Resources, Conservation & Recycling 179 (2022) 106130 Contents lists available at ScienceDirect Resources, Conservation & Recycling journal homepage: www.elsevier.com/locate/resconrec Full length article Extended use and optimization of struvite in hydroponic cultivation systems Verónica Arcas-Pilz a, Felipe Parada a, Martí Rufí-Salis a, Gaia Stringari a, c, Ramiro González a, Gara Villalba a, b, *, Xavier Gabarrell a, b a Sostenipra Research Group (2017 SGR 1683), Institut de Ciència i Tecnologia Ambientals ICTA-UAB, (CEX2019-0940-M), Z Building, Universitat Autònoma de Barcelona (UAB), Campus UAB, Bellaterra, Barcelona, 08193 Spain b Department of Chemical, Biological and Environmental Engineering, Universitat Autònoma de Barcelona (UAB), Campus UAB, Bellaterra, Barcelona, 08193 Spain c DISTAL—Department of Agricultural and Food Sciences and Technologies, Alma MaterStudiorum University of Bologna, 40127 Bologna, Italy A R T I C L E I N F O A B S T R A C T Keywords: Hydroponic systems are an attractive form of urban agriculture due to their low weight load, inert substrate Struvite conditions, and overall better control of plant nutrition and growth. However, gaining urban food sovereignty Urban Agriculture cannot be at the cost of increasing environmental impacts, such as eutrophication and nonrenewable resource Life Cycle Assessment depletion, associated with phosphorus fertilizer use. Struvite, a wastewater byproduct, is a potential slow- Phosphorus depletion Hydroponic releasing P source that can serve as a substitute for mineral P fertilizer. In this study, we explored the ade­ quacy struvite in hydroponic systems, testing different quantities (5 g, 10 g and 20 g per plant) compared with monopotassium phosphate for pepper and lettuce hydroponic production. The results show competitive pro­ ductions for both crops with the use of struvite, especially during the first lettuce harvest (225.5 g, 249.9 g, 272.6 g, and 250 g for 5 g, 10 g, 20 g and control, respectively) where a greater struvite dissolution was seen. Although all struvite treatments in pepper show low phosphorous content in the biomass, yields do not deviate greatly from the control (3.6 kg, 4.3 kg, 7.5 kg and 5.3 kg for 5 g, 10 g, 20 g and control, respectively). The environmental performance of all lettuce treatments showed a reduction in all impact categories, especially freshwater eutrophication and mineral resource scarcity, except for marine eutrophication. All impact categories were reduced for all pepper treatments with 10 g and 20 g of struvite. When the results are extrapolated to a full year of production, we find that the slow dissolution of struvite can sustain competitive production with an initial 20 g, with less impact in all categories except marine eutrophication 1. Introduction improve water savings since a more controlled environment can be ensured with more accurate irrigation systems as well as water recir­ Urban agriculture (UA) has the potential to significantly increase culation depending on the installation (Parada et al., 2021). As shown by food security in cities (Toboso-Chavero et al., 2019). Increasing green Appolloni et al. (2021) among 92 cases of urban agriculture identified areas in urban landscapes have been gaining popularity, and with new from 2011 to 2019 a 46% produced with a soilless system. In addition to technologies, greening and food production have been taken to building increasing food sovereignty, UA can promote biodiversity, CO2 capture roofs, facades and even indoors (Despommier, 2013; Specht et al., 2014; and pollination (Baró et al., 2014; Camps-Calvet et al., 2016; Ayers and Appolloni et al., 2021). In particular, soilless agriculture is highly Rehan, 2021) but can also have negative effects, such as the extensive attractive in urban settings because of the reduced weight load on use of mineral and synthetic fertilizers (Sanjuan-Delmás et al., 2018). building structures, inert substrate conditions, and overall control of Soilless agriculture does not contemplate the addition of nutrients plant nutrition and health (Walters and Stoelzle Midden, 2018; Vinci through the substrate but through a nutrient solution given with the and Rapa, 2019), as well as because it provides an alternative to irrigation system (EI-Kazzaz, 2017; Sambo et al., 2019). Previous work contaminated soils. Soilless production can also be a beneficial system to on life cycle assessment of hydroponic production systems shows that, * Corresponding author at: Sostenipra Research Group (2017 SGR 1683), Institut de Ciència i Tecnologia Ambientals (CEX2019- 000940-M), Z Building, Uni­ versitat Autònoma de Barcelona (UAB), Campus UAB, Bellaterra, Barcelona,08193 Spain. E-mail address: gara.villalba@uab.cat (G. Villalba). https://doi.org/10.1016/j.resconrec.2021.106130 Received 25 September 2021; Received in revised form 22 December 2021; Accepted 23 December 2021 Available online 2 January 2022 0921-3449/© 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). V. Arcas-Pilz et al. Resources, Conservation & Recycling 179 (2022) 106130 while these fertilizers secure direct nutrient uptake by the plant, their releasing fertilizer struvite has been deemed a solution not only for the production, extraction, use and disposal are known to have adverse supply of this nutrient in agriculture but also to avoid further phosphate consequences for the surrounding natural ecosystems (Sanjuan-Delmás rock extraction and an increase of P in wastewater streams and water et al., 2018; (Rufí-Salís and Calvo, 2020; Rufí-Salís and Petit-Boix, cycles (Bradford-Hartke et al., 2015). 2020). Alternatives for the fertilization in soilless agriculture are gain­ Struvite has already been tested in a variety of agricultural soils ing interest being aquaponic systems most know for the efficient use of (Latifian et al., 2012; Li et al., 2019), obtaining a wide range of results in fish debris as nutrient for crop production reducing the potential impact crop growth and yield for a diverse range of plants, as shown by Ahmed of the production system (Graber and Junge, 2009; Chen et al., 2020). et al. (2018). Although results vary among different crops, a common However, aquaponic installations can entail a great initial investment perception is the slow solubility of granulated struvite; therefore, its and call for an additional production fish and therefore a greater need most common application is in the form of pulverized struvite (Degryse for maintenance and skill (Baganz et al., 2020). et al., 2017). Struvite dissolution has been tested before in continuously The extraction of phosphate rock, the main source of phosphorus (P) stirred pots and at controlled temperature, showing a lower dissolution for fertilizer use, has become a necessity for modern agriculture and is an rate than triple superphosphate (TSP) (Ariyanto et al., 2017; Rech et al., indispensable nutrient for plant growth and animal feed (Rahman et al., 2018). 2014). However, phosphate rock deposits are limited due to the slow Further experiments testing struvite dissolution have demonstrated regeneration rate of their cycle compared to carbon or nitrogen, already the importance of medium pH and plant root proximity (Massey et al., generating supply shortages due to increasing prices and unequal dis­ 2009; Achat et al., 2014; Talboys et al., 2016; Degryse et al., 2017). This tribution (Alewell et al., 2020). In recent decades, estimations have been proximity can provide greater access to dissolution mechanisms from made regarding imminent depletion if extraction continues at the pre­ the plant that can make the P available, such as the exudation of organic sent rate (Cordell et al., 2009), which can be drastically shortened by soil acids to lower the pH of the soil or substrate (Rech et al., 2018). This erosion caused by unsustainable production practices (Alewell et al., slow dissolution has been seen to hide crop development in early stages, 2020). corresponding to still early growth of the plant root. All the P extracted is mostly “lost” from agricultural lands and live­ Although information on the use of struvite as fertilizer is already stock management through surface and underground runoff (Carpenter available, its use in soilless agriculture is still scarce. Since hydroponic and Bennett, 2011; Rahman et al., 2014). This one-way nutrient flow has systems enable better control of plant nutrition but are designed to use increased fourfold since preindustrial times (Liu et al., 2008; Villalba chemical fertilizers, the use of struvite in exchange for the mineral et al., 2008; Carpenter and Bennett, 2011; Alewell et al., 2020) and phosphorous used in soilless agriculture has the potential to reduce its contributes to great ecosystem damage, such as eutrophication, espe­ environmental burdens. First approaches have been made to identify the cially in freshwater environments (Cordell and White, 2014). suitability of struvite in hydroponic production as well as it’s combi­ While this ever-growing thread demands better management of P nation with biological amendments like rhizobium showing promising sources, there is possible recovery from an ongoing loss of nutrients results (Arcas-pilz, 2021a). The P emissions to water seam to decrease occurring daily in our wastewater treatment plants (WWTPs) (Cordell significantly with the use of struvite compared to mineral derived P and White, 2013; Harder et al., 2019). These nutrients contained in fertilizers while other studies reveal even greater productions with wastewater sludge are disposed and managed mostly in complex pro­ struvite (Arcas-pilz, 2021b; Carreras-Sempere et al., 2021). Previous cesses due to the high content of heavy metals, pathogens and other work identifies the low solubility of struvite as a potential burden for compounds, making it a toxic residue (Panizza and Cerisola, 2001; plant uptake while it could also ensure reduced P emissions and longer Rahman et al., 2014). While direct application of sewage sludge to the productions over time. soil is practiced in several countries, it’s application can entail a bad With this knowledge on struvite the question arises if urban agri­ management of the soil, due to over application for P fertilization as well culture can directly profit from the nutrients generated in their imme­ as the increase of pathogens and heavy metals into the soil, and the diate surroundings and strive for expansion without shifting the potential problematic of social acceptance of this practice due to the environmental damage to the generation of greater water and air emitting odors (Pradel et al., 2020). Countries like Sweden have seen a emissions in urban settings. For this purpose, the production of crops for reduction of the unwanted toxic metals since the 1970 and have started longer periods was proposed to understand the struvite nutrient to regard sewage sludge as a potential nutrient provider and soil discharge in time for short and long cycle crops. amendment, still only 20% of the sludge is applied in arable land The following experiment was performed to analyze the potential of (Kirchmann et al., 2017). struvite in providing P in hydroponic production systems by testing In recent decades, research has been conducted on the shift from a struvite solubility and uptake in granular form for two different crops: removal to a recovery approach in urban water cycles in terms of nu­ pepper plants (a highly P-demanding crop with a long growth cycle) and trients, not only for their further use in other production sectors but also lettuce (shorter cycle). In addition to the dissolution and uptake anal­ to prevent their environmental damage in their disposal (Harder et al., ysis, this work also focuses on the nutrient discharge into water, 2019; Rufí-Salís and Brunnhofer, 2020). covering the potential reduction of P loss into water bodies. To express One of the byproducts in these sewage treatment plants is magne­ the environmental burden of the struvite fertilizer and the mineral sium ammonium phosphate (MgNH4PO4•6H2O), a crystal commonly phosphate fertilizer, the collected information was used to perform an called MAP struvite or just struvite. Struvite is not a new material; its environmental analysis for the produced vegetables using the life cycle precipitation was first documented in Los Angeles (Borgerding, 1972), analysis assessment (LCA) to determine the environmental footprint but it was approached as a great problem since its precipitation occurs during the timeframe of this experiment. The assessment was extrapo­ spontaneously at a 1:1:1 molar ratio of magnesium (Mg2+), ammonium lated to a one-year production period to simulate these fertilization (NH4+) and phosphate (PO43− ) and under suitable pH conditions techniques for a longer time to maximize the P available in the struvite (8.5–9.5) (Buchanan et al., 1994; Bouropoulos and Koutsoukos, 2000; placed in the substrate. Le Corre et al., 2009). The purging of this uncontrolled struvite pre­ cipitation can be the cause of additional expenses due to damaged 2. Materials and methods equipment that needs replacement or increased labor costs (Stratful et al., 2004). Since then, technologies aimed at struvite removal through 2.1. ICTA-UAB greenhouse induced precipitation in WWTPs have unveiled a product with great fertilizer potential. The following experiment was performed in the ICTA-UAB inte­ The possibility of P recovery from wastewater in the form of the slow grated rooftop greenhouse in the Universitat Autònoma in Barcelona 2 V. Arcas-Pilz et al. Resources, Conservation & Recycling 179 (2022) 106130 from June to September (2020). The production system was hydroponic The short cycles of lettuce lasted a total of 27 days after transplanting using individual pots and perlite as the substrate (see picture in sup­ (DAT). Once the plants were harvested, a new seedling was planted in porting information). Nutrients were given through fertigation, mixing the same pot (14∅ I 13 cm high for lettuce and 25∅ I 20 cm high for concentrated nutrient solutions (NS) with harvested and filtered rain­ pepper). For each treatment, two pods were removed after each cycle to water (RW) in a proportion of 1:100 (NS:RW) through a drip irrigation take substrate samples. Pepper plants were planted parallel to the first system with a 2 L h − 1 flow. To ensure sufficient irrigation, the amount lettuce crop until the harvest of the third lettuce cycle (81 DAT), as of drained water was determined daily and maintained ca. ~30% of the shown in Fig. 1. Pepper fruit harvests were made weekly once produc­ incoming irrigation with increasing or decreasing irrigation time. The tion started, accounting for a total of four harvests before finalizing the growing frame consisted of 4 m x 0.5 m wide tables with the capacity to experiment. On the other hand, lettuce yields were weighed after each grow two crop lines. Between tables, a distance of 1 m was given from cycle, generating three harvests. plant to plant. To obtain a more accurate understanding of the possible yield vari­ ations among lettuce samplings, 15 pots of each lettuce treatment were labeled with a reference letter (from A to O) maintained throughout the 2.2. Crop growth and treatments experiment. Relative yields obtained could then be traced back to the corresponding pot, therefore allowing precise appreciation of possible In this experiment, the determination of struvite dissolution and production changes. uptake was carried out on two different crops, Lettuce (Lactuca sativa L.) The yield produced by the pepper plants was obtained in four har­ and Pepper (Capsicum annum L.). The lettuce and pepper plants were vests. The total fruit weight recorded for each treatment was obtained as obtained from a nearby nursery in early growth stage with the first the sum of all four harvests. The number of fruits produced in each growth of the true leaf’s (about 7 cm tall for lettuce and about 10 cm tall harvest was also accounted for to estimate the weight per fruit. for pepper), which were then planted in the greenhouse inside perlite filled pots. The perlite was previously wettened with water to ensure a 2.3. Plant sampling methods better handling and provide moisture to the plants during the trans­ planting process. The treatments were arranged in rows to facilitate For each lettuce cycle, the fresh and dry weights of each plant were irrigation and leachate sampling as well as drainage measurement. Each measured. After harvest, a random sample of four plants for each row represents a treatment with a different struvite quantity ranging treatment was dried at 60 ◦ C until a constant weight was achieved (ca. 7 from 5 g (named 5LE for lettuce, 5P for pepper) and 10 g (named 10LE days). for lettuce, 10P for pepper) to 20 g (named 20LE for lettuce, 20P for At the end of the experiment, all pepper plants were harvested and pepper), including a control treatment (named CLE for lettuce, CP for weighed. The pepper plants were separated into leaves, stems and roots, pepper). All crops fertilized with struvite received P-deficient NS, while removing all flowers. Additionally, we quantified the leaf weight, the control treatment was irrigated with conventional NS (NS specified number, area index (LAI), stem weight and length. The latter was in supplementary information Table 1). To maximize the contact be­ measured accounting for the main central stem without considering tween struvite and the plant, the granules were placed close to the root ramifications. However, these ramifications were considered when once the seedlings were transplanted into the pots. weighing the stem. The LAI was obtained using the scanned images of In the case of the lettuce crop, each treatment consisted of 28 plants 25% of the leaf fresh weight and further processed with a Python script arranged in two rows, making two replicates of 14 lettuce plants (as indicated in the Supplementary material) (relating the number of distributed randomly for each treatment, while for the pepper crop, pixels per leaf area) to give the total area of each pepper plant (Garrido eight plants were arranged in simple rows (Fig. 1). et al., 2020; Ribalta-Pizarro et al., 2021). A sample of five plants per During the experiment, several sensors were used to record the cli­ treatment was also dried in an oven at 60 ◦ C until reaching a constant matic conditions inside the greenhouse. Humidity and temperature were dry weight (ca. 7 days). In the case of the fruit, a sample of pepper pods recorded with a CS215 Campbell Scientific, and radiation was recorded was taken from each treatment after every fruit harvesting. Fresh and with a pyranometer (L202 by Hukseflux). dry weights were measured following the same procedure as for the All water flows were measured daily throughout the experiment. The plant biomass. irrigated water was quantified through water meters installed in the Before drying all plant biomass was rinsed with Elix water and dried irrigation system, while the volume of drained water was measured on to avoid any potential external contamination. buckets at the end of each crop line. Samples of incoming and outgoing Once the dry weights for lettuce, pepper fruit and biomass were water were taken three times a week for each treatment. To ensure good quantified, the samples were ground for further analysis, consisting of irrigation conditions, the pH and EC for these water samples were digestion with concentrated HNO3 in a single reaction chamber micro­ measured immediately after collecting daily samples (Supplementary wave to be then analyzed for total P concentration using optical spec­ information Figs. 2 and 3). trometry (ICP-OES). Substrate samples were transferred to a polypropylene sampling pot Table 1 after thoroughly mixing the perlite from the pot in a clean container. For Average yield (g/ per plant) expressed as fresh weight (FW) and dry weight each treatment, two samples were taken at the end of each of the three (DW) for all three harvests at 27 DAT, 54 DAT and 81 DAT. Significant differ­ production cycles. After taking the perlite samples, they were placed on ences (p< 0.05) between treatments marked with different letter (a,b,c). aluminum trays and dried at 60 ◦ C for 72 h. Once dry, the samples were Average 1st 2nd 3rd weighed and ground for total P determination using the method detailed Yield Harvest Harvest Harvest (g/per before. plant) Irrigation and leached water samples were proportionally unified FW DW FW DW FW DW into weekly samples considering a volume ratio. These samples were 5LE 225.5c 9.1c 224.9b 9.8b±2.2 133.3a 5.5a±1.6 filtered through a 0.22 mm filter and analyzed with ionic chromatog­ ± 43.2 ± 52.1 ± 28.1 ±1.5 raphy (ICS-200) for nitrite and nitrate contents. The Mg and P contents 10LE 249.9b 10.1b 251.7a 10.9a±2.4 139.8a 5.8a±1.6 ± 35.2 ±1.4 ± 56.9 ± 31.2 in the water samples were analyzed with ICP-OES (Perkin-Elmer Optima 20LE 272.6a 10.9a 261.4a 11.4a±2.6 149.6a 5.8a±3.0 4300DV). ± 32.1 ±1.3 ± 59.2 ± 56.4 To calculate the struvite dissolution rate, the amount of P found in CLE 250.0b 10.1b 279.0a 12.1a±1.5 137.8a 5.4a±2.0 the plant and leachate was calculated for each treatment (for lettuce, ± 26.6 ± 33.5 ± 35.7 each cycle was taken separately). The obtained quantity was assumed to ±1.1 3 V. Arcas-Pilz et al. Resources, Conservation & Recycling 179 (2022) 106130 Fig. 1. Experimental outline (A) shows the distribution of the pepper and lettuce treatments along the laboratory greenhouse (C= control treatment). The exper­ imental timeline (B) shows the total duration of the experiment and the duration of the pepper and lettuce cycles (DAT= days after transplanting). The System boundaries for the environmental analysis (C) show the scope of the analysis within the dotted line. Fig. 2. Amount of P (g) accumulated in the lettuce biomass(above) and perlite (below) for treatments 5LE, 10LE, 20LE and CLE in all three harvests (1, 2, 3) given in total g per plant (in the case of perlite g per pot). be the dissolved P from the struvite and divided by the liters irrigated to and 14,044 (ISO 2006). the crop. The dissolution rate was then plotted against the initial struvite amount given to the plant. For the second and third lettuce cycles, the 2.4.1. Goal and scope and inventory initial struvite was assumed to be the remaining struvite in the perlite An environmental assessment of the fertilization method was made, after the previous crop cycle. comparing the environmental load to produce 1 kg of fresh lettuce and pepper pods considering the incoming fertilizers and outgoing emissions to water and air, as shown in Fig. 1(C). The life cycle assessment (LCA) 2.4. Environmental assessment tool was used to determine the environmental impact of fertilization for all treatments, which was calculated with Simapro 9.1 software, using The Life Cycle Assessment (LCA) used to determine the environ­ the Ecoinvent 3.7 database to account for the background mental impact of the irrigation system followed the ISO norms 14,040 4 V. Arcas-Pilz et al. Resources, Conservation & Recycling 179 (2022) 106130 Fig. 3. P amount (g) in pepper biomass, fruit, and perlite for treatments 5P, 10P, 20P and CP at 81 DAT given in total g per plant. environmental information and the ReCiPe midpoint impact assessment were based on the author’s expertise and previous literature focusing on method (Huijbregts et al., 2017). The scope of this attributional LCA was the impacts of fertilizers in soilless systems and the use of struvite defined as cradle to grave since the production, transport, use and (Brentrup et al., 2004; Sanjuan-Delmás et al., 2018; Rufí-Salís and disposal stages for the different fertilizers were considered. On the other Brunnhofer, 2020). hand, the greenhouse infrastructure and production system were not Global warming, expressed in kg CO2 eq. was chosen due to the included in the system boundaries, focusing only on the impact of the documented relevance of greenhouse gas emissions from the production use of struvite as fertilizer. of fertilizers, as well as their transport, and due to the direct emissions occurring at the plant level (Hasler et al., 2015; Chatzisymeon et al., 2.4.2. Life cycle incentory 2017). This case is especially true for nitrous oxide (N2O). Thus, we The inventory for the LCA was comprised by the obtained data from considered the additional nitrogen given in the form of ammonia the experiment. The fertilizer applied was obtained from the nutrient through struvite. The proportional fraction of ammonia released in each solution and irrigation amount controlled daily through the water me­ treatment, determined through direct measurement in the leachates, ters installed in the irrigation system. The incoming irrigation as well as was considered when calculating the emission factor as well as the ni­ the leachate water was analyzed to obtain the N and P emitted to water trogen given in the irrigation. For the same reason, TA (kg SO2 eq.) was (in the form of NO2− , NO3− and PO4− ). From the incoming irrigation the also chosen to reflect the direct emissions due to the application of calculation of the N emissions to air in the form of ammonia (NH3), ammonia as well as other acidifying agents generated during trans­ nitrous oxide (N2O) and nitrogen oxides (NOx) following the emission portation and manufacturing of fertilizers (Hasler et al., 2015). FE (kg N factors established by Montero et al. (2011). eq.) and ME (kg P eq.) have been regarded as the most relevant impact The generation of struvite was accounted for in the environmental categories when analyzing fertilization methodologies, especially assessment based on the production of the commercial house Ostara®. considering nitrogen and phosphorous (Hasler et al., 2015; Chatzisy­ For the production of 1 kg of struvite, the additional chemical input in meon et al., 2017; Vatsanidou et al., 2020). These impact categories are the precipitation stage was 0.4239 kg MgCl, 0.766 kg NaOH for pH especially relevant in this study since slow struvite dissolution can stabilization and 0.523 kWh energy applied for mixing and aeration provide insight into the possible reduction of P leaching into fresh and (Amann et al., 2018). Impacts related to wastewater treatment, such as marine waterbodies, and again, the addition of N through struvite can an improvement of the effluent or the additional technology required for also be reflected in the leachate quantities. FRS (kg oil eq.) was added as P removal to the sludge line, were excluded from the system boundaries a relevant impact category to reflect fossil energy-related emissions that of this LCA. The transport for all fertilizers accounted for 50 km from the could arise due to struvite precipitation and transport compared to greenhouse. The struvite transport, on the other hand, was estimated to mineral P. Finally, MRS (kg Cu eq.) was chosen to reflect the extraction be 30 km, which corresponds to the approximate distance to the two of finite mineral resources, especially focused on phosphate rock nearest WWTPs of the city (EDAR Besós and EDAR Llobregat), although extraction versus the recycling and reuse of phosphorous in the form of they currently are not producing or selling struvite. struvite. The environmental assessment was made for a single plant pot, taking into account its fertilization and production. These results can 2.5. Statistical analysis then be further extrapolated to greater production. The detailed in­ ventory and processes can be seen in the supplementary information Shapiro–Wilk test p > 0.05 was used to test the normality of the data, (Table 5) while homogeneity of variance was tested with Levene’s test p > 0.05. Duncan’s multiple range test was used to assess the statistical signifi­ 2.4.3. Environmental impact assesment cance of treatments when parametric criteria were validated. For The impact categories selected were global warming (GW), terres­ nonparametric data, the Kruskal–Wallis test was used. The significance trial acidification (TA), freshwater eutrophication (FE), marine eutro­ between the treatments is marked with different letters (a, b and c). phication (ME) and mineral resource scarcity (MRS). These selections 5 V. Arcas-Pilz et al. Resources, Conservation & Recycling 179 (2022) 106130 3. Results Table 3 Total yield obtained in four pepper fruit harvests (55 DAT, 62 DAT, 72 DAT and 3.1. Crop growth and yield production 81 DAT) for each treatment. The yield given in g while an average fruit weight given with g/ fruit. The resulting productions for the three lettuce cycles can be observed Harvest Treatment Total Fruit Fruit Weight/ in Table 1. Here, we can appreciate the average yields obtained for all Weight (g) number (nr) number (g/ fruit) three harvests and all treatments for their fresh and dry weights. Further 1ST HARVEST 5P 1261.9 22 57.4 information on the specific production within the marked pots (A to O) (55 DAT) can be seen in Table 2 in the supplementary information. 10P 1402.7 21 66.8 We identified a general decrease in yield during the third harvest, 20P 2155.5 25 86.2 most likely due to a remarkable decrease in the overall temperature CP 1479.0 22 67.2 2ND HARVEST 5P 911.0 18 50.6 during 54 DAT and 81 DAT in contrast to the previous crop cycles (1 (62 DAT) DAT to 54 DAT). This variation in the climatic conditions can be 10P 1240.0 19 65.3 observed in Fig. 1 in the supplementary information with the recordings 20P 1833.4 28 65.5 of humidity, radiation and temperature during all three cycles. While CP 1597.0 21 76.0 3RD HARVEST 5P 528.0 9 58.7 temperatures still ranged between 20 ◦ C and 25 ◦ C, the sudden change in (72 DAT) comparison to the previous two crop cycles may have caused a delay in 10P 759.0 10 75.9 lettuce growth. 20P 1649.2 18 91.6 While no great differences can be seen in the overall yield of our CP 1046.0 13 80.5 lettuce cycles, the close monitoring of our pots can give us the variability 4TH HARVEST 5P 860.0 20 43.0 (81 DAT) of the obtained yield for the lettuces grown with the same initial stru­ 10P 940.0 21 44.8 vite. This finding means that from the same pot, we can monitor the 20P 1860.0 29 64.1 obtained yield in all three cycles. Table 2 in the supplementary infor­ CP 1225.0 19 64.5 mation provides us with such information showing a general decline in TOTAL 5P 3560.9 69 51.6 10P 4341.7 71 61.2 production, with the most acute decrease in yield in the 5LE treatment 20P 7498.1 100 75.0 with a − 11% difference between the first harvest and the second. On the CP 5347 75 71.3 other hand, the decline for treatments 10LE and 20LE was less pro­ nounced, with − 2% for both. In the case of pepper plant growth and production, Tables 2 and 3 remaining in the substrate were analyzed and plotted in Fig. 2. Here, we provide insight into the differences spotted between treatments. Table 2 can appreciate a great difference between the struvite fertilization provides the main measurements made of the pepper plants at the end of treatments and the control, since the nutrient content in the perlite the experiment (81 DAT). slowly increases over time for the latter, while the P content in the While no significant differences were seen for the stem weight, an struvite treatments fluctuates and slowly decreases due to its dissolu­ increase in the fresh and dry weight was observed with the increasing tion. Here, again, a much greater variability in the results was observed amount of struvite applied. The same increase was noted for the leaf for the second cycle. weight, number, and LAI, showing significant differences in all but the Fig. 3 depicts the P content in pepper biomass, fruit and perlite, latter. The control treatment generally showed greater values in all showing great variation between struvite fertilization treatments and measurements apart from one, the plant stem length. the control. While our treatments showed a low P content of 1.2 mg/g in The yield produced by the pepper plants (Table 3) showed a greater leaves and 0.7–0.8 mg/g in the plant stem, giving ranges of 0.02 to 0.03 total weight for the 20P treatment. While the total number of fruits was g of P in the total dry biomass, the control treatment showed values also higher for the 20P treatment, the weight per fruit did not differ within adequate ranges of 2.1 mg/g (0.08 g of P in the total dry greatly from that of the other treatments. biomass). The amount of P in the harvested fruits reveals the differences between treatments based on the great mobility within the plant. Fruits 3.2. Nutrient content in plant biomass and substrate are an ultimate sink of the phosphorous content in plants, and this result is reflected with a very clear relation to the struvite treatment. The great The results shown in Fig. 2 depict the P content in the lettuce crop variability seen in these results derives from the great difference found after 27 days of growing in the greenhouse for all treatments. The between harvests within the same treatment, while the first pepper fruit amount of P found in the lettuce biomass is directly related to the harvest contained greater P concentrations, the third suffered a great amount of struvite given, being lowest for the 5LE treatment followed by reduction for all treatments, even the control (Supplementary informa­ the 10 L and finally 20LE treatments. The amounts of P found in the 5LE tion Figure 7). Finally, the amount of P found in perlite responds to the and 10LE treatments decrease noticeably over time in the second and initially given struvite. third cycles, while the 20LE treatment does not experience a great reduction during the second cycle but rather on the third cycle. It is 3.3. Phosphorous content in the leachate important to point out that the results found for the second crop cycle show a much greater variability than the first and third ones. The resulting phosphorous concentrations found in the leachates The remaining struvite content in the perlite and therefore P were calculated for the total outgoing water weekly per plant, Table 2 Mean values of pepper plant biomass measurements made in the 81 DAT. Stem weight and Leaf weight given in g for their fresh weight (FW) and dry weight (DW). *The LAI calculated in cm2. Significant differences (p< 0.05) between treatments marked with different letter (a,b,c). TREATMENT Stem weight (g/plant) Leaf weight (g/plant) Stem length (cm) Leaf number (nr) LAI* FW DW FW DW 5P 169.3a 32.2a 131.9a 21.2a 99.1ab 110.1a 2.5 a 10P 184.0a 35.2a 166.7ab 26.0ab 107.1b 127.3ab 3.3 a 20P 204.1a 40.9a 195.2bc 31.4b 96.0ab 138.9ab 3.8 a CP 220.1a 43.9a 236.2c 35.5b 90.0a 156.3b 4.1 a 6 V. Arcas-Pilz et al. Resources, Conservation & Recycling 179 (2022) 106130 generating the patterns found in Fig. 4. The accumulation of P in the remained similar. The remaining phosphorous in perlite remained high leachates for the lettuce and pepper crops can be seen in supplementary in the 5LE and 10LE treatments, while more than half was reduced in the information Fig. 6. 20LE treatment. We also appreciate an accumulation of P in the perlite The results for the lettuce crop show the discharge of phosphorous of the CLE treatment. during all three cycles, recognizing a clear pattern before and after each For pepper, the biomass flows were divided between the fruits pro­ harvest. This pattern was highly noticeable for the CLE treatment, where duced and the generated biomass on the day the plants were cut and the phosphorous content in the leachates decreased with the growth of weighed. Here, we can appreciate the great quantity of phosphorous the plant and rose once the plant was harvested and replaced with a found in the pepper fruits, which equals the total phosphorous found in seedling. This same pattern can be observed for all struvite fertilization the plant leaves and stem. The total biomass showed a great difference treatments for lettuce, finding greater amounts for 20LE and less for between the CP treatment and the struvite fertilization treatments, 10LE and 5LE. revealing a much greater P content in the control. Due to the greater The phosphorous content in water, on the other hand, differs greatly irrigation needs of pepper plants compared to lettuce plants, the CP when observing the CP and the 5P, 10P and 20P treatments. The biomass treatment received an overall greater amount of P through irrigation growth, climatic conditions and subsequent irrigation amount define the compared to the CLE treatments. Therefore, although the P in the perlite loss of phosphorous in the CP treatments, showing an overall decrease in and leachates is lower in CP than in the CLE treatments in terms of the concentration with a peak at approximately Day 37 after trans­ percentage, the absolute amounts are greater. In the case of the pepper planting. All treatments with struvite showed very low concentrations in plants fertilized with struvite, the P in the leached water was similar and the leachates, especially after 20 DAT. even smaller than the amount found in the lettuce crop. The outgoing P in the leachates of the pepper plants was 10, 19 and even 35 times lower 3.4. Phosphorous balance than that in the control treatment (CP) for the 20P, 10P and 5P treat­ ments, respectively. The results obtained in the previous sections enable us to generate The calculated dissolution rates for the applied struvite in lettuce and the nutrient balance for P during these cycles for all treatments. This pepper are shown in Fig. 5. The struvite dissolution was estimated by the understanding helps us estimate the P flows into the plant, substrate and P contained in the water leachates as well as P in the plant biomass. This water. These nutrient balances were calculated for the P flows in the dissolution has a direct impact on the P uptake by the plant that was lettuce and pepper crops (Table 4) and averaged to obtain data for one estimated as the P contained in the P biomass. The results for lettuce plant. In addition, the water balances per plant are given in Figures 9 show greater dissolution with a greater initial amount of struvite. The and 10 in the supplementary information. The nutrient balance is sub­ dissolution of the struvite was also higher during the first lettuce cycle jected to potential inaccuracies given through the sampling of substrate, (marked with number 1 in the figure), showing smaller differences be­ water and biomass and the generation of mean values for all samples tween the second and third cycles (marked with 2 and 3, respectively). generating approximate values close to 100%. The dissolution rate found in the pepper crop was smaller than that in The balance for lettuce gives an overall picture of the obtained re­ the lettuce crop but followed the same pattern as seen before, with sults of the phosphorous flows into the plant biomass as well as leach­ greater dissolution with higher amounts of struvite. ates. Compared to the control treatment, the phosphorous flow into the outgoing water was approximately 10 to 14 times lower for the 10LE and 5LE treatments, respectively, while the flow into the plant biomass Fig. 4. Total phosphorous (Total g) found in the leachate water from 4 DAT to 74 DAT in Lettuce and Pepper crops for treatments with 5 g, 10 g and 20 g of struvite as well as Control treatments (CP and CLE) irrigated with KPO4H2. 7 V. Arcas-Pilz et al. Resources, Conservation & Recycling 179 (2022) 106130 Table 4 Phosphorous balance per plant for the lettuce and pepper crop for treatments 5LE, 10LE, 20LE, CLE, 5P, 10P, 20P and CP. Initial P given through struvite and NS. Biomass 1, 2 and 3 corresponding to the 3 Lettuce cycles at 27 DAT, 54 DAT and 81 DAT. Pepper total biomass corresponds to total P found in Fruit Production and Biomass.* For the biomass the amount of P from the root was also included with the root DW and root phosphorous content obtained from literature (Xu et al., 2004; Pereira-Dias et al., 2018; Erel et al., 2019). TREATMENT INITIAL P BIOMASS 1* BIOMASS BIOMASS 3* TOTAL PERLITE LEACHATES BALANCE 2* BIOMASS LETTUCE Struvite g % g % g % g % g % g % % /NS g 5LE 0.625 0.047 8 0.047 8 0.035 6 0.130 21 0.443 71 0.022 4 95 10LE 1.25 0.064 5 0.050 4 0.042 3 0.157 13 0.885 71 0.028 2 86 20LE 2.5 0.077 3 0.079 3 0.064 3 0.221 9 1.196 48 0.085 3 60 CLE 1.049 0.071 7 0.081 8 0.044 4 0.196 19 0.384 37 0.318 30 86 INITIAL P PRODUCTION BIOMASS* TOTAL PERLITE LEACHATES BALANCE BIOMASS PEPPER Struvite g % g % g % g % g % % /NS g 5P 0.625 0.029 5 0.069 8 0.099 12 0.561 90 0.014 2 104 10P 1.25 0.055 4 0.076 4 0.130 9 0.904 72 0.026 2 83 20P 2.5 0.086 3 0.092 3 0.178 6 1.602 64 0.051 2 72 CP 2.595 0.106 4 0.189 7 0.295 11 0.709 27 0.501 19 57 Fig. 5. Correlation of the dissolution rate of the struvite P and initial P given in Lettuce and Pepper for treatments 5LE, 10LE, 20LE, 5P, 10P and 20P For lettuce all three cycles are taken in account, marked as 1, 2 and 3 respectively. 3.5. Life cycle assessment ME, although related to the emissions to water, also does not decrease substantially for the struvite-treated crops due to its relation to Fig. 6 shows the results for lettuce, and Fig. 7 shows the results for nitrogen emissions, which are sustained for all treatments. Furthermore, the environmental assessment of the fertilization treatments. Since only we can observe that although a reduction of the impacts is occurring for the fertilization of the crops was considered for the analysis (Fig. 1), all the 20LE treatment, this reduction is most likely not a consequence of a differences will be related to the use of struvite instead of monop­ reduced N emission to water but due to greater yields obtained; on the otassium phosphate (MKP) in the form of KPO4H2, leaving out the lab­ other hand, treatment 5LE is overshadowed by the lower yields gener­ oratory infrastructure and auxiliary equipment, as well as the end-of-life ated and a proportionally greater N emission due to the smaller plant processes. growth. The obtained results for six out of seven impact categories show that The results obtained for the pepper crop indicate a considerably fertilization with struvite has lower impacts than the control, and for the abrupt decrease in the emissions in all impact categories for treatments cases of ET, MRS, FRS and GW, impacts are also reduced as we increase 10P and 20P in comparison to the CP treatment. In comparison to the the amount of struvite applied. In terms of eutrophication, FE, which is lettuce crop, the ME was severely reduced for these two treatments. The directly related to the emissions to water, had the greatest impact on the 5P treatment with lower production rates and therefore lower FU ex­ control irrigated with mineral P, followed by 20LE, which was the periences much greater values for all impact categories except FE and treatment with the highest quantity of struvite per plant. MRS, which are slightly below the control treatment CP in the latter. 8 V. Arcas-Pilz et al. Resources, Conservation & Recycling 179 (2022) 106130 Fig. 6. Impact assosciated to the system fertilization for 3 consecutive lettuce productions for 81 days. The obtained emissions have been calculated in relation to the resulting yield as FU. GW (global warming), TA (Terrestrial acidification), FE (Freshwater eutrophication), ME (Marinewater eutrophication), ET (Ecotoxicity), FRS (Fossile resource scarcity), MRS (Mineral resource sarcity). *Emissions to air were based on the emission factors of Ammonia, N2O and NOx obtained for the applied nitrogen; the emissions to water were directly obtained from the nitrogen and phosphorous detecte in the water leachate. Impact assosciated to the system fertilization for pepper production during 81 days. The obtained emissions have been calculated in relation to the resulting yield as functional unit. GW (global warming), TA (Terrestrial acidification), FE (Freshwater eutrophication), ME (Marinewater eutrophication), ET (Ecotoxicity), FRS (Fossile resource scarcity), MRS (Mineral resource sarcity). *Emissions to air were based on the emission factors of Ammonia, N2O and NOx obtained for the applied nitrogen; the emissions to water were directly obtained from the nitrogen and phosphorous detected in the water leachate. Overall emissions for pepper production were lower than those unit only accounting for the obtained yield. found for the three lettuce cycles combined. This finding can be explained by the greater weight obtained with pepper production, making a direct comparison between crops difficult with a functional 9 V. Arcas-Pilz et al. Resources, Conservation & Recycling 179 (2022) 106130 4. Discussion present study. While the pepper plants did not reach adequate ranges of P in the biomass with struvite fertilization, the lettuce crops did not 4.1. Is struvite a good fertilizer for hydroponic production? differ greatly from the control treatment, especially for 10LE and 20LE. This information reinforces the idea of further reusing the given struvite The results show that the long cycle of pepper and short cycles of for consecutive cycles within the same substrate with short cycle crops, lettuce fertilized with struvite did not differ greatly from each other in such as lettuce. the uptake and use of P. We identified that the amounts accumulated in On the other hand, the dissolution rate seems to be greater during the the plant biomass between treatments with the same struvite quantity first plant cycle in all lettuce treatments. The struvite crystal composi­ (5LE and 5P, 10LE and 10P, 20LE and 20P) did not change substantially. tion and available P could be more prone to dissolve earlier, progres­ This information reveals that little to no effect on struvite uptake can be sively reducing the dissolution rate with consecutive plant cycles. This attributed to the crop cycle duration or needs. This second idea is same dissolution trend was seen by Rech et al. (2018) when observing reinforced by the level of P found in the pepper biomass, corresponding the P concentration in the soil solution of wheat and soybean crops with to low concentrations and mirrored in the fruit P content (Hochmuth the fertilization of three different struvite types. Concentrations of P et al., 2018). Although a clear P deficiency is shown in the plant biomass were recorded for 40 days, showing a decrease and stability by the end nutrient content, no such deficiency can be traced in the plant physi­ of the experiment. The close dissolution rate of the second and third ology or production capacity (Zelia et al., 2017). Pepper fruit production cycles could indicate this point of stability. increases with the given struvite, as well as leaf production and growth, showing significant differences that indicate the relevance of the given 4.2. Does the use of struvite reduce the environmental burden of struvite amount to the plant. Related to the findings of Talboys et al. hydroponic production? 2016 in 90-day experiments with struvite-fertilized crops, the amount of P taken by the plant is lower in the case of struvite but does not affect the The environmental analysis showed that the 5LE and 5P treatments final yield, being very similar to the more soluble triple superphosphate had the highest impacts since they had the lowest yields. On the other (TSP). This finding has been attributed to the struvite residual value in hand, the greater use of struvite can also generate a greater discharge of the substrate in comparison to TSP, enabling P uptake by the plant P into the water system compared to treatments with less applied stru­ during a sustained timespan. vite. This finding is reflected in the case of the lettuce crops for the ME The leachate P for lettuce and pepper plants was also shown to be a and FE impact categories. While greater yields were achieved for the great indicator of the slow solubility of the fertilizer and increased with 20LE treatment, greater P and N leachates were generated, increasing greater water flow when lettuce was harvested. The higher water de­ the environmental footprint in comparison to the other struvite treat­ mands of the pepper plants could therefore have been a defining factor ments. Smaller crop growth in the case of 5LE and 5P can also increase contributing to low struvite dissolution, as seen in Fig. 5. Although the the amount of leachates and discharge of N to the environment. This plants had sufficient irrigation indicated by the daily water drainage, the finding has been observed both for lettuce and pepper, where smaller leaching of phosphorous into the drained water only increased during crop growth leads to greater water and nutrient discharge. However, the the early stages of plant growth until 20 DAT. Once temperatures start to P discharge in the struvite treatments was always lower than that in the rise and drainage is reduced, the emissions of P into the drainage are also control and thus impacted freshwater eutrophication. reduced. Although greater temperatures have been seen to increase The impact of the struvite production compared to the monop­ struvite solubility (Rahaman et al., 2006; Ariyanto et al., 2017), its use otassium phosphate seem significantly smaller, being most noticeable in as a fertilizer unveils that irrigation plays a major role in plant phos­ GW and FRS. The production of monopotassium phosphate on the other phorus uptake (Lunt et al., 1964; Turner, 1985; Silber et al., 2005). hand has a large impact on the MRS as predicted, due to the extraction of The greater variability obtained in the second lettuce cycle can also the finite phosphate rock. The impact of monopotassium phosphate is be attributed to the increasing temperatures enabling a greater disso­ also noticeable in the ET, TA and FRS categories, responding to the lution of struvite in the perlite substrate as well as the slight reduction of emissions of chemical agents into the environment for the extraction and the pH from the nutrient solution increasing the struvite solubility transport to site. The overall impacts seem to be more dominated by the (Ariyanto et al., 2017). The uptake and use of the plant could have been production emissions associated to potassium sulfate, being present in affected by the delay in the irrigation adjustment to meet the plant almost all IC due to its major role in the nutrient solution. needs. Takin in account the influence of the struvite slow solubility to The capacity of struvite dissolution, which has been attributed to reduce the emissions of P to water as well as the reduction of the impacts different factors in previous literature, like the plant rhizosphere associated to the production of monopotassium phosphate, a great exudation (Kamilova, 2006; Khademi, 2010; Dakora and Phillips, 2002; reduction of the impacts of fertilization can be seen. Talboys et al., 2016), plant growth stage (Degryse et al., 2017) and plant While the pepper crop shows a clear reduction in emissions related to needs. These factors for greater struvite dissolution have not been re­ fertilization with the use of 10 g and 20 g of struvite, sustained pro­ flected in these results, indicating a reduced uptake from the pepper duction is unclear due to the low content of P in plants. While the pro­ plant compared to the lettuce crop. The idea of plant rhizosphere duction of pepper continues and demands on P can increase, its exudation being important for struvite dissolution was also questioned dissolution and uptake might not be sufficient in time. On the other by Rech et al., 2018, who demonstrated the inefficiency of hand, the lettuce needs were covered for all three cycles for all treat­ low-concentration root exudates to solubilize granular struvite. ments, showing a P content similar to that of the control treatment. The Overall, the quantity of P in the plant biomass (9–21% and 6–12% of idea of sustained production for longer periods of time corresponds to the applied P for lettuce and pepper, respectively) as well as the P the findings of Bonvin et al., 2015 and Rech et al., 2018 urging for the leachate (2–4% and 2% of the applied P for lettuce and pepper, definition of the residual value of the remaining struvite after the initial respectively) in both crops indicate that the amount of dissolved P is crop production. very small. This information is reinforced by the analysis of the perlite To understand the environmental impact of one year of lettuce cy­ substrate, indicating that a large amount of struvite remains undissolved cles, several assumptions were made. To generate the nine-cycle sce­ in the substrate. This effect was also seen in previous literature with nario (Supplementary Information Table 4) that would correspond to other crops, such as soybean and wheat (Rech et al., 2018) and common yearly production, the three initial cycles for our three treatments were bean (Arcas-pilz, 2021a). These low dissolution percentages coincide taken as references to generate correlations for the P uptake in biomass with dissolutions in media with pH values ranging from 7.5 to 8 (Tal­ from the initial P given, as well as the potential yield produced with the boys et al., 2016; Rech et al., 2018), which were mainly found in the P content in the plant biomass (Supplementary Information Fig 8). 10 V. Arcas-Pilz et al. Resources, Conservation & Recycling 179 (2022) 106130 Further on, the error detected in this last correlation was subjected to a well as a reduction and adjustment of the nutrient solution N content sensitivity analysis adding a standard deviation of a total 46% to the with the addition of struvite. yield production for a 9 cycle production of lettuce in all treatments. The Further loss into the environment can be assessed with a specific control was also given a standard deviation of 10%, taking in account analysis of the struvite nitrogen emission factor to the air in the form of that the P fertilization was consistent over time. ammonia, N2O and NOx in soilless systems, which is strongly encour­ The P loss through the leachates was estimated from the average aged to determine the GW impact more accurately. This result has been obtained in all treatments due to its direct relation with irrigation. With viewed as both interesting and necessary research to understand the following prediction, the total biomass content of yearly production whether slow dissolution can discourage emission to air or if the as well as the resulting yield and emissions to water were obtained to composition of N struvite in the form of ammonia will further induce further extend the environmental outcome (Fig. 8). The control treat­ processes of nitrification and denitrification in the substrate. ment was estimated with the generated yields and emissions from the The findings in this work point out that the successful reuse of three initial cycles. All other fertilizers for all treatments were based on struvite in hydroponic production is possible and is been growing in the NS used for the three initial cycles extended for nine production importance, even being used in the fertirrigation for other crops cycles. The obtained emissions were then divided by the obtained total achieving equal results to conventional fertilizers (Carreras-Sempere yields. et al., 2021). Similar work has been made with source separated urine, The LCA for the year’s production with the same initial struvite integrated into hydroponic production as nutrient source, and also using shows a slight emission increase for all ICs, especially for the 5LE and phytoremediation systems for yellow water treatment (Ikeda and Tan, 10LE treatments. The changes observed indicate that the productions 1998; Yang et al., 2015; Simha and Ganesapillai, 2017; Volpin et al., obtained for the 5LE and 10LE treatments decrease to a point where the 2018). These works have found promising results on the reuse of urine functional unit is reduced and consequently emissions are increased. On although its application can be considered controversial (Simha et al., the other hand, control treatment yields were sustained in time and 2017). maintained close to identical emissions of the three lettuce cycles. The This new way to find circularity in urban ecosystems is deemed as prospective production obtained for the 20LE treatment was similar to necessary and imposed specially in the waste treatment sector. The ca­ that of the control, obtaining results that reduced the environmental pacity to find an added value to the outcome of urban waste can help emissions for all impact categories compared to the control treatment achieve new environmental goals like the compulsory recovery of P in except ME. certain regions of the EU (Kratz et al., 2019). The local P recuperation The 20LE treatment maintains the capacity for competitive produc­ and local administration can increase the local resilience to P pricing tion in time compared to the other struvite treatments which can also be and distribution; therefore the P precipitation and struvite production seen in the sensitivity analysis in fig 11 in the supplementary informa­ should be encouraged in WWTP. tion, staying below the control emissions in lower production scenarios, especially for FE and MRS. This, however, implies a potential greater 5. Conclusions emission to water, as reflected in the FE and ME impact categories generated by the leaching of the struvite containing N and P. The use of The main conclusions drawn from the present experiment can be the discharged water for less demanding crops (Rufí-Salís and Calvo, divided into two main aspects, one regarding the production and uptake 2020) can further reduce nutrient leaching into the urban water cycle as by the plants and the second regarding the environmental benefits when Impact assosciated to the system fertilization for 9 consecutive lettuce productions. The obtained emissions have been calculated in relation to the prospective yield as FU. GW (global warming), TA (Terrestrial acidification), FE (Freshwater eutrophication), ME (Marinewater eutrophication), ET (Ecotoxicity), FRS (Fossile resource scarcity), MRS (Mineral resource sarcity). *Emissions to air were based on the emission factors of Ammonia, N2O and NOx obtained for the applied nitrogen; the emissions to water were directly obtained from the nitrogen and phosphorous detected in the water leachate. 11 V. Arcas-Pilz et al. Resources, Conservation & Recycling 179 (2022) 106130 compared to the use of mineral fertilizer. We found that the three cycles Supplementary materials of lettuce treated with 20 g of struvite had the highest and most sus­ tained overall yield, although such a high struvite concentration resul­ Supplementary material associated with this article can be found, in ted in very slow dissolution. In this case, 50% to 70% of the struvite the online version, at doi:10.1016/j.resconrec.2021.106130. remains undissolved in the substrate after three lettuce cycles, indi­ cating great potential for further consecutive production cycles. Esti­ References mations of a year’s worth of lettuce production with the same initial struvite indicate sustained production similar to the control, while Achat, D.L., et al., 2014. Plant-availability of phosphorus recycled from pig manures and dairy effluents as assessed by isotopic labeling techniques. Geoderma 232–234. production for all struvite treatments apart from 20LE would be https://doi.org/10.1016/j.geoderma.2014.04.028, 24–33. reduced. While no signs of P deficiency can be seen in the pepper plants, Ahmed, N., et al., 2018. Struvite recovered from various types of wastewaters: even when obtaining a greater production, the P content was regarded as characteristics, soil leaching behaviour, and plant growth. Land Degradation and Dev. 29 (9), 2864–2879. https://doi.org/10.1002/ldr.3010. very low due to the slow struvite solubility. Pepper production was Alewell, C., et al., 2020. Global phosphorus shortage will be aggravated by soil erosion. successful in the three-month experiment, although longer production Nat. Commun. 11 (1), 4546. https://doi.org/10.1038/s41467-020-18326-7. cycles were not tested. Amann, A. et al. (2018) ‘Environmental impacts of phosphorus recovery from municipal wastewater’, Resour. Conserv. Recycl. doi: 10.1016/j.resconrec.2017.11.002. The environmental outcome of the experiment shows a general Appolloni, E. et al. (2021) ‘The global rise of urban rooftop agriculture : a review of reduction in the environmental impacts, especially regarding the use worldwide cases’, 296. doi: 10.1016/j.jclepro.2021.126556. and emission of P for freshwater eutrophication and mineral resource Arcas-pilz, V., et al., 2021b. Recovered phosphorus for a more resilient urban agriculture : assessment of the fertilizer potential of struvite in hydroponics. Science scarcity. The production of 20LE is sustained over time, therefore of the Total Environment 799. https://doi.org/10.1016/j.scitotenv.2021.149424. reducing the emissions below the control treatment except for ME. The Arcas-Pilz, V., et al., 2021a. Assessing the environmental behavior of alternative greater N emissions to water associated with the ME can be reduced by fertigation methods in soilless systems: the case of Phaseolus vulgaris with struvite adjusting the nutrient solution, considering the N delivered by the and rhizobia inoculation. Sci. Total Environ. 770, 144744 https://doi.org/10.1016/ j.scitotenv.2020.144744. struvite. The findings of this study further encourage the use of struvite Ariyanto, E., Ang, H.M., Sen, T.K., 2017. The influence of various process parameters on in hydroponic production due to the capacity of sustained production of dissolution kinetics and mechanism of struvite seed crystals. J. Institution of shorter and longer cycle crops as well as the reduction of the environ­ Engineers (India): Series A. Springer India 98 (3), 293–302. https://doi.org/ 10.1007/s40030-017-0212-4. mental impacts compared to mineral fertilizer, such as MKP. Ayers, A.C., Rehan, S.M., 2021. Supporting bees in cities: how bees are influenced by local and landscape features. Insects. https://doi.org/10.3390/insects12020128. 6. Author contributions Baganz, G., et al., 2020. Profitability of multi-loop aquaponics: year-long production data, economic scenarios and a comprehensive model case. Aquac. Res. 2711–2724. https://doi.org/10.1111/are.14610. All authors were responsible for the conception and design of the Baró, F., et al., 2014. Contribution of ecosystem services to air quality and climate study. V. Arcas-Pilz, M. Rufí-Salís, F. change mitigation policies: the case of urban forests in Barcelona, Spain’. Ambio. https://doi.org/10.1007/s13280-014-0507-x. Parada, G. Villalba and X. Gabarrell conceived the original idea for Bonvin, C., et al., 2015. Plant uptake of phosphorus and nitrogen recycled from synthetic the study. V. Arcas-Pilz, G. Stringari, R. source-separated urine. Ambio 44 (2), 217–227. https://doi.org/10.1007/s13280- R. Gonzalez and F. Parada set up, supervised and acquired the data 014-0616-6. Borgerding, J., 1972. Phosphate deposits in digestion systems. J. Water Pollut. Control for the experimental tests. V. Arcas-Pilz processed and analysed the data. Fed. 44 (5), 813–819. V. Arcas-Pilz took the lead in writing the manuscript. All authors criti­ Bouropoulos, N.C., Koutsoukos, P.G., 2000. Spontaneous precipitation of struvite from cally revised the draft for important intellectual content. All authors aqueous solutions. J. Crystal Growth. North-Holland 213 (3–4), 381–388. https:// gave their final approval to the manuscript. doi.org/10.1016/S0022-0248(00)00351-1. Bradford-Hartke, Z., et al., 2015. Environmental benefits and burdens of phosphorus recovery from municipal wastewater. Environ. Sci. Technol. 49 (14), 8611–8622. Declaration of Competing Interest https://doi.org/10.1021/es505102v. Brentrup, F., et al., 2004. Environmental impact assessment of agricultural production systems using the life cycle assessment (LCA) methodology II. The application to N The authors declare that they have no known competing financial fertilizer use in winter wheat production systems. Eur. J. Agronomy 20 (3), 265–279. interests or personal relationships that could have appeared to influence https://doi.org/10.1016/S1161-0301(03)00039-X. the work reported in this paper. Camps-Calvet, M., et al., 2016. Ecosystem services provided by urban gardens in Barcelona, Spain: insights for policy and planning. Environ. Sci. Policy. https://doi. org/10.1016/j.envsci.2016.01.007. Acknowledgements Carpenter, S.R., Bennett, E.M., 2011. Reconsideration of the planetary boundary for phosphorus. Environ. Res. Letters. https://doi.org/10.1088/1748-9326/6/1/ 014009. The authors are grateful to the Spanish Ministry of Economy, In­ Carreras-sempere, M. et al. (2021) ‘Use of Recovered Struvite and Ammonium Nitrate in dustry and Competitiveness (Spain) for the grant awarded to V. Arcas- Fertigation in Tomato (Lycopersicum esculentum) Production for Boosting Circular Pilz (FPI-MINECO 2018), to the National Commission for Scientific and Sustainable Horticulture’, (November). doi: 10.3390/agriculture11111063. Chatzisymeon, E., Foteinis, S., Borthwick, A.G.L., 2017. Life cycle assessment of the and Technological Research (Chile) for the grant awarded to F. Parada environmental performance of conventional and organic methods of open field (PFCHA-CONICYT 2018 – Folio 72180248) and to the Generalitat de pepper cultivation system’, International Journal of Life Cycle Assessment. Int. J. Catalunya for the grand awarded to G. Stringari (2020 FI_B 01004). This Life Cycle Assessment 22 (6), 896–908. https://doi.org/10.1007/s11367-016-1204- work was supported by the “María de Maeztu” program for Units of 8. Chen, P., et al., 2020. Comparative life cycle assessment of aquaponics and hydroponics Excellence in R&D [ CEX2019- 000940-M]. in the Midwestern United States. J. Clean. Prod. 275, 122888 https://doi.org/ The research leading to these results has received funding from the 10.1016/j.jclepro.2020.122888. European Research Council under the European Horizon2020 research Cordell, D., Drangert, J.O., White, S., 2009. The story of phosphorus: global food security and food for thought. Global Environ. Change 19 (2), 292–305. https://doi.org/ and innovation program under grant agreement No 862663 (FoodE) as 10.1016/j.gloenvcha.2008.10.009. well as, ERC grant agreement n◦ 818002 URBAG, awarded to Gara Cordell, D., White, S., 2013. Sustainable phosphorus measures: strategies and Villalba. Finally, we would like to thank Secretaria d’Universitats i technologies for achieving phosphorus security. Agronomy. https://doi.org/ 10.3390/agronomy3010086. Recerca del departament d’Empresa i Coneixement de la Generalitat de Cordell, D., White, S., 2014. Life’s bottleneck: sustaining the world’s phosphorus for a Catalunya for the grand awarded under the n◦ AGAU 2020 PANDE food secure future. Annu. Rev. Environ. Resour. https://doi.org/10.1146/annurev- 00021. The publication reflects the author’s views. The Research Ex­ environ-010213-113300. Le Corre, K.S., et al., 2009. Phosphorus recovery from wastewater by struvite ecutive Agency (REA) is not liable for any use that may be made of the crystallization: a review. Critical Rev. Environ. Sci. Technol. Taylor & Francis Group. information contained therein. https://doi.org/10.1080/10643380701640573. 12 V. Arcas-Pilz et al. Resources, Conservation & Recycling 179 (2022) 106130 Degryse, F., et al., 2017. Dissolution rate and agronomic effectiveness of struvite Rahman, M.M., et al., 2014. Production of slow release crystal fertilizer from fertilizers – effect of soil pH, granulation and base excess’, Plant and Soil. Plant Soil wastewaters through struvite crystallization - A review. Arabian J. Chem. King Saud 410 (1–2), 139–152. https://doi.org/10.1007/s11104-016-2990-2. University 7 (1), 139–155. https://doi.org/10.1016/j.arabjc.2013.10.007. Despommier, D., 2013. Farming up the city: the rise of urban vertical farms. Trends Rech, I., et al., 2018. Solubility, diffusion and crop uptake of phosphorus in three Biotechnol. 31 (7), 388–389. https://doi.org/10.1016/j.tibtech.2013.03.008. different struvites. Sustainability (Switzerland) 11 (1). https://doi.org/10.3390/ EI-Kazzaz, A., 2017. Soilless Agriculture a New and Advanced Method for Agriculture su11010134. Development: an Introduction. Agricultural Res. Technol. Open Access J. https:// Ribalta-Pizarro, C., Muñoz, P., Munné-Bosch, S., 2021. Tissue-specific hormonal doi.org/10.19080/artoaj.2017.03.555610. variations in grapes of irrigated and non-irrigated grapevines (Vitis vinifera cv. Erel, R. et al. (2019) ‘Root development of bell pepper (Capsicum annuum L.) as affected “Merlot”) growing under mediterranean field conditions. Front. Plant Sci. https:// by water salinity and sink strength’. doi: 10.3390/plants9010035. doi.org/10.3389/fpls.2021.621587. Garrido, M., et al., 2020. The adjustment of Prosopis tamarugo hydraulic architecture Rufí-Salís, M., Brunnhofer, N., et al., 2020. Can wastewater feed cities? Determining the traits has a homeostatic effect over its performance under descent of phreatic level in feasibility and environmental burdens of struvite recovery and reuse for urban the Atacama Desert. Trees - Structure and Function. https://doi.org/10.1007/ regions. Sci. Total Environ. Elsevier B.V. 737, 139783 https://doi.org/10.1016/j. s00468-019-01899-2. scitotenv.2020.139783. Graber, A., Junge, R., 2009. Aquaponic Systems: nutrient recycling from fish wastewater Rufí-Salís, M., Calvo, M.J., et al., 2020. Exploring nutrient recovery from hydroponics in by vegetable production. Desalination. Elsevier B.V 246 (1–3), 147–156. https://doi. urban agriculture: an environmental assessment’, Resources, Conservation and org/10.1016/j.desal.2008.03.048. Recycling. Elsevier 155, 104683. https://doi.org/10.1016/j.resconrec.2020.104683 Harder, R., et al., 2019. Recycling nutrients contained in human excreta to agriculture: (November 2019). pathways, processes, and products. Crit. Rev. Environ. Sci. Technol. 49 (8), 695–743. Rufí-Salís, M., Petit-Boix, A., et al., 2020. Recirculating water and nutrients in urban https://doi.org/10.1080/10643389.2018.1558889. agriculture: an opportunity towards environmental sustainability and water use Hasler, K., et al., 2015. Life cycle assessment (LCA) of different fertilizer product types. efficiency? J. Clean. Prod. Elsevier 261, 121213. https://doi.org/10.1016/J. Eur. J. Agronomy. Elsevier B.V. 69, 41–51. https://doi.org/10.1016/j. JCLEPRO.2020.121213. eja.2015.06.001. Sambo, P., et al., 2019. Hydroponic solutions for soilless production systems: issues and Hochmuth, G., et al., 2018. Plant tissue analysis and interpretation for vegetable crops in opportunities in a smart agriculture perspective. Front. Plant Sci 10. https://doi.org/ Florida. Horticultural Sci. Department, UF/IFAS Extension, 2012(4/11/2012), p. 55. 10.3389/fpls.2019.00923. Huijbregts, M.A.J., et al., 2017. ReCiPe2016: a harmonised life cycle impact assessment Sanjuan-Delmás, D., et al., 2018. Environmental assessment of an integrated rooftop method at midpoint and endpoint level’, International Journal of Life Cycle greenhouse for food production in cities. J. Clean. Prod. 177, 326–337. https://doi. Assessment. Int. J. Life Cycle Assessment 22 (2), 138–147. https://doi.org/10.1007/ org/10.1016/j.jclepro.2017.12.147. s11367-016-1246-y. Silber, A., et al., 2005. High fertigation frequency and phosphorus level: effects on Ikeda, H., Tan, X., 1998. Urea as an organic nitrogen source for hydroponically grown summer-grown bell pepper growth and blossom-end rot incidence. Plant Soil 270 tomatoes in comparison with inorganic nitrogen sources. Soil Sci. Plant Nutr. 44 (4), (1), 135–146. https://doi.org/10.1007/s11104-004-1311-3. 609–615. https://doi.org/10.1080/00380768.1998.10414484. Simha, P., et al., 2017. Farmer attitudes and perceptions to the re–use of fertiliser ISO (2006) ISO 14044:2006 -Environmental management e life cycle assessment - products from resource–oriented sanitation systems – The case of Vellore. South requirements and guidelines. India’, Sci. Total Environ. Elsevier B.V. 581–582. https://doi.org/10.1016/j. Buchanan, J.R., J, R., Mote, C.R., C, R., Robinson, R.B., R, B, 1994. Thermodynamics of scitotenv.2017.01.044, 885–896. Struvite Formation. Trans. ASAE. Am. Society of Agricultural and Biol. Eng. 37 (2), Simha, P., Ganesapillai, M., 2017. Ecological sanitation and nutrient recovery from 617–621. https://doi.org/10.13031/2013.28121. human urine: how far have we come? A review. Sustainable Environ. Res. Elsevier Kirchmann, H. et al. (2017) ‘From agricultural use of sewage sludge to nutrient Ltd 27 (3), 107–116. https://doi.org/10.1016/j.serj.2016.12.001. extraction : a soil science outlook’, pp. 143–154. doi: 10.1007/s13280-016-0816-3. Specht, K., et al., 2014. Urban agriculture of the future: an overview of sustainability Kamilova, F., et al., 2006. Organic acids, sugars, and L-tryptophane in exudates of aspects of food production in and on buildings. Agric. Human Values 31 (1), 33–51. vegetables growing on stonewool and their effects on activities of rhizosphere https://doi.org/10.1007/s10460-013-9448-4. bacteria. Molecular Plant-Microbe Interactions 19 (3), 250–256. https://doi.org/ Stratful, I., Scrimshaw, M.D., Lester, J.N., 2004. Removal of struvite to prevent problems 10.1094/MPMI-19-0250. associated with its accumulation in wastewater treatment works. Water Environ. Khademi, Z., et al., 2010. Organic acids differ in enhancing phosphorus uptake by Res. 76 (5), 437–443. https://doi.org/10.2175/106143004X151491. Triticum aestivum L.-effects of rhizosphere concentration and counterion. Plant and Talboys, P.J., et al., 2016. Struvite: a slow-release fertiliser for sustainable phosphorus Soil 334 (1), 151–159. https://doi.org/10.1007/s11104-009-0215-7. management? Plant Soil 401 (1–2), 109–123. https://doi.org/10.1007/s11104-015- Kratz, S., Vogel, C., Adam, C., 2019. Agronomic Performance of P recycling Fertilizers 2747-3. and Methods to Predict it: a Review. Nutrient Cycling in Agroecosystems. Springer Toboso-Chavero, S., et al., 2019. Towards productive cities: environmental assessment of Netherlands. https://doi.org/10.1007/s10705-019-10010-7. the food-energy-water nexus of the urban roof mosaic. J. Ind. Ecol. 23 (4), 767–780. Latifian, M., Liu, J., Mattiassona, B., 2012. Struvite-based fertilizer and its physical and https://doi.org/10.1111/jiec.12829. chemical properties. Environ. Technol. (United Kingdom). https://doi.org/10.1080/ Turner, L.B., 1985. Changes in the phosphorus content of capsicum annuum leaves 09593330.2012.676073. during water stress. J. Plant Physiol. https://doi.org/10.1016/S0176-1617(85) Li, B., et al., 2019. Phosphorous recovery through struvite crystallization: challenges for 80079-1. future design. Sci. Total Environ. Elsevier B.V. 648, 1244–1256. https://doi.org/ Vatsanidou, A., et al., 2020. Life cycle assessment of variable rate fertilizer application in 10.1016/j.scitotenv.2018.07.166. a pear orchard. Sustainability (Switzerland) 12, 17. https://doi.org/10.3390/ Liu, Y., et al., 2008. Global phosphorus flows and environmental impacts from a SU12176893. consumption perspective. J. Ind. Ecol. 12 (2), 229–247. https://doi.org/10.1111/ Villalba, G., et al., 2008. Global phosphorus flows in the industrial economy from a j.1530-9290.2008.00025.x. production perspective. J. Ind. Ecol. 12 (4), 557–569. https://doi.org/10.1111/ Lunt, O.R., Kofranek, A.M., Clark, S.B., 1964. Nutrient availability in soil: availability of j.1530-9290.2008.00050.x. minerals from magnesium ammonium phosphates. J. Agric. Food Chem. https://doi. Vinci, G., Rapa, M., 2019. Hydroponic cultivation: life cycle assessment of substrate org/10.1021/jf60136a005. choice. British Food J. 121 (8), 1801–1812. https://doi.org/10.1108/BFJ-02-2019- Massey, M.S., et al., 2009. Effectiveness of recovered magnesium phosphates as fertilizers 0112. in neutral and slightly alkaline soils. Agron. J. https://doi.org/10.2134/ Volpin, F., et al., 2018. Simultaneous phosphorous and nitrogen recovery from source- agronj2008.0144. separated urine: a novel application for fertiliser drawn forward osmosis. Montero, J.I. et al. (2011) ‘Environmental and economic profile of present greenhouse Chemosphere. Elsevier Ltd, 203 482–489. https://doi.org/10.1016/j. production systems in Europe’, (January). chemosphere.2018.03.193. Panizza, M., Cerisola, G., 2001. Removal of organic pollutants from industrial Walters, S., Stoelzle Midden, K., 2018. Sustainability of urban agriculture: vegetable wastewater by electrogenerated Fenton’s reagent. Water Res. https://doi.org/ production on green roofs. Agriculture 8 (11), 168. https://doi.org/10.3390/ 10.1016/S0043-1354(01)00135-X. agriculture8110168. Parada, F., et al., 2021. Optimizing irrigation in urban agriculture for tomato crops in Xu, G., et al., 2004. Integrated effect of irrigation frequency and phosphorus level on rooftop greenhouses. Sci. Total Environ. 794. https://doi.org/10.1016/j. lettuce: p uptake, root growth and yield. Plant Soil 263 (1), 297–309. https://doi. scitotenv.2021.148689. org/10.1023/B:PLSO.0000047743.19391.42. Pereira-Dias, L., et al., 2018. Different root morphological responses to phosphorus Yang, L., et al., 2015. Application of hydroponic systems for the treatment of source- supplies in grafted pepper. Bulletin of University of Agricultural Sciences and separated human urine. Ecol. Eng. Elsevier B.V. 81, 182–191. https://doi.org/ Veterinary Medicine Cluj-Napoca. Horticulture 75 (1), 59. https://doi.org/ 10.1016/j.ecoleng.2015.04.013. 10.15835/buasmvcn-hort:001217. Zelia, S.A., et al., 2017. Symptoms of Macronutrients deficiency in sweet pepper Pradel, M., et al., 2020. Environmental performances of production and land application (Capsicum annuum L.). Agrociencia Uruguay. of sludge-based phosphate fertilizers—A life cycle assessment case study’, Dakora, F. D. and Phillips, D. A. (2002) ‘Root exudates as mediators of mineral Environmental Science and Pollution Research. Environ. Sci. Pollut. Res. 27 (2), acquisition in low-nutrient environments Author ( s ): Felix D . Dakora and Donald A 2054–2070. https://doi.org/10.1007/s11356-019-06910-4. . Phillips Source : Plant and Soil , FOOD SECURITY IN NUTRIENT-STRESSED Rahaman, M.S., et al., 2006. Exploring the determination of struvite solubility product ENVIRONMENTS : EXPLOITING PLANTS ’ GENE’, 245(1), pp. 201–213. from analytical results. Environ. Technol. 27 (9), 951–961. https://doi.org/10.1080/ 09593332708618707. 13