The Anticoagulant Treadmill: VKORC1 Resistance, Secondary Poisoning, and a Policy Instrument That Did Not Work

1. Introduction: a cycle rather than a progression

The conventional account of rodenticide development is a story of progress. Warfarin and its contemporaries arrived in the early 1950s and transformed rodent control. When resistance appeared, chemists responded with a second generation of more potent compounds. Where those in turn began to fail, a third tier of options emerged. Each step is presented as a solution to the problem created by the step before it.

This paper argues the sequence is better read as a treadmill. Each escalation solved the immediate efficacy problem and simultaneously deepened two others: it increased the persistence of the compound in the bodies of animals that were never the target, and it left the underlying selection pressure entirely intact, so that the next round of resistance was not merely possible but scheduled.

That framing is not ours alone. Researchers examining exposure in Iberian raptors state it directly: the difficulty of managing widespread use of second-generation anticoagulants in agricultural and urban environments is compounded because resistance in target species drives the continuous use of baits, which exacerbates exposure risk for non-target wildlife.¹ Elsewhere the same literature notes that resistance leads to greater environmental contamination as rodenticides persist and bioaccumulate in non-target species, exacerbating the risks of secondary poisoning and biomagnification in the food chain.¹

2. How anticoagulants kill

The mode of action is worth stating precisely, because both the resistance mechanism and the secondary poisoning problem follow directly from it.

Vitamin K epoxide reductase, the enzyme complex encoded in part by VKORC1, regenerates the reduced form of vitamin K. That reduced form is an essential precursor for gamma-glutamyl carboxylase, which catalyses a post-translational modification of vitamin K dependent proteins required for the cessation of bleeding.²

Anticoagulants inhibit the reductase. The supply of functional clotting factors falls, and the animal dies of haemorrhage over a period of days rather than immediately. That delay was originally a design feature: it prevents bait shyness, because the rodent does not associate the meal with the illness.

2.1 Why the delay creates the wildlife problem

The same delay is the origin of the ecotoxicology. A rodent that has consumed a lethal dose remains alive and increasingly impaired for several days. During that window it is unusually easy to catch. Predators are therefore not merely exposed to poisoned prey at random. They are preferentially exposed, because the poisoning process makes the prey easier to take.

When consumed in lethal or sublethal doses, these compounds bioaccumulate in the liver, becoming a risk for non-target predators.³

3. The VKORC1 mechanism

Since the 1950s, resistance to anticoagulant rodenticides in several rodent species has emerged through single nucleotide mutations in the vitamin K epoxide reductase complex subunit 1 gene, often located in its exon 3.⁴

3.1 The characterised alleles

The best-characterised mutations are the amino acid replacements Tyr139Cys, Tyr139Phe and Leu120Gln, which are known to mediate resistance to all first-generation anticoagulants and to some of the second-generation compounds, specifically bromadiolone and difenacoum.⁴

That final clause carries most of the operational content in this paper. Resistance is not a binary state in which a rodent either responds to anticoagulants or does not. It is compound specific. A population carrying Y139C may be fully resistant to warfarin, substantially resistant to bromadiolone and difenacoum, and still susceptible to brodifacoum or difethialone.

3.2 Why this shapes practice

The consequence is that anticoagulant failure in the field is a diagnostic event, not simply a reason to escalate. The specific pattern of what works and what does not carries information about which allele is present, and therefore about which compounds remain available.

In practice the diagnostic step is skipped. An operator experiencing poor control moves to the most potent available product, which is rational at the level of the individual job and is exactly the behaviour that drives the treadmill at the level of the population.

3.3 Mutations as a proxy for control failure

Because the genetics are well characterised, VKORC1 genotyping has become a monitoring tool. Studies have examined the prevalence of single nucleotide polymorphisms in urban rat populations specifically to inform the effectiveness of anticoagulant use, sequencing exon 3 across samples collected from urban settings.² One such study analysed 130 rat tail samples comprising 83 Rattus norvegicus and 47 of the Rattus rattus complex, collected between late 2016 and the end of 2019.²

Another analysed 243 tail tips of house mice from mainland Portugal, southern Spain, the Azores and Madeira, sequencing all three exons of VKORC1.⁵ The method is neither exotic nor expensive, which becomes relevant in §12.

4. Standing variation and the speed of the response

One finding in this literature deserves more attention than it usually receives, because it explains why anticoagulant resistance appeared so quickly and why it will continue to do so.

The adaptive alleles in R. norvegicus have been shown to segregate in the ancestral environment, and these formed the genetic basis for rapid resistance evolution against warfarin, with several natural allelic VKORC1 variants conferring resistance.⁴

4.1 What this means

Resistance did not have to wait for a new mutation. The variation was already present in the population before warfarin was invented, presumably maintained for reasons unrelated to rodenticides. Selection did not create the alleles. It sorted them.

4.2 The titles tell the story

One paper in this literature is titled, pointedly, as an enquiry into novel mutations in the VKORC1 gene of wild rats and mice, framed as a response to fifty years of selection pressure by warfarin.⁶ That is a fair description of what the global rodent control enterprise has been doing, viewed as an experiment in evolutionary biology.

5. The historical sequence

The timeline is compact and the intervals are short.

First-generation anticoagulants were introduced in the early 1950s, and resistance to them emerged later that same decade.⁵ More precisely, resistance was first detected in brown rats in 1958 and in house mice in the early 1960s, both in the United Kingdom.⁷

Since that initial observation, resistance has been reported worldwide: across many European countries, in the United States, in Canada, in Japan and in Australia.⁷

The emergence of resistance to the first generation, comprising warfarin, diphacinone, coumatetralyl and chlorophacinone, led to the development of second-generation molecules in the 1970s and 1980s: bromadiolone, difenacoum, flocoumafen, brodifacoum and difethialone.⁷ Second-generation rodenticides were then developed, with resistance to them being reported in the late 1970s.⁵

Read the intervals. Roughly eight years from introduction to documented resistance for the first generation in rats. Roughly the same for the second generation. The chemistry changed. The waiting period did not.

6. Escalation and its ecological price

The response to resistance has been consistent across seven decades: deploy something stronger. The literature is explicit about the consequence. Research linking resistance to VKORC1 mutations has led to the use of more toxic and environmentally harmful compounds.⁵

6.1 The specific substitution

Because Y139C and its relatives confer resistance to bromadiolone and difenacoum but not necessarily to the most potent second-generation compounds,⁴ the practical response to resistance is substitution toward brodifacoum and difethialone.

This substitution is documented in the field. Researchers examining Madeira note that reduced effectiveness of bromadiolone may account for the high levels of difethialone detected there, used as an alternative to a less effective anticoagulant.³

6.2 Why persistence is the variable that matters

The compounds that resistance pushes operators toward are precisely those with the longest hepatic half-lives. A predator eating a rodent poisoned with a first-generation compound receives a dose that clears comparatively quickly. A predator eating a rodent poisoned with brodifacoum receives a dose that persists, and that accumulates with each subsequent poisoned prey item.

The bioaccumulation is observable in the age structure of contaminated wildlife. Older raptors show a higher likelihood of accumulating greater concentrations of residues, consistent with bioaccumulation through consumption of multiple rodent prey over their lifespan.⁸

7. The evidence from raptors

Birds of prey are the sentinel group for this problem, and the monitoring data are stark.

7.1 Detection rates

In Portuguese raptor monitoring, the Eurasian eagle-owl (Bubo bubo), the common kestrel (Falco tinnunculus) and the tawny owl (Strix aluco) had anticoagulant detection rates exceeding ninety per cent.⁸

In a related study of raptors from continental and insular regions of the Iberian Peninsula, results revealed a high prevalence, above eighty per cent, of second-generation anticoagulants, with brodifacoum and bromadiolone the most frequent. Multiple second-generation detections in the same individual were also common, at approximately fifty per cent.¹

7.2 Species and landscape effects

Exposure is not uniform. Species and region significantly influence the likelihood of exposure, with kestrels showing a greater probability of exceeding 100 nanograms per gram wet weight than buzzards.¹ Raptors from insular territories were ten times more likely to have higher concentrations than those from continental areas.¹

The island effect is mechanistically unsurprising. Small land areas with intensive rodent control and restricted prey ranges concentrate exposure. It is also a warning about closed systems generally, which is relevant to any setting where rodent control is intensive and the predator population is constrained.

7.3 Population-level consequences

The question of whether residue burdens translate into population effects is harder, but there is evidence. Work in this area reports that annual abundance of common kestrels is negatively associated with second-generation anticoagulant rodenticides,⁹ and separate probabilistic work has attempted to establish toxicity threshold values for the lethal effect of these compounds in free-living British raptors.⁹

High exposure rates have also been documented in predatory birds in intensively managed landscapes in Denmark.⁹ This is not a regional anomaly.

7.4 The sentinel argument

Researchers make a further point that has practical value. Residues detected in top predators serve as a proxy for the level of environmental anticoagulant exposure, and may also reflect the level of resistance in pest rodents.³ Kestrels and barn owls, being widespread, are identified as effective sentinels for monitoring.⁸

This closes the logical loop of the paper. Raptor residue levels are a readout of both halves of the treadmill at once, because high residues indicate both heavy deployment and, by inference, the resistance that is driving it.

8. A policy instrument that did not work

This section concerns a single finding that we regard as the most important result in this literature for anyone thinking about regulation.

8.1 The finding

A legal restriction on second-generation anticoagulant bait concentrations came into effect in 2018. Examining exposure in Iberian raptors afterwards, researchers report that the restriction did not significantly impact exposure levels.¹

8.2 Why a concentration limit might fail

The paper does not model the mechanism, so what follows is our reasoning rather than a reported finding. Several explanations are consistent with the data.

Total dose is a function of consumption, not concentration. A rodent feeding on lower-concentration bait may simply feed for longer or more often, arriving at a comparable liver burden. Anticoagulants are chronic toxicants and the exposure that matters is cumulative.

Resistance raises consumption directly. A resistant animal survives longer while continuing to feed, which increases the total quantity ingested before death.

Predator burden integrates across many prey. A raptor's liver concentration reflects dozens of prey items over months. Reducing the dose per item does not obviously reduce the integral if the number of poisoned items available stays constant, and resistance keeps that number high.

Substitution effects. Restricting concentration may push operators toward more applications or toward compounds outside the restricted set.

All four mechanisms share a common structure: the regulation acted on concentration while the harm is driven by total deployment, and total deployment is driven by resistance. This is why we argue the two problems cannot be separated.

9. Resistance in the predators themselves

There is a further development in this literature that is genuinely unsettling and that has not reached practitioner awareness at all.

Given the high prevalence and concentrations of anticoagulants detected in non-target vertebrates including carnivorous Mustelidae, researchers hypothesised that secondary exposure through feeding on poisoned prey may itself cause selection along the food chain, and that VKORC1-based resistance might therefore have evolved in the predators of rodents.⁴

Using mustelid-specific primers for direct sequencing of genomic DNA, they studied VKORC1 polymorphisms in 115 mustelids of five species, comprising pine marten, stone marten, weasel, stoat and polecat, obtained from northern Denmark. The work yielded six sites with nonsynonymous amino acid polymorphisms in exon 3, alongside several synonymous ones.⁴

9.1 The implication

If secondary exposure is strong enough to act as a selective force on predators, then rodenticide deployment is not merely contaminating the food chain. It is applying evolutionary pressure to species that were never targeted, in a direction nobody chose.

We note carefully that the cited work reports polymorphisms and a hypothesis, not demonstrated resistance. The presence of nonsynonymous variation in exon 3 of a predator's VKORC1 is suggestive and it is not proof. But the question has now been asked with data attached, and that is a meaningful change.

10. Alternatives and partial answers

The treadmill argument invites the obvious question of what else is available.

10.1 Cholecalciferol

One line of work examines cholecalciferol, vitamin D3, which kills by inducing hypercalcaemia rather than by anticoagulation and therefore does not select on VKORC1. Researchers examined the effect of a low concentration of cholecalciferol in brodifacoum bait on bait consumption by Norway rats and on control success in both laboratory and field studies.¹⁰

Cholecalciferol caused a strong stop-feed effect after two days in the laboratory study.¹⁰ Field trials applied bait containing either 25 mg per kg brodifacoum alone, or 25 mg per kg brodifacoum combined with 100 mg per kg cholecalciferol, with infestations assessed before and after treatment, rats radio-tagged, carcasses searched for during treatment, and each rat genotyped to determine resistance status conferred by VKORC1.¹⁰

The stop-feed effect is the interesting variable. A rodent that stops consuming bait after two days ingests less total active ingredient, which in principle reduces the residue available to a predator that later eats it. This is a route to reducing secondary exposure that operates on consumption rather than on concentration, which is precisely the variable that §8 suggests the regulatory approach missed.

10.2 The limits of substitution

We would caution against reading any of this as a clean escape. Cholecalciferol has its own non-target profile, particularly for dogs. More fundamentally, the treadmill described in this paper is a structural consequence of relying on toxicant deployment against a fecund, genetically variable species with standing adaptive variation. Substituting the toxicant changes which gene is under selection. It does not remove the selection.

The interventions that genuinely escape the cycle are the unglamorous ones: exclusion, harbourage removal, and sanitation. They are also the ones no manufacturer has a commercial reason to promote.

11. Why resistance management and wildlife protection are one problem

The two constituencies concerned with anticoagulants rarely speak to each other. Pest management professionals discuss resistance as an efficacy problem. Conservation biologists discuss secondary poisoning as a contamination problem. They are describing two ends of the same causal chain.

Resistance reduces field efficacy. Reduced efficacy increases the quantity and persistence of compound deployed. Increased deployment raises predator exposure. None of this reduces the selection pressure, so resistance continues to increase.

The practical consequence is that measures which genuinely reduce resistance selection also reduce wildlife exposure, and measures which reduce wildlife exposure only by limiting concentration may do neither.¹ A programme that achieves control through exclusion and sanitation, using toxicant sparingly and with genotype-informed product selection, serves both objectives at once.

We would go further. The pest management industry has a stronger interest in raptor conservation than it realises, because raptors are both a free rodent control service and, as the sentinel argument in §7.4 establishes, a readout on the state of its own resistance problem.

12. The Manitoba position

What can be said about this province specifically is limited, and the limitation is itself the point.

Anticoagulant resistance has been reported in Canada.⁷ We have found no published VKORC1 survey of Manitoba or Winnipeg rodent populations, and no published raptor residue monitoring for this province.

Manitoba's position is worth stating accurately, because it is commonly confused with Alberta's. Alberta has maintained an essentially rat-free province since 1950 through a control zone along its eastern border with Saskatchewan.¹¹ Manitoba has no equivalent programme. Norway rats reached Saskatchewan in the 1920s and spread northwest from there,¹¹ which means they passed through this province first, and they are established here.

That makes the resistance question more pressing rather than less. Manitoba has established commensal rat populations receiving anticoagulant exposure from agricultural, municipal and commercial sources, with no published record of what alleles those populations carry.

The method is not difficult. The studies cited here sequenced exon 3 of VKORC1 from tail tip samples, in one case 130 samples collected opportunistically from urban settings over three years.² That is within reach of a provincial collaboration between pest management operators, who handle the carcasses routinely, and a university laboratory.

We would consider this the second most valuable piece of unfunded local research in Manitoba pest management, after bed bug resistance genotyping. It would tell operators which compounds remain useful, and it would give regulators a factual basis for any future restriction.

13. Limitations and open questions

The evidence base is heavily European. The raptor monitoring cited is predominantly Portuguese, Spanish, Danish and British.¹⁸⁹ Species composition, land use and regulatory regimes differ from Canada. The mechanisms transfer; the prevalence figures should not be assumed to.

Detection rate is not the same as harm. A ninety per cent detection rate establishes exposure, not mortality or population effect. The link to population-level consequence rests on a smaller body of work.⁹

Our explanation for the 2018 result is inference. The finding that the concentration restriction did not significantly affect exposure is reported.¹ The four candidate mechanisms in §8.2 are our reasoning and are not tested in the cited work.

Predator resistance is hypothesis, not finding. The mustelid work reports VKORC1 polymorphisms and proposes food-chain selection as an explanation.⁴ It does not demonstrate functional resistance, and we have been careful not to claim it does.

Chart values are thresholds. The raptor detection figure plots reported thresholds such as "exceeding ninety per cent" as point values for legibility. Readers needing exact figures should consult the sources.

14. Conclusion

Seventy years of anticoagulant rodenticide use has produced a well-documented cycle. First generation compounds arrived in the early 1950s and met resistance within roughly eight years.⁵⁷ Second generation compounds followed in the 1970s and met resistance within roughly the same interval.⁵⁷ The resistance rests on a small set of VKORC1 mutations, principally in exon 3, which confer protection against all first-generation compounds and against some of the second.⁴ The adaptive alleles were already present in the population before the compounds existed.⁴

The escalation that resistance drives has a price that is paid by species that were never the target. Detection rates exceed ninety per cent in several sentinel raptor species,⁸ second-generation prevalence exceeds eighty per cent in Iberian raptors with multiple simultaneous detections in roughly half of individuals,¹ and burden increases with age in a pattern consistent with lifetime bioaccumulation.⁸

The regulatory instrument most commonly reached for, limiting the concentration of active ingredient in bait, was implemented in 2018 and did not significantly change exposure levels.¹ That result should reshape the conversation. It suggests the harm is driven by total deployment rather than by dose per bait, and total deployment is driven by resistance.

The conclusion we draw is that the industry's efficacy problem and the conservation sector's contamination problem are the same problem, and that the measures which actually break the cycle are the ones that reduce reliance on toxicant altogether. For Manitoba the immediate step is smaller and entirely achievable: find out what alleles are actually present in local rodent populations, because at the moment nobody knows, and every product decision being made in this province is being made without that information.

References

1. Martin-Cruz, B. et al. Differential exposure to second-generation anticoagulant rodenticides in raptors from continental and insular regions of the Iberian Peninsula. Environmental Pollution. Source for SGAR prevalence above 80%, brodifacoum and bromadiolone as most frequent, multiple detections at approximately 50%, kestrel versus buzzard exceedance of 100 ng/g ww, the ten-fold insular effect, the finding that the 2018 legal restriction on SGAR bait concentrations did not significantly impact exposure levels, and the statement that resistance in target species drives continuous baiting. https://www.sciencedirect.com/science/article/pii/S0269749124017494
2. VKORC1 mutations in rodent populations of a tropical city-state as an indicator of anticoagulant rodenticide resistance. Scientific Reports (2022), 12, s41598-022-08653-8. Source for the vitamin K epoxide reductase and gamma-glutamyl carboxylase pathway, and for the 130-sample urban rat survey sequencing exon 3 of VKORC1. https://www.nature.com/articles/s41598-022-08653-8
3. Toxic legacy: the hidden impact of anticoagulant rodenticides on Portuguese raptors. Science of the Total Environment. Source for hepatic bioaccumulation, the sentinel-proxy argument linking raptor residues to both environmental exposure and rodent resistance levels, and the substitution of difethialone where bromadiolone effectiveness is reduced. https://www.sciencedirect.com/science/article/pii/S0048969725021874
4. The potential of VKORC1 polymorphisms in Mustelidae for evolving anticoagulant resistance through selection along the food chain. Source for exon 3 mutation location, the Tyr139Cys, Tyr139Phe and Leu120Gln alleles and their compound-specific coverage, the segregation of adaptive alleles in the ancestral environment, and the 115-mustelid five-species survey yielding six nonsynonymous polymorphic sites. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6715177/
5. Carromeu-Santos, A. et al. (2023). Widespread distribution of rodenticide resistance-conferring mutations in the Vkorc1 gene among house mouse populations in Portuguese Macaronesian islands and Iberian Atlantic areas. Science of the Total Environment, 900, 166290. Source for the FGAR introduction and resistance timeline, SGAR resistance reported in the late 1970s, the 243-sample three-exon survey, and the observation that resistance has led to use of more toxic and environmentally harmful compounds. https://pubmed.ncbi.nlm.nih.gov/37586516/
6. Rost, S., Pelz, H.-J., Menzel, S., MacNicoll, A.D., Leon, V., Song, K.-J., Jaekel, T., Oldenburg, J. & Mueller, C.R. (2009). Novel mutations in the VKORC1 gene of wild rats and mice, a response to 50 years of selection pressure by warfarin? BMC Genetics, 10, 4. doi:10.1186/1471-2156-10-4 https://www.sciencedirect.com/science/article/pii/S0160412026000577
7. Adaptative evolution of the Vkorc1 gene in Mus musculus domesticus is influenced by the selective pressure of anticoagulant rodenticides. Source for first detection of resistance in brown rats in 1958 and house mice in the early 1960s in the UK, the worldwide spread including Canada, the first-generation compound list, and the 1970s and 1980s development of the second-generation molecules. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5395456/
8. Carromeu-Santos, A. et al. Portuguese raptor monitoring. Source for detection rates exceeding 90% in Bubo bubo, Falco tinnunculus and Strix aluco, the identification of kestrels and barn owls as effective sentinel species, and the age-related bioaccumulation pattern. https://www.sciencedirect.com/science/article/pii/S004896972304915X
9. Probabilistic approach reveals the toxicity threshold values of free-living raptors in Great Britain for the lethal effect of second-generation anticoagulant rodenticides. Environment International. Includes reference to the negative association between annual kestrel abundance and SGARs (Ecotoxicology, 30, 560-574) and to high exposure rates in Danish predatory birds (Archives of Environmental Contamination and Toxicology, 63, 437-444). https://www.sciencedirect.com/science/article/pii/S0160412026000577
10. The stop-feed effect of cholecalciferol (vitamin D3) and the efficacy of brodifacoum combined with cholecalciferol in Y139C-resistant Norway rats (Rattus norvegicus). Source for the two-day stop-feed effect, the 25 mg/kg brodifacoum and 100 mg/kg cholecalciferol field trial design, radio-tagging, carcass recovery and VKORC1 genotyping methodology. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9948778/
11. Government of Alberta. History of rat control in Alberta. Source for the establishment of the Rat Control Zone along the Alberta-Saskatchewan border in the early 1950s, the arrival of Norway rats in Saskatchewan in the 1920s, their northwestward spread at roughly 24 km per year, and Alberta's maintenance of essentially rat-free status. Note that this programme is Alberta's; no equivalent exists in Manitoba. https://www.alberta.ca/history-of-rat-control-in-alberta