More Stallions, Same Gene Pool: What a Brazilian Friesian Study Reveals About Breeding Diversity

A Friesian breeder can choose a mating with a low five-generation inbreeding coefficient and still contribute to a gene pool that is becoming narrower. That may sound contradictory, but it is one of the most important lessons to emerge from a recent pedigree study of Friesian horses in Brazil.

The 2026 study, "Pedigree-Based Assessment of Genetic Structure and Disease-Associated Variants in Friesian Horses in Brazil," does not demonstrate that Brazilian Friesians are genetically distinct from the broader population. What it provides is a useful real-world example of how a population can contain hundreds of horses and many names in its pedigrees, yet still depend heavily on a remarkably small number of genetic contributors. For breeders, health advocates, and those who remain skeptical that genetic concentration is a meaningful problem in the Friesian breed, that distinction deserves closer attention.

A Population Can Look Larger Than It Is

Researchers examined the pedigrees of all 416 Friesians registered in Brazil between 2003 and 2024. By tracing those pedigrees through international databases, they expanded the pedigree to 1,647 horses, including 1,231 horses in the ancestral population. Five Brazilian horses were excluded from the statistical analysis because missing parents or grandparents could have produced artificially low inbreeding values, leaving a reference population of 411 horses.

The analysis identified 1,306 total ancestors and 58 founders. On paper, that sounds like a substantial genetic base, but those ancestors did not contribute equally to the present population.

The researchers calculated an effective number of ancestors of only 10.19. This does not mean that the population literally descends from just ten horses. It means the ancestral contributions were so uneven that the remaining pedigree diversity in the current population was approximately equivalent to what would be expected if only about 10 ancestors had contributed equally.

The estimated effective population size was similarly low, at 32.6. Effective population size is not a head count. It estimates how large a population behaves genetically after accounting for unequal reproductive contributions and the rate at which genetic variation is being lost. A population may contain hundreds or thousands of registered horses, but if only a small proportion of them produce most of the next generation, it behaves genetically as though it were much smaller.

The reason for those low effective numbers becomes clearer when sire use is examined. Just three stallions sired 148 of the 411 horses, representing 36% of the reference population, and only 14.4% of the males had recorded offspring. Brazil therefore had hundreds of registered Friesians and more than a thousand horses represented in the expanded pedigree, but a much smaller group made a meaningful genetic contribution. Counting horses or pedigree names alone would greatly overstate the diversity actually being maintained.

Four Measurements, Four Different Questions

Part of the confusion surrounding Friesian breeding is that several related measures are often discussed as though they are interchangeable. They are not, because each answers a different question.

• Five-generation inbreeding coefficient. This value estimates the inbreeding expected in a proposed foal from ancestors shared by the sire and dam within the first five generations. It is a practical tool for identifying and avoiding close recent matings, but it does not include all known pedigree generations in the horse's ancestry. The KFPS advises the five-generation inbreeding coefficient not to exceed 5%.

  
  

• Full pedigree inbreeding coefficient. This estimate uses all available generations in an individual horse’s recorded pedigree. It reflects the accumulated shared ancestry expected in that horse, although it remains dependent on the depth and accuracy of the recorded pedigree. This number is not currently published for KFPS horses. KFPS foals born in 2022 had an average full pedigree inbreeding coefficient of 17.0%

  
  

• Kinship percentage (mean kinship). This estimates how closely related an individual horse is, on average, to a defined reference population. A lower-kinship stallion carries ancestry less common in the current population and may therefore offer greater value in maintaining or improving the population's genetic diversity. As of 2022, the population mean kinship was 17.8%.

  
  

• Inbreeding increase per generation. This measures how much additional inbreeding is accumulating in the population over approximately one generation (10 years). Recent research estimated that the current KFPS population inbreeding increase per generation was 0.72% over the 2013 through 2022 period, whereas the preferred population management goal is below 0.5% per generation.

A low five-generation inbreeding coefficient can help prevent close mating, but it does not prove that the mating preserves uncommon ancestry within the breed.

---

In the 2024 study "Evaluation of Breeding Strategies to Reduce the Inbreeding Rate in the Friesian Horse Population: Looking Back and Moving Forward", Steensma and colleagues calculated inbreeding using all known pedigree generations. KFPS foals born in 2022 had an average full pedigree inbreeding coefficient of 17.0%, while population mean kinship was 17.8%. A proposed foal can therefore have a low five-generation coefficient while still inheriting the much deeper shared ancestry already present throughout the population.

There is important historical context behind the KFPS emphasis on recent generations. The Friesian studbook has been closed for generations, and the breed has experienced severe population bottlenecks combined with disproportionately large contributions from influential ancestors. Breeders today cannot reverse genetic diversity that was lost many generations ago, but they can avoid adding excessive close inbreeding in the next foal. The five-generation coefficient is best understood as a breeding-management tool designed to guide the mare owner's stallion selection for a breeding, rather than as a complete description of all the inbreeding present in a horse.

Full pedigree calculations also have limitations. The oldest known horses in a pedigree must eventually be treated as founders, and those founders are generally assumed to be unrelated even when their true relationships are unknown. Pedigrees also differ in depth and completeness, so a horse with more recorded ancestry may reveal more shared connections than a horse whose pedigree contains gaps.

The Brazilian researchers demonstrated this problem directly when they excluded horses with missing parents or grandparents because incomplete pedigrees could create artificial inbreeding values of zero. Steensma et al. found that the KFPS pedigree was unusually complete in recent decades, but even a deep pedigree produces an estimate based on recorded ancestry rather than a perfect measure of every genetic relationship. Greater transparency could be achieved by presenting the five-generation inbreeding coefficient results alongside full pedigree inbreeding and kinship information, with a clear explanation of what each figure represents.

An alternative is a genomic inbreeding value, which measures how much matching DNA a horse actually inherited from common ancestors through both parents rather than estimating inbreeding from recorded relationships alone. The KFPS 70K SNP test reads tens of thousands of DNA markers across the genome, and those data can be used to identify long matching segments, called runs of homozygosity. This makes genomic values more directly applicable to an individual horse than pedigree coefficients, which estimate expected inheritance. Although the KFPS has collected 70K SNP data from foals since 2023, routine use would still require a sufficiently large and representative reference population, validated standards for calculating and interpreting results, integration into breeding tools, and funding for analysis, database development, and testing of horses without existing SNP data. Genomic inbreeding is therefore technically possible in the future with the data already being collected, but a dependable, population-wide tool is likely years away if pursued by the KFPS.

The Popular Sire Effect in Brazil

The Brazilian study found that just three stallions sired 148 of the 411 horses analyzed, representing 36% of the reference population. Only 14.4% of the males had recorded offspring, compared with 32.2% of the females. Once horses with no offspring were removed from the calculation, breeding males averaged considerably more offspring than breeding females. This pattern is called the popular sire effect.

When one stallion produces a large number of offspring, his genetics rapidly become common. If his sons, daughters, grandsons, and other close relatives are also selected heavily, the influence of that family can expand for generations.

---

Popular sire use is not necessarily the result of irresponsible breeding. A stallion may be heavily selected because he has exceptional movement, conformation, character, sport performance, offspring results, or commercial appeal. Artificial insemination also allows a single stallion to breed mares over enormous distances, but the population-level consequence remains the same: the more offspring one stallion produces, the less room remains for other male lines to contribute. Any desirable variants he carries become widespread, but so may recessive disease variants or other risks that were not known when he entered breeding.

The overuse of popular sires also makes genetic drift more powerful. Genetic drift is the random increase, decrease, or loss of genetic variants from one generation to the next. It occurs in every population, but its effects are stronger when the effective population is small and only a limited number of families contribute substantially. A rare family line or useful genetic variant may disappear simply because few descendants are selected for reproduction, while other variants may become common because they happen to be carried by families receiving the most breeding opportunities. The popular sire effect creates the imbalance, and genetic drift can then accelerate the loss of some variants and the spread of others within the narrowed population.

How Brazil Compares With the Wider KFPS Population

The Brazilian result is striking, but it should not be compared too casually with the wider KFPS population. The Brazilian figure represents the cumulative contribution of three stallions to 411 horses born over the study period, whereas the KFPS data reported by Steensma et al. reflect annual sire contributions. These are not identical calculations, but both studies demonstrate that the number of available stallions is not the same as the number contributing meaningfully to the next generation.

Steensma et al. reported that 322 different sires produced at least one registered Friesian foal in 2022. Only 89 were approved KFPS studbook stallions, but those approved stallions sired 3,547 of the year’s 3,985 foals, or 89% of the entire foal crop. Most of the remaining sires produced only one foal, so counting all 322 sires creates an impression of much broader participation than actually occurred.

Use was also distributed unequally within the approved group. The ten most-used approved stallions collectively produced approximately 35% of all foals born that year. Had all 89 approved stallions contributed equally, the ten leading sires would have produced only about 11%, rather than roughly three times that share.

The KFPS population has improved substantially compared with its history. In 1980, the ten most-used stallions produced approximately 75% of that year’s foals, while by 2022 the figure had fallen to about 35%. However, researchers concluded that this decline resulted mainly from an increase in the number of approved stallions, not from truly equal use of the stallions available.

More stallions do not automatically lead to greater diversity if the additional stallions each produce only a small number of foals while the same popular stallions continue to sire a large percentage of the population.

---

What the Research Found Would Reduce Inbreeding

Steensma et al. explored various methods to reduce inbreeding through computer simulations based on the KFPS population structure. Increasing the number of breeding stallions from 90 to as many as 150 had little effect on the projected inbreeding increase when the unequal pattern of sire use remained in place. Even when the larger sire population was combined with a limit of 80 offspring per stallion per year, the projected inbreeding increase fell only from 0.55% to 0.51% per generation.

By contrast, selecting new breeding stallions according to mean kinship was much more effective. Limiting selection to stallions with below-average kinship to the recent foal population reduced the projected inbreeding increase per generation from approximately 0.66% to 0.33%. Stratified breeding limits, which allowed greater use of lower-kinship stallions and progressively less use of higher-kinship stallions, reduced the projected increase to 0.43% per generation and performed better than a single general quota.

Breeding limits are most effective when they do more than reduce popular-sire use; they must redirect breedings toward stallions whose genetics are less common in the population

---

The implication is not that quality should be abandoned in favor of rarity. A stallion should still meet appropriate expectations for health, character, function, and breed type. Steensma et al. acknowledged that strict mean-kinship selection could exclude some high-kinship candidates with strong breeding values, which helps explain why the most effective population strategy may be difficult to apply within a system that also rewards inspection results, marketability, and rapid progress in selected traits. Population-wide relatedness nevertheless has value and should be considered alongside more familiar selection criteria.

Diversity Depends on Distribution

The most useful lesson from Brazil is not that breeders should panic over one country’s inbreeding percentage, nor is it that the answer is simply to approve more stallions. The lesson is that diversity depends on how breeding opportunities are distributed. A population may contain hundreds of horses, dozens of available stallions, and thousands of names in its pedigrees while still depending heavily on a small number of ancestors.

A breeder may select a mating below the KFPS five-generation inbreeding recommendation of 5% while choosing a stallion whose genetics are already widespread throughout the population. The individual foal’s five-generation inbreeding coefficient and the population’s genetic diversity are connected, but they do not answer the same question. The first asks how closely related the foal’s parents are within recent generations, while the second asks whether the mating preserves genetic material that is becoming uncommon or simply adds more of what the breed already has in abundance. Responsible breeding requires both perspectives.

Brazil offers a small-scale view of how quickly a gene pool can become concentrated. The wider KFPS population shows that the same popular-sire pressure can persist even when many approved stallions are technically available, and the research indicates that preserving diversity requires more than simply approving additional stallions. It also requires attention with regard to how heavily individual stallions and closely related families contribute, as well as recognition of the population value carried by less-related bloodlines.

For the future of the Friesian horse, the important question is not simply how many stallions the breed has. It is about how many are truly contributing, how closely related they are, and whether today’s breeding decisions leave the next generation with more or fewer genetic options.

References

Botelho Rocha, I. A., Araujo, F., & Patterson Rosa, L. (2026). Pedigree-based assessment of genetic structure and disease-associated variants in Friesian horses in Brazil. Journal of Equine Veterinary Science, 160, 105860. https://doi.org/10.1016/j.jevs.2026.105860.

Steensma, M. J., Doekes, H. P., Pook, T., Derks, M. F. L., Bakker, N., & Ducro, B. J. (2024). Evaluation of breeding strategies to reduce the inbreeding rate in the Friesian horse population: Looking back and moving forward. Journal of Animal Breeding and Genetics, 141, 668–684. https://doi.org/10.1111/jbg.12872.

Royal Friesian Horse Studbook. (2023, October 6). Meer dan 2000 haarmonsters in Schotland onderzocht. https://kfps.nl/nieuws/meer-dan-2000-haarmonsters-in-schotland-onderzocht/.

Royal Friesian Horse Studbook. (2026). Hengsteninformatie maart 2026. https://kfps.nl/app/uploads/2026/02/Hengsteninformatie-2026-LR-2.pdf