The data for this study was obtained from the Synergistic Plant and Animal (SYNBREED) project (www.synbreed.tum.de). The research included a total of 217 individuals from eight different chicken breeds. Among these, there were 101 individuals from four large chicken breeds (BHxx, Cosch, LSxx, Orge) and 116 individuals from four bantam chicken breeds (Abwa, FZxx, GBxx, SBxx), detailed information can be found in (Table 1).

The genotyping of DNA samples was conducted using the Affymetrix Axiom TM 600K Whole Genome Chicken Genotyping Array, which comprises over 580,000 single nucleotide polymorphism (SNP)s [11]. The genotyping array was annotated using the Gallus_gallus-5.0 reference genome [12], resulting in annotations for a total of 579,621 SNPs.

Firstly, 27,416 SNPs that were duplicates, ambiguous annotations, or were located on the sex chromosomes were excluded from the SNP array. Subsequently, the remaining 552,205 SNPs underwent additional quality control using PLINK 1.9 [13]. The quality control criteria included a genotyping missing rate<0.01 and a minor allele frequency>0.05, a total of 425,164 SNPs were retained. For the subsequent analysis of population structure, an additional linkage disequilibrium filtering is necessary. The filtering parameters, including window size, step size, and r2, are set as “50, 5, 0.2”. Finally, 45,294 SNPs were retained for the population structure analysis.

The PCA results show the contribution rates of the first two principal components: PC1 (20.10%) and PC2 (13.93%). PC1 effectively separates the large chickens from the bantam chickens, the large chickens cluster together with closer genetic distances within the group, the bantam chickens show relatively more dispersed clustering and greater genetic distances within the group. It is worth noting that ABwa and GBxx cluster together, this observation aligns well with the findings from the phylogenetic tree analysis (Figure 1). In both the phylogenetic tree and ancestry component analysis, the FZxx, SBxx and BHxx populations are observed to have two subpopulations. However, in the PCA, only the FZxx population exhibits clear separation into two subpopulations, while the SBxx and BHxx populations do not show separation.

Through Admixture analysis (Figure 2), the optimal fit curve for population clustering was calculated, revealing that the lowest error was observed at K = 10. At K = 2, the SBxx population was first separated. At K = 3, the BHxx and ORge populations was separated. At K = 4, the COsch and ABwa populations were separated. At K = 6, the LSxx population was separated. At K = 7, the SBxx population differentiated into two subpopulations. At K = 8, the FZxx population also divided into two subpopulations. At K = 9, the BHxx population further subdivided into two subpopulations. It is noteworthy that at K = 10, At K = 10, GBxx remains undifferentiated and exhibits a mixed ancestry from several breeds, even at K = 15, GBxx still does not show distinct differentiation. Understanding the relationship of population structure is highly significant for determining population differentiation, carrying out genetic improvement of varieties and selective breeding.

The study examined the distribution of ROH islands in populations of large and bantam chickens and visualized them based on SNP positions on chromosomes (Figure 3). In large chickens, 48 ROH islands were identified; in bantam chickens, 56 ROH islands were identified. We focused on the selected ROH islands for analysis and identified 2 genes related to skeletal development and weight regulation in large chickens, and 6 genes in bantam chickens (Table 2). Interestingly, we observed overlapping regions between the identified ROH island regions on chromosome 2 and chromosome 8 in large chickens, which coincided with the detected selection signatures and encompassed the INHBA and LEPR. Similarly, there was an overlapping region on chromosome 14 in bantam chickens, annotated to the NOG2. Analysis of ROH islands can help us understand the selected regions and genes of chickens with different body sizes under long-term natural and artificial selection.

To identify candidate genomic regions under selection during the growth and development processes in large and bantam chicken populations, this study utilized the Fst and π radio methods to detect selection signatures. The top 5% threshold was employed as the selection criterion. If a region was significantly detected by both methods, it was considered a genuine selected region. (Figure 4) illustrates the Manhattan plots for different chicken body sizes, generated using the Fst and π methods, with the top 5% threshold serving as the selection criterion. (Figure 5) depicts the selected regions in different chicken body sizes using the Fst-π radio method. In large chicken population, 701 selected regions were identified, while in bantam chicken population, 1115 selected regions were detected.

The region of chromosome 2 (81099018–81269619) in bantam chickens, specifically within the GRB10, exhibits a strong selection signature. Haplotype analysis was conducted on this region, and the results are depicted (Figure 6). In this region, there are differences in genotypes at various loci between large and bantam chickens.

To understand the gene and functional differences in the selected regions between large and bantam chickens, this study annotated the 701 selected regions detected in the large chicken population, annotations for 322 genes. Similarly, the 1115 selected regions detected in the bantam chicken population were annotated, with annotations for 447 genes. The annotated genes were subjected to GO functional enrichment analysis and KEGG pathway analysis to gain insights into their functional features. The results of the enrichment analysis are presented in (Figure 7). In the large chicken population, a total of 16 enriched GO terms were identified. These terms are primarily associated with embryonic skeletal system morphogenesis, negative regulation of cell growth, calcium ion binding, and positive regulation of cell proliferation. In the bantam chicken population, a total of 20 enriched GO terms and 2 enriched KEGG pathways were obtained. The annotated GO terms are mainly related to insulin receptor binding, positive regulation of mitotic nuclear division, positive regulation of activated T cell proliferation, and positive regulation of the mitogen-activated protein kinase (MAPK) pathway. The annotated KEGG pathways are associated with autophagy and progesterone-mediated oocyte maturation. By selection signatures analysis and enrichment analysis, genes related to chicken body size and growth and development were identified, which laid a foundation for the subsequent development of molecular markers.

The results of population structure analysis show that large chickens cluster together prominently and exhibit close genetic distance, which is primarily associated with their origins and population history [22]. Bantam chickens exhibit relatively more dispersion, however, GBxx and ABwa cluster together and occupy the same branch on the evolutionary tree, which is consistent with a findings reported [23]. In the analysis of evolutionary tree and ancestral composition, SBxx, FZxx, and BHxx are all divided into two subpopulations. This is in line with the fact that each of SBxx (golden feathers and silver feathers), FZxx (white feathers and black feathers), and BHxx (light-colored feathers and dark-colored feathers) has two feather color phenotypes [24]. Population separation happens due to the fact that different populations possess distinct genotypes at the same locus. These different genotypes symbolize different genetic information and have diverse gene functions and roles, thereby giving rise to differences at both the phenotypic and genetic levels. It is worth noting that in the admixture analysis, The GBxx bloodline is relatively complex and has been involved in genetic flow events with other breeds, which is consistent with its use and history [23].

In this study, we focused on specific genomic regions associated with chicken skeletal development and weight regulation, identifying a total of 8 relevant genes. Among them, the INHBA, LEPR, and NOG2 overlapped with the genes identified by selection signatures. INHBA, located on chromosome 2 in large chickens, is potentially involved in the hedgehog signaling pathway, which plays a crucial role in the formation of growth plates in long bones [25]. LEPR, was discovered on chromosome 8 in large chickens and is important for weight regulation. In bantam chickens, we found the HOXD family (HOXD3, HOXD4, HOXD8, HOXD9) on chromosome 7, which has a significant impact on vertebrate axial skeletal development [26]. IHH is also located on chromosome 7 in bantam chickens and collaborates with parathyroid hormone-related protein to regulate the development of growth plates and long bones [27]. Additionally, during early cartilage development, IHH is a major source of the hedgehog signaling pathway, exhibiting proliferative expression in limb buds and eventually differentiating into skeletal chondrocytes. Mutations in the IHH can lead to defective cartilage development, affecting the proliferation of cartilaginous tissue and bone formation [25]. NOG2, located on chromosome 14 in bantam chickens, is a bone morphogenetic protein (BMP) antagonist. Currently, its function and mechanism of action in chickens are not well understood. We suppose that it may interact with BMPs in chicken skeletal development, regulating the proliferation and differentiation of bone cells, thereby influencing normal bone development in chickens.

The genomic regions under selection were determined by selecting the top 5% Fst and π ratio and taking their intersection. This approach combines Fst and π ratio, provides more robust and reliable results [28]. The candidate genes identified were further investigated using GO and KEGG enrichment analysis. Notably, several genes associated with chicken body size and growth were discovered in the analysis conducted on large chickens. HOXA 10 is a member of the homeobox gene family, with a high degree of conservation. While the current understanding of HOXA10 function primarily focuses on embryonic development, there is research indicating its vital role in adult skeletal formation and bone healing [29]. In large chickens, this gene has been positively selected, suggesting it may impact on skeletal formation and development influence chicken body size. A study revealed that the allele B of the POU1F1 gene is likely to positively affect the growth traits on Nanyang cattle [30]. Similarly, POU1F1 in native Korean chicken breeds showed that POU1F1 polymorphism significantly influenced the growth and growth curve characteristics of chickens [31]. This shows that the POU1F1 gene has a great relationship with the weight and growth and development of the animals. Overexpression of BMP10 in mice has result in excessive myocardial growth and the formation of hyper trabeculation during embryonic development [32]. Transgenic mice with myocardial overexpression of BMP10 show a significant reduction of 50% in heart size, as well as decreased body weight and size at one month of age; in a manner similar to mice, overexpression of BMP10 in zebrafish leads to shortened body length, this implies that BMP10 plays a crucial role in determining body size in vertebrates [33]. There are reports suggesting that BMP10 also plays a role in inducing cell apoptosis, proliferation and growth [34].

Insulin-like growth factor I (IGF1) is a selected gene in bantam chickens and is known to affect body size in mice and humans [35]. Additionally, previous report has demonstrated that IGF1 allele is a major determining factor for smaller body size in dogs [2], which is consistent with the findings of this study. Other studies have indicated that IGF1 promotes longitudinal bone growth through its “insulin-like” metabolic effects, leading to the enlargement of chondrocytes [36], thereby influencing growth, development, and changes in body size. The synthesis and metabolic functions of growth hormone are primarily mediated by IGF1, suggesting that IGF1 is the main determinant of cellular growth [37]. Rare mutations in the human IGF1 result in severe growth inhibition and intellectual disability [38]. Mice with a knockout of the IGF1 were born with only 60% of normal birth weight, and a few of the surviving adult mice even exhibit a body size less than one-third of that of normal mice [39], this indicates that IGF1 in animal body size development. Insulin-like growth factor II (IGF2) is a crucial growth factor in embryonic and placental development, and its deficiency leads to severe growth retardation. IGF2 primarily functions through the IGF1 receptor in embryos and an unidentified receptor in the placenta [40]. Additionally, IGF2 is indispensable for bone development as it stimulates the proliferation and differentiation of bone cells. The down regulation of IGF2 is likely the underlying cause of the reduced bone mass observed during cortisol treatment [41]. The growth of developing mouse embryos is predominantly controlled by IGF2 [42]. Moreover, IGF2 was the first identified imprinted gene, and when targeted IGF2 deletion is paternally transmitted, mouse embryos inherit only the inactive maternal allele, resulting in impaired development at birth [43]. Conversely, overexpression of IGF2, achieved through disruption of inhibitory IGF2R [44]or transactivation of IGF2 [45], the fetus will display excessive growth. This suggests that IGF2 can directly or indirectly influence changes in body size by affecting both bone development and embryonic growth. GRB10 encodes a growth factor receptor-binding protein called Growth Factor Receptor-Binding Protein 10. This protein plays a crucial role in interacting with insulin receptors and insulin-like growth factor receptors, such as IGF1R and IGF2R. Overexpression of certain isoforms of GRB10 can lead to the inhibition of tyrosine kinase activity, ultimately resulting in growth suppression. GRB10 is also classified as an imprinted gene, in mice, the inactivation of the paternal allele of GRB10 results in mice with enhanced aggression. When the maternal allele is inactive, the mice exhibit fetal overgrowth and are noticeably larger than their wild-type littermates, this demonstrates the gene’s role as a growth suppressor [46,47]. The gene may potentially influence changes in body size by affecting it through the inhibition of growth.

Finally, we have identified several genes associated with chicken body size as well as growth and development (Table 3). This offers a foundation for the functional analysis of relevant genes, the development of molecular markers, and the breeding of new varieties. This study also has some limitations. This analysis is carried out using natural populations, and it is impossible to ensure that the inherent genetic characteristics of different varieties will not interfere. However, when selecting and grouping varieties, we have taken this into account to ensure that it will not affect the analysis results. In general, this study will serve as a reference for analyzing the genetic mechanism of chicken body size. In future research, relevant causal mutation sites will be further identified to develop molecular markers for chicken body size and drive the development and progress of molecular breeding.