Introduction

Peri-urban agriculture (peri-UA) provides the benefits of conventional agriculture, such as open-air, soil-based crop production with high yields and extensive fields, as well as the potential to reduce the environmental impacts normally associated with traditional food supply systems through circular nutrient strategies (CNSs)1. Replacing distant linear-nutrient sources, such as traditional mineral fertilizers produced from nonrenewable ores requiring energy-intensive processes2 for local, circular ones, such as nutrients recovered from anthropogenic waste streams in cities, is one of the system-wide transformations required to create a more sustainable urban food production paradigm1,3.

In addition to reducing the carbon footprint1, the increased use of circular nutrients in peri-UA could ameliorate the transgressed planetary boundary of biogeochemical flows for nitrogen and phosphorus4 by maximizing the circulation of nutrients in the anthropogenic system and minimizing their loss to the environment5. Currently, nutrients flow linearly from ores (for P and K) or air (for N) through agriculture and humans6 into waste systems7, ending up in local water ecosystems and causing pollution, such as marine and freshwater eutrophication6,7. In the best cases, waste management systems capture a fraction of these nutrients in the shape of byproducts requiring further treatment, use or disposal, such as compost (19% of global solid waste is composted/recycled8) or wastewater treatment plant sludge. An even smaller fraction is recovered in higher-value products, such as struvite (a P-rich compound), which returns to peri-UA fields9. Therefore, coupling urban waste management systems and peri-UA systems through circular nutrients is thought to help create opportunities to increase agricultural productivity and deliver other societal co-benefits, such as increasing food sovereignty3,10,11, improving health in cities12, helping achieve “net-zero nutrient” cities5 and providing ecosystem services, such as freshwater, food, and fuel, in cities13.

Waste‒food systems linked through circular nutrients have been found to be technically, economically, socially, and environmentally feasible3,12,14,15,16,17,18,19,20,21,22, depending on population size, crop pattern, waste stream composition, and infrastructure of waste treatment facilities. The city crop pattern, i.e., the distribution of specific crops along the peri-UA areas, determines the demand for nutrients and describes the areas where the recovered nutrients will most likely be used3,14 given economic constraints. Therefore, peri-UA areas represent the largest market for circular nutrients. Furthermore, the choice of a specific nutrient removal and recovery (NRR) technology determines the amount, concentration and chemical form of recovered nutrients23, which further affects plant growth, productivity, soil-nutrient product dynamics24, and field emissions to air25,26, and water. Recovered nutrients in the form of sludge, compost, concentrated nutrient products, etc., interact differently with local soil conditions and local ecosystems13,27 when used in the field, in addition to influencing the growth and productivity of crops differently. For example, the application of sludge from wastewater systems with high levels of P is limited to soils with low P contents to avoid environmental problems, such as eutrophication28.

Research has focused primarily on the technical aspects and their environmental impacts of NRR technology22,29. For example, the impacts of the materials and resources used by technologies, their capacity to recover nutrients, and their feasibility in urban environments under economic or technical constraints3,14 have been studied. Fewer studies have investigated the consequential impacts of the transition from linear to circular nutrients in a systematic way12. Integrated assessments of waste and food urban systems to understand the system-level consequences of the transition from linear to circular nutrients are lacking13,16,22. These types of studies provide essential knowledge for policy development and sustainable urban planning, in relation to e.g. United Nations Sustainable development goal (SDG) 11, and could resolve questions such as, what are the cross-scale (e.g., city, region, globe) and cross-disciplinary (e.g., waste management and agriculture) changes in the transition to circular nutrient use in peri-UA10? What are the consequences of the alteration of nutrient flows through the city in waste streams, both avoided or induced, and the replacement of linear sources of nutrients by recovered nutrients that come from constrained sources? What are the consequences of changes in the markets of nutrient products, including the markets of mineral fertilizers, and of incipient markets for circular products obtained from NRR technology? In general, system-level changes due to the upscaled implementation of NRR technology and their link with city systems delivering food should be addressed under a “bigger picture” system analysis22.

In this study, we develop a framework for prospective regionalized life cycle assessment (LCA) of peri-UA to understand the local (direct) and transboundary30 (indirect) environmental impacts of food provision using linear and circular nutrient supply while considering 1) the city conditions, i.e., the geographically explicit crop pattern; 2) the technical specifications of relevant waste treatment systems of the city as well as nutrient recovery potential through NRR technology; and 3) the regionalized impact of nutrients emitted to the local environment, i.e., finely regionalized eutrophication characterization31. We created a python-based tool that is driven by the spatial distribution of crops in a city and accounts for the LCA impacts of linear (mineral fertilizers) and circular products made in the relevant NRR technologies (explained below for the case study). Because NRR technologies are expected to be fully deployed in the future, we include prospective calculations of the impact for linear and circular systems in 2050. For this purpose, two prospective life cycle inventory (pLCI) databases based on global scenarios of socioeconomic development, i.e., shared socioeconomic pathways32 (SSP2 for the case study), are used: 1) the prospective business-as-usual (BAU) scenario, with a temperature increase from preindustrial levels of 3.5 °C by 2100, and 2) the prospective 2-degree scenario (RCP2.6), with a temperature increase from preindustrial levels of 2.0 °C by 2100. System-level dynamics are assessed by comparing the linear supply of nutrients to peri-UA production from mineral fertilizers against a set of CNS scenarios, where nutrients come from NRR technologies in the form of circular products, which for our case study are three: 1) struvite (mostly containing P and smaller amounts of N) recovered from a wastewater treatment plant (WWTP), 2) ammonium salts (containing N) and struvite (as a source of P) both recovered from a WWTP, and 3) compost (containing N, P, and K) recovered from organic municipal solid waste (OMSW). The comparison investigates avoided, additional and substituted activities in waste treatment systems given the full-scale implementation of NRR technologies12 (Fig. 3) and considers that circular nutrients are constrained resources limited by the amount of nutrients in waste streams. This latter consideration means that the total demand for N, P and K cannot be covered by circular nutrients and that mineral fertilizers cover a fraction of the demand in the CNS scenarios. The contribution of each nutrient source per scenario is detailed in Table 1. Finally, the framework is illustrated for the current and projected peri-UA areas of the Metropolitan Area of Barcelona (AMB for its acronym in Catalan) in Spain, as outlined by the AMB urbanistic master plan (PDUM for its acronym in Catalan)33. See Table 1 for all the systems and scenarios analysed in the case study. In addition to presenting the tool and framework, the aim of this study is to quantify the key life cycle impacts of peri-UA annual food production in a city and understand whether circular nutrients can help improve the environmental sustainability of urban food production and quantify trade-offs. The functional unit is the annual food production of crops in the peri-UA areas under different nutrient sources. The impacts of focus are climate change (GWP100), marine (ME) and freshwater (FE) eutrophication, abiotic resource depletion (ADP) and water consumption (WC).

Table 1 Key characteristics of the scenarios analysed in this study
Full size table

Results

Circular nutrients can contribute to more environmentally sustainable peri-UA

We find that nutrient circularity reduces the carbon footprint of peri-UA1 as well as other environmental impacts, depending on the city crop pattern, the circular product or NRR technologies employed and the state of local ecosystems (Fig. 1). For the case study of the AMB, we find that the footprints for GWP100 and ADP are lower per annual food production in all the CNS scenarios than they are in the linear scenario, whereas for ME, FE and WC, some scenarios lead to an increase in impacts. Figure 1 shows that no CNS alone reduces all the impacts considered, which reflects the need to complement various strategies to optimize benefits. Each CNS has trade-offs. For example, phosphate recovery in the form of struvite slightly increases FE compared with FE in the linear scenario (Fig. 1) because large quantities of NaOH and MgCl2 are required for the elutriation process of phosphorus (Supplementary Note 3). NaOH is mostly produced in countries where coal is burned for electricity generation and where coal mining contributes to FE impacts. These impacts outweigh the FE impacts of producing P mineral fertilizers. In contrast, FE reductions of 4% can be expected in the prospective BAU scenario for 2050 because struvite recovery reduces the use of electricity at WWTPs because less sludge is produced and dewatered34, and up to an 8% reduction in RCP2.6 in 205033 because electricity for NaOH production will be less fossil fuel dependent.

Fig. 1: Life cycle impact assessment results for regionalized freshwater (FE_Reg) and marine (ME_Reg) water eutrophication at the Metropolitan Area of Barcelona (AMB), climate change (GWP100), water consumption (WC) and abiotic resource depletion (ADP) for the total functional unit for current and projected peri-UA areas in the AMB.
figure 1

Three circular nutrient strategy (CNS) scenarios are shown in the columns: Struvite, Compost and Struvite + Recovered ammonium salts. Three prospective scenarios are shown in the rows: Current (ecoinvent v3.9.1 cutoff model), prospective BAU in 2050, prospective 2 degree (RCP2.6) in 2050 (prospective scenarios use prospective ecoinvent database based on the IMAGE SSP2 BAU 2050 and RCP2.6 2050 scenarios, created using Premise tool v1.8.2); yr= year; MinFert= mineral fertilizer (linear system). Green check marks indicate that the impacts of the CNS scenario are lower than in the linear system by the percentage shown. Red x marks indicate that the impacts of the CNS scenario are higher than in the linear system by the percentage shown.

Full size image

The application of compost (Supplementary Note 4) has significant benefits between 64 and 71% in terms of GWP100 and greater than 100% for FE, as shown in Fig. 1, but for ME impacts, there is an increase of approximately 3%, as shown in Fig. 1. The AMB’s largest composting site, Ecoparc 135,36, has a mechanical biological treatment plant adapted with a separate line using anaerobic digestion prior to composting, which processes 61% of the solid residues of the metropolitan area. Therefore, a fraction of the remaining residues induces landfilling, and a fraction of the organic residues are composted, avoiding landfilling of organic matter. The balance between the impacts of these two fractions plays a key role in why some impacts decrease and some increase. Moreover, the low concentration of nutrients in compost (Supplementary Note 7) results in little replacement of mineral fertilizers, i.e., approximately 40% for P demand, 3% for N demand, and 6% for K demand (Table 1), slightly varying with the peri-UA areas considered, whether current or projected (Table 1). In the ammonium salt + struvite scenario, the FE impacts are greater than those in the linear system by 17% for current areas and 14% for projected areas and are greater than those in the struvite-only scenario, where the increase is 1% (Fig. 1). This occurs because, in addition to using NaOH and MgCl to recover phosphorus, sulfuric acid is used to recover ammonium in the form of ammonium sulfate, and its production contributes largely to the impacts of the circular product, outweighing the impacts of producing mineral N. Despite the fact that recovered ammonium salts are a concentrated source of N, i.e., 18.8% of the N content, only 43% and 36% of the N demand can be supplied from recovered N in ammonium salts and struvite, respectively (Table 1). WC slightly increases in the ammonium salt scenario, as the production of sulfuric acid is a hotspot for water use (Fig. 1).

The dominance of direct or indirect impacts varies for each circular nutrient strategy

Cities can influence direct environmental impacts due to the local production of food but also transboundary (indirect) impacts30 resulting from food production supply chains, such as mineral fertilizer production37. Using circular products changes the production of nutrients used in the cultivation of food (indirect impacts) and may change the emissions generated onsite (direct impacts). Onsite emissions depend on product interactions with the soil and environment in the field28,38. For example, struvite appears to have a low solubility39; thus, phosphate leaching from struvite is expected to be less than that from P mineral fertilizers, as assumed here, despite the few studies available on the emissions caused by circular products onsite. Furthermore, the status of local ecosystems also plays a role in the impact of emissions. In this study, the regionalization of eutrophication impacts31 reflects the ecosystem specificities of the city and thanks to our spatial analysis we can see where they are taking place e.g. in El Prat de Llobregat area for ME (Supplementary Note 14). The AMB is a coastal city with two key river basins (Llobregat and Besòs) of variable flow where onsite nitrate emissions are very relevant for ME impacts because emissions end up rapidly in the ocean, and phosphate emissions are less impactful given short residence times in surface waters31 (Supplementary Note 12). Bearing in mind all these dynamics, direct and indirect impacts vary for each CNS.

For the AMB linear and CNS scenarios, FE and ADP are dominated by indirect impacts, whereas ME and WC are dominated by direct impacts, and GWP100 has an even share of both impacts (Supplementary Note 13). FE indirect impacts dominate in all the scenarios, indicating that the production of linear and circular products is more relevant than the local phosphate emissions to water from the fertilized soil. The regionalization of phosphate emissions leads to a smaller characterization factor of phosphate emissions to river basins31 than the default factor from the ReCiPe method40 used to characterize phosphate emissions in the supply chain of linear and circular products (Supplementary Note 12). The impact of phosphate emissions in local ecosystems accounts for the short residence time and insignificant sedimentation of P in local basins. In contrast to FE, the ME impact is dominated by direct emissions for all scenarios given the significant nitrate leaching emissions, even in the struvite scenario, because this recovered N source provides only 3% of the total N demand. In contrast to phosphate leaching, nitrate leaching has a greater regional characterization factor in ME than does ReCiPe because N emissions are transported rapidly to the ocean31. Thus, direct nitrate emissions are weighted more heavily on the total impacts than are N-related emissions contributing to ME impacts and occurring in supply chain activities, such as the production of N mineral fertilizers, the production of chemicals used for struvite and ammonium salt recovery, and landfilling, among the most relevant activities. Direct nutrient flows in urban agriculture systems may vary significantly across different locations, which can directly influence the spatial distribution of environmental effects. We capture part of this effect with the regionalized characterization factors. Although, the resolution of the factors is already high, it could be even higher in order to capture, at a much finer level, effects of nutrient flows to the environment. However, this would imply an analysis of the ecosystem almost at a plot level. Furthermore, because we consider no changes in these direct emissions in the future, we see the same behaviour of ME impacts for all scenarios, including prospective ones.

Direct water consumption due to local irrigation also dominates the total impacts for all scenarios in comparison with water consumed indirectly for all processes, such as liquid ammonia production for mineral fertilizer scenarios or sulfuric acid production for ammonium salt recovery. It is of outmost importance for sustainable urban planning to base decisions on spatially resolved water consumption impacts (Supplementary Note 14) to identify hotspots of water inefficiencies, its relation to N2O emissions from leaching, water extraction from basins with water scarcity, etc. Finally, the impact of climate change is mostly due to the direct emission of N2O from the onsite application of mineral N fertilizer, recovered ammonium salts, compost, and struvite. We apply Tier 2 emission factors in the case of mineral fertilizers, ammonium salts and struvite (Supplementary Note 8), but we can expect a significant degree of uncertainty for these emissions, as they depend on interactions of the product with the soil as well as on management practices41. For example, studies have shown that, depending on the crop and soil type, N2O emissions can decrease from 40% to 58% with struvite application42. Field measurements could help reduce these uncertainties for circular products, which are not yet broadly applied in peri-UA. The results for the AMB suggest that the most effective actions to reduce the impacts of ADP, FE and GWP100 should attempt to improve the efficiency and resource intensity of nutrient production technologies when possible, whereas for ME and WC, tackling the field application of products and practices would be more effective in reducing these impacts.

Geo-explicit life cycle modelling aids spatial planning to promote high productivity and lower impacts

The results can be further exploited to investigate the yield‒impact relationships for various crops to aid spatial planning of peri-UA and promote more sustainable productivity. We illustrate this by using a nine-colour palette to contrast the yield with the various LCA impacts calculated by the Python module. Figure 2 shows the GWP100 and ME impacts superposed with the crop yield per plot of the current peri-UA area in the AMB. We select these impacts because they have different contributions of direct and indirect impacts. For ME, direct impacts dominate, whereas for GWP100, direct and indirect impacts contribute more or less equally.

Fig. 2: Map of food production and associated impacts.
figure 2

Map for normalized food production (a) for current peri-UA areas in the Metropolitan area of Barcelona (AMB), Spain. Further this map is overlayed with the maps for normalized impacts of Marine eutrophication (ME) (b), and for climate change (GWP100) in each scenarios except compost (c) and climate change (GWP100) for compost current (d), BAU (e) and 2-degree scenario (f). Food yields and impacts per plot are normalized to the maximum value per parameter. Colour pattern consists of nine-colours organized in a grid from low to high for yield on y-axis and impact in x-axis. Data for food production and impacts is divided into three classes using equal intervals.

Full size image

Figure 1 quickly provides an overview of the plots and crops that are most efficient in terms of producing high yields while generating low impacts. These areas are shown in pink. The compost applied to vegetables and herbaceous crops delivers the lowest GWP100 impacts while providing the highest yields. Figure 2c shows the GWP100 results for the linear system with mineral fertilizers, whereas Fig. 2d, e, and f show the results for the compost scenarios with various socioeconomic systems. The change from purple to pink illustrates the reduction in GWP100 impacts for some areas with vegetables and herbaceous crops. Compost replaces little N, P and K in mineral fertilizer; thus, high yields are expected to be maintained, but its production avoids significant impacts from landfilling residues; thus, the overall impact is significantly reduced. Nonetheless, some plots with cereals and non-citric fruits present an increase in the GWP100 impact when transitioning from the linear to the circular compost system under the BAU socioeconomic scenario, shown as a change from grey to light blue (Fig. 2c–e).

The scenarios of struvite and struvite + ammonium salts show the same normalized GWP100 and ME impacts in relation to normalized yields as the linear system for all socioeconomic scenarios. Therefore, for struvite and ammonium salts, the yield‒impact relationship remains unchanged compared with that of the linear system, but the absolute impact may increase or decrease depending on the socioeconomic scenario. Although this finding was expected, as yields remain unchanged in all scenarios and only the impact change is evaluated, it is important to investigate both the relative and absolute results of the CNS scenarios compared with those of the linear scenario.

Urban planning could benefit from evaluating different crop patterns under the scope of their impacts—yield geo-explicit performance, promoting a pattern where high yields and low impacts are dominant but also bearing in mind that the absolute impacts should be reduced for the different CNSs available to the city. Besides, for further aiding of urban planning, the locations of impacts hotspots are known. For our case study, promoting the use of compost for herbaceous crop and vegetable production around the Llobregat River basin area, and in particular around el Prat de Llobregat, is a clear outcome from the analysis, with the goal of reducing the carbon footprint and increasing the circularity of the food produced in the area.

The future economy largely determines the effectiveness of circular nutrient strategies

The use of pLCI databases to explore future socioeconomic configurations of the global economy highlights the effect of a changing economy on the life cycle impacts of various scenarios, even if the productive process itself remains the same in time (as assumed here). Developments in the electricity and chemical production sectors affect the impacts of the production of mineral fertilizers and circular products. Therefore, their relative performance is also affected by these future socioeconomic developments. Far from predictions, pLCI databases enable an explorative exercise elucidating epistemological uncertainty of what the future could bring43. On the basis of social narratives of the future that are further simulated in integrated assessment models (IAMs) and further integrated into LCI databases44, one can learn from possible life cycle dynamics that emerge while new products are compared with existing products in various future contexts. However, careful interpretation of these results should be performed, given limitations in sectoral, technological and impact coverage and given the early development of pLCI databases45. Furthermore, future impacts should be considered in ecosystems in the future, where water46, resources and the climate will also look different. This aspect is outside the scope of this study, but it is an essential future development needed.

For the case study, the footprints of all circular scenarios are smaller than those of mineral fertilizers for the current and prospective databases (Supplementary Note 5), except for the impacts of WC. This means that nutrient recovery and application in peri-UA areas are always more environmentally beneficial than the use of mineral fertilizer, even in the potential future economy of cleaner energy and production processes. The comparison of footprints calculated with current databases and with prospective databases shows that for a decarbonized economy, i.e., 2-degree scenarios, all impacts for all products decrease, except for ADP impacts. In the case of a BAU economy, some impacts increase (ADP, ME, and FE), whereas others decrease/stay the same (GWP100 and WC). The increase in ADP, especially for fertilizers in the linear scenario, reflects how abiotic resources are becoming scarcer, emphasizing the need for a transition. Characterization factors for ADP are also vulnerable to temporal changes in production and reserves47, although we keep them constant for assessments in the future. In addition, mineral fertilizer production will most likely also evolve. New technologies for ammonia production have been estimated to change the global ammonia market and reduce climate impacts from 11–70% compared with the market mix modelled in current LCI databases2. Such reductions could favour linear systems using mineral fertilizers in 2050 over circular products, but most notably, mineral fertilizer would still be required to cover the demand for nutrients, as shown in all the CNS scenarios.

Discussion

In this study, we present a tool and framework based on prospective regionalized life cycle assessment, which calculates the environmental impacts of peri-UA production for a city with different strategies for nutrient supply. Linear and theoretical circular nutrient systems are considered in terms of their life cycle and as a shift in waste management in the present and in the future. Circular nutrients depend on constrained waste streams, and as such, system-level dynamics induced from their recovery should be carefully considered, e.g., alterations in nutrient flows in the linear system causing new impacts and/or increased demand for the resources necessary to recover nutrients. Our methodological framework helps explore such system-level dynamics in a geographically explicit manner and offers an instrument for urban planning that can be further explored for other cities, with their specific crop patterns, their own portfolio of recovery technologies, and local ecosystem characteristics. With city-specific data, the analysis should be adaptable and reproducible, even for cities with different municipal waste treatment realities, for example, different levels of sophistication of technologies for solid and wastewater treatment. Because the tool builds on the life cycle inventories of these technologies, if city data exist to create an LCA for these technologies, as well as geo-explicit data on peri-UA areas and their nutrient and water requirements, we believe that the tool can be adapted. Some of these data, e.g., crop nutrient and water requirements, could be obtained from the case study we present here, as nutrient and water requirements for specific crops may not vary much among locations with similar climate.

We find that circular nutrient strategies can contribute to sustainable urban food production. Trade-offs depend on 1) the distribution of specific crops in peri-UA areas, which determines the demand and largest market for circular nutrients; 2) city municipal waste treatment facilities, which determine the feasibility and deployment of specific nutrient recovery technologies and the nutrient recovery potential and quality of waste streams; and 3) the state of local ecosystems in the urban environment, which determines the intensity of the impact of emissions from different sources of nutrients used in peri-UA. These aspects should be included in any analysis to understand the most appropriate CNS for each city. In addition, the moment in time in which recovery technologies are expected to be fully functional plays an important role because the state of the future economy, represented here with socioeconomic scenarios, largely determines the effectiveness of circular nutrient strategies (CNSs) in comparison with linear alternatives in the future. The cleaner the economy is, the more challenging it is to perform environmentally better than the linear systems already in place.

For our case study, some CNSs led to more improvements than others in specific impacts, but no CNS alone led to improvements in all studied impacts with respect to a linear system where nutrients come from mineral fertilizers. Direct and indirect impacts depend on the CNS; thus, effective actions to improve each impact are different, with some affecting the supply chain of nutrients and others affecting practices in the field. For the AMB, struvite has the fewest trade-offs and has the greatest benefits for most of the studied impacts. Ammonium salts and struvite recovery from WWTPs increase FE and WC but improve ME, GWP100 and ADP. In the AMB, replacing N mineral fertilizer use is essential to reduce direct and indirect impacts. Because struvite and compost are not concentrated sources of N, they are not the best options for replacing N mineral fertilizer. The use of recovered ammonium salts should address this shortcoming; however, it comes at the cost of the impacts of its own production, such as water consumption for sulfuric acid production, and the constraint amount that can be recovered from the WWTP evaluated. The compost had the best yield‒impact relationship for GWP100, whereas struvite and struvite + ammonium salts had the same yield‒impact relationship as did the linear system for GWP100 and ME, yet with different absolute impacts. Promoting the use of compost in herbaceous crops and vegetables appears to be a clear strategy for the AMB to address the carbon intensity and increase the circularity of nutrients used for local food production.

We have explored the impacts of a transition to nutrient circularity in a geo-explicit manner through the lens of life cycle impacts, ignoring important considerations, such as soil health (e.g., N‒C dynamics and pollutant presence27), biodiversity status, and exceedance of biogeochemical planetary boundaries. Moreover, socioeconomic, ethical, and cultural factors are essential in assessing peri-UA48,49, which we have not included here. For example, social inclusion, job and business creation, engagement of citizens in food systems, improvements in dietary habits, therapeutic and recreational experiences, etc., are important aspects of peri-UA, especially compared with conventional agriculture, as peri-UA offers benefits in these domains that may outweigh environmental trade-offs. These are also aspects beyond the considerations of the environmental LCA not included in the present analysis but that could be included via other methodologies, such as social LCA and life cycle costing in the fashion of a life cycle sustainability assessment (LCSA).

Moreover, this study examined two crop patterns without exploring crop diversification to match nutrient demand and supply. Future research could evaluate relevant crop patterns for the same areas, perhaps with all stakeholders involved, and use the tool as an instrument to find the least impactful pattern. For a complete comparison of current and projected peri-UA, the latter should account for the avoided impacts of producing additional food elsewhere, which could be included in future research.

The Python module has been coded flexibly and is publicly available so that users can include other nutrient sources, such as manure and agricultural residues, while considering available technologies and nutrient recovery potentials under constrained waste streams. Adding environmental impacts, such as ecotoxicity and the emission of heavy metals to soil, could further elucidate other trade-offs relevant for CNSs, such as compost, which may be hampered by the presence of these substances. Cities with different sanitation management practices and infrastructures may require adapted assessments. Local ecosystem conditions must also be considered, as urban soils with high pollutant levels cannot use low-quality circular nutrient sources.

Finally, the assumption of fixed yields, despite the different nutrient sources and emission factors to air and water for circular products, could be improved with onsite measurements to add robustness to the analysis, e.g., evaluation of yield impacts. We provide default values in the tool, which can be a good first approximation for improvements.

Methods

Prospective regionalized life cycle assessment (LCA)

We use prospective regionalized LCA50,51,52,53,54,55 to determine the environmental impacts resulting from the upscaled deployment of nutrient removal and recovery (NRR) technologies and the application of recovered nutrients in the peri-UA areas of a city. In the case study, the regional scope of the LCA is the Metropolitan Area of Barcelona (AMB), in Spain, which has a total land cover of 5,568 ha dedicated to peri-UA and has plans to potentially expand to 25%33. The regionalized LCA54,55,56 allows us to work at high resolution, displaying variability in impacts spatially by crops, which vary depending on location and cultivation practices31 adjusting to local environmental conditions54. For example, we have regionalized characterization factors for marine and freshwater eutrophication while accounting for the characteristics of the water bodies in the AMB31. For other impacts, such as water consumption, we use ReCiPe 2016 v1.03, midpoint (H)40 family of impact assessment methods; for climate change, we use IPCC AR6 GWP10057 values; and for abiotic depletion potential, we use the CML v4.847 family of methods. Moreover, future changes in the impacts of the supply chain of nutrients, e.g., given the decarbonization of various economic sectors, such as the energy sector, which could alter the impacts of mineral fertilizer production, are studied via prospective LCI (pLCI) databases44,45. This is a cradle-to-field LCA although this is not a strict definition for the circular strategies considered.

System boundaries and conceptual framework

The system boundaries encompass the supply and demand of nutrients for peri-UA to compare linear and circular nutrient supply systems (Fig. 3). In the linear system, nutrients are not recycled through the city (Fig. 3, in green), whereas in the circular system, nutrients flow from municipal waste management to peri-UA areas of the city in the form of recovered products (Fig. 3, in grey). The demand for nutrients in both linear and circular systems depends on the crop pattern of the peri-UA areas in the AMB41 (Fig. 3, orange map). The current pattern of crops and the most likely projected pattern of crops are outlined by the AMB urbanistic master plan (PDUM for its acronym in Catalan)33. We have complemented this geographic information with data on crop nutrient contents, crop water coefficients, climate parameters and irrigation system parameters, among other key attributes, described in detail in the Supplementary Note 1. With these data, the demand for nutrients and for water per year per plot for the whole AMB is determined (Supplementary Note 2). Below, we describe these procedures.

Fig. 3: System boundaries for linear (in grey) and circular (in green) nutrient systems.
figure 3

The functional unit is the annual food production within the Metropolitan Area of Barcelona (AMB), Spain. Activities in dashed boxes are equal for the linear and circular systems and are therefore excluded.

Full size image

With respect to nutrient supply, in the linear system, there are no NRR technologies in place, and the provision of N, P and K to peri-UA areas is assumed to come fully from mineral fertilizer products. To represent their production, we use the LCA database ecoinvent V3.9.1 cut-off model58 and the markets for inorganic nutrients for Spain. The results for this system are presented under the MinFert scenario (Table 1). Under the circular systems, a set of city-specific NRR technologies set at OMSW treatment plants and at WWTPs are assumed to be implemented to their full theoretical nutrient-recovery potential. The provision of N, P and K to peri-UA areas is assessed in scenarios where the potential from each technology is studied (one scenario per technology, Table 1). For the AMB case, technologies are selected given their current viability, which is either some level of deployment already or under consideration. These methods include the following: 1) composting of OMSW (compost scenario), which has already been deployed to a large degree36; 2) struvite crystallization from the phosphorus elutriation process at a WWTP (struvite scenario), which is under research; and 3) the ion exchange process combined with a hollow fibre membrane contactor (HFMC) for nitrogen recovery in the form of ammonium salts (ammonium salt scenario), which is also under research. The circular products are struvite, compost and ammonium salts, which are all recovered from existing municipal solid or wastewater treatment facilities.

To account for system-level dynamics, we look into the differences between the linear and the circular systems’ nutrient supply12, which may lead to three situations: 1) activities that remain the same in both systems, 2) activities that are additional in the circular system compared with the lineal system and 3) activities that are avoided or substituted in the circular system compared with the linear system. For activities remaining the same between both systems, we excluded their modelling, as they entail no change. Struvite and circular ammonium salts include the following (Fig. 3, in grey dashed lines): preliminary and primary treatment, secondary treatment and sludge treatment. In the case of compost, the collection of waste is excluded, as this happens in both systems without any change (Fig. 3, in grey dashed lines).

Additional activities in the circular system include those necessary for the implementation of NRR technologies. In the case of struvite and ammonium salts recovered at WWTPs, additional energy, chemicals, and feedstocks and the transport of chemicals and products to the peri-UA areas (distribution) are included. To derive these inventories, we use the pilot data and modelling framework from the LIFE ENRICH project34. This project implemented pilot-scale processes to recover struvite and ammonium salts in a WWTP in Murcia, Spain. The results from this pilot were modelled and adapted to 2022 WWTP flows and nutrient data from Baix Llobregat in Barcelona, Spain. This plant treats a quarter of the wastewater in the AMB. The Supplementary Note 3 shows the final life cycle inventories considered for struvite and ammonium salt production. For compost, additional activities include the aerobic and anaerobic production of compost and its distribution to the peri-UA areas. Primary data were collected for 2016 from three functional composting plants in the AMB, with the capacity to treat approximately 85,000 tons of OMSW, of which Barcelona Zona Franca Ecoparc 135,36 is the largest. Data for water, energy, compost transport emissions, avoided landfilling, and refused waste disposal (landfilled) are included (Supplementary Note 4).

Avoided activities are those that occur in linear systems and do not occur in circular systems because of the reconfiguration of nutrient cycles, such as avoiding N and P emissions to the ocean and avoiding sludge end-of-life (EOL) treatment due to nutrient recovery in WWTPs. In our case study, there were no avoided emissions to the ocean despite the recovery of nutrients. Because of environmental regulation, it is expected that the effluent to the ocean should carry similar N and P contents in both systems. Additionally, N and P are recovered from currently recirculated streams within the WWTP, i.e., during sludge treatment and dewatering; thus, the N and P contents in the effluent are not expected to change. For WWTP sludge EOL, the recovery of nutrients, particularly P in struvite34, may alter the amount and composition of sludge, which may no longer fulfil fertilization functions when applied in fields. For the AMB, we assumed that recovering N and P in the form of struvite and ammonium salts from wastewater do not lead to significant changes in the sludge composition and that would not induce additional use of N and P in the form of mineral fertilizer by current sludge users. The first assumption follows the findings of the ENRICH project34, where a change in P content is not expected in the sludge because P is recovered from the liquid phase of sludge treatment at the WWTP. However, uncertainty remains around the P content of the recirculated water to the WWTP and its effect on the sludge P content. The second assumption follows personal communication with sludge users outside of the AMB59,60, which stresses that less N and particularly P content in sludge would only increase the product value given the soil conditions of agricultural lands where it is currently applied. Furthermore, avoided activities for OMSW include avoiding the landfilling of organic residues that are composted in the circular system, which are included in the compost scenario and lead to negative footprints for most compost impacts. The Supplementary Note 5 shows the impact intensities for all the impacts considered, with current and pLCI databases, per product.

Scenario characteristics

The main characteristics of the four scenarios analysed that represent the linear and circular systems are shown in Table 1. The scenarios are defined in terms of the product of interest (linear or circular), the crop pattern, the database representing the global economy and the way N, P and K are provided.

URBAG python-based tool

To calculate the impacts for the four scenarios, we developed a python-based tool available on GitHub61. It consists of a series of Jupyter notebooks that are run in a specific order. Initially, the goal and scope are set; second, the inventories are determined; third, the LCIA is calculated, and four results are visualized for interpretation (Fig. 4). The calculated impacts are direct and indirect. Direct impacts are emissions occurring onsite at the peri-UA plots. Indirect impacts are caused by the supply chain/end-of-life of the feedstocks used in the field to produce the food, and they arise from the raw materials, production, transport, packaging, and EOL/disposal of these feedstocks. The sum of the direct and indirect impacts is the total life cycle impact of the linear and circular systems, which are then further compared.

Fig. 4
figure 4

General methodology implemented in the python tool (covered here in the grey boxes) under the Life Cycle Assessment (LCA) framework.

Full size image

Initially, from the crop pattern, the total city N, P and K requirements, water extraction, total food production and total peri-UA area are calculated (Fig. 4, grey box in goal and scope). The Supplementary Note 2 presents the maps and totals for these parameters.

Life cycle inventories (LCI)

Furthermore, inventories for yearly food production are calculated at the plot level. The following inputs are considered for all the scenarios: N, P and K inputs in the form of various products (according to the scenario and product specifications) and crop water consumption. The following outputs are included again for all scenarios: emissions of N2O to air, NH3 to air, NOx to air, NO3- to water and PO43- leaching, and nutrients removed during harvest, which are equal to the crop nutrient requirements. To determine nutrient-related inputs and outputs, four steps are followed in the tool (Fig. 4, LCI in green): 1) read in the crop nutrient content from the maps and determine crop nutrient requirements (Supplementary Note 6); 2) determine the supply of nutrients from different products according to the scenario under analysis and consider that circular nutrients are constrained resources and, as such, their production is limited to the amount of wastewater and solid waste generated annually (Supplementary Note 7); 3) calculate emissions using product- and substance-specific emission factors (Supplementary Note 8); and 4) balance the inputs and outputs with specific rules so that mass is preserved (Supplementary Note 9). For water-related inputs and outputs, different considerations are made (Fig. 4, LCI in blue): 1) read from the map the monthly, precipitation and evapotranspiration data in the map, as well as monthly crop water coefficients and irrigation and distribution efficiencies; 2) determine crop water requirements; 3) determine effective precipitation; 4) calculate irrigation demand; 5) calculate water extraction; and 6) convert water extraction in mm per month to yearly volumes of water consumed to use it as an indicator of direct water consumption in the LCIA (Supplementary Note 10). The results for the inventories are shown in the Supplementary Note 11.

Life cycle impact assessment (LCIA)

Finally, the LCIA results are calculated per plot by summing the characterized direct and indirect impacts. For the characterization of direct emissions, characterization factors for all substances emitted onsite are read via the python tool. The characterization factors depend on the impact and the family of methods used for each impact and for ME and FE impacts, which are regionalized considering the characteristics of the water basins in the AMB (Supplementary Note 12).

For indirect impacts, the tool reads in the precalculated impact intensities for all products, e.g., kg CO2eq per kg N in mineral fertilizer (Supplementary Note 5). These are calculated using the current and pLCI databases generated with the PREMISE tool v1.8.244. We precalculate the impacts of the inventories for the linear and circular products via 1) the ecoinvent v3.9.1 cut-off model as an LCI database and 2) two prospective versions of the ecoinvent v3.9.1 cut-off generated using the global socioeconomic and climate scenarios from the IMAGE62 IAM (Supplementary Note 5). The results for the total impacts of the AMB for direct and indirect impacts for all scenarios, including current and prospective LCIA results and for the current and projected peri-UA areas, are presented in the Supplementary Notes 13 and 14 in a geographically explicit visualization.