Environmental and health impacts of electric service vessels in the recreational boating industry

The recreational boating industry in the United States has grown continuously since 2011, and participation in boating activities is also increasing among Americans (NMMA 2019). While exerting a positive effect on coastal economies (Stoll et al. 1988), the rise in popularity of recreational boating in the United States and worldwide presents numerous challenges in managing air and water pollution, preserving aquatic habitats, and safeguarding access to waterways. Underlying all of these is the need to control the release of untreated human sewage produced on recreational vessels. The uncontrolled release of human sanitary waste has numerous detrimental effects on aquatic ecosystems, including nutrient loading and eutrophication (Vargas-González et al. 2014; Lapointe et al. 2015), the formation of toxic algal blooms (Lapointe et al. 2015; van Beusekom 2018), and the dispersal of human hormones (often in the form of prescription drugs) that disrupt fish development and reproductive physiology (Hallgren et al. 2014; Adeel et al. 2017). Untreated sewage release can also have negative effects on human health by exposing swimmers and beachgoers to pathogens (Soller et al. 2014; Gregory et al. 2019) and inhibiting access to coastal resources (Betancourt et al. 2014).

Because of the damage that unregulated sewage discharge can cause to natural environments and human health, recreational and commercial vessels operating near United States coastlines are required by law to ‘pump out’ untreated sewage from their holding tanks into land-based sewage treatment infrastructure (33 USC 1322; 40 CFR 25). To do this, recreational boats may dock at shoreline-based stationary facilities. Alternatively, pump-out boats, which navigate within and between marinas to collect waste from other vessels, offer a versatile solution for managing the sewage generated from recreational boating activities. In the United States, federal funding for state- and locally-administered pump-out programs, allocated through the Clean Vessel Act (CVA) and managed by the United States Fish and Wildlife Service (USFWS), provides support for the operation and maintenance of pump-out boats (16 USC 777c).

Most pump-out boats are the size of a small- or medium-length recreational vessel, and virtually all are powered by gasoline engines. These engines contribute to air, water, and noise pollution around marinas; to greenhouse gas emissions and ozone depletion in the atmosphere; and to the detrimental burden of recreational boating on aquatic ecosystems (Lloret et al. 2008; Hansen et al. 2019). Electrically-powered pump-out and other service boats could provide a more sustainable alternative to gasoline-powered vessels. However, the environmental and health effects of operating electrically-powered recreational service vessels have never been quantified. Furthermore, because many service providers in the United States recreational boating sector are federally funded and publicly owned (including CVA-funded pump-out boat operators), the influence of policy and regulation on the adoption of low-impact boating technology needs to be explored.

In this study, the environmental, health, social, and policy impacts of converting to electric pump-out service vessels were assessed. Life-cycle assessments of electrically-powered and gasoline-powered pump-out boats were analyzed, and national surveys of pump-out boat operators and state-level CVA-funded pump-out boat program coordinators were conducted. The network of CVA coordinators served to gauge attitudes towards electric technology adoption for the recreational boating service sector as a whole. Boats that provide public services other than recreational pump-outs – such as those that monitor water quality, collect and control waterway pollution, and patrol high-use waterways – may experience similar barriers and benefits to adoption as do the pump-out vessels studied in this work. The multifaceted approach presented here enabled the identification of numerous challenges and opportunities that exist for widespread adoption of electric technology in the recreational boating service sector.

Two surveys were developed to assess the social and policy impacts of electric technology adoption for recreational service vessels. One was formulated for state-level coordinators of the CVA (50 CFR 85). The second was developed for operators of individual pump-out boat programs, and was designed to quantify the operating and management costs of pump-out boats. Both surveys included questions aimed at assessing attitudes and perceived barriers towards electric pump-out boats.

The state coordinator survey was distributed via email once weekly for three weeks to a list of CVA state coordinators maintained by the States' Organization for Boating Access (SOBA). Distribution of the CVA state coordinator survey began on February 21st, 2018, and respondents were only permitted to complete it once. In order to recruit respondents for the survey of individual program operators, respondents to the state coordinator survey were asked to provide contact information for program operators in their state. Individual survey requests were then sent to those program operators for whom contact information had been obtained. The first emails to individual program operators were sent out on February 28th, 2018. One follow-up request was sent to operators who did not respond to the initial recruitment email.

A life-cycle assessment (LCA) was conducted using the methodology outlined in protocol 14044 of the International Organization for Standardization (ISO) (Rebitzer et al. 2004; Curran 2013). It was hypothesized that the boat's propulsion method (i.e. internal combustion engine or electric motor using electricity generated from various power grid sources) and the heaviest boat components would dominate the environmental and health profile of recreational service vessels. The LCA's scope was therefore limited to investigating the influences of propulsion method and boat hull origin on the environmental and health impacts of a pump-out boat over its lifetime. Models of pump-out boats were constructed to isolate the effects of these variables (Table 1). The reference configuration for the LCA is an industry-standard pump-out boat with a recycled aluminum hull and a gasoline-powered engine. The modeled electric vessel derives 10% of its electric energy from onboard solar panels and obtains the remainder (90%) from power grid infrastructure. It was assumed that all boat configurations modeled are functionally equivalent; that is, all pump-out boats have the same lifespan and operating constraints, and perform the same number of pump-outs over the courses of their lifetimes.

In partnership with the Connecticut East Shore District Health Department (Branford, CT) and local stakeholders, Pilots Point Marina (Westbrook, CT) designed and constructed the first full-size (7.5 m length overall), fully electric pump-out boat in September 2018, to the Health Department's specifications. The specifications provided by the Marina (personal communication) enabled compilation of a system bill of materials (SBOM) for this vessel, which served as the prototype electric pump-out boat for this study. Using the manufacturers' estimated component lifetimes for the electric boat's propulsion system (onboard solar panels, batteries, and electric motors) enabled inclusion of replacement components in the SBOM for the electric boat. Specifically, it was estimated that half of the boat's 2 m² solar panels, two of its six batteries, and one of its two electric motors would need to be replaced during its ten-year service lifetime.

The electric boat's motor and battery sub-assemblies were modeled by scaling analogous electric automobile components to the final weights of those used in the boat constructed by the Marina (Burnham et al. 2006). A conventional gasoline-powered pump-out boat was modeled in the same manner, replacing electric-power-related sub-assemblies with components for an outboard internal combustion motor. The lifetime energy needs of the electric boat were modeled using operational parameters specified by the Marina, while the lifetime fuel needs of the gasoline-powered boat were estimated from the results of the pump-out boat operators' survey (see Survey development and distribution in the Methods section).

The Sustainable Minds software (http://www.sustainableminds.com/, Cambridge, MA) was used to evaluate the associated environmental discharges from all materials and processes identified in the SBOMs. Sustainable Minds collates data on the environmental discharges associated with the manufacture, use, and disposal of product configurations modeled by the user, and is compliant with ISO protocol 14044 (ISO14044 2016). The software then uses the United States Environmental Protection Agency's (USEPA) Tool for Reduction and Assessment of Chemicals and other Environmental Impacts, version 2.1 (TRACI 2.1) to normalize environmental discharges to a set of predefined impact categories (Gloria et al. 2007; USEPA 2015b). These environmental and health impacts were interpreted in the context of priorities voiced by pump-out service providers through the national surveys.

The CVA state coordinators' survey responses enabled the assessment of attitudes and concerns regarding the adoption of electric pump-out boats. It also enabled the recruitment of individual pump-out program operators for a second survey aimed at quantifying the operational and financial characteristics of pump-out boating across the United States. Many findings from the operator survey (including average days of use per annum of pump-out boats, average annual gasoline usage, and average engine replacement frequency) were incorporated into the LCA of gasoline-powered and electric boats.

During survey distribution (February to March 2018), 65 CVA state coordinators were registered with SOBA. Thirty-two completed surveys were collected from them, yielding a 49% response rate. Twenty-seven states were represented; two responses were received from Illinois, Kansas, Massachusetts, New Jersey, and North Carolina, respectively. Numerical responses from the two respondents in the same state were averaged, and non-numerical responses were counted as separate responses.

Of the 27 states that responded, 16 operated both pump-out boats and stationary pump-out stations (Table 2). The sizes of pump-out boat programs varied considerably, with the number of pump-out boats operated by each state ranging from 1 to 66 (mean: 15) (Figure 1). The number of CVA-funded pump-out programs in each state – including both pump-out boat programs and stationary services – ranged from 1 to 96 (mean: 11). CVA state coordinators leverage many different strategies to encourage recreational boaters to pump out their sewage rather than discharge it into waterways. When advertising their programs and distributing promotional materials, they often focus on the environmental and recreational benefits of vessel pump-outs by emphasizing the potentials for water quality improvements (52% of respondents), for swimming and beach-going (52%), and for recreational shellfishing (44%). Less emphasis is placed on commercial and economic benefits associated with vessel pump-outs (15%), although the presence of human waste in waterways can cause substantial economic losses in coastal regions by disrupting tourism and commercial fishing operations (Rabinovici et al. 2004; Soller et al. 2014).

CVA state coordinators most frequently identified lower carbon emissions (41% of respondents) and reduced air and water pollution (41%) as the primary benefits of converting pump-out boat fleets to electric power (Figure 2(a)). Financial and operating performance concerns dominated the anticipated barriers to adopting electric boat technology (Figure 2(b)). 41% of respondents cited the high implementation cost as a major barrier and 30% cited budget uncertainty, suggesting that year-to-year public funding fluctuations may make such service providers averse to adopting high-cost electric technologies. 30% of state coordinators indicated that using an electric vessel in open water (i.e. between service points at marinas along coastal regions) could present an operational performance barrier, and 30% also cited concern over the boats' charge capacities. Public regulations also generated substantial concerns. 11% of respondents noted that regulatory hurdles could hinder the adoption of electric technology, and two additional state coordinators commented that they could not stipulate the propulsion method of their state-funded boats.

Attitudes towards adopting electric boat technology varied widely among CVA state coordinators (Figure 2(c)). Sixteen respondents answered an optional question asking them to gauge their state's receptiveness towards converting to electric pump-out vessels. Among these, 44% believed that their state would be neutral towards the adoption of this technology. The CVA state coordinator survey results suggest that pump-out boat program managers are aware of the environmental benefits associated with electric technology, but have wide-ranging concerns about the financial and operational performance aspects arising from adoption.

A major concern among pump-out boat service providers is the cost of converting to electric technology. CVA state coordinators most frequently cited the cost of implementation as a barrier to adoption (Figure 2(b)). To assess the basis for this, the operating and maintenance costs of gasoline-powered and electric pump-out boats were compared using data from the national surveys. Currently, operators spend an average of US $5,370 annually per gasoline-powered pump-out boat (Figure 3). US $4,000 of this is devoted to maintenance items associated with using a gasoline engine. The remaining US $1,370 is spent on maintenance items that would also be incurred for an electric boat (including maintenance of the pump-out unit, the application of antifouling paint, and hull repairs). By contrast, the annual operating and maintenance costs of an electric boat with onboard solar panels for supplemental charge was estimated to be US $3,220 per boat per year, or 40% less than the gasoline-powered average. The majority of cost savings comes from the use of electricity (US $400 annually) instead of gasoline (US $1080 annually) as an energy source, as well as from the lower maintenance costs for an electric system (US $1,450 for an electric motor versus US $2,220 for a gasoline-powered engine).

To test the hypothesis that propulsion method influences the lifetime health and environmental impacts of pump-out boats, LCA models of gasoline-powered and electric boats were constructed. The gasoline-powered boat uses an internal combustion engine, and responses from the pump-out boat operator survey indicated that a typical boat consumes an average of 1,360 liters of fuel annually. The electric boat is powered using electricity generated from coal, oil, solar, nuclear, or wind energy, and its annual energy requirements were estimated to be 2,100 kWh. The greenhouse gas emissions per pump-out were assessed for each boat configuration, assuming that pump-out performance was identical for all models.

The LCA indicated that the gasoline-powered reference configuration generates 7.7 kilogram equivalents of CO₂ (kg CO₂-eq) per pump-out, amounting to 37,800 kg CO₂-eq over its ten-year service lifetime (Figure 4). Use-stage carbon emissions for the gasoline-powered boat account for 95% of all carbon emissions over its entire life cycle. Electric pump-out boats powered with non-renewable energy sources – coal and oil – have lower carbon footprints than does the gasoline-powered equivalent at 5.4 and 4.1 kg CO₂-eq per pump-out, respectively. Using renewable electricity, however, leads to dramatic reductions in the boat's use-stage carbon footprint. Electricity derived from infrastructural solar energy reduces carbon emissions to 0.90 kg CO₂-eq per pump-out, and nuclear- or wind-derived electricity leads to carbon emissions of 0.70 or 0.69 kg CO₂-eq per pump-out, respectively. For the wind-powered electric pump-out boat, use-stage carbon emissions account for 54% of lifetime carbon emissions.

These results highlight the profound impact that energy source has on the lifetime greenhouse gas footprints of pump-out boats. To study these effects in real-world settings, the actual per-pump-out carbon emissions of an electric vessel operating in the United States and drawing electricity from regional mixed resource grids were estimated. Using the 2016 version of the USEPA's Emissions & Generation Resource Integrated Database (eGRID) tool, it was estimated that a present-day electric pump-out boat in the northeast (NEWE) region would emit 1.6 kg CO₂-eq per pump-out (USEPA 2015a). 50% of the electricity produced in the NEWE region derives from natural gas (USEPA 2016), whose combustion emits some 50–70% less greenhouse gas per unit of electricity produced than coal combustion (Weisser 2007; Sovacool 2008). Another 3% of the region's electricity derives from non-renewable fossil fuels such as oil and coal, and the remaining 47% is generated from near carbon-neutral sources including nuclear energy, hydroelectric power, and biomass (USEPA 2016). A boat identical to that modeled in the LCA would emit 2.7 kg CO₂-eq per pump-out using electricity generated on New York's Long Island (NYLI) grid, only 30 km from the NEWE-based grid. Within the electricity grid with the lowest carbon footprint in the United States – that of upstate New York (NYUP) — the boat would emit as little as 1.1 kg CO₂-eq per pump-out. Within the country's worst performing electrical grid – the Wisconsin region along Lake Michigan (MROE) — the same vessel would emit 3.5 kg CO₂-eq per pump-out.

The USEPA's TRACI 2.1 framework was used to compare boat configurations over eight environmental and health impact categories (Figure 5(a)) (Gloria et al. 2007; USEPA 2015b). TRACI-calculated impact scores were normalized to the total impact score for the reference configuration. Constructing a gasoline-powered pump-out boat by repurposing an existing hull led to modest impact reductions, while manufacturing a new hull from primary aluminum raised the boat's adverse environmental and health impact by 40%. This is primarily due to the release of carcinogenic compounds attributable to aluminum mining and refining (Wesdock & Arnold 2014). Regardless of electricity source, all electric pump-out boat configurations modeled had lower total impacts than a comparable gasoline-powered vessel. Powering the boat using electricity generated from non-renewable fossil fuels led to smaller reductions in environmental and health impact than powering it using nuclear, wind, or solar electricity.

While all electric boat configurations had lower total impacts than gasoline-powered boats, they performed more poorly than the gasoline-powered counterpart with regard to particular environmental and health effects (Figure 5(b)). Electric pump-out boats powered with non-renewable electricity contributed equally or greater than gasoline-powered vessels to ocean acidification (sulfur dioxide release), respiratory health effects (PM_2.5 release), smog production (nitrous oxide and volatile organic compound release), and CO₂ release during manufacture. For electric boats powered using solar, nuclear, or wind electricity, the greatest impact reductions occurred with regard to ecotoxicity (measured via comparative toxic units for ecosystems), eutrophication (nitrogen release), global warming (lifetime CO₂ emissions), ozone depletion (chlorofluorocarbon release), fossil fuel depletion (surplus energy generated from non-renewable sources), smog, and CO₂ emissions during use. These findings confirm the role that energy sources play in determining the impacts of recreational service vessels on the environment and human health.

The LCA indicated that electric pump-out boat manufacture led to the emission of 55% more CO₂ into the atmosphere than gasoline-powered boat manufacture (Figure 5(b)). To understand the basis for this, the manufacturing-stage CO₂ emissions of electric and gasoline-powered pump-out boats with recycled aluminum hulls were analyzed (Figure 6). The electric boat's lithium-ion batteries played the largest role in increasing its carbon footprint during construction. Synthesizing organometallic compounds, including lithium hexafluorophosphate, for the batteries has a high carbon cost relative to other electric vessel manufacturing processes.

Electrically powered vessels represent a promising opportunity to reduce the adverse environmental and health effects of recreational boating services. However, the feasibility and impact of converting service vessels to electricity remain uncharacterized. In this study, national-scale surveys and LCAs of pump-out boats were conducted to shed light on the environmental, health, social, and policy factors affecting the conversion of service vessels to electric technology. The study was focused on pump-out boats because there exists an accessible national network of pump-out service providers, and models of gasoline-powered and electric pump-out vessels could be generated using available data. Narrowing the study's scope to pump-out boats enabled precise appraisal of the benefits, barriers, and consequences associated with adopting electric technology in the recreational boating service sector. A wide range of attitudes exists among pump-out boat service providers towards the conversion to electric technology in the United States. The LCA demonstrated that electricity-powered pump-out vessels have lower adverse environmental and health impacts than gasoline-powered boats, regardless of electricity source. The analysis also revealed that boats powered using non-renewable electricity sources show only modest reductions in lifetime greenhouse gas emissions relative to the gasoline-powered standard.

The CVA state coordinator survey (Figure 2) indicates that administrators in the recreational boating service sector are aware of the reduced environmental impacts of electric boat technology, a perception confirmed empirically with the LCA (Figures 4 and 5). CVA state coordinators identified implementation costs (coupled with budget uncertainty) as the greatest barrier to entry for electric pump-out boats. Costs notwithstanding, 31% of survey respondents also indicated that their states would be either ‘not receptive’ or ‘not at all receptive’ to adopting electric pump-out boats, suggesting that cultural and policy norms may hinder the transition to low-impact boating technology even if cost concerns are allayed. Motivating this transition may thus require initiatives extending beyond cost reduction or technology development and into policy action.

A national-scale pump-out boat operator survey enabled quantification of pump-out programs' operational characteristics, and estimation of the costs associated with operating and maintaining gasoline-powered pump-out boats in the United States. The operational cost analysis indicated that existing electric pump-out boat technology can address the financial concerns (and concerns associated with budget uncertainty) voiced by CVA state coordinators (US $3,220 per year for electric boats versus US $5,370 per year for gasoline-powered boats) (Figure 3).

The costs associated with decommissioning an existing gasoline-powered pump-out boat and replacing it with an electric vessel are likely too high for many pump-out programs – the electric boat upon which the LCA is based cost approximately US $200,000 to design and build (personal communication). The pump-out boat operator survey, by contrast, indicated that a conventional gasoline-powered pump-out boat costs US $59,600, on average, to replace (Table 2). This replacement cost may be an underestimate of the cost of an entirely new gasoline-powered pump-out boat, since many boat components – for example, navigation systems, safety equipment, and the pump-out unit – can be salvaged and reused. The electric pump-out boat on which the LCA was based, on the other hand, used all new components, and a significant share of its expense came from fixed costs for research and development. It is expected that reduced research and development costs, as well as economies of scale, will lower the price of electric pump-out boat construction in the future.

The LCA enabled identification of important design and operating issues for minimizing service vessels' detrimental impacts on the environment and human health. Holding all other design considerations constant, charging electric vessels using combinations of wind, solar, and/or nuclear electricity would lead to an order of magnitude reduction in the per-pump-out carbon footprint compared to the conventional gasoline-powered standard (Figure 4). This finding is similar to previously reported results on the effects of electricity source on the lifetime emissions for automobiles (Bauer et al. 2015; Onat et al. 2015; Elgowainy et al. 2016; Woo et al. 2017), and emphasizes the dominant role that energy sources play in determining total greenhouse gas emissions of personal-scale transportation. The electric pump-out vessel models were based on the assumption that onboard solar panels generate 10% of needed electricity, with the remaining electricity derived from infrastructural (i.e. grid) sources. Because of this, the use-stage carbon footprint of an electric pump-out boat with no onboard solar panels (i.e. entirely dependent on infrastructural power) is likely underestimated by approximately 10%. However, it should be noted that in the first year of operation of the prototype boat, the solar panels provided well over 10% of its electricity needs (personal communication).

Converting a pump-out boat to electric power leads to substantial reductions in a range of adverse environmental and health impacts, regardless of the electricity source used to charge the vessel (Figure 5(b)). The largest reductions result from the use of low-carbon electricity, including solar, nuclear, and wind power sources. However, conversion to electric power also shifts the balance of environmental and health effects relative to those of a gasoline-powered equivalent. The LCA results suggest that, for electric vessels charged using solar, nuclear, or wind power, the smallest reductions in TRACI-defined impact categories occur for emissions capable of causing ocean acidification, for carcinogens, and for particles that cause respiratory health issues. For low-carbon electricity-powered vessels, environmental and health effects (particularly carcinogen release) arise predominantly from the manufacture of batteries and electric motors, not from use-stage emissions associated with electricity generation. Even a pump-out boat using the current mix of renewable and non-renewable electricity available in the northeastern United States offers a substantial reduction in use-stage carbon emissions compared to that of a conventional gasoline-powered boat (Figure 4). As with electric automobiles, the remaining carbon and other emissions – along with their environmental and health effects – are non-local and occur primarily at electricity generation and electric motor construction sites (Tessum et al. 2014; Holland et al. 2016).

Pump-out boats provide a crucial service for recreational boaters by collecting the sewage generated on vessels that lack onboard waste treatment systems. The work presented here comprises a comprehensive study of pump-out boating practices in the United States. The range of attitudes expressed among pump-out boat administrators suggests that the widespread adoption of electric technology requires addressing the numerous environmental, social, financial, technological, and policy factors that govern decision making in the recreational boating service sector. This is similar to the constraints placed on the transition to electric-powered automobiles (Egbue & Long 2012; Bakker & Trip 2013).

The CVA program coordinator survey showed that awareness of electric technology's environmental benefits is high, but that aversion to its costs and doubts about budget stability hinder its adoption among pump-out boat service providers (Figure 2). Bringing the technology to scale is thus as much an issue of implementing suitable incentives through policy interventions as of further developing a technology that has demonstrated environmental and health advantages over the gasoline-powered standard. Encouraging the adoption of electric technology in the boating service sector – through financial subsidies at the federal (i.e. CVA) and/or state levels – could lead to immediate reductions in the sector's carbon footprint. Doing so would also lead to significant short- and long-term operational cost savings for service vessel operators.

This study demonstrates that sustainable change in the recreational boating service sector will require federal administrators, state CVA coordinators, and local stakeholders to design and demand initiatives that incentivize the adoption of existing electric technologies through partnerships with local boat builders, public relations campaigns, and realignments in federal funding structures. Electric boat technology can grow to become the preferred alternative for pump-out boats, other boating service sector vessels, and recreational vessels themselves. Operators of publicly owned fleets, state and local government administrators, and marina managers should all consider adoption of electric technology as an option to reduce the boating industry's carbon footprint, and its adverse environmental and health impacts, immediately.

MP and RD formulated the study. JC, ER, and JS developed the surveys. CH conducted the life-cycle analysis. CH, JC, and ER analyzed the data and generated figures. CH, MP, and RD wrote the manuscript. All authors reviewed the manuscript, provided feedback, and approved the manuscript for submission.

The Connecticut East Shore District Health Department, which MP directs, received a grant from the Connecticut Department of Energy and Environmental Protection, using funds received from the United States Fish and Wildlife Service, for the purpose of construction of an electric pump-out boat. This boat served as the prototype electric pump-out boat described in this study. A Provisional Patent Application for the prototype electric pump-out boat has been filed for a consortium of the East Shore District Health Department, the U.S. Fish and Wildlife Service, the Connecticut Department of Energy and Environmental Protection, and the Pilots Point Marina.

The Yale Climate Change and Health Initiative (New Haven, Connecticut) supported this work through a grant from the Overlook International Foundation. The United States Fish & Wildlife Service and the Connecticut Department of Energy and Environmental Protection (grant number 2017–190) also supported this work. The Connecticut East Shore District Health Department and its community partners provided additional funding.

We extend sincere thanks to Kate Hughes Brown of the Connecticut Department of Energy and Environmental Protection, and Lisa van Alstyne of the United States Fish and Wildlife Service for valuable input throughout this study and for assistance in distributing nationwide surveys to CVA program coordinators and pump-out boat operators. We also thank Jeremy Maxwell of Pilots Point Marina for providing information regarding the design and construction of an electric pump-out boat; Thomas Swarr of the Yale School of Forestry and Environmental Sciences for expertise on life-cycle assessments; Brianna Weller of the East Shore District Health Department for providing information on regional pump-out boat programs; and Libby Yranski of the States' Organization for Boating Access for assisting with survey distribution.

All relevant data are included in the paper or its Supplementary Information.