Salmonella control in animal feed: what actually worked
Somewhere between 200 CFU and nothing. That is the range where most Salmonella arguments in feed milling actually happen, and it is also the range where every control measure starts to look unreliable.
A challenge trial recently put three commercial anti-Salmonella products against the same four-serovar cocktail in two real matrices, and the spread in outcomes was wide. Poultry feed treated with a formaldehyde product came back negative by most probable number. One hundred percent. Poultry feed treated with a formic acid product dropped from 6.4 to 0.1 MPN per gram, a 98.4 percent reduction. Dry dog food treated with a bacteriophage product did nothing measurable. Above 6.4 MPN per gram going in, above 6.4 coming out. Direct plating even went the wrong way, from 38 to 75 CFU per gram.
Enriched shotgun metagenomics told the same story in the same order: 99.9 percent down for formaldehyde, 97.7 percent for formic acid, 33.1 percent for the phage. Unenriched sequencing did not tell that story at all, and for the formic acid product it said Salmonella had gone up. Knowing why is more useful than any of the individual verdicts.
Why Salmonella control in animal feed is harder than it looks
Nontyphoidal Salmonella is responsible for an estimated 1.28 million illnesses, 12,500 hospitalizations and 238 deaths a year in the United States alone. That makes it the deadliest of the major foodborne pathogens tracked there, and it is not close.
Animal feed is a known route. The category covers pet food, livestock feed, raw materials and ingredients, and it has been tied to recalls and outbreaks more than once. Surveys keep finding the organism in ingredients and in finished feed across the entire production chain, not parked at one dirty step somewhere in the middle.
So Salmonella control in animal feed is not a checkbox. It is a persistent hazard you have to hold down in a product that is, by composition, an invitation to anything that can grow in it.
Three things get in the way. Salmonella survives for days to weeks in dry feed. Contamination arrives intermittently and at low levels rather than in one identifiable slug. And some of the additives used for control interfere with the methods used to prove they worked.
The three additives and why each was tested where it was
Three anti-Salmonella feed additives carry most of the conversation. Pet food Salmonella control has the fewest cleared options of the three, which is part of why a bacteriophage preparation was tested there.
Table 1. The three anti-Salmonella products at a glance
| Product class | What it is | Matrix tested | Application rate as used | How it was applied |
| Bacteriophage | A lytic phage suspension labeled at a minimum of 10 to the power of 10 PFU per milliliter | Dry dog food kibble | 2 ml per pound, roughly 4 million PFU per gram | Sprayed on as a 1x working solution while the kibble was mixed |
| Formaldehyde | A liquid formaldehyde preparation diluted to a 20 percent solution before use | Poultry feed pellets | 3 g of product per kg of feed, equal to 2.7 g per kg of formaldehyde | Sprayed on in a chemical hood, then held overnight |
| Formic acid | A liquid formic acid preparation | Poultry feed pellets | 4 kg per metric ton | Sprayed on in a chemical hood, then held overnight |
Notice something about that table. Nothing was tested in the wrong matrix on purpose, but nothing was cross-tested either. The assignment is regulatory, not scientific.
Formaldehyde is cleared under 21 CFR 573.460 as an antimicrobial agent for keeping complete feeds or feed ingredients Salmonella negative for up to 21 days. Formic acid sits under 21 CFR 573.480 as a feed acidifying agent for complete swine and poultry feed, and it works by pushing pH down. Bacteriophage preparations have nine GRAS notices for Salmonella control in human food plus one recent no-objection letter covering pet food use as a processing aid. Nobody sprayed formaldehyde onto kibble here because no label allows it.
That leaves you comparing three products inside the matrices they are cleared for. Performance, floor conditions and regulatory reality are tangled together in every number below, and no amount of sequencing untangles them.
One more thing about 21 CFR 573.460. “Salmonella negative for up to 21 days” is a maintenance claim. Read it carefully, because it is not a sterilization claim, and the trial data does not suggest sterilization either.
Two hazards sit next to any efficacy discussion. The first is masking, where formaldehyde and organic acids interfere with culture recovery and hand you false negatives that look exactly like efficacy. Any efficacy claim built on plates alone deserves suspicion for precisely that reason, and it is why the molecular half of this trial was worth running. The second is safety and access. Worker exposure to formaldehyde is a live occupational concern, and the European Union refused to authorize formaldehyde as a feed additive. Effective and available are two different questions.
How the challenge trial was run
Eight experimental groups per additive. Once you understand how those eight arise, every result table reads differently.
Start with two 1 kg portions. One gets a gram of four-serovar Salmonella cocktail in nonfat dry milk, targeting about 1,000 CFU per gram, mixed by shaking while the jar rotates. The other gets a gram of sterile milk powder. Split each portion in half for treated and untreated, and you have four subgroups. Split all four again into sets tested directly versus enriched, and you have eight. Each of those finally becomes triplicate 25 g test portions.
Enrichment ran in buffered peptone water at 35 degrees Celsius for 24 hours, following the culture enrichment protocol used in regulatory screening. Unenriched portions went into phosphate buffered saline instead, then were shaken, settled and spun down.
Table 2. Salmonella challenge conditions in dry dog food and poultry feed
| Condition | Dry dog food | Poultry feed |
| Physical form | Complete kibble | Maintainer pellets, 15 percent protein, 3 percent fat, 9 percent fiber |
| Background aerobic plate count | 3.4 x 10 to the power of 1 CFU/g | 1.6 x 10 to the power of 3 CFU/g |
| Salinity target versus achieved | Target 1,000 CFU/g, achieved roughly 2.2 to 75 CFU/g | Same target and same shortfall |
| Inoculum | Four serovars: Salmonella Enteritidis, Salmonella Montevideo, Salmonella Senftenberg, Salmonella Typhimurium | Same four-serovar cocktail |
| Test portion size | 25 g, triplicate | 25 g, triplicate |
| Enrichment | Buffered peptone water, 35 degrees Celsius, 24 hours | Same protocol |
| Quantification | Five-tube three-level most probable number covering 25 g, 2.5 g and 0.25 g | Same MPN scheme |
| Detection limit, MPN | Above 6.4 MPN per gram | Above 6.4 MPN per gram |
| Detection limit, plating | 10 CFU per gram | 10 CFU per gram |
| Molecular screening | LAMP Salmonella detection, time to positive recorded | Same assay |
The four serovars were Enteritidis, Montevideo, Senftenberg and Typhimurium, and that choice was deliberate. Nearly half of Salmonella positive feed ingredients in an earlier survey carried two serovars, Montevideo and Senftenberg among them. Enteritidis and Typhimurium lead human illness in the United States. A single-serovar challenge would have flattered the product and told you less.
Now look at the target-versus-achieved row. Counts taken after inoculation but before treatment landed one to two logs under 1,000 CFU per gram. The cocktail was already dying before any additive touched it. In dry matrices the steepest drop happens inside the first 24 hours, and this trial took no interval samples, so nobody knows when it happened.
The additives were therefore challenged against a falling population. More realistic. Much harder to score.
What culture found: two clear wins and one null result
Table 3. Salmonella reduction by culture
| Product and matrix | Untreated | Treated | Reduction by MPN | Direct plating result |
| Bacteriophage in dry dog food | Above 6.4 MPN per g | Above 6.4 MPN per g | None detected | Untreated 38 CFU/g, treated 75 CFU/g |
| Formaldehyde in poultry feed | 2.2 MPN per gram | Not detected | 100 percent | Not performed, detection threshold too high |
| Formic acid in poultry feed | 6.4 MPN per gram | 0.1 MPN per gram | 98.4 percent | Not performed, detection threshold too high |
Formaldehyde cleared it. Formic acid came within a rounding error. The phage did nothing you could measure, and plating roughly doubled.
Every uninoculated group came back negative by MPN regardless of treatment, which is the reason the inoculated results mean anything. Direct plating was skipped for the chemical treatments because a 10 CFU per gram detection limit cannot resolve anything below what MPN already resolved. Comparing them there would have wasted effort.
Salmonella die-off in feed deserves another mention. It is exactly the sort of detail that quietly undermines a challenge study. The steepest Salmonella reduction in dry food and feed happens in the first day. This study timed nothing, so any claim about when the population collapsed is unavailable, and you are left scoring an additive against a moving target.
Sequencing: where it agreed with culture and where it did not
Here is the methodology lesson, and it is the part people misread.
Table 4. Salmonella relative abundance measured by shotgun metagenomics
| Product and matrix | Unenriched, untreated | Unenriched, treated | Enriched, untreated | Enriched, treated | Relative change, unenriched | Relative change, enriched |
| Bacteriophage in dry dog food | 97.2 percent | 96.3 percent | 73.5 percent | 49.2 percent | Down 0.9 percent | Down 33.1 percent |
| Formaldehyde in poultry feed | 2.3 percent | 0.7 percent | 24.3 percent | 0.02 percent | Down 69.6 percent | Down 99.9 percent |
| Formic acid in poultry feed | 4.9 percent | 7.3 percent | 57.1 percent | 1.3 percent | Up 49.0 percent | Down 97.7 percent |
Table 5. Salmonella relative abundance measured by 16S rRNA gene amplicon sequencing
| Product and matrix | Unenriched, untreated | Unenriched, treated | Enriched, untreated | Enriched, treated | Relative change, unenriched | Relative change, enriched |
| Bacteriophage in dry dog food | 36.1 percent | 32.6 percent | 29.4 percent | 20.8 percent | Down 9.7 percent | Down 29.3 percent |
| Formaldehyde in poultry feed | 2.3 percent | 1.2 percent | 28.3 percent | 0 percent | Down 47.8 percent | Down 100 percent |
| Formic acid in poultry feed | 3.0 percent | 3.4 percent | 31.7 percent | 3.2 percent | Up 13.3 percent | Down 89.9 percent |
Every percentage in those two tables is a relative change in proportion. That single sentence is most of the story.
Start with the unenriched columns. Unenriched means DNA pulled straight from the feed with no growth step, which is the closest thing to ground truth about what is physically in the bag.
For Salmonella in dry dog food, unenriched shotgun metagenomics read 97.2 percent before treatment and 96.3 percent after. A 0.9 percent relative move. On that evidence you would conclude the phage did essentially nothing, and here you would be right, by a route that would not reliably reach the right answer anywhere else.
In unenriched poultry feed treated with formic acid, Salmonella relative abundance went up: 4.9 to 7.3 percent by shotgun, 3.0 to 3.4 percent by amplicon sequencing. Culture said the same product had just removed 98.4 percent of viable Salmonella. Both are true. Neither contradicts the other.
The two sequencing approaches agreed on direction throughout, which is reassuring given how differently they work. Amplicon sequencing reads one marker gene and costs little per sample. Shotgun metagenomics reads everything present and costs considerably more. They landed in the same place, with amplicon data tending to report higher Bacillota abundance across the board.
The result that looks backwards
Worth slowing down here. This one is quotable, and easy to quote wrongly.
A formic acid product killed 98.4 percent of Salmonella by culture and made Salmonella a bigger share of the reads at the same time.
Relative abundance is a proportion. When a treatment knocks down competitors faster than it knocks down the pathogen, the pathogen loses ground and gains share together. Pseudomonas and Pantoea dominate the natural poultry feed community. Push them down and everything else rises proportionally, including the thing you were aiming at.
For Salmonella control in animal feed, the rule writes itself: a rise in relative abundance is not evidence of failure. Only culture, or a viability-corrected molecular method, can tell you whether the count moved. This trial used no viability dyes such as propidium monoazide, so it cannot separate live cells from dead cell DNA, and says so plainly.
The corollary matters just as much. Unenriched sequencing cannot answer the question you actually care about, whether a handful of viable cells survived. It lacks the sensitivity, no matter how deep you go, because the target is a fraction of a very large community.
Enrichment is the step that made the methods agree
Put the 24 hour enrichment step back in and the picture rearranges completely.
Enriched shotgun data shows relative Salmonella abundance down 99.9 percent for formaldehyde, 97.7 percent for formic acid and 33.1 percent for the phage. Enriched amplicon data gives the same ranking at 100 percent, 89.9 percent and 29.3 percent. Both now agree with what culture said from the beginning.
Only culture-enriched metagenomes reproduced the culture result. The explanation is arithmetic rather than philosophical. Enrichment grows whatever few survivors remain until they dominate the reads. Remove essentially everything and there is nothing to amplify. Leave a few cells behind and there is plenty. Enrichment turns an invisible difference into an obvious one.
Loop-mediated isothermal amplification agreed with both sequencing approaches and did it far faster. Time to positive stayed under 12 minutes in every group except formaldehyde treated enriched poultry feed, where it never crossed the 20 minute cutoff. For a screening assay that is a remarkably clean signal, and it argues for putting enrichment plus LAMP into routine additive evaluation rather than treating it as a research curiosity.
One caution attaches to all three molecular methods. LAMP and sequencing both returned Salmonella signals in sporadic uninoculated replicates whatever their treatment or enrichment status, though enriched samples showed fewer. The likely source is dead cell DNA already sitting in the product. Read any molecular detection as “Salmonella DNA is present,” not “viable Salmonella is present.”
How much sequencing effort this took
Table 6. Output of the two sequencing approaches
| Measure | 16S rRNA gene amplicon sequencing | Shotgun metagenomics |
| Average raw reads | 1.2 million, ranging from 0.2 to 3.5 million | 22 million, ranging from 1.8 to 87 million |
| Average reads after filtering | 440,000, ranging from 97,000 to 1.2 million | 16 million, ranging from 1.2 to 58 million |
| Read retention rate | 37.0 percent | 73.8 percent |
| Reads mapping to bacteria | 86.1 percent of final reads | 56.1 percent of final reads |
| Taxa detected at 0.1 percent or above | 7 phyla, 73 families, 168 genera | 7 phyla, 70 families, 127 genera |
| Platform | MiSeq, v3 reagent kit, 600 cycles, V3-V4 region | NextSeq 500, high output v2 kit, 300 cycles |
| Second opinion needed | Chloroplast and mitochondrial reads removed | Standard plus reference database, abundance re-estimated |
Two things worth taking away. This depth belongs to characterization work, not release testing; you would not build a verification program on it. And there is the plant DNA problem. Roughly 21 percent of unenriched poultry feed amplicon reads were chloroplast and roughly 22 percent were mitochondrial, and all of it went into the bin before the bacterial community could be read at all. Budget for losing about 40 percent of your reads on plant-based matrices.
DNA concentration spanned from undetectable on a high-sensitivity assay in every unenriched dry dog food group to more than 400 nanograms per microliter in the richest enriched poultry feed group. That spread tells its own story.
What was living in the feed before anything else happened
Table 7. Dominant taxa in the natural animal feed microbiota
| Matrix and group | Dominant phyla | Dominant families | Dominant genera |
| Dry dog food, uninoculated untreated unenriched | Bacillota, Pseudomonadota, Actinomycetota | Clostridiaceae, Vibrionaceae, Thermaceae | Clostridium, Vibrio, Pseudomonas |
| Poultry feed, uninoculated untreated unenriched | Pseudomonadota, Bacillota, Actinomycetota | Pseudomonadaceae, Erwiniaceae, Enterobacteriaceae | Pseudomonas, Pantoea, Kosakonia |
| Dry dog food, inoculated, unenriched | Pseudomonadota, driven by the inoculum | Enterobacteriaceae | Salmonella |
| Poultry feed, inoculated, unenriched | Unchanged from the uninoculated profile | Unchanged from the uninoculated profile | Pseudomonas and Pantoea remain dominant |
Clostridium in pet food sits on a different footing from Pseudomonas in poultry feed, mostly because the two products carry very different amounts of biomass. Dry dog food carried a background plate count of about 34 CFU per gram. Poultry feed carried about 1,600. Nearly two logs of difference, and it explains most of what follows.
With so little resident biomass, adding Salmonella at even a modest level reorganizes the whole community. Dominance flipped from Clostridium to Salmonella on inoculation alone, treated or untreated. In poultry feed, where plenty of biomass was already present, inoculation barely registered as a compositional change.
That same thinness explains why every unenriched dry dog food sample came back too dilute to measure on a high-sensitivity fluorometric assay and still produced usable libraries. Extrusion runs at 90 to 200 degrees Celsius under 290 to 1,160 psi, which is extremely good at removing bacteria and stripping DNA along with them. Low biomass work like this really wants extraction blanks and negative controls. None were included.
Before enrichment, neither inoculation nor treatment shifted the natural poultry feed community significantly. After enrichment, both did. Keep that sequence in mind before drawing conclusions from enriched data about what lives in feed.
What enrichment did to the community
Table 8. How culture enrichment reordered the animal feed microbiota
| Matrix | Dominant before enrichment | Dominant after enrichment | Note |
| Dry dog food | Clostridium | Lysinibacillus, family Bacillaceae | Bacillota dominates in both states |
| Poultry feed, formaldehyde untreated | Pseudomonas and Pantoea | Clostridium | Family Clostridiaceae after enrichment |
| Poultry feed, formaldehyde treated | Pseudomonas and Pantoea | Bacillus | Treatment changed which enriched taxon won |
| Poultry feed, formic acid | Pseudomonas and Pantoea | Clostridium | Same endpoint as untreated formaldehyde groups |
Look at how much larger the enrichment effect is than the treatment effect. A day in buffered peptone water took dry dog food from Clostridium to Lysinibacillus, and took poultry feed from Pseudomonas and Pantoea to Clostridium. Treatment nudged things around inside that picture. Enrichment chose the picture.
So treat enrichment as part of the measurement, not as a neutral step between sample and answer. Any statement about the feed microbiota built on enriched data describes the enrichment broth at least as much as the feed.
Diversity analysis said the same thing in aggregate. Treatment reduced richness and evenness in almost every comparison, and enrichment reduced it further, with a few exceptions among enriched shotgun libraries. Amplicon data cleared significance in two places: the Shannon index between uninoculated enriched groups, and Pielou evenness between inoculated enriched groups. The usual ecological account fits well enough. Acidifiers create conditions that suit acid-tolerant organisms and eliminate the rest, which is a filter, not a general suppression. Even so, clustering still grouped samples by their original matrix after inoculation, treatment and enrichment had all been applied. The starting community keeps shaping everything downstream.
Which genera actually moved
Table 9. Differentially abundant genera identified by shotgun metagenomics
| Comparison | Product | Genera that differed significantly | Significance |
| Enriched inoculated, untreated versus treated | Bacteriophage | Salmonella | P below 0.01 |
| Enriched inoculated, untreated versus treated | Formaldehyde | Clostridium | P below 0.01 |
| Enriched inoculated, untreated versus treated | Formaldehyde | Staphylococcus, Bacillus | P below 0.05 |
| Enriched inoculated, untreated versus treated | Formic acid | Bacillus, Clostridium, Streptomyces, Salmonella | P below 0.05 |
| Unenriched inoculated, untreated versus treated | Formaldehyde | Exiguobacterium, Klebsiella, Acinetobacter, Burkholderia, Streptomyces, Pantoea, Enterococcus, Salmonella | P below 0.05 |
| Unenriched inoculated, untreated versus treated | Formic acid | Lysinibacillus, Klebsiella | P below 0.05 |
| Enriched uninoculated, untreated versus treated | Formaldehyde | Clostridium, Bacillus | P below 0.05 |
Two results cleared the stricter threshold: Salmonella itself in phage treated dry dog food, and Clostridium in formaldehyde treated poultry feed. Both below P of 0.01, and both make mechanistic sense for what those products are supposed to do.
The rest of the table is better used for interpretation than for verdicts. Several genera on it are common, hardy environmental organisms, and the amplicon dataset flagged just one significant difference across all three comparisons in uninoculated material, Bacillus in formaldehyde treated poultry feed. The two sequencing approaches agreed more on direction than on what reached significance, which is a fair warning against leaning too hard on a single table like this one.
Practical takeaways for Salmonella control in animal feed
- Verify chemical treatments with culture first, then confirm with an enriched molecular method. Unenriched metagenomics will mislead you, and in this data set it reported the second most effective product as making things worse.
- Never read relative abundance as a pathogen count. A proportional rise after treatment usually means competitors died faster, not that your target survived.
- Treat enrichment as part of the measurement rather than neutral preparation. Choose it deliberately and say so when reporting, because it changes which taxa come out dominant.
- Bacteriophage in pet food has to be validated for delivery to target cells first. Expect performance to be strongly matrix dependent, and do not trust an application rate printed on a label until someone has confirmed it.
- Weigh formaldehyde efficacy against worker exposure and market access in the same decision. The European Union declined to authorize formaldehyde as a feed additive, so efficacy alone will not settle procurement.
- Do not mistake a low background plate count for a safety benefit. Dry dog food gave Salmonella nothing to compete with, which made each surviving cell more consequential rather than less.
- Build interval sampling into any challenge study. Salmonella die-off in dry matrices is steepest in the first 24 hours, and without timed samples you cannot say when it happened or whether the additive did anything at all.
- Inoculate with multiple serovars at realistic levels. Half of positive feed ingredients in one survey carried two serovars, and single-serovar cocktails flatter every control measure tested against them.
- Plan for plant DNA loss on plant-based matrices. Roughly 40 percent of unenriched poultry feed amplicon reads were chloroplast or mitochondrial and had to be discarded.
- Add viability dyes such as propidium monoazide when the question is about live cells. Molecular methods on their own cannot answer it.
Limitations worth knowing before you act on this
One matrix per product, no cross-testing. Defensible on regulatory grounds and frustrating on scientific ones, since the matrices differ in nutrient composition, water activity, background community and physical form, any of which could account for part of the performance gap. A same-matrix comparison would settle it, and does not exist yet.
One poultry feed formulation, one dry dog food formulation, artificially contaminated rather than naturally. Application rates, holding conditions and storage period were all single-point, and nothing followed the product out to the 21 days that the formaldehyde regulation talks about.
No extraction blanks, no negative controls. That matters most where it was hardest, in the ultra-low biomass dry dog food samples where a contamination signal looks exactly like a real one.
No viability discrimination anywhere. Every molecular detection in this work means DNA present, not cells alive.
And the achieved inoculum came in one to two logs low, with no interval sampling to place it in time. Both make the test more realistic. Both make the numbers harder to score.
Bottom line
Chemical treatment of poultry feed worked by culture and by enriched sequencing, with formaldehyde ahead of formic acid, and one of the two cleared detectable Salmonella outright. The bacteriophage produced no measurable reduction in dry dog food, and since every strain in the cocktail showed clean inhibition zones against the same phage stock on a plate, the likely culprits are formulation, strain specificity and delivery rather than phage biology itself failing.
The finding that will outlast the product verdicts is methodological. In feed like this, only culture-enriched sequencing agreed with culture. Enrichment was doing the work people usually assume sequencing depth is doing.
FAQ
Which anti-Salmonella additive works best in animal feed?
In the available trial data, a formaldehyde product performed best, eliminating detectable Salmonella from inoculated poultry feed entirely by most probable number. A formic acid product followed closely at a 98.4 percent reduction. A bacteriophage product applied to dry dog food produced no measurable reduction. All three rankings apply only within the tested matrix.
Why did the bacteriophage fail in dry dog food?
No reduction showed up by either most probable number or direct plating, though every Salmonella strain in the cocktail formed clear inhibition zones against the same phage stock on agar. That points away from susceptibility and toward delivery. In a dry, low-moisture matrix applied by spraying, phage particles may simply fail to reach bacterial cells.
What is Salmonella masking in feed testing?
Masking is the interference of formaldehyde and organic acids with culture recovery of Salmonella, producing false negatives that look like successful treatment. It is why an efficacy claim resting solely on plate counts deserves suspicion. Enriched molecular methods or viability-corrected assays are the practical way to check whether a negative result is real.
Why did formic acid increase Salmonella relative abundance while cutting counts by 98.4 percent?
Relative abundance is a proportion, not a count. Formic acid suppressed competing genera such as Pseudomonas and Pantoea faster than it suppressed Salmonella in the sequencing snapshot, so Salmonella gained share while losing absolute numbers. Culture measured the kill. Sequencing measured the reshuffling. Both findings are internally consistent.
Does shotgun metagenomics agree with Salmonella culture results?
Only after culture enrichment in this trial. Enriched shotgun libraries showed relative abundance drops of 99.9 percent for formaldehyde, 97.7 percent for formic acid and 33.1 percent for the phage, matching the culture ranking. Unenriched libraries did not agree, and reported Salmonella rising in formic acid treated poultry feed.
Why is culture enrichment used before sequencing animal feed?
Because it converts an invisible difference into a visible one. Enrichment grows whatever few survivors remain until they dominate the reads, so a treatment that removed everything yields nothing while one that left a few cells yields plenty. The cost is that the community you sequence now reflects the enrichment broth as much as the feed.
What bacteria live in poultry feed naturally?
Pseudomonas and Pantoea dominate the natural poultry feed microbiota, followed by Kosakonia, with Pseudomonadota as the leading phylum. Multiple independent surveys agree on Pseudomonas as a core community member. Both genera stayed dominant through inoculation and chemical treatment in unenriched samples, and only gave way to Clostridium after enrichment.
What bacteria dominate dry dog food?
Clostridium leads the natural community in dry dog food, followed by Vibrio and Pseudomonas, though total counts are tiny at about 34 CFU per gram. Extrusion at 90 to 200 degrees Celsius removes most bacteria and their DNA. That thin background is why inoculation alone flipped dominance to Salmonella before any treatment was applied.
How does formic acid control Salmonella in poultry feed?
It works as a feed acidifying agent. Lowering pH creates conditions that favor acid-tolerant organisms while suppressing or eliminating others, acting as an ecological filter rather than a general disinfectant. In this trial the effect was a 98.4 percent reduction by most probable number, close to but behind formaldehyde treatment.
How does formaldehyde control Salmonella in animal feed?
Formaldehyde is cleared for maintaining complete feeds or feed ingredients Salmonella negative for up to 21 days, applied here at 3 grams of product per kilogram of feed. It produced a 100 percent reduction in inoculated poultry feed, the strongest result in the trial. Worker safety concerns and the European Union refusal to authorize it remain relevant constraints.
How long can Salmonella survive in dry pet food and feed?
Days to weeks, depending on strain and conditions. Survival studies consistently find the steepest die-off within the first 24 hours, after which decline slows considerably. In this trial counts had already fallen one to two logs below the inoculation target before any treatment was applied, which is typical for dry matrices rather than anomalous.
What is LAMP testing for Salmonella in feed?
Loop-mediated isothermal amplification is a DNA screening assay that reports a time to positive value. In this work every group except formaldehyde treated enriched poultry feed returned a positive below 12 minutes, agreeing with both sequencing approaches. It is fast and cheap, but without viability treatment it detects DNA rather than living cells.
