In this experiment, two recycled fertilizers were chosen due to promising results in earlier studies: one was a struvite containing P fertilizer (0.66% N, 6.98% total-P of which< 1% is water soluble, 0.24% K, STR) produced from biogas digestate with a new procedure called NuTriSep, the second one was a thermic-chemically treated sewage sludge-ash (7.97% total-P, 5.9% Fe, 5.1% Al, SSA) from the ASH -DEC procedure (Hermann and Schaaf, 2016). The NuTriSep procedure separates P, N, and K from the digestate to enable the application of individual nutrients orientated on actual crop demand. We tested in this trial the P recyclate. Both soils received the same amount of P (50 mg P kg^-1 dry substrate). The recycled fertilizers did not apply a considerable amount of N or K, that is why the fertilizer amount of N and K was not adopted.

Regarding the BSs, we used commercial products or products that will soon be released to the local market to increase the chance of a successful inoculation of the microbial BSs and an effective action in the soil. We have chosen to include non-microbial and microbial biostimulants to be able to directly compare their effectiveness. For the microbial biostimulants, we only included microbial consortia as they were shown to be more effective than single strains (Herrmann et al., 2022). Therefore, we included two different plant extracts, one humic acid, and two microbial consortia. The available details of the products as well as the applied dosages can be found in Table 1 . The microbial products contained microorganisms with known P solubilization ability (Elhaissoufi et al., 2022). To ensure an effective and visible action of these products, we used a higher dosage than the recommendation of the manufacturers, which may have influenced the results of the trial. For the microbial consortia inoculation, the application rate of both BSs was adopted to reach the same colony-forming unit per gram soil which was set to 10⁸ colony-forming units (cfu) g^-1 soil as successfully implemented in earlier experiments (Nkebiwe et al., 2016). This corresponded to a 100 times higher dosage compared to the recommendation of the manufacturer. To make the results comparable between non-microbial and microbial products, we also chose for the incubation trial without plants for the non-microbial products a 100 times higher dosage than the recommendation of the manufacturer. However, because the plant extracts may become phytotoxic (manufacturers advice) if the applied concentrations are too high, we reduced the application amount of the non-microbial BSs to a 10 times higher dosage compared to the recommendation of the manufacturer in the pot trial which included plants. Nutrient additions due to the low application amount are negligible for the BSs.

In the incubation trial, we tested two soils with different P contents, while in the pot trial we only used the soil with the lower P content. Relevant information of the soils can be found in Table 2 . Both soils were retrieved from the topsoil (0-20 cm) and were sieved with a ø5 mm-sieve after airdrying.

Summarizing the above-mentioned treatments, the incubation trial consisted of 36 treatments (2 soils x 6 BSs including one control x 3 fertilizer treatments including one unfertilized control), and the pot trial included 12 treatments (1 soil x 4 BSs including one control x 3 fertilizer treatment including one unfertilized control). In both cases, a randomized complete block design was used with 5 and 4 blocks for the incubation and pot trial, respectively.

The incubation trial was set up for 8 weeks in incubation cabinets with a temperature of 20°C. 750 g dry soil was each filled in 1 L glasses and watered with distilled water 1 week before the start of the experiment to reach a water holding capacity of 50% which corresponds to 34% (soil 1) and 28% (soil 2) water content. BSs and fertilizers were given at the same time by applying the necessary amounts and thoroughly mixing the soil. The liquid amount given to each of the treatments was balanced by giving distilled water to treatments receiving a lower amount of liquids. The glasses were covered with a thin plastic layer to avoid water loss but to guarantee aerobic conditions in the glasses at the same time. The water content was checked gravimetrically once per week and water was added accordingly.

Soil samples were collected 5 times during the experiment: 2 days, 1 week, 2 weeks, 4 weeks, and 8 weeks after fertilization. For this, soil was watered and mixed again, and a representative sample was taken. Afterwards, when watering, the weight loss due to the sampling was accounted for. For the analysis of microbial biomass P and phosphatase activity, fresh samples were taken 1 week and 4 weeks after fertilization and were stored at 4°C until further analysis.

The greenhouse trial was set up for 6 weeks in the greenhouse facilities at University of Hohenheim from May 2022 to July 2022. Temperature and relative humidity in the greenhouse were recorded automatically several times during the day and night. The average temperature during the growing period was 26.6°C, and the relative humidity was on average 48.3%. The substrate used for this experiment consisted of 70% of dry low P soil (P-Class A) and 30% of quartz sand (ø 0.6-1.2 mm) on a weight basis. Besides P fertilizers, the substrate was fertilized with 150 mg N (kg dry substrate)^-1 in the form of Ca(NO₃)₂, 200 mg K (kg dry substrate)^-1 in the form of K₂SO₄, and 100 mg Mg (kg dry substrate)^-1 in the form of MgSO4. All these fertilizers were homogeneously mixed with the soil and 6 kg of the substrate (dry matter) was filled in Mitscherlich-pots.

We used the maize (Zea mays L.) cultivar Stabil (KWS SAAT SE & Co. KG, Einbeck, Germany). Three seeds were sown and then thinned to one after emergence. BS were applied to the soil close to the seeds. The initial plan was to apply the BS amount once after sowing (in the seed hole) and again after germination (close to the plant after thinning). Since germination of the first set failed and we had to re-sow the seeds, we decided to double the BS application upon the germination of the second planting. This was to make up for any BS concentration lost when taking out seeds/plants. Throughout the experiment, moisture was kept at 70% water holding capacity, and pots were gravimetrically watered 2 to 3 times per week depending on the weather.

Soil samples were taken 1 day after fertilization, post-emergence (6 days after sowing), at the 3-leaf stage (12 days after sowing), at the 6-8 leaf stage (32 days after sowing), and at final harvest at the 9-leaf stage (48 days after sowing), respectively. During the growing period of maize, soil was taken from the first 10 cm of the pots close to the plants and mixed thoroughly. Fresh soil samples were taken for the analysis of microbial biomass P and phosphatase activity at final harvest, which were stored at 4°C until further analysis.

Statistical analysis was implemented with SAS Software (Version 9.4. SAS Institute Inc., Cary, NC, USA). To correct for potential variations within one soil analysis method due to different measuring time points, a control soil was always analyzed together with the soil samples. In the statistical model, the control soil was, then, added as a covariate, and blocks were included as a random effect in order to eliminate that error. Our full linear mixed model for the unrepeated case was as follows:

where µ is the intercept, b is the random effect of the ith block, s is the linear correction factor based on the control soil, α, β, γ are the effects of jth soil, the kth BS and the lth fertilizer. y_ijkl is the mean with a homogeneous and normal distributed residual error e_ijkl. Because in the pot trial only one soil was used, the parameter vectors for soil and all respective interaction terms were excluded.

The final model was then chosen based on the results of the ANOVA by removing factors with non-significant effects from the model.

In the case of repeated measurements, a time effect was included in the model. To consider the correlation among the observation time points, the best fitting variance-covariance matrix was chosen based on the AIC criterion (Wolfinger, 1993). For CAL-P measurements in the pot, leaf area, and stem diameter, time was considered as a metric variable since a simple polynomial model could be fit which was able to explain the data sufficiently. In any other case, time was included as a categorical variable.

Illustrations of the graphs were implemented with the R package ggplot2 (Wickham, 2016). Fisher’s LSD test of pairwise differences (Fisher, 1936) were performed based on the final reduced model and are displayed with letters in the graphs.

In the current incubation test, CAL-P was strongly dependent on soil, fertilizer, BSs, and the sampling time ( Supplementary Table 1 ). CAL-P showed a decreasing trend for the fertilizer treatments till the fourth sampling time, 28 days after fertilizer application while the unfertilized control maintained its CAL-P content over time ( Figure 1A ). Thereby, the STR treatment declined faster in the beginning than the SSA leading to a convergence at the end of the experiment of both fertilizer treatments’ CAL-P content in soil. In the case of the moderate P-deficient soil (Class B), the difference between the two fertilizers was no longer significant at the end of the experiment with a content of 42.5 mg (kg dry soil)^-1 for the SSA treatment and 46.3 mg (kg dry soil)^-1 for the STR treatment.

The addition of humic acids to the soil resulted in a slightly higher CAL-P content for the unfertilized control in the moderately P deficient soil (Class B), and for the fertilized treatments in the highly P-deficient soil ( Figure 1B ). All other BSs did not vary substantially from the untreated control and the fertilization effect was overall predominant.

While the microbial biomass P was not significantly affected by any treatment ( Supplementary Table 1 , data shown in Supplementary Figure 1 ), the acidic and alkaline phosphatase activity were influenced by the fertilizers and the soil ( Supplementary Table 1 ). The acidic phosphatase activity expressed as p-nitrophenol development in dry soil over time was increased by 13.4 µg g^-1 h^-1 by the struvite containing P fertilizer compared to the unfertilized control independent of the soil ( Figure 2A ). The alkaline phosphatase activity showed a significantly higher activity (by 6.5 µg g^-1 h^-1) when fertilized with P ( Figure 2B ). Both enzymes were more stimulated in the soil with the lower P content ( Figures 2A, B ).

The soil pH did not reflect any relevant changes by a BS application or by the fertilizer application which could result in a higher plant P availability ( Supplementary Table 3 ).

BSs had no significant effect on the growing speed of the plants ( Supplementary Table 2 ). All plants started growing at the same rate, but by the second measurement time, the STR and SSA treatments were significantly higher than the unfertilized control ( Figure 3B ). Likewise, BSs had no significant effect on leaf area and stem diameter. Differences among fertilizer treatments were minuscule at the start, while the SSA and STR started to diverge from the unfertilized control treatment by the second measurement time. At harvest, the SSA and STR treatments did not vary significantly from one another regarding any plant trait ( Figures 3A–D ).

Fertilizing the plants increased the P content and concentration of the plants, while the fertilizers did not differ significantly from each other ( Figures 4A, B ). BSs did not affect root growth significantly, and only the fertilization effect was visible. Consistently, the two fertilizers increased root growth compared to the control but did not differ from each other ( Figures 5A, B ).

Concerning the soil analysis, CAL-P was not affected by the BS application, but by time and fertilization ( Supplementary Table 2 ). At the beginning of the trial, all fertilization treatments varied significantly from each other with the STR treatment leading to the highest CAL-P content of 38.6 mg (kg dry soil)^-1 followed by the SSA treatment with 24.1 mg (kg dry soil)^-1 ( Figure 6 ). The slopes of the two fertilizers are not significantly different from each other leading to a parallel decrease over time for the fertilizer treatments while the control stayed stable at the same level of around 7 mg (kg dry soil)^-1.

The BSs had a statistically significant effect on microbial biomass P ( Supplementary Table 2 ). The application of the SSA resulted in a considerable drop of the microbial biomass P in the control and extracts, which could not be observed with humic acids and microbial consortium ( Figure 7 ). When the soil was inoculated with the BS “M2” the application of SSA led to an increase in microbial biomass P (15.3 mg P (kg dry soil)^-1) compared to the other fertilization treatments (10.5 mg P (kg dry soil)^-1 for the unfertilized control and 10.1 mg P (kg dry soil)^-1 for the STR) ( Figure 7 ). Regarding the enzyme activity, BSs and fertilization had no significant effect on the acidic or alkaline phosphatase activity ( Supplementary Table 2 ).

The pH was significantly affected by BS application (p = 0.0157) and fertilizer treatment (p<0.0001). While the fertilizers slightly increased the pH by 0.2 from 4.9 to 5.1, the plant extract and the humic acid decreased the pH by 0.1 compared to the control (5.0), respectively ( Supplementary Table 4 ).

BSs failed to ensure a promotion of plant growth under P limiting conditions with and without the supply of recycled fertilizer products. The reliability of BS product is limited due to their nature of indirectly influencing the soil which is also highlighted by existing studies in which BSs improved P fertilizer use efficiency of recycled fertilizers (Mpanga et al., 2018; Wollmann and Möller, 2018) and others in which they did not (Lekfeldt et al., 2016; Wollmann et al., 2018; Jastrzębska et al., 2019). The soil had a strong P limitation. Previous studies confirmed that a starter fertilization of P under strongly P-limiting conditions is needed to ensure or improve the action of various beneficial microorganisms such as AMF and P-solubilizing microorganisms (Treseder and Allen, 2002; Schütz et al., 2017) but also for active natural compounds such as humic acids and seaweed extracts (Li et al., 2022). However, the fertilization of P led to a significant increase in soil P turnover activity, P availability, and plant performance. The recycled P fertilizers nourished the plants sufficiently with P as P concentrations in shoots are above the critical threshold (Bergmann and Neubert, 1976) which may also explain the lack of action by the applied BSs.

Even though in the pot and incubation trial the two fertilizers differed clearly concerning their increase in the CAL-P content in soil, both can be considered equivalently effective regarding their impact on the plant growth parameters as well as nutrient uptake. Similar results were also obtained by previous studies (Nanzer et al., 2014; Severin et al., 2014; Wollmann et al., 2018) and it is widely recognized that CAL-P is not a direct predictor of plant growth and plant P uptake, especially when using alternative P sources (Duboc et al., 2017).

One main result regarding BS application to the soil is the higher CAL-P content in the case of the application of humic acids while other BSs did not influence the CAL-P content. Previous studies underlined this result by achieving a higher P availability when humic acids are applied together with fertilizers (Wang et al., 1995; Du et al., 2013). Humic acids can directly act against P adsorption onto minerals to the soil by competing with P for binding sites, by the dissolution of bound P, and by the inhibition of P precipitation (Du et al., 2013; Perassi and Borgnino, 2014; Gerke, 2021). Thus, these mechanisms can explain the correlation of the applied humic acid amount and the P availability in soil (Yuan et al., 2022). The application of humic acids may be more reliable than the inoculation of microorganisms which may not establish in soil or colonize the rhizosphere (Podile et al., 2014) or by an insufficient P solubilizing capability to deliver nutrients to the plants (Raymond et al., 2021). 2021). However, the impact of humic acids was not consistent in both soils in our study. This soil-dependent behavior was also observed in previous studies (Maluf et al., 2018) and could be related to soil pH (Gerke, 2021), clay contents (Maluf et al., 2018), and possibly other edaphic factors. Nevertheless, the effect of humic acids on CAL-P was possibly overshadowed in the pot trial by the effect of the plant and fertilizers. With higher amounts of applied humic substances, the impact on P mobilization could be further enforced and possibly be also relevant in plant trials like implemented in previous studies (Hua et al., 2008).

The microbial biomass P was not significantly altered without the plant in the incubation trial while a significant BS and SSA interaction was detected in the pot trial. This is an indicator of the necessity of a plant for the BSs to establish or interact with soil nutrients, especially for the microorganisms which need a carbon-rich environment to grow. In the pot trial, a sharp drop of microbial biomass P was observed when pots were fertilized with SSA. Previous studies have also detected a similar response when plants were fertilized with P (Clarholm, 1993) while in other studies microbial biomass P was positively affected by a P fertilization (Shi et al., 2020). However, it seems that humic acids and the microbial consortia have the capability to prevent such a decrease. Other studies revealed that seaweed extracts and humic acids shaped microbial community structure in soil but also on plant roots (Renaut et al., 2019; Yuan et al., 2022) and plant extracts could stimulate microbial degradation of organic material (Chen et al., 2002). Thus, also plant extracts can potentially alter the microbial community. Increasing the application frequency and the dosage of plant extracts may be a strategy to induce effects in microbial communities (Alam et al., 2013). The combination of the microbial consortia 2 and the SSA treatment even appeared to favor an establishment of the microbes while this was not the case for other fertilizers and microbe combinations. Despite this significant impact on the microbial biomass P, plant traits were not affected by the BS application.

Phosphatase activity in soil was not affected by BS application. In the contrary, previous studies have shown a reduction of phosphatase activity in soil by humic acids (Malcolm and Vaughan, 1979; Yuan et al., 2022) and a stimulation of phosphatase activity by mycorrhization of plants (Tarafdar and Marschner, 1994; Joner et al., 2000), by phosphate solubilizing bacteria (Kim et al., 1997) and by the addition of seaweed extracts (Chen et al., 2023). However, in none of the cases, the phosphatase activity was measured without plants. In our trial with plants, the influence on BSs may have been masked by a stronger influence of plants at harvest stage due to dense rooting of the pots. However, other studies also concluded that mobilization of P turnover processes in soil could not be identified as an important mechanism contributing to improved P availability by biostimulants, but rather the promotion of root growth (Raymond et al., 2021; Nkebiwe et al., 2024). In the incubation trial, we detected a fertilization effect on acid and alkaline phosphatase activity, while this was not recognizable in the pot trial. In the literature, consistency for fertilization effects on phosphatase activity and the correlation of soil P content is not given. For example, Garg and bahl (2008) found an increase in acid phosphatase activity with increasing soil P content, while Spiers and McGill (1979) concluded the opposite or could not find an impact of fertilization on phosphatase activity. In our case, it can be assumed that a lower P-availability in the soil itself increases the phosphatase activity potential in soil as plants and microorganisms require a higher mobilization capability with a lower P-availability. A short-term fertilization, however, can stimulate phosphatase activity in soil. Another decisive factor whether phosphatase activity is stimulated or not by P addition, is the P source provided via fertilization. Highly soluble P fertilizers decreased phosphatase activity while non-soluble P fertilizers did not affect or slightly increased the phosphatase activity (Redel et al., 2019). Both used recycled fertilizers are characterized by a low immediate solubility in soil, which could explain the stimulation of phosphatase activity by both fertilizers. As above mentioned, plants densely rooted the pots at the end of the trial, potentially masking an impact of the treatments, in the pot trial. Nevertheless, it can also be argued that available P content in soil is not a good predictor of phosphatase activity in soil. Margalef et al. (2017) could demonstrate that a correlation of acid phosphatase activity and available P content is not given, and that the enzyme activity is more connected to the available organic P content in soil.