The simultaneous removal of the mixture of pesticides was assayed in 50 mL polypropylene tubes containing 5 g of biomixture. Each tube was spiked with the commercial formulations (30 mg kg⁻¹ each, nominal concentration), manually homogenized and incubated in the dark at (25 ± 1) °C; for each biomixture (screening phase), triplicate tubes were withdrawn for analysis at each sampling point (0, 2, 6, 9, 15, 21, and 30 d) for pesticide quantification. During the optimization phase an analogous procedure was followed, using only the CF containing biomixture at the compositions indicated in Figure 1A; duplicate tubes were prepared per composition, except for the composition corresponding to the central point of the CCD (five replicates); pesticide concentrations were determined at time zero and after a treatment period of 30 d. Additional unitary systems containing 300 g biomixture were sampled at times 0, 15 and 30 d to perform ecotoxicological analysis with D. magna and Lactuca sativa in the screening phase.

The mineralization of ¹⁴C-chlorpyrifos was determined through ¹⁴CO₂ production in biometric flasks, as described in Huete-Soto et al. (2017a). Briefly, 50 g of each biomixture were spiked (triplicates) with either commercial chlorpyrifos (30 mg kg⁻¹) or the mixture of pesticides (30 mg kg⁻¹ each), plus ¹⁴C-chlorpyrifos (5,000 dpm g⁻¹) and incubated at 25°C; triplicate blanks without pesticides were included for each biomixture. The KOH (0.25 M) contained in the CO₂ traps was withdrawn at selected times and replaced with the same amount of fresh KOH during a period of 44–46 d. Activity of ¹⁴C in the ¹⁴CO₂ produced from the mineralized pesticide was analyzed in the KOH samples as described in section 2.5.2.

Microbial respiration was determined in the biometric flasks (section 2.4.2), using 10 g biomixture with no addition of pesticides; NaOH (0.55083 M) was added in the CO₂ trap. Analysis is described in section 2.5.3.

Extraction of pesticides was carried out following a method described by Ruiz-Hidalgo et al. (2014), which employs a mixture of water and acidified acetonitrile (formic acid 1% v/v) as the extractant. Carbofuran-d₃ and linuron-d₆ were added as surrogate and internal standard, respectively. Analysis of sample extracts was performed by LC–MS/MS using ultra high performance liquid chromatography (UPLC-1290 Infinity LC, Agilent Technologies, CA) coupled to a triple quadrupole mass spectrometer (model 6,460). The chromatographic separation method is described in Chin-Pampillo et al. (2015). Selected transitions, limit of detection (LOD) and limit of quantification (LOQ) for the analytes are shown in Supplementary Table S2. Remaining pesticide concentrations were used to calculate the percentage of pesticide removal. Removal rate constants and pesticide half-lives (DT₅₀) were estimated in Python 3.9 using either a three parameter (Eq. 2) or a simple modified (Eq. 3) first order exponential decay model:

Where y is the pesticide concentration, t is time elapsed, y₀, a,b,c are parameters of the model. Removal rate constants were analyzed by means of ANOVA tests to compare regression lines using the InfoStat.

Scintillant liquid (10 mL) was added to 2 mL aliquots from the removed KOH solution samples and the ¹⁴C activity from the trapped ¹⁴CO₂ was measured by liquid scintillation (Beckman LS6000SC counter). Mineralization was estimated by comparing the total cumulative ¹⁴CO₂ activity evolved from the biomixture and the activity of the ¹⁴C-chlorpyrifos initially added; data was modeled in Python 3.9.

The CO₂ captured as Na₂CO₃ was precipitated with BaCl₂, and the remaining NaOH was titrated with HCl; basal respiration was calculated as described in Pell et al. (2006).

Elutriates obtained from biomixture samples were prepared according to the protocol EPA-823-B-01-002 (US EPA, 2001), by adding 40 mL distilled water to 10 g samples followed by mechanical shaking (1 h) and centrifugation (10 min at 3500 rpm). The immobilization test in D. magna was performed as described by Lizano-Fallas et al. (2017). Briefly, ten daphnid neonates (triplicates) were exposed to 10 mL of elutriate dilutions in moderately hard reconstituted water (21 ± 1°C for 48 h); immobility of neonates was determined and assumed as equivalent to mortality. The relative concentration of the sample that resulted in 50% of immobilization in the daphnids (EC₅₀) was calculated using Python 3.9.

The phytotoxicity of the samples was determined with seed germination tests in lettuce (Lactuca sativa var. Georgia) (US EPA, 1996). Relative seed germination (SG), relative root elongation (RE) and germination index (GI) were determined using 10 seeds exposed to elutriate samples (5 mL), after 6 d of incubation in darkness at 22°C. These parameters were determined by comparison to germination controls obtained by exposure to distilled water and were calculated as described elsewhere (Lizano-Fallas et al., 2017).

Respiration of biomixtures was assessed as a preliminary indicator of potential pesticide removal capacity. Respiration data revealed significantly higher initial rates in the CH-biomixture (10.2 ±0.2 mgCO₂ kg⁻¹·h⁻¹), compared to RH- (3.83 ± 0.34 mgCO₂ kg⁻¹·h⁻¹) and CF-biomixtures (1.47 ± 0.34 mgCO₂ kg⁻¹·h⁻¹). Respiration rate is usually considered as a marker of biological activity, which correlates with the capacity of soils to transform organic pollutants (Pascual et al., 2000). Nonetheless, in biomixtures this parameter has been previously determined to estimate the effect of antibiotics of agricultural use on these matrices (Castillo-González et al., 2017; Huete-Soto et al., 2017a).

The mineralization of ¹⁴C-chlorpyrifos was determined as another indicator of pesticide transformation capacity of the matrix. Significantly higher final mineralization values (after 45 d) were achieved in the CF-biomixture in both cases (17.7 ± 6.2% and 15.0 ± 6.2% for chlorpyrifos alone or in the mixture containing the 12 target pesticides, respectively), compared to the other biomixtures (up to 6.5% in the CH-biomixture and 11.0% in the RH-biomixture) (Figure 2). Other studies have reported comparable mineralization values in analogous biomixtures, ranging from 10 to 17.6% after 62 d for ¹⁴C-chlorpyrifos (Castillo-González et al., 2017; Huete-Soto et al., 2017a) and 13% after 28 d for ¹⁴C-carbofuran (Jiménez-Gamboa et al., 2018). The mineralization capacity, however, did not correlate with the respiration rates, and conversely, the highest mineralization was observed in the biomixture with the lowest respiration rate. Consequently, parameters directly related to pesticide removal/transformation should be prioritized over matrix respiration when evaluating the performance biomixtures.

The removal profiles for individual pesticides during the treatment in three different biomixtures are shown in Figure 3. Overall, the CF-biomixture exhibited the best removal performance, achieving a significantly lower final total concentration of pesticides after 30 d of treatment, followed in descendant order by the CH- and RH-biomixtures. Estimated DT₅₀ values for individual pesticides are shown in Table 2; given that DT₅₀ concentrations were experimentally reached only for five pesticides, no DT₅₀ values were estimated in the other cases; instead, removal values (%) by the end of the experiment are reported. Although the elemental composition of the biomixtures was not determined, the differences among their C/N ratios are exclusively related to the lignocellulosic substrate (as the proportion and origin of soil and compost is the same). Given that the C/N for CF (~83) is much higher than those for RH (~23) and CH (~30) (Shemekite et al., 2014; AboDalam et al., 2022), the CF-biomixture has a higher C/N value. The biomixtures with high C/N ratios are known to favor the activity of white rot fungi and their enzymatic machinery, which may enhance the oxidation of organic micropollutants (Yang et al., 2013); this observation might at least partially explain the best pesticide removal observed in this matrix.

Half of the evaluated pesticides belonged to the group of triazole fungicides. For these compounds (epoxiconazole, hexaconazole, propiconazole, tebuconazole and triadimefon), most of the removal took place during the first 3–4 days of treatment. However, the removal rate decreased by day five, reaching constant values in most cases, which ultimately did not permit the estimation of DT₅₀ values, except in the case of triadimefon whose DT₅₀ values (lowest DT₅₀ of 4.2 d in CF-biomixture) were below the reported elimination in soil (DT₅₀ 26 d; Lewis et al., 2016). In general, triazoles are usually considered as persistent compounds, and their removal in soil ranges from 35 d to 365 d for the remaining compounds employed in this study (Lewis et al., 2016). On the other hand, triazole removal in biomixtures has revealed contrasting results, i.e., from DT₅₀ = 19 d (Delgado-Moreno et al., 2017) to negligible elimination for tebuconazole (Murillo-Zamora et al., 2017), or cases of faster (epoxiconazole, DT₅₀ = 61 d; Fogg et al., 2003), similar (hexaconazole, DT₅₀ = 187 d); or slower (propiconazole, DT₅₀ = 160 d; Spliid et al., 2006) elimination compared to reports in soil. It should also be considered that removal behavior might change depending on the combination of pesticides simultaneously occurring in a matrix (Huete-Soto et al., 2017b), likely due to the ecotoxicological effects of the pesticide mixture (and their metabolites) on the potentially degrading microbial communities (Bose et al., 2021).

The concentration profile for triadimenol differs from the other triazoles, as accumulation was observed in the CH- and RH-biomixtures, while initial removal followed by accumulation was determined in the CF-biomixture. Such behavior can be explained as triadimenol is the main transformation product of triadimefon in soil (Hou et al., 2022). Hence, data suggest that faster initial removal of triadimenol in the CF-biomixture was then masked by the transformation of triadimefon (and the consequently production of triadimenol); on the contrary, the accumulation rate of newly produced triadimenol was always higher than triadimenol removal rates in the other biomixtures, thus resulting in its net accumulation. Longer treatment periods are recommended to estimate the extent of triadimenol removal in the biomixtures, more importantly considering that its DT₅₀ in soil is around ten-times longer than that of triadimefon (26 d versus 250 d; see Table 2) (Hou et al., 2022).

Three compounds (fluazifop-p-butyl, oxamyl, thiophanate-methyl) showed almost complete removal after short treatment periods. Fluazifop-p-butyl showed shorter DT₅₀ values (estimated DT₅₀ = 0.3 d in all biomixtures) compared to its removal in soil (Lewis et al., 2016). In the case of oxamyl the removal rates (DT₅₀ ranging from 0.3 d in the CH-biomixture to 2.32 in the CF- biomixture) were higher than previously reported in soil and biomixtures (Osman et al., 2009; Lewis et al., 2016). Similarly, the dissipation rate for thiophanate-methyl (DT₅₀ ranging from 0.85 d in the CF-biomixture to 1.44 d in the RH-biomixture) was within the removal values in soil (Lewis et al., 2016), but shorter than reported by El-Sebai et al. (2023) in a rice husk-compost-based biomixture.

The removal of carbendazim, although not complete, exhibited DT₅₀ values ranging from 12.6 d (CF-biomixture) to 20.7 d (RH-biomixture); they were in general faster than observed dissipation rates in soil, but similar or longer than estimated in other biomixtures (Castillo-González et al., 2017; Huete-Soto et al., 2017b; Lescano et al., 2022). Moderate removal was observed for the organophosphate chlorpyrifos (maximum of 47.4% after 30 d in the CF-biomixture), whose elimination in biomixtures has shown DT₅₀ values ranging from 9 d to 69 d (Kravvariti et al., 2010; Masís-Mora et al., 2019; Pérez-Villanueva et al., 2022). The described accumulation of the chlorpyrifos metabolite 3,5,6-trichloropyridinol (TCP) in biomixtures (Tortella et al., 2012) is known to potentially delay chlorpyrifos degradation due to the toxic effect TCP exerts on degrading microbial communities (Bose et al., 2021). Finally, imidacloprid is highly persistent both in soil (Lewis et al., 2016) and biomixtures; reports in the latter indicate a DT₅₀ value of 83 d (Lescano et al., 2022) or even negligible dissipation (Huete-Soto et al., 2017b). Although some extent of removal was achieved in this work (up to 35.6% after 30 d in the CF-biomixture), imidacloprid dissipation reached a more or less constant value after 15 d. Significantly longer treatment periods are likely to be required to remove such highly persistent compounds, or the use of more than one BPS at on-farm level, to treat more recalcitrant compounds (Acosta-Sánchez et al., 2020).

Ecotoxicological assays were performed in D. magna and Lactuca sativa to monitor the removal of pesticides in the biomixtures. The immobilization test in D. magna (Figure 4A) showed a steady significant detoxification in the biomixtures during the treatment period, shifting from initial EC₅₀ values ranging from 0.65–0.77% to 2.70–3.17% after 15 d and to 6.57–6.67% after 30 d, in a detoxification pattern that did not exhibit significant differences among the three biomixtures. Despite differences in the pesticide removal performance of biomixtures, the similarity in their ecotoxicological outcome supports the versatility of these matrices, in which detoxification may be also related to adsorption processes during the aging of the biomixtures (Castillo et al., 2008). Nevertheless, the detoxification due to transformation of the parent pesticides remains the most desired process from the bioremediation point of view, as long as toxic transformation products are not formed or accumulated. Similar detoxification towards D. magna was also observed in compost-based biomixtures used for the treatment of seven fungicides (including carbendazim, epoxiconazole, tebuconazole, and triadimenol) (Murillo-Zamora et al., 2017), a mixture of triazines plus chlorpyrifos (Lizano-Fallas et al., 2017) or mixtures of carbamates (Rodríguez-Rodríguez et al., 2017). On the contrary, fluctuations in the toxicity towards D. magna with no net detoxification were described during the treatment (115 d) of mixtures of up to nine pesticides (including four used in this work) (Huete-Soto et al., 2017b); moreover, increased toxicity has been reported, which may take place during the formation of highly toxic transformation products or due to the accumulation of specific pesticides in cases of several applications on the biomixture (Acosta-Sánchez et al., 2020).

Among the treated pesticides, chlorpyrifos is by far the most toxic to D. magna (EC₅₀ = 0.00010 mg L⁻¹; Lewis et al., 2016), and is likely the most important compound within the mixture to shape the final ecotoxicological outcome in the matrix, while carbendazim (EC₅₀ = 0.15 mg L⁻¹) and oxamyl (EC₅₀ = 0.319 mg L⁻¹) follow in toxicity. All these three compounds were significantly removed in the biomixtures, which might explain the overall decrease in their residual toxicity. Nonetheless, the final toxicity will depend on the interactions between the complex mixture of pesticides and their transformation products (Hernández et al., 2017). The adsorption of the pesticides during the aging of the matrix might also decrease their bioavailability (and their biological removal) (Castillo et al., 2008), which may contribute to the observed detoxification. Even though detoxification was achieved in this work, longer treatments are recommended to further decrease the residual toxicity.

Germination tests in Lactuca sativa showed a similar trend to that observed in daphnids: a significant detoxification in the CF- and RH-biomixtures after 30 d of treatment, with respect to the initial toxicity (Figure 4B). No detoxification was observed in the CH-biomixture, as the initial GI value (higher compared to the other matrices) did not differ from the GI at times 15 d or 30 d; such unexpected high initial GI could be related to specific pesticide adsorption to the CH-biomixture, that hindered pesticide release during the aqueous extraction. The decrease in the toxicity towards Lactuca sativa has been described during the removal of herbicides in biomixtures (Acosta-Sánchez et al., 2020) and during the treatment of pesticide mixtures (Huete-Soto et al., 2017b). As in the case of the test in D. magna, final toxicity values were not significantly different regardless of the biomixture. For this reason, although detoxification was achieved, the ecotoxicological data did not play a key role in defining the biomixture of best performance for further optimization assays.