Embedding circularity into the transition towards sustainable agroforestry systems in Peru

− 1 (i.e.,50%oftheNrequirements)shouldcomefromcompost, while for pastures, the requirement was 40 kg P ha − 1 . We found that composting could lead to large greenhouse gas (GHG) reductions compared with the current waste disposal methods (i


Introduction
Tropical forests are global hotspots of carbon storage and biodiversity, vital for local livelihoods, and important for local, regional and global climate regulation (D'Almeida et al., 2007;Davidson et al., 2012;Bradshaw et al., 2009). However, they are being increasingly deforested at alarming rates, with agricultural expansion being the primary driver of forest loss (Gibbs et al., 2010;Morton et al., 2006). One of the key actions proposed to achieve a food system reconfiguration towards a more climate-resilient future is to avoid agricultural expansion in carbon-rich landscapes (Steiner et al., 2020). Narratives associated with halting deforestation have also been present at the COP26, where global leaders committed to working "collectively to halt and reverse forest loss and land degradation by 2030 while delivering sustainable development and promoting an inclusive rural transformation" (COP26, 2021). Although these high-level recommendations and agreements are valuable to set global policy and research agendas, it is critical to zoom into national efforts to explore what strategies are being used to avoid agricultural expansion into forests and how to boost them for effective change.
The national context of Peru is relevant in this discussion considering that it has the second-largest area of Amazon Forest, and the fourth largest tropical area worldwide, with 76 million hectares of forest (Asner et al., 2017). The Amazon forest occupies nearly 60% of Peru's national territory. Although it is the least populated area of the country, it is home to a wide diversity of indigenous peoples and hosts 20% of Peru's agricultural units (INEI, 2012a). Peru has been regarded historically as a low deforestation country (Anderson et al., 2018), in comparison to other Amazonian nations (e.g. Colombia or Brazil) and countries in South East Asia (Hoang and Kanemoto, 2021). Deforestation rates, however, have increased substantially in the 21st century. While in 2002 the annual humid primary forest loss was approximately 80,000 ha, by 2020 it had risen to nearly 200,000 ha (Global Forest Watch, 2021). Unlike Brazil, where most Amazon deforestation was historically driven by large-scale agriculture (Morton et al., 2006) and more recently by cattle ranching (dos Santos et al., 2021), in Peru nearly 80% of the forest loss events between 2001 and 2015 have been directly linked to small-scale events (<5 ha), whereas the remaining 20% is related to an array of different land use changes (Finer and Novoa, 2017), including medium-and large-scale agriculture (Vijay et al., 2018), alluvial gold mining (Espejo et al., 2018), road expansion (Larrea-Gallegos and Vázquez-Rowe, 2022) or coca cultivation (Bax and Francesconi, 2018), among others.
To achieve a 'zero net deforestation rate by 2030' and meet climate compromises, Peru aims to halt small-scale forest loss driven by smallholder farmers informally settled in public forest lands by promoting agroforestry. The strategy seeks to integrate farmers into the formal economy and improve their livelihoods by granting them access rights via agroforestry concessions (referred to as CUSAF in this document) on the condition that the farmer will avoid further deforestation, as well as the application of agroforestry-based land use and soil and water conservation practices (SERFOR, 2017). By mid-2021, 45 concessions have been granted and 500 applications were in process of recognition (SPDA, 2021). In total, it is expected that more than 120,000 smallholders, including coffee farmers but also maize, banana and cacao farmers (Robiglio et al., 2018), could benefit from this mechanism. Although CUSAF concessions are seen as a groundbreaking legal initiative, their success to effectively stop deforestation has been questioned due to the serious obstacles that farmers face (e.g., precarious public services in education, infrastructure and finance) to sustain their livelihoods, manage soil and water resources, and conserve forests (Pokorny et al., 2021).
Soil conservation and nutrient management are key aspects for the long term productivity of agroforestry-based systems, especially in acidic tropical soils (Szott and Kass, 1993). Although it is known that nutrient cycling is favored by the capture of nutrients from deep soil layers by trees' deep and shallow roots and their return to the topsoil via litter decomposition (Alegre et al., 2017;Nair, 1994), some systems can be prone to nutrient depletion and subsequent low productivities in the absence of fertilization (Alfaia et al., 2004). In Peru, coffee, cocoa and livestock smallholder tropical systems are generally low-input and organic by-default (Jezeer and Verweij, 2015;Somarriba and Lopez-Sampson, 2018). Diverse studies have identified poor soil management practices among farmers, leading to soil mining and soil degradation (Pokorny et al., 2021;Van-Heurck et al., 2020). Once soils have been mined and yields reduced, primary forests are cleared for more new fertile lands (Barham and Weber, 2012). In some areas, this trend has been reversed by farmers enrolling in voluntary certification standards. In such systems, farmers have adopted management practices such as the application of soil amendments and fertilizers, and systematic pruning of plants ( Barham and Weber, 2012). The application of soil amendments (i.e., compost, biochar, manure) has also been proved to be beneficial to allow recultivation in degraded soils, and to increase the carbon sequestration potential of agroforestry systems (Gay-des-Combes et al., 2017;Shrestha et al., 2018). As the CUSAF mechanism expects to restore degraded lands (i.e., mainly pastures) and to enroll smallholders in formal agroforestry production systems, the use of soil organic amendments could become imperative for the program to succeed.
In Peru, food loss and waste (FLW) can reach 45% of the total food produced (Bedoya-Perales and Dal'Magro, 2021). Without appropriate valorization strategies, most of this waste ends up in mismanaged disposition sites. In fact, the Peruvian waste management system is still transitioning from the use of open dumpsters to landfills . So far, the implementation of existing and prospective circular practices that target the utilization of organic waste before it reaches landfills or open dumpsters have not been prioritized. Instead, there have been proposals to mitigate GHG emissions associated to waste management via technological solutions involving energy recovery or anaerobic digestion in landfills (Vázquez-Rowe et al., 2021). Composting is a waste valorization practice that allows the reutilization of nutrients contained in biowaste as fertilizers. Quantifying compost volumes that could be obtained from biowaste, and its GHG mitigation potential, is key to start the discussion on circular solutions for waste management, but also to assess the potential provisioning of circular fertilizers for climate-resilient agricultural systems such as the CUSAF agroforestry systems.
Given the potential role that soil organic amendments might play in sustaining the establishment of agroforestry systems under the CUSAF mechanism, and the need for alternative FLW valorization strategies to reduce the carbon footprint of the Peruvian waste management sector, in the current study we explore the potential of valorizing FLW as compost to be used as soil amendment in agroforestry systems. To this aim, the objectives of the current study are to: i) calculate GHG emissions associated to compost elaboration compared to current waste management strategies; and, ii) estimate the total area that could be fertilized with compost under the CUSAF mechanism.

Estimation of the potential area for agroforestry concessions CUSAF
The geographical scope of the analysis was concentrated in the four Peruvian regions that hold more than 65% of coffee lands that could benefit from CUSAF, based on the estimations provided by Robiglio et al. (2018). These regions were Amazonas, Cajamarca, San Martín and Junín. To estimate the potential area of agroforestry concessions under the CUSAF mechanism, we used the criteria defined in Peruvian legislation (SERFOR, 2017). These criteria were: i) be within the zoning for agroforestry or silvopastoril or restoration activities according to the most recent approved forestry zoning (i.e. criteria for forestry zoning is described in SERFOR (2017)); ii) not overlap with national and regional conservation areas, private lands, forestry concessions or indigenous territories; iii) individual concessions cannot be larger than 100 ha; iv) be within lands of public domain; and v) have agroforestry systems under production within the area. Our analysis started with the zoning criteria. The agroforestry, silvopastoral and restoration zoning was only used for the region of San Martín (MINAM, 2020) as it was the only national region with an approved forestry zoning. For the remaining regions, we used the areas classified for different agricultural and restoration uses (see Table S1 for details) based on the approved Ecological and Economic zoning (SINIA, 2020). Subsequently, from the areas previously selected, we excluded the areas with other land use rights, namely national and regional conservation areas (SERNANP, 2022), communal lands (MINCUL, 2022), and forestry concessions (SERFOR, 2019) (i.e., criteria ii). To refine the area for lands that can sustain agroforestry systems (i.e., criteria v), we used the Geobosques study area (Vargas et al., 2019) to select the areas that corresponded to humid tropical forests within the Amazon basin. Data from the latest National Agricultural Census (INEI, 2012a) was then used to estimate the agricultural units without property rights (i.e., informal tenure) and those smaller than 100 ha (criteria iii and iv assuming that all lands without legal tenure were of public domain) at a district level. The resulting CUSAF area included forest lands and arable lands (see Table 1). To estimate the extension of CUSAF lands which were either coffee plantation areas or pastures, we selected the CUSAF areas that overlapped with the national agricultural land use surface for year 2018 (Livia et al., 2021) and, thereafter, multiplied, at a district level, the overlapped area with the proportion of harvested coffee and cultivated grassland area for 2018 (SEIA-MINAGRI, 2021). The selection of 2018 as the reference year was linked to the fact that it was the most recent year with data availability; however, it should be noted that a certain level of uncertainty is possible due to changes in the spatial distribution of crops ever since. A graphical representation of all the steps is presented in Fig. S1. The total surface area for coffee cultivation and pastures per region for 2018 (i.e., accounting for the coffee lands within and outside the CUSAF areas) was also extracted from SEIA-MINAGRI (2021) and shown in Fig. 2 as a referential comparison value to CUSAF areas.

Food loss and waste and GHG emissions associated to current waste management
The amount of FLW and GHG emissions resulting from current waste management systems was obtained from Vázquez-Rowe et al. (2021). This study used data from the national food purchase survey (INEI, 2012b) to determine the average diet composition in the main city of each of the 24 Peruvian regions. The amounts of FLW were calculated using the percentage of each food type lost or wasted through different stages of the food system (i.e., production, handling and storage, processing, distribution and consumption) described by Gustavsson et al. (2011) for South America. However, GHG emissions were only calculated for FLW generated in the distribution and consumption stages, as these are the streams that follow a municipal collection and disposal pathway. Other food loss flows in the early stages of the food supply chain were not considered given their heterogeneous nature and distribution, usually not reaching waste disposition sites (Vázquez-Rowe et al., 2021).
For the current study, we selected four cities: Cajamarca, Chachapoyas, Huancayo and Tarapoto (Fig. 1). These cities are the most populous located in the CUSAF regions chosen and were selected assuming that each city will supply compost to the CUSAF lands of its own region to minimize transportation. We selected the GHG emissions of the predominant waste management technology in each city, being landfilling for Tarapoto and Chachapoyas, and open dumping for Cajamarca and Huancayo. GHG emissions, as described in Vázquez-Rowe et al. (2021) were modelled using the EASETECH software and considering the local climatic conditions in each city. GHG emissions per tonne of biowaste are presented in Table 2. For this study, we followed the IPCC guidelines (Bogner et al., 2007) and biogenic CO 2 emissions were excluded. The reason for this exclusion is linked to the fact that CO 2 emissions from biomass sources are compensated by changes in biomass stocks in land use, land use changes and through forestry (Bogner et al., 2007). Given that EASETECH considers N 2 O emissions as negligible in dumping and landfilling activities, CH 4 was the only GHG considered. The characterization factors provided by IPCC (2013) were used to convert GHG emissions to CO 2 eq.

Quantification of compost amounts and GHG emissions
The potential compost quantities for each city were estimated considering a default yield of 400 kg of compost per tonne of biowaste input (Boldrin et al., 2009). GHG emissions were estimated assuming open window compost systems which are common in tropical areas due to their low implementation costs. GHG emissions were estimated based on CH 4 and N 2 O emissions resulting from the composting process, as well as electricity and fuel inputs. All default data were obtained from the literature as shown in Table 3.
GHG emissions and electricity use per tonne of biowaste composted were obtained from the United Nations Framework Convention on Climate Change methodological tools for clean development projects (UNFCC, 2017). These values have been carefully and conservative selected by UNFCC from high quality studies and are used as standard values in different studies (Mertenat et al., 2019;Yeo et al., 2020) given that they represent estimates that describe the direct GHG emissions from low technology composting systems in low-and middle-income countries. We also used the GHG emissions, electricity use and fuel use from the upper range reported by Boldrin et al. (2009). We selected the upper range to be conservative. The CO 2 eq emissions associated to electricity and fuel production in Peru were assumed to be 0.17 kg CO 2 eq kWh −1 and 0.40 kg CO 2 eq kg −1 diesel, respectively, and the assumed fuel combustion burned in an agricultural machinery was 3.16-5.45 kg CO 2 eq kg −1 diesel. These values were obtained from Ecoinvent 3.8 (Wernet et al., 2016), except the minimum value for fuel combustion which was obtained from the European Environmental Agency (EEA, 2019). The specific names of the activities extracted from Ecoinvent are presented in Table S2. We used the GWP 100 of 27.75 CO 2 eq for CH 4 and 265 for N 2 O (IPCC, 2013) to be in line with the estimations of Vázquez-Rowe et al. (2021). Given that standard emission values were used, our estimations do not consider the effect that the different temperature and humidy conditions in each of the cities could cause on the time of composting neither the final nutrient content.

Fertilization potential with compost
Total nutrient contents in compost, fertilizer replacement values and corrected nutrient contents per unit of compost are presented in Table 4. Based on the corrected nutrient contents, we estimated how much land could be fertilized based on specific requirements for coffee shade plantations and for the establishment of silvopastoril systems in degraded soils with pastures. Coffee fertilizer requirements were assumed to be 180 kg N ha −1 , 35 kg P ha −1 and 158 kg K ha −1 and were obtained from Agrorural, a subdivision of the Peruvian Ministry of Agriculture that provides technical assistance to peruvian farmers (Agro Rural, 2018). The fixed fertilization requirements are referential, as fertilization requirements will change depending on factors such as coffee age, variety and soil type. These factors were not accounted for in our analysis. The amount of compost needed was estimated based on fulfilling 25% and 50% of the N requirements ha −1 in a growing coffee plantation (i.e., 45 kg N ha −1 for 25% of the N needs and 90 kg N ha −1 for 50% of the N needs). We opted to only half of the N requirements with compost as previous research has shown that the application of compost at 25% and 50% of the N coffee requirements improves soil chemical characteristics, while at a dose of 100% different soil parameters (e.g., pH, P, and K, Ca and Mg antagonism) get unbalanced (Martins Neto et al., 2020). Fertilization requirements for the establishment of silvopastoril systems in degraded pastures were assumed Table 1 Estimation of total CUSAF areas and the agricultural and non-agricultural areas.

Region
Total CUSAF area to be 40 kg P ha −1 . This amount has been used in the successful establishment of silvopastoril systems in degraded pastures of the Peruvian Amazon (Alegre et al., 2017).

Limitations
The potential CUSAF areas were estimated based on the current available data, but as data for some CUSAF requirements were not available or were outdated, our estimations might change with more updated and precise data and, therefore, should be used and interpreted with caution. For instance, the forestry zonification for Amazonas, Cajamarca and Junín, was not available, nor was the share of agricultural units with agroforestry systems. Outdated data included the extension of areas without legal tenure and agricultural units smaller than 100 ha, as these were obtained from the last agricultural census, nearly 10 years ago. The fertilization requirements for the lands were standard and did not distinguish between soil types.
The estimation of FLW faces the same limitations described by Vázquez-Rowe et al. (2021), namely rough FLW ratios for the different stages of the food system, outdated food purchase data from 2008 to 2009, and the exclusion of the impacts associated to the removal of impurities (i.e., plastics, metal, etc.) to have a clean organic fraction. Furthermore, GHG emissions for compost elaboration, as well as the compost N content and fertilizer replacement values, are standard values from the literature and, therefore, the outcomes might be different when measured values are used. Lastly, this study is framed purely on a biophysical basis and does not consider the economic (i.e., investment in infrastructure, education, creation of new supply chains) and social implications (i.e., household waste segregation, reduction of FLW) that will imply to compost FLW that is currently landfilled or dumped.  Table 2 GHG emissions associated to the predominant waste disposal method, population and total amount of biowaste produced for each city. GHG emissions were recalculated from Vázquez-Rowe et al. (2021) excluding biogenic CO 2 emissions, as well as CO 2 emissions from infrastructure and transport activities at the disposition sites. In landfills with venting, gases are not collected, while in landfills with flaring a capture system allows a controlled combustion of the landfill gas (LFG). Population was obtained from the last national census (INEI, 2017  3. Results and discussion

GHG emissions associated to composting
The estimated direct GHG emissions associated to composting ranged between 121 and 131 kg CO 2 eq tonne −1 biowaste. Expressed per tonne of compost GHG emissions ranged between 302 and 327 kg CO 2 eq ton −1 compost on wet basis. These emissions are in a similar or higher than what has been previously reported in the literature. For instance, Mertenat et al. (2019) reported 111 kg CO 2 eq ton −1 biowaste, while Vergara and Silver (2019) reported 57 kg CO 2 eq ton −1 waste. Yang et al. (2013) reported 161 kg CO 2 eq ton −1 dry matter compost or 60 kg CO 2 eq ton −1 compost on a wet basis (assuming 37% DM). The net GHG emissions of composting could be even lower if avoided emissions from waste management, and enhanced C sequestration from land application to compost are included. Although applying organic amendments to soils can increase soil respiration by approximately 20%, carbon losses are substantially offset by the increases in above and below net primary productivity up to 2.1 +/− 0.8 Mg C/ha to 4.7 +/− 0.7 Mg C/ha (mean +/− SE), respectively in a 3-year period (Ryals and Silver, 2013).
The estimated direct emissions calculated for composting are far lower than the 1354-2530 kg CO 2 eq ton −1 biowaste for landfilling with and without gas recovery, and the 2019-2136 kg CO 2 eq ton −1 biowaste for deep dumping estimated by Vázquez-Rowe et al. (2021). For every tonne of biowaste composted instead of landfilled or dumped, at least 1223-2409 kg CO 2 eq could be avoided. These differences are mainly linked to CH 4 emissions. While for composting CH 4 emissions per tonne of biowaste ranged between 1.5 and 2 kg tonne −1 biowaste, for dumping they ranged between 73 and 78 kg, and for landfilling between 49 kg (i.e., with flaring) and 73 kg (i.e., with venting only). During composting, CH 4 emissions are limited due to the aerobic conditions caused by the frequent aeration process needed for composting, while for landfilling and deep dumping anaerobic conditions favor the generation of CH 4 .
Compared to the overall GHG emissions of the current waste management system in each of the four cities selected, composting could lead to large reductions in GHG emissions (Fig. 3), leapfrogging in terms of waste management technologies and fostering deep decarbonization pathways beyond Peru's nationally-determined contributions (NDCs) within the Paris Agreement (Geels et al., 2017;Vázquez-Rowe et al., 2019). In fact, when accounting for all cities, composting could reduce waste-associated GHG emissions by more than 92,000 t CO 2 eq per year and provide the food system with more than 20,000 t of compost to be used as soil amendment.

Fertilization potential for CUSAF lands
The area of coffee production lands that could be fertilized with the assumed compost nutrient content providing 50% of the N requirements is limited. If 50% of the N needs for coffee production are provided by compost, only 279-2902 ha would be fertilized (see Table S3). This number is far from the 86,786 ha of CUSAF coffee producing lands in the four regions, and represents only 3% of CUSAF coffee producing surface area when the maximum N content in compost was considered (Fig. 4). The fertilization of pastures with compost at a rate of 40 kg P ha −1 is also limited (Fig. 4, Table S3). Only in the case of the region of Junín, there would be enough compost if the maximum P content is considered. This relates to the small extension of CUSAF pastures in this region (i.e., 911 ha, see Fig. 2), which were easily fulfilled with the potential quantities of compost produced in the region. In the remaining cases, the limited fertilization potential of compost is linked to three factors.
First, organic fertilizers such as compost from biowaste have low N concentrations, requiring high doses per ha to meet the nutrient needs of coffee plantations (Martins Neto et al., 2020). The assumed N concentration of the compost from biowaste was 0.1 to 1.3% (fresh weight), which accounts for a minimum and maximum N fertilizer replacement value (see Table 4). Without the correction for the fertilizer replacement value, total N content would be 0.6 to 2.2% on fresh basis, and the proportion of fertilized lands could increase up to 5.5%. Other studies also report N values ranging from 0.7 to 1.3% for compost produced from food waste from municipal waste collection systems (Hwang et al., 2020). Other compost types that were mixed with livestock manure have N contents above 2.4% (Raviv et al., 2013) supplied by the ammoniacal-N contained in manure. To increase the N content of compost obtained from biowaste, the compost could be mixed with guano which contains 10-14% of N (Agro Rural, 2018) and which is currently used by some coffee producers, especially those certified under voluntary schemes (Barham and Weber, 2012). In addition to guano or mineral fertilizers, establishing coffee intercropping with nitrogen-fixing legumes could add more than 44 kg N ha −1 (Mendonça et al., 2017).
The second factor has to do with the amount of FLW used to estimate the available compost biomass. In Peru, FLW can reach 45% of the total food produced (Bedoya-Perales and Dal'Magro, 2021). However, from this large share of food wasted, 80% takes place during the agricultural production, storage and processing, and the remaining 20% during the distribution (10%) and consumption stages (10%) which we used to estimate the compost biomass (Vázquez-Rowe et al., 2021). This trend is different to regions such North America or Europe where nearly 40% of the FLW take places during the consumption stage (Gustavsson et al., 2011). While we only used the 20% of FLW because that is the waste that is managed in urban disposal sites (i.e., open dumps, landfills), it is key to identify how the remaining 80% is being used. It is likely that a portion of the crop residues and wasted crops are already used as animal feed contributing to circular economy, and that a portion is also burnt on fields; however, data on the overall volumes of reused and burnt biomass are lacking. For the coffee sector, several certification standards (e.g., Rainforest Alliance) or public-private partnerships (e.g., Coffee Alliance for Excellence) are requesting producers the composting of coffee residues (i.e., husk, pulp) on farm to be reused as a soil amendment. While coffee residues can provide valuable nutrients for coffee plants when composted with other sources (i.e., green and animal manure) (Dawid, 2018), these residues are not always generated at the farm given that the coffee from Peruvian smallholders gets processed in other areas.
Finally, in our analysis we only accounted for the compost that could be produced in the cities of the main coffee regions and did not account for potential amounts of compost that could be produced in rural settlements and other cities. This is an important consideration given that one third of the population is settled in the capital of Peru, Lima. Only in Lima, more than 200,000 t of compost could be obtained, which could be used to fertilize up to 3 to 35% of the CUSAF coffee lands. However, transporting that compost would imply the emission of 0.17 kg CO 2 eq per tonne and kilometer (Wernet et al., 2016). The transport emissions would lead to additional GHG emissions ranging from 50 to 200 kg CO 2 eq per tonne of compost, Table 4 Minimum and maximum total nutrient contents in compost (on a fresh basis), fertilizer replacement values equivalent to mineral fertilizers, and corrected nutrient content (on a fresh basis) to be equivalent to mineral fertilizers. Values were obtained from Boldrin et al. (2009 equivalent to 15-60% more emissions, depending on the distance from Lima. Defining whether producing such transport emissions are worth will depend on two main issues. On the one hand, if adding soil amendments to forest can indeed help to reduce deforestation and, consequently, avoid additional emissions associated to forest loss. On the other hand, if this pathway would be more socially and economically favorable than fertilizing agricultural lands in the surrounding agricultural areas of Lima, where large-scale and export-oriented agriculture largely depends on imported chemical fertilizers and efforts are slowly being implemented to rely on higher fractions of organic fertilizers.

Implications and policy outlook
Although compost from biowaste tends to have low concentrations of N, it is worth considering that biowaste compost can have multiple benefits for soil health and crop productivity in the long term. The benefits of applying organic amendments such as compost to crops are multiple and include higher resistance to droughts, increase in water use efficiency, reduction of soil loss by erosion (Martínez-Blanco et al., 2013;O'Connor et al., 2021;Shiralipour et al., 1992), increase in microbial activity, and increase of N supply over time (Diacono and Montemurro, 2010). In addition, in acidic soils such as those present in the Amazon, phosphorus deficiency is common because P is mainly adsorbed to Al, Fe and to a lesser extent with Ca (Reis et al., 2011). Applying compost can therefore increase P levels substantially and make the soil less acid, making this nutrient more bioavailable for plant nutrition (Martins Neto et al., 2020). Future biophysical assessments should determine at a food systems level the fertilizer potential of biowaste compost mixed with available N-rich organic amendments (e.g., guano, animal manures, green manure), and long-term field trials should be performed to evaluate their effect on crop growth and soil health. Such efforts have been done in Brazil (Martins Neto et al., 2020), but should be replicated in other Latin American countries.
Appropriate nutrient management and soil health is not only fundamental for the success of CUSAF but is also directly linked to Peru's NDCs (Gobierno del Perú, 2018) on implementing robust coffee agroforestry systems. In fact, the Peruvian government aims at reducing GHG emissions linked to coffee and cocoa production by 1.4 Mt. CO 2 eq by 2030, based on a set of improved agricultural and processing practices, which is aligned with the objectives of the current study.
The use of fertilizer inputs is not a common practice among smallholders. In coffee farms from the region of San Martín inputs only accounted for 11% of the total costs, of which fertilizers were the most important cost (83%). However, the largest costs were associated with land (44%) and labor (38%) (Jezeer et al., 2018). Thus, as farmers do not traditionally invest in fertilizers nor do they always have access to them (e.g., absence of local suppliers, lack of financial systems), future efforts targeting the adoption of fertilization practices for CUSAF (i.e., via compost and other sources) should contemplate the implementation of subsidies or other type of incentives to make them accessible. These subsidies could come as part of the in-kind rewards associated to REDD mechanisms (Montoya-Zumaeta et al., 2021) used in some of the Peruvian coffee growing regions. It is important though, to avoid rebound effects for further agricultural expansion due to the expected better accessibility to fertilizers.
Although the potential land fertilized with compost reported in this study is limited, composting municipal biowaste could lead to drastic GHG emissions reductions compared to the current dumping and landfilling practices (Fig. 3), an environmental benefit that is directly linked to Sustainable Development Goal (SDG) 13 on climate action. Segregation of organic waste for composting and the construction of landfills with flaring technology have been both included in the list of Peruvian NCDs (Gobierno del Perú, 2018). However, Peru's transition from dumpers is moving towards landfills, and in most cases flaring or more advanced technologies (e.g., energy recovery or anaerobic digestion) have been discarded. These decisions will constraint Peru's efforts to reduce GHG emissions and meet its emission targets. Moreover, given the long dumping and recent landfilling culture associated to waste disposal, a change towards other circular valorization strategies will need consumer and political willingness. Consumer willingness is needed to reduce shares of food waste and achieve large rates of unavoidable organic waste segregation at households, as a key action to comply with SDG 12.3 on halving FLW by 2030, Political willingness is needed to path the rules for waste segregation and collection for different stakeholders (e.g., consumers, restaurants, markets food processors), and to promote the creation of transfer stations for organic waste collection. An example of the latter are the strict rules that the state of California will impose if household waste is not segregated (Latimes, 2021). In addition, there should be incentives available for the creation of new public or private companies that will be in charge of the recovery and circular valorization, and new rules for those waste

Conclusions
In this study we explored the environmental and fertilization potential of an urban-rural circular system in which compost obtained from municipal biowaste is used to fertilize concessions of coffee agroforestry systems   4. Percentage of coffee production and pasture lands within CUSAF that could be fertilized for each region, and for the four regions (i.e., total). Bars show the mean and error bars minimum and maximum values. If the horizontal line at 100% is overpassed, it means that there is enough compost to fertilize all CUSAF lands for the specific crop. Specific values are presented in Table S3. and silvopastoral systems in Peru. The amount of compost that can be generated in the cities close to the agroforestry and silvopastoral systems considered in this study is limited, as it only reaches 0.3 to 3.7% of the envisioned agroforestry concession areas. However, composting municipal biowaste to be used as soil amendment can substantially mitigate GHG emissions compared to the current dumping and landfilling practices. To implement the latter, an enabling environment is needed that can drive changes in consumers behavior and in stakeholders participating in the waste segregation and collection process. We conclude that composting municipal biowaste should be adopted to decarbonize current Peru's waste management system, but its mainstream use as fertilizer will require mixing it with N-rich sources an assessing if bringing compost from other regions could have positive impacts in term of GHG emissions reductions.
CRediT authorship contribution statement

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.