ABSTRACT
The study assessed the effect of soil slope on Mycobacterium avium subsp. paratuberculosis transport into rainwater runoff from agricultural soil after application of M. avium subsp. paratuberculosis-contaminated slurry. Under field conditions, 24 plots of undisturbed loamy soil 1 by 2 m2 were placed on platforms. Twelve plots were used for water runoff: 6 plots at a 3% slope and 6 plots at a 15% slope. Half of the plots of each slope were treated with M. avium subsp. paratuberculosis-contaminated slurry, and half were not treated. Using the same experimental design, 12 plots were established for soil sampling on a monthly basis using the same spiked slurry application and soil slopes. Runoff following natural rainfall was collected and analyzed for M. avium subsp. paratuberculosis, coliforms, and turbidity. M. avium subsp. paratuberculosis was detected in runoff from all plots treated with contaminated slurry and one control plot. A higher slope (15%) increased the likelihood of M. avium subsp. paratuberculosis detection but did not affect the likelihood of finding coliforms. Daily rainfall increased the likelihood that runoff would have coliforms and the coliform concentration, but it decreased the M. avium subsp. paratuberculosis concentration in the runoff. When there was no runoff, rain was associated with increased M. avium subsp. paratuberculosis concentrations. Coliform counts in runoff were related to runoff turbidity. M. avium subsp. paratuberculosis presence/absence, however, was related to turbidity. Study duration decreased bacterial detection and concentration. These findings demonstrate the high likelihood that M. avium subsp. paratuberculosis in slurry spread on pastures will contaminate water runoff, particularly during seasons with high rainfall. M. avium subsp. paratuberculosis contamination of water has potential consequences for both animal and human health.
INTRODUCTION
Mycobacterium avium subsp. paratuberculosis is the causative agent of paratuberculosis, or Johne's disease, in domestic and wild ruminants (1). Johne's disease is an economically important disease characterized by chronic intestinal inflammation, diarrhea, progressive weight loss, emaciation, and death (2, 3). M. avium subsp. paratuberculosis is consistently found in people with Crohn's disease, suggesting that this agent is potentially zoonotic (4). The first report of M. avium subsp. paratuberculosis in Chilean cattle was in 1958 (5). At present, roughly 40% of Chilean dairy herds are considered infected (6).
The main transmission route for ruminants is by ingestion of food or water contaminated with feces from infected animals, including ingestion of M. avium subsp. paratuberculosis from contaminated pastures (3). Once M. avium subsp. paratuberculosis is shed in feces from the host, it is capable of surviving for prolonged periods in the environment (7). When the organisms reach the soil surface after slurry application, they interact in complex ways with the soil matrix (13).
As in other countries, dairy farming has intensified in Chile. Along with this, slurry application to pastures has gained popularity (8). Slurry is defined as a mixture of livestock feces and urine diluted with rainfall or facility cleaning water and may contain animal bedding and feed. It is applied to agricultural soils to enrich the soil, improve soil structure, and increase soil buffering capacity and biological activity due to its high content of organic matter and nutrients (8). Slurry application can potentially spread a variety of bacterial, viral, and protozoan animal pathogens which may persist and to which susceptible grazing animals may be exposed (9, 10, 11). J�rgensen reported that M. avium subsp. paratuberculosis survived in slurry pits from cattle farms for up to 98 days at 15�C (12); however, little is known about M. avium subsp. paratuberculosis survival after slurry application to soil.
The first study under field conditions on the fate of M. avium subsp. paratuberculosis in agricultural soils after application of contaminated slurry (13) confirmed that the bacterium tends to remain on the grass and soil surface layers, moving very slowly through deeper soil layers. From that work it seems that M. avium subsp. paratuberculosis is likely to move across the soil surface with rainfall runoff and potentially contaminate surface water. Local conditions of rainfall and ground slope as well as soil type could influence the rate of M. avium subsp. paratuberculosis movement. Different methods of slurry application to soils, such as slurry injection, could mitigate this potential problem. The present study aimed to evaluate the effect of rainfall intensity and soil slope on M. avium subsp. paratuberculosis contamination of runoff water after application of M. avium subsp. paratuberculosis-contaminated slurry to grassland soil.
MATERIALS AND METHODS
Experimental design.An experiment conducted under field conditions using 24 plots of undisturbed loamy soil was carried out at a farm located in the Lagos region of Chile from 1 June to 1 October 2011. Soil plots mounted at two different inclines were artificially contaminated with M. avium subsp. paratuberculosis-spiked liquid bovine manure slurry. Water runoff and soil core samples were collected to assess the movement of M. avium subsp. paratuberculosis into water runoff and persistence in soil over time. Concurrent climate data were also collected.
Soil plots.The pasture where soil was dug up had not been grazed by livestock for at least 4 years. Plots were obtained by digging around the perimeter of a 1-m by 2-m plot to a depth of 20 cm with a shovel. Each soil plot was gently removed from the pasture, placed on a sheet of plywood, and moved to the experimental platform at the study site (Fig. 1). Soil plots were mounted on wooden platforms which were angled to obtain a slope of 3% or 15%. The plots were exposed to ambient temperatures and rainfall. All water runoff was collected in a trough mounted at the bottom edge of the slope and directed to 1.0-liter polyethylene containers. All materials used were new but not sterilized.
Soil plot experimental platform. (A) Diagram of the platform and soil plot with key components labeled; (B) photograph of the platforms with soil plots.
Treatment groups.Twelve plots were used for water runoff sampling: 6 plots at 3% slope and 6 plots at 15% slope. Half of the plots of each slope were treated with M. avium subsp. paratuberculosis-contaminated slurry and half were not treated, allowing triplicates for each slope-slurry treatment combination (Fig. 2A). In parallel, using the same experimental design, 12 plots were established for soil sampling on a monthly basis using the same spiked slurry application and soil slopes (Fig. 2B). Prior to excavation of each soil plot, a number was drawn out of a bag to assign that plot an experimental location (1 to 24). As slurry and slope were predetermined by location, all plots were therefore randomized to treatment combinations.
Experimental design. (A) Soil plots for runoff water collection; (B) soil plots for soil sampling. P, plot number; S, plots with slurry applied (black); WS, plots without slurry applied (gray); percentages are soil slopes.
Preparation and quantification of the M. avium subsp. paratuberculosis inoculum.M. avium subsp. paratuberculosis ATCC 19698 was incubated in Middlebrook 7H9 liquid medium supplemented with 10% oleic acid, albumin, dextrose, and catalase (OADC; BD Diagnostics, Franklin, NJ) and 2 mg/liter of mycobactin J (Allied Monitor, Fayette, MO) for 1 month at 37�C in 250-ml ventilated culture flasks containing 40 ml medium. On the basis of weekly spectrophotometer readings, cultures were harvested when they were in exponential growth and had an optical density at 600 nm of 1.00, equating to roughly 108 CFU/ml of M. avium subsp. paratuberculosis (14).
M. avium subsp. paratuberculosis-contaminated slurry.Slurry was collected from the manure storage lagoon of a local commercial dairy farm. The slurry was culture negative for M. avium subsp. paratuberculosis prior to spiking. The slurry (168 liters) was harvested and then spiked with a liquid culture of M. avium subsp. paratuberculosis (ATCC 19698) and thoroughly mixed to achieve an estimated final concentration of at least 106 M. avium subsp. paratuberculosis bacteria/ml of slurry.
Slurry application.Using a recommended slurry application rate of 70,000 liters per hectare, each 1- by 2-m plot was irrigated with 14 liters of M. avium subsp. paratuberculosis-spiked slurry (4% dry matter) using a manual sprinkler to evenly apply the slurry across the surface of each soil plot. Control plots received no slurry application.
Water runoff collection.After a natural rainfall event, water moving along the soil surface (water runoff) dropped from the edge of the soil plot into a polyvinyl chloride (PVC) trough located on the lower (downhill) side of the plot (Fig. 1). The PVC trough was protected from direct rainfall. The surface runoff of each plot was collected in a sterile 1-liter polyethylene container. Containers were observed following each rainfall event and in the case of heavy rainfall events were monitored more frequently. After collecting runoff for analysis, the same container was returned to each respective plot (Fig. 3).
Runoff sampling flowchart for slurry-treated (black) and nontreated (gray) soil plots. UACh, Universidad Austral de Chile; INIA Remehue, Institute for Agricultural Research, Remehue Research Station.
At each sampling, the runoff volume was measured using 1.0-liter buckets. Only runoff samples with ≥60 ml were used for quantitative analyses of bacteria and turbidity. Samples that were processed for bacterial detection and that had sample volumes of less than 60 ml were included in binary assessments of the presence/absence of bacteria but not in assessments of bacterial concentration. In instances of very high rainfall, a 60-ml mixed sample was collected at multiple intermediate time points. Runoff samples were coded and stored in sterile plastic tubes at a temperature of 5�C for no more than 3 days until submission to the Paratuberculosis Lab at the Instituto de Bioqu�mica y Microbiolog�a, Universidad Austral de Chile, for detection and quantification of M. avium subsp. paratuberculosis and total coliforms.
Soil sample collection.Soil was sampled monthly for M. avium subsp. paratuberculosis detection using a soil core sampler which was flame sterilized between samples. The sampling site in each soil plot was selected from the upper third of the slope, and between 5 and 10 g of material was taken from the first 5 cm of soil. The samples were coded, kept at room temperature, and submitted within 3 days to the Paratuberculosis Lab at the Instituto de Bioqu�mica y Microbiolog�a, Universidad Austral de Chile, for bacteriologic analysis.
Turbidity analysis.From the 60-ml runoff sample, 20 to 25 ml was used for turbidity analysis. Runoff turbidity was determined using a nephelometric technique at the Institute for Agricultural Research, Remehue Research Station. The results were reported as nephelometric turbidimetry units (NTU) (15).
M. avium subsp. paratuberculosis extraction and culture.Prior to inoculation into MGIT ParaTB culture medium (BD Diagnostic Systems, Franklin, NJ), runoff samples were processed following protocols for fecal samples, with several modifications noted. In brief, runoff water samples (approximately 30 ml, keeping 1 ml for total coliform analysis) were first centrifuged at 900 � g for 30 min at room temperature. This was followed by a spore germination and decontamination step which was shortened to only 6 h. Each processed sample was then inoculated into a tube of MGIT ParaTB medium (BD Diagnostic Systems, Franklin, NJ) containing supplement and antibiotics, according to the manufacturer's protocol. Soil samples were processed in the same way. Each MGIT ParaTB medium tube contained 7 ml of modified Middlebrook 7H9 broth base with mycobactin J, 500 μl of egg yolk suspension (Becton, Dickinson, Sparks, MD), 100 μl of VAN cocktail (vancomycin, nalidixic acid, and amphotericin; Sigma-Aldrich), and a fluorescent oxygen indicator embedded in silicon at the bottom of the tube. The final concentrations of antibiotics were 10 μg/ml vancomycin, 40 μg/ml amphotericin B, and 60 μg/ml nalidixic acid. Each inoculated MGIT tube was inserted in an MGIT 960 instrument (BD Diagnostic Systems, Franklin, NJ) and incubated at 37�C for 49 days. Tubes signaling positive by day 49 were removed and tested for the presence of M. avium subsp. paratuberculosis by IS900 PCR. Tubes not signaling positive by that time were considered negative for M. avium subsp. paratuberculosis.
DNA extraction.Signal-positive MGIT tubes were removed from the MGIT 960 instrument, inverted three times to mix the contents, and briefly vortexed. An aliquot of 200 μl from the middle of the tube was transferred to a 1.5-ml centrifuge tube and centrifuged at 5,000 � g at room temperature for 5 min. The supernatant of each tube was discarded. The pellet was disrupted by pipetting with a mixture of 500 μl lysis buffer (2 mM EDTA, 400 mM NaCl, 10 mM Tris-HCl, pH 8.0, 0.6% SDS) and 2 μl proteinase K (10 μg/μl) and then transferred to a bead-beating tube (BioSpec Products, Inc., Bartlesville, OK) containing 200 μl of 0.1-mm zirconia/silica beads (BioSpec Products, Inc., Bartlesville, OK). The tubes were incubated in a heating block at 56�C for 2 h. Following this, the tubes were shaken in a cell disrupter (MiniBeadbeater-8; BioSpec Products) at 3,200 rpm for 60 s and then cooled on ice for 10 min. To remove foam and beads from the inner walls, tubes were centrifuged at 5,000 � g for 30 s. A brief vortexing of the samples made sure that any DNA that stuck to small solid particles was not lost when the lysate was transferred. The entire liquid content of the bead-beating tube (500 μl) was transferred to a 1.5-ml microcentrifuge tube, and 500 μl of 100% ethanol was added. The tube was left to stand at room temperature for 2 min and then vortexed for 5 s and again centrifuged at 18,000 � g at room temperature for 5 min. Following removal of the supernatant, the resulting pellet was resuspended in 200 μl 70% ethanol by pipetting and centrifuged once more at 18,000 � g for 5 min. The supernatant was removed, and the tube was left open for 10 min at room temperature to allow evaporation of the remaining ethanol. The pellet was then resuspended in 50 μl of sterile distilled DNase-free water and incubated in a dry heating block at 100�C for 5 min with the lid closed. Finally, samples were centrifuged at 18,000 � g for 30 s to remove solids. A 25-μl aliquot of the final DNA extract was transferred to a new 1.5-ml tube and stored at −20�C until analysis by PCR.
Real-time PCR confirmation of M. avium subsp. paratuberculosis.A real-time PCR was performed, where the target was the insertion element IS900. The PCR mixture included 5 μl DNA template, 10 μl 2� TaqMan Universal Master Mix (Roche, Indianapolis, IN), 0.2 μM IS900 primers, 0.1 μM probe (Roche, Indianapolis, IN), and water for a total volume of 20 μl. Primer sequences for IS900, which amplified a 63-nucleotide fragment of the IS900 gene target, were 5′-GACGCGATGATCGAGGAG-3′ (left) and 5′-GGGCATGCTCAGGATGAT-3′ (right). The probe sequence was TCGCCGCC. The reactions were carried out in a Roche LightCycler (version 2.0) system under the following standard conditions: one cycle at 95�C for 10 min; 45 cycles with three steps of 95�C for 10 s, 60�C for 30 s, and 72�C for 1 s; and a final cooling step at 40�C for 30 s. Negative and positive (Mycobacterium avium subsp. paratuberculosis ATCC 19698) PCR controls, as well as DNA extraction negative and positive controls, were included.
Quantification of M. avium subsp. paratuberculosis.M. avium subsp. paratuberculosis quantification was performed in liquid culture using time to detection (TTD), provided by the Bactec MGIT 960 software, using methods already described (14) and modified (16). Briefly, a standard curve was established by plotting serial dilutions of aggregate-free M. avium subsp. paratuberculosis against TTD (Fig. 4). Aggregate-free M. avium subsp. paratuberculosis suspensions were obtained by sonication and filtration, and organisms were quantified by visual counting under a microscope using a Neubauer chamber (17). Thereafter, MGIT tubes were inoculated with the same serial dilutions for TTD determination. The relationship between M. avium subsp. paratuberculosis cell counts (CCs) and TTD was fitted to the following equation: log10 CC = span � e−k � TTD + plateau, where span is the difference between TTD at time zero and that at the plateau, k is the degree of decay for the log10 CC, and plateau is the value for log10 CC curve flattening. Estimates for span (12.23), k (−0.09644), and plateau (0.9467) were obtained in a previous study (16) by a nonlinear least-squares approach using GraphPad Prism (version 4.03) software (San Diego, CA).
Standard curve for M. avium subsp. paratuberculosis quantification, i.e., relationship between the bacterial concentration in the specimen and TTD in Bactec MGIT ParaTB medium read using a Bactec MGIT 960 instrument.
Coliform counts.Runoff water samples were homogenized by vortexing and then serially diluted with water. One milliliter of the 10−2 dilution was plated on Petrifilm plates (3M Microbiology Products) (18) and incubated at 37�C for 24 h. Plates were read visually and interpreted according to the manufacturer's instructions, and results were expressed as the number of CFU/ml of runoff water sample.
Ambient environmental conditions.Environmental conditions during the experimental period were obtained from the official records of the INIA Remehue Agrometeorological Station at the location of the experiment (latitude, 40�31′S; longitude, 73�03′W; altitude, 73 m above sea level). The data obtained were minimum and maximum daily temperatures (1.5 m above the ground), evaporation, rainfall intensity, and cumulative rainfall beginning on 1 June 2011 and continuing for 4 months (Table 1).
Summary of environmental conditions observed during the study period
Data analysis.Analyses were performed in SAS software (version 9.3; SAS Corporation, Cary, NC). Predictors evaluated were design variables (slope, slurry application), study duration, rainfall on the day that runoff was collected, and rainfall during the interval between the previous runoff collection and the sample collection day. Although the initial layout of the soil plots included sorting of plot slopes and slurry application by row, no effort was made to randomize treatments within rows. The excavated 1- by 2-m turf plots collected were randomly assigned across the 24 plots through simple random sampling. Therefore, this analysis was performed without considering this to be a blocked design.
Turbidity values were log transformed prior to use. Bacterial counts were log transformed, and for regression analyses, these values were rounded up to the nearest integer. Pearson correlations were calculated for log-transformed turbidity and bacterial counts. The relationship between sample turbidity and presence/absence of M. avium subsp. paratuberculosis or coliforms was analyzed via logistic regression (PROC LOGISTIC procedure). Turbidity was not included in the multivariate analyses, as only a subset of samples was tested for turbidity due to low sample volumes.
Separate analyses were performed to determine whether the probability of positive samples and bacterial counts were influenced by study factors and selected environmental factors. Multivariate generalized estimating equation (GEE) models with repeated statements in the PROC GENMOD procedure were run for the presence of bacteria using a logistic link and binomial distribution. Models for transformed bacterial counts used a log link and either a negative binomial or a Poisson distribution. Prior to multivariate model construction, independent analyses were performed to determine parameters for inclusion in initial multivariate models. Independent analyses for logistic models were performed without repeated measures using PROC LOGISTIC, while independent analyses for bacterial counts accounted for repeated sampling via GEEs in PROC GENMOD.
Initial multivariate models included all predictors from independent analyses with P values of <0.20 and were subsequently reduced by backward stepwise regression. The final model structure was determined by the significance of the predictor variables (P < 0.05); however, all design variables were forced into the final models. Correlation structures (independent, errors are uncorrelated across samples; exchangeable, errors are equally correlated across all samples; autoregressive, error correlations decrease exponentially across samples; banded Toeplitz [3 bands], error correlations for samples within two of the selected samples are the same for each different distance and are zero between samples further away) were compared by quasilikelihood under the independence model criterion (QIC) values. Selection of negative binomial or Poisson models was based on QIC.
RESULTS
Environmental conditions.In June and July (winter months in Chile), a decrease in maximum daily temperatures and evaporation was observed. Progressively lower minimum temperatures and increasing daily rainfall were recorded from June to August (Table 1).
M. avium subsp. paratuberculosis recovery.There were 481 runoff samples collected and tested for M. avium subsp. paratuberculosis of a total of 492 possible at 42 sampling time points. For 11 individual plot-time point combinations, no runoff was available for evaluation. Seventy-five samples without sufficient volume were processed for the presence of bacteria only and not bacterial concentration or turbidity. Of the 481 samples, 95 (19.8%) were positive for M. avium subsp. paratuberculosis and 134 (27.9%) were positive for coliforms. The number of plots with runoff testing M. avium subsp. paratuberculosis positive and coliform positive over time are presented, along with the number of plots testing positive for M. avium subsp. paratuberculosis from soil (Fig. 5 and 6).
Results for days 0 to 110 after slurry application. (A) Rainfall intensity (mm/day); (B) water runoff volume (ml); (C) runoff turbidity (NTU); (D) rate of coliform detection (number of soil plots testing positive); (E) rate of M. avium subsp. paratuberculosis detection (number of soil plots testing positive). The major horizontal axes show the days PSA (range, days 1 to 110). The minor horizontal axes show the sample number (1 to 40). Black bars, slurry-treated soil plots; open bars, nontreated soil plots.
Number of runoff samples and soil plot samples positive for M. avium subsp. paratuberculosis and coliforms over time. Closed circles and dashed black line, M. avium subsp. paratuberculosis in soil; open gray circles and dotted gray line, coliforms in runoff; open inverted triangles and solid black line, M. avium subsp. paratuberculosis in runoff.
Samples were used up to sampling time point 42 (Fig. 5). Water runoff from all slurry-treated plots was culture positive for M. avium subsp. paratuberculosis at least once, and M. avium subsp. paratuberculosis was detected in water runoff at 7 to 106 days post-slurry application (PSA), with the highest number of plots testing positive in runoff on days 25 to 47 PSA (Fig. 5E). The highest concentration of M. avium subsp. paratuberculosis in runoff was found from slurry-treated plot 1 on day 39 PSA (6.2 � 106/ml), plot 8 on day 37 PSA (5.0 � 106/ml), and plot 9 on day 41 PSA (3.8 � 106/ml). In total, M. avium subsp. paratuberculosis was recovered from runoff of all 6 slurry-treated plots, and counts averaged 1.5 � 103 � 1 � 101/ml. Control plot 5 was not treated with slurry, but M. avium subsp. paratuberculosis was recovered from runoff on 10 separate instances at counts ranging from 1.6 � 101 to 4.5 � 103 CFU/ml.
A total of 84 soil core samples were tested for M. avium subsp. paratuberculosis, of which 13 (15.5%) were found to be positive; all were only from slurry-treated plots. M. avium subsp. paratuberculosis was isolated from soil core samples of 5/7 slurry-treated plots at 29 days PSA and 2/6 slurry-treated plots at 97 days PSA (Fig. 6).
Analytical results.The correlation between log M. avium subsp. paratuberculosis counts and log coliform counts for 426 samples with both values was weak (0.19; P < 0.001, Pearson correlation). Turbidity values were not available for 5 samples with volumes of ≥60 ml. When evaluated independently, a 1-unit increase in log turbidity increased the odds of detecting both M. avium subsp. paratuberculosis (odds ratio [OR], 1.30; 95% confidence interval [CI], 1.05 to 1.60; P = 0.016) and coliforms (OR, 1.94; CI, 1.57 to 2.40; P < 0.001). For each 1-unit increase in log turbidity, the expected log coliform count increased (OR, 1.54; CI, 1.25 to 1.90; P < 0.001). However, turbidity did not influence the M. avium subsp. paratuberculosis concentration (OR, 1.07; CI, 0.82 to 1.39; P = 0.62).
M. avium subsp. paratuberculosis.Independent analyses indicated that slope (P = 0.040), slurry (P < 0.001), and study duration (P = 0.001) but not rainfall event volume or rainfall volume between samplings (P > 0.2) should be included in preliminary multivariate models for the probability of detecting M. avium subsp. paratuberculosis. Backward stepwise selection in generalized estimating equation-based models considering repeated samples from specific plots resulted in all three predictors being included in the final model. Overall, multivariate models with the Toeplitz correlation structure had the lowest QIC. The banded Toeplitz covariance structure indicates that samples close together in time within a plot are correlated; however, the correlation does not decay exponentially with sampling distance. The probability of detecting M. avium subsp. paratuberculosis in runoff increased with slope and slurry application and declined with study duration (Table 2).
Odds of M. paratuberculosis positivity in runoff
From 426 runoff samples with sufficient volume for bacterial analysis, independent analyses indicated that slurry (P = 0.025), study duration (P < 0.001), volume of rainfall at the runoff event (P = 0.006), and volume of rain between runoff events (P = 0.014) influenced the M. avium subsp. paratuberculosis concentration in runoff. The design variable slope was forced into multivariate models (P = 0.42). All predictors included with the exception of slope were significant in the final multivariate model with a Toeplitz correlation structure, indicating a correlation between samples taken consecutively which did not decay exponentially with the following sample. A Poisson distribution was selected over the negative binomial on the basis of the lower QIC. M. avium subsp. paratuberculosis log counts had a tendency to be higher at a 15% slope than a 3% slope; however, this effect was not significant (P = 0.101) (Table 3). Plots with contaminated slurry applied had a 10-fold higher log count of M. avium subsp. paratuberculosis recovered than plots without slurry application. In 1 week, the log10 M. avium subsp. paratuberculosis count decreased 0.93 times. While a 1-ml increase in rainfall at sample collection reduced log counts 0.97-fold, conversely, rainfall between sample collection time points increased counts 1.03-fold.
M. paratuberculosis counts in runoffa
While slurry application and the volume of rainfall between runoff events increased the M. avium subsp. paratuberculosis concentration in runoff, there was a trend toward increased bacterial counts at a higher volume. The volume of daily rainfall and study duration decreased M. avium subsp. paratuberculosis concentrations in runoff (Table 3).
Coliforms.Independent analyses indicated that slurry (P = 0.002), duration (P < 0.001), and rainfall on the current day (P = 0.001) should be included in preliminary multivariate analysis for detecting coliforms in runoff, while rainfall between runoff events and slope were not associated with the probability of detecting coliforms (P > 0.20). Following backward selection, the multivariate model with the Toeplitz correlation structure indicated that slope was not associated with detection of coliforms; however, there was a trend toward increased detection of coliforms with slurry application. The probability of detection decreased with study duration and increased with the volume of rainfall events (Table 4).
Odds of coliform positivity in runoff
Coliform concentrations.Independent analyses indicated that slurry application (P = 0.02), duration of study (P < 0.001), and daily rainfall (P < 0.001) should be included in multivariate models but that slope and rainfall during an interval (P > 0.20) did not influence the concentration of coliforms in runoff. Slope was forced into multivariate models as a design element. A Poisson distribution was selected over the negative binomial on the basis of the lower QIC. The best-fitting models used an exchangeable correlation structure, indicating that the degree of correlation between coliform readings was related more to overall plot characteristics than to immediate events (i.e., all measurements for a plot were generally similar, but the relationship between two samples in a plot was not more similar if those samples were consecutive or chosen at random). Slope was not related to coliform count, as the incidence rate ratio was not different between high- and low-sloped plots (P = 0.577). Plots with slurry applied had log10 coliform counts 2.14 times higher than those of untreated plots. A 1-week increase in study duration decreased log10 coliform counts 0.87 times relative to those in the previous week, while a 1-ml increase in rainfall increased log10 coliform counts 1.02 times (Table 5).
Coliform counts in runoffa
DISCUSSION
Southern Chile, with abundant grassland and moderate temperatures, is an ideal environment for raising dairy cattle. It receives 1,200 mm to 3,200 mm of rainfall annually, and this rainfall is concentrated in the winter season (June to August), when daily temperature lows are 2 to 4�C and daily highs are 10 to 12�C. These conditions result in a high rate of rainfall runoff water, particularly from steeply sloped soils (8).
Manure from dairy farms is commonly applied to soils as a slurry to dispose of the manure and to improve soil fertility. Contained in the slurry are multiple potential animal and human pathogens, such as Salmonella, Escherichia coli O157, Yersinia spp., M. avium subsp. paratuberculosis, and many others (10, 13). Thus, it is important to understand the dynamics of pathogen survival and movement in soil and rainwater runoff.
Previous studies using lysimeters showed that M. avium subsp. paratuberculosis does not percolate down into the soil. Rather, it is retained at the soil surface (13, 19). Thus, we hypothesized that M. avium subsp. paratuberculosis would be susceptible to being washed off the soil surface and into runoff after a rain event. M. avium subsp. paratuberculosis movement was studied in situ using plots of undisturbed soil placed on platforms, allowing control of the soil slope and facilitating runoff collection. The use of slurry contaminated with known quantities of M. avium subsp. paratuberculosis allowed precise measurement of the effects of several environmental parameters and variables on M. avium subsp. paratuberculosis detection.
Slurry application tended to influence the likelihood of coliform detection in runoff (P = 0.002 in independent analysis and P = 0.061 in multivariate analysis). One plot, plot 5, that was not treated with M. avium subsp. paratuberculosis-contaminated slurry yielded positive M. avium subsp. paratuberculosis cultures from runoff on several occasions. Other than accidental contamination of the plot, we have no good explanation for these findings. We did not exclude data from plot 5 when doing statistical analyses.
Soil slope, 15% versus 3%, significantly affected the presence of M. avium subsp. paratuberculosis in rainwater runoff from soil plots where M. avium subsp. paratuberculosis-contaminated slurry was applied (P = 0.035). However, the concentration of M. avium subsp. paratuberculosis in runoff was not significantly associated with soil slope (P = 0.10), though it could be considered a trend. In contrast, neither coliform presence nor quantity in runoff was influenced by soil slope.
Rainfall intensity affected both the likelihood of finding coliforms and the quantity of coliforms in runoff (P < 0.001). In contrast, neither M. avium subsp. paratuberculosis presence nor M. avium subsp. paratuberculosis concentration increased (actually, the concentration decreased) with rainfall intensity (mm rainfall/day).
The different effects of both slope and rainfall intensity on M. avium subsp. paratuberculosis bacteria compared to those on coliforms were surprising and collectively suggest that the dynamics of M. avium subsp. paratuberculosis and coliform movement across a soil surface and into runoff are influenced differently by physical or biochemical interactions with soil or water. There are major differences in the cell wall structure of mycobacteria and Gram-negative bacteria. These differences likely affect adsorption to soil particles mediated by van der Waals electrostatic forces or by hydrophobic interactions with the cell surface (19, 20). Adsorption to soil would also be affected by soil pH (20). In southern Chile, acidic soils are frequent, with pH values normally ranging from 5.1 to 6.0. These acidic soils are also high in organic matter and are of volcanic origin (8).
No association between M. avium subsp. paratuberculosis counts and coliform counts was seen. This is consistent with the aforementioned findings indicating that M. avium subsp. paratuberculosis and coliform movement into runoff is influenced differently by soil or water. As the study proceeded, starting with application of M. avium subsp. paratuberculosis-contaminated slurry, coliforms were detected first and M. avium subsp. paratuberculosis were detected some days later. The numbers of plots testing positive for coliforms and M. avium subsp. paratuberculosis clearly show this pattern of delayed M. avium subsp. paratuberculosis appearance (Fig. 6). The practical implication is that coliforms may not serve as good indicator organisms for prediction of M. avium subsp. paratuberculosis contamination of runoff water.
Runoff water turbidity was associated with an increased probability of detecting coliforms (P < 0.001) and was also directly related to the concentration of coliforms (P < 0.001). Runoff turbidity was associated with an increased probability of detecting M. avium subsp. paratuberculosis (P = 0.016) but was not related to the concentration of M. avium subsp. paratuberculosis (P = 0.62). These findings further support the observed differences in the movement of coliforms and M. avium subsp. paratuberculosis into water running off the surface of soil during a rainfall event. They also indicate that runoff turbidity could be used as a screening test to select samples with a higher probability of coliform or M. avium subsp. paratuberculosis contamination of runoff.
Daily rainfall negatively affected the concentration of M. avium subsp. paratuberculosis in runoff (P < 0.035), while the total volume of rain falling between runoff sample collections positively affected M. avium subsp. paratuberculosis concentrations (P < 0.022). Rainfall in the interval between sample collections is an indirect measure of soil saturation. In water-saturated soil, the adhesion forces between M. avium subsp. paratuberculosis and soil particles may weaken, allowing the organism to more easily be carried across the soil surface with water runoff (21).
M. avium subsp. paratuberculosis and coliforms were detected in runoff up to 106 and 112 days after the start of the study, respectively. M. avium subsp. paratuberculosis was also recovered from soil samples from all slurry-treated plots, but no soil samples were M. avium subsp. paratuberculosis culture positive beyond day 100 post-slurry application. This roughly coincided with the end of the rainy season. Study duration was strongly correlated with a decreasing detection rate and the quantity of both M. avium subsp. paratuberculosis and coliforms (P < 0.001).
M. avium subsp. paratuberculosis is now frequent in dairy cattle globally. Manure from dairy herds and, therefore, M. avium subsp. paratuberculosis are commonly spread on land. This study demonstrates that M. avium subsp. paratuberculosis will contaminate water running off soils on which M. avium subsp. paratuberculosis-contaminated slurry has been applied. The studies further suggest that coliform counts may not be a good predictor of M. avium subsp. paratuberculosis contamination of water runoff because coliforms behave differently than M. avium subsp. paratuberculosis, possibly due to different physical and chemical interactions with soil and water. M. paratuberculosis contamination of water has potential consequences for both animal and human health (22).
ACKNOWLEDGMENTS
This work was supported by FONDECYT grant 11100200. R. M. Mitchell is supported by a USDA NIFA postdoctoral fellowship (2011-67012-30725).
FOOTNOTES
- Received 25 February 2013.
- Accepted 24 March 2013.
- Accepted manuscript posted online 29 March 2013.
- Copyright � 2013, American Society for Microbiology. All Rights Reserved.
