HMGA2 (pony)

What it does

Crossing two of them

Its gene carrot

Gene key

horsegenetics.hmga2

Priority

85

Alleles

p N

Outcomes

pony, pony-mixed, pony-plain

Coat effect

none — every outcome is a wild type

Default allele

N

Wild population

p(p) = 0.25

Founder draws

1 × nextFloat()

Catalogue entries

1 — it never changes what a horse looks like

The combination table

The hearts are the whole point

A pony is not a broken horse, it is a hardy one. The small end of the size range has to buy something or nobody would ever breed toward it — and a locus whose only effect is “worse” is a locus a player fixes once and never thinks about again. Four hearts against a sixth of a jump and a hundredth of a horse’s speed is a real decision.

Gene and locus

HMGA2 is a chromatin-associated architectural transcription factor with three AT-hook regions that bind the minor groove of AT-rich DNA and help regulate proliferation and differentiation programmes. It affects body size across species — humans, mice, dogs, horses.

The equine change sits in the first and highly conserved AT-hook. Glycine is neutral; glutamate is negatively charged, and DNA is negatively charged too, so the substitution is mechanistically plausible as a DNA-binding disruption. In an electrophoretic mobility-shift assay the mutant AT-hook peptide had greatly reduced DNA-binding affinity. That is unusually strong functional support for causality rather than mere linkage. Frischknecht et al., PMC4608717

Size, which is the well-evidenced part

The A allele is size-reducing; the reference G allele goes with taller stature. In the original 110 genotyped Shetlands, each A copy was associated with an average reduction of about 9.5 cm at the withers, and every Shetland under 87 cm in that cohort was homozygous for the size haplotype. When the genotype was accounted for, the ECA6 height signal disappeared — strong evidence that this variant is that QTL.

A later multi-breed Australian study found that after accounting for breed, genotype explained about 54.9% of remaining height variation across the sampled ponies, 44% of within-breed variation in Shetlands and 21% in Welsh ponies. In that cohort A/A ponies averaged 10.4 cm shorter than G/A and 12.6 cm shorter than G/G, while G/A and G/G did not differ significantly. PMC10229368

What “build” means here

The data are strongest for height, not for detailed conformation. The variant probably reduces overall developmental growth, so a smaller-bodied animal follows — but nothing shows that it independently produces a particular head or neck shape, shoulder or croup angle, limb-to-body ratio, bone circumference, muscle distribution, hoof form, or “pony type” versus refined type. Some build consequences follow from being smaller; the allele does not create an athletic conformation.

Height is polygenic. Major loci include NCAPG on ECA3 (the mod’s LCORL), ZFAT on ECA9, GH on ECA11 and HMGA2 on ECA6; in one cross-breed analysis four loci together explained a large share of size variation. So an HMGA2 genotype is informative, not a complete height prediction — which is also how the mod uses it, as one of two size loci pulling against each other rather than as the answer.

Inheritance

The Shetland study supported a largely additive / semi-dominant quantitative effect — roughly 9.5 cm per copy — and the Welsh cohort likewise favoured an additive model. The 2023 mixed-pony study found a pattern closer to recessive for height, with A/A clearly shorter and the other two indistinguishable. That probably reflects breed composition, allele frequencies, unequal genotype groups, modifiers and sample size rather than overturning the additive model.

The practical reading: a large-effect quantitative allele whose expression is breed- and background-dependent, not a Mendelian test that guarantees a number of centimetres. Two G/A parents give the usual 25% / 50% / 25% genotype probabilities, and A/A offspring are expected to average shorter than their siblings while still varying a great deal individually.

Health — and the one place the mod inverts the sign

The most important health finding is an association between the A allele and higher insulin in ponies. That matters because persistent hyperinsulinaemia and insulin dysregulation are major risk factors for endocrinopathic laminitis.

In 294 Welsh ponies, the A allele correlated strongly with shorter height (r = −0.75) and moderately with baseline insulin (r = 0.32), insulin after an oral sugar test (0.25), non-esterified fatty acids (0.19) and triglycerides (0.22). The ECA6 region accounted for roughly 40% of estimated genetic variation in height and roughly 20% in baseline insulin in that population. PMC6430908

In a six-breed Australian study of 236 ponies with an A-allele frequency of 62%, A/A and G/A ponies had basal insulin about 4.3 and 2.7 µIU/mL higher than G/G — genotype accounting for about 7.1% of the explained variation. Meaningful, and far from the whole story: farm and management, cresty-neck score, age, sex, breed, diet, body condition, exercise, season and many other genes all contribute.

An A allele is not a diagnosis of equine metabolic syndrome, insulin dysregulation or laminitis. It is a risk-associated factor found in pony studies, and it should never be used alone to call a pony metabolically unhealthy. The unresolved question is whether p.G28E itself changes metabolic physiology — true pleiotropy — or whether another linked, unidentified variant on the same selected ECA6 haplotype does. Researchers flagged this explicitly as open.

There is no good evidence that the allele directly causes skeletal dysplasia or disproportionate dwarfism, lethal developmental disease, a specific orthopaedic syndrome, immune deficiency, poor fertility, cancer predisposition, or laminitis independent of insulin dysregulation and management. HMGA2 has important cancer and growth biology in other species — abnormal overexpression is tied to tumour biology in humans — and that is not evidence that the naturally occurring pony allele causes equine cancer.

Coat, speed, jumping

Coat. No evidence that c.83G>A controls base colour, grey, any dilution, white spotting, roan, appaloosa patterning, mane and tail colour, feathering, or coat texture, length and shedding. Colour lives at MC1R, ASIP and the dilution and spotting loci; HMGA2 is not among the validated pigment genes. Where a colour looks associated with the allele in one population, breeders selected both small size and that colour in the same lines.

Speed. No published evidence of a direct effect on sprint speed, acceleration, maximal velocity, racing performance, gait mechanics, aerobic capacity, muscle-fibre type, heart and lung capacity or recovery. There is an indirect relationship, because size changes stride length, mass, leverage and what discipline suits — but smaller is not inherently faster or slower. A small efficient pony can be very quick over a short distance or agile on rough ground, while a larger horse has stride-length advantages at an open gallop. Biomechanics and training, not an HMGA2 performance effect. Do not confuse this with MSTN: HMGA2 is a body-size locus.

Jumping. No controlled study shows that genotype predicts free-jumping score, scope, carefulness, technique, bascule, competition results, tendon or ligament injury risk, or soundness under jumping load. Stature does influence the practical fit between a pony and a fence height, a rider, a stride distance and a division — a shorter A/A pony has a shorter stride and different distance needs, which is not the same as less jump ability.

The mod charges 0.008 speed and 0.02 jump per copy, described in the source as “shorter stride”. That is the indirect, biomechanical reading above rather than a claim about the gene, and it is the honest way round.

Breed distribution

Overwhelmingly a pony allele, with frequencies differing sharply by country, breed section and sampled population. Small samples are not breed prevalence.

The mod uses p(p) = 0.25 across all founders. Against a real distribution that runs from 0.5% in large breeds to 100% in Welsh Section A, that number is doing something different from any of them: it is setting how often a player meets a pony at all. Given that the mod has no breed structure at founder time, a single middling frequency is the only thing available — and it is the clearest example on the wiki of why per-breed allele weighting would change the whole feel of the genetics.

What is settled, and what is not

Strongly supported. c.83G>A / p.G28E on ECA6 is a functional, major-effect quantitative variant for reduced height in several pony populations; it disrupts an AT-hook DNA-binding domain and reduces DNA binding in vitro; the A allele is common in some small pony populations and rare in sampled full-sized horses; and it is associated with higher basal insulin and other metabolic measures.

Plausible, not settled. Whether the insulin association is caused directly by p.G28E or by a separate linked variant on the selected haplotype; whether inheritance is consistently additive across breeds — studies support additive, recessive-like and, in one insulin analysis, dominant-like patterns depending on cohort and phenotype; the exact effect size in any individual breed or line; and whether other HMGA2 variants explain small stature in further pony populations, Debao ponies included.

Not supported. p.G28E as a coat-colour gene, as a direct jumping gene, as a direct speed or race gene, as a stand-alone EMS or laminitis diagnostic, or as a deterministic adult-height forecast from one genotype.