When registrations rise but the breeding population falls
Research note · Registry measures
Abstract
An increase in new registrations does not necessarily mean that a breed’s existing breeding population has grown. We compared female registrations with the estimated breeding female population for 2021 and 2025 in the UK’s official breed inventory. Of 20 native horse breeds in the release, 17 have nonzero 2021 values and complete values for both measures in both years. The two measures changed in opposite directions for six of those 17 breeds. In three, each change exceeded 5% in opposite directions. This is a descriptive comparison of published records, not a claim that registrations caused the population change.
Two numbers, two questions
The inventory’s female registrations count females newly registered in a year. Its estimated breeding female population counts living females that have produced at least one foal within the previous ten years, using the method described by the UK Department for Environment, Food & Rural Affairs (Defra). One is an annual flow into the records; the other describes a larger, longer-lived group. The 2025 statistical release explains the definitions and notes that the equine population method was introduced in 2025.
A watch on new registrations can be useful without making it a population census. We wanted to measure how often these two readings would give opposite directional headlines for the same breed and period.
Data and method
We used Defra’s machine-readable 2025 inventory CSV, released with its statistical notice. We selected rows where species = Horses and the release marks the breed native = Y. For each breed, we read females and fempop for 2021 and 2025. The first field corresponds to female registrations; the second is the estimated breeding female population. We calculated each percentage change as (2025 value ÷ 2021 value − 1) × 100.
We excluded a breed from the directional comparison if either endpoint was missing or its 2021 denominator was zero. That leaves 17 of 20 native breeds. Dartmoor Hill Pony, Percheron, and Thoroughbred lack at least one required breeding-population endpoint in this release. A direction is “opposite” only when one percentage change is positive and the other negative; a zero change is neither. We also counted cases where both changes were more than 5% in magnitude in opposite directions, so a very small movement would not drive the stronger result.
Results
| Comparison, 2021 to 2025 | Native horse breeds |
|---|---|
| Listed in the release | 20 |
| Complete for both measures at both endpoints | 17 |
| Registration and breeding-population changes in opposite directions | 6 |
| Opposite directions, each greater than 5% in magnitude | 3 |
All three cases meeting the stronger threshold had more female registrations in 2025 but a smaller estimated breeding female population:
| Breed | Female registrations, 2021 → 2025 | Change | Estimated breeding females, 2021 → 2025 | Change |
|---|---|---|---|---|
| Cleveland Bay Horse | 8 → 11 | +37.5% | 120 → 110 | −8.3% |
| Shetland Pony | 872 → 940 | +7.8% | 6,906 → 6,053 | −12.4% |
| Shire Horse | 110 → 128 | +16.4% | 1,060 → 973 | −8.2% |
The full 20-breed comparison shows the other three opposite-direction cases and the excluded breeds. Defra’s own report presents the breeding-population changes; our added comparison asks whether the registration series points the same way. The Shetland Pony example illustrates the difference plainly: a 2025 registration increase over 2021 is true, and a decline in the estimated breeding female population over those same years is also true.
What this means for watching records
For a breeder or registry, a trend needs its unit, observation period, and source. A newly registered female, a living female that has foaled, and an active sire are different records of a breeding program. They should not be silently combined into one “breed growth” indicator. A useful watch brings a change to attention, then lets a reader inspect which measure moved and which records support it.
This is relevant to how AnimalTrace presents search and update results. Search can define a group of records to follow; a saved watch can bring additions or changes back for review. Interpreting those changes still requires the right denominator. We would describe a rise in registered animals as a registration trend, and reserve a breeding-population statement for data that actually estimates that population. The distinction also matters when comparing lines: more recorded descendants may reflect more complete reporting as well as more breeding activity.
Limits and reproducibility
These are UK native horse breeds in one official release, not a sample of all horse breeds or registries. The percentage for a breed with a small starting count can be large after only a few additional registrations: Cleveland Bay rose from eight to eleven. Endpoint comparisons do not describe every intervening year or explain why either measure changed. The breeding-female figure is an estimate under Defra’s equine definition; it is not a genomic diversity measure or a count of all living females. The release notes a 2025 method change, so historical comparisons should be read in light of its methodology. We report counts rather than a significance test because this is an audit of the listed breeds, not a random sample.
- Source and definitions: Defra’s 2025 FAnGR statistical release.
- Source data: Machine-readable inventory CSV, SHA-256
08a3e6aac3e7a86d9abc25fd7cd810d3be64f53c0820c78f7173b9e3968fda1e. - Derived data and method: 20-breed comparison and standard-library Python script. Download the source CSV and pass its path and an output path to the script to repeat the analysis.
Source information is attributed to Defra and used under the Open Government Licence v3.0. This AnimalTrace research note has not undergone external peer review.