The growing adoption of wastewater (WW) reuse for agricultural irrigation has emerged as a strategic response to global water scarcity (Christou et al., 2024). Treated wastewater (TWW) offers a reliable alternative to freshwater, supporting crop production in water-stressed regions (Heyde et al., 2025; Ofori et al., 2021). However, despite advances in WW treatment technologies, residual micropollutants—including antibiotics, disinfectants, and heavy metals—along with antibiotic-resistant bacteria (ARB) carrying antibiotic resistance genes (ARGs), may persist in WW even after treatment (Hendriksen et al., 2019; Matesun et al., 2024; Pärnänen et al., 2019; Pazda et al., 2019; Rizzo et al., 2013). The co-occurrence of these micropollutants with bacteria harbouring mobile genetic elements (MGEs) and integrons facilitates the horizontal transfer of ARGs, contributing to the emergence of new ARBs (Berglund et al., 2023; Di Cesare et al., 2016; Heuer and Smalla, 2012; Manaia et al., 2018; Murray et al., 2024; Narciso-da-Rocha et al., 2018) and potentially creating reservoirs of antibiotic resistance in agricultural soils (Han et al., 2022). Once introduced into agricultural soils, WW-borne micropollutants can alter both physicochemical and microbial soil properties (Becerra-Castro et al., 2015; Siebe and Fischer, 1996) and exert a selective pressure on soil and plant-associated microbial communities, favouring the survival of resistant strains, and promoting the persistence and dissemination of ARGs (Larsson and Flach, 2021). This raises serious concerns about the potential transmission of resistance traits into the food chain (Bengtsson-Palme et al., 2023; Blau et al., 2018; Christou et al., 2017; FAO and WHO, 2019; Gekenidis et al., 2018; Hirt, 2020; Owino et al., 2020; Pan and Chu, 2017).

A notable long-term WW irrigation case study is the Mezquital Valley in central Mexico, where untreated wastewater (UWW) from Mexico City has been used for agricultural irrigation for over a century, resulting in sustained environmental exposure to a wide range of pharmaceuticals, heavy metals, and disinfectants among others (Dalkmann et al., 2012; Heyde et al., 2021; Lucho-Constantino et al., 2005; Siebe and Fischer, 1996; Siemens et al., 2008). Soils in the region exhibit high abundances of potentially harmful bacteria and elevated levels of ARGs and MGEs, including tetracycline (tetW, tetQ), streptomycin (aadA), sulfonamide (sul1, sul2), ciprofloxacin (qnrB and qnrS) resistance genes, class 1 integrons (intl1), and IncP-1 plasmid-backbone genes (korB) (Broszat et al., 2014; Broszat and Grohmann, 2017; Dalkmann et al., 2012; Jechalke et al., 2015; Lüneberg et al., 2018). With the operation of the Atotonilco WW treatment plant (WTTP) in 2017, a gradual shift from irrigation with UWW toward TWW is now underway.

The effects of this transition on ARG dynamics remain uncertain. In our previous soil microcosm study using three representative Mezquital Valley soils, relative ARG and MGE abundances were primarily driven by antibiotics and disinfectants in irrigation water, whereas soil type shaped bacterial community composition, with water quality showing no significant effect (Soufi et al., 2025). It remains unclear whether similar patterns would emerge in more complex soil-plant systems. Few studies have investigated the effects of WW irrigation on the occurrence and dissemination of ARGs and MGEs in soil-plant systems. Some have reported that water quality influenced the diversity, distribution and abundance of ARGs along the soil-plant continuum, typically showing a decline in ARG prevalence from soil to rhizosphere, phyllosphere, fruits and beans (Bhattacharjee et al., 2024; Cerqueira et al., 2019a, 2019b; Phan et al., 2024). Conversely, others have found minimal or no impact of WW irrigation on ARG levels both in soils and plants (Cerqueira et al., 2019c; Ofori et al., 2025), suggesting that other factors such as the crop type and agricultural management practices may play a more decisive role in determining ARG abundance.

In addition to increasing antimicrobial resistance, WW irrigation can disrupt natural soil-plant microbial communities and introduce human pathogens into plant microbiomes, posing significant risks to human health (Alegbeleye et al., 2018; Bhattacharjee et al., 2024). Several studies have demonstrated that WW irrigation influences the composition and diversity of root and plant associated bacterial communities (Bigott et al., 2022; Cerqueira et al., 2019a, 2019b; Man et al., 2020; Zolti et al., 2019). Cerqueira et al. (2019c), further observed that TWW had a smaller or negligible effect on soil and plant microbiomes compared to organic or fecal amendments

Building on our previous findings, we hypothesized that contaminants, i.e. antibiotics and disinfectants, partly released from soil due to the shift in irrigation water quality, drive the selection and spread of bacteria carrying ARGs and MGEs, with varying effects along the soil-plant continuum in a soil-type dependent manner. To test this, we conducted a greenhouse experiment using monolithic soil columns of two representative soil types (Leptosols and Vertisols) from the Mezquital Valley planted with cilantro (Coriandrum sativum) and irrigated for eight weeks with UWW and TWW spiked or not with a mixture of antibiotics and disinfectants. The influence of irrigation water quality on ARGs and MGEs dynamics was evaluated in total community (TC)-DNA using real-time qPCR, while the effects on the bacterial community composition in soil–plant systems were assessed by 16S rRNA gene amplicon sequencing analysis.

Bacterial 16S rRNA gene copy numbers ranged from 8.32 to 9.88 log₁₀ gene copies/L across the different water samples (Fig. S4). UWW exhibited slightly higher average counts (9.27 ± 0.46) than TWW (9.17 ± 0.51), but the difference was not statistically significant (p = 0.84; Wilcoxon test) (Fig. S4). All targeted ARGs and MGEs were detected in all WW samples, except the Staphylococcus pSK1 plasmid, found in one TWW sample (Fig. S4). The highest relative abundances were observed for sul1 and intI1 genes (Fig. 1). sul2, tetA, tetM genes, IncP-1 plasmids, and ermB genes were detected at lower relative abundance (Fig. 1). Unexpectedly, ARG and MGE distribution patterns showed no statistically significant differences between UWW and TWW samples (Fig. S5, PERMANOVA, R² = 0.20, p = 0.1053). Although all ARG and MGE relative abundances were consistently higher in UWW compared to TWW, these differences were not significant (p > 0.095; Wilcoxon test; Fig. 1).

In soil, bacterial 16S rRNA gene copy numbers were slightly higher in samples from stained soil (10.97 ± 0.31 log₁₀ 16S rRNA gene copies per g dry soil) compared to unstained soil (10.80 ± 0.16 log₁₀ 16S rRNA gene copies per g dry soil) (Fig. S6). However, this difference was not significant (p > 0.05; linear regression with bootstrap values) (Fig. S6). In stained soil, copy numbers were lower under spiked WW irrigation than unspiked, though not statistically significant (p > 0.05; linear regression with bootstrap values). In unstained soil, irrigation with different water qualities had no significant effect (p > 0.05; linear regression with bootstrap values) (Fig. S6). All analysed ARGs and MGE genes were detected across all soil samples, with varying relative abundances within the same treatment, likely due to the heterogeneous nature of the soil column cores. The highest relative abundances in stained and unstained soil were observed for tetA, sul1, sul2, intI1 genes, followed by IncP-1 plasmids, ermB, and ermA genes (Fig. 2 and Fig. S7A, B, C, and D). ARG and MGE profiles were significantly distinct between stained soil and unstained soil (Fig. S8; PERMANOVA, R² = 0.06, p = 0.003), with a significant influence of the spiking of the irrigation water (Fig. S8; PERMANOVA, R² = 0.09, p = 0.0001, Fig. 3A, PERMANOVA, R² = 0.12, p = 0.0021, and Fig. 3B, R² = 0.11, p = 0.0015). Specifically, samples from stained soil showed higher relative abundance of sul1 and sul2, intI1 genes, IncP-1, and Enterococcus plasmids compared to samples from unstained soil. Differences were significant for IncP-1 and Enterococcus plasmids (p = 0.0408 and p = 0.02018, respectively, linear regression with bootstrap values) (Fig. 2 and Fig. S7A, B, C, and D). Irrigation with spiked WW, either UWW or TWW, significantly increased the relative abundances of intI1, sul1, sul2, and tetA genes in all soil compartments (p = 0.01692, p = 0.01274, p = 0.04706, and p = 0.00022, respectively, linear regression with bootstrap values) (Fig. 2 and Fig. S7A, B, C, and D).

In the cilantro rhizosphere, the highest relative abundances of ARGs and MGEs were found in Leptosol soils for tetA, intI1, and sul2 genes (Fig. 2 and Fig. S7E and F). sul1, IncP-1 plasmids, and qnrA, qnrS genes were also found in high relative abundance in the rhizosphere of both soil types. The Staphylococcus plasmids pSK1 and pI258-like, as well as Enterococcus plasmids were either absent or only found in very few samples (Fig. 2 and Fig. S7E and F). ARG and MGE profiles in the rhizosphere were only influenced by the soil type (Fig. 3C, PERMANOVA, R²=0.38, p = 0.0001) with Leptosol columns exhibiting higher relative abundances of most studied ARGs and MGEs compared to Vertisol columns, except for the sul1 resistance gene and Inc18 plasmids (Fig. 2 and Fig. S7E and F). Spiked-WW irrigation significantly increased the relative abundance of tetA (p = 0.04944, linear regression with bootstrap), intI1 (p = 0.04588, linear regression with bootstrap), and sul2 genes (p = 0.00382, linear regression with bootstrap values) in the rhizosphere of both soil types (Fig. 2 and Fig. S7E and F).

In the phyllosphere, the highest relative abundance of ARGs and MGEs across the two soil types was observed for tetA, sul2, and sul1 genes, although high variability was found among the different samples. The Staphylococcus plasmids pSK1 and pI258-like were either detected in very few samples or not (Fig. 2 and Fig. S7G and H). Irrigation with spiked WW did not affect the overall ARG and MGE distribution patterns compared to unspiked WW (Fig. 3D, PERMANOVA, R²=0.05, p = 0.1325). However, spiked-WW irrigation led to a significant increase in the relative abundance of specific ARGs, namely ermA (p = 0.01688, linear regression with bootstrap values) and qnrA (p = 0.02168, linear regression with bootstrap values) in cilantro phyllosphere from both soil types (Fig. S7G and H).

First, the bacterial community composition of the different irrigation water types was characterized. Bacterial community compositions varied significantly between UWW and TWW samples, (Fig. S9, PERMANOVA, R²=0.36, p = 0.0081). The most prevalent bacterial phyla identified in both UWW and TWW samples were Proteobacteria (dominated by Gammaproteobacteria and Alphaproteobacteria), Bacteroidota (represented by Bacteroidia), Actinobacteriota (mainly represented by Actinobacteria), followed by Firmicutes (predominantly Clostridia), and Verrucomicrobiota (Fig. S10A and S9B). WW treatment significantly increased the relative abundance of the phylum Proteobacteria (p = 0.016; Wilcoxon test) while the relative abundance of the phyla Bacteroidota, Actinobacteriota, and Firmicutes tended to decrease but only significantly for Actinobacteriota (p = 0.016; Wilcoxon test). Nitrosomonas, Romboutsia, Simplicispira, Sediminibacterium, Leucobacter, Pedobacter, Reyranella, Acidovorax, and Microbacterium were among the 15 most abundant ASVs (Fig. S11). Among these, Leucobacter, an unclassified Alcaligenaceae, Acidovorax, and Microbacterium showed significantly higher relative abundances in UWW compared to TWW. In contrast, an unclassified Rhizobiales, Simplicispira (ASV135 and ASV159), and Reyranella had significantly higher relative abundances in TWW compared to UWW (p < 0.032; Wilcoxon test) (Fig. S11).

Given the distinct ARG and MGE profiles and higher levels in stained soils than in unstained soils, bacterial community composition was analysed only in soils exposed to preferential flow path water, rhizosphere, and phyllosphere. In stained soil samples, the bacterial communities were primarily shaped by soil type (Fig. 4A, PERMANOVA, R²=0.23, p = 0.0001), although the spike level also had a significant influence (Fig. 4A, R²=0.05, p = 0.0334). Unexpectedly, the type of the irrigation water did not have a significant effect (Fig. 4A, R²=0.04, p = 0.1118). ARG/MGE profiles were weakly but significantly correlated with Bray-Curtis distances of bacterial communities (Mantel rho = 0.2595, p = 0.0014). The bacterial phyla Proteobacteria and Actinobacteriota exhibited the highest relative abundance across all soil samples. Proteobacteria were higher in Leptosols irrigated with unspiked WW (50.48% ± 6.53%) than with spiked WW (39.33% ± 7.08%), although the differences were not significant (p > 0.05; Dunn´s test with Benjamini-Hochberg correction; Fig. S12A). Actinobacteriota relative abundance ranged from18% to 60% across both Leptosol and Vertisol samples. In contrast, other phyla such as Chloroflexi, Acidobacteriota, and Bacteroidota were present at lower relative abundances (below 16%) (Fig. S12A). Methylophilus, Nocardioides, Methylotenera, Gaiella, and Intrasporangium were among the most abundant genera in stained soils (Fig. S12A).

Rhizosphere bacterial communities were influenced by the soil type (Fig. 4B, PERMANOVA, R²=0.16, p = 0.0001), and the spiking to the irrigation water (Fig. 4B, R²=0.06, p = 0.0021). Water type (Fig. 4B, R²=0.05, p = 0.0143) had a modest yet statistically significant effect. However, the correlations between ARG/MGE profiles and rhizosphere bacterial community Bray-Curtis distances were not significant (Mantel rho = 0.02934, p = 0.3721). Proteobacteria dominated rhizosphere samples (relative abundance of 83% ± 7%), while Actinobacteriota, Bacteroidota, Chloroflexi, and Firmicutes were less abundant (< 1–23%) (Fig. S12B). Pseudomonas was the dominant genus, displaying significantly higher relative abundance in the rhizosphere samples from Leptosols (49% ± 9%) compared to Vertisols (40% ± 9%) (p = 0.0088; Wilcoxon test with Benjamini-Hochberg adjustment) (Fig. S13B).

In contrast to the rhizosphere and stained soil, phyllosphere bacterial communities showed no significant response to the soil type, spiking to the irrigation water or water type (soil type: R²=0.03, p = 0.8690; spike level: R²=0.03, p = 0.8135; water type: R²=0.04, p = 0.0951; Fig. 4C). These results were confirmed by the correlation between ARG/MGE profiles and the Bray-Curtis distances from phyllosphere bacterial communities, which was negative and non-significant (Mantel rho=-0.1097, p = 0.7619). The phyllosphere was dominated by members of the Proteobacteria phylum (64% ± 22%), and Firmicutes (34% ± 21%) (Fig.S12C) with Exiguobacterium, Pseudomonas, Pantoea and Acinetobacter among the most abundant genera (Fig. S13C).

To investigate whether the spiking of the irrigation water and water type influenced the relative abundance of specific ASVs from stained soil, rhizosphere, and phyllosphere, a logistic linear regression analysis was conducted (Fig. 5 and Table S2).

In stained soils, the relative abundance of ASVs affiliated with the class Alphaproteobacteria—six from the order Rhizobiales, one from Caulobacterales, two from the phylum Actinobacteriota (one Actinobacteria and one Thermoleophilia), and two from the class Bacteroidia (order Flavobacteriales) was significantly higher in soil columns irrigated with spiked WW compared to those irrigated with non-spiked WW. Conversely, multiple ASVs within Gammaproteobacteria—notably Methylibium, Methylotenera (family Methylophilaceae), and other Burkholderiales, as well as ASVs from Xanthomonadales, Diplorickettsiales, and two Alphaproteobacterial ASVs—Azospirillum and Hyphomicrobium were less abundant under spiked WW irrigation (Fig. 5 and Table S2).

With respect to the water type, only one ASV assigned to Methylotenera (Gammaproteobacteria) showed a significantly higher relative abundance in soil columns irrigated with UWW compared to TWW, while the relative abundances of several ASVs related to Pseudoxanthomonas, Chryseobacterium, Herpetosiphon, and Hyphomicrobium within the Gammaproteobacteria, Bacteroidia, Chloroflexia and Alphaproteobacteria, respectively, were significantly lower in UWW-irrigated columns (Fig. 5 and Table S2).

In rhizosphere samples, several ASVs affiliated with the class Gammaproteobacteria were differentially affected by spiked WW irrigation. Notably, within the order Enterobacterales, three ASVs showed significantly higher relative abundances after irrigation with spiked WW compared to non-spiked WW, whereas other ASVs showed significantly lower relative abundances. Within the order Pseudomonadales, two ASVs showed significantly higher relative abundances, while another was significantly lower. Several ASVs within the order Burkholderiales also increased following irrigation with spiked WW. Beyond Gammaproteobacteria, significantly higher relative abundances were also observed for ASVs affiliated with Bacteroidia and Polyangia. By contrast, ASVs from the class Alphaproteobacteria, Actinobacteria, Deinococci, and Bacilli exhibited significantly lower relative abundances under spiked WW irrigation (Fig. 5 and Table S2).

Regarding the effect of the water type on rhizosphere taxa, a similar trend was observed. Several gammaproteobacterial ASVs showed significantly higher relative abundances in rhizosphere samples irrigated with UWW compared to samples irrigated with TWW, whereas others exhibited significantly lower relative abundances in UWW-irrigated soils compared to TWW-irrigated soils. ASVs belonging to the Enterobacterales were significantly more abundant after the addition of UWW compared to TWW, while others were less abundant. Similarly, within the Alphaproteobacteria, one Rhizobiales showed significantly higher relative abundances after irrigation with UWW compared to TWW, while relative abundances of one Rhizobiales and two Caulobacterales were smaller in rhizosphere of UWW-irrigated rhizosphere soil than TWW-irrigated rhizosphere soil. In addition to the order Enterobacterales, several ASVs affiliated with the order Burkholderiales and Xanthomonadales, and one ASV belonging to the class Bacteroidia displayed significantly higher relative abundances after the addition of UWW compared to TWW. Conversely, ASVs affiliated to the order Pseudomonadales within the Gammaproteobacteria, one ASV belonging to the Actinobacteria and one ASV within the Chloroflexia exhibited significantly lower relative abundances after irrigation with UWW compared to TWW (Fig. 5 and Table S2).

In phyllosphere samples, no ASVs significantly responded to the different water qualities. However, one ASV affiliated to the genus Massilia, belonging to the order Burkholderiales within the Gammaproteobacteria, showed a significantly higher abundance in Vertisol soils compared to Leptosol soils (Table S2).

The study of plant-associated and soil microbiomes has attracted increasing scientific attention due to concerns that agricultural environments—particularly those irrigated with WW—may serve as reservoirs and transmission pathways for ARB, ARGs and MGEs (Larsson and Flach, 2021). The co-occurrence of antibiotics, heavy metals and disinfectants—alongside ARB harbouring ARGs—may favour ARG dissemination through MGEs and integrons in both bulk soil and rhizosphere, potentially enabling the colonization of edible plant tissues by bacteria carrying these elements (Becerra-Castro et al., 2015; Cerqueira et al., 2019a; Han et al., 2016; Piña et al., 2020; Zolti et al., 2019). In this context, our study evaluated the impact of antibiotics, disinfectants, and changes in irrigation water quality on the abundance and diversity of ARGs and MGEs, as well as on the bacterial community composition in the soil, rhizosphere, and phyllosphere of cilantro grown in monolithic soil columns under controlled greenhouse conditions.

Analysis of UWW and TWW samples showed a noticeable decrease in COD and TOC following treatment. In contrast, the concentrations of several investigated antibiotics and the total dissolved nitrogen remained unchanged or were even higher after WW treatment. The latter two phenomena are mainly attributed to the back-transformation of pharmaceutical metabolites—excreted by humans—into their original forms during wastewater treatment (Kumar et al., 2022), and to the mineralization of particulate organic N in UWW that was not captured by the analysis of N_tot since only dissolved nitrogen was assessed. Bacterial 16S rRNA gene copy numbers in UWW and TWW samples were comparable, with only slightly lower levels in TWW. A similar observation was reported in our previous study on influents and effluents from the same WWTP (Soufi et al., 2025) and has also been documented elsewhere (Oliveira et al., 2016; Rafraf et al., 2016; Rodríguez et al., 2021). The minor reduction in bacterial 16S rRNA gene copy numbers between the UWW and TWW can be attributed to the Atotonilco WWTP's operational strategy of maintaining nutrient concentrations in the TWW, which is beneficial for local agricultural use (Garduño-Jiménez et al., 2023). In contrast, other studies reported significant reductions in bacterial abundances following treatment (Pallares-Vega et al., 2019; Pärnänen et al., 2019). All analysed WW samples harboured the investigated ARGs and MGEs, except the Staphylococcus pSK1 resistance plasmid. Sulfonamide resistance gene sul1 exhibited the highest relative abundance, followed by class 1 integron intI1 gene, consistent with earlier reports (Soufi et al., 2025). The sul1 gene is frequently associated with the 3´end of class 1 integrons and commonly detected on diverse conjugative plasmids. Class 1 integrons foster plasmid diversification and are widely distributed in bacteria across various environments (Berendonk et al., 2015; Chen et al., 2015; Gillings et al., 2015; Haenelt et al., 2023; Heuer et al., 2011; Heuer and Smalla, 2012; Shintani et al., 2023). TWW showed only a slight reduction in ARGs and MGE levels compared to UWW, highlighting the limited effectiveness of conventional treatment in removing bacteria carrying these elements, which often remain in the final effluent (Rafraf et al., 2016; Rodríguez et al., 2021). The most abundant phylum in both UWW and TWW was Proteobacteria followed by Bacteroidota and Actinobacteriota, consistent with prior findings in WW studies (Azli et al., 2022; Li et al., 2023; Lira et al., 2020). The relative abundance of Proteobacteria was significantly higher in TWW as reported by Numberger et al. (2019) and Soufi et al. (2025), reflecting their ecological resilience and metabolic versatility in degrading organic matter (Atasoy et al., 2023; Tyagi et al., 2024; Yoo and Lee, 2021). In contrast, Bacteroidota, Actinobacteriota, and Firmicutes were less abundant, aligning with earlier findings (An et al., 2018).

In soils, bacterial 16S rRNA gene copy numbers were slightly higher in stained soil exposed to preferential flow path water compared to unstained soil, likely due to nutrients introduced with WW (Becerra-Castro et al., 2015; Mola et al., 2024). However, the bacterial 16S rRNA gene copy numbers were lower after irrigation with spiked WW, possibly due to antimicrobial effects of the spiked substances (Cleary et al., 2016; Katsivelou et al., 2023; Markowicz et al., 2021). ARG and MGE profiles differed significantly between stained and unstained soils, with stained soils exhibiting higher ARG abundances. This aligns with previous findings suggesting that flow pathways can serve as hotspots for the occurrence and accumulation of resistance genes in soil (Lüneberg et al., 2018). Spiking the irrigation water significantly increased ARGs and MGEs relative abundances across stained and unstained soil samples, consistent with our previous work (Soufi et al., 2025) and supporting prior research indicating that prolonged WW irrigation promotes ARG enrichment via selective pressure exerted by persistent micropollutants (Franklin et al., 2024; Jechalke et al., 2015; Kampouris et al., 2021).

Additional analysis on the soil bacterial community composition in stained soil revealed that soil type was the main influencing factor, in agreement with our previous findings (Soufi et al., 2025) and earlier studies (Krause et al., 2020; Mola et al., 2024; Obayomi et al., 2021). Furthermore, the irrigation with spiked water caused a significant but modest effect on the bacterial community. This observation was supported by the Mantel test, which revealed a weak yet significant correlation between soil bacterial community composition and ARG/MGE profiles—an effect also observed in our previous study (Soufi et al., 2025)—suggesting that shifts in ARG and MGE profiles may be caused by the soil bacterial community, underscoring the importance of sampling preferential water flow path soil (Kim et al., 2008; Lüneberg et al., 2018; Reichenberger et al., 2002; Salazar-Ledesma et al., 2018; Villamizar and Brown, 2017; Yinghu Zhang et al., 2018). In contrast, the irrigation with UWW or TWW had no significant impact on soil bacterial community composition. This finding contrasts with studies reporting pronounced microbial shifts due to irrigation water quality (Onalenna and Rahube, 2022), but is in line with previous research indicating that soil microbial communities may exhibit resilience to variations in irrigation water quality (Becerra-Castro et al., 2015; Lüneberg et al., 2018). Proteobacteria and Actinobacteriota were the most abundant soil bacterial phyla, aligning with previous reports on their prevalence in WW-irrigated environments (Gallego et al., 2021; Ibekwe et al., 2018; Xu et al., 2020). Proteobacteria are known for their role in spreading ARGs and MGEs (Hu et al., 2016), while Actinobacteria are recognized for their capacity to produce a wide array of secondary metabolites and for harbouring ARGs (D’Costa et al., 2011).

The effects of spiking UWW or TWW used for irrigation were further assessed for bacterial ASVs responding within stained soils. Several ASVs affiliated to different taxa (e.g. Rhizobiales, Actinobacteria, and Bacteroidia) were significantly more abundant in stained soil samples after irrigation with spiked WW compared to those irrigated with non-spiked WW. These taxa exhibit high metabolic versatility in degrading organic pollutants and tolerance to chemical stressors (Behera and Das, 2023; Izabel-shen et al., 2022; Porras-Socias et al., 2024; Teng et al., 2015; Wang et al., 2022; Zhang et al., 2017). Conversely, thirteen ASVs within Gammaproteobacteria, including four Methylibium (family Comamonadaceae) and three Methylophilaceae (two of which affiliated with Methylotenera), showed lower relative abundances, contrary to previous studies reporting their capacity for pollutant degradation (Li et al., 2024; Rissanen et al., 2017; Tiwari et al., 2019; Zhang et al., 2017). The observed reduced abundances suggest that specific taxa, particularly within Gammaproteobacteria may be more sensitive to micropollutant exposure (Gallego et al., 2021).

Irrigation with UWW led to a significantly higher relative abundance of an ASV affiliated with Methylotenera in stained soils compared to TWW. In our previous study, two Methylotenera ASVs—among the most abundant in both WW and soil—also increased following WW irrigation, suggesting they were likely introduced via the irrigation water or benefited from it. Members of the Methylophilaceae family have been associated with methanol-driven denitrification in WWTPs (Zhang et al., 2017), suggesting their proliferation in our study may be linked to substrate availability and nitrogen cycling under UWW irrigation. In contrast, four ASVs within the classes Gammaproteobacteria, Bacteroidia, Chloroflexia, and Alphaproteobacteria—affiliated with the genera Pseudoxanthomonas, Chryseobacterium, Herpetosiphon, and Hyphomicrobium, respectively—showed significantly lower relative abundances in stained soils due to irrigation with UWW compared to TWW. Pseudoxanthomonas, and Chryseobacterium, have previously been described in contaminated environments (Huda et al., 2022; Kundu et al., 2014; Patel et al., 2012; Szoboszlay et al., 2008; Velazquez-Meza et al., 2025) and are known to carry multiple ARGs and class 1 integrons (Fu et al., 2021; Kieffer et al., 2022; Klimkaitė et al., 2023; Selvaraj et al., 2022; Zhang et al., 2024). Hyphomicrobium, reported in soils and WW habitats (Huang et al., 2024; Vuilleumier et al., 2011), plays a key role in methanol and C1-compound degradation (Layton et al., 2000), while Herpetosiphon is known for bioactive compound production and occurs in soil and decomposing organic matter (Livingstone et al., 2018; Pan et al., 2017). Their lower relative abundances in soils irrigated with UWW than in those irrigated with TWW likely reflect competitive exclusion by fast-growing, copiotrophic taxa favoured under nutrient-rich UWW conditions (Koch, 2001).

In the cilantro rhizosphere, soil type was the only driver of ARG and MGE profiles, with Leptosol (sandy loam) exhibiting higher ARG levels—particularly tetA, sul2, and intI1—compared to Vertisol (clay), likely due to differences in soil texture and physicochemical properties influencing pollutants binding and sequestration. Marano et al. (2019) reported positive correlations between intI1 abundance in irrigation water and sandy loam soils, but not in clay-rich soils, supporting the idea that soil characteristics strongly influence ARG distribution through differences in the microbial community composition and micropollutant bioavailability (Jechalke et al., 2014; Song et al., 2024; Wang et al., 2020; Zeng et al., 2025; Zhang et al., 2021). In our study, irrigation water type (UWW vs. TWW) did not influence the abundance of ARG and MGE levels in the cilantro rhizosphere, consistent with Cerqueira et al. (2019c), but contrasting Bhattacharjee et al. (2024), who reported an enrichment of ARG diversity in the soil-plant-earthworm continuum following irrigation with municipal WW, whether spiked with antibiotics or not. Nevertheless, spiking the irrigation water further increased the abundance of tetA, sul2, and intI1 in the rhizosphere across both soil types, confirming that the presence of elevated concentrations of antibiotics and disinfectants in the environment promote ARG abundance and diversity (Phan et al., 2024). Similarly, Shen et al. (2019), reported higher intI1 occurrence in lettuce roots irrigated with pharmaceutical-contaminated water.

In our study, rhizosphere bacterial communities were predominantly shaped by soil type, with dominant phyla including Proteobacteria, Actinobacteria, and Bacteroidetes, and a high abundance of Pseudomonas consistent with previous studies (Berg and Smalla, 2009; Echeverry-Gallego et al., 2024; Kouadri, 2024; Philippot et al., 2013; Schreiter et al., 2014; Yuping Zhang et al., 2018). Elevated Pseudomonas abundance in the rhizosphere may reflect long-term WW irrigation. While soil type remained the dominant driver, spiking the irrigation water and water type also influenced the rhizosphere bacterial communities, though to a lesser extent. Water treatment was also reported to contribute to microbial shifts in the root-associated microbiome of tomato and lettuce (Zolti et al., 2019). These findings align with reports showing that pharmaceuticals in irrigation water can alter root microbiomes, including changes following irrigation with WW or fertilizer solution spiked with pharmaceuticals (Bhattacharjee et al., 2024; Bigott et al., 2022; Cerqueira et al., 2020; Man et al., 2020; Phan et al., 2024; Shen et al., 2019).

The impact of spiking irrigation water on rhizosphere bacterial ASVs was evaluated in the present study and revealed a significant enrichment of several gammaproteobacterial ASVs—affiliated with Citrobacter and Rheinheimera, Pseudomonas, and members of the Burkholderiales, including Achromobacter, Methylophilus, Variovorax, Acidovorax, and Pseudorhodoferax—as well as one additional ASV from Bacteroidia and one Polyangia, in response to irrigation with spiked WW. Several studies have documented that Citrobacter—widely detected in WW and treated effluents—and Rheinheimera species (both Enterobacterales) harbour multiple ARGs, often carried on conjugative plasmids, making them notable reservoirs and potential vectors of antimicrobial resistance in WW-irrigated contexts (Belachew et al., 2018; Fu et al., 2020b, 2020a; Hem et al., 2024; Ota et al., 2022; Tesfaye et al., 2019). Consistent with our results, Zolti et al. (2019) reported an increased abundance of Pseudomonadales in the rhizosphere of lettuce and tomato plants irrigated with TWW. Members of the order Burkholderiales have been identified as original environmental reservoirs of trimethoprim resistance genes (Kneis et al., 2024). Within this order, the genera Achromobacter, Methylophilus, and Variovorax—commonly found in contaminated environments—are known to degrade both antibiotics, aromatic compounds, and pesticides as well as to remove heavy metals in WW and biofilm reactors (Liang and Hu, 2021; Mahale et al., 2024; Ren et al., 2025; Sniegowski et al., 2011; Sun et al., 2022). Interestingly, ASV138, affiliated with Acidovorax—a member of Burkholderiales and previously detected among the 15 most abundant ASVs in UWW—also showed a significantly higher relative abundance in the rhizosphere, suggesting possible introduction via UWW irrigation. Additionally, several studies investigating soils amended with manure or irrigated with WW have reported increased abundance of Bacteroidetes (including Bacteroidia), a group frequently linked to the carriage of ARGs such as tetQ, ermF, and cfxA (Chen et al., 2019; Gatica and Cytryn, 2013; Niestępski et al., 2020; Zhang et al., 2019). To the best of our knowledge, this is the first study to report an increase in members of Polyangia and Pseudorhodoferax—a recently described member isolated from activated sludge (Bruland et al., 2009)—in rhizosphere of cilantro plants irrigated with WW.

In contrast, several ASVs showed a lower relative abundance in the rhizosphere after irrigation with spiked WW compared to those irrigated with non-spiked WW. Among those, Sphingomonas (order Sphingomonadales)—a genus widely distributed in nature and known to thrive in xenobiotic-contaminated environments and degrade various pollutants (Aulestia et al., 2021; Bai et al., 2016; Coronado et al., 2012; Mulla et al., 2016; Stolz, 2009; Yoon et al., 2009)—showed a marked decline in our study, likely reflecting strain-specific sensitivity to chemical stressors. While Cerqueira et al. (2020) reported no change in Sphingomonadales abundance after irrigation with antibiotic spiked water, our results showed a significant reduction in three ASVs affiliated with the genus Sphingomonas following irrigation with spiked WW. Similarly, Shen et al. (2019) observed reduced root bacterial diversity with antibiotic-contaminated irrigation water compared to those irrigated with fertilizer solution free of antibiotics, supporting the impact of such chemicals on microbial communities.

The most notable shifts in the rhizosphere bacterial communities following irrigation with UWW compared to TWW were increases in relative abundance of six ASVs affiliated to Enterobacterales, Burkholderiales, Xanthomonadales, Flavobacteriales, and Rhizobiales. Notably, ASV857, assigned to the genus Acidovorax within Burkholderiales, exhibited significantly higher relative abundances in rhizosphere due to irrigation with UWW. This might be relevant, as a metagenome-assembled genome (MAG) of Acidovorax from spinach rhizosphere irrigated with TWW carried a MexJK-OprM efflux pump, conferring resistance to β-lactams, sulfonamides, and other antibiotics (Bhattacharjee et al., 2024).

Several other ASVs showed significantly lower relative abundances in the rhizosphere following irrigation with UWW than in those irrigated with TWW, including ASVs from the orders Enterobacterales, Pseudomonadales, Caulobacterales, Rhizobiales, Chloroflexales, and Pseudonocardiales. Despite the high abundance of Enterobacterales and Pseudomonadales in WW (Azli et al., 2022; Numberger et al., 2019), their lower relative abundances in the rhizosphere likely reflect selective chemical pressures and intense microbial competition (Lagos et al., 2015). These results align with previous research demonstrating that irrigation water quality significantly shapes the rhizosphere bacterial community, with TWW irrigation promoting Actinobacteria decline and Gammaproteobacteria enrichment (Zolti et al., 2019).

In cilantro phyllosphere samples, all analysed ARGs and MGEs were detected, with exception of the pSK1- and pI258-like plasmids, which were either rare or below the detection limit. In contrast, Bhattacharjee et al. (2024) did not recover any ARG in MAGs from the phyllosphere of spinach and radish plants irrigated with municipal WW containing antibiotics using a metagenomic approach, highlighting the importance of using complementary cultivation methods or qPCR to detect ARGs present in low-abundance bacteria or plasmid-associated ARGs that may be missed by metagenome analyses. Overall, spiking, WW type, and soil type did not significantly impact the ARG and MGE profiles in phyllosphere of cilantro. These results align with previous findings by Cerqueira et al. (2019c), who reported minimal influence of irrigation water quality on ARG abundance in plants. However, in our study, spiking the irrigation water led to increased relative abundances of some of the targeted genes, notably ermA and qnrA. Similarly, Shen et al. (2019) observed that ARG profiles were responsive to pharmaceutical exposure, though the patterns were not consistent. The detection of ARGs on edible plants raises potential, though still uncertain, public health concerns related to human exposure through consumption (Blau et al., 2018; Reid et al., 2020; Zhang et al., 2022).

The present study showed that the bacterial community composition in the phyllosphere was not significantly influenced by soil type, the spiking of the irrigation water or WW type. Our findings are consistent with previous studies reporting little to no influence of irrigation water quality on plant-associated microbiome composition (Cerqueira et al., 2019a) and diversity (Phan et al., 2024). The dominant phyla identified were Proteobacteria and Firmicutes, followed by Actinobacteriota, and Bacteroidota (Sohrabi et al., 2023). Among the dominant genera were Exiguobacterium, Pseudomonas, Pantoea and Acinetobacter—bacteria known for producing diverse antimicrobial secondary metabolites that promote plant health and enhance nutrient cycling (Alattas et al., 2024; Lv et al., 2022; Marfetán et al., 2023; Mujumdar et al., 2023; Raio and Puopolo, 2021). However, some species within these genera are also recognized as important plant pathogens (Coutinho and Venter, 2009; Xin et al., 2018) and human opportunistic pathogens (Antunes et al., 2014; Chen et al., 2017; Dutkiewicz et al., 2016; Sanz-García et al., 2021), posing potential risks for plant disease and human opportunistic infections (Ma et al., 2025).

In phyllosphere samples, none of the ASVs showed a significant response to the irrigation with spiked WW or water type. However, one ASV affiliated with the genus Massilia, responded significantly to the soil type, showing higher relative abundance in the phyllosphere of cilantro grown in Vertisol soils compared to Leptosol soils. This genus was also detected in the study of Bhattacharjee et al. (2024), where a carbapenem resistance gene was detected in a MAG belonging to the genus Massilia and assembled from metagenomes of bulk soil planted with spinach and irrigated with treated municipal WW.

A multivariable approach was employed to assess how the presence of antibiotics and disinfectants in WW and a shift in irrigation regime—from UWW to TWW—affects the dynamics of ARGs, MGEs, and bacterial communities within a soil–plant system, using monolithic soil columns cultivated with cilantro. Our results demonstrate that spiking the irrigation water with antibiotics and disinfectants significantly increased the relative abundances of several ARGs and MGEs across all soil types, as well as in the cilantro rhizosphere and phyllosphere. Spiking emerged as the most important driver shaping ARG and MGE distribution in soils, indicating that micropollutants in irrigation water play a critical role in determining the antibiotic resistance profile and dynamics along the soil–plant continuum. Contrary to our hypothesis, irrigation with either TWW or UWW did not significantly alter ARG or MGE abundances in the soil, cilantro rhizosphere, or phyllosphere. While the soil type was the primary factor influencing the bacterial communities of both stained soil and cilantro rhizosphere, spiking the irrigation water also had a significant effect. In contrast, water type alone influenced only the rhizosphere bacterial community. Irrigation with different water qualities led to significant changes in specific members of the classes Alphaproteobacteria and Gammaproteobacteria. These findings, obtained under controlled greenhouse conditions over a short duration, provide valuable insights into the effects of irrigation water quality and WW-borne pollutants on the dissemination and persistence of ARGs in plant–soil systems. They also offer a scientific basis for large-scale field experiments under real-world agricultural scenarios. Our results emphasize the importance of managing WW quality and implementing advanced treatment technologies to reduce the spread of ARGs in agricultural soils, thereby supporting more sustainable irrigation practices. Future research involving chemical analysis of pollutant bioaccumulation in soil and plants, along with cultivation-based isolation and characterization of ARBs and plasmid capture experiments, will contribute to further elucidate the role of WW pollutants in the spread of environmental antibiotic resistance within soil–plant systems.