https://orcid.org/0000-0002-9179-9860
Abstract
House flies (Musca domestica L.) are filth-breeding pests of urban and rural environments around the world. Frequenting microbe-rich substrates for nutritional and reproductive needs, house flies pose a risk to human and animal health through their carriage and transmission of pathogenic and antimicrobial resistant bacteria (AMR). Adult house flies were collected from Kansas beef and dairy cattle operations to assess factors influencing bacterial abundance and AMR incidence flies. Aerobic culturable bacteria and suspected coliforms (SC) were enumerated from fly homogenate cultured on nonselective (tryptic soy agar) and selective (violet-red bile agar VRBA) media, respectively. Unique morphotypes of SC isolates were screened for tetracycline resistance and tested for resistance to 4 additional antibiotics to identify multi-drug resistant (MDR) isolates. Female house flies carried greater abundances of both culturable bacteria and SC than male flies. Abiotic factors such as ambient and soil temperatures correlated with culturable bacteria and SC abundances in flies, but farm type correlated only with SC abundance and trends of resistance phenotypes observed in SC isolates. Male and female flies from both farm types carried one or more AMR and MDR SC isolates (73.02% AMR and 31.09% MDR). The majority of AMR and MDR bacteria were Escherichia/Shigella sp., which possessed the widest range of phenotypic resistance variability found in our study. Our results further emphasize the role house flies play in harboring bacteria of risk to human and animal health and identified factors of potential use for the development of strategies to mitigate house fly transmission of bacterial pathogens and AMR within confined cattle operations.
Introduction
House flies are worldwide pests of humans and animals that opportunistically utilize a variety of microbe-rich substrates for nutritional and developmental needs. Over 200 different species of microbes have been identified in wild house fly surveys (Nayduch and Burrus 2017), and a single fly can carry up to 100 different potential pathogens (Greenberg 1973). House fly interactions with microbe-rich substrates begin immediately after hatching as larvae require microbes in their diet to develop properly (Schmidtmann and Martin 1992, Zurek et al. 2000). Microbe-fly interactions continue into adulthood with house flies frequently exploring microbe-rich substrates for nutrition and, in the case of females, oviposition (Shah et al. 2016). Environments which produce copious amounts of microbe-rich substrates, like manure produced from confined cattle operations, are both conducive to house fly proliferation and rich sources of microbes for fly acquisition during oviposition. The persistent quest for oviposition sites may drive females to have more frequent exposure to microbe-rich substrates, resulting in greater bacterial loads than male flies (Neupane et al. 2020).
Antimicrobial resistance (AMR) in bacterial pathogens is an increasing concern for livestock producers, especially those keeping animals in confined spaces such as dairies and feedlot operations. As one of the largest consumers of antibiotics important for both human and animal health (FDA-CVM 2021), cattle industries are at risk of losing a cost-effective tool for combatting common bacterial pathogens that impact animal wellness and production yields. Surveillance of reservoirs of AMR bacteria and antimicrobial resistance genes (ARGs) within these operations is important for developing management strategies to efficiently mitigate the development and spread of AMR. Researchers have identified numerous AMR bacteria and pathogens in environmental sources associated with livestock production, such as cattle manure, feed, and water (Guo et al. 2021, Ma et al. 2021). Despite current efforts to manage these substrates and limit antibiotic usage in livestock, AMR remains a rising threat to human and animal health (Naghavi et al. 2024). Therefore, methods to monitor AMR incidence, emergence, and prevalence are still urgently needed to mitigate this threat.
Insects, particularly house flies, are believed to bridge AMR bacteria between environmental sources in livestock production and, as such, could help assess AMR prevalence in these facilities (Zurek and Gosh 2014, Nayduch and Burrus 2017, Onwugamba et al. 2018, Nayduch et al. 2023). Although house flies frequently interact with microbe-rich substrates within the environment and actively proliferate within large-scale cattle operations, information on their role in the prevalence and transmission of AMR bacteria is limited. Flies have been shown to carry and transmit pathogenic and AMR bacteria, including from cattle manure to fresh produce, in laboratory assays (Talley et al. 2009, Thomson et al. 2021). Additionally, AMR bacteria have been isolated from house flies collected at livestock operations, with a single fly capable of carrying multiple species of AMR bacteria (Chakrabarti et al. 2010, Onwugamba et al. 2018, Neupane et al. 2020, Geden et al. 2021). However, prior studies have focused on house fly carriage of limited AMR bacterial species of interest, particularly pathogens. Many non-pathogenic bacteria, including flora such as coliforms, can serve as rich sources of ARGs that can be horizontally transferred to other bacterial species (D’Costa et al. 2006, Wellington et al. 2013). Biotic and abiotic drivers affecting the house fly’s carriage and dissemination of AMR bacteria, particularly non-pathogens, in livestock production settings is unknown. Using a culture-based approach, we evaluated the overall abundance of bacteria and the prevalence of suspected coliforms, including those with AMR, carried by house flies in confined beef and dairy cattle production environments. Additionally, biotic and abiotic factors, such as fly sex, farm type, location, collection date, and climate, were investigated to assess their potential influence on bacteria and AMR prevalence and abundance in house flies.
Materials and Methods
Sample Collection
Adult house flies (Musca domestica L.) were collected from a dairy and a beef feedlot cattle operation in each of 3 counties in northeastern Kansas, USA (Riley, Marion, and Washington). Beef and dairy operations in each county were within 25 km of each other, and facility capacity, total number of animals present, feed type, and manure management strategies were recorded for each site during each visit. Flies were sampled from all sites on the same collection date every other week beginning in mid-August and ending in early October of 2019. Historical weather data for climate variables measured hourly by the Manhattan, Rock Springs, and Washington (Kansas Mesonet 2021) weather stations were recorded for beef and dairy sites within Riley, Marion, and Washington counties, respectively (Supplementary Table S1).
Using disinfected sweep nets soaked overnight in 10% bleach, house flies were collected randomly from animal feed bunks in front of 2 to 3 pens at each site and placed into a sterile petri dish on ice for immobilization. Six male and 6 female flies per facility and collection date (n = 352 total flies) were placed individually into sterile 1.5 ml microcentrifuge tubes using gloves and forceps. Forceps were disinfected by immersion in 70% ethanol between samplings. Tubes containing flies remained on ice during transport to the laboratory and were processed within 8 h of collection.
Bacterial Culture and Enumeration
In the laboratory, each fly was homogenized with a sterile plastic pestle in 500 µl of sterile phosphate-buffered saline (PBS; 1.37 M NaCl, 27 mM KCl, 101.4 mM NaH2PO4, 17.6 KH2PO4, pH 7.4; Fisher Scientific, Pittsburg, PA, USA). After homogenization, an additional 500 µl of PBS was added and the sample was briefly vortexed. Homogenate was 10-fold serially diluted and 100 µl of each dilution was spread-plated in duplicate on tryptic soy agar (TSA; Becton Dickinson and Company, Sparks, MD, USA) for culturable aerobic bacteria enumeration, and in duplicate on violet-red bile agar (VRBA; Becton Dickinson and Company, Sparks, MD, USA) for coliform enumeration. Culture plates were incubated at 37 °C for 18 to 24 h, after which colony forming units (CFUs) were enumerated for all plated dilutions on TSA and VRBA. Only pink or red CFUs indicating selection of lactose-fermenting Gram-negative bacteria were counted on VRBA (hereafter “suspected coliforms (SC)”).
Selection, Susceptibility Testing, and Identification of Suspected Coliform Morphotypes
Suspected coliforms (observed lactose fermentation) with differing morphologies (hereafter “SC isolates” representing SC morphotypes used for AMR testing) from each collection date were picked from VRBA cultures with a sterile inoculating loop, quadrant streaked for isolation onto a new VRBA plate and incubated at 37 °C for 24 h. Prior to storage, a single colony from each SC morphotype was quadrant streaked onto TSA and incubated at 37 °C for 24 h. Subsequently, a colony from TSA was selected and used to inoculate 5 ml tryptic soy broth (TSB; Becton Dickinson and Company, Sparks, MD, USA), which was incubated at 100 rpm at 37 °C for 4 to 6 h. Cultures were cryo-preserved as 25% glycerol stocks and stored at −80 °C until resuscitation for downstream identification and AMR susceptibility testing.
To initially screen SC isolates for suspected tetracycline resistance, glycerol stocks of SC isolates were partially thawed at room temperature before adding 100 µl to 5 ml of TSB and incubating for 6 to 8 h at 37 °C and 100 rpm. A loop of culture was streaked onto VRBA and incubated for 18 to 24 h at 37 °C. Isolated colonies from each sample were subcultured onto TSA and incubated at 37 °C for another 24 h to confirm purity. An isolated colony was streaked onto a quadrant of Mueller-Hinton (MH; Becton Dickinson and Company, Sparks, MD, USA) agar supplemented with tetracycline (16 µg/ml) which is the CLSI MIC breakpoint concentration for resistance in Enterobacterales (CDC 2019). SC isolates were considered potentially tetracycline-resistant if growth was observed after incubation for 24 to 48 h at 37 °C and selected for later disk diffusion susceptibility testing. Potentially tetracycline-resistant SC isolates were subcultured onto TSA before creating a new glycerol culture stock.
SC isolates which grew on tetracycline-infused MH agar were later resuscitated for disk diffusion to confirm tetracycline resistance via CLSI standards and investigate susceptibility to 4 other antibiotics of veterinary importance using the Kirby-Bauer disk diffusion method (Hudziki 2009). In preparation for antibiotic susceptibility testing, glycerol stocks were cultured onto VRBA, followed by streaking for isolation onto TSA. Isolated colonies from TSA were added to sterile saline and compared with a 0.5 Remel McFarland Equivalence Turbidity Standard (Remel, Lenexa, KS, United States) to achieve desired inoculum turbidity. Saline inoculums were spread onto a MH (Remel, Lenexa, KS) plate before adding a disk of each of the 5 following antibiotics: 16 µg tetracycline (Becton Dickinson and Company, Sparks, MD), 30 µg florfenicol (Nufluor; VetLab, Palmetto Bay, FL, USA), 5 µg enrofloxacin (Baytril; VetLab, Palmetto Bay, FL, USA), 10 µg ampicillin (VetLab, Palmetto Bay, FL, USA), and 30 µg ceftiofur (Naxcel; VetLab, Palmetto Bay, FL, USA). Plates were incubated at 37 °C for 18 h and the zones of inhibition measured (mm) for each antibiotic. Zones were interpreted for susceptibility utilizing the 2020 Clinical and Laboratory Standards Institute (CLSI) VET01S Enterobacterales zone diameter breakpoints (CLSI 2020).
Identification of Antimicrobial Resistant Suspected Coliform Isolates
Single- or multi-drug resistant SC isolates were identified to genus via 16S rRNA Sanger sequencing. Isolates were resuscitated from 50 µl of glycerol stock onto VRBA and TSA as outlined above for tetracycline resistance screening methods. An isolated colony was picked from the TSA with a sterile toothpick and added to 50 µl of nuclease free water inside a 0.2-ml 96-well plate. The plate was boiled at 100 °C for 10 to 15 min in a thermal cycler (Bio-Rad T100 Thermal Cycler, Life Science, Hercules, CA, USA). Tubes were centrifuged for 3 to 5 min at 13,000 RPM to pellet cell debris, after which 20 µl of supernatant was stored as DNA template. Using universal primer pair 8F and 806R targeting the 16S rRNA gene (Relman et al. 1992, Turner et al. 1999), PCR reaction mixtures were prepared and 16S rRNA gene amplification performed as described previously (Neupane et al. 2019). PCR products were sent for Sanger sequencing by the University of Arizona Genetics Core (Tucson, AZ, USA). Bases were called and reads assigned quality scores by analyzing the chromatographs with phred (Ewing and Green 1998) and CodonCode Aligner v.10.0 (CodonCode Corporation, www.codoncode.com) using the standard default parameters. Quality trimming of reads was performed using fastp v.0.22.0 (Chen et al. 2018) with default sliding window of 4 and front mean and cuttail mean quality scores of 25. Trimmed reads were then assigned to the genus level with ≥80% confidence threshold using Ribosomal Database Project (RDP) Classifier (Wang et al. 2007). Any remaining SC isolates with reads which were unassignable for failure to meet the above standards were instead identified using matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) at the Kansas State University Veterinary Diagnostic Laboratory (Manhattan, KS, USA). Open access to the repository hosting our workflow code for classifying Sanger sequences of 16srRNA gene is available online (https://github.com/vlpickens04/Sanger_Phred_Code).
Statistical Analysis
Serial dilution plates with CFUs in the range of 30 to 300 CFUs were enumerated and included in the quantitative data. Number of CFUs were averaged between plated dilutions and replicates of fly samples (n = 176 house flies/farm type; n = 352 total flies). For some samples, no dilutions yielded colony growth within the relevant range for either plated replicate, with colonies either too numerous to count (TNTC; > 300 CFUs) or too few to count (TFTC; < 20 CFUs). In these cases, samples underwent the following adjustments. Samples where dilutions skipped from TNTC down to TFTC were assigned 301 CFUs for the highest dilution at which TNTC was recorded. Samples with dilutions that jumped from TFTC to no growth were assigned 29 CFUs for the highest dilution with TFTC assigned. Only samples which yielded no growth for all plated dilutions were assigned 0 CFUs so that samples were not misrepresented as flies without any culturable bacteria. For all samples, CFUs from each replicate were averaged for cultures on TSA and VRBA and log10 transformed before statistical analysis.
All statistical analyses were performed in JMP Pro Statistical Software (JMP, Version 16.2, SAS Institute Inc., Cary, NC, USA). Log-transformed CFUs were analyzed using the linear regression and ANOVA approaches in the Standard Least Squares Personality of Fit Model to initially investigate the effect of fly sex, farm type, county, collection date, and their interactions on culturable bacteria and SC abundances in house flies. Pairwise post hoc comparisons of least squares means estimates were calculated for the comparisons of fly sex across farm types and locations (farm type × county) on bacterial abundances, as well as the comparisons of collection date and the interaction of fly sex and collection date using the Tukey honestly significant difference (HSD) test.
The effect of collection date on culturable bacteria and SC abundances carried by house flies and their possible associations with climate variables were additionally explored. First, an initial variable reduction was performed by deriving principal components of the measured climate variables and bacterial enumerations via the principal component analysis (PCA), multivariate methods platform in JMP 16.2. Original variables observed in PCA to exhibit large loading values in the first 2 principal components were reserved and used as regressors in backwards regression via the least-squares means platform in JMP. Further variable reduction occurred in backwards regression step with removal of variables where P ≤ 0.05, resulting in a final set of variables for prediction model exploration. Collection date and county were excluded from the regression model to reduce redundancy, as the same weather data was used for sites within the same county on the same collection date and were therefore confounding variables to the model. Fixed effects of fly sex, farm type, and their interactions were combined with selected continuous climate variables and used to predict culturable bacterial and SC abundances of house flies.
Contingency Analysis (Fit Y by X platform, JMP 16.2) was used to further explore the associations between SC abundance and antimicrobial resistance of SC isolates from house flies. Log transformed SC abundance data for each fly was assigned a category based upon the ranges of SC abundance observed for house flies in a zero-inflated Poisson distribution (1 = 0.00 to 0.50, 2 = 0.51 to 2.90, 3 = 3.00 to 3.99, 4 = 4.00 to 4.99, 5 = 5.00 to 5.99, 6 = ≥ 6.00; in Log10(CFU + 1) per fly). Flies possessing SC abundance in category 1 were removed to eliminate confounding interpretations for the dependence of SC isolate resistance profiles in house flies since this category was created for the outlying flies carrying 0 or too few CFUs for proper quantification (which were assigned a CFU value of 29), and no SC isolates were taken from these fly samples to assess AMR despite bacterial growth on culture plates. House flies from which SC isolates were tested for antimicrobial susceptibility were assigned to 2 different categories: (i) the number of antibiotics to which SC isolates the fly carried were resistant and (ii) the type of antibiotics to which SC isolates carried by the fly were resistant. To investigate the association between mean abundance and AMR characteristics of SC carried by house flies, the relative frequency of house flies carrying each category of SC abundance and SC isolates belonging to the number and type of antibiotic resistance categories was discerned by the Fisher’s Exact Test. For all statistical tests, P ≤ 0.05 was considered significant.
Results
Factors Affecting Culturable Bacteria and Suspected Coliform Abundances in House Flies
Culturable bacteria and SC were enumerated from 352 adult house flies for this study (n = 176 house flies/farm type). Irrespective of farm type, collection date, county, or fly sex, culturable bacterial CFUs ranged from 3.35 × 103 to 3.01 × 107 CFUs/fly (mean ± SEM: 1.57 ± 0.20 × 106 CFUs/fly) and SC CFUs ranged from 0 to 3.01 × 107 CFUs/fly (mean ± SEM: 1.91 ± 0.25 × 105 CFUs/fly). Overall, female house flies carried significantly greater abundances of both culturable bacteria (mean ± SEM: 2.38 ± 0.34 × 106; F = 83.48, df = 1, P < 0.0001) and SC (mean ± SEM: 3.05 ± 0.47 × 105; F = 52.34, df = 1, P < 0.0001) than male flies (culturable bacteria mean ± SEM: 7.84 ± 2.05 × 105; SC mean ± SEM: 7.84 ± 1.64 × 104). No observable growth of SC CFUs was recorded for 14 house flies, of which 11 were male house flies. Comparisons of culturable bacteria and SC abundances in male and female flies from each location are summarized in Supplementary Table S2. An overall effect of county and collection date was also observed on abundances of both culturable bacteria (county: F = 17.45, df = 2, P < 0.0001; date: F = 11.20, df = 4, P < 0.0001) and SC (county: F = 9.27, df = 2, P = 0.0001; date: F = 17.46, df = 4, P < 0.0001) in house flies. Farm type did not significantly affect culturable bacterial abundance in house flies (F = 0.64, df = 1, P = 0.42) but SC abundance was significantly greater in flies from beef operations than flies from dairies (F = 5.92, df = 1, P = 0.02).
The interaction between collection date and county significantly affected the mean abundances of both culturable bacteria and SC in house flies (culturable: F = 4.00, df = 8, P = 0.0002; SC: F = 2.39, df = 8, P = 0.02) and the interaction between collection date and location (farm type × county) also had a significant effect (culturable: F = 4.64, df = 8, P < 0.0001; SC: F = 3.75, df = 8, P = 0.0003). The abundance of culturable bacteria recovered in house flies also was significantly affected by interactions between collection date and fly sex (F = 7.19, df = 4, P < 0.0001), county and fly sex (F = 3.32, df = 2, P = 0.04), and county, fly sex, and location (farm type × county) (F = 2.77, df = 8, P = 0.006). In contrast, only the interaction of collection date, county, and fly sex (F = 2.17, df = 8, P = 0.04) significantly affected SC abundance of house flies. No overall location effect (farm type × county) was observed for culturable bacteria (F = 1.95, df = 2, P = 0.14) or SC (F = 0.80, df = 2, P = 0.45) abundance.
The interaction between farm type and fly sex did not significantly influence the abundance of culturable bacteria (F = 2.85, df = 1, P = 0.09) or SC (F = 3.65, df = 1, P = 0.06) in house flies. While mean SC abundance was significantly greater in male house flies from beef versus dairy sites, mean culturable bacteria abundance in males did not differ between farm types (Fig. 1A and C). Female house fly mean culturable bacteria and SC abundances did not significantly differ between beef or dairy farms but were greater than male flies both within and between farm types (Fig. 1A and C). Within each location, the culturable bacteria and SC abundances in female house flies was not always significantly greater than male flies (Fig. 1B and D). Within collection date, abundances of culturable bacteria and SC in house flies overall (ie irrespective of sex) were significantly lower in flies collected on later dates, when lower average ambient and soil temperatures were observed (Fig. 2A and C). However, within fly sex, significant differences in culturable bacteria and SC abundances did not always follow this same trend across the collection dates (Fig. 2B and D). Female house flies generally carried greater culturable bacteria and SC abundances than male flies within collection dates, although these differences were not always statistically significant (Fig. 2B and D). Female house flies carried significantly more culturable bacteria than male flies on 3 of the 5 collection dates (Aug 13, Aug 27, and Sep 24) and greater abundance of SC on 2 of the 5 collection dates (Aug 13 and Sep 24).
Mean abundance of culturable bacteria (grown on TSA, see text) and suspected coliforms (SC; grown on VRBA, see text) in male and female house flies within farm types (A, C) and locations (B, D). Differing letters in (A) and (C) indicate significant differences in pairwise comparisons of least square means (Tukey HSD, P < 0.05). For pairwise comparisons of least square means in (B) and (D): * P < 0.01, ** P < 0.001, *** P < 0.0001. Box and whisker graphic overlay of the mean abundance values provides evaluation of the full range of CFUs, where the diamonds are the raw mean, solid middle bars are the median values, and the bars extending from the center represent the interquartile range.
Relationship between abundance of culturable bacteria (grown on TSA, see text) and suspected coliforms (SC; grown on VRBA, see text) in flies irrespective of sex (A, C) and delineated by sex (male and female; B, D) with ambient and soil temperatures. Raw mean ± SEM is shown. Differing letters indicate significant differences in pairwise comparisons of least square means (Tukey HSD, P < 0.05).
Environmental Predictors for Mean Bacterial Abundances Carried by House Flies
Following reduction via PCA and backwards regression to remove insignificant variables (P > 0.05; Supplementary Fig. S1), a least square fit of the multivariate prediction generated models for predicting the mean abundances of culturable bacteria and SC in house flies using the parameters outlined in Table 1. Separate models were generated for predicting the abundance of culturable bacteria and SC in house flies based upon the listed parameters.
Climate variables used for prediction plots of factors influencing mean abundance of culturable bacteria and suspected coliforms carried by house flies.
| Parameter | Estimate | Standard Error | t Ratio | Prob > |t| |
|---|---|---|---|---|
| Culturable Bacteria | ||||
| Intercept | −0.6554 | 0.9883 | −0.66 | 0.5076 |
| Fly Sex [Female] | 0.3498 | 0.0436 | 8.02 | < 0.0001 |
| Collection Soil Temp Avg | −0.2526 | 0.0572 | −4.42 | < 0.0001 |
| Collection Avg Temp | 0.2618 | 0.0485 | 5.40 | < 0.0001 |
| 3D Avg Humidity | 0.0732 | 0.0150 | 4.87 | 0.0038 |
| 3D Avg Precipitation | 0.3498 | 0.0436 | 8.02 | <0.0001 |
| Suspected Coliforms | ||||
| Intercept | −1.4365 | 1.4127 | −1.02 | 0.3099 |
| Farm Type [Dairy] | −0.1414 | 0.0600 | −2.35 | 0.0191 |
| Fly Sex [Female] | 0.4100 | 0.0600 | 6.83 | < 0.0001 |
| Collection Avg Temp | 0.2852 | 0.0537 | 5.31 | < 0.0001 |
| 3D Days of Precipitation | 1.6464 | 0.4864 | 3.38 | 0.0008 |
| 3D Avg Humidity | 0.0856 | 0.0271 | 3.16 | 0.0017 |
| 3D Avg Precipitation | 0.0361 | 0.0076 | 4.71 | <0.0001 |
| 1W Avg Soil Temp | −0.3757 | 0.0899 | −4.18 | < 0.0001 |
| Parameter | Estimate | Standard Error | t Ratio | Prob > |t| |
|---|---|---|---|---|
| Culturable Bacteria | ||||
| Intercept | −0.6554 | 0.9883 | −0.66 | 0.5076 |
| Fly Sex [Female] | 0.3498 | 0.0436 | 8.02 | < 0.0001 |
| Collection Soil Temp Avg | −0.2526 | 0.0572 | −4.42 | < 0.0001 |
| Collection Avg Temp | 0.2618 | 0.0485 | 5.40 | < 0.0001 |
| 3D Avg Humidity | 0.0732 | 0.0150 | 4.87 | 0.0038 |
| 3D Avg Precipitation | 0.3498 | 0.0436 | 8.02 | <0.0001 |
| Suspected Coliforms | ||||
| Intercept | −1.4365 | 1.4127 | −1.02 | 0.3099 |
| Farm Type [Dairy] | −0.1414 | 0.0600 | −2.35 | 0.0191 |
| Fly Sex [Female] | 0.4100 | 0.0600 | 6.83 | < 0.0001 |
| Collection Avg Temp | 0.2852 | 0.0537 | 5.31 | < 0.0001 |
| 3D Days of Precipitation | 1.6464 | 0.4864 | 3.38 | 0.0008 |
| 3D Avg Humidity | 0.0856 | 0.0271 | 3.16 | 0.0017 |
| 3D Avg Precipitation | 0.0361 | 0.0076 | 4.71 | <0.0001 |
| 1W Avg Soil Temp | −0.3757 | 0.0899 | −4.18 | < 0.0001 |
3D = Collection date and 2 d prior to the collection.
1W = Collection date and 6 d prior to the collection.
Climate variables used for prediction plots of factors influencing mean abundance of culturable bacteria and suspected coliforms carried by house flies.
| Parameter | Estimate | Standard Error | t Ratio | Prob > |t| |
|---|---|---|---|---|
| Culturable Bacteria | ||||
| Intercept | −0.6554 | 0.9883 | −0.66 | 0.5076 |
| Fly Sex [Female] | 0.3498 | 0.0436 | 8.02 | < 0.0001 |
| Collection Soil Temp Avg | −0.2526 | 0.0572 | −4.42 | < 0.0001 |
| Collection Avg Temp | 0.2618 | 0.0485 | 5.40 | < 0.0001 |
| 3D Avg Humidity | 0.0732 | 0.0150 | 4.87 | 0.0038 |
| 3D Avg Precipitation | 0.3498 | 0.0436 | 8.02 | <0.0001 |
| Suspected Coliforms | ||||
| Intercept | −1.4365 | 1.4127 | −1.02 | 0.3099 |
| Farm Type [Dairy] | −0.1414 | 0.0600 | −2.35 | 0.0191 |
| Fly Sex [Female] | 0.4100 | 0.0600 | 6.83 | < 0.0001 |
| Collection Avg Temp | 0.2852 | 0.0537 | 5.31 | < 0.0001 |
| 3D Days of Precipitation | 1.6464 | 0.4864 | 3.38 | 0.0008 |
| 3D Avg Humidity | 0.0856 | 0.0271 | 3.16 | 0.0017 |
| 3D Avg Precipitation | 0.0361 | 0.0076 | 4.71 | <0.0001 |
| 1W Avg Soil Temp | −0.3757 | 0.0899 | −4.18 | < 0.0001 |
| Parameter | Estimate | Standard Error | t Ratio | Prob > |t| |
|---|---|---|---|---|
| Culturable Bacteria | ||||
| Intercept | −0.6554 | 0.9883 | −0.66 | 0.5076 |
| Fly Sex [Female] | 0.3498 | 0.0436 | 8.02 | < 0.0001 |
| Collection Soil Temp Avg | −0.2526 | 0.0572 | −4.42 | < 0.0001 |
| Collection Avg Temp | 0.2618 | 0.0485 | 5.40 | < 0.0001 |
| 3D Avg Humidity | 0.0732 | 0.0150 | 4.87 | 0.0038 |
| 3D Avg Precipitation | 0.3498 | 0.0436 | 8.02 | <0.0001 |
| Suspected Coliforms | ||||
| Intercept | −1.4365 | 1.4127 | −1.02 | 0.3099 |
| Farm Type [Dairy] | −0.1414 | 0.0600 | −2.35 | 0.0191 |
| Fly Sex [Female] | 0.4100 | 0.0600 | 6.83 | < 0.0001 |
| Collection Avg Temp | 0.2852 | 0.0537 | 5.31 | < 0.0001 |
| 3D Days of Precipitation | 1.6464 | 0.4864 | 3.38 | 0.0008 |
| 3D Avg Humidity | 0.0856 | 0.0271 | 3.16 | 0.0017 |
| 3D Avg Precipitation | 0.0361 | 0.0076 | 4.71 | <0.0001 |
| 1W Avg Soil Temp | −0.3757 | 0.0899 | −4.18 | < 0.0001 |
3D = Collection date and 2 d prior to the collection.
1W = Collection date and 6 d prior to the collection.
The equation for the best-fitting regression model to predict the mean abundance of culturable bacteria in house flies at beef and dairy operations is as follows:
Where Ycb is the predicted mean Log10(CFU + 1) of culturable bacteria per fly,
Xs is the average soil temperature of the collection date
Xa is the average ambient temperature of the collection date
Xh is the average humidity of the 3 d prior
Xp is the average precipitation (cm) for the 3 d prior
Xf is the condition for adding or subtracting 0.35 based upon the fly sex
◦ if “female” + 0.35
◦ if “male”—0.35
Based on the model, a positive relationship was predicted between culturable bacteria abundance, average ambient temperatures of the collection date, and average humidity and precipitation of the 3 d prior to collection (Equation 1). However, greater abundances of culturable bacteria were negatively associated with higher average soil temperatures on the date of collection. Additionally, culturable bacteria abundance for female flies are predicted to be greater than those of male house flies under the same conditions.
In contrast, the equation for the best-fitting regression model to predict the mean abundance of SC in house flies at beef and dairy operations is as follows:
Where Ycol is the predicted mean Log10(CFU + 1) of SC per fly,
Xa is the average ambient temperature of the collection date
Xd is the number of days with precipitation of the 3 days prior
Xh is the average humidity of the 3 days prior
Xw is the average soil temperature of the week prior
Xp is the average precipitation (cm) for the 3 days prior
Xf is the condition for adding or subtracting 0.41 based upon the fly sex
◦ if “female” + 0.41
◦ if “male”—0.41
Xt is the condition for adding or subtracting 0.14 based upon the location farm type
◦ if “dairy” + 0.14
◦ if “beef”—0.14
A negative relationship was predicted between SC abundance and average soil temperatures in the week prior to collection (Equation 2). Alternatively, higher average ambient temperatures on the date of collection, higher average humidity and precipitation for the 3 d prior to collection, and more days with precipitation in the 3 d prior were predicted to increase culturable bacteria carried by house flies. The model predicted greater SC abundance in female house flies than male flies and greater SC abundance in flies from dairies than those from beef operations.
Notably, fly sex had the strongest influence in predicting mean abundances of both culturable bacteria and SC in house flies, followed by average ambient temperatures on the collection date (Table 1). However, ANOVA indicated a lack of fitness for the prediction profiler models of both the mean culturable bacteria (F = 3.0904, df = 24, P < 0.0001) and SC (F = 1.7371, df = 52, P = 0.0025) for flies, suggesting other potential coefficients unmeasured in this study exist that could improve accuracy of future models.
Antimicrobial Susceptibility and Identities of Suspected Coliform Morphotype Isolates
A total of 661 SC isolates were selected from the VRBA cultures of adult house flies collected at beef (n = 412 isolates) and dairy (n = 249 isolates) operations (Supplementary Table S3). These SC isolates were selected from 215 of the 352 house flies collected in this study and ranged from 1 to 15 SC isolates per fly. Following the agar-based tetracycline resistance screening, 260 beef SC isolates and 115 dairy SC isolates were suspected to be tetracycline resistant and underwent disk susceptibility testing. Despite growing on tetracycline-infused agar, 50.1% (n = 188) SC isolates suspected to be tetracycline-resistant were not resistant to tetracycline via Kirby-Bauer disk testing. In total, 132 (31.8%) beef SC isolates and 55 (22.1%) dairy SC isolates were tetracycline resistant. However, 94 (25.1%) of the SC isolates found susceptible to tetracycline were instead resistant to ampicillin, 2 SC isolates were dual-resistant to florfenicol and ampicillin, and one SC isolate was dual-resistant to florfenicol and ceftiofur.
Of 661 SC isolates selected from house flies, 43.0% (n = 284/661) were resistant to one or more antibiotics. Of these 284 AMR SC isolates, 218 (76.8%) SC isolates were single-drug resistant to either tetracycline (n = 124; 56.9%) or ampicillin (n = 94; 43.1%) (Supplementary Table S4). Single-resistant SC isolates were largely Escherichia/Shigella (n = 99) and Klebsiella (n = 50) species, with 96.0% (n = 95/99) of Escherichia/Shigella isolates possessing tetracycline resistance and 98.0% (n = 49/50) of Klebsiella isolates possessing ampicillin resistance. The AMR phenotype and genera of single-resistant SC isolates were variable between beef and dairy operations (Fig. 3), and individual sites within the same farm type (Supplementary Table S4). Tetracycline-resistant SC isolates cultured from house flies collected at both beef and dairy operations included genera, such as Citrobacter, Escherichia/Shigella, Kluyvera, and Providencia. Tetracycline-resistant SC isolates from dairy house flies additionally included Enterobacter and beef house flies additionally had tetracycline-resistant Klebsiella, Leclercia, Proteus, Pseudescherichia, Serratia, and Siccibacter isolates. Ampicillin-resistant SC isolates from house flies at both farm types included Escherichia/Shigella, Klebsiella, Kluyvera, and Serratia. Additionally, Citrobacter, Morganella, and Pseudocitrobacter represented ampicillin-resistant isolates from dairy house flies and Enterobacter, Kosakonia, Pantoea, and Raoultella represented ampicillin-resistant isolates from beef house flies.
Number (n) of SC isolate genera cultured from house flies from beef and dairy operations with tetracycline (TET) or ampicillin (AMP) resistance phenotypes.
Sixty-six SC isolates were resistant to two or more antimicrobials and characterized as multi-drug resistant (MDR), including 8.0% (n = 20/249) of SC isolates from flies from dairy operations and 11.2% (n = 46/412) SC isolates from flies from beef operations. Bacterial genera and MDR phenotypes of SC isolates from flies were variable between sites within the same farm type (Supplementary Table S5). Of the 187 tetracycline-resistant SC isolates from house flies, 64 (34.2%) SC isolates had additional resistance to florfenicol, enrofloxacin, ceftiofur, or ampicillin. Two MDR SC isolates did not have tetracycline resistance via Kirby-Bauer testing but were dual-resistant to florfenicol and ceftiofur or ampicillin. Dual resistance to tetracycline and ampicillin was the most common MDR phenotype for SC isolates from flies at both beef and dairy operations (n = 33/66; 50%) and had the widest variety of bacterial genera associated with this phenotype (Fig. 4): Enterobacter (1), Escherichia/Shigella (7), Klebsiella (10), Kluyvera (1), Morganella (4), Proteus (2), Providencia (2), Raoultella (2), Serratia (4). Triple resistance to tetracycline, florfenicol, and ampicillin was the next most common MDR phenotype (n = 18/66; 27.3%), although this phenotype was only observed in the genera Escherichia/Shigella and Pseudescherichia. Additionally, Escherichia/Shigella were associated with the widest variety of MDR phenotypes (Fig. 4). SC isolates from flies collected at beef operations had a wider variety of MDR phenotypes than SC isolates from flies collected at dairies (Fig. 4). Enrofloxacin resistance was only observed in 2 SC isolates recovered from 2 different female house flies at the Marion beef operation, one of which was a Pseudomonas isolate that was resistant to all 5 antibiotics.
Number (n) of SC isolate genera with MDR profiles found in house flies from beef and dairy operations. Resistance phenotypes: T = tetracycline; F = florfenicol; A = ampicillin; C = ceftiofur; E = enrofloxacin.
Prevalence of House Flies Carrying AMR Isolates
No particular category for the mean abundance of SC CFUs enumerated from a fly was found to significantly increase the probability of whether SC isolates from a fly were resistant to one (P = 0.6536) or multiple (P = 0.4493) antibiotics. Overall, 73.02% (n = 157/215) of adult house flies from which SC isolates were susceptibility tested carried at least one AMR SC isolate and 26.05% (n = 56/215) carried at least one MDR SC isolate. In one case, as many as 9 different AMR SC isolates were recovered from a single male fly collected at a beef operation. No apparent difference between male or female house flies carrying AMR and MDR SC isolates at these operations was observed, but a larger proportion of house flies collected at beef operations carried AMR (77.31%) and MDR (31.09%) SC isolates than house flies from dairy operations (67.71% AMR; 19.57% MDR) (Fig. 5).
Prevalence (%) of house flies carrying at least one SC isolate with resistance to ≥1 (AMR) or ≥ 2 (MDR) antibiotics. Beef: n = 61 male, n = 58 female; Dairy: n = 46 male; n = 50 female.
Discussion
This study investigated potential biotic and abiotic factors affecting the abundances of culturable aerobic and suspected coliform (SC) bacteria carried by house flies in Kansas beef and dairy confined cattle operations. Similar to prior findings (Neupane et al. 2020), female house flies in this study overall had significantly greater abundances of culturable aerobic and SC bacteria than male flies. Female flies also carried significantly greater abundances of culturable aerobic and SC bacteria within farm types, although this trend was not always significant within location or collection date. While female house flies are on average larger than male flies, body size is not considered correlated to the mean abundances of bacteria carried by house flies (Neupane et al. 2020). Rather, the frequent interaction of female house flies with microbe-rich substrates to seek out potential oviposition sites (Thomson et al. 2017), as well as their nutritional needs required for egg development (Hanski 1987) and feeding preferences (Neupane et al. 2023), are believed to influence their trophic activities and subsequent tendency to carry greater mean abundances of bacteria than male flies.
Coliforms are highly abundant in animal manure, which is frequented by female flies both for oviposition and as a food source (Thomson et al. 2017). Considering that manure accessibility and management practices often differ between beef and dairy facilities, we therefore predicted farm type would most likely affect SC abundance in female flies. However, farm type significantly influenced SC abundance in only male house flies, with male flies from dairy sites carrying significantly lower SC abundance compared to male flies collected from beef sites. Some male flies from 2 of the 3 dairy sites (Marion and Washington) did not have any detectable SC. Interestingly, SC abundance in male flies collected from Riley dairy were similar to male flies collected from the beef facilities. This is likely because the Riley dairy is located close to the Riley beef site (less than 600 m), and shares site management and manure practices. The abundance of SC among female flies from dairy and beef locations did not differ, indicating that female flies may persistently interact with manure, potentially for oviposition and feeding, irrespective of the different manure accessibility or management practices associated with these 2 types of cattle operations. Alternatively, males may only have ephemeral, infrequent interactions with manure compared to females who inspect manure frequently for oviposition purposes. Further, in contrast to female flies, which have a higher tendency to feed on a variety of food resources, male flies prefer sugar-rich foods when presented with a variety of food resources (Neupane et al. 2023). Therefore, male flies within cattle operations may interact with sources of SC other than manure that differ in availability, consistency, or management resulting in variable SC levels. Cattle feed is one such substrate where the formulation can differ substantially between operations to best meet desired end animal product (volume and quality of milk vs. weight gained as muscle and fat). These feed ingredients not only vary in attractiveness to house flies but also in their ability to support coliform growth. For example, fresh steam-flaked corn, a carbohydrate-rich cattle feed ingredient, has previously been identified as a site for coliform contamination by house flies within a beef cattle feedlot (Ghosh and Zurek 2015). The combination of these factors could influence SC accumulation by male flies at beef compared dairy sites. Our study only sampled 6 sites in a relatively small geographic area, and future studies should further investigate the sex-specific difference in SC acquisition and carriage in by house flies when feeding on different food sources across a variety of locations.
Date of collection significantly influenced culturable bacteria and SC abundances in both male and female flies, in contrast to previous studies (Neupane et al. 2020). In the previous study, flies were only sampled over a 1.5-month period mid-summer, whereas in the current study, flies were collected over a 2-mo period from summer into early fall. A shorter time frame within a single season may have more stable climate conditions that preclude meaningful impacts of collection date and associated climatic variables on bacterial abundances. In the current study, significant differences in bacterial abundances in flies over time were only observed later in our collection period when climate variables such as average daily ambient and soil temperatures were decreasing. This may be a result of increased interactions between house flies and environmental sources, as activities of adult flies in dairy cattle operations have been observed to increase with higher temperatures and flies may be therefore acquiring more bacteria (Zahn and Gerry 2020). Additionally, temperature is a known factor in bacterial propagation and the succession of bacterial communities within soil and other environments (Ratkowsky et al. 1982, Zhang et al. 2005) and impacts diversity and richness of ARGs within substrates that house flies frequent in confined cattle operations (Wang et al. 2022). Our findings suggest that local ambient and soil temperature and humidity may assist in predicting bacterial abundances in house flies at confined cattle operations and could be used to identify conditions during which house fly management may be more critical for mitigating pathogen dissemination. Sobur et al. (2022) found evidence of higher seasonal temperatures resulting in higher occurrences of S. aureus and MRSA isolates cultured from house flies in hospital-associated areas, as well as the isolates’ resistance to individual and multiple antibiotics tested. A knowledge gap remains regarding the relationship between temperature or other climate factors and the ability of house flies to transmit pathogens and AMR bacteria.
Of the 215 flies from which SC isolates were tested for resistance to antibiotics in this study, 157 (73%) carried one or more AMR isolates, and one fly carried as many as 9 different AMR isolates. Since our methods did not test the AMR of every morphotype cultured from each fly, the proportion of house flies carrying AMR bacteria may be higher than our data reflects. Although our subjective method of selecting distinct morphotypes of suspected coliforms from house fly cultures limits a quantitative estimation of AMR bacteria carried by house flies, our results still provide strong evidence that most house flies in confined cattle operations are carrying AMR bacteria and therefore are contributing to AMR persistence and spread in these environments. We found that the prevalence of AMR/MDR SC was similar between male and female house flies, which is further supported by the lack of correlation between the mean abundance of SC and carriage of AMR or MDR SC isolates by house flies. However, it is worth noting that the lack of SC was observed more often in male than female flies. Nevertheless, while female flies carried overall greater abundances of bacteria and may present the highest risk as either reservoirs or transmitters of AMR bacteria, AMR/MDR SC bacteria were found in both male and female flies from all locations.
A higher proportion of isolates from house flies collected from beef operations were tetracycline-resistant than those isolated from flies at dairies. Additionally, MDR isolates from beef operations possessed a wider variety of MDR phenotypes to the antibiotics tested than those from dairy operations. House flies from beef operations carried bacteria with resistance to florfenicol, enrofloxacin, and ceftiofur at higher rates than flies from dairies likely due to restrictive usage of these antimicrobial drugs in dairy production. Florfenicol and enrofloxacin are not labeled for use in female dairy cattle > 20 mo (FDA-HHS 2023), and therefore are unlikely to be administered. The prohibition of extralabel use of enrofloxacin and ceftiofur also may contribute to lower prevalence of resistance to these drugs in bacteria from both beef and dairy cattle systems (FDA-HHS 1997, 2012). In addition, feedlots source cattle from multiple locations while dairies usually operate as a closed herd with animals bred onsite. The constant influx of animals with differing treatment histories and geographic origins may also contribute to feedlots showing a wider variety of MDR phenotypes. Interestingly, bacterial abundances and AMR profiles carried by house flies from the Riley dairy and beef facilities were more similar to each other than our other sites of the same farm type. We believe the close proximity of the Riley beef and dairy sites to each other (< 0.5 km) and other livestock nearby (swine, equine, and poultry teaching and research facilities) contributed to this phenomenon and may indicate cross contamination and widespread fly movement among these livestock facilities, as they are well within observed flight distances of house flies (Geden et al. 2021).
AMR SC isolates cultured from house flies in this study belonged to 16 different bacterial genera. Importantly, we identified 6 genera (Kosakonia, Morganella, Pseudocitrobacter, Pseudescherichia, Raoultella, Siccibacter) from which AMR isolates had not previously been reported from house flies collected in food or animal facilities (Nayduch et al. 2023). Of the 12 genera assigned to MDR SC isolates cultured from house flies, we further identified MDR Morganella sp., Pseudescherichia sp., and Raoultella sp. from flies in both beef and dairy facilities. However, Escherichia/Shigella sp. were the most common AMR and MDR SC isolates cultured from house flies, were carried most frequently by house flies, and carried the widest variety of AMR profiles to the 5 antibiotics tested. Members of this taxonomic group include food-borne pathogens such as Escherichia coli, which commonly possess MDR genes that can be horizontally transferred (Hunter et al. 1992, Rovira et al. 2019), including within the gut of house flies (Petridis et al. 2014). Numerous studies have also identified culturable AMR/MDR Escherichia/Shigella sp. carried by house flies collected from other livestock facilities, including swine (Literak et al. 2009, Cervelin et al. 2018, Fukuda et al. 2018, Wadaskar et al. 2021), equine (Dolejska et al 2011), poultry (Poudel et al. 2019, Akter et al. 2020, Wadaskar et al. 2021), broiler (Solà-Ginés et al. 2015, Fukuda et al. 2018, Poudel et al. 2019), buffalo (Wadaskar et al. 2021), sheep (Wadaskar et al. 2021), and goat (Wadaskar et al. 2021). Consequentially, Escherichia/Shigella sp. may pose the greatest risk for the transmission and prevalence of AMR and MDR bacteria in house flies at cattle operations.
There are a few caveats important to consider when interpreting the results of our study. First, non-coliforms were selected during our attempts to select coliforms and included genera, such as Pseudomonas, Serratia, Providencia, Kluyvera, and Proteus, indicating difficulty in visual selection of coliforms from VRBA and likely overestimation of true coliforms in CFU estimates as well. However, this is why we used the term “suspected coliforms” throughout our study. Nonetheless, a number of the SC isolates belong to non-coliform genera that contain species that are pathogenic and/or known for their affinity to carry and transfer MDR to other bacteria (ie Pseudomonas), and still provide valuable data on fly carriage of taxa that can serve as a source of AMR for other bacterial pathogens within cattle operations. Second, we used CLSI guidelines for Enterobacterales on all isolates to help interpret disk susceptibility results. A few isolates were later identified as belonging to taxonomic groups outside of Enterobacterales. CLSI has interpretation guidelines for only a few other Gram-negative bacteria (Pseudomonas aeruginosa, Bordetella bronchiseptica, Mannheimia haemolytica, Pasteurella multocida, Actinobacillus pleuropneumoniae, Histophilus somni), and we did not have species-level identification to later correct resistance interpretations of Pseudomonas sp. isolates. We ultimately retained AMR interpretations for non-Enterobacterales to provide data on bacteria from flies that may likely be contributing to AMR and to help select AMR isolates for downstream analysis. We are currently performing whole genome sequencing of MDR isolates from this study and this will provide additional genetic evidence for suspected AMR in these isolates. Finally, because identifying a morphotype is subject to the experimenter’s discretion, we likely underreported AMR/MDR phenotypes carried by flies. Current work by our group incorporates metagenomic analyses of whole house fly homogenates which can detect both pathogens and AMR genes carried in or on house flies. The combination of the culture-based work we present here, and these molecular analyses will provide a more comprehensive picture of the role flies play in the carriage and movement of pathogens and AMR in cattle operations.
Together, the results of this study further emphasize the threat house flies pose to both animal and human health. House fly sex, farm type, and climate can significantly affect the abundance of bacteria and prevalence of AMR carried by house flies and therefore can be used as predictors for risk modeling. These parameters also should be considered for integrated AMR management and mitigation practices in confined animal feeding operations such as beef feedlots and dairies.
Acknowledgments
We especially thank the dairy and beef cattle producers from this study for providing access to their facilities for fly collections. We also thank Brianna Davis for assistance in sample collection, as well as Dr. Yoonseong Park and Dr. Justin Talley for helpful feedback on study design. Funding was provided by Central Life Sciences and USDA-ARS NP104 project number 3020-32000-018-000D. Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture. USDA is an equal opportunity employer.
Author contributions
Victoria Pickens (Conceptualization [Equal], Data curation [Lead], Formal analysis [Equal], Investigation [Lead], Methodology [Equal], Software [Equal], Validation [Equal], Visualization [Equal], Writing - original draft [Lead], Writing - review & editing [Equal]), Brandon Hall (Investigation [Supporting], Writing - review & editing [Equal]), Kathleen Yeater (Formal analysis [Equal], Methodology [Equal], Software [Equal], Validation [Equal], Visualization [Equal], Writing - original draft [Supporting], Writing - review & editing [Equal]), Tanya Purvis (Investigation [Supporting], Methodology [Equal], Writing - review & editing [Equal]), Edward Bird (Formal analysis [Equal], Methodology [Equal], Software [Equal], Validation [Equal], Visualization [Equal], Writing - original draft [Supporting], Writing - review & editing [Equal]), Grant Brooke (Investigation [Supporting], Writing - review & editing [Equal]), Cassandra Olds (Conceptualization [Equal], Project administration [Equal], Resources [Equal], Supervision [Equal], Validation [Equal], Visualization [Equal], Writing - original draft [Supporting], Writing - review & editing [Equal]), and Dana Nayduch (Conceptualization [Equal], Data curation [Supporting], Formal analysis [Equal], Funding acquisition [Lead], Investigation [Lead], Methodology [Equal], Project administration [Equal], Resources [Equal], Supervision [Equal], Validation [Equal], Visualization [Equal], Writing - original draft [Supporting], Writing - review & editing [Equal])
Conflicts of interest. None declared.
Data Availability
Raw CFU counts, climate data, MALDITOF-MS isolate identities, and antimicrobial susceptibilities supporting the findings of this study are publicly available at the Ag Data Commons (DOI:10.15482/USDA.ADC/27089242). All trimmed Sanger sequence reads of AMR isolates are publicly available at GenBank (PQ636534 - PQ636762).
