This is the submitted version of the article: Arcas Pilz, Verónica; Rufí Salís, Martí; Parada, Felipe; [et al.]. «Recovered phosphorus for a more resilient urban agriculture : assessment of the fertilizer potential of struvite in hydroponics». Science of the total environment, Vol. 799 (Dec. 2021), art. 140424. DOI 10.1016/j.scitotenv.2021.149424 This version is available at https://ddd.uab.cat/record/248799 under the terms of the license 1 Recovered phosphorus for a more resilient urban agriculture: 2 assessment of the fertilizer potential of struvite in hydroponics 3 4 Verónica Arcas-Pilza,1, Martí Rufí-Salísa,b,1, Felipe Paradaa, Anna Petit-Boixc, 5 Gara Villalbaa,b, Xavier Gabarrella,b,* 6 7 a Sostenipra Research Group (SGR 01412), Institute of Environmental Sciences and Technology 8 (MDM-2015-0552), Z Building, Universitat Autònoma de Barcelona (UAB), Campus UAB, 08193 9 Bellaterra, Barcelona, Spain 10 b Department of Chemical, Biological and Environmental Engineering, Universitat Autònoma de 11 Barcelona (UAB), Campus UAB, 08193 Bellaterra, Barcelona, Spain 12 c Chair of Societal Transition and Circular Economy, University of Freiburg. Tennenbacher Str. 13 4, 79106 Freiburg i. Br. (Germany) 14 1 These authors contributed equally to this research 15 Corresponding author: xavier.gabarrell@uab.cat 16 Abstract 17 Urban agriculture (UA) is a means for cities to become more resilient in terms of food 18 sovereignty while shortening the distance between production and consumption. 19 However, UA still intensively depends on the use of fertilizers, which relies on 20 depleting non-renewable resources such as (P) and causes both local and global 21 impact for its production and application. With the aim to reduce such impacts and 22 encourage a more efficient use of nutrients, this study assesses the feasibility of using 23 struvite precipitated from an urban wastewater treatment plant as the unique source 24 of P fertilizer. To do so, we apply various quantities of struvite (ranging from 1 to 20 25 g/plant) to the substrate of a hydroponic Phaseolus vulgaris crop and determine the 26 yield, water flows and P balances. The results show that treatments with more than 27 5g of struvite per plant produced a higher yield (maximum of 181.41 g/plant) than 28 the control (134.6 g/plant) with mineral fertilizer (KPO4H2). On the other hand, P 29 concentration in all plant organs was always lower when using struvite than when 30 using chemical fertilizer. Finally, the fact that different amounts of struvite remained 31 undissolved in all treatments denotes the importance to balance between a correct 32 P supply to the plant and a decrease of P lost through the leachates, based on the 33 amount of struvite and the irrigated water. The findings of this study show that it is 34 feasible for UA to efficiently use locally recovered nutrients such as P to produce local 35 food. 36 Graphical Abstract: 37 38 1. Introduction 39 Meeting the food demand of the ever-growing urban population is a global 40 challenge. Since food provision to cities is highly dependent on long and complex 41 supply chains, the distance between production and consumption points has 42 extensively increased. This prevents nutrient recycling, while emitting huge amounts 43 of greenhouse gases due to long-distance transport (Rees and Wackernagel, 1996; 44 Thomaier et al., 2015). In this sense, moving towards more robust and resilient food 45 systems should be a priority in the following years (European Commission, 2020). To 46 do so, alternatives that narrow the distance between production and consumption 47 points have already been reported, being urban agriculture one of the most 48 prominent (Deelstra, 1987). However, this the yield (Astee and Kishnani, 2010; Rufí- 49 Salís et al., 2020b). On the other hand, the use of local fertilizers is still very limited, 50 and often reduced to the use of compost (Thomaier et al., 2015). 51 The case of phosphorus (P) fertilizers is of great relevance, since P is primarily 52 obtained from non-renewable phosphate rocks. Moreover, previous studies quantify 53 that 80% of the available stock of phosphate rocks is being used in the production 54 of fertilizers (Shu et al., 2006). Since half of the world’s current economic phosphate 55 resources will have been used up by the end of the 21st century (Steen, 1998) the 56 European Union recognizes P as a critical resource (European Comission, 2014). 57 Among its recommendations, a planned amendment of the fertilizer regulation 58 encourages P recovery from local sources by enforcing a shift towards a more circular 59 use of nutrients (European Comission, 2016). 60 In this sense, urban wastewater treatment plants (WWTPs) are well-known sources of 61 secondary P. WWTPs have already been addressed as a potential alternative to 62 importing mineral fertilizers (e.g. de-Bashan and Bashan, 2004; Kern et al., 2008; Shu 63 et al., 2006). Struvite, also known as magnesium ammonium phosphate (MAP with 64 the formula NH4MgPO4·6H2O) is a crystalline precipitate that has been gaining 65 popularity as a way to recover P from wastewater. To induce its precipitation a molar 66 ratio of 1:1:1 for magnesium (Mg2+), ammonium (NH4+) and phosphate (PO43-) is 67 needed, under specific pH conditions (8.5-9.5) (Bouropoulos and Koutsoukos, 2000; 68 J. R. Buchanan et al., 1994; Le Corre et al., 2009). Originally the precipitation of struvite 69 was associated to a major concern in WWTP being the cause of equipment damaging 70 causing labor and infrastructure costs (Borgerding, 1972; Doyle et al., 2003; Stratful 71 et al., 2004). Struvite forced precipitation has gained attraction since the 90’s, not 72 only to avoid infrastructure damage but also as a P recovery technique (Doyle et al., 73 2003). This process has been studied and improved in the past years making it a 74 more efficient precipitation process (Le Corre et al., 2009; Li et al., 2019; Sena and 75 Hicks, 2018). 76 In terms of application, the properties of struvite as an effective source of nutrients 77 (P-PO43-, N-NH4+ and Mg-Mg2+) for plants (Li and Zhao, 2003) and its low solubility 78 in water (0.018g·100ml-1 at 25ºC) (Bridger et al., 1961) make it a slow-releasing 79 valuable fertilizer that can reduce economic costs in agriculture (Rahman et al., 2014). 80 However, only limited literature has explored the application of struvite in agricultural 81 facilities. For example, Antonini et al. (2012), Uysal et al. (2014), Gell et al. (2011) and 82 Liu et al. (2011) assessed the maize performance of struvites with different 83 characteristics and origins in different soils. In a review made by Li et al. (2019) we 84 can see that almost all struvite trials found that vegetables grown with struvite had 85 the same -or even improved- performance compared to controls with conventional 86 fertilizers. 87 Creating a closed-loop, waste-to-resource system such as that of struvite recovery 88 within the city limits and not applying it at this scale seems contradictory within the 89 concept of urban metabolism. In this sense, the synergy between struvite precipitation 90 in urban WWTPs and urban agriculture seems worth exploring considering the 91 potential of the latter to blurry the lines between waste and resource within urban 92 areas (Ferreira et al., 2018; Rufí-Salís et al., 2020a; Smit and Nasr, 1992). This article 93 aims to assess the potential of struvite precipitated in a WWTP as a fertilizer within 94 the framework of urban metabolism. Based on experimental and analytical results 95 performed on a Phaseolus vulgaris crop grown in a hydroponic rooftop greenhouse, 96 we determine the implications of fertilization with struvite in terms of yield, water 97 flows and P balances and provide recommendations to further improve the 98 performance of this waste-to-resource fertilizer. 99 100 2. Methodology 101 This section describes the materials and methods used in our analysis. We first present 102 the system under study (section 2.1) along with the fertilization and experimental set- 103 up (section 2.2). The experimental and analytical assessment of the P balances is 104 defined in section 2.3, whereas sections 2.4 and 2.5 present the validation test set-up 105 and its most relevant results, respectively. Finally, section 2.6 presents the 106 configuration of the determination test, which will finally provide the results for this 107 study. 108 109 2.1 Characterization of the system 110 The present study was conducted in a rooftop greenhouse on the ICTA-ICP building, 111 located in the campus of the Universitat Autònoma de Barcelona, 15km away from 112 Barcelona. The building is equipped with a 900m2 rainwater harvesting system that 113 stores water in a 100m3 tank. Most of this rainwater is used in the rooftop greenhouse 114 (122.8m2) to irrigate crops with a hydroponic system, i.e. mixing water with nutrients 115 before providing the solution through a dripping system (2 L/h) to the perlite 116 substrate bags (40L capacity). The perlite substrate has a pH of 7, an electrical 117 conductivity of 0.09 dS·m-1, a granulometry of [0-6] mm and 4 plants can be planted 118 in each bag. 119 2.2 Fertilization and experimental set-up 120 Struvite granules were obtained from Aarhusvand A/S company from Aarhus, 121 Denmark. This company distributes fertiliser grade struvite under the name 122 PhosphorCareTM, recovered using the PhosphogreenTM technology. This technology is 123 based on a fluidized bed reactor that creates the specific conditions to precipitate 124 struvite through the addition of magnesium chloride, sodium hydroxide and air. The 125 final struvite granules have a size range of 0.5-1.5 mm. 126 Common bean plant (Phaseolus vulgaris var. Pongo) was chosen as the crop for this 127 study, planting nursery plants (approximately 10-14 days old). To apply the struvite 128 to the plants, we considered different possibilities. Mixing it with the nutrient solution 129 was discarded because the system could not benefit from the slow-release 130 characteristics of struvite. Thus, we choose to directly apply the granules to the plant 131 roots. Considering this option, we designed a system that consisted on mixing perlite 132 with struvite inside a low-density polyethylene perforated bag with holes of no more 133 than 1 mm diameter (Figure SM1 in the Supplementary Materials). At the same time, 134 this system allows the interaction between struvite granules and roots and avoids the 135 depletion of undissolved struvite into the leachates. 136 Two different experiments were carried out: the validation test and the determination 137 test, both of them using double growing lines with 8 substrate bags each (Figures 138 SM2 and SM15 in the Supplementary Materials). For control treatments, the nutrient 139 solution applied to the crops in milligrams per litre was KPO4H2 – 136, KNO3 – 101, 140 K2SO4 – 217.5, Ca(NO3)2 – 164, CaCl2 · H2O – 111, Mg(NO3)2 – 148.3, Hortilon – 10, 141 and Sequestrene – 10. In treatments with struvite, the mineral P source, KPO4H2 in 142 this case, was excluded from the initial nutrient solution. All other mineral fertilizers 143 were maintained. 144 2.3 Phosphorus balances 145 To account for the P balances, Equation 1 was calculated on a plant basis for every 146 control and struvite treatment. Figure SM1 in the Supplementary Material shows a 147 diagram of the perforated bag with the elements displayed in Equation 1. 148 𝑃𝑁𝑆 + 𝑃𝑆𝐼 = 𝑃𝐿𝑉 + 𝑃𝑆𝑇 + 𝑃𝐵𝑁 + 𝑃𝑆𝐹 + 𝑃𝐿𝐼𝑋 + 𝑃𝐴𝐶 (Equation 1) 149 150 In Equation 1, P represents mass of phosphorus. PNS is the amount of mineral P 151 supplied through the irrigation system during all the crop cycle. PSI is the amount of 152 P in the form of struvite applied at the beginning of the test. PLIX is the amount of P 153 in the leachates during all the crop cycle. PLV, PST, and PBN, represent P uptake by 154 leaves, stem and beans, respectively. PSF is the amount of remaining undissolved P in 155 the form of struvite at the end of the test, plus the P adsorbed in the perlite granules. 156 Finally, PAC is the amount of dissolved P accumulated in the water retained in the 157 substrate at the end of the crop. Three different biomass and substrate sampling 158 dates were used in every test: 26, 54 and 78 days after planting (DAP) for the 159 validation test and 23, 51 and 72 DAP for the determination test. 160 The initial nutrient concentration of the substrate was verified to be negligible at the 161 beginning of the experiment through atomic spectroscopy and elemental analysis. 162 Samples of the fertilizer solution were collected directly from the drippers placed in 163 the perlite bags. Leachate samples were taken from plastic drainage buckets placed 164 on one side of each line. To determine the PNS and PLIX, the respective samples were 165 collected three times per week and externally analyzed using ICP-OES atomic 166 spectroscopy (Optima 4300DV by Perkin-Elmer). PSI was quantified summing the 167 amount of perlite in a specific bag with the amount of struvite that was applied, 168 considering weights obtained by drying two struvite samples and two perlite samples 169 at 105°C in a furnace until reaching constant weight (reached after 3 days). PSF was 170 quantified differently in each test. In the validation test, all 4 samples for a specific 171 treatment were homogenized after extracting the roots, using distilled water to 172 separate the struvite granules from the roots. After this process, two random samples 173 were dried at 105°C in a furnace until reaching constant weight and externally 174 analyzed using ICP-OES atomic spectroscopy. On the other hand, in the determination 175 test, roots were shredded, homogenized and integrated within every individual 176 substrate sample. Then, a fraction of these samples was dried and analyzed using the 177 same method as in the validation test. 178 PLV, and PST were determined based on the nutrient content of every plant separately. 179 Leaves and stem were separated, sorted into paper envelopes and dried in a furnace 180 at 65°C until reaching constant weight (reached after 7 days) before analyzing 181 externally the concentration of P through ICP-OES atomic spectroscopy. The same 182 methodology was applied to determine the PBN, with randomly chosen 500-gram 183 bean samples being processed for every treatment. The P analytical results obtained 184 for the beans were multiplied by the production obtained in every treatment to 185 comply with the balances in a plant basis. 186 2.4 Validation test set-up and justification 187 From September 13th until December 3rd, 2018, 10 double growing lines were used 188 (totalling 320 plants), distributing the treatments as showed in Figure SM2 of the 189 Supplementary Material. The aim of this experiment was to validate and keep track 190 of different parameters of the system. First, to check that the small, perforated bag 191 did not have negative consequences to crop development. To do so, we split the 192 control lines into two different treatments, VCB and VC0, using standard nutrient 193 solution with and without the bags, respectively. Secondly, to check the correct 194 development of bean plants with struvite in a hydroponic system, we applied different 195 struvite amounts per plant: 5, 10, 15, 20 and 25g corresponding to the treatments 196 tagged as V5, V10, V15, V20 and V25, respectively. Additionally, a treatment with no 197 struvite was tagged as V0. These amounts of struvite were based on previous 198 experiments done with the same crop species and variety in hydroponic cultivation 199 that accounted for P uptake (Rufí-Salís et al., 2020c). One week after the first harvest, 200 KPO4H2 was added in the nutrient solution of struvite treatments until the end of the 201 harvest to ensure a good nutrition to the plants during the production period, which 202 is highly demanding in P (e.g. Bender et al. 2015; Kouki et al. 2016; da Silva et al. 203 2019). 204 2.5 Validation test results 205 2.5.1 Production and phenological stages 206 The production results for the control treatments VCB and VC0 showed that the 207 perforated bag did not have any effect on the correct crop development and yield 208 (Figure 1 and Figure SM5), as the yields from the different lines do not differ between 209 them (VC0_2 187.54±69.35; VCB_1 186.15±84.01 g/plant). Even though treatment 210 VC0_1 generated more yield (224.84±91.84 g/plant), it could be attributed to the fact 211 that it was an exterior cropping line facing the border and thus received more 212 radiation. Similarly, VCB_2 also produced more yield (195.45±88.63 g/plant) than its 213 replicate (VCB_1) although no significant differences where determined by the end of 214 the experiment. 215 216 Figure 1: Production (g/plant) of the control treatments in the validation test, with (VCB) and 217 without (VC0) perforated bags for each harvest. Same letters (a,b) indicate no significant 218 difference (p>0.05) between treatment for each harvest time. Sample size for harvests 1, 2, 3, 4 219 and 5 (35- 54 DAP) correspons to n=28 plants, for harvests 6, 7 and 8 (63-78 DAP) n=24 plants 220 per treatment. 221 On the other hand, treatments with struvite (Figure SM3 and SM4 of the 222 Supplementary Material) exerted a similar yield than the control treatments at the 223 end of the crop. The treatment with the highest quantity of struvite (V25) had the 224 highest production median (203.85 g/plant), while the treatment with the lowest 225 quantity of struvite (V5) had the highest mean (216.15±93.54 g/plant). On the other 226 hand, the treatment without struvite produced a really low yield (7.19±4.49 g/plant). 227 The similarities in terms of yield between all struvite treatments at the end of the 228 cycle may be related to the additional mineral fertilization during the production 229 phase. Moreover, we can see that struvite treatments produced more than the control 230 in the first 3 harvests (35, 39 and 42 DAP) (Figure SM6 in the Supplementary Material). 231 This effect is similarly observed for the phenological stages (Figures SM7 to SM11 in 232 the Supplementary Material). For the parameters that were quantified in different 233 dates (number of leaves (Fig. SM7), side shoots (Fig. SM8). Open flowers (Fig. SM9) 234 and floral buttons (Fig. SM10)), we can see that the treatments with struvite not only 235 had a correct early stage development, but also develop plant organs earlier than in 236 control treatments. 237 238 2.5.2 Water 239 We applied more water in struvite treatments (125 L/plant) than to the control (94.76 240 L/plant) to ensure a proper dissolution of this fertilizer (Fig. SM12 – Supplementary 241 Material). However, we can see in Figure SM13 in the Supplementary Material that if 242 no control is taken over the water that is being irrigated, leachates emitted by the 243 struvite treatments with higher concentrations (28.9 mg/L – V25) of this fertilizer tend 244 to be similar to those of the control treatments. Obviously, this behaviour can only 245 be observed before the irrigation with mineral P added during the harvesting process. 246 Parallelly, we can see that the perforated bag mechanism did not affect the P 247 concentration in the leachates between the control treatment C0 and CB. 248 249 2.5.3 Phosphorus content 250 Figure SM14 of the Supplementary Material shows the P content in the different plant 251 organs as well as the content in the substrate, described as “undissolved”. P content 252 in the stem show low variability along all treatments, with V25 having the highest 253 (0.083±0.020g P) and V0 the lowest (0.008±0.002g P) at the end of the crop cycle. A 254 great P accumulation was observed in the low production of the V0 treatment with 255 a content of 0.107±0.005g P (54 DAP) in beans, which was even higher than the 256 highest observed in the control for VC0 (0.094±0.013g P –54 DAP), although the 257 greater content was found for treatment V25 with 0.172±0.023g P (54 DAP). The V0 258 treatment doesn’t show P results in leaves for 54 and 78 DAP because no leaves 259 remained in the plant at the sampling time. For this same reason, there is a lack of 260 data in beans for 78 DAP. Finally, concentration in beans for struvite treatments was 261 similar to the one observed in the control. For all plant organs, a pattern in the 262 accumulation of P in the plant tissue can be observed. In the first sampling all 263 treatments show a rather low accumulation with greater content for plants with 264 greater struvite quantities, in some cases also for the control treatments. For the 265 second sampling, a bigger content difference can be seen with an acute increase of 266 the V25 P content, especially for the stems and leaves. Finally, at 78 DAT, these 267 differences between treatments even out and only treatment V0 remains significantly 268 reduced. This last part however, does not correspond to the undissolved P in the 269 perlite, where the P content in the substrate directly responds to the amount of 270 struvite given, being always higher for the V25. The control treatments receive the P 271 through irrigation making the existing content in the substrate comparably small. 272 2.6 Determination test set-up 273 From September 16th until November 27th, 2019, 8 double growing lines were used 274 (totalling 256 plants), distributing the treatments as showed in Figure SM15 of the 275 Supplementary Material. The determination test was designed based on the results 276 of the validation test. The treatment distribution was randomized throughout the 277 Greenhouse avoiding the influence of climatic conditions. Thus, the struvite 278 treatments were recalculated, applying per plant: 1, 2.5, 5, 7.5, 10, 15 and 20g 279 corresponding to the treatments tagged as S1, S2.5, S5, S7.5, S10, S15 and S20, 280 respectively. Struvite amounts below 5g were applied based on the yield and P 281 content performance in the validation test for V5. Since we found that the perforated 282 bag did not affect plant development, we only used one control treatment, tagged 283 as CB, which used the same perforated bag as the struvite treatments. Moreover, 284 considering the yield and phenological findings in the validation test, we decided not 285 to apply mineral P fertilizer to the struvite treatments at any point, being struvite the 286 only source of P to the plants. 287 2.7 Statistical analysis 288 The analysed data was tested for normality using the Shapiro-Wilk test p>0.05. 289 Further on, the Levene’s test p>0.05 was used to determine homogeneity of variance. 290 Once these parameters were validated the Duncan’s multiple range test was used to 291 assess the statistical significance of treatments. On the other hand, non-parametric 292 data were analysed for significance using the Kruskal-Wallis test. The significance 293 between the treatments was marked with different letters in each plot. All statistical 294 analyses were made with the R studio software. 295 296 3. Results 297 This section presents the results of the determination test. Section 3.2 shows the 298 production of the control and struvite treatments. Section 3.2 presents the results in 299 terms of amount and concentration of the water flows. Section 3.3 displays the 300 findings related to the P amount in the substrate and the undissolved struvite. Finally, 301 Section 3.4 zooms in and shows the P concentration in the different plant organs. 302 303 3.1 Yield 304 Figure 2 (and SM16 in the Supplementary Material for the final total yield) shows the 305 results of the accumulated yield per number of harvests, being the sixth harvest (71 306 DAP) the final one before uprooting the plants. Only treatments S1 (78.9 g/plant) and 307 S2.5 (128.1 g/plant) had lower yields than the control treatment (134.6 g/plant), being 308 the first significantly lower. On the other hand, all other treatments with 5g of struvite 309 or above produced more than the control treatment, demonstrating the potential of 310 struvite to produce similar or even higher yields than with mineral fertilizer, as 311 reported by Li et al. (2019). 312 313 Figure 2: Comparison of accumulated production of fresh bean per plant (g/plant) per treatment 314 for each harvest time. Same letters (a, b, c, d) indicate no significant difference (p>0.05) between 315 treatment for each harvest time. Sample size for all harvests is n=24 plants per treatment. 316 317 As we can see in Figure 2, it was not until the second harvest (42 DAP) that great 318 differences were observed between the S1 yield and the other struvite treatments, 319 while a decrease in S2.5 yield was observed between the 4th and 5th harvest, 57 and 320 64 DAP, respectively. Regarding the control treatment, the first harvest produced 321 lower yield (6.31±5.71 g/plant) than even the S1 struvite treatment (9.98±8.51 322 g/plant). 323 324 Figure 3: Distribution of accumulated production per plant per harvest (DAP = days after plantin) 325 of different treatments. Sample size for all harvests is n=24 plants per treatment. 326 This fact reinforces the idea that the application of struvite could be beneficial for 327 early stage plant development, as the validation test showed better behaviour in 328 struvite than in control in phenological variables. This fact could be related to the 329 NH4+ supply by struvite, which could benefit the plant root balance when combined 330 with nitrate supply (Marschner, 1995). The fact that previous literature suggests that 331 NH4+ supply to common bean could be harmful for plant development (Chaillou et 332 al., 1986; Guo et al., 2007) could be related to the amount of NH4+ supplied. Because 333 struvite does not only enable a slow release of P but also of NH4+, reaching NH4+ 334 accumulation to harmful levels seems improbable. 335 In terms of distribution, yields show an asymptote behaviour among treatments, 336 where S20 produces the highest yield (g/plant) (181.41±66.16) and S1, the lowest 337 (78.94±34.23). Figure SM16 in the Supplementary Material shows how treatment S10 338 was detected as the exception for this tendency in terms of mean production 339 (150.50±56.10), probably related to bias parameters like shapes in the greenhouse or 340 a non-homogenic distribution of struvite in the perlite bag. However, boxplots 341 represented in Figure 3 shows how the median of the final amount of yield harvested 342 for S10 (155.70) follows the tendency, while not presenting outliers in the distribution. 343 3.2 Water 344 Figure SM17 of the Supplementary Materials shows that the irrigated water in the 345 control and the struvite treatments was the same (42.5 litres per plant), while Figure 346 4 shows the accumulated P during the entire cycle in the different water streams. The 347 quantity of P present in the control streams is much bigger than the one in the 348 struvite streams, with the former irrigating and leaching 2.07 and 1.41 g of P per 349 plant for the entire crop cycle, respectively. The fact that the P leachates are one 350 order of magnitude smaller when using struvite (maximum of 0.03 g of P per plant 351 in S20) could be related to the slow-release characteristic of struvite reported in the 352 literature. A clear benefit of this finding is a decrease in both P depletion and 353 freshwater eutrophication related to the leachates flow. Moreover, if the leachates of 354 struvite treatments do not contain a large amount of P, it means that most of the 355 struvite has been whether taken up by the plant or remains undissolved in the 356 substrate. 357 When comparing Figure SM13 and Figure SM18 of the Supplementary Material, we 358 can see that P release by struvite is highly dependent on the input water flow, 359 represented in Figures SM12 and SM17 for the validation and determination test, 360 respectively. Because the irrigated water was three times less in the determination 361 test (125.2 against 42.5 litres per plant, respectively), the P observed in the leachates 362 is less than in the validation test, considering the period where P was not supplied 363 through mineral fertilizer in the validation test. 364 Differences are observed within the struvite treatments in Figure 4, highly dependent 365 on the quantity of struvite that was applied at the beginning of the crop. Treatments 366 S1 and S2.5 stopped emitting P in the leachates just 14 DAP, which could have 367 triggered P deficiencies. On the other hand, treatments S15 and S20 were the only 368 struvite treatments that did not stop emitting P to the leachates flow. Figure 4: Distribution of accumulated phosphorus in the irrigation and leachates of different treatments. Rcon: P in the control irrigation stream. 369 3.3 Substrate and undissolved struvite 370 Figure 5 shows the distribution of P among all possible input and outputs considered 371 in the system. At the end of the crop cycle, the control treatment supplied more P 372 (2.07 g of P per plant) than the treatment with the highest amount of struvite (S20 - 373 1.90 g of P per plant). Most of the P supplied in the control treatments is discharged 374 (68%), while in the struvite treatments it still remains in the substrate. 375 376 Figure 5: P distribution among all water, biomass and substrate flows and compartments. This 377 amount of struvite at the end of the crop could be recovered, or the same substrate with struvite 378 could be used for a successive crop. 379 3.4 Biomass 380 In terms of biomass, we can see that the concentration in percentage (Figure SM19 381 – Supplementary Material) in all organs increases with the quantity of struvite applied 382 to the treatment, having S15 and S20 similar concentrations in the leaves (0.70±0.13 383 and 0.67±0.18, respectively) and stem (0.50±0.09 and 0.44±0.12, respectively). 384 However, the control treatment with mineral fertilizer presented higher concentrations 385 of P than all struvite treatments, also in beans (0.73±0.04). This is especially relevant 386 in the case of beans, where the P deficiency in this organ directly affects the 387 nutritional value of the product that is going to reach the market. 388 4. Discussion 389 Studying isolated parameters in agriculture only shows part of the big picture. In this 390 sense, this section will discuss the results and tendencies that were found regarding 391 the inputs and outputs for the control and struvite treatments and provide 392 recommendations to practitioners based on our findings. 393 Treatments S1 and S2.5 had lower yields than the control treatments, establishing a 394 clear relationship between the yield and possible P deficiencies in these treatments. 395 However, struvite remains undissolved in all treatments, even though the production 396 and the distribution of P among plant organs was different between treatments 397 (Figure 2). The fact that we have undissolved struvite even in treatments S1 and S2.5 398 shows that the limitation is not only related to the quantity of struvite available, but 399 also its dissolution (Figure 5 and Figure SM18 Supplementary Material). 400 Because the irrigated water was three times lower in the determination test, the P 401 observed in the leachates is lower than in the validation test, considering the period 402 where P was not supplied through mineral fertilizer. Moreover, there is a significant 403 amount of P accumulated in the substrate bag at the end of the treatment in the 404 control test. This stored P will be depleted if a successive crop is planted, since the 405 small nursery plants will not benefit from all of it due to the lower needs of a smaller 406 plant. With the addition of irrigation the accumulated nutrients in the perlite bag 407 would eventually be moved to the leachates. By applying struvite (and verified by the 408 small amount of P in the leachates in struvite treatments) this P is not stored and 409 thus, not lost. 410 Based on the findings of this study, a well-designed struvite crop cycle needs to take 411 into account two essential parameters. First, the quantity of struvite, considering that 412 the quantity that remains undissolved at the end of the crop can be used again for 413 a successive cycle. Second, the irrigation management, considering that if we modify 414 this variable to increase the dissolution of struvite granules, we would also be 415 increasing the P in the leachates. Moreover, since previous studies highlighted the 416 effect of the surface area of the granules on the solubility of slow-release fertilizers 417 (Chien and Menon, 1995; Gell et al., 2011; Li et al., 2019), the size used in our study 418 (0.5-1.5mm) seems adequate for the balance between P supply and P lost through 419 the leachates. Literature with higher sizes reported solubility problems that affected 420 early plant development (Talboys et al., 2016), while studies using lower sizes or 421 powder do not report these problems (Achat et al., 2014; Antonini et al., 2012; Bonvin 422 et al., 2015; Gell et al., 2011). Additionally, the use of nursery plants is preferable since 423 the struvite low dissolution has been reported to be a disadvantage when providing 424 P to feed the transition from seeds to nursery plants (Talboys et al., 2016). 425 Struvite supply per plant should always be above 5g for Phaseolus vulgaris, 426 considering that more quantity of struvite would release more P into the leachates, 427 but ensure that P is available for plants. On the other hand, we should also account 428 for the nutritional value of the beans, considering the ultimate function is to produce 429 yield. In this sense, P in the biomass was a variable where the control treatment had 430 a better performance than struvite treatments. Only S15 and S20 reach a similar P 431 amount to the control in all plant organs. For this reason, a quantity between 15 and 432 20g of struvite, a responsible irrigation management and growing successive crops 433 with the same substrate constitutes the best option to grow a well-designed struvite 434 bean crop cycle. 435 436 4. Conclusions 437 On the way towards resilient cities, the recovery of scarce resources that can be 438 utilised within the urban boundaries will play an important role, especially in the food 439 vector. This study assessed the performance of the potential application of struvite 440 recovered from WWTPs in hydroponic bean crops to diminish the need for external 441 resources in urban agriculture. Three main conclusions could be drawn from this 442 analysis. 443 First, applying struvite in hydroponics crops equals and even increases the yield 444 compared to mineral fertilizer while diminishing P losses in the leachates, contributing 445 to both less nutrient depletion and eutrophication potential. In this sense, a quantity 446 above 5g/plant of struvite was observed to be enough for correct bean plant 447 development and yield production. 448 Second, the input water flow was relevant in supplying enough P to the plants 449 through dissolution using struvite. On the other hand, a correct water irrigation 450 management is relevant to diminish P losses through overdissolution. Therefore, a 451 balance between these two potential problems should be one of the key parameters 452 when growing crops with struvite. 453 Third, a great quantity of struvite remains undissolved at the end of the crop in all 454 treatments. In this sense, planting a successive cycle or recovering the struvite of the 455 substrate could be alternatives to not losing valuable fertilizer. 456 Based on the findings presented in this paper, we believe that future research should 457 focus on three different aspects. First, the role of NH4+ supplied by struvite on plant 458 development during the first production phase. Second, the performance of crops if 459 successive cycles are grown using the same undissolved struvite in hydroponic 460 systems. Third and finally, the modelling of P release by struvite based on quantity 461 applied and input water flow. 462 463 Acknowledgements 464 The authors are grateful to Universitat Autònoma de Barcelona for awarding a 465 research scholarship to M. Rufí-Salís (PIF-UAB 2017), to the Spanish Ministry of 466 Economy, Industry and Competitiveness (Spain) for the grant awarded to V. Arcas- 467 Pilz (FPI-MINECO 2018), and to the National Commission for Scientific and 468 Technological Research (Chile) for the grant awarded to F. Parada (PFCHA-CONICYT 469 2018 – Folio 72180248). This work was supported by the Spanish Ministry of Economy, 470 Industry and Competitiveness (AEU/FEDER) [CTM2016-75772-C3-1-R] and the “María 471 de Maeztu” program for Units of Excellence in R&D [MDM-2015-0552]. A. Petit-Boix 472 thanks the German Federal Ministry of Education and Research for the financial 473 support of the research group “Circulus - Opportunities and challenges of transition 474 to a sustainable circular bio-economy”, grant number 031B0018. 475 The research leading to these results has received funding from the European 476 Research Council under the European Horizon2020 research and innovation program 477 under grant agreement No 862663 (FoodE) as well as, ERC grant agreement 478 n° 818002 URBAG, awarded to Gara Villalba. 479 480 481 Author contributions 482 All authors were responsible for the conception and design of the study. V. Arcas- 483 Pilz, M. Rufí-Salís, A. Petit-Boix, G. Villalba and X. Gabarrell conceived the original idea 484 for the study. M. Rufí-Salís, V. Arcas-Pilz and F. Parada set up, supervised and acquired 485 the data for the experimental tests. M. Rufí-Salís and V. Arcas-Pilz processed and 486 analysed the data. M. Rufí-Salís and V. Arcas-Pilz took the lead in writing the 487 manuscript. All authors critically revised the draft for important intellectual content. 488 All authors gave their final approval to the manuscript. 489 490 References 491 Achat, D.L., Sperandio, M., Daumer, M.-L., Santellani, A.-C., Prud’Homme, L., Akhtar, 492 M., Morel, C., 2014. Plant-availability of phosphorus recycled from pig manures 493 and dairy effluents as assessed by isotopic labeling techniques. Geoderma 232– 494 234, 24–33. https://doi.org/10.1016/j.geoderma.2014.04.028 495 Antonini, S., Arias, M.A., Eichert, T., Clemens, J., 2012. Greenhouse evaluation and 496 environmental impact assessment of different urine-derived struvite fertilizers 497 as phosphorus sources for plants. Chemosphere 89, 1202–1210. 498 https://doi.org/10.1016/j.chemosphere.2012.07.026 499 Astee, L.Y., Kishnani, N.T., 2010. Building Integrated Agriculture: Utilising Rooftops 500 for Sustainable Food Crop Cultivation in Singapore. J. Green Build. 5, 105–113. 501 https://doi.org/10.3992/jgb.5.2.105 502 Bender, R.R., Haegele, J.W., Below, F.E., 2015. Nutrient uptake, partitioning, and 503 remobilization in modern soybean varieties. Agron. J. 107, 563–573. 504 https://doi.org/10.2134/agronj14.0435 505 Bonvin, C., Etter, B., Udert, K.M., Frossard, E., Nanzer, S., Tamburini, F., Oberson, A., 506 2015. Plant uptake of phosphorus and nitrogen recycled from synthetic source- 507 separated urine. Ambio 44, 217–227. https://doi.org/10.1007/s13280-014-0616- 508 6 509 Borgerding, J., 1972. Phosphate deposits in digestion systems. J. Water Pollut. 510 Control Fed. 44, 813–819. 511 Bouropoulos, N.C., Koutsoukos, P.G., 2000. Spontaneous precipitation of struvite 512 from aqueous solutions. J. Cryst. Growth 213, 381–388. 513 https://doi.org/10.1016/S0022-0248(00)00351-1 514 Bridger, G.L., Salutsky, M.L., Starostka, R.W., 1961. Metal Ammonium Phosphates as 515 Fertilizers. 140th Meet. Am. Chem. Soc. 1–19. 516 Chaillou, S., Morot-Guadry, J., Lesaint, C., Salsac, L., Jolivet, E., 1986. Nitrate or 517 ammonium nutrition in french bean. Plant Soil 91, 363–365. 518 Chien, S.H., Menon, R.G., 1995. Factors affecting the agronomic effectiveness of 519 phosphate rock for direct application. Fertil. Res. 41, 227–234. 520 https://doi.org/10.1007/BF00748312 521 da Silva, D.A., Tsai, S.M., Chiorato, A.F., da Silva Andrade, S.C., de Fatima Esteves, 522 J.A., Recchia, G.H., Carbonell, S.A.M., 2019. Analysis of the common bean 523 (Phaseolus vulgaris L.) transcriptome regarding efficiency of phosphorus use. 524 PLoS One 14, 1–27. https://doi.org/10.1371/journal.pone.0210428 525 De-Bashan, L.E., Bashan, Y., 2004. Recent advances in removing phosphorus from 526 wastewater and its future use as fertilizer (1997-2003). Water Res. 527 https://doi.org/10.1016/j.watres.2004.07.014 528 Deelstra, T., 1987. Urban Agriculture and the Metabolism of Cities. Food Nutr. Bull. 529 9, 1–3. https://doi.org/10.1177/156482658700900210 530 Doyle, J.D., Oldring, K., Churchley, J., Price, C., Parsons, S.A., 2003. Chemical Control 531 of Struvite Precipitation. J. Environ. Eng. 129, 419–426. 532 https://doi.org/10.1061/(ASCE)0733-9372(2003)129:5(419) 533 European Comission, 2016. Paket zur Kreislaufwirtschaft. Verordnung des 534 Europäischen Parlaments und des Rates mit Vorschriften für die Bereitstellung 535 von Düngeprodukten mit CE-Kennzeichnung auf dem Markt und zur Änderung 536 der Verordnungen 0084. 537 European Comission, 2014. Critical raw materials | Internal Market, Industry, 538 Entrepreneurship and SMEs [WWW Document]. 539 European Commission, 2020. A Farm to Fork Strategy for a fair, healthy and 540 environmentally-friendly food system 1–13. 541 Ferreira, A.J.D., Guilherme, R.I.M.M., Ferreira, C.S.S., Oliveira, M. de F.M.L. de, 2018. 542 Urban agriculture, a tool towards more resilient urban communities? Curr. 543 Opin. Environ. Sci. Heal. 5, 93–97. https://doi.org/10.1016/J.COESH.2018.06.004 544 Gell, K., Ruijter, F.J. d., Kuntke, P., Graaff, M. de, Smit, A.L., 2011. Safety and 545 Effectiveness of Struvite from Black Water and Urine as a Phosphorus Fertilizer. 546 J. Agric. Sci. 3, p67. https://doi.org/10.5539/jas.v3n3p67 547 Guo, S., Kaldenhoff, R., Uehlein, N., Sattelmacher, B., Brueck, H., 2007. Relationship 548 between water and nitrogen uptake in nitrate- and ammonium-supplied 549 Phaseolus vulgaris L. plants. J. Plant Nutr. Soil Sci. 170, 73–80. 550 https://doi.org/10.1002/jpln.200625073 551 J. R. Buchanan, J.R., C. R. Mote, C.R., R. B. Robinson, R.B., 1994. Thermodynamics of 552 Struvite Formation. Trans. ASAE 37, 617–621. 553 https://doi.org/10.13031/2013.28121 554 Kern, J., Heinzmann, B., Markus, B., Kaufmann, A.C., Soethe, N., Engels, C., 2008. 555 Recycling and Assessment of Struvite Phosphorus from Sewage Sludge. Agric. 556 Eng. Int. CIGR J. 10, 13. 557 Kouki, S., Abdi, N., Hemissi, I., Bouraoui, M., Sifi, B., 2016. Phosphorus fertilization 558 effect on common bean ( Phaseolus vulgaris L .) -rhizobia symbiosis 25, 1130– 559 1137. 560 Le Corre, K.S., Valsami-Jones, E., Hobbs, P., Parsons, S.A., 2009. Phosphorus recovery 561 from wastewater by struvite crystallization: A review, Critical Reviews in 562 Environmental Science and Technology. Taylor & Francis Group . 563 https://doi.org/10.1080/10643380701640573 564 Li, B., Boiarkina, I., Yu, W., Huang, H.M., Munir, T., Wang, G.Q., Young, B.R., 2019. 565 Phosphorous recovery through struvite crystallization: Challenges for future 566 design. Sci. Total Environ. 648, 1244–1256. 567 https://doi.org/10.1016/j.scitotenv.2018.07.166 568 Li, X.., Zhao, Q.L., 2003. Recovery of ammonium-nitrogen from landfill leachate as a 569 multi-nutrient fertilizer. Ecol. Eng. 20, 171–181. https://doi.org/10.1016/S0925- 570 8574(03)00012-0 571 Liu, Y., Rahman, M.M., Kwag, J.-H., Kim, J.-H., Ra, C., 2011. Eco-friendly Production of 572 Maize Using Struvite Recovered from Swine Wastewater as a Sustainable 573 Fertilizer Source. Asian-Australasian J. Anim. Sci. 24, 1699–1705. 574 https://doi.org/10.5713/ajas.2011.11107 575 Marschner, H., 1995. Mineral nutrition of higher plants. San Diego Academic P, 889. 576 https://doi.org/http://dx.doi.org/10.1006/anbo.1996.0155 577 Rahman, M.M., Salleh, M.A.M., Rashid, U., Ahsan, A., Hossain, M.M., Ra, C.S., 2014. 578 Production of slow release crystal fertilizer from wastewaters through struvite 579 crystallization - A review. Arab. J. Chem. 7, 139–155. 580 https://doi.org/10.1016/j.arabjc.2013.10.007 581 Rees, W., Wackernagel, M., 1996. Urban ecological footprints: Why cities cannot be 582 sustainable - and why they are a key to sustainability. Environ. Impact Assess. 583 Rev. 16, 223–248. https://doi.org/10.1016/S0195-9255(96)00022-4 584 Rufí-Salís, M., Brunnhofer, N., Petit-Boix, A., Gabarrell, X., Guisasola, A., Villalba, G., 585 2020a. Can wastewater feed cities? Determining the feasibility and 586 environmental burdens of struvite recovery and reuse for urban regions. Sci. 587 Total Environ. 737, 139783. https://doi.org/10.1016/j.scitotenv.2020.139783 588 Rufí-Salís, M., Petit-Boix, A., Villalba, G., Ercilla-Montserrat, M., Sanjuan-Delmás, D., 589 Parada, F., Arcas, V., Muñoz-Liesa, J., Gabarrell, X., 2020b. Identifying eco- 590 efficient year-round crop combinations for rooftop greenhouse agriculture. Int. 591 J. Life Cycle Assess. https://doi.org/10.1007/s11367-019-01724-5 592 Rufí-Salís, M., Petit-Boix, A., Villalba, G., Sanjuan-Delmás, D., Parada, F., Ercilla- 593 Montserrat, M., Arcas-Pilz, V., Muñoz-Liesa, J., Rieradevall, J., Gabarrell, X., 594 2020c. Recirculating water and nutrients in urban agriculture: An opportunity 595 towards environmental sustainability and water use efficiency? J. Clean. Prod. 596 261, 121213. https://doi.org/10.1016/J.JCLEPRO.2020.121213 597 Sena, M., Hicks, A., 2018. Life cycle assessment review of struvite precipitation in 598 wastewater treatment. Resour. Conserv. Recycl. 599 https://doi.org/10.1016/j.resconrec.2018.08.009 600 Shu, L., Schneider, P., Jegatheesan, V., Johnson, J., 2006. An economic evaluation of 601 phosphorus recovery as struvite from digester supernatant. Bioresour. Technol. 602 97, 2211–2216. https://doi.org/10.1016/J.BIORTECH.2005.11.005 603 Smit, J., Nasr, J., 1992. Urban agriculture for sustainable cities: using wastes and idle 604 land and water bodies as resources. Environ. Urban. 4, 141–152. 605 https://doi.org/10.1177/095624789200400214 606 Steen, I., 1998. Phosphorus availability in the 21st Century: Management of a non- 607 renewable resource. Phosphorus and Potassium 25–31. 608 Stratful, I., Scrimshaw, M.D., Lester, J.N., 2004. Removal of Struvite to Prevent 609 Problems Associated with its Accumulation in Wastewater Treatment Works. 610 Water Environ. Res. 76, 437–443. https://doi.org/10.2175/106143004X151491 611 Talboys, P.J., Heppell, J., Roose, T., Healey, J.R., Jones, D.L., Withers, P.J.A., 2016. 612 Struvite: a slow-release fertiliser for sustainable phosphorus management? 613 Plant Soil 401, 109–123. https://doi.org/10.1007/s11104-015-2747-3 614 Thomaier, S., Specht, K., Henckel, D., Dierich, A., Siebert, R., Freisinger, U.B., Sawicka, 615 M., 2015. Farming in and on urban buildings: Present practice and specific 616 novelties of Zero-Acreage Farming (ZFarming). Renew. Agric. Food Syst. 30, 43– 617 54. https://doi.org/10.1017/S1742170514000143 618 Uysal, A., Demir, S., Sayilgan, E., Eraslan, F., Kucukyumuk, Z., 2014. Optimization of 619 struvite fertilizer formation from baker’s yeast wastewater: Growth and nutrition 620 of maize and tomato plants. Environ. Sci. Pollut. Res. 21, 3264–3274. 621 https://doi.org/10.1007/s11356-013-2285-6 622 623