behave differently in steep versus flatter landscapes.The role of vegetation is likely the most uncertain and may affect various plastic types in different ways as shown by the contrasted impacts of vegetation.A macroplastic sampling campaign performed in eight Italian rivers showed that plastic shape and size were not significantly different between vegetated and unvegetated sampling sites (Cesarini and Scalici 2022).In contrast, a plastic litter survey performed in Red Sea mangrove forests showed that the distribution of plastic litter in the mangrove was dependent on plastic shape and dimension (Martin et al 2019).

Importance of extreme events
Extreme rainfall events are known to increase riverine loads for, e.g.nitrogen (Lu et al 2020) or suspended sediment (Biggs et al 2022) as well as plastics (Jang et al 2014, Mihai 2018, Roebroek et al 2021).Extreme rainfall events promote soil erosion (Defersha andMelesse 2012, Garcia 2016) and therefore have the potential to remobilize buried litter or plastic components that were still 'in-use' .Since the frequency of extreme rainfall events are expected to increase with global warming (Garcia 2016, Ge et al 2019, 2021), total annual plastic export is also expected to increase due to larger remobilization of buried plastics.
While the current estimate that the 2018 annual plastic flux was 2.6 times larger than the average because of the July 2018 flood is highly uncertain, it illustrates the importance of extreme events in annual plastic loads.Plastic exports during extreme events are challenging to measure.In fact, measuring fluxes in river is bound to provide highly variable estimates because of the variability in hydrology.One solution is to combine bottom-up estimates, i.e. inventories in natural archives, and top-down approaches at the catchment level (e.g.Boucher et al 2019, Clayer et al 2021).Sampling river sediment in Northwest England before and after a major flooding, Hurley et al (2018) showed that the event exported approximately 70% of the microplastic stored on the river beds.These inventories are also needed in various compartments of the catchment to constrain the stocks of plastic on land, in vegetation, or on riverbanks (e.g.Martin et al 2019, Cesarini andScalici 2022).Plastic mobilization during extreme events might also complicate our understanding of the relationship between river plastic load and discharge.In fact, correlations between river discharge and the plastic flux are never above 0.5, and strongly vary between catchments (Roebroek et al 2022).While Roebroek et al (2022) suggest that human behaviors might be driving some of the seasonality and temporal variability observed in plastic fluxes, our modeling highlights that various thresholds for plastic mobilization from different stores are also a likely explanation for temporal variability.In fact, each hydrological event is associated with its own discharge to macroplastic flux relationship (figure 7).Over the period 2010-2020, the correlation between modeled river discharge and plastic fluxes was also below 0.5 for the Imus River (figure 7), even though MPW inputs have no temporal variability except a linear increase (figure S1) and hydrology is the main driver of plastic transport.Hence, the apparent mismatch between river discharge and plastic fluxes can be the result of the combination of mobilization and transport processes with various trigger thresholds.Riverine plastic loads likely depend more on plastic availability than on river flow magnitude, as shown elsewhere (Haberstroh et al 2021b), but plastic availability is likely influenced by human behaviors in intricated ways in addition to wind and surface runoff.In summary, extreme events likely play a major role in plastic export to sea (Lebreton and Borrero 2013, Murray et al 2018, Roebroek et al 2021).However, our ability to predict future aquatic plastic pollution strongly depends on a better understanding of plastic mobilization and retention processes.

Impact of mitigation measures: stricter enforcement of RA 9003
INCA-Macroplastics further provides a conceptual framework in which our understanding of the impact of mitigation measures can be projected.Here, we exemplified the impact of a stricter enforcement of RA 9003 than in practice by lowering mismanagement by 30% compared to the baseline (figure S1(b) and (c)).Results show that stricter enforcement would have had a significant impact on plastic fluxes to the sea, and on land stocks.Due to the relatively longer residence time of plastic estimated with the LandAcc calibration (table 4), compared to the RiverAcc, the impact of improved (>+30%) waste management on export fluxes was limited to about 8% decrease for LandAcc over 1990-2020 (table 2).On the other hand, for the RiverAcc calibration this impact reached 22% over 1990-2020.However, the strongest impact for LandAcc was on land stocks, with a 43 and 27% decrease in free and buried plastic, respectively (table 2).Despite large uncertainty on impact of stricter enforcement of RA 9003, results from both calibrations show that the nearly 30% decrease in plastic emissions does not translate into a comparable decrease in river plastic loads.In fact, in addition to steadily increasing waste production per capita, lag times and legacy stocks of plastic litter on land or in the river delay and attenuate the impact of the measure.
Quantifying the impact of management and mitigation measures is currently bound to high uncertainties related to loosely constrained model parameters and lack of proper calibration and validation datasets.In addition, poor quality and gaps in waste data involve a lack of a credible baseline (Edelson et al 2021).In the present study, waste data was collected at the barangay, or barangay-cluster level and were spatially highly variable and likely vary over time (figure S2).Uncertainties in plastic waste mismanagement are not generally well understood outside of the scientific community; thus, data collection efforts to improve MPW statistics, mostly relying on the nonscientific community, are under-prioritized (Edelson et al 2021).

Conclusion
INCA-Macroplastics is the first spatio-temporally and macroplastic-type specific catchment model able to simulate macroplastic mobilization and transport through various compartments of a catchment.The application of INCA-Macroplastics to the Imus River catchment provides a benchmark study to build on and challenge our current knowledge of plastic transport within a catchment.With INCA-Macroplastics, sources of uncertainty in macroplastic fluxes and stocks throughout the catchment can be quantified.
Modeling results from the Imus River suggest that the catchment's response time can be decades because of unsound disposal of past waste and remobilization during extreme events, in addition to present pollution.However, the residence time of plastic litter within a catchment should be carefully assessed to confirm these findings and a method similar to the Deep Dive approach could be highly valuable (Falk-Andersson 2021).Similar to microplastic pollution dynamics in rivers (de Carvalho et al 2022), our modeling study suggests that urbanization and flood episodes create 'hot spots' and 'hot moments' of plastic pollution while it also highlights a likely complex relationship between hydrology and riverine plastic loads related to mobilization thresholds and intricate pools and fluxes dynamics.
High uncertainty in model parameters, lack of calibration and validation datasets as well as poor quality input data on MPW are currently the largest barriers to widespread application of INCA-Macroplastics.While the two formers lay mostly in the scientific community's field, the latter depends on collaboration with international to local authorities.Legally-binding agreements to address plastic pollution are needed globally, but the most effective intervention options are likely to be tailored to local conditions, including monitoring of MPW (Alpizar et al 2020).In this context, policies efficiencies studies, e.g.Diana et al (2022), will be of great importance to inform modeling efforts and represent the impact of regulatory instruments in various cultural and environmental contexts.In this multi-scale challenge, communication and collaboration will highly benefit from a plastic management modeling tool tailored to local environments such as INCA-Macroplastics.

Figure 1 .
Figure 1.Imus River catchment map showing subcatchments, macroplastic sampling locations, river discharge monitoring site (a), population density in each barangay (b) and land use (c).

Figure 2 .
Figure 2. Flow diagram (a) and illustration (b) of main pools and fluxes of INCA-Macroplastics.Model formulation is described in detail in text S1.

Figure 3 .
Figure 3. Observed macroplastic abundance (a, b and d), weight (c) and weight concentrations (e) as a function of distance to river outlet determined by various sampling methods as well as simulated weight concentrations with default parameters (both LandAcc-empty boxes-and RiverAcc-filled boxes-included) and equivalent discharge conditions and seasons in 2020 (e).Panel f shows the location map of the monitoring sites (as in figure1).Observed series showing significant decreasing trends with distance to river outlet are highlighted with a star (Pearson coefficient; P < 0.05).

Figure 4 .
Figure 4. Comparison of modeled and observed river discharge at Daang Hari Bridge (a) along with modeled surface runoff in urban and natural land use types (b and c).
need for catchment models to tackle plastic pollution Hydrology is one of the dominating drivers of plastic fluxes to the sea (van Emmerik et al 2022).To successfully tackle plastic pollution, interventions should thus be scaled to the natural boundaries of a hydrological catchment.In the multi-disciplinary challenge of quantifying plastic fluxes and their responses to various formal and informal interventions, catchment models represent the best assets to help select the best mitigation strategy and predict its impact.INCA-Macroplastic is such a model, since it encompasses spatiotemporally explicit and plastic-type specific simulations, including on-land and in-river plastic burial, mobilization, and transport.It enables the integration of state-of-the-art hydrological modeling (PERSiST; Futter et al 2014) with recent plastic transport modeling approaches like Newbould et al (2021) or Mellink et al (2022).The strengths of INCA-Macroplastics lay in the fact that it encompasses all components of the catchment, it can handle the representation of strict policy enforcement or other interventions, it is driven by readily available data (weather, population counts, solid waste production rates and composition) and it enables calibration and validation against a range of observations, i.e. river concentrations, counts on land or riverbanks gathered with standardized methods under known hydrological conditions.Spatio-temporal explicit models are key tools to complement observations for the estimation of plastic loads, travel time, and transport pathways as well as to test and improve our understanding of macroplastic transport and deposition dynamics considering weather and river flow variability.Given the accessibility and flexibility of its host, the Mobius framework (Norling et al 2021, Mobius 2019/2022), INCA-Macroplastics can easily be tailored to the expanding plastic knowledge.In addition, the integrated nature of INCA-Macroplastics allows for joint uncertainty

Figure 5 .
Figure 5. Boxplot of daily plastic mass exported (a), wind speed (b) and Imus River discharge (c) grouped by month over 2015-2020.

Figure 6 .
Figure 6.Modeled macroplastic litter on land (buried and free), in river (on riverbanks and in river-vegetation) and exported to the sea following two scenarios (historical and RA 9003 enforcement) and two calibrations (LandAcc and RiverAcc).Note the log scale on the y-axis in panels (e)-(j).

Figure 7 .
Figure 7. Modeled daily plastic export to sea as a function of daily Imus River discharge between 2010 and 2020.Each year is highlighted by a different color.Two extreme events in July 2018 and August 2013 are also highlighted.

Table 1 .
Calibration statistics for river water flow predictions in Imus River.

Table 2 .
Fate of macroplastics in Imus River catchment and final recipient as of 2020.

Table 3 .
Total effect indexes of selected parameters from Monte Carlo simulation for both LandAcc and RiverAcc calibrations.Process and parameters are described in detail in text S1.Parameters selection for Monte Carlo sensitivity analysis is described in text S6.

Table 4 .
Residence time of macroplastics in the compartments of Imus River catchment estimated over 2015-2020.