Genetic Testing Limitations in Warmblood Breeding: Rethinking the Role of DNA, EBVs, and Aggregate Scores

Top: Helgstrand's Stallion Revolution Bottom: Bocoy Stable's Hanoverian Mare Furstin Vera

Warmblood breeding has always balanced art and science, but in recent years the scale has tipped decisively toward data. Genetic testing, genomic breeding values, and aggregate indices now promise to quantify what was once judged primarily by eye, experience, and lineage. Stallions are marketed with numerical precision. Young horses are evaluated before they have ever been backed. Decisions that once relied on intuition now appear to be guided by objective measures. Yet beneath this growing confidence in data lies a more complicated reality. The biology of performance horses has not changed simply because our tools for measuring it have improved. The question facing modern breeders is not whether genetic testing is useful—it clearly is—but whether it is being interpreted correctly, and whether its limitations are being fully acknowledged.

At its core, genetic evaluation in warmbloods is an attempt to predict potential. Traditional estimated breeding values (EBVs) rely on pedigree, progeny performance, and competition results to isolate the heritable portion of a trait. More recently, genomic breeding values (GEBVs) have incorporated DNA markers, allowing breeders to estimate genetic merit earlier in a horse’s life. Aggregate indices—often referred to as total merit scores or AGR values—combine multiple traits into a single number intended to simplify selection. The appeal of these systems is obvious. Breeding is slow, expensive, and uncertain. Anything that promises earlier and more accurate selection is inherently attractive. However, the assumption that these numbers represent a definitive measure of quality is where problems begin.

“Breeding is slow, expensive, and uncertain”

One of the most important—and often overlooked—limitations of genetic testing in sport horses is the relatively low heritability of performance itself. Traits such as show jumping success, dressage performance, or even rideability are influenced by many genes, each contributing a small effect, and are heavily shaped by environment. Training, rider skill, management, nutrition, and opportunity all play substantial roles in determining outcomes. In statistical terms, heritability estimates for performance traits are modest, meaning that a large proportion of variation cannot be explained by genetics alone. This has a critical implication: genetic values describe probabilities, not guarantees. A horse with an exceptional breeding value for jumping is not destined to become a Grand Prix competitor, just as a horse with average values is not excluded from success. Genetics establish a range of potential, but they do not dictate where within that range an individual will fall.

Despite this, aggregate scores often create the impression of precision. By condensing complex genetic information into a single number, they offer clarity and comparability—but at the cost of nuance. These indices are constructed using weighted combinations of traits, and those weights vary between studbooks and breeding programs. One system may prioritize scope and power, another rideability and temperament. As a result, the same horse could rank differently depending on which index is used, even though its underlying genetics remain unchanged.

The issue is not that these indices are incorrect, but that they are contextual. They reflect specific breeding goals and specific datasets. When breeders compare scores across systems or treat them as universal measures of quality, they risk making decisions based on mismatched assumptions.

The reliability of any genetic evaluation also depends on the quality of the data behind it. In warmblood breeding, this remains a significant constraint. Unlike dairy cattle or poultry, where production traits can be measured precisely and consistently, sport horse data is inherently variable. Competition results are influenced by rider ability, opportunity, and career management. Conformation scoring and rideability assessments, while structured, retain elements of subjectivity. Even within a single discipline, standards can differ across countries and levels of competition. Genomic predictions do not eliminate these challenges; they depend on them. DNA markers are associated with traits based on observed relationships within a population. If the underlying performance data is inconsistent or biased, the resulting genomic predictions will reflect those imperfections. This is why genomic tools tend to be most powerful for traits that are easier to measure objectively and across large populations, and less reliable for complex, subjective characteristics such as temperament or trainability.

A Ultrasound Image of a Franklin x Skovens Rafael Pregnancy bred by Bocoy Stables (Vet work by Dr. Erin Newkrik of Well-Grove Equine in Loxahatchee, FL)

Another subtle but important limitation lies in how data is distributed. Breeding values often draw heavily from elite performance records, yet top-level sport represents a small and highly selective subset of the population. Horses that reach international competition have already passed through multiple filters—training quality, financial backing, and opportunity among them. Relying too heavily on this data can introduce bias, as it does not necessarily represent the broader genetic population.

Paradoxically, including data from lower levels of competition can improve the accuracy of genetic evaluation by providing a more complete and representative picture. However, such data is often underutilized or inconsistently recorded.

While these limitations suggest caution, they do not imply that genetic testing lacks value. In some areas, its contribution is both clear and indispensable. Screening for recessive disorders such as Warmblood Fragile Foal Syndrome has fundamentally improved breeding outcomes, allowing carriers to be managed responsibly and preventing the production of affected foals. Similarly, monitoring inbreeding coefficients and genetic diversity has become increasingly important as certain sire lines dominate the population. These applications highlight where genetic testing is most effective: in areas where traits are well-defined, directly linked to specific genetic variants, and less influenced by environmental factors. In contrast, the further a trait moves toward complexity—particularly those involving behavior, performance, or durability—the more cautious interpretation becomes necessary.

Rayna de Bocoy (Revolution x Sir Gregory) a resulting filly from an embroy transfer bred by Bocoy Stables, Vet work by Dr. Luis Cadena of Ocala, FL

The current challenge in warmblood breeding is not simply over-reliance on DNA, but uneven reliance. In some cases, breeders place excessive weight on aggregate scores, selecting stallions based on rankings without fully considering the individual mare, the compatibility of traits, or the long-term implications for genetic diversity. In other cases, valuable genetic information—particularly related to health and soundness—is underutilized or overlooked.

Perhaps the most significant risk is the gradual shift from evaluating horses to evaluating numbers. A breeding value is an abstraction, derived from models and assumptions. It does not capture the entirety of the horse in front of us—the subtle qualities of balance, mindset, and adaptability that often define success. When selection decisions become too heavily anchored in numerical rankings, there is a danger of narrowing the gene pool and losing traits that are difficult to quantify but essential in practice. A more balanced approach recognizes genetic data as one component of a broader decision-making framework. It is most useful when it informs, rather than dictates, choices. Breeding decisions benefit from integrating genomic information with careful observation of phenotype, knowledge of family lines, and an understanding of how specific traits interact. The mare, in particular, remains a central and sometimes underappreciated influence, contributing not only genetics but also developmental environment and behavioral tendencies.

“DNA can reveal hidden risks, clarify probabilities, and accelerate selection, but it cannot fully predict the outcome of a breeding decision.”

Ultimately, the promise of genetic testing in warmblood breeding lies not in replacing traditional evaluation, but in refining it. The most effective use of these tools comes from understanding both their strengths and their limitations. DNA can reveal hidden risks, clarify probabilities, and accelerate selection, but it cannot fully predict the outcome of a breeding decision.

Sport horse breeding remains, in part, an exercise in uncertainty. No model can account for every variable, and no index can capture every quality that contributes to success. The goal, therefore, is not to eliminate uncertainty, but to manage it intelligently—using genetic data where it is robust, questioning it where it is less certain, and always grounding decisions in the reality of the horse itself.

In this sense, the future of warmblood breeding will not be determined by how much data is available, but by how thoughtfully it is used.