Genetic marker analysis of RAPGEF6 gene and covariance structure modelling for egg

Abstract

Background

Egg production sustainability is economically critical for layer farms. The Marandi chicken, an indigenous breed from northwestern Iran (Azerbaijan region), possesses significant potential for efficient egg production, yet its genetic potential remains underexplored. This study aimed to investigate the association between single nucleotide polymorphisms (SNPs) in the RAPGEF6 gene and productive and reproductive traits in the Marandi chicken population.

Methods

Performance and reproductive data were collected from 153 Marandi hens. Genotyping was performed using PCR–RFLP. Data were analyzed using two complementary approaches in SAS: a repeated-measure model (with eight covariance structures) for repeated egg weight measurements, and Generalized Linear Models (GLM) for the remaining growth and performance traits. Genotype frequencies for the studied SNPs were determined. Post-hoc statistical power analysis was conducted using GLMPOWER in SAS to evaluate the ability of the current sample size to detect observed effect sizes.

Results

A significant association between one SNP and Maturity Weight (P = 0.014) was found in the present work, while no significant associations were found for the other 16 traits (P > 0.05). However, post-hoc power analysis revealed that the study had only 5% power to detect this effect, indicating that this finding is statistically unstable. The ARMA (1,1) model demonstrated good performance in accounting for the repeated egg weight measurements.

Conclusions

Polymorphisms in the RAPGEF6 gene may influence growth traits in Marandi chickens, offering insights into the genetic underpinnings of performance. However, given the single-SNP design, the small sample size, and the severely limited statistical power (5%), all findings should be interpreted as exploratory and preliminary, warranting validation in independent, well-powered populations. No recommendations for marker-assisted selection can be made based on the current data. The optimal ARMA (1,1) covariance structure enhanced repeated-record analysis precision for egg-laying performance, facilitating a more consistent and reliable assessment of egg production patterns.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12864-026-13373-8.

Introduction

Indigenous chickens serve as the primary foundation of the genetic diversity in the poultry industry. With increasing human population, the demand for egg production, which serves as an affordable source of protein, has become strategically important for food security. Egg production traits are controlled and regulated by several genes and environmental factors, necessitating extensive genomic studies in poultry to identify candidate genes. Identifying single nucleotide polymorphisms (SNPs) and genes associated with egg weight and quantity during a laying cycle is essential. Genes and their receptors within the hypothalamic-pituitary–gonadal (HPG) axis, linked to reproductive hormones, play a crucial role in the reproduction and egg-laying rates of birds [1].

Merabet et al. [2] noted that polymorphisms in the PRLHR, GHSR, and SALL3 genes positively influenced egg production in White Leghorn hens. Research by Du et al. [3] on IPB-D1 chicken breed indicated that a mutation in the prolactin gene (PAL) at exon 5, specifically at locus T8052C, serves as a significant marker for egg production. Studies by Azmal et al. [4] highlighted that polymorphism in the IGF-IR gene in Chinese Yellow and White Peking quails significantly affected egg production and quality [5]. Additionally, Yang et al. [6] found a correlation between the FAM184B gene on chromosome 4 in Yellow Chinchilla chickens and egg weight, indicating that the weight at the onset of laying directly impacts peak egg production [6]. The RAPGEF6 gene is a protein-coding gene located on chromosome 13 in birds, which plays a crucial role in guanine exchange and GTP-dependent protein binding. Polymorphisms of the RAPGEF6 gene at loci are considered excellent SNPs for improving egg production in native Chinese Jinghong chickens [7]. This gene spans chromosome 13:16,092,830–16195620 encompassing five transcripts with 25 to 30 exons, and encoding proteins with 1510 to 1567 amino acids.

The Marandi chicken is a native and ornamental layer breed in Azerbaijan (north-west) of Iran characterized by excellent egg-laying performance and a layer-type body. The Marandi chicken body and legs are covered with black feathers that shine dark green under sunlight. The Marandi chicken produces her first egg at 21 to 23 weeks of age and demonstrates excellent egg-laying capabilities. Marandi is a well-adapted breed to mountainous and cold climates and is disease-resistant, and is also raised in other temperate and tropical regions of Iran.

Repeated measures studies in poultry and livestock record-keeping have opened new avenues in this field of research. One of the challenges in mixed models with repeated measures is obtaining an appropriate covariance structure. The term repeated measures in the analysis of mixed models refers to controlling the implemented covariance structure based on residuals or errors [8–10]. In General Liner Model (GLM) and repeated measures methods, errors are considered independent, while in the Proc Mixed procedure in SAS, these errors exhibit variable variances. The remarkable difference between the Mixed model and GLM appears in the estimated variances derived from the least square means differences [11]. These models incorporate both fixed and random effects [12], where random effects represent the variance and covariance of observed records, while fixed effects reveal the expected outcomes of observed data [13]. This covariance and variance structure relate to the components of the observed data. Measurable parameters such as weight, blood parameters, metabolic products, nutrient absorption capabilities, therapeutic drugs, and hormones can be included.

Improving egg production in indigenous chickens is of significant economic importance to breeders, as it directly impacts the egg production cycle. The weight and number of eggs produced over a specific period are traits influenced by genetic and environmental factors, and selecting certain SNPs that affect these economically important traits can lead to substantial improvements.

To date, no studies have investigated egg production monitoring and egg weight analysis of the RAPGEF6 gene at locus 16,120,088 in Marandi hens, nor has any study reported on identifying covariance structures using mixed models with repeated egg-laying records in poultry. This research could contribute to a better understanding of the genetic impact on egg weight and production quantity in this native breed.

Hypothesis and Objectives.

It is hypothesized that polymorphisms in the RAPGEF6 gene are associated with egg production traits in Marandi chickens, and that a repeated measure model utilizing an Auto Regressive Moving Average (ARMA (1,1)) covariance structure will optimally describe the temporal dependencies in repeated egg weight data. The objectives of this study were to Identify RAPGEF6 gene polymorphisms at locus 16,120,088 in Marandi chickens, to determine associations between RAPGEF6 genotypes and production traits (maturity weight, egg weight, and egg number) and to develop and evaluate an optimal covariance structure for analyzing repeated egg weight measurements using a repeated measure model.

Materials and methods

Fertile egg collection and rearing practice

A total of 645 fertilized eggs was collected during 2023 from six cities in northwestern Iran: Azarshahr, Urmia, Osku, Karaj, Gogan, and Marand. The incubation of the eggs took place at the Savaran hatchery, located 35 km west of Tabriz, resulting in the hatching of 405 one-day-old chicks, which were then transferred to the Khalat Pushan Research Center at the University of Tabriz. Figure 1 addressed the map of the six fertile egg collection locations within Iran during the present experiment.

Fig. 1.

Map of the six fertile egg collection locations within Iran during the present study

At 9 weeks of age, the hens were separated from the roosters, resulting in 176 selected Marandi hens. The selection process was made according to the Marandi breed specifications (all-black body feathers, feathered legs, simple comb, black and gray skin color, black and yellow beak color, white ear color, black and white nail color, white eye rings, tail feathers, crest, and no beard). Unselected birds were removed from the study, and the selected Marandi chickens were raised on the floor until 15 weeks of age. At 16 weeks of age, 153 Marandi hens were transferred into individual cages arranged in three tiers across two separate rearing halls with nearly identical rearing conditions. Lighting and vaccination programs were conducted according to commercial layer protocols, adhering to all animal welfare regulations. From the onset of egg production (i.e., 5% of the flock that took place at 21 weeks of age), a restricted daily diet of 100 g per hen was provided, as was outlined in Table 1.

Table 1.

Ingredients and chemical composition of the diet fed to Marandi chickens at age 5 months

Calculated composition	%	Ingredient	%
Dry Matter	88.66	Corn, grain	51.00
Metabolizable. Energy(kcal/g)	2.85	Soybean meal (48%CP)	27.50
Crude Protein	16.49	Barley (> 10%CP)	8.00
Calcium	3.50	Dicalcium phosphate	1.79
Available Phosphorus	0.48	Soy Oil	2.00
Na	0.17	Limestone	4.23
Cl	0.25	Oyster shells	4.00
Choline	1.57	Common Salt	0.35
Folate	0.98	Vitamin Premix*	0.25
Lys	0.89	Mineral Premix*	0.25
Met	0.60	DL-Methionine	0.38
Saa	0.90	L-Lysine HCL	0.01
Hr	0.85	L-Threonine	0.24
Rp	0.21
Arg	1.06

Vitamin Premix* and Mineral Premix* per kg of feed: vitamin A 11,000 IU; vitamin D3 3,500 IU; vitamin E 100 IU; vitamin K3 5 mg; vitamin B12 0.03 mg; riboflavin, 12 mg; nicotinic acid 55 mg; pantothenic acid 15 mg; folic acid 2 mg; pyridoxine 6 mg; thiamine 3 mg; d-biotin 0.3 mg; choline 300 mg; Cu 10 mg; I 2 mg; Fe 50 mg; Mn120 mg; Zn 100 mg; Se 0.3 mg

Performance traits

Daily egg production was recorded individually from each hen, with data recorded from the start of week 21 to the end of week 39.

The performance traits of the birds assessed included the “total eggs count” (TEC) produced during the recording period, “egg weight” (EW), “average eggs weight” (AEW), “cumulative weight of the first 50 eggs” (50EW), “average weight of the first 50 eggs” (A50E), “egg laying rate” (ELR), “egg mass” (EM), “feed conversion ratio” (FCR), and “egg weight loss” (EWL).

The following equations were used to calculate the egg laying rate (ELR) and egg mass (EM) of hens;

To calculate the FCR, the consumed feed (Kg) of each hen was measured and divided by the weight of eggs (Kg) produced by that hen. To assess the EWL during shelf storage, six eggs per hen were collected over an eight-day period, starting at week 34 of age, and weighed twice. First measurement at 2:00 PM on laying day, and the 2nd is 48 h later, using a digital scale with an accuracy of 0.01 g. The eggs were stored in sealed plastic containers that allowed airflow in a room with a temperature fluctuating between 15 and 20 °C and a relative humidity of 30 percent. The EWL was analyzed as the average loss of egg weight of the six recorded eggs.

Growth related traits

Growth traits such as “hatch weight” (HW), “maturity weight” (MW), “estimated hatch weight” (EHW), “age at inflection point” (AIP), “weight at inflection point” (WIP), “ratio of maturity weight to weight at inflection point” (MW/WIP), “ratio of weight at inflection point to age at inflection point” (WIP/AIP) and hen weight at first laying” (WFL).

The mentioned growth-related traits were retrieved fitting the Richards function on the individually recorded chicken weights across the rearing period as follow;

In which: W_t = weight of the chicken; W₀ = hatch weight; W_f = maturity weight; b = Maturity index or coefficient related to growth (per days); t = day and n = shape parameter (n = -1).

Reproductive traits

The reproductive traits that were studied in the present work included the “age at first laying” (AFL).

DNA extraction from hen blood

At 36 week of age, blood samples were collected from the jugular vein of the hens using sterile disposable syringes, containing 0.2 CC of the anticoagulant EDTA, with a total of 1.5 cc of blood collected. The samples were then transferred to specific microtubes for each hen and stored in a laboratory freezer at the Molecular Genetics Laboratory of the Department of Animal Science, University of Tabriz, Tabriz, Iran, at -25 °C until DNA extraction. DNA extraction was performed using a modified salt method [14]. The quality of the extracted DNA was assessed by electrophoresis on a 0.8% (w/v) agarose gel, and for quantification, 15 samples were measured using a Nano Drop (Bio Tek) Citation device at Tabriz Medical School. In our experiments, we did not have any cases of sacrifice or anesthesia.

PCR amplification

Forward and reverse primers for the 16,120,088 locus of the RAPGEF6 gene were designed according to the template [8]:

• F: 5`-GCACGAATATGGTTTTTCAG- 3` and R: 5`-AGGGCTCCTTGGCTTCCATT-3`

These primers were synthesized by BGI Company, Beijing, China. Each PCR sample consisted of 15 µl, including 5.6 µl of ddH₂O, 7.5 µl of master mix, 0.45 µl of each primer (F and R), and 1 µl of template DNA from each bird. The PCR was conducted in a Bio-Rad T100 thermocycler with the following temperature program: incubation at 95 °C for 5 min, followed by 35 cycles of denaturation at 95 °C for 30 s, annealing at 54.6 °C for 30 s, extension at 70 °C for 30 s, final extension at 72 °C for 7 min, and storage at 25 °C for 30 s. The PCR products were then run on a 1.5% agarose gel for analysis and stained with ethidium bromide.

Genotyping by PCR–RFLP

For genotyping the 261 bp fragment of the RAPGEF6 gene at locus 16,120,088, the restriction enzyme MnII from Thermo Scientific was used. For each sample, 10.03 µl of ddH2O, 4 µl of Tango buffer, 1 µl of MnII enzyme, and 5 µl of PCR sample were mixed in a microtube and incubated at 37 °C for 20—24 h for enzyme digestion. The resulting fragments were observed by electrophoresis on a 3% agarose gel, and after staining with ethidium bromide, images were captured using a gel documentation system for genotyping identification at the specified locus.

Sequencing of PCR products

Six PCR samples of the RAPGEF6 gene were sent to BGI Company, China, for Sanger sequencing to validate the genotypic patterns of the agarose gel.

Genotype–phenotype association

The association between the genotypes of the RAPGEF6 (Rap Guanine Nucleotide Exchange Factor 6) gene at 13:16,120,088–16120089 locus and the repeatedly measured egg weights recorded over 17 weeks was conducted using a repeated-measure model with genotype included in the model as fix effect. The analysis was carried out using the Mixed procedure of SAS statistical software (version 9.4, 2017) as follow:

Model 1:

In which, represents the repeatedly measured egg weight records, μ is the average eggs weight, (l =1–2) is the effect of the hall) the term “hall” refers to rooms divided into two separate units located 500 m apart, each unit consisting of three rooms measuring 8 × 6 m. Each room is hygienic and equipped with all semi-modern poultry farming equipment(, is the effect of the city (m=1–6), is the effect of the RAPGEF6 genotype (k =1–3), is the effect of hen age at first laying, is the effect of the hen weight at first laying, is the random effect associated with the i th hen that assigned to k th genotype, l th hall and m th city, and is the random error.

All statistical analyses were performed using SAS (version 9.4). To evaluate the genotype–phenotype associations, two complementary approaches were applied: a repeated-measure model (MIXED) for repeated egg weight records, and Generalized Linear Models (GLM) for the remaining growth and performance traits, as detailed in Models 2 to 6 below.

To evaluate the association between the genotypes of the RAPGEP6 gene and other studied traits, calculations were conducted using the GLM method and various statistical models as follow;

Model 2: = μ + + + +

Model 3: = μ + + + + +

Model 4: = μ + + + + +

Model 5: = μ + + + + +

Model 6: = μ + + + +

All growth-related traits and egg performance traits including TEC, A50E, ELR, EM, and FCR were analyzed using Model 2. WFL and AFL were analyze using Models 3 and 4, respectively, A50E and 50EW traits were analyzed employing Model 5, and AEW was analyzed using Model 6, respectively. In the models y represents the phenotypic value, is the effect of cage nested within the hall and the remaining factors were as above.

To calculate significant differences and perform multiple comparisons, Tukey's test was used with a significance threshold of p-value < 0.05. The Hardy–Weinberg equilibrium test and calculation of allele and genotype frequencies were conducted using POPGENE v1.32 software.

Statistical power analysis

Post-hoc statistical power analysis was conducted using the GLMPOWER procedure in SAS software (SAS Institute Inc., Cary, NC, USA) to evaluate the ability of the current sample size (n = 153) to detect the observed effect sizes for each trait. The analysis was performed at a two-sided alpha level of 0.05, based on the observed standard deviations derived from the data and the group sample sizes. The minimum detectable effect size (MDES) at 80% power was also estimated for comparisons between genotype groups.

Covariance structure matrix

Mixed models with repeated measures in statistics offer high flexibility for analyzing unbalanced data [15]. Estimation methods for variance and covariance components in linear mixed models include Restricted Maximum Likelihood (REML), Maximum Likelihood (ML), and Minimum Variance Quadratic Unbiased Estimation (MIVQUE) [16].

Various (co)variance structures were tested including Ante dependence (ANTE (1)), Auto regressive (1) (AR (1)), Heterogeneous AR (1) (ARH (1)), ARMA (1,1), Compound symmetry (CS), Heterogeneous CS (CSH), Non-diagonal factor analytic (FA1(2)), Toeplitz (TOEP), Banded Toeplitz (TOEP (1)), Heterogeneous TOEP (2) (TPEPH (2)), Heterogeneous TOEP (TPEPH), General linear (LIN (1)), Unstructured (UN), and Variance components (VC).

To determine the best-fitting covariance structure four criteria were employed including - 2 log likelihood (- 2 Log L), Akaike’s Information Criterion (AIC), corrected AIC (AICc), and Bayesian Information Criterion (BIC). Finally, the most appropriate model with the highest goodness of fit criteria (i.e., lowest value of AIC, BIC, and AIC_C) were selected for the association analysis.

Ownership of the birds

We would like to highlight here that our flock belongs to the main nucleus, which is being raised in Tabriz University and the animal science department research center for study and research purposes.

Results

Descriptive statistics for all measured traits are presented Table 2. Marandi hens produce eggs with an average weight of 44.94 g (range: 29.3–57.25 g). According to the egg size classification in Australia, New Zealand, and South Africa, the Marandi chicken takes place in the medium category.

Table 2.

Descriptive statistics for the studied traits on 153 hens

Traits	Mean	Minimum	Maximum	Standard Deviation
Performance traits
Total egg count (TEC)	66.86	2	108	20.34
Egg weight (EW; gr)	44.68	14.67	75.32	50.40
Average egg weight (AEW; gr)	44.94	29.3	57.25	3.89
Feed conversion ratio (FCR)	2.91	1.54	9.34	1.00
Eggs weight loss (EWL; gr)	0.51	0.03	1.66	0.217
First 50 Eggs weight (50EW; gr)	2039.26	70.67	2605.13	457.46
Average weight of first 50 eggs (A50E; gr)	43.86	29.3	57.28	4.83
Egg laying rate (ELR; %)	59.59	1.78	96.42	18.09
Egg mass (EM)	1.34	1.04	2.03	0.122
Growth-related traits
Hatch weight (HW; gr)	31.96	21.01	41.02	3.83
Maturity weight (MW; gr)	1747.18	1026.02	2921.60	309.45
Estimated hatch weight (EHW; gr)	27.25	16.12	49.35	7.86
Age at inflection point (AIP; day)	74.66	55.29	107.58	9.35
Weight at inflection point (WIP; gr)	675.84	444.02	996.23	102.19
Ratio of maturity weight to weight at inflection point (MW/WIP)	2.72	1.05	4.50	0.52
Hen weight at first laying (WFL; gr)	1590.10	1030.02	2168.01	234.86
Reproductive traits
Age at first laying (AFL; day)	175.23	147.00	223.00	16.11

Allelic and genotypic frequencies

Genotyping via PCR–RFLP was performed to identify polymorphisms in the RAPGEF6 gene at the intronic SNP located at chromosome 13:16,120,088 (intron 7). The digestion site corresponds to the MnII restriction enzyme cleavage site located at position 16,120,095. The RAPGEF6 transcript (ENSGALT00010038615.1) comprises 30 exons and 29 introns. After enzymatic digestion, fragments of 184 bp and 77 bp were visualized on a 3% agarose gel (Fig. 2).

Fig. 2.

Electrophoresis of the RAPGEF6 gene at 16120088 in Marandi chickens on a 3% agarose gel, based on PCR-RFLP genotyping using the MnII restriction enzyme. The resulting genotypes include the homozygous wild genotype TT, which shows an uncut fragment of 261 bp, the heterozygous genotype TC (261, 184 bp), and the homozygous genotype CC (184, 77 bp). The symbol of“Cont” represents a pool of several digested samples with no prior knowledge about their genotypes

Three genotypes were identified: CC (homozygous C), TC (heterozygous), and TT (homozygous T). The observed genotypic frequencies were 0.329, 0.396, and 0.275 for CC, TC, and TT, respectively, corresponding to allelic frequencies of 0.527 (C) and 0.473 (T). According to the Hardy–Weinberg equilibrium test (χ² = 1.8, P > 0.05), the studied Marandi hen population did not deviate significantly from equilibrium, with expected genotype frequencies of 0.277 (CC), 0.499 (TC), and 0.234 (TT). Genotype TT was selected as the baseline genotype, and the effects of other genotypes were evaluated relative to this genotype. In Table 3, the estimated effects of the genotypes, with 95% confidence intervals relative to the reference genotype (TT), are presented for each of the studied traits examined.

Table 3.

Comparison of least square means of egg laying traits between the different genotypes of RAPGEF6 gene

Traits	TT (n=40)	TC (n=57)	CC (n=47)	P-value£	Estimated effects€		95% confidence limits
							Low	High	Low	High
					TC	CC	TC		CC
Performance traits
Total eggs count (TEC; number)	65.86	64.80	63.87	0.9026	-1.06	-1.98	-9.72	7.60	-10.70	6.73
Egg weight (EW; gr)	43.83	43.92	44.11	0.0600	0.44	0.28	0.171	0.715	0.005	0.560
Average eggs weight (AEW; gr)	45.20	45.13	44.67	0.7079	-0.07	-0.52	-1.53	1.39	-1.99	0.95
Feed conversion ratio (FCR)	3.08	2.76	3.01	0.3125	-0.33	-0.07	-0.80	0.15	-0.56	0.42
Eggs weight loss (EWL; gr)	0.49	0.51	0.50	0.8287	0.026	0.027	-0.036	0.069	-0.0457	0.063
First 50 Eggs weight (50EW; gr)	2004.4	2043.3	2023.3	0.8989	38.56	18.89	-129.2	206.33	-149.67	187.45
Average weight of first 50 eggs (A50E; gr)	44.13	44.09	43.68	0.7695	-0.42	-0.44	-1.49	1.41	-1.90	1.02
Egg laying rate (ELR; %)	58.80	57.85	57.03	0.9026	3.91	3.95	-8.68	6.79	-9.55	6.01
Egg mass (EM)	1.329	1.333	1.347	0.7152	0.004	0.0175	-0.04	0.05	-0.03	0.06
Growth related traits
Hatch weight (HW; gr)	31.06	31.36	32.02	0.3317	0.30	0.96	-1.07	1.66	-0.39	2.31
Maturity weight (MW; gr)	1647a	1826.06c	1763.33b	0.0144	179.00	116.27	59.01	298.99	-8.78	241.33
Estimated hatch weight (EHW; gr)	26.94	23.76	27.66	0.1462	-3.18	0.72	-7.78	1.43	-3.82	5.27
Age at inflection point (AIP; day)	72.74	72.06	73.01	0.9282	-0.95	-0.27	-6.19	4.29	-5.45	4.49
Weight at inflection point (WIP; gr)	616.40	635.48	655	0.3189	19.09	39.00	-32.35	70.52	-12.28	90.29
Ratio of maturity weight to weight at inflection point (MW/WIP)	2.80	2.86	2.73	0.3716	0.06	-0.07	-0.15	.0.26	-0.28	0.13
Hen weight at first laying (WFL; gr)	1552.3	1623.2	1617.2	0.3607	70.91	64.92	-33.34	175.17	-39.95	169.78
Reproductive traits
Age at first laying (AFL; day)	174.74	176.46	176.38	0.8628	1.68	1.61	-4.96	8.33	-5.08	8.30

^€ Genotype TT was selected as the baseline genotype, and the effects of other genotypes

£ means within the same line with different letters are significantly different (p< 0.05). Post-hoc power analysis revealed that the study had only 5% power to detect the significant effect observed for Maturity Weight (MW), and substantially lower power for the remaining 16 traits. Therefore, all reported differences should be interpreted as exploratory and preliminary. The minimum detectable effect size at 80% power for comparisons between TT and TC genotypes was estimated to be approximately 110 g for growth traits and 15 eggs for production traits

A T → C transition was detected at nucleotide 16,120,088 of the RAPGEF6 gene and confirmed by Sanger sequencing of PCR products (Fig. 3). Based on the sequencing results shown in Fig. 3, the chromatograms indicate that chickens M1, M2, and M5 had the heterozygous TC genotype. In contrast, chickens M3 and M4 displayed the homozygous TT genotype, while chicken M6 showed the homozygous CC genotype.

Fig. 3.

Sanger sequence of a fragment of RAPGEF6. Chromatogram profiles of six Marandi chickens sequenced by BGI (China), showing the T/C SNP at locus 16,120,088. Chickens M1, M2, and M5 exhibit the heterozygous TC genotype; M3 and M4 display the homozygous TT genotype; and M6 shows the homozygous CC genotype. The highlighted nucleotide represents the identified SNP position

The least squares mean comparisons of RAPGEF6 genotypes across the 17 studied traits, derived from the GLM, revealed that only Maturity Weight (MW) showed nominal significance at P = 0.05, while the remaining 16 traits showed no significant differences.

Among the 17 traits analyzed, only Maturity Weight (MW) showed a statistically significant difference among genotypes (P = 0.014). However, despite this nominal significance, the post-hoc power analysis indicated that the study had only 5% power to detect this effect (Power = 0.05), suggesting that this finding is statistically unstable and highly sensitive to sample size and variance. All other 16 traits showed no significant differences (P > 0.05), which is likely attributable to insufficient statistical power rather than a true absence of genetic effects.

Covariance structure matrix

The estimation of (co)variance structures for egg weight in Marandi hens from 21 to 39 weeks, using the REML method with the Proc Mixed Repeated command, revealed that the Autoregressive Moving Average (ARMA (1,1)) method provided the best fit among the eight tested structures. This was determined by the lowest values for the statistical fit criteria: AIC (45,867.8), AICc (45,867.8), and BIC (45,876.9). All models converged successfully.

A comparison of least square means for egg weight (EW) across repeated measures from 21 to 39 weeks in Marandi hens, analyzed using the Tukey test and Proc Mixed command with the optimal ARMA (1,1) covariance structure, indicated no significant differences among genotypes of the RAPGEF6 gene at locus 16,120,088. However, individuals with the CC genotype produced heavier eggs than other genotypes.

The ARMA (1,1) model’s superior fit (AIC = 45,867.8, AICc = 45,867.8, BIC = 45,876.9) was evident when compared to other structures, such as Heterogeneous Compound Symmetry (CHS) with an AIC of 45,867.8 and Compound Symmetry (CS) and AR [1] which yielded significantly higher information criteria. Consequently, the ARMA (1,1) structure was selected for subsequent analyses, including the estimation of genetic covariance components and the evaluation of associations between genotype and phenotypic traits. A comprehensive summary of all model criteria, parameter counts is detailed in Table 3.

Table 4.

Covariance structure of statistical model for egg production in Marandi hens during 21 to 39 weeks of age

Covariance structures	Parms€	AIC£	AICC a	BICΩ
Banded Toeplitz (TOEP (1))	t	55,409.7	55,409.7	55,412.8
Heterogeneous TOEP(TPEPH)	2t−1	55,178.6	55,181.0	55,509.0
Heterogeneous compound symmetry (CHS)	t + 1	46,590.0	46,592.4	46,923.3
Autoregressive Moving Average (ARMA (1,1))	3	45,867.8	45,867.8	45,876.9
Autoregressive (1) (AR (1))	2	48,232.7	48,232.7	48,238.8
Compound Symmetry (CS)	2	48,359.0	48,359.0	48,365.1
Variance Components (VC)	q	55,409.7	55,409.7	55,412.8
Heterogeneous TOEP (2) (TPEPH (2))	t + q−1	51,212.9	51,215.3	51,546.2

^€ number of covariance parameters in the structure. . 

t denotes the overall dimension of the covariance matrix. All covariance structures were converged successfully

£Akaike Information Criterion (AIC)

^aAkaike Information Criterion correct (AICC), and Bayesian

^ΩInformation Criterion (BIC) Information Criterion

Discussion

Egg production improvement and the optimization of reproductive traits are central objectives in laying hen breeding programs. These traits are typically quantitative, regulated by multiple genes, and exhibit low to moderate heritability, generally ranging from 0.10 to 0.35 for egg number and up to 0.45 for egg weight [17, 18]. The consistent results between Sanger sequencing and PCR–RFLP genotyping for the RAPGEF6 gene confirming high genotyping accuracy validate the reliability of the genomic data used in this study and will yield specific heritability estimates for the traits studied. Several studies, particularly in areas such as production traits in livestock and poultry, or physiological characteristics in humans, involve multiple measurements taken from the same individual over time [19]. These data are referred to as repeated records or repeated measures. Crucially, each measurement within an individual is correlated with other measurements from the same individual, rather than being independent. Consequently, accurate modeling necessitates the precise determination of the covariance matrix among these records [20].

Genotypic frequencies and population dynamics

The identification of CC, TC, and TT genotypes with allelic frequencies of 0.527 (C) and 0.473 (T) provides a baseline for the genetic structure of Marandi hens. Interestingly, the Marandi population was in Hardy–Weinberg equilibrium (χ² = 1.8), which demonstrated a lack of intensive artificial selection for this specific locus. In contrast, Jing Hong chickens show significantly higher frequencies of the T allele and TT/TC genotypes [4]. This discrepancy is likely due to the rigorous commercial selection programs applied to Jing Hong birds for production efficiency, whereas Marandi hens, as an indigenous breed, have been shaped primarily by natural selection and environmental adaptation. The equilibrium in Marandi hens implies that the RAPGEF6 polymorphism has not yet been exploited as a selection criterion, offering a potential window for future genetic investigation.

Comparative analysis of production traits

The association analysis revealed that while some traits showed numerical differences, many did not reach statistical significance threshold. For instance, in Jing Hong hens, the RAPGEF6 locus is a useful marker for the Egg Laying Rate (ELR), with the TC genotype being superior [4]. However, in Marandi hens, although the CC genotype had a higher ELR, the differences were not significant. This suggests that the impact of RAPGEF6 may be breed-specific or influenced by different genetic backgrounds.

Regarding egg weight, the wild-type TT genotype was slightly superior in Marandi hens, though not significantly. This mirrors findings in other species; for example, investigating polymorphism in the cGF-1R gene of Peking quail [21] or the Ovocalyxin-32 gene in native hens [22, 23] has shown more pronounced effects on egg weight. Similarly, the Growth Hormone (GH) gene is an essential gene for avian growth and production [24], yet like our RAPGEF6 results, studies on cGH in Nagoya chicks often show non-significant differences in egg numbers [25–27].

Maturity weight and biological mechanisms

In the present study, a nominally significant association was observed between the RAPGEF6 SNP and Maturity Weight (MW) in Marandi hens (P = 0.014). However, post-hoc power analysis was performed using the GLMPOWER procedure in SAS (with n = 153 fixed), and it was revealed that only 5% power was available to detect this effect, which is far below the conventionally acceptable threshold of 0.80. Therefore, this finding must be interpreted with great caution and cannot be regarded as definitive evidence of a genuine genetic association. This low power also clarifies that the genetic effect of this marker is weak, representing a poor marker index. Furthermore, given that only a single SNP was analyzed and that egg production and growth traits are typically controlled by numerous QTLs and are highly polygenic, the observed effect even if real is unlikely to account for a substantial proportion of the phenotypic variation in these traits [28].

Modeling and covariance structure

The use of mixed models for repeated measures such as egg production and weight over time requires an accurate specification of the covariance structure to ensure valid statistical inference. In this study, the ARMA (1, 1) structure was identified as the best-fit model (with the lowest AIC and BIC). The ARMA (1, 1) model captures temporal patterns in the data more effectively than simpler AR or MA models because it combines both the memory of previous observations (Autoregressive) and the persistence of random shocks (Moving Average). Repeated measurements are inherently correlated; by using ARMA (1, 1), we accounted for this non-independence and the potentially uneven time intervals in biological growth, leading to more precise estimates of genotype effects compared to studies using simpler structures like UN or CSH [19, 29, 30].

Study limitations, implications, and future research directions

Although GLM was employed for the analysis of most growth and performance traits, and repeated-measure models (with eight covariance structures) were employed for repeated egg weight records, the primary limitation of the present study remains the relatively small sample size (n = 153), which substantially reduced the statistical power required to detect significant differences, particularly in traits with low variability such as ELR and FCR. Post-hoc power analysis indicated that the study had only 5% power to detect the significant effect observed for MW, and considerably lower power for the remaining traits, severely limiting the interpretability of both the positive and negative findings. Furthermore, although our focus was on a specific SNP within the RAPGEF6 gene, egg production and growth traits are complex and polygenic in nature, governed by multiple genes and quantitative trait loci (QTL); consequently, a single marker cannot adequately account for the phenotypic variation in these traits. Additionally, potential confounding factors, such as subtle variations in micro-environmental conditions or epistatic interactions with other growth-related genes (e.g., STAT5B or ESR), were not fully controlled for, which may have masked the individual effect of the SNP in question. Moreover, our findings have not been validated in an independent population; yet the efficacy of marker-assisted selection (MAS) depends on the power of QTL detection and unbiased estimates of QTL effects, and estimates derived from the same data used for detection are often inflated, leading to overly optimistic assessments of MAS efficiency.

Consequently, no practical recommendations for breeding programs can be made based on the current data, and the present findings should be regarded merely as exploratory and hypothesis-generating, providing a foundation for future research. To consolidate these results, future investigations should focus on three key axes: first, functional genomics specifically, validating the expression levels of RAPGEF6 in ovarian and pituitary tissues to elucidate its biological role; second, multi-locus analysis by expanding the genetic panel to include markers in other growth-related genes (such as GH, ESR, and STAT5B) to evaluate epistatic interactions; and third, large-scale validation testing these associations in larger, independent, multi-generational populations with adequate statistical power (e.g., thousands of individuals) to confirm or refute the genuine impact of this locus on growth traits before any practical application in selection programs is considered.

Conclusion

Native chicken breeds in Iran, particularly the Marandi chicken, possess significant potential for reproductive, performance, and growth traits. This study was focused on the molecular analysis of the RAPGEF6 gene within the Marandi population. For the statistical analyses, GLM were employed for the evaluation of growth and production traits, while repeated-measure models with eight covariance structures among which ARMA (1,1) was identified as the optimal were applied to analyze repeated egg weight records. No significant differences were observed for most genotype comparisons derived from GLM analyses; however, the TT genotype was associated with a numerically favorable maturity weight among the assessed genotypes. Nevertheless, due to the single-SNP design, small sample size, and severely limited statistical power (5%), this observation should be interpreted as preliminary and exploratory rather than conclusive. Consequently, this SNP cannot yet be proposed as a reliable marker for breeding programs, and validation in larger, independent populations is essential before any practical application can be considered. Notably, the ARMA (1,1) structure was identified as the optimal covariance model for analyzing repeated records of egg-laying performance, improving the evaluation of egg production data, which is considered a valuable methodological contribution of this study.

Abbreviations

50EW

Cumulative weight of the first 50 eggs

A50E

Average weight of the first 50 eggs

AEW

Average egg weight

AFL

Age at first laying

AIP

Age at inflection point

AR

Autoregressive

AR (1)

Autoregressive (1)

ARH (1)

Heterogeneous autoregressive (1)

ARMA (1,1)

Autoregressive Moving Average (1,1)

CS

Compound Symmetry

CSH

Heterogeneous Compound Symmetry

EHW

Estimated hatch weight

ELR

Egg laying rate

EM

Egg mass

EW

Egg weight

EWL

Egg weight loss

FA1(2)

Non-diagonal factor analytic

FCR

Feed conversion ratio

GLM

Generalized Linear Model

GLMPOWER

SAS procedure for power analysis

HW

Hatch weight

MA

Moving Average

MAS

Marker-assisted selection

MW

Maturity weight

MW/WIP

Ratio of maturity weight to weight at inflection point

POPGENE

Population Genetic Analysis software

PROC MIXED

SAS procedure for mixed models

RAPGEF6

Rap Guanine Nucleotide Exchange Factor 6

TEC

Total eggs count

TOEP

Toeplitz

TOEP (1)

Banded Toeplitz

TOEPH

Heterogeneous Toeplitz

TOEPH (2)

Heterogeneous Toeplitz (2)

UN

Unstructured

VC

Variance components

WFL

Hen weight at first laying

WIP

Weight at inflection point

Authors’ contributions

MTZ: Data curation; Formal analysis; Methodology; Writing—review & editing; Writing—original draft; investigation; Software; Project administration; Resources. KH: Project administration; Validation; Supervision; Methodology; Conceptualization; Writing—review & editing. SA: Software; Formal analysis; Methodology; Validation; Project administration; Conceptualization. A J: Validation; Supervision; Visualization. MO: Validation; Supervision; Investigation.

Funding

This research did not receive any specific funding.

Data availability

The GenBank accession numbers for nucleotide sequences on the NCBI site are as follows: BankIt3029886 seqzarrinniaM1 PX673850 BankIt3029890 seqHasanpur-zarrinniaM2 PX673851 BankIt3029896 seqAlijani-zarrinniaM3 PX673852.

Declarations

Ethics approval and consent to participate

Ethical approval for this study was obtained from the Institutional Review Board (IRB) of Tabriz University (Ethical Code: IR.TABRIZU.REC.1404.133), which ensures compliance with animal welfare regulations.

Consent for publication

Not applicable. This study involved experimental animals (laying hens), and no identifying personal or clinical data were included.

Competing interests

The authors declare no competing interests.