2.1 Experimental design

The greenhouse pot experiment was maintained for eight months. The soil was collected from the topsoil layer (0-15 cm depth) of a field in the Faculty of Agronomy (34°36'48" S, 56°12'54" W). The soil was a Mollisol with 5.0%, 70.9%, and 24.1% of sand, silt, and clay, respectively. The soil pH was 6.4, and the exchangeable cation content was K 2.0, Ca 14.0, Mg 3.1, and Na 0.3 cmolc kg^-1.

Pots containing 4 kg of homogenized soil were sown withFestuca arundinaceacv. Tacuabé (four plants per pot), which seeds were supplied by the National Seeds Institute (INASE). The experiment had a randomized block design with three replicates, according to Illarze⁴. The N fertilizer treatments included four applications to the soil at an equivalent rate of 50 kg N ha^-1, i.e., 20 mg N kg of pot soil^-1 consisting of (1) two-lagoon stabilized dairy effluent (LDE), (2) raw dairy effluent (RDE), (3) synthetic nitrogen fertilizer (urea), and (4) a non-amended soil (control). All applications were performed in solution, adjusting the same final volume per pot, and the same amount of tap water (pH 6.9, electric conductivity (EC) 2 mS cm^-1) was added to the control. The first application was at seeding. Consecutive fertilization/harvesting cycles of 45 days were performed three times, simulating the typical management of tall fescue forage cuts. The final cut was done two months after the last application of treatments. The tall fescue forage was cut manually at 5 cm height.

Dairy effluents were collected from the dairy farm of Centro Regional Sur from the Faculty of Agronomy, with geographic coordinates 34°36'47.83" S and 56°12'54.00" W. The collection of effluents and their characterization is described in Illarze⁴. Although variable in composition throughout the year, dairy effluents had a total solid content lower than 2% and suspended solids less than 1%. The organic carbon (OC) content of RDE was higher than that of LDE except in the January samples⁴. The RDE applied in the experiment presented an average pH of 8.0, 1325 mg L^-1 of total carbon, 371 mg L^-1 of total N, and 98 mg L^-1 of total ammoniacal N, and LDE had an average pH 8.4, 425 mg L^-1 of total carbon, 194 mg L^-1 of total N, and 63 mg L^-1 of total ammoniacal N.

The selected farm followed the usual production practices in the country and used beta-lactam antibiotic agents to treat bovine mastitis.

2.2 Chemical soil analysis

At the end of the experiment, after two months of the fourth application of DE, soil samples were collected from each pot, and chemical properties were determined. The soil was dried at 45 °C until a constant weight was reached. The total exchangeable bases, Ca, Mg, K, and Na, were extracted with ammonium acetate buffered at pH 7¹⁶. The K and Na were determined by atomic emission spectrometry, and Ca and Mg by absorption spectrometry. The available P content was determined by the Bray 1 method¹⁷.

2.3 Soil basal respiration

Soil samples were collected immediately after and at 7, 30, and 45 days from DE applications and stored at 5 °C before analysis.

The soil basal respiration was measured by titration according to Öhlinger¹⁸ with slight modifications. 15 g of soil were incubated in a closed vessel for 10 days at 25 °C in the dark. The CO₂ was trapped in NaOH solution (0.25 M) and titrated against 0.1 HCl after freshly prepared saturated BaCl₂ (0.5 M) was added. The respiration rate was expressed as mg of CO₂ g^-1 dry soil d^-1. Cumulative basal respiration until 45 days of each application date was calculated by adding all mg CO₂ g^-1 dry soil d^-1 mean values for each sampling date multiplied by the number of days between these dates.

2.4 Chemical analysis of tall fescue

The aboveground biomass obtained at the final herbage cut was dried at 65 °C for 48-72 h (until constant weight) and ground to pass through a 0.5-mm mesh. After that, the total N content was determined by the Kjeldahl method¹⁹. P, K, Ca, Mg, and Na were extracted with dilute HCl (20%) in samples calcined at 550 °C for 5 h. The molybdenum blue method was used to determine the total P with ascorbic acid²⁰. The Ca and Mg elements were determined by atomic absorption spectrometry, and K and Na by atomic emission spectrometry²¹.

2.5 Quantification of pathogenic bacterial indicators

Total coliforms and E. coli numbers were determined in DE by a ten-fold dilution series of 10 mL of effluent in 90 mL of 0.01 M phosphate buffer, followed by agitation for 30 min in an orbital shaker at 200 rpm. The diluted samples were plated in triplicate on 3M™ E. coli/ Coliform Petrifilm™. After incubation at 37 °C for 24 h, characteristic blue colonies were counted as E. coli and red as total coliforms (CFU/100 mL).

Total coliforms and E. coli numbers were determined in the soil immediately after DE application and 45 days after each application, and expressed as log₁₀ CFU g^-1 dry soil. For data analysis, the number of viable bacteria below the detection limit (10 CFU g^-1 soil) was set at 1 log CFU g^-1 dry soil. The E. coli and total coliforms survival results were calculated as C/Co values, where Co is the applied concentration in the irrigation solution and C is the concentration in the soil after 45 days of each application of dairy effluents.

2.6 DNA extraction and quantification of ARGs

Total DNA was extracted from the dairy effluents and soil samples using the Qiagen DNeasyPowerSoil Pro Kit (Qiagen, Carlsbad, CA, USA). For soil, 0.25 g samples were processed according to the manufacturer’s instructions. In dairy effluent samples, 50 mL were centrifuged in 50 mL Falcon tubes at 10,000 × g for 10 min at room temperature to obtain pellets before bead beating.

The DNA concentration and purity were determined with NanoDrop® 2000c UV-vis spectrophotometry (Thermo Scientific, USA). The extracted DNA samples were stored at −20 °C before PCR amplification.

Three beta-lactam resistance genes (blaTEM, blaOXA, and blaSHV) were chosen to assess the presence of the genes in the samples and their quantification.

Primers for each gene were those published in Dallenne and others²². For blaTEM: forward primer 5’-CAT TTC CGT GTC GCC CTT ATT C-3’ and reverse primer 5’- CGT TCA TCC ATA GTT GCC TGA C-3’; for blaOXA: forward primer 5’-GGC ACC AGA TTC AAC TTT CAA G-3’ and reverse primer 5’-GAC CCC AAG TTT CCT GTA AGT G-3’; for blaSHV: forward primer 5’-AGC CGC TTG AGC AAA TTA AAC-3’ and reverse primer 5’-ATC CCG CAG ATA AAT CAC CAC-3’.

Purified PCR products for each of the genes analyzed were used as templates for the calibration curves. The calibration curves consisted of five ten-fold dilutions of the template; each dilution point was amplified by triplicate. The DNA concentration of the templates was obtained by spectrofluorometry (QuBit^TM, ThermoFischer).

The qPCRs for each gene were run separately and not in a multiplex fashion. The Maxima SYBR Green qPCR Master Mix was used (ThermoFischer). The mix for each qPCR assay contained 2 μL of the extracted DNA and 10 mM of each primer. Runs were performed on a Rotor-Gene-Q Pure Detection machine manufactured by QIAGEN. The cycling conditions were an initial denaturing step of 10 minutes at 95 °C, followed by 40 cycles of 15 seconds at 95 °C, 20 seconds at 60 °C, and 20 seconds at 72 °C; the fluorescence was measured at the end of each cycle. A melting curve was performed at the end of the run beginning fluorescence measurement at 72 °C and incrementing 1 °C in each step until 95 °C.

2.7 Statistical analysis

One-way analysis of variance (ANOVA) followed by Fisher’s least significant difference (LSD) test was used to examine the effects of treatments on the soil chemical composition, plant yield and nutrition, bacterial indicators, and the abundance of ARGs. Each parameter was tested for normality of distribution and variance homogeneity using Shapiro-Wilk’s test and Levene’s test, respectively. When transformed data, such as pH, was also not normal, it was analyzed using the non-parametric Krustal-Wallis test followed by a Wilcoxon Rank Sum test to determine significant differences between treatments.

Data were expressed as means ± standard error. All statistical analyses were performed using Infostat²³, and the significance level was considered at p<0.05.

3.1 Effect of DE applications on soil chemical properties and plant biomass and nutrition

In general, the soil application of organic amendments like DE increases its fertility, especially N, P, and K ^{_{24)(25)26)(27)} 28}. In this pot trial, with a N fertilization dose equivalent to an annual 200 kg N ha^-1, the content of exchangeable cations, such as Na, K, and Mg, increased in the soil after LDE applications relative to the control (Table 1). Additionally, Na and NO₃ ^--N content increased in soil with RDE applications relative to the control soil (Table 1).

High concentrations of K present a challenge in achieving an agronomic balance of nutrients when applying effluent to land²⁹, potentially causing an imbalance in Mg and Ca in animals. A study on the land application of two types of DE demonstrated that both plant biomass and soil K levels increased under effluent treatments, as the amount of K applied in DE exceeded tall fescue maintenance requirements at rates above 100 kg N/ha/year³⁰. The increase of K levels in soils after the addition of LDE may then pose a problem for land application; however, in the current case, the nutrient content in the plant biomass with applications of DE did not exceed that of the control or the urea treatments (Table 2). The grass stagger index (GSI = (K⁺)/((Ca²⁺) + (Mg²⁺))), used as an index of the nutritive quality of forage, was very low and never exceeded 2.2, pointed as the risk of hypomagnesemia³¹.

Long-term application of effluents may lead to the accumulation of exchangeable Na in the soil. Although monovalent cations are generally less strongly held on cation exchange sites than divalent ones and do not pose a risk to the receiving soil, a decrease in exchangeable Ca and Mg has been reported³¹. This effect cannot be detected in this pot trial but must be considered in the field.

While this study focused on N, dairy manure also contained high levels of P, although the final soil P content was not modified by DE addition (Table 1). In this case, the high nutrient status of the soil used for this pot trial could have masked the effect of DE nutrient input on the soil.

Dairy effluents typically exhibit alkalinity, and numerous works indicate that effluent irrigation increases soil pH in acid soils³². In this study, the soil pH was close to neutrality, and the application of DE did not affect soil pH relative to the control soil (Table 1). Conversely, urea fertilization led to a decrease in soil pH. This acidification effect due to excessive use of synthetic N fertilizer has previously been reported³³.

The percentage of OC input from RDE generally surpassed that from LDE⁴, and the application of DE contributes to a long-term increase in the soil OC content³². Even though our results did not reveal an increase in soil OC content (Table 1), the cumulative basal respiration at the last application of LDE was significantly higher. There was a similar tendency for RDE in the preceding two applications (Figure 1).

Figure 1: Cumulative basal respiration of soil with the application of raw dairy effluent (RDE) or lagoon dairy effluent (LDE), urea, and non-amended soil at the four-application events. Columns represent means ± standard errors (n=3). Columns with different letters indicate a significant difference between treatments on each sampling date (P<0.05) 

The cumulative basal respiration of the soil, considering all four applications in the control and urea treatment, averaged less than 40 mg CO₂ g^-1 dry soil, whereas in the LDE and RDE treatments, it ranged from 46 to 50 mg CO₂ g^-1 dry soil. This finding aligns with Liu and others³⁴, who reported that soil basal respiration with organic amendments like farmyard manure and green manure was higher than with NPK treatment.

Soil respiration, widely employed to quantify microbial activities in soil ^_35)(36) , generally exhibits a positive correlation with soil OM content³⁶. In accordance, the application of these two DE in the same pot experiment showed an enhancement in microbial activity⁴.

The application of organic materials with a low dry matter content, as observed in our case, resulted in few measurable changes in soil properties in the short term³⁷ and may induce a priming effect, leading to the mineralization of native soil OM and the potential decrease of soil OC³⁸. Therefore, the increase in soil basal respiration obtained in this study may be attributed to the priming effect induced by the OC input from DE.

The tall fescue yield at the fourth cut of the experiment was higher for RDE than for LDE application (Table 2). However, neither the N content of the plant biomass nor the N uptake after repeated applications of DE differed from urea fertilization or the control (Table 2). In this same pot trial, the authors of the present study verified that the repeated applications of these DE did not result in detectable changes in N uptake or yield (sum of four herbage cuts)⁴. Furthermore, this study evaluated the chemical composition of other crucial macronutrients in plant biomass that showed similar levels between treatments. These findings once again highlight the basal nutrient status of the soil as the factor responsible for masking the effect of DE or urea fertilization on plant nutrient content and yield.

3.2 Bacterial indicators of soils with DE applications

By determining the number of pathogenic bacterial indicators in DE and assessing the spread and survival in the soil, we aimed to estimate the health risks associated with the land disposal of DE.

Previous research has demonstrated that the number of fecal bacteria in manure slurries can be reduced through long-term storage³⁹. Consistent with this, our current study revealed that the total coliforms and E. coli numbers were higher for RDE than for LDE, except for the September sampling date (Figure 2). The difference in total coliform number between both effluents reached its maximum in November (2.0E+08 vs 1.7E+06), and for E. coli in January and March (on average 3.5E+07 vs 2.0E+05). These results indicate a substantial removal of enteric bacterial pathogens through lagoon storage.

Figure 2: Numbers of total coliforms (a) and E. coli (b) in raw dairy effluent (RDE) or lagoon dairy effluent (LDE) at the four-sampling dates. Columns represent means ± standard errors (n=3). Columns with different letters indicate a significant difference between dairy effluent types at each sampling date (p<0.05) 

The seasonal variability in the chemical composition of both DE has been reported in our previous work⁴. For the September sampling (end of the cool season), RDE did not differ from LDE in dry matter content and recorded lower total C than in the other sampling dates. This may account for the lack of differences in the number of photogenic bacterial indicators at that date.

Upon the application of DE to the soil, significantly higher E. coli numbers were accounted for RDE-treated soil compared to the control and urea treatments (Figure 3).

Figure 3: Average number of total coliforms (a) and E. coli (b) in soil with the application of raw dairy effluent (RDE) or lagoon dairy effluent (LDE), urea or control, immediately after and 45 days from applications. Columns represent means ± standard errors (n=3). Columns with different letters indicate a significant difference between treatments for each day (P<0.05) 

After 45 days from its application, E. coli number decreased by nearly two orders of magnitude (from 2.5E+04 to 2.3E+02), yet it remained significantly higher than in the other treatments (Figure 3). On the other hand, the count of total coliforms did not rise at the time of DE applications and remained nearly constant after 45 days of applications. Nevertheless, soil treated with RDE had a significantly higher number of total coliforms than the control after 45 days of DE applications (Figure 3).

Table 3 shows the survival of E. coli introduced with the effluents in the soil after 45 days of each application. E. coli was undetectable after repeated LDE applications but was counted in the pots treated with RDE. No differences were determined in the survival of total coliforms for both DE applications (Table 3). Several studies have indicated that coliforms can survive and proliferate in extraintestinal environments, and their role as indicators has been questioned⁴⁰.

Table 3: Survival of E. coli and total coliforms in fescue pot soil with the application of raw (RDE) or lagoon dairy effluent (LDE) for each of the four applications. Ratio of number of pathogenic bacteria indicators after 45 days of application to number applied with effluents 

Results from a field lysimeter study showed that the land application of the treated effluent can lead to significant reductions in E. coli leaching compared with the untreated original DE⁴¹.

It has been established that dairy effluent application to soil generally promotes the numbers and survival of enteropathogenic bacteria ^_30)(42) . Our results suggest a low-level persistence of enteric bacteria following RDE application to the soil, indicating that more than 45 days are required for the complete elimination of indicator bacteria.

In Uruguay, a wait time after DE application of 21-30 days before grazing is recommended, and grazing is advised to be avoided with categories of less than 1-year-old and pre-calving cows⁴³. The present results confirm the pertinence of this recommendation.

3.3 Occurrence of ARGs in DE and amended soil

The fate of ARGs spread after DE applications to soil was assessed by targeting three beta-lactam ARGs. Resistance to beta-lactam antibiotics is an increasing concern, and beta-lactamase production is the most common mechanism for this drug resistance²².

A growing number of cases of multidrug-resistant bacteria producing beta-lactamases have been reported in Uruguayan dairy farms ^{_14)(44) 45}. Numerous studies have proved that DE serves as a source of ARGs and may play a key role in the dissemination to the soil ^{_46)(47) 48}.

The presence of blaTEM and blaOXA was detected in at least one sampling date in RDE or LDE (Table 4), while blaSHV was not detected. However, neither of the beta-lactam ARGs could be later observed in the DE-amended soils at the end of the experiment (data not shown). These results indicate that DE applications did not lead to the dissemination of these ARGs in the soil. The decline over time following applications suggests that microbes possessing beta-lactam ARG had somewhat limited survival potential in the soil environment⁴⁶, probably due to their lack of adaptation to the soil characteristics and the inhibition by soil indigenous bacteria ^_49)(50) .

Table 4: Abundance of blaTEM and blaOXA genes in raw (RDE) or lagoon dairy effluent (LDE) at four-sampling dates (mean n=3 ± S.E) 

n.d = not detected

In this study, the blaTEM gene was present in November and March in RDE samples, and in LDE it was also detected in September (Table 4). On average, the blaTEM abundance was 3.5 × 10³ and 8.8 × 10² copies mL⁻¹ for LDE and RDE, respectively, with no significant differences. The blaOXA was only present in November in RDE samples, and its abundance was not significantly different from LDE. On average, the blaOXA abundance was 1.6 × 10³ and 2.1 × 10³ copies mL⁻¹ for LDE and RDE, respectively.

Neither of the beta-lactam ARGs was detected in the DE in January, the warmest sampling date (Table 4). According to these findings, Schages and others⁵¹ reported a seasonal effect of ARG abundance in wastewater, with ARG abundance negatively correlated with warmer temperatures. Furthermore, the authors found that antibiotic-resistant bacteria were also more prevalent in the colder seasons.

The occurrence of both beta-lactam ARGs was more frequent in LDE than in RDE. This was expected, as the RDE sampling represents one day flush of the dairy barn floor, while the LDE sampling represents annual dairy effluents disposal, where excreted antibiotics can continue exerting selection pressure for antibiotic-resistant genes⁵². So, lagoon storage does not remove ARGs in this case, although qPCR does not provide information on the expression of these genes⁵³.

Further studies are mandatory to determine the presence of other ARGs that confer resistance to other antibiotics used in the country to treat mastitis. However, our initial work is promising regarding the release of DE to pastures where the beta-lactam ARGs could not be detected.