Kazakhstan’s sheep portfolio underpins much of the country’s mutton and wool production, yet several of its principal breeds remain genomically uncharacterised. The aim of this study was to characterise the genomic diversity, population structure, and global phylogenetic placement of three economically important Kazakh breeds and to determine whether they constitute separate gene pools requiring independent management. We present the first genome-wide SNP characterisation to include the Degeres (DE), the Etti Merino (EM), and the Kazakh fat-tailed coarse-wool (KKG) breeds simultaneously. A total of 1497 animals (DE = 354, EM = 642, KKG = 501) sampled across seven production households were genotyped and, after quality control, analysed at 42,279 SNPs, of which 22,766 LD-pruned markers were used for principal component analysis and AMOVA. We applied principal component analysis (PCA), pairwise F_ST, analysis of molecular variance (AMOVA), neighbour-joining phylogenetics, model-based ancestry estimation (ADMIXTURE), and Hill-number diversity profiling, and projected the breeds against the global Ovine SNP50 HapMap panel (74 reference breeds, 2819 animals; 37,685 shared SNPs). All three breeds retained uniformly high within-breed diversity (expected heterozygosity 0.413–0.417) with fixation indices at or near zero. AMOVA partitioned 94.03% of variance within breeds (Φ_ST = 0.060, p < 0.001). PCA, phylogeny, and ADMIXTURE concordantly resolved three breed-specific clusters at K = 3, with a maximum interbreed F_ST of 0.038 within the study dataset. Against the global panel, EM was genetically closest to Merino and Merino-derived reference breeds (pooled F_ST = 0.017) and substantially more distant from Southwest Asian sheep (F_ST = 0.045), whereas DE and KKG showed the reciprocal pattern (F_ST = 0.027 and 0.020 to Southwest Asia, 0.052 to the Merino group). DE additionally displayed the heterozygote excess and partial admixture expected of an incompletely consolidated composite. These results delineate three distinct gene pools and carry direct implications for breed management and the conservation of genomic diversity in Kazakhstani sheep.

Background/Objectives: The Kazakh horse comprises several locally adapted types and ecotypes of the Eurasian Steppe. The Adai is its western ecotype, kept year-round on desert pasture in Mangystau, Western Kazakhstan, and recognised as a separate breed in 2024 without genetic evaluation. We asked whether that label corresponds to a genetic unit and whether the population has remained closed. Methods: We genotyped 533 horses recorded as Adai, 500 traced to five farms in Mangystau, against 988 reference horses of 45 breeds on 36,193 shared autosomal SNPs. Structure was assessed by principal component analysis, pairwise F_ST with block-jackknife intervals, neighbour-joining trees and cross-validated individual assignment; introgression per animal by an f₄ statistic against an empirical null of 223 reference-steppe horses, with the ancestry scale calibrated on 1800 simulated crosses built from real genotypes; and eight body measurements were modelled within farms. Results: The Adai was not separable from the Central Asian steppe complex as a population: pairwise F_ST was 0.004–0.008 against Tuva, Yili, Sanhe and Mongolian horses, and two Adai farms differed from each other as much as an Adai farm differed from a reference-steppe breed. Individual assignment was nonetheless accurate against those breeds (86–100% pairwise), but the same classifier separated two herds of Adai at 69–85%, so assignment accuracy on thousands of markers tracks marker number rather than breed identity. Against the empirical null, 15.6% of Adai exceeded the 95th percentile and twelve exceeded the maximum. The estimator carries a floor of 3.7 percentage points on an unadmixed Adai; on the calibrated scale, 9.6% of animals reached the ancestry expected from one Thoroughbred great-grandparent and 4.2% that expected from one grandparent (15.8% and 4.6% uncalibrated). Introgression was concentrated: one farm carried 42% of its horses at a great-grandparent equivalent or more against 0–9% at the other four, a difference that survived every analysis accounting for family structure. Ancestry showed no within-farm association with any of eight body measurements, with or without age in the model. Conclusions: The Adai is not a differentiated population but part of a continuous steppe complex, and carries undocumented ancestry of Thoroughbred affinity concentrated in particular holdings. As managed, it cannot be certified from a document or a phenotype. Genomic verification at registration is a plausible instrument, but the marker panel it would require was not developed or validated here.