A survey of honey bee drones and dandelion flowers screened tetracycline resistance genes and found tetB and tetM at high rates. The work covered 13 targets in 45 samples, pairing sequencing with gene-specific PCR. Oxytetracycline was not tested in the experiment. The useful takeaway is a tetracycline-class reading: oxytetracycline is part of the same class whose gene-mediated resistance mechanisms these targets describe.
Tetracycline resistance genes in drone and flower samples
This was a cross-sectional look at resistance markers in two linked settings. The first is the honey bee drone, sampled whole-body at four points during development. The second is the dandelion flower gathered from the apiary surroundings. The goal was to describe which resistance genes occur and how often, not to measure what a drug did.
That difference in purpose shapes how the numbers should be read. The survey screened a set of targets and reported how common each one was. Bees and flowers were never given a product, and no organism was challenged with a measured dose. Every value below is therefore a carriage value: what was found, rather than what a treatment achieved.
The term tetracycline does two jobs here. It names a drug class and it names a family of resistance genes. The survey measured the gene side. The class framing comes from the fact that the mechanisms these genes encode are shared across the tetracycline class. Oxytetracycline sits in that class. Calling the result a class-level signal keeps the observation and the interpretation properly aligned.
Study design and the sample set
The design paired next-generation sequencing of the bacterial community with targeted PCR for resistance genes. Drones were collected at four developmental stages and flowers came from the same apiary surroundings. Three biological replicates were taken per group, for 15 samples altogether. The prevalence calculation was done at the replicate level across 45 PCR readings.
Item | Detail |
Drone sample types | Whole-body drones at four stages: ULD (unsealed larvae), SLD (sealed larvae), PD (pupae), AD (adults) |
Flower sample type | Taraxacum officinale flowers (TO, dandelions) from the surrounding apiary environment |
Biological replicates | n = 3 per group |
Total sequencing samples | 15 (12 drone plus 3 dandelion) |
Samples for ARG prevalence | 45 (36 drone plus 9 dandelion, replicate-level PCR results) |
Sequencing approach | V3-V4 segment of the 16S rRNA gene, next-generation sequencing, DADA2 processing |
ARG detection | Target-specific PCR for 13 genes |
Primer annealing temperatures | 48-66 C |
PCR product sizes | 111-1115 bp |
Community statistics | Kruskal-Wallis with post-hoc tests; PERMANOVA and PERMDISP on weighted and unweighted UniFrac |
The two readouts were different in scope. Sequencing described the entire bacterial community, while PCR answered a narrow question about 13 named genes. Looking at both at once lets a reader compare a community shift with a gene pattern, without blending the two measurements into one.
Sequencing output and amplicon sequence variants
Depth was high for the number of samples involved. Fifteen libraries yielded millions of reads. Cleaning and denoising removed a large share of that, leaving a smaller set for analysis. DADA2 then split the retained data into amplicon sequence variants, which are exact sequence types rather than loose clusters.
Sequencing metric | Value |
Samples | 15 |
Raw reads | 6,852,660 |
Raw reads per sample, range | 393,510-540,236 |
Mean raw reads per sample | 456,844 |
High-quality reads retained | 3,055,573 |
Average retained sequences per sample | 203,705 |
Total ASVs across all samples | 1,724 |
Richness varied a great deal from one sample to the next. An unsealed larva sample gave the lowest count and a dandelion sample the highest. The cumulative totals also differed by group, with flowers holding the largest pool and unsealed larvae the smallest.
Stage | Cumulative ASVs | Average ASVs (+/- SD) |
ULD (unsealed larvae) | 182 | 60.67 +/- 23.35 |
SLD (sealed larvae) | 396 | 132.00 +/- 33.18 |
PD (pupae) | 238 | 79.33 +/- 5.51 |
AD (adults) | 286 | 95.33 +/- 13.05 |
TO (dandelions) | 207.33 +/- 3.79 (average only) | 207.33 +/- 3.79 |
Individual observed ASVs ran from 40 in ULD3 to 210 in TO2, so the dandelion group sits at the top of the richness range in this data.
Alpha diversity across developmental stages
Richness was not the same across groups. Several stage comparisons reached significance for observed ASV richness. The Shannon index, which accounts for both richness and evenness, picked out just one difference.
Diversity measure | Significant contrasts | Threshold |
Observed ASV richness | ULD vs SLD; SLD vs PD; drone stages vs TO | p < 0.05 |
Shannon diversity | ULD vs TO only | p < 0.05 |
The main richness changes came at the larval-to-sealed-larval step and in the move from bees to flowers. Evenness contributed little beyond that. Kruskal-Wallis with post-hoc testing produced these comparisons.
Beta diversity and community separation
Community makeup was compared with UniFrac distances, weighted and unweighted. The overall test came out significant in both cases, meaning the communities across groups were not alike.
Beta-diversity analysis | Result |
Overall PERMANOVA, unweighted UniFrac | pseudo-F = 3.28, p = 0.002 |
Overall PERMANOVA, weighted UniFrac | pseudo-F = 4.19, p = 0.001 |
Pairwise comparisons after correction | Not significant, q > 0.05 |
PERMDISP, unweighted UniFrac | F = 0.343, p = 0.463 (not significant) |
PERMDISP, weighted UniFrac | F = 2.643, p = 0.016 (significant heterogeneity) |
Within-group UniFrac distance, highest | Adult drones, 0.437 +/- 0.126 |
Within-group UniFrac distance, lowest | Dandelions, 0.013 +/- 0.006 |
PCoA unweighted UniFrac | PCoA1 52.6%, PCoA2 13.5% of variation |
PCoA weighted UniFrac | PCoA1 42.9%, PCoA2 26.9% of variation |
A significant overall result sits beside pairwise contrasts that lost significance after correction. The groups differ together, in other words, but separating any one pair is difficult once the correction is applied. The weighted PERMDISP result indicates uneven spread within groups, which fits the wide adult-drone distance and the tight dandelion distance.
Taxonomy from phylum to genus
Fourteen phyla showed up, with four of them dominating. The leading phylum depended on the group.
Phylum | Group where dominant | Relative abundance |
Pseudomonadota | ULD drones | 91.7% |
Pseudomonadota | PD drones | 65.8% |
Pseudomonadota | AD drones | 71.6% |
Bacillota | SLD drones | 71.2% |
Pseudomonadota | Dandelions | 59.5% |
Bacillota | Dandelions | 40.0% |
Actinomycetota remained low in most groups but reached roughly 0.4% to 9.2% overall, peaking in sealed larvae at 9.2% and pupae at 5.1%. At finer resolution the survey separated 86 families and 177 genera.
Group | Dominant family or families | Dominant genera |
ULD drones | Enterobacteriaceae 89.6% | Frischella 64%, Gilliamella 25.7% |
PD drones | Enterobacteriaceae 56.3% | Frischella 49.5%, Gilliamella 6.8% |
AD drones | Enterobacteriaceae 50.2% | Frischella 22.7%, Gilliamella 26.9% |
SLD drones | Lactobacillaceae 26% | Lactobacillus 20%, Bombilactobacillus 3.8%, Apilactobacillus 2.3% |
Dandelions | Pseudomonadaceae 40.4%, Streptococcaceae 39.9%, Enterobacteriaceae 16.7% | Pseudomonas 40.4%, Lactococcus 40.0% |
The two flower families make up about 80% of the dandelion community. The drone groups center instead on Enterobacteriaceae, with sealed larvae leaning toward lactic acid bacteria.
Species-level patterns
Thirty-nine distinct species were found. Sealed larvae carried the most at 23, including three seen nowhere else in that group. Dandelions had eight species unique to flowers. Four species occurred in all four drone stages, and seven were shared between drones and flowers.
Species-level observation | Count or detail |
Distinct species identified | 39 |
Species in SLD, highest group | 23 |
Species exclusive to SLD | 3 |
Unique species in dandelions | 8 |
Species present across all drone stages | Gilliamella apicola, Frischella perrara, Snodgrassella alvi, Aquabacterium parvum |
Species shared between drones and dandelions | 7 |
The seven shared species form a small but genuine link between the two settings, which matters when reading resistance markers that appear in both.
The resistance gene panel and how it was read
Target-specific PCR was used to detect resistance genes. Thirteen targets were screened, with annealing temperatures from 48-66 C and products from 111 to 1115 bp. Each replicate was scored for each gene, and prevalence was calculated over 45 PCR results.
Panel attribute | Value |
Genes screened | 13 |
Gene targets | aac_aph, blaZ, sul1, sul2, tetB, tetC, tetD, tetH, tetM, tetW, tetY, tlrB, tlrD |
Annealing temperature range | 48-66 C |
Product size range | 111-1115 bp |
Prevalence denominator | 45 samples (36 drone plus 9 dandelion) |
Targets not detected in any sample | tetY, tlrB |
Eleven of the 13 targets appeared at least once. The two that never did were tetY and tlrB. That absence is a finding in its own right, not a hole in the panel.
Core prevalence: tetB, sul2, sul1 and tetM
Four genes stood apart. Two came from the tetracycline group and two from the sulfonamide group, and their rates were close to one another while sitting far above most of the rest.
Gene | Class group | Prevalence | 95% CI |
tetB | Tetracycline | 88.9% | 76.5-95.2% |
sul2 | Sulfonamide | 86.7% | 73.8-93.7% |
sul1 | Sulfonamide | 73.3% | 59.0-84.0% |
tetM | Tetracycline | 68.9% | 54.3-80.5% |
The two leading tetracycline resistance genes here, tetB and tetM, are why a class-level reading helps. They encode mechanisms that shield the bacterial ribosome from tetracycline-class drugs. The class that contains oxytetracycline acts on that same ribosomal target, so high carriage of tetB and tetM is a class-relevant observation. It is not a measurement of oxytetracycline doing anything in these bees. No bee in this survey received oxytetracycline or any other drug.
The confidence bands are broad for several rows, which is what a modest replicate count tends to produce. The ordering remains clear, and tetB was the single most common target across all 45 samples.
Stage-by-stage tetracycline gene patterns
Prevalence shifted with development and with sample type. tetB appeared in every unsealed larva, sealed larva and pupa sample before falling in adults and flowers. tetM went the other way, reaching its peak in adult males.
Gene | ULD | SLD | PD | AD | TO |
tetB | 100% | 100% | 100% | 77.8% (95% CI 45.3-93.7%) | 66.7% (95% CI 35.4-87.9%) |
tetM | 66.7% | 77.8% | 33.3% (95% CI 12.1-64.6%) | 100% (95% CI 70.1-100%) | Not reported for TO |
tetC | 33.3% | 55.6% (95% CI 26.7-81.1%) | Not reported | Not reported | 66.7% |
tetD | Not reported | 22.2% (95% CI 6.3-54.7%) | 22.2% (95% CI 6.3-54.7%) | Not reported | 100% (95% CI 70.1-100%) |
tetH | 22.2% | Not reported | Not reported | Not reported | 66.7% |
tetW | 22.2% | 22.2% | Not reported | Not reported | 88.9% (95% CI 56.5-98.0%) |
Two patterns stand out. tetM hit 100% in adults even while tetB declined in the same group. Flowers also carried high figures for several tetracycline targets, with tetD at 100% and tetW at 88.9%. The flower setting is not a passive backdrop; it holds a real share of the same gene-mediated resistance mechanisms that concern the tetracycline class.
Sulfonamide and other resistance genes
The sulfonamide genes were widespread and several other targets turned up at lower levels. A few genes were confined to flowers.
Gene | Prevalence detail |
sul1 | 73.3% overall; 100% in ULD; 100% in TO; 33.3% in PD |
sul2 | 86.7% overall (95% CI 73.8-93.7%) |
blaZ | 66.7% in TO; 33.3% in SLD |
aac_aph | 33.3% in TO only; 6.7% overall (95% CI 2.3-17.9%) |
tlrD | 33.3% in TO only; 6.7% overall (95% CI 2.3-17.9%) |
The sulfonamide genes did not differ significantly from the two leading tetracycline genes. They did sit clearly above aac_aph, blaZ, tetC, tetD, tetH, tetW and tlrD. Fisher's exact tests with Benjamini-Hochberg adjustment supplied those comparisons.
Distinct targets and significant group differences
The count of different genes per group ranged widely, from the broadest set in flowers to the narrowest in adults.
Sample type | Distinct ARG targets detected |
TO (dandelions) | 11 of 13 |
ULD | 7 |
SLD | 8 |
PD | 5 |
AD | 4 |
Three contrasts held up after correction.
Gene | Contrast | Values | Adjusted q |
sul1 | ULD and TO higher than PD | 100% vs 100% vs 33.3% | q = 0.045 for both |
tetD | TO higher than SLD and PD | 100% vs 22.2% vs 22.2% | q = 0.006 for both |
tetW | TO higher than ULD and SLD | 88.9% vs 22.2% vs 22.2% | q = 0.038 |
The direction is steady: flowers outrank the bee stages for several targets. That keeps the environmental sample at the center of any conversation about where these markers persist.
What a tetracycline-class reading means for oxytetracycline
Here the evidence meets the product context, and the line between them needs to stay sharp. The survey detected genes, not a drug effect. Oxytetracycline is a tetracycline-class molecule, and that class shares the gene-mediated resistance mechanisms represented by tetB and tetM. That is the whole connection.
Reading at class level still has practical value. When a resistance gene is common in a setting, every member of the drug class that depends on the same target is part of that picture. For oxytetracycline, the relevance is class membership, not a measured result in these bees. The survey gives no dose, no kill curve and no treatment outcome that involves oxytetracycline.
The sound conclusion is narrow but useful. Widespread tetracycline resistance genes in bees and flowers mean the class-level resistance background exists in those settings. Watching that background is a fair part of resistance surveillance. Statements about how one molecule performs would call for a different kind of study.
Limits of the survey
Two limits were reported as part of the work, and both bound how far the numbers can be stretched.
Limit | Consequence |
Limited biological replicates, n = 3 per group | Lower statistical power for pairwise comparisons after multiple-testing correction |
Whole-body sequencing rather than gut-only sampling | Reduced comparability with worker-bee gut studies |
The small replicate count explains why the overall PERMANOVA result was significant while pairwise comparisons were not. The whole-body choice explains why these values should not be set beside studies that sampled only the gut. Neither limit changes the prevalence counts, but both frame the interpretation.
Bottom line
The survey recorded tetracycline resistance genes at high rates in honey bee drones and dandelion flowers. tetB reached 88.9% and tetM 68.9% across 45 samples, with the sulfonamide genes sul2 and sul1 close behind. Communities differed between groups, flowers held the widest gene set, and the environmental samples often showed the highest figures. oxytetracycline was never tested in this work. Its place here is as one member of the tetracycline class whose shared mechanisms these targets describe, making the data a useful class-level surveillance signal and nothing beyond that.
FAQ
What are tetracycline resistance genes?
Tetracycline resistance genes are DNA markers that let bacteria withstand tetracycline-class drugs. Common examples include tetB, tetC, tetD, tetH, tetM, tetW and tetY. They work by protecting the bacterial ribosome or by moving the drug out of the cell. The survey measured the carriage of these markers in bee and flower samples. It did not measure drug response in a treated animal.
Did the survey test oxytetracycline?
No. The survey screened resistance genes by PCR and described bacterial communities by sequencing. It did not dose bees or flowers with oxytetracycline or any other drug. Oxytetracycline matters here only because it belongs to the tetracycline class. The gene-mediated mechanisms found in the samples are mechanisms that this class shares. Any claim about how oxytetracycline performed would require a separate study.
How common was tetB in the samples?
tetB was the most prevalent target in the full set of 45 samples, at 88.9% with a 95% CI of 76.5-95.2%. It was detected in every unsealed larva, sealed larva and pupa sample. It fell to 77.8% in adults and 66.7% in dandelions. The drop in later stages is one of the clearer stage-related patterns in the survey.
Which resistance genes were most prevalent?
Four genes led the panel. tetB was at 88.9% and sul2 at 86.7%, followed by sul1 at 73.3% and tetM at 68.9%. The two tetracycline resistance genes and the two sulfonamide genes did not differ significantly from one another. All four sat well above most of the remaining targets, which ran down to 6.7% for aac_aph and tlrD.
What does a tetracycline-class reading mean?
It means interpreting a gene finding at the level of the drug class rather than a single product. tetB and tetM encode mechanisms that protect the ribosome from tetracycline-class drugs. Oxytetracycline belongs to that class and shares the same target. A high carriage rate is therefore class-relevant. It is not evidence about how one specific molecule worked in these bees.
How many samples and reads were sequenced?
Fifteen samples produced 6,852,660 raw reads, with 393,510 to 540,236 reads per sample and a mean of 456,844. After filtering, 3,055,573 high-quality reads remained, averaging 203,705 sequences per sample. DADA2 resolved 1,724 ASVs across the whole set. The prevalence analysis used 45 replicate-level PCR results from 36 drone and 9 dandelion samples.
What is an amplicon sequence variant?
An amplicon sequence variant, or ASV, is an exact DNA sequence type resolved from sequencing data. DADA2 was used to identify them here. The survey found 1,724 ASVs in total. Observed counts per sample ranged from 40 in one larva sample to 210 in a dandelion sample. ASVs give a finer view of community makeup than broad clustering does.
Did bee stage change the bacterial community?
Yes at the overall level. PERMANOVA was significant for both unweighted and weighted UniFrac, with p = 0.002 and p = 0.001. Observed richness differed between several stages and between drones and flowers. Pairwise contrasts were not significant after correction, so individual stage pairs are hard to separate. Adult drones also showed much wider within-group distances than dandelions.
Why are sulfonamide genes important here?
Sul1 and sul2 were among the most common targets, at 73.3% and 86.7%. They did not differ significantly from the leading tetracycline genes. Sul1 was higher in unsealed larvae and flowers than in pupae, with q = 0.045 for both contrasts. Their abundance shows that the resistance picture is not limited to one drug class in these samples.
What were the main limits of the survey?
Two limits were reported. Replicates were few, at n = 3 per group, which reduced power for pairwise comparisons after multiple-testing correction. Sequencing covered the whole body rather than the gut alone, which limits comparison with worker-bee gut studies. These limits affect interpretation, not the prevalence counts themselves.
Do tetracycline resistance genes in bees matter for animal health?
They are a surveillance signal. High carriage of tetracycline resistance genes in bees and flowers shows that the class-level resistance background is present in those settings. That matters because the same gene-mediated mechanisms can affect any member of the drug class. The data do not show disease or treatment failure. They describe where resistance markers are found.
How was ARG prevalence measured?
Prevalence came from target-specific PCR across 13 genes. Each replicate was scored for the presence of each target, and rates were calculated over 45 samples. Annealing temperatures ran from 48-66 C and products ranged from 111 to 1115 bp. Confidence intervals were reported for key genes, and group contrasts used Fisher's exact tests with Benjamini-Hochberg correction.