Ability of unsterilized recycled manure solids bedding to support growth of Klebsiella pneumoniae and Escherichia coli

Abstract

Although recycled manure solids (RMS) bedding is used on dairy farms, it could allow bacterial growth when contaminated by feces and thus increase the incidence of clinical mastitis in cows. The objective of this study was to describe bacterial growth in three different types of RMS bedding, as well as in sand, when samples were experimentally inoculated with Escherichia coli and Klebsiella pneumoniae. Two 3-day trials were conducted, during which treatments included inoculating bedding samples with E. coli and K. pneumoniae, as well as no inoculation. The trial was repeated 3 times for each bedding sample on each day. Samples were incubated at 15°C for 3 d and bacterial counts were measured every day. After inoculation, there was no significant K. pneumoniae or E. coli growth phase during the trial in those RMS samples that were prepared either in a container or in a heap. Recycled manure solids and sand samples prepared in a rotary drum, however, showed a similar active growth phase of K. pneumoniae during the first 24 h of the trial. Moreover, a significant E. coli growth phase was observed in the samples of sand bedding in the first 24 h. The 3 different types of RMS bedding samples did not react in a similar manner to coliform inoculation. No active growth phase was observed in bedding samples already containing a high bacterial concentration following inoculation with coliforms.

Introduction

Recycled manure solids (RMS) bedding is commonly used on dairy farms, despite its less desirable microbiological characteristics. It has been demonstrated that cows housed on RMS bedding were at a much greater risk [7.0; 95% confidence interval (CI): 2.0, 24.6] of experiencing clinical mastitis (CM) due to Klebsiella pneumoniae than cows on straw bedding (1).

This finding confirmed anecdotal reports from farmers and stakeholders who are often concerned about the incidence and severity of environmental CM episodes in herds using RMS bedding. One hypothesis to explain the higher incidence of clinical mastitis due to Klebsiella spp. in RMS-bedded herds is that RMS bedding would allow rapid growth of Klebsiella spp. when contaminated with fecal material. This would quickly expose the cows’ teats to a high risk of infection.

Godden et al (2) studied the ability of RMS bedding to promote the growth of environmental bacteria and found that RMS showed the highest potential for growth of K. pneumoniae when compared to wood shavings or sand (recycled or clean). In that study, however, bedding was first sterilized and then inoculated with coliforms (either Enterococcus faecium or K. pneumoniae) and monitored over time. A common belief about composting is that the intrinsic microbiome will impede bacterial growth after an external inoculation. Therefore, the results of a study in which RMS bedding was sterilized before inoculation may not accurately represent a real-life situation. In fact, cows on a farm would contaminate bedding that already contains a diverse and abundant microbiome.

The objective of our study was to describe bacterial growth in RMS and sand bedding when inoculated with coliforms, but without prior sterilization of the bedding. We hypothesized that RMS would allow a greater growth rate of coliforms than sand, especially during the first 24 h following inoculation.

Materials and methods

Collection of bedding samples

A convenience sample of 3 herds bedded on RMS was chosen from 27 such herds recruited for a previous study (1). The selection criteria for these herds have been described in earlier studies (1–4). Briefly, herds needed to be located within 250 km of the University of Montreal’s Faculty of Veterinary Medicine in Saint-Hyacinthe, Quebec, Canada and to have used RMS as the primary bedding for their lactating cows for at least 6 mo at the time samples were taken from 2018 to 2019.

In January 2020, 3 RMS herds were visited to collect samples for the regrowth trial. These herds were chosen based on their proximity to the facilities of the Research and Development Institute for the Agri-Environment (IRDA) in Saint-Bruno-de-Montarville, Quebec and to represent different methods of preparing bedding, i.e., maturation in a heap, a closed container, or a rotary drum. A fourth herd using sand bedding was conveniently recruited for comparison purposes.

Bedding samples were collected in a clean plastic bag from the center of the storage pile just before the bedding was used in stalls. Samples were kept on ice during transport and stored at 2°C to 8°C in the lab facilities until the trials, which began within 24 h of sample collection.

Experimental setup

Two trials were conducted in controlled conditions at the IRDA laboratory. In each trial, samples (sand and RMS from 3 farms) were separated into sterile filter bags (25 g per aliquot). Treatments included bacterial inoculation to demonstrate bacterial growth following contamination of the bedding by cows or no inoculation to demonstrate bacterial growth in unused bedding, followed by various incubation periods of 0, 1, 2, or 3 d at 15°C, which was chosen to represent the average temperature on dairy farms. For each bedding sample, 3 replicates were carried out per treatment group, inoculated or not, per incubation time (0, 1, 2, or 3 d), and per trial. Two different trials were conducted, which resulted in 6 replicates for each condition (Figure 1).

Figure 1.

Experimental design for a study investigating growth of E. coli and of K. pneumoniae in bedding samples inoculated with 150 μL of a solution containing E. coli and K. pneumoniae (to mimic contamination by cows’ feces) or with 150 μL of sterile water (as control).

Preparation of inoculum

Three different isolates of E. coli (ID #31100908, #10109458, and #10109199) and of K. pneumoniae (ID #10214985, #41518533, and #41518519) were obtained from the Mastitis Pathogen Culture Collection of the Mastitis Network for use in this study (5). These isolates have been used and described in previous studies (6,7).

The isolates were all obtained from milk samples collected from the affected mammary gland on the day clinical mastitis was diagnosed. The cows sampled ranged from first to 7th lactation, were from 4 different farms located in Alberta, Quebec, and the Atlantic provinces, and their clinical mastitis episode occurred on Days 1 to 298 of their lactation.

The day before bedding samples were inoculated, inoculum solutions containing a mixture of approximately 9 Log₁₀ colonyforming units (CFUs)/mL of the E. coli or K. pneumoniae isolates were prepared as follows. One full inoculating loop of each of the 3 previously tryptic soy agar (TSA)-grown strains (35°C for 24 h) was transferred into a bottle containing 500 mL of tryptic soy broth (TSB) and warmed at 35°C for 3 h. Escherichia coli and K. pneumoniae inocula were prepared separately. Bottles were incubated at 35°C overnight (12 h) at 120 × g on an orbital shaker.

Inoculation and incubation of recycled manure solids

Half the aliquots of a given bedding sample were inoculated with 75 μL of each inoculum preparation, in addition to what the bedding sample may already contain, thus representing an addition of approximately 6 Log₁₀ CFUs/g of E. coli and 6 Log₁₀ CFUs/g of K. pneumoniae per aliquot. Aliquots for the non-inoculated group received 150 μL of sterile water to ensure that each sample had the same moisture content. All aliquots were incubated at 15°C for up to 3 d and a set of analyses was carried out on Days 1, 2, and 3 post-inoculation.

Analytical procedures

For each type of bedding, basic physicochemical measurements, as well as bulk density and dry matter (DM) content, were obtained before inoculation. Briefly, 30 g of each material were weighed on foil plates (wet mass) and dried at 105°C for 24 h before weighing again (dry mass). Dry matter content of the material was calculated by taking the ratio of its dry mass to its wet mass. For bulk density, a 100-cm³ jar filled with bedding without compacting was weighed. Density was calculated by dividing the total mass by 100 cm³. Assays were run in triplicate.

Bacterial analyses were carried out every day. Counts of E. coli and presumptive K. pneumoniae were determined as follows. Briefly, 225 mL of 0.1% peptone water was added to 25 g of solid material and thoroughly mixed in a stomacher for 60 s; 1 mL of this solution was transferred into 9 mL of 0.1% peptone water and a series of 1:10 dilutions was completed. For each dilution, 1 mL was plated on a 3M Petrifilm E. coli/Coliform Count Plate (3M Microbiology Products, St. Paul, Minnesota, USA) according to the manufacturer’s instructions for E. coli counts.

The agar associated with the first dilution (25 g of sample in 225 mL of peptone water) represented 0.1 g of sample, the second dilution represented 0.01 g of sample, and so on. For counting Klebsiella spp., 50 μL of each dilution was also directly plated on MacConkey Agar No. 3 (Oxoid, Basingstoke, UK) and incubated at 35°C for 24 h. For this analysis, the agar associated with the first dilution (25 g of sample in 225 mL of peptone water) represented 0.005 g of sample, the second represented 0.0005 g of sample, and so on.

Dilutions that presented from 30 to 300 pale pink mucoid colonies were counted as presumptive Klebsiella spp. 3M Petrifilm was incubated at 35°C for 48 h and dilutions presenting 15 to 150 blue colonies associated with entrapped gas were counted as E. coli. If no agars presented the adequate range, dilutions presenting 1 to 14, for E. coli, or 1 to 30, for Klebsiella spp. were used for counting. For both analyses, CFUs counted on agars were reported per gram of sample by dividing the counted colonies by the associated sample quantity (wet-weight basis).

The limits of detection for E. coli and Klebsiella were 10 CFUs/g and 200 CFUs/g, respectively. This is associated with 1 CFU present on the first dilution. For each analytical method, negative and positive controls were used with each batch of samples tested. For positive controls, ATCC standard strains (E. coli 25922 and K. pneumoniae 13883) were added in 0.1% peptone water, plated, and incubated on separated agars as mentioned for samples.

Statistical analysis

Bacterial concentrations were reported on a dry-weight basis and on a volumetric basis, as suggested by Gabler et al (8). Briefly, CFUs per gram of sample were reported on a dry-weight basis or on a volumetric basis by dividing the number of CFUs by the sample’s dry matter content or the bulk density, respectively.

When comparing only one type of bedding and its variations in bacterial count over time, bacterial counts on a weight basis can be appropriate. When comparing 2 or more different types of bedding, e.g., RMS and sand, the variation of material density in bedding types may bias the analyses. Since the amount of bedding added to stalls is related to the volume needed to cover the bottom of the stall and since this volume represents the potential exposure of a cow’s teats to pathogens, bacterial count per volume is increasingly reported in cm³ when analyzing bedding (9,10,11).

Bacterial counts were normalized using log-transformation. Linear mixed regression models were used to describe the effect of coliform inoculation of different beddings on bacterial counts over time. Two mixed models were constructed, one for the K. pneumoniae counts and another for the E. coli counts. The outcome was the least square mean bacterial count (log₁₀ CFU/g or log₁₀ CFU/cm³) and the predictors were treatment (inoculated or not), time (Days 0 to 3), and bedding sample (closed container RMS, heaped RMS, rotary drum RMS, or sand).

The following interaction terms were also included in the models: i) time and treatment; ii) time and bedding sample; and iii) time, treatment, and type of bedding sample. A random intercept of bedding samples was also added to capture the variations due to triplicate samples realized in all assays. Significance level was set at P < 0.05. Statistical analyses were carried out with SAS Version 9.4 using the Proc Mixed procedure (SAS Institute, Cary, North Carolina, USA). These datasets and SAS scripts are publicly available at https://doi.org/10.5683/SP3/RM6Y48

Results

The physicochemical characteristics of the bedding samples used are provided in Table I. Briefly, the characteristics of the 3 types of RMS bedding samples were relatively similar. For example, dry matter content (DM) varied from 32.5 to 44.3%, bulk density varied from 0.19 to 0.29 g/cm³, pH ranged from 8.6 to 9.2, and organic matter content varied from 86.9 to 94.2%. Conversely, for the sand sample, DM was 96.3%, density was 0.96 g/cm³, pH was 7.4, and organic matter content was 0.2%.

Table I.

Physicochemical analyses of bedding samples.

Bedding type	DM (%)	Density (g/cm3)	pH	Total N (mg/kg)	N-NH4 (mg/kg)	Ashes (%)	Organic matter (%)	Total organic C (%)	N-NO3 (%)
Container RMS	35.4	0.29	9.2	5195	305	5.8	94.2	47.1	11.3
Heaped RMS	32.5	0.19	9.0	5266	313	7.5	92.5	46.2	5.7
Rotary drum RMS	44.3	0.21	8.6	6731	496	13.1	86.9	43.5	12.1
Sand	96.3	0.96	7.4	0	0	99.8	0.2	0.1	0.1

DM — Dry matter; N-NH₄ — Ammonium; N-NO₃ — Nitrate nitrogen; N — Nitrogen; RMS — Recycled mature solids.

Klebsiella pneumoniae model

Results of the K. pneumoniae models using a volumetric- or weight-based analysis were relatively similar. For simplicity, only volumetric-based analyses will be discussed, although both volumetric- and mass-based analyses are provided in Figure 2.

Figure 2.

Estimated concentration (least square means estimates and 95% CIs) of K. pneumoniae for different types of bedding, inoculated or not with K. pneumoniae, over a 3-day period. Results are presented both on a volumetric basis (in log₁₀ CFUs/cm³; A) and on a dry-mass basis (in log₁₀ CFUs/g; B).

On Day 0, the K. pneumoniae counts of the 4 untreated bedding samples were all statistically different from each other (Figure 2), with closed-container RMS having the largest estimated [least square mean (LSM) estimates] concentration of K. pneumoniae (3.6 log₁₀ CFUs/cm³; 95% CI: 3.3, 3.8), followed by heaped RMS (2.5 log₁₀ CFUs/cm³; 95% CI: 2.2, 2.7), sand (2.0 log₁₀ CFUs/cm³; 95% CI: 1.7, 2.2), and rotary-drum RMS (1.5 log₁₀ CFUs/cm³; 95% CI: 1.2, 1.7). An active K. pneumoniae growth phase was observed in closed-container RMS during the first 48 h of the trial, whereas the active-growth phase of heaped RMS was completed in 24 h. There was no significant K. pneumoniae growth in the rotary-drum RMS or in sand during the entire assay.

Once inoculated, closed-container and heaped RMS did not undergo any significant K. pneumoniae growth phase (LSM estimates ranging from 5.8 to 6.2 log₁₀ CFUs/cm³ for closed container and 5.9 to 6.2 log₁₀ CFUs/cm³ for heaped). However, a similar active K. pneumoniae growth phase was observed in the rotary-drum RMS and in sand during the first 24 h of the trial (LSM from 6.0 to 6.6 log₁₀ CFUs/cm³ for rotary drum and 6.6 to 7.2 log₁₀ CFUs/cm³ for sand) before entering a stationary phase for the last 48 h.

Escherichia coli model

For E. coli, results of the mass-based and volumetric-based analyses were similar. At the beginning of the trial, E. coli contents were significantly different in all untreated beddings (Figure 3), with closed-container RMS having the greatest E. coli counts (LSM estimates of 3.9 log₁₀ CFUs/cm³, 95% CI: 3.6, 4.3). Heaped and rotary-drum RMS had significantly lower E. coli counts (LSM estimates of 2.1 log₁₀ CFUs/cm³, 95% CI: 1.7, 2.4; and 1.7 log₁₀ CFUs/cm³, 95% CI: 1.3, 2.0, respectively). Sand had the lowest initial E. coli estimated counts at 1.2 log₁₀ CFUs/cm³ (95% CI: 0.9, 1.5).

Figure 3.

Estimated concentration (least square means estimates and 95% CIs) of E. coli for different types of bedding, inoculated or not with E. coli, over a 3-day period. Results are presented both on a volumetric basis (in log₁₀ CFUs/cm³; A) and on a dry mass-basis (in log₁₀ CFUs/g; B).

During the 3-day trial, E. coli counts decreased over time in all untreated beddings, except for heaped RMS where an initial growth phase was observed during the first 24 h (from an LSM of 2.1 on day 0 to 2.8 log₁₀ CFUs/cm³ on Day 1), followed by a decrease in the counts over the next 2 d (LSM of 2.4 on day 2 and 2.2 log₁₀ CFUs/cm³ on Day 3).

When inoculated with E. coli, all RMS beddings had similar initial counts (approximatively 6 log₁₀ CFUs/cm³) and did not undergo any significant E. coli growth throughout the trial. In sand bedding, however, significant growth of E. coli was observed in the first 24 h of the study, going from 6.5 log₁₀ CFUs/cm³ (95% CI: 6.1, 6.8) to 7.0 log₁₀ CFUs/cm³ (95% CI: 6.6, 7.3). A stationary phase was then observed in sand for the last 48 h of the trial.

Discussion

In this study, we observed that samples of RMS bedding that were prepared using different methods had different initial concentrations of coliform. For K. pneumoniae counts, an active growth phase was observed in inoculated rotary-drum RMS and sand samples during the first 24 h of the assay. However, only the sand sample demonstrated an active E. coli growth phase during the first 24 h of the assay.

Our results differ from those of an earlier study by Godden et al (2) in which RMS appeared to best support the growth of K. pneumoniae after inoculation. The preparation methods of the bedding used in our study, however, varied considerably from those used in this earlier study, as they studied RMS beddings made of post-digested manure solids. We hypothesized that, after an anaerobic treatment of 30 d, the bacterial count of the ready-to-use bedding may have been lower than that of the RMS beddings used in our study. Most importantly, in the Godden et al (10) study, bedding was sterilized before inoculation.

In our study, the greatest growth phase generally occurred in the beddings with the lowest initial bacterial concentrations. Indeed, after accounting for the type of bedding, the initial K. pneumoniae concentration was a significant predictor of the growth of K. pneumoniae over the next 24 h (analyses shown as supplementary materials at https://doi.org/10.5683/SP3/RM6Y48).

For example, in samples that were not inoculated with K. pneumoniae, a 1-unit increase in the initial log₁₀ CFUs/cm³ was associated on average with a 0.65 log₁₀ CFUs/cm³ (95 CI: 0.45, 0.85) decrease in the subsequent 24-hour growth, which illustrated the lower K. pneumoniae 24-h growth in samples with a high initial concentration of K. pneumoniae. We hypothesize that a similar phenomenon explains the dramatic growth phase in the first 24 h of the Godden et al (2) study when the absence of an intrinsic bacterial population allowed a considerable growth of the bacterial inoculum.

Our study highlighted the differences of the various RMS bedding samples in their ability to support bacterial growth. It should be noted that, with only one sample of each bedding type, we cannot comment on how the method used to prepare and store the RMS bedding, i.e., closed container, heaped, or rotary drum, would have affected the initial bacterial load or bacterial growth following inoculation. We can only conclude that all the RMS beddings did not react in a similar manner.

In our study, the closed-container and heaped RMS samples had relatively high initial K. pneumoniae counts. These counts further increased during the study, even though there were no additional inoculations. With an already high bacterial concentration, these 2 bedding samples did not experience a significant active K. pneumoniae growth phase when inoculated. The rotary-drum RMS and sand samples were similar, however, with very low K. pneumoniae counts in the untreated bedding. Moreover, these latter samples did not undergo any natural bacterial growth during the trial. Once inoculated, however, a significant active growth phase of K. pneumoniae occurred in these two bedding samples during the first 24 h of the trial.

These results suggest a potential saturation effect for K. pneumoniae in some samples. Apparently, bedding samples with low initial bacterial counts were able to support growth of K. pneumoniae, whereas bedding samples with already high counts of K. pneumoniae did not. We could still hypothesize that it would be better for cows to spend a number of hours on bedding with an initially lower count of K. pneumoniae than on bedding with an already high concentration when put into the stalls. Once contaminated with K. pneumoniae from fecal material, however, this initial difference may no longer be relevant.

In the E. coli model, when bedding samples were not inoculated (untreated), the closed-container RMS samples were similar to the heaped RMS samples. Again, the results of the rotary-drum RMS samples were similar to those of the sand samples. Overall, in the untreated bedding samples, the E. coli counts decreased over time, which suggests a poor survival rate of E. coli in this environment. Once inoculated, RMS samples did not show any significant E. coli growth during the assay. The sand samples, however, had an active growth phase of E. coli in the first 24 h.

During this study, we did not confirm our initial hypothesis that the growth rate of bacteria would be greater in RMS than in sand. In our trials, RMS bedding samples were not associated with a greater growth rate than sand samples when inoculated with coliforms. However, our in-vitro study presents some limitations and does not exactly reflect the farm environment.

Contamination with animal feces, urine, and milk may be more frequent in a farm environment. Moreover, in our trial, we chose to incubate the inoculated bedding samples at exactly 15°C to approximate an average barn temperature. Cows lying down in a barn, however, generate heat, which could influence bacterial growth. In addition, the temperature in the barn and the bedding is likely to vary during the day and throughout the year.

As expected, the first 24 h of the trial was the period during which an active growth phase could be observed, but only for the sand samples (both E. coli and K. pneumoniae counts) and for the rotary-drum RMS samples (K. pneumoniae counts). It may be of interest in future to monitor bacterial growth more closely, i.e., targeting multiple time points, during the first 24 h, in order to determine whether management practices, such as frequency of removing soiled bedding, would reduce exposure of the cows’ teats to pathogens.

In the future, the methodology used in this study could be applied to samples of RMS bedding prepared using different methods, but from a larger number of farms. This would help in evaluating whether the preparation method, i.e., closed container, heaped, rotary drum, or other methods, affects the initial bacterial load of the bedding and/or its subsequent bacterial growth rate.

In conclusion, samples of RMS bedding prepared using different methods had different initial concentrations of K. pneumoniae and E. coli. When looking at K. pneumoniae counts, inoculated rotary-drum RMS samples and sand samples experienced an active growth phase during the first 24 h of the assay. However, only the sand sample demonstrated an active E. coli growth phase during the first 24 h of the assay. The results of our study suggest that different types of RMS bedding are not all equivalent in terms of initial concentration and growth of coliforms following inoculation.

Funding Statement

This work was funded by Novalait, the Consortium de recherche et innovations en bioprocédés industriels au Québec (CRIBIQ; grant number #2015-044-C17), the Fonds Québécois de la recherche sur la nature et les technologies (grant number #2017-LG-201835), and the Natural Sciences and Engineering Research Council of Canada (grant number #CRDPJ 499421-2016). Annie Fréchette received funding and support from the Natural Sciences and Engineering Research Council of Canada’s Collaborative Research and Training Experience Program in Milk Quality, the Canadian Dairy Commission, Fondation Agria, and Op+Lait.