Heritability Estimates in Japanese Thoroughbreds
The thoroughbred racehorse has been repeatedly bred to produce horses of the highest quality. While we believe that every year there is significant misallocation of resources in the industry with matings that simply do not stand a chance to producing a superior racehorse, and that mares are not selected on nearly as heavily as needed for true genetic improvement, the traits of speed and stamina, which may be related to optimum race distance, serve as a important indicator of success and are how we define the breed.
Heritability of performance is an important factor to understand. Just because we breed to a particular stallion, or use a particular mare for a mating because of their racing performance, it must be understood at first that heritability, as estimated for various measures such as handicap ratings and racing time has ranged between 30-40% (or h2 = 0.3 to 0.4) for handicap rating and racing times between 10-20% (or h2 = 0.1 to 0.2). As a side note, this is one reason why we have never understood why breeders in America like to use peak Beyer Speed figures (or any other speed figure), a figure that is based on the performance time of the day, as a guide to what stallion they should be using, or what mare they should be buying at auction. These are figures based on time, which has been shown to be lowly heritable (Formal Gold at a peak Beyer of 126 should be a much better stallion than A.P Indy at 114!). What the industry needs is a breeding value figure or a figure that shows the performance merit of one horse compared to all other horses, not just those that showed up on the day. Anyway, we digress....
Teruaki Tozaki and his team at the Department of Molecular Genetics in the Laboratory of Racing Chemistry in Japan have just released another very interesting paper in the Journal of Animal Breeding and Genetics. Previously Tozaki and his team had identified some variants within genes on Chromosome 18 associated with racing performance in Japanese thoroughbreds and this paper leans on those findings a little and confirms what they found genetically on a statistical basis. The paper is a study of the heritability estimates of racing performance in Japanese Thoroughbred Racehorses and evaluated a linear and non-linear model of the heritability of Lifetime Earnings and Lifetime Ranking using 3927 horses with their pedigrees traced back three generations. Summarily, the study found that at h2 = 0.25, or 25%, the heritability obtained from a non-linear model based on formal Japan Racing Association ranking was much higher than that obtained from a linear model h2 = 0.11, or 11%, equally the heritability of Lifetime earnings from the linear model was 12% and the non-linear model 19% . While establishing the heritability of both these measures was important in itself, the apparent variety of heritability estimates on a statistical basis correlates to their previous study in that it may indicate the existence of genes affecting ranking, and thus racing performance, in Thoroughbred racehorses in Japan. When the researchers categorized the horses in terms of race performance and separated out non-winning from winning horses, by removing the non-winning horses out of the equation, heritability estimates in the non-linear model jumped to as high as 0.34 (34%) in the various groups used. This again suggests on a statistical modeling basis that, in addition to other factors that may affect performance (injury, laryngeal paralysis, etc) non-winning horses may possess undesirable genetic factors that affect performance.