Evaluation of bedding treatments to inhibit Klebsiella pneumoniae growth in wood shavings and sand Authors: Gabriella Harb, Samuel Rowe, Ruth N Zadoks Published: 2026 ABSTRACT Graphical AbstractSummary: Environmental mastitis pathogens, such as Klebsiella, can proliferate in dairy bedding, yet the influence of bedding type and treatment use on bacterial survival remains unclear. Sand and wood shavings treated with a commercially available bedding conditioner, ZorbiFresh (Daviesway DASCO), or a disinfectant, Doxall (Wintermute Biomedical Australia Pty Ltd.), were examined after inoculation with Klebsiella pneumoniae under controlled laboratory conditions. Although laboratory-based, these results suggest that bedding characteristics and treatment choice influence Klebsiella growth in bedding, highlighting the need for further on-farm investigation. Mastitis remains one of the most common and economically significant diseases for the global dairy industry. Associated costs differ between dairy production systems and may include a loss of up to $39,000 annually in milk quality premiums for a 300-cow herd, along with reduced milk production, discarded milk, culling, increased husbandry time, veterinary treatments, and increased mortality risk (Dairy Australia, 2024). Gram-negative organisms pose a significant challenge due to their potential for severe clinical outcomes. Coliform bacteria, including E. coli and Klebsiella spp., are the most common gram-negative pathogens causing clinical mastitis (Kurban et al., 2022). Klebsiella spp., particularly K. pneumoniae, a very frequently isolated species (Schukken et al., 2012; Massé et al., 2020; Payros et al., 2026), are commonly associated with environmental transmission. Organic material in the dairy environment, including manure, urine, and leaked milk, acts as an important reservoir for transmission of these pathogens to the udder (Munoz et al., 2006; Giannattasio-Ferraz et al., 2022; Song et al., 2023). In recent decades, dairy systems have been shifting from pasture-based to intensive production, accompanied by a shift from contagious and predominantly gram-positive mastitis pathogens to environmental, often gram-negative bacteria (Ruegg, 2017). Globally, the prevalence of mastitis-causing Klebsiella spp. has increased over time (Song et al., 2023). In Australia, K. pneumoniae has frequently been identified in intensive herds through clinical mastitis investigations (Rowe et al., 2024). Unlike predominantly contagious mastitis pathogens, bedding may be an original source of Klebsiella in the case of wood-based products (Bagley et al., 1978) or become contaminated with manure containing Klebsiella (Munoz et al., 2006). The number of bacteria at the teat end is highly correlated with those in cattle bedding (Hogan et al., 1989), emphasizing the importance of bedding and manure management in mastitis control. Despite its importance, both small and large farms face constraints in bedding management. Conventional approaches, such as frequent bedding replacement and stall cleaning, are resource- and labor-intensive, costly, and should be material specific, which may be impractical. Small farms may lack capital or labor flexibility to invest in machinery or high-quality bedding materials (Gargiulo et al., 2018). Larger farms incur higher infrastructure and maintenance costs for handling, processing and disposing of greater bedding volumes (Ferraz et al., 2020), even as the cost per cow declines (Smith et al., 2017). Bedding type influences environmental pathogen load, with studies covering both organic and inorganic bedding materials, including wood-based products, sand, and mattresses (O'Keefe et al., 2013). Organic materials like shavings are readily available, comfortable, and inexpensive, but when wet, they provide an environment conducive to bacterial growth (Hogan et al., 1989). In contrast, materials such as sand are more expensive and labor-intensive, but their low organic matter content means they are considered favorable for udder health (Hogan et al., 1989; Kristula et al., 2005). These materials are classified as inorganic, which can imply that they remain free of organic matter, even during use. Compared with wood shavings, new and recycled sand can reduce the risk of Klebsiella mastitis, but poorly managed sand bedding, whether it be new or recycled when first introduced to cows, may still contribute to exposure of the udder to mastitis pathogens (Kristula et al., 2005; Godden et al., 2008; Pilch et al., 2021). To mitigate bacterial growth in bedding materials, 2 broad categories of products exist: conditioners and disinfectants. Conditioners are specifically used to manage moisture and bacterial growth in dairy bedding through drying agents, antimicrobial additives, microbial and enzyme inoculants, pH buffers, or ammonium binders (Duniere et al., 2024). However, their effectiveness across materials and against Klebsiella spp. is variable (Hogan et al., 2007; Duniere et al., 2024). Conversely, disinfectants have strong bactericidal properties in vitro and are often validated for equipment or surface sanitation (Lajhar et al., 2017), but are rarely evaluated as bedding treatments. We hypothesized that applying a disinfectant and a conditioner bedding treatment to wood shavings and recycled sand would reduce the K. pneumoniae load, given their effectiveness in vitro or against other gram-negative bacteria. Therefore, the primary objective of this study was to evaluate and compare the efficacy of bedding conditioners in wood shavings and sand against K. pneumoniae by quantifying their effect on bacterial load in bedding samples after a single application. From May to September of 2025, this laboratory study assessed the efficacy of a commercially available bedding conditioner and a surface disinfectant. A preliminary medium selection phase was conducted to choose an appropriate selective or differential culture medium for visually identifying K. pneumoniae in the presence of other bacteria. A treatment-experiment phase followed, using the previously selected culture medium to assess K. pneumoniae in sand and wood shavings when treated with disinfectant or bedding conditioner. Experiments were conducted at the University of Sydney's Infectious Disease Laboratory in Camden. For the medium selection phase, various isolates from cattle mastitis origin, including K. pneumoniae (Rowe et al., 2024; n = 10), Citrobacter (n = 6), Enterobacter (n = 1), and Escherichia coli (n = 12), as well as K. pneumoniae of fecal origin (n = 12) were retrieved from the cryopreserved biorepository of the Infectious Disease Laboratory, University of Sydney, Camden. These isolates were selected to compare growth characteristics among phylogenetically related genera and strains that may be present in bedding materials, milk, or manure, and enable assessment of how different media can distinguish Klebsiella spp. from other closely related Enterobacteriaceae. Isolates were initially recovered by streaking onto nonselective, nutrient-rich agar (Mueller-Hinton [MH] agar; Thermo Scientific, Oxoid, USA) with sterile, disposable inoculating loops and incubated aerobically at 37°C for 24 h. Phenotypic appearance of all isolates was then compared on 4 media (listed below) commonly cited in the literature for use in isolating Klebsiella spp. from clinical and environmental samples. This was done to identify the most suitable screening media for detecting Klebsiella in bedding samples, which may harbor mixed bacterial populations. Each recovered isolate was streaked from the MH agar onto MacConkey agar (Thermo Scientific, Oxoid, USA) with ampicillin (MacA; Astral Scientific, Australia), cystine lactose electrolyte deficient agar (CLED; Thermo Scientific, USA), and Simmons citrate agar (Merck, Germany) with inositol (Merck, Germany; SCAI). We also evaluated the combination of SCAI with ampicillin (SCAI + Amp) to determine whether it would improve discrimination of Klebsiella compared with MacA or SCAI alone. A new sterile, disposable inoculating loop was used each time an isolate was transferred from MH to a new plate. All plates were incubated aerobically at 37°C for 24 h. Following incubation, observable differences in colony growth, in terms of the presence and proportion of the plate covered by colonies, were compared across genera and media types. Growth, precipitation, and mucoidy were scored visually on an ordinal scale, where 0 indicated no observable growth, precipitate formation, or mucoid phenotype; 1 indicated colonies covering less than 50% of the plate, precipitate formation around the minority of colonies, or a somewhat mucoid appearance; and 2 indicated colonies covering equal to or more than 50% of the plate, precipitate formation around the majority of colonies, or obvious mucoid appearance of colonies. Colony elevation was described as domed, flat, or pointed based on visual appearance. These assessments were used to compare the ability of each medium to support the growth of Klebsiella isolates (sensitivity) while limiting or differentiating the growth of nontarget genera (including closely related Enterobacteriaceae that share similar colony characteristics) (specificity). Known isolate identities, based on prior laboratory classification, were used as a reference to establish characteristic growth patterns of Klebsiella on each medium for subsequent visual classification. This phase was not intended to provide a formal evaluation of diagnostic sensitivity or specificity, but rather to guide selection of an appropriate medium for the treatment-experiment phase based on relative performance under controlled conditions. Phenotypic features observed in > 90% of Klebsiella isolates and in <10% of nontarget isolates were interpreted as relatively sensitive and specific indicators for Klebsiella detection, respectively. Although this approach does not provide definitive identification of Klebsiella at the colony level, it enables consistent and practical differentiation based on morphology. Performance results, including growth scores (Figure 1a), colony elevation (Figure 1b), and mucoidy (Figure 1c), for CLED, MacA, and SCAI are summarized in Figure 1. For clarity, results for SCAI + Amp are not shown because they were highly similar to those obtained on SCAI across all assessed phenotypic characteristics. Performance results showed that all media supported a high level of growth of K. pneumoniae from milk and feces, with colonies typically covering more than 50% of each plate. In contrast, there was an observable decrease in the percentage coverage by Citrobacter and Enterobacter on SCAI and SCAI + Amp compared with other media, and Escherichia coli did not grow on these media. All E. coli isolates grew on CLED and MacA. (a) Mean growth scores (±SD) of Klebsiella from milk (n = 11) or feces (n = 12), Citrobacter/Enterobacter spp. (n = 14), and Escherichia coli (n = 12) on MacConkey agar with ampicillin (MacA), cystine lactose electrolyte deficient agar (CLED), and Simmons citrate agar with inositol and without ampicillin (SCAI), where 0 = no growth, 1 = growth covering <50% of the streaked area, and 2 = growth covering ≥50% of the streaked area. (b) Number of isolates exhibiting different colony elevation shapes by genus and medium. (c) Number of isolates exhibiting degrees of mucoidy by genus and medium. Precipitate formation was observed only in Escherichia on MacA, providing a simple means of distinguishing this species from Klebsiella, Citrobacter, and Enterobacter. Observable differences in colony elevation varied by genus and medium type. Klebsiella consistently displayed domed colonies across all media, with 95% to 100% of isolates displaying the domed phenotype. This feature was interpreted as a sensitive microbiological indicator for detecting Klebsiella. On SCAI and SCAI + Amp, Citrobacter and Enterobacter isolates also displayed domed colonies, indicating low specificity and the risk of false-positive results on SCAI, regardless of ampicillin addition. In contrast, none of the 19 non-Klebsiella isolates produced the domed phenotype on CLED or MacA, indicating high specificity on these media. Only a subset of Klebsiella isolates formed mucoid colonies (10/22), and this occurred exclusively on CLED, to varying degrees. The absence of mucoid colony formation among non-Klebsiella isolates on CLED demonstrates high specificity for Klebsiella detection. Sensitivity, however, was limited because not all Klebsiella isolates had the mucoid phenotype on CLED. Given these results, MacA was the most practical medium for phenotypic identification of Klebsiella isolates. In the treatment-experiment phase, 3 bedding treatments were evaluated: a conditioner (ZorbiFresh Active [Daviesway DASCO]; >99% [wt/wt] zeolite and <1% [wt/wt] oxone), a disinfectant (Doxall [Wintermute Biomedical Australia Pty Ltd.]; 15% arginine monodecanoate), and an untreated control. These were tested across 2 bedding materials (wood shavings and sand) in a 2 × 3 factorial arrangement. Bedding samples were evenly allocated to treatments after manually mixing the jars to achieve randomization and then assigned numeric codes (1–54). Treatments were applied in blocks within each bedding material. Outcome assessment was not blinded because the investigator applied the visually distinct treatments. Wood shavings were new, and sand was collected from recycled sand piles at a freestall dairy herd. The 6 combinations were tested in technical triplicates (replicate samples within each experimental run) across 3 independent biological replicates (conducted on different days using freshly prepared inocula, manure slurry, and bedding). This was done for E. coli (strain ID: IDL.000492), feces-derived K. pneumoniae (strain ID: MKS 707), and milk-derived K. pneumoniae (strain ID: DA22 0149), for a total of 54 read-outs per strain (6 treatment combinations × 3 technical replicates × 3 biological replicates). Isolates were selected based on their consistent appearance, typical of their species, during the medium selection phase. The E. coli was included as a control strain because the conditioner has proven effectiveness against E. coli (Blue Pacific Minerals, 2025). Fecal and milk K. pneumoniae were included to reflect some of the heterogeneity present in Klebsiella populations. The experiment was repeated once for milk-derived Klebsiella to ensure unexpected culture results were not an artifact, yielding a second set of 54 read-outs. To prepare samples for the treatment-experiment phase, each 60-mL specimen jar received 4.5 g of wood shavings or sand, weighed using a calibrated balance, and 0.5 mL of manure slurry inoculated with E. coli or K. pneumoniae from suspensions standardized to a 0.5 McFarland turbidity standard by visual comparison with a reference. This standard was used to provide a widely accepted, reproducible estimate of bacterial cell density across treatments. The exact concentration of the bacterial suspension was determined using a 10-fold dilution series, with three 10-μL drops of each concentration plated on one-quarter of a MacConkey agar plate without ampicillin (Mac) for E. coli or MacA for K. pneumoniae. The manure slurry was prepared using 8 g of refrigerated dairy cattle feces (freshly voided, obtained from a local dairy farm, with no sterilization performed to simulate natural contamination of bedding) mixed with 8 mL of sterile saline containing one of the test isolates. An additional 8 mL of sterile saline was then added to the slurry to assist in homogenization. The bacterial suspension-manure mixture was shaken until uniform, and 0.5 mL of the mixture was pipetted directly onto the bedding material in each specimen jar, aiming to achieve an even surface distribution before incubation. Conditioner was added to the relevant jars as a dry powder at a rate of 0.6 g, distributed across the surface and base of the bedding layer, representing 10 times the manufacturer's recommended dosage. Disinfectant (15%) was added to the relevant jars using a syringe to evenly distribute 1 mL across the bedding surface, in accordance with the manufacturer's directions, to apply a uniform surface layer. These higher-than-label dosage approaches aimed to ensure that any lack of effect could not be attributed to underdosing, recognizing that additional experimentation with lower doses would be needed in future studies if the desired effects were observed. Control jars contained bedding and manure slurry with the bacterial inoculum, but no bedding conditioner or disinfectant was added. All jars were incubated at 37°C for 18 h at 85% relative humidity as measured using ThermoPro TP50 Digital Hygrometers positioned at the base of the tubs. The outcome of this study was cfu per bedding-sample jar. At 18 h postinoculation, each jar of bedding received 45 mL of PBS prepared using 1 Gibco PBS tablet in 500 mL of MilliQ ultrapure water and was shaken vigorously for 5 s to resuspend the bacteria. A 1:10 dilution was prepared by transferring 100 μL from the bedding suspension into 900 μL of PBS. From this dilution, 100 μL was spread-plated onto Mac for E. coli or MacA for K. pneumoniae for quantification. Plates were incubated aerobically at 37°C for 24 h. The number of colonies with an appearance typical of E. coli (semi-opaque, flat, precipitate-forming, and pink) or K. pneumoniae (shiny, domed, regular-edged, pink, no-precipitate) was counted manually and recorded along with total colony counts. Counts for all plates recorded as “too numerous to count” were capped at 2,500 cfu. This threshold represents the upper countable limit for the plating method and was selected as a tool for retaining biologically relevant samples that reflect high growth in the dataset. Statistical analyses were performed in R (version 2023.03.1+446; R Core Team, 2018). Data were screened for outliers using graphical methods. Because all values were biologically plausible and consistent with experimental variation, all data points were retained for analysis. An analysis log is available at http://rpubs.com/gabriellaharb/1360253. Sample size for the experimental phase was estimated using the “epi.sscompc” function in the epiR package (Stevenson, 2025). Variance estimates were derived from a field dataset of used bedding samples (n = 160; Rowe et al., 2019), with a pooled standard deviation of 1.3 log10 units as a conservative estimate. Assuming a minimum detectable difference of 1 log10 unit between treatment and control (α = 0.05, 80% power), 3 independent experimental units per treatment group were required within each bedding type (biological replicates), with each unit measured in triplicate as technical replication. Analyses were initially stratified by isolate and by species. Data were assessed for normality and homogeneity of variance using the Shapiro-Wilk and Levene's tests. Natural log transformations of colony count data [log(x + 1)] were used to improve normality as assessed by quantile-quantile plots and handle values of zero. Therefore, log-transformed data were used for model fitting. A linear mixed-effects model (LMM) was fitted using the lme4 package (Bates et al., 2015), with bedding material (wood shavings or sand), treatment (control, conditioner, or disinfectant), and their interaction as fixed effects. Biological experiment replicate was included as a random effect to account for between-replicate variability. Model diagnostics were assessed using the performance package (Lüdecke et al., 2021). Using the emmeans package (Lenth, 2016), model-predicted colony counts were generated and back-transformed to the response scale. These represent the model-estimated means for each treatment combination and summarize treatment effects while accounting for data variability. For each isolate or species, pairwise comparisons between conditioners and bedding types were conducted with Tukey-adjusted P-values. A sensitivity analysis was conducted to evaluate the robustness of the K. pneumoniae model assumptions and the stability of results. This included assessing LMM performance when the upper limit on colony counts was increased to 5,000 and 10,000. A negative binomial generalized linear mixed model (GLMM) was also tested. Alternative model diagnostics and outputs were compared with those of the original LMM model to evaluate the consistency of treatment effects. Significant conditioner and disinfectant effects on the number of E. coli and K. pneumoniae colonies were detected in both bedding types. Model-predicted colony counts for both K. pneumoniae strains were broadly similar across bedding types and treatments (see analysis log), so observations were pooled at the species level. For E. coli, the conditioner reduced counts in wood shavings, consistent with prior laboratory evidence of efficacy, whereas it showed little evidence of effect in sand. The disinfectant reduced E. coli counts in sand but not in wood shavings. For K. pneumoniae, both treatments reduced counts in sand. However, in wood shavings, the conditioner showed no evidence of effect, and the disinfectant increased counts. Detailed statistical comparisons are provided in the following text. In wood shavings, the conditioner supported significantly lower predicted E. coli colony counts (6.1, 95% CI: 0.5 to 32) than either the control (1,507, 95% CI: 324 to 6,993; P < 0.001; Figure 2a) or the disinfectant (365, 95% CI: 77 to 1,685; P < 0.001; Figure 2a). The predicted control and disinfectant counts did not differ significantly in wood shavings (P = 0.097; Figure 2a). In sand, the disinfectant yielded the lowest predicted E. coli colony counts (1.7, 95% CI: −0.4 to 12), which were significantly lower than those for the control (2,500, 95% CI: 538 to 11,598; P < 0.001; Figure 2a) and the conditioner (1,769, 95% CI: 381 to 8,209); P < 0.001; Figure 2a). Compared with the control, the conditioner did not significantly reduce E. coli counts (P = 0.865; Figure 2a). Model-predicted Escherichia coli (a) and Klebsiella pneumoniae (b) colony counts across bedding treatment and bedding material groups. Points represent model-predicted means, and whiskers indicate 95% CI. Brackets denote significant pairwise differences between treatments based on linear mixed models with Tukey adjustment for multiple comparisons. For K. pneumoniae, the disinfectant supported significantly higher predicted colony counts (1,254, 95% CI: 338 to 4,657) than both the control (214, 95% CI: 57 to 795; P < 0.001; Figure 2b) and the conditioner (164, 95% CI: 43 to 612; P < 0.001; Figure 2b) in wood shavings. Predicted control and conditioner counts did not differ significantly in wood shavings (P = 0.779; Figure 2b). In sand, the disinfectant yielded the lowest predicted colony counts (3.5, 95% CI: 0.2 to 15.7), which were significantly lower than those for the control (475, 95% CI: 127 to 1,765; P < 0.001; Figure 2b) and conditioner (157, 95% CI: 42 to 587; P < 0.001; Figure 2b). Compared with the control, the conditioner also significantly reduced K. pneumoniae counts in sand (P = 0.015; Figure 2b). The model sensitivity analysis performed for the K. pneumoniae model showed that increasing the colony count upper limit from 2,500 to 5,000 or 10,000 did not alter the direction or significance of treatment effects. The adjusted intraclass correlation coefficient (ICC) for the main model was 0.415, indicating that 41.5% of the variation in K. pneumoniae colony counts was due to differences between experimental replicates, and suggests a moderate level of agreement in terms of how consistently the model distinguishes between groups (bedding treatments and bedding materials). This value remained similar for the higher-upper limit models, with an adjusted ICC of 0.425 when the limit was 5,000 and 0.428 when the limit was 10,000. Results of the negative binomial GLMM were largely consistent with the log-transformed LMM for K. pneumoniae, whereby the comparison between the conditioner and the control in sand bedding yielded a P-value just above 0.05 in the GLMM (P = 0.057), while results were significant in the LMM (P = 0.015). The results of this study only partially support the hypothesis that the conditioner and disinfectant would reduce bacterial growth in the bedding materials. The effects of treatments were highly dependent on the bedding substrate. In addition, treatment effects were not consistent between E. coli and Klebsiella. Wood-based substrates have previously been shown to support the growth of Klebsiella spp., particularly under wet conditions (Hogan et al., 1989; Zdanowicz et al., 2004). Organic matter in wood shavings readily absorbs moisture from biological materials. This retained moisture acts as a carrier for dissolved acids, alkalis, and organic compounds, allowing them to diffuse through the bedding matrix and create buffered microenvironments that dampen rapid pH shifts to different degrees (Munir et al., 2019). Weak acid-based disinfectants, such as the disinfectant used in this study, rely on environmental pH to remain in their nonionized and lipid-soluble form. Under acidic conditions, this chemical state enables diffusion of the acid across bacterial cell membranes and disruption of cellular functions (Lambert and Stratford, 1999), but when in a moist, buffered bedding environment, the effect may be dampened. This may explain why the disinfectant did not reduce the growth of E. coli or K. pneumoniae in wood shavings. The increased growth of K. pneumoniae wood shavings treated with disinfectant was unexpected, which is why the experiment with milk-derived K. pneumoniae was repeated. Results were consistent across both iterations and across K. pneumoniae from feces and milk, implying this is a reproducible effect rather than an erroneous outcome. However, because cfu was the outcome measured in this study, the observed increase may reflect not only bacterial proliferation but also changes in bacterial recovery from the bedding substrate. It is not clear how growth is enhanced, although increased access to nutrients within wood shavings due to exposure to disinfectant could, in theory, be part of the explanation. Acidic conditions have been shown to weaken wood structure and promote hemicellulose breakdown (Usino et al., 2023), potentially making nutrients more accessible to bacteria such as K. pneumoniae. In commercial sand-bedded systems, acidifying agents have been applied as disinfectants (Kristula et al., 2008). In the present study, the disinfectant was highly effective in reducing model-predicted E. coli and K. pneumoniae counts in sand. This finding is biologically plausible, as sand is a dry, nutrient-poor substrate that resembles the hard surfaces for which disinfectants are typically designed. In contrast to wood-based bedding, sand provides limited protection to bacteria through organic matter or moisture retention, potentially enhancing disinfectant efficacy. However, in interpreting these results, it must be noted that the disinfectant was applied at a concentration substantially higher than standard commercial recommendations, and the recycled sand used was visually clean before the addition of manure containing E. coli or K. pneumoniae. This approach, using recycled sand, was selected to better reflect on-farm conditions, where bedding is rarely sterile. Under field conditions, sand bedding may contain organic matter before use, with continual deposition of manure, urine, and milk once cows are present. This organic load can markedly reduce disinfectant performance (Rowbotham and Ruegg, 2016). Therefore, although disinfectant application was effective under controlled laboratory conditions, its efficacy in commercial sand-bedded systems may be substantially lower and requires further validation under practical farm conditions. This study clearly demonstrated that treatments do not affect all coliforms equally. Consistent with prior laboratory testing (Blue Pacific Minerals, n.d.), the conditioner reduced E. coli counts in wood shavings, confirming that the bedding inoculation and incubation method used here can replicate known conditioner responses. However, the conditioner had no effect on K. pneumoniae in the shavings. Commonly, K. pneumoniae possesses a thicker polysaccharide capsule than E. coli (Amako et al., 1988; Phanphak et al., 2019; Singh et al., 2022), which can improve the persistence of the bacteria in nutrient-rich, organic materials like wood shavings (Xu et al., 2024) and provides protection from active ingredients in bedding treatments (Wei et al., 2024). The difference observed for the disinfectant treatment in wood shavings, with no effect on E. coli and enhanced growth of K. pneumoniae, are likely to have a different biological basis. Klebsiella spp. naturally inhabit soil, water, and plant matter, so they are adapted to use plant-derived carbon (Bagley, 1985). For example, because of the enzymatic expression of cellulase-like hydrolases, K. pneumoniae can use a broad range of these wood-derived carbohydrates (Barbosa et al., 2020). In contrast, E. coli is primarily enteric and does not metabolize complex, lignin-associated carbohydrates efficiently (Pang et al., 2020), which may be why counts in wood shavings with the disinfectant were only increased for K. pneumoniae. This aligns with reports that outbreaks of Klebsiella mastitis have repeatedly been linked to wood-based bedding. (Hogan et al., 1989; Sampimon et al., 2006), whereas E. coli is often associated with exposure to moisture and manure (Freu et al., 2023). This study employed a controlled inoculation approach with a replicated design. However, several limitations must be acknowledged. All bacterial suspensions for bedding inoculation were prepared to the 0.5 McFarland standard by visual comparison with a McFarland reference, ensuring consistent turbidity and providing biologically comparable inocula for relative treatment comparisons. Although visual McFarland standardization is widely used, it does not necessarily reflect actual bacterial loads on farms and does not guarantee identical starting concentrations in vitro. Visual assessment of turbidity relative to the colorless standard can be influenced slightly by culture characteristics, such as colony color, and human perception. This was accounted for in the LMM so that treatment comparisons were made after adjusting for concentration differences. The bedding inoculation model was validated with E. coli following the main K. pneumoniae experiments. Although validation with E. coli supported the biological relevance of results, performing these validation steps alongside the K. pneumoniae experiments would have reduced the risk of undetected procedural weaknesses influencing treatment effects. Applying conditioners at concentrations higher than recommended to bedding in small specimen jars also does not accurately reflect true farm bedding management practices. As this study was conducted under controlled in vitro conditions, the extent to which these findings translate to commercial dairy systems remains uncertain. Therefore, field studies are required to confirm the presence of these effects under farm-specific conditions. Ideally, such studies would incorporate frequent bedding disturbance, contamination with milk, feces, and urine, more frequent conditioner application rates, and fluctuations in temperature and humidity. Additionally, the LMM approach provided robust estimates of treatment effects, but the GLMM sensitivity analysis revealed a minor difference in significance patterns. The inherent variability in bacterial count data is reflected in this result, suggesting that although the trends are robust, distributional assumptions between models may influence pairwise contrasts to a minor degree. In conclusion, the efficacy of bedding treatments is not universal and depends on the bedding material and the targeted bacterial species. Conditioner and disinfectant both reduced counts of K. pneumoniae in sand. By contrast, the conditioner had no effect on K. pneumoniae counts in wood shavings, whereas the disinfectant even increased counts. These findings highlight the importance of targeted bedding management strategies that match products to specific substrates and pathogens rather than assuming uniform efficacy. Further studies would be required to test these interactions under farm-specific conditions and determine the mechanisms behind bacterial growth inhibition or improvement. This study provides evidence that certain bedding treatments have the potential to limit bacterial proliferation when matched correctly to the substrate and pathogen, but that results cannot necessarily be extrapolated from one bacterial genus to another or from one substrate type to another. Notes This study received no external funding. This article did not involve studies with human or animal subjects and did not require Institutional Animal Care and Use Committee or Institutional Review Board approval. The authors have not stated any conflicts of interest. Nonstandard abbreviations used: CLED = cystine lactose electrolyte deficient agar; GLMM = generalized linear mixed model; ICC = intraclass correlation coefficient; LMM = linear mixed-effects model; Mac = MacConkey agar without ampicillin; MacA = MacConkey agar with ampicillin; MH = Mueller-Hinton; SCAI = Simmons citrate agar with inositol; SCAI + Amp = Simmons citrate agar with inositol and ampicillin. REFERENCES