Mass balance calculations were done for the simplified agri-food system (Figure 1) at the national scale using data sourced from the literature and the FAOSTAT database for the year 2018 (Table 1). For the regional scale, we aggregated data sourced from the Bolivian Agricultural Census (2014) to calculate the nutrient flows for each agro-ecological, region as the census is performed at the municipality level. We assumed that the contribution of each region to the national nutrient budget remained constant until 2018. Detailed calculations can be found in the supplementary information.

Data calculated in the mass balance were processed with the PostGIS extension for PostgreSQL and then plotted as choropleth maps with QGIS.

At the national level, the total nutrient requirements in relation to crop yields obtained in the year 2018 (i.e., current crop yield; see Section 2.2.1) were 395 and 63 kt for N and P, respectively. The total nutrient inputs to the agri-food systems after accounting for nutrient losses during application amounted to 363 kt for N and 55 kt for P. This suggests that there are apparent deficits of 32 kt of N (8% of total crop requirement after accounting for biological fixation) and 7.7 kt of P (12% of total crop requirement) (Table 3). However, if all crops received optimal nutrient inputs so that their yields were maximised (i.e., optimum crop yield; see Section 2.2.1), then 605 kt of N and 97 kt of P would be required (Table 4). This means that the real nutrient deficit was much larger, 110 kt of N and 33 kt of P (Figure 2, Figure 3).

Although the top five crops in total yield (i.e., in tonnes) were sugarcane, soybean, maize, potato, and sorghum (Table 2), the five crops that demanded the largest N and P inputs were soybean, maize, sorghum, rice, and wheat (47, 31, 30, 13, and 10 kt for N, and 38, 5, 6, 2.7, and 1.8 kt for P, after accounting for biological nitrogen fixation and losses during application; see Table 3). The differences between the nutrient input required for maximising crop yields (Table 4) and the nutrient input received by crops in year 2018 (Table 3) were smallest for sorghum (6%) and soybean (28%), both of which are processed mainly to produce animal feed, but much larger (52% on average) for the staple crops wheat, rice, and maize.

Three agro-ecological regions (Chiquitania-Pantanal, Chaco, and the Valleys) accounted for nearly 90% of the nutrient demand (Table 3). About 65% of the cultivated area in Bolivia is found in the Chiquitania-Pantanal region (Table 2), which alone accounts for 80% of the nitrogen and phosphorus demand of the country (Table 3). This is because soybean, sorghum, and rice (three out of the five most cultivated and most nutrient-demanding crops grown) can only be cultivated in low-altitude agro-ecological regions (Figure 4). In contrast to Chiquitania-Pantanal, the Valleys region accounts for 10% of cultivated area but only 6 and 5% of the N and P demands of the country, respectively.

In five out of the seven regions (excluding Altiplano and the Valleys), there are deficits of both the nutrients. The deficit is largest in Chiquitania-Pantanal, where there is a shortfall of 50 kt N and 37 kt P, followed by Chaco (15 kt N and 3.1 kt P) and Yungas-Chapare (deficit of 3.4 kt N but surplus of 0.1 kt P). In both the Valleys and the Altiplano, there are nutrient surpluses of approximately 41 kt of N and 33.5 kt of P.

According to the mass balance at the regional scale, there is discrepancy between the demand for nutrients and the nutrients harvested for some crops. For example, sugarcane is one of the five most cultivated crops in Amazonia, Chiquitania-Pantanal, and Llanuras, but is not among the top five nutrient-demanding crops, and it is the fifth largest crop in terms of nitrogen demand in Chiquitania-Pantanal (Table 3). This is also the case for tomato cultivation in the Valleys and for pineapple cultivation in Yungas-Chapare. Soybeans accounted for about 17% of the total nitrogen applied in Chiquitania-Pantanal, where almost all the country’s soybean crop is cultivated. Although it is a common approach when calculating nutrient demand (INIAF, ANAPO, CIAB, 2019) to focus on crops with the highest harvests, our findings suggest that this can be inaccurate in countries and regions where crops with high water content and low nutrient demand are produced at large scale.

The main nitrogen input at national level was biological nitrogen fixation (204 kt, accounting for 47% of the total input) followed by manure (25%), crop residues (19%), and synthetic fertilizer (8%). For phosphorus, the main input was manure (29 kt, accounting for 53% of total input), followed by synthetic fertilizers (26%) and crop residues (21%). Therefore, the nutrient circularity of the Bolivian agri-food system (i.e., the proportion of nutrients recirculated to the nutrient inputs added during crop cultivation; see Eq. 4) was estimated to be 44% for nitrogen and 74% for phosphorus (Table 3).

In the two regions where there is surplus of both nutrients, the main nutrient input was animal manure (swine in the Altiplano and chicken in the Valleys), accounting for >65% of the total input (Table 3). In the two regions with the largest nutrient deficits (Chiquitania-Pantanal and Chaco), soybean was the most nutrient-demanding crop but also the main source of nitrogen input to the soil because of biological fixation (72 and 51% of the nitrogen inputs for Chiquitania-Pantanal and Chaco, respectively), while the crop residues left over after harvest were the main source of phosphorus (57 and 78% for Chiquitania-Pantanal and Chaco, respectively) (Table 3).

In regions with nutrient surpluses, the proportion of nutrients recirculated back to the food system was high, >95% in the Altiplano and >71% in the Valleys, respectively. This is likely because of the wider availability of manure across both these agro-ecological regions in comparison to that in the other five regions (Figure 5). In contrast, the estimated nutrient circularity in the two regions with the largest nutrient deficits was low, 24–67% for Chiquitania-Pantanal and 49–83% for Chaco.

In Bolivia, about 39 kt of nitrogen and 5 kt of phosphorus are available in human excreta (Figure 6). If all of the excreta were to be collected and recycled back to cropland, 100% of the nitrogen deficit and 64% of the phosphorus deficit of the current system (Table 3) could be met. Overall, the nutrients in excreta are equivalent to 20% of the nitrogen demand and 8% of the phosphorus demand of the agri-food system (Table 5). At the regional level, the replacement potential of excreta is higher in regions with large populations and low crop production. For example, 60% of the country’s population lives in the Altiplano and the Valleys, where only 16% of crop production occurs, so recycling either animal manure alone (the current practice; see Table 3) or human excreta alone in these regions can meet the majority of the nutrient demand (Table 5). Outside the Altiplano region, excreta would not be enough to meet nitrogen or phosphorus requirements. In Chiquitania-Pantanal, less than 10% of nutrient requirements could be met by recycling excreta.

The findings of this study suggest that 395 kt N and 65 kt P were required to support crop production in Bolivia for the year 2018. According to our mass balance calculations, there was an apparent deficit of 32 kt N and 8 kt P in crop production, which we cannot account for from the available data. The starting point of our calculations was the FAOSTAT (FAO, 2022) database where Bolivia reports the total yields of 45 different crops grown in the country. Thus, the total amounts of N and P in agricultural harvests are known, as are the total nutrient inputs to the food system in the form of biological fixation, manure, and synthetic fertiliser application. Therefore, it is unclear why the mass balance shows a nutrient deficit and/or how that deficit is being met.

There are several possible reasons for this deficit. First, the deficit may have been met with nutrient stocks in soil that become available during land use change (planned or unplanned conversion of forest land to agricultural land). There is evidence of deforestation in Bolivia linked to large-scale farming of industrial crops in agro-ecological zones where agriculture is practiced in close proximity to forest land (i.e., Amazonia, Chaco, and Chiquitania-Pantanal) (Müller et al., 2012; Ormachea, 2018). Second, the deficit could have been met through traditional agricultural practices (e.g., crop rotation, fallowing, etc.) for which there are no data, and which thus could not be accounted for. If these explanations hold, it would mean that the agri-food system is operating at a deficit that will ultimately result in depletion of soil nutrient stocks.

We also estimated that the true deficit in the Bolivian agri-food system is much higher than what the mass balance suggests, given the following two aspects. Firstly, if all crops received optimal nutrient application rates that maximised their yields, an additional 220 kt N and 32 kt P would be required. Secondly, the mass balance at the national level assumes that in regions that have nutrients in excess of the local demand, the surplus is collected and applied in regions that have a net deficit. If this is not the case, then there is an additional deficit of 53 kt N and 40 kt P. In the Chiquitania-Pantanal and Chaco regions alone, where 75% of the crop production occurs, the local deficit is approximately 15% for nitrogen but 70% for phosphorus. Addressing this deficit is vital because according to projections by Rivero Lobo and Aliaga Lordermann (2014), the production of proteins, fruits, and vegetables in Bolivia for 2018 could only meet 60–70% of the nutritional requirements of its population.

We found that less than half of the nitrogen applied or fixed in the cropland was recirculated in the food system, whereas three-quarters of the phosphorus was recirculated. However, there is great variation in nutrient circularity between the different agro-ecological zones. In zones where animal manure is the main nutrient input and livestock density is high, the circularity is also high (>0.9). This does not mean that all of the available nutrients are recycled, as there are surpluses of 41 kt N and 34 kt P in these zones, primarily in the Valleys (Table 3). Even if the surplus nutrients were to be recycled locally to optimise crop yields, there would still be excesses of 21 kt N and 28 kt P (Table 4). The surplus nutrients are equivalent to 40% of the N deficit and 75% of the P deficit in Chiquitania-Pantanal. Therefore, recycling nutrients from regions with a net surplus to regions with a net deficit could be one way to reduce the overall deficit, as well as to improve nutrient circularity and crop yields.

As in other developing countries that aim to increase food production, the agricultural strategy in Bolivia has focused on improving the access of farmers to synthetic fertilisers (INIAF, ANAPO, CIAB, (2019); Mnthambala et al., 2021). In 2017, a state-owned petrochemical plant was inaugurated in Bolivia to produce urea and ammonia. In 2022, the plant is expected to produce 600 kt of urea, of which about 7% will be sold locally and the rest exported to other Latin American countries (McNelly, 2020; BNAmericas, 2021). However, considering the true national N deficit (110 kt; see Table 4), urea fertiliser at the current application rate can only meet 20% of the N required to obtain optimal crop yields. About 40% of the industrial urea production would need to be sold domestically to have zero nitrogen deficit in the agri-food system. As of today, there is no domestic production of synthetic phosphate fertiliser in Bolivia, but the petrochemical plant intends to also produce monoammonium phosphate and diammonium phosphate.

Despite the domestic sale of urea, there are still total N and P deficits of 109 kt and 33 kt, respectively, in the agri-food system. If all the human excreta in Bolivia were to be collected and returned to cropland, the deficits can be reduced to 70 kt N and 28 kt P. This would significantly increase circularity, as excreta is a renewable source of nutrients. It would also improve the resilience in the food system against global events, e.g., disruptions in fertiliser supply chains seen during the COVID-19 pandemic (Amjath-Babu et al., 2020) and rise of fertiliser prices against a backdrop of rising natural gas prices (Hebebrand and Laborde, 2022).

Since >70% of the population in Bolivia lives in cities and urban areas, it could be possible to collect the majority of the nutrients available in excreta. For example, the population density in La Paz and El Alto is > 9,000 inhabitants/km² (Parés-Ramos et al., 2013). However, much of the economic feasibility of transporting these nutrients to croplands surrounding the cities will depend on how excreta are collected, treated, stored, and applied (Simha et al., 2020). The feasibility of this increases if water is removed from excreta to concentrate the nutrients (Trimmer and Guest, 2018). A wide range of technologies are available to convert human excreta (Harder et al., 2019) or source-separated fractions like human urine (Vasiljev et al., 2022) and faeces (Musazura and Odindo, 2022) to concentrated fertilisers.

In this study, we assumed that 100% of the human excreta-derived nutrients can be recovered and returned to cropland. In practice, the recovery efficiency would be likely lower, especially depending on the choice of technology used to treat excreta or source-separated excreta fractions (Harder et al., 2019). Many technologies like alkaline dehydration are currently being optimised to recover >98% of the nitrogen and phosphorus contained within human urine (Vasiljev et al., 2022). However, our findings showed that even if excreta-derived nutrients are applied at current manure-like efficiencies (i.e. 50%), they would significantly improve the nutrient circularity of the agri-food system (Table 5).