Estimation of genetic parameters for growth traits of white-feather broilers and research on genotype-environment interaction

Abstract

This study aimed to estimate the variance components of growth traits in white-feathered broilers under different environments and detect the genotype-by-environment interaction (G × E) effects of the same trait across various environments. Using growth performance records of white-feathered broilers from 6 batches under 3 environments, five key growth traits were analyzed: 45-day weight (45DW), weight gain (WG), feed conversion ratio (FCR), feed intake (FI), and mid-term metabolic weight (MWT), encompassing a total of 8,651 phenotypic records. Detailed quality control and grouping were performed on the original phenotypic data. Subsequently, variance components analysis of the five growth traits across the three feeding environments with different energy and protein levels was conducted using the mixed linear model in SPSS software. The genetic correlations of the same trait between different environments were then estimated using the multi-trait animal model in ASReml software as indicators for detecting G × E effects. The results showed that the established quality control criteria effectively eliminated abnormally distributed phenotypic values. In the variance components analysis, the genotype-by-environment interaction (G × E) effects had extremely significant impacts on all growth traits. These effects not only altered broiler growth performance due to variations in feeding environments with different energy and protein levels but also stemmed from differences in environmental adaptability among different families. The bivariate model revealed the strongest G × E effects between breeder feed and high-energy broiler feed, followed by those between low-energy broiler feed and the other two feeds, with most G × E effects being significant. In conclusion, the genetic performance of the same growth trait varied across different feeding environments, and significant G × E effects existed between some feeding environments. Therefore, environmental differences and G × E effects on genetic progress should be considered when conducting genetic improvement of growth traits in white-feathered broilers.

Introduction

As the world's most populous country, China has enormous chicken consumption and output. In China, poultry meat accounts for 23% of total meat production, with chicken contributing 60% of poultry meat output, making it the country's second-largest meat product. Broilers have production advantages including a short production cycle, high feed conversion rate, and high meat yield, enabling more meat production with less land and feed. According to the data of China Statistical Yearbook, the per capita consumption of chicken increased from 8.9 kg in 2017 to 12.4 kg in 2023, with an annual growth rate of 6.6% (Zhang et al., 2025). According to data from the United States Department of Agriculture, global broiler production in 2023 was 102,389,000 tons, an increase of 549,000 tons from 2022, of which white-feathered broilers accounted for about 70% (Zhuang et al., 2025). Although China's white-feathered broiler industry holds a significant position in terms of consumer market and production scale, the autonomous and controllable capacity of its core industrial segments still faces severe challenges. For a long time, domestic white-feathered broiler breeding has experienced discontinuity due to historical reasons, resulting in insufficient accumulation of superior genetic resources and heavy reliance on imported germplasm. In recent years, with the development and promotion of indigenous varieties such as "Shengze 901," "Guangming 2," and "Ward 188," the market share of domestic germplasm has gradually increased; however, as of 2025, it remains only 28%, with 72% of germplasm demand dependent on imports. This renders the industrial chain vulnerable to external shocks such as disease outbreaks, geopolitical tensions, and logistical disruptions. Meanwhile, the R&D investment intensity of domestic breeding enterprises is insufficient (accounting for less than 2% of sales revenue, compared with 5–8% at the international advanced level), and the application of key technologies such as genomic selection and biosecurity purification lags significantly behind international giants, indicating that the bottleneck of weak breeding foundation has not been fundamentally resolved.

In broiler breeding, growth traits are among the most important functional traits of white-feathered broilers. They are moderate-to-low heritability traits, meaning they are significantly influenced by factors other than genetics. Broiler growth mainly relies on nutrients provided by feed. Feed contains a variety of key components that not only maintain broilers' life activities but also promote cell growth and development, and regulate their physiological functions, thereby ensuring the full exertion of broilers' health and growth performance. The nutritional level of feed is crucial for broilers' health, which directly affects the quality of their growth traits. The energy content (Massuquetto et al., 2020) and protein content (Nikoletta et al., 2021) in broiler feed are the main factors affecting broiler growth traits. Energy is the most fundamental requirement for broiler growth. Zhao et al. (2025) studied the effect of dietary energy levels on broiler growth performance and found that feeding low-energy diets has negative impacts on broiler weight gain and feed utilization. Hu et al. (2021) also reported in their research on dietary energy levels and broiler growth performance that broilers fed high-energy diets have higher average daily gain (ADG) than those fed low-energy diets. The protein and amino acid contents in feed also play important roles in broiler growth traits. Yalçin et al. (1996) studied the effect of dietary protein content on broiler growth performance and showed that feeding low-protein diets to cockerels aged 0-3 weeks leads to slow weight gain. Abou-Elkhair et al. (2020) found that a reduction in dietary crude protein (CP) content results in a linear decrease in broiler growth performance. However, when reducing protein content while adding an appropriate amount of amino acids (Law et al., 2018), there is no significant negative impact on broiler growth performance (Kriseldi et al., 2018; van Harn et al., 2019). Meanwhile, when formulating reduced-crude protein (CP) diets for broilers, the accessibility of feed ingredients plays a critical role in determining how efficiently nutrients are utilized, which ultimately impacts the birds' growth performance(Kareem et al., 2025).

The phenotype of an organism is not only influenced by its genotype but also by environmental effects and genotype-by-environment (G × E) interactions. The ability to exhibit different phenotypes and alter traits in response to various environmental factors is known as phenotypic plasticity (Agrawal, 2001). Another ability is homeostasis, which allows organisms to adapt to environmental influences without changing their inherent traits. Maintaining homeostasis is particularly crucial for the health of organisms. The definition of environment is broad, encompassing internal and external environments, such as cell type, tissue morphology, physiological state, disease factors, climatic conditions, geographical location, and feeding levels (Liu et al., 2019; van der Laak et al., 2016; Zwald et al., 2001). Compared with traditional single-environment or single-trait analysis, G × E enables a deeper understanding of the relationships between an individual’s phenotype, genes, and living environment. Numerous studies published in recent years have aimed to incorporate G × E components into livestock breeding to obtain more valuable traits or identify additional loci associated with complex quantitative traits(Berghof et al., 2018; Rauw and Gomez-Raya, 2015). For example, in G × E research on beef cattle, relevant production traits (such as milk yield and milk quality (Bohlouli et al., 2021; Carrara et al., 2021), growth traits (such as birth weight and weaning weight) (Toghiani et al., 2020), and some functional traits (such as fertility) (Schmid et al., 2021) have been considered. In livestock studies, most production and economic traits are determined by quantitative traits, which are often most susceptible to environmental factors. Studies on cattle using methods such as environmental genetic correlation models (Guzzo et al., 2018; Pfeiffer et al., 2016; Shabalina et al., 2021) multi-trait models (Schmid et al., 2021), and reaction norm models (Huquet et al., 2012) have found that G × E may affect important economic traits such as milk yield, milk quality (Shariati et al., 2007), and fertility (Strandberg et al., 2009), manifested as phenotypic differences across different environments. Additionally, previous studies have indicated the presence of G × E effects on important economic traits in sheep, including body weight (Santana et al., 2013; Steinheim et al., 2008) backfat thickness (McLaren et al., 2014), weaning weight, and milk yield (Abousoliman et al., 2020). In poultry production, considering genotype-by-environment (G × E) effects can facilitate the selection and breeding of poultry breeds with strong adaptability based on environmental conditions, thereby improving production performance and economic benefits. Meanwhile, it can avoid problems such as unstable genetic improvement effects or resource waste caused by ignoring environmental factors, realizing precision breeding and sustainable development.

Materials and methods

Phenotype collection

The data used in this study were derived from a white-feathered broiler population with the same generation selection and consistent genetic background. Families with low feed conversion ratio (FCR) were selected for mating, and the 28-day body weight of sample white-feathered broilers was recorded. The experiment was conducted using single-cage rearing. The broilers were fed three types of diets with different energy and protein levels (breeder feed, high-energy broiler feed, and low-energy broiler feed) under the same other environmental conditions. White-feathered broilers at 28 days of age were divided into three dietary treatment groups: Batch 1 was fed breeder diet (crude protein 14.00%, metabolizable energy 11.7649 MJ/kg), Batches 2, 3, and 4 were fed low-energy broiler diet (crude protein 20.08%, metabolizable energy 12.9791 MJ/kg), and Batches 5 and 6 were fed high-energy broiler diet (crude protein 19.83%, metabolizable energy 13.7327 MJ/kg). These three diets represented different nutritional levels, with energy and protein levels serving as environmental variables: breeder diet represented low-energy and low-protein levels, while low-energy broiler diet and high-energy broiler diet represented medium-energy high-protein and high-energy high-protein levels, respectively, to evaluate the effects of different nutritional levels on growth traits. Each batch was raised during different time periods. Except for dietary energy and protein levels, other rearing conditions (housing environment, management practices, sufficient water supply, etc.) were maintained consistently. The experimental period lasted from 28 to 45 days of age, with ad libitum feeding and free access to water. Feed intake (FI) and feed conversion ratio (FCR) were measured using the single cage as the experimental unit. Daily feed consumption of each bird was recorded, and body weight and feed intake were measured at the end of the experiment to calculate each trait.

• (1). 45-day weight (45DW): Measured using an electronic scale with a precision of 0.1 g. Residual feed in the trough was removed 12 h before weighing, and the body weight at 45 days of age was recorded in grams (g).

• (2). Weight gain (WG). Measured using an electronic scale with a precision of 0.1 g. Residual feed in the trough was removed 12 h before weighing, and the body weight at 28 days of age was recorded. WG was calculated as the weight increment from 28 to 45 days of age, in grams (g).

• (3). Feed intake (FI): The feed consumption from 28 to 45 days of age was measured using an electronic scale with a precision of 0.1 g, in grams (g).

• (4). Feed conversion ratio (FCR): Calculated as $FCR=\frac{FEEDW}{WG}$, with the unit of (g feed/g gain).

• (5). Mid-term metabolic weight (MWT): Calculated as ${\left(\frac{28DW+45DW}{2}\right)}^{0.75}$, in grams (g).

It should be noted that in this study, "batch" refers to groups of broilers raised in the same chicken house during different periods using the same feed formula. Therefore, differences between batches mainly reflect non-nutritional environmental fluctuations (such as seasonal changes, temperature and humidity variations, and minor differences in management practices), rather than changes in nutritional levels.

The nutritional components of the three different diets are presented in Table 1.

Table 1.

The nutritional components of the three different diets.

Feed Type	Crude Protein	Metabolizable Energy
Breeder Feed	14.00%	11.7649 MJ/kg
Low-Energy Broiler Feed	20.08%	12.9791 MJ/kg
High-Energy Broiler Feed	19.83%	13.7327 MJ/kg

Data quality control and grouping

The grouping criterion for outlier removal was based on each individual value of every trait, with values rounded to four decimal places. All values were continuous, with each value forming its own group rather than being broadly categorized. Frequency anomaly peaks were identified for specific values, and when frequencies were abnormally high, the associated families were removed—specifically, the paternal half-sib family containing that individual was excluded from the same batch.

The phenotypic quality control for each trait included three steps: (1) retaining phenotypic data within three standard deviations; (2) screening for data with abnormal phenotypic distributions by family; (3) selecting families with no fewer than 10 individuals. The determination of outliers was based on the relative difference (RD; ③) between the frequency of a value in a certain family (probability, Pro; ①) and the average frequency of the four adjacent values of that value (probability nearby, Pro_n; ②). If the RD of the value was greater than 0 (i.e., Pro was higher than Pro_n; ④), it was determined that there was an abnormal peak in the distribution of the trait phenotype of the family under the feed condition at that value. It should be noted that the grouping criterion for outlier removal was based on each individual value of every trait, with values rounded to four decimal places. All values were continuous, with each value forming its own group rather than being broadly categorized. By identifying frequency anomaly peaks for specific values, cases with significantly high frequencies resulted in the removal of all individuals from the paternal half-sib family containing that value—that is, the entire paternal half-sib family to which the abnormal individual belonged was excluded from the same batch. Subsequently, the frequency of phenotypic outliers in each family was calculated. If the frequency of phenotypic outliers obtained based on family data was higher than Pro_n based on the batch data, the phenotypic distribution of the trait in that family was determined to be abnormal, and all data from that family were excluded.

After outlier removal, to explore the impact of genetic relationships among cockerels on G × E detection, records with no fewer than 10 offspring per paternal half-sib family were retained. For each trait, the obtained data were divided into three groups for subsequent analysis: (1) all data (5 datasets); (2) data by each diet type (15 datasets); (3) merged data of two different diet types (15 datasets).

Statistical models

• (1). Single-environment Analysis of Variance (ANOVA) model

Assuming there are $F$ families, $B$ batches, and $N$ individuals per batch, the linear model for this experiment is as follows:

$$\text{xijl}=\mu +\text{Bi}+\text{Fj}+\text{FBij}+\text{eijl}$$

Where $xijl$is the observation value of the $j-\text{th}$ family in the $i-\text{th}$ batch, $\mu$ is the overall mean, $Bi$ is the effect of the $i-\text{th}$ batch, $Fj$ is the effect of the $j-\text{th}$ family, $FBij$ is the interaction effect between the $j-\text{th}$ family and the $i-\text{th}$ batch, and $eijl$ is the residual error.

• (2). Multi-environment Analysis of Variance (ANOVA) model

Suppose there are S environments, each containing B batches, with F families and N individuals per batch, the linear model for this experiment is as follows:

$$x_{ijkl}=\mu +S_i+B/S_{i\left(j\right)}+F_k+{\left(FB/S\right)}_{ijk}+F{S}_{ik}+{ϵ}_{ijkl}$$

Where $x_{ijkl}$ is the observation value of the l-th individual from the k-th family in the j-th batch under the $i-th$ environment, $\mu$ is the overall mean, $S_i$ is the effect of the $i-th$ environment, $B/S_{i\left(j\right)}$ is the effect of the$j-th$ batch within the $i-th$ environment, $F_k$ is the effect of the $k-th$ family, ${\left(FB/S\right)}_{ijk}$ is the interaction effect between the $k-th$ family and the$j-th$ batch within the $i-th$ environment, $F{S}_{ik}$ is the interaction effect between the $k-th$ family and the $i-th$ environment, and ${ϵ}_{ijkl}$ is the residual error.

Analysis of variance was performed using SPSS 26 software. Based on the above model, the restricted maximum likelihood (REML) method was used to estimate the variance components, including family variance components (${\sigma }_F^2$), family × batch variance components (${\sigma }_{FB}^2$), family × batch/environment variance components (${\sigma }_{FB/S}^2$), family × environment variance components (${\sigma }_{FS}^2$), and residual error variance components (${\sigma }_E^2$).

• (3). Genetic Parameters in White-Feathered Broilers

To estimate genetic parameters for various traits in white-feathered broilers, this study employed a univariate animal model based on Datasets (1), (2), and (3) to estimate genetic parameters for the overall data, single environments, and pairwise environment combinations, respectively. The model is presented as follows:

$$y=X\beta +Zu+Ws+e$$

where $y$ is the vector of phenotypic observations; $\beta$ is the vector of fixed effects, including only batch effect; $u$ is the vector of random additive genetic effects with $u\sim N\left(0,A{\sigma }_{\alpha }^2\right)$ , where A is the numerator relationship matrix constructed from pedigree data; s is the vector of random sire half-sib family × environment interaction effects with $s\sim N\left(0,I{\sigma }_s^2\right)$; and e is the vector of random residual effects with $e\sim N\left(0,I{\sigma }_e^2\right)$ . $X$, Z, and $W$ are the corresponding incidence matrices. In single-environment analysis, environment was defined as batch, and the model evaluated family stability across different batches through the family × batch interaction effect. In multi-environment analysis, environment was defined as feed type (breeder feed, low-energy broiler feed, and high-energy broiler feed), and the model quantified genotype × environment (G × E) interaction effects through the family × feed type interaction effect. Variance components were estimated using restricted maximum likelihood (REML), and heritability was calculated as $h^2=\frac{{\sigma }_{\alpha }^2}{\left({\sigma }_{\alpha }^2+{\sigma }_s^2+{\sigma }_e^2\right)}.$

• (4). Genetic Correlation and Genotype-Environment (G × E) Interaction

To evaluate the genotype-environment (G × E) interaction effect of the same trait across different environments, this study adopted the two-trait animal model to estimate the genetic correlations between each pair of environments based on Dataset (3). The model is presented as follows:

$$y=Xb+Zu+e$$

Where $y$ is the vector of individual phenotypic values, $y=\left[\begin{array}{c}{y}_1\\ y_2\end{array}\right]$; $X=\left[\begin{array}{cc}{X}_1& 0\\ 0& X_2\end{array}\right]$ is the incidence matrix of fixed effects; $b=\left[\begin{array}{c}{b}_1\\ b_2\end{array}\right]$ is the vector of fixed effects; $Z=\left[\begin{array}{cc}{Z}_1& 0\\ 0& Z_2\end{array}\right]$ is the incidence matrix relating individuals to their additive genetic effects;$u=\left[\begin{array}{c}{u}_1\\ u_2\end{array}\right]$ is the vector of individual additive breeding values; and $e=\left[\begin{array}{c}{e}_1\\ e_2\end{array}\right]$ is the vector of residual errors. The model can be rewritten as follows:

$$\left[\begin{array}{c}{y}_1\\ y_2\end{array}\right]=\left[\begin{array}{cc}{X}_1& 0\\ 0& X_2\end{array}\right]\left[\begin{array}{c}{b}_1\\ b_2\end{array}\right]+\left[\begin{array}{cc}{Z}_1& 0\\ 0& Z_2\end{array}\right]\left[\begin{array}{c}{u}_1\\ u_2\end{array}\right]+\left[\begin{array}{c}{e}_1\\ e_2\end{array}\right]$$

Where the subscripts 1 and 2 denote different environments: $X_1$ and $X_2$ are the fixed effects incidence matrices for Environment 1 and Environment 2, respectively; $Z_1$ and $Z_2$ are the incidence matrices for additive genetic effects; and $u_1$ and $u_2$ are the vectors of additive breeding values for Environment 1 and Environment 2, respectively.

The pedigree information consisted of 160 paternal half-sib families, with pedigree records including only individuals and their parents (one generation of pedigree). The pedigree completeness was 100%. The additive genetic relationship matrix ($A$) was constructed based on the pedigree information using the algorithm of Henderson. The distributions of random effects are assumed as follows:

$$\left[\begin{array}{c}{u}_1\\ u_2\end{array}\right]\sim N\left(0,A\otimes \left[\begin{array}{cc}{\sigma }_{u_1}^2& {\sigma }_{u_1{u}_2}\\ {\sigma }_{u_1{u}_2}& {\sigma }_{u_2}^2\end{array}\right]\right);\left[\begin{array}{c}{e}_1\\ e_2\end{array}\right]\sim N\left(0,I\otimes \left[\begin{array}{cc}{\sigma }_{e_1}^2& 0\\ 0& {\sigma }_{e_2}^2\end{array}\right]\right)$$

Where ${\sigma }_{u_1{u}_2}$ is the additive genetic covariance between the two environments. The permanent environmental covariance and residual covariance between the two environments are assumed to be 0. This model is used to measure the presence of genotype-by-environment (G × E) interaction through the genetic correlation of the same trait across different populations under similar genetic backgrounds. If the genetic correlation is less than 0.8, it is considered that there exists a genotype-by-environment (G × E) interaction between the populations (Martin et al., 2023).

Results and analysis

Phenotypic data quality control

After QC, phenotypic data counts for different traits under each feed environment are shown in Table 2. Batch 1 had a relatively low retention rate: the retention rate of 45DW was 69.87%, and the FCR and WG did not exceed 73%. FI retention for Batch 2only 69.43%. The low retention rate was due to the removal of outliers during quality control and the requirement that each half-sib family should retain no fewer than 10 individuals. In contrast, Batches 2 and 4 had higher data retention rates, with the retention rates of multiple indicators approaching or exceeding 97%. In addition, the retention rate of feed conversion ratio (FCR) in Batch 4 was only 88.45%, while the retention rate of FCR in Batch 6 was 94.52%, indicating large fluctuations in retention rates across different traits and batches. Overall, the retention rate of data after quality control reflects the quality and consistency of the original data. A lower retention rate may be caused by the selection of half-sib families, while a higher retention rate indicates a more uniform data distribution. Such differences in data retention rates among batches suggest that genetic background and rearing conditions may jointly affect phenotypic stability, which should be fully considered in subsequent genetic evaluation models to improve the accuracy of parameter estimation. Fig. 1

Table 2.

Quality control results of growth and feed efficiency traits under each environment.

Batch	Raw Data (Individuals)	Quality-Controlled Data (Individuals)					Retention Rate after Quality Control (%)
		45DW(g)	FCR (g feed/g gain)	WG(g)	MWT(g)	FI(g)	45DW(g)	FCR (g feed/g gain)	WG(g)	MWT(g)	FI(g)
1	1706	1235	1229	1222	1192	1192	72.39%	72.04%	71.63%	69.87%	69.87%
2	1796	1756	1739	1756	1763	1247	97.77%	96.83%	97.77%	98.16%	69.43%
3	1212	1184	1072	1177	1185	1172	97.69%	88.45%	97.11%	97.77%	96.70%
4	1508	1483	1473	1476	1483	1473	98.34%	97.68%	97.88%	98.34%	97.68%
5	1298	1027	1187	1196	1207	1200	79.12%	91.45%	92.14%	92.99%	92.45%
6	1131	760	1069	1068	1076	1070	67.20%	94.52%	94.43%	95.14%	94.61%

Fig. 1.

Flowchart and example of quality control for outliers in phenotypic distribution.

Due to the low data retention rate and large volume of deleted data under the breeder feed feeding environment, Fig. 2 shows the quality control process for abnormal distribution of feed conversion ratio (FCR) data under the high-energy broiler feed environment. It can be seen that Step 3 of the quality control screened out most individuals. By examining the original data distribution, it was found that there were abnormal peaks in the data under this feeding environment (e.g., FCR = 1.60). After quality control, the data distribution became relatively consistent. The elimination of abnormal peaks ensures that the phenotypic data conforms to the assumption of normal distribution, which lays a reliable foundation for subsequent genome-wide association analysis and genomic selection. Fig. 3

Fig. 2.

Data retention volume at each quality control step for traits and quality control of abnormal distribution of 45DW across 6 batches. a. Number of retained individuals after each quality control step for phenotypes across different batches: Step (1) retains phenotypic data within three standard deviations; Step (2) eliminates outliers; Step (3) retains half-sib families with no less than 10 individuals; b. Original data distribution of feed conversion ratio (FCR) in white-feathered broilers fed breeder feed; c. Data distribution of FCR in white-feathered broilers fed breeder feed after quality control.

Fig. 3.

Diagram showing significant differences in phenotypes among different environments.

Descriptive statistics of growth traits in white-feathered broilers

Descriptive statistical analysis was performed on the growth traits of white-feathered broilers under three different environments, and the results are shown in the following table. The table lists the mean, standard deviation (SD), and phenotypic coefficient of variation (CV) for each environment and the overall population, which are used to evaluate the central tendency and dispersion of each trait. These descriptive statistics provide a basic understanding of phenotypic characteristics and variation patterns, which is essential for subsequent genetic analysis and environmental effect evaluation.

The growth performance of half-sib families of white-feathered broilers is shown in Table 3. There were differences in growth performance under different feeding environments: Under the breeder feed environment: The means of 45-day weight (45DW), feed conversion ratio (FCR), weight gain (WG), mid-term metabolic weight (MWT), and feed intake (FI) were 2935.60 g, 1.73, 1729.17 g, 307.24 g, and 2988.68 g, respectively. The coefficient of variation (CV) of each indicator ranged from 3.28% to 5.65%, with the smallest overall variation and the most stable group growth performance. Under the low-energy broiler feed environment: The means of 45DW, FCR, WG, MWT, and FI were 3226.86 g, 1.56, 2013.09 g, 323.37 g, and 3131.53 g, respectively. The CV of each indicator ranged from 4.40% to 8.92%, and the means of 45DW and WG were the highest among all environments. Under the high-energy broiler feed environment: The means of 45DW, FCR, WG, MWT, and FI were 3152.92 g, 1.52, 1928.22 g, 319.56 g, and 2929.08 g, respectively. The CV of each indicator ranged from 4.67% to 9.40%, and the mean of FCR was the lowest among all environments. Across all environments (including all samples): The means of 45DW, FCR, WG, MWT, and FI were 3160.79 g, 1.58, 1944.81 g, 319.84 g, and 3051.34 g, respectively. The CV of each indicator ranged from 4.68% to 10.13%, with the largest overall variation and the worst group growth consistency. 45DW and WG were the highest under the low-energy broiler feed environment and the lowest under the breeder feed environment. FCR was the lowest under the high-energy broiler feed environment and the highest under the breeder feed environment. MWT was the highest under the high-energy broiler feed environment and the lowest under the breeder feed environment. FI was the lowest under the breeder feed environment and the highest under the low-energy broiler feed environment. It can be seen that the half-sib families of white-feathered broilers showed different growth differences under each environment due to the influence of energy and protein levels in the feed. These results reflect obvious genotype-environment interaction effects, suggesting that the same genetic material exhibits different phenotypic expressions under different nutritional conditions, which is of great importance for targeted breeding and feeding optimization.

Table 3.

Descriptive statistics of growth and feed efficiency traits in white-feathered broilers.

Environment	Number of families paternal half-sib families	Number of families maternal half-sib families	Number of individuals	Trait	Mean	Standard Deviation	Phenotypic Coefficient of Variation
Breeder Feed	89	589	1192	45DW	2935.6	127.36	4.34%
				FCR	1.73	0.08	4.63%
				WG	1729.17	97.7	5.65%
				MWT	307.24	10.09	3.28%
				FI	2988.68	125.84	4.21%
Low-Energy Broiler Feed	153	1220	3792	45DW	3226.86	207.2	6.42%
				FCR	1.56	0.1	6.44%
				WG	2013.09	179.57	8.92%
				MWT	323.37	14.21	4.40%
				FI	3131.53	209.78	6.70%
High-Energy Broiler Feed	98	612	1787	45DW	3152.92	214.58	6.81%
				FCR	1.52	0.08	5.42%
				WG	1928.22	181.2	9.40%
				MWT	319.56	14.91	4.67%
				FI	2929.08	225.04	7.68%
All Environments	160	1337	6771	45DW	3160.79	224.19	7.09%
				FCR	1.58	0.12	7.32%
				WG	1944.81	196.92	10.13%
				MWT	319.84	14.96	4.68%
				FI	3051.34	223.96	7.34%

The figure shows the significant phenotypic differences in various growth traits of white-feathered broilers under different feeding environments: 45-day weight performs the best under the low-energy broiler feed environment and across all environments (marked as a), while it is slightly poorer under the breeder feed and high-energy broiler feed environments (marked as b); feed conversion ratio is higher under the breeder feed environment and across all environments (marked as a), and lower under the low-energy and high-energy broiler feed environments (marked as b); feed intake is lower under the breeder feed environment and across all environments (marked as a), and higher under the low-energy and high-energy broiler feed environments (marked as b); mid-term metabolic weight and weight gain perform the best under the breeder feed environment and across all environments (marked as a), while they are slightly poorer under the low-energy and high-energy broiler feed environments (marked as b). This indicates that the feeding environment has a significant impact on the growth performance of broilers, and there are obvious differences in trait performance under different environments. The significant environmental effects observed suggest that genetic evaluation and selection should be carried out in combination with specific feeding environments to improve the accuracy and efficiency of broiler genetic improvement programs.

Analysis of variance for growth traits in white-feathered broilers

Analysis of variance was performed on the growth traits of white-feathered broilers under three different environments, and the results are shown in the following Table 4. The Table 4 lists the degree of influence of different effects on traits under each environment. The breeder diet group contained only a single rearing batch; therefore, the batch effect and family × batch interaction effect could not be estimated and were excluded from the model.

Table 4.

Analysis of variance for growth and feed efficiency traits of half-sib families of white-feathered broilers in a single environment.

Environment	Source	Trait	Sum of Squares	Degrees of Freedom	Mean Square	F-Value	P-Value
Breeder Feed	Family	45DW	3030283.078	88	34435.035	2.323	0
		FCR	1.264	87	0.015	2.51	0
		WG	1791996.417	87	20597.66	2.369	0
		MWT	16049.645	84	191.067	2.01	0
		FI	2130295.6	84	25360.662	1.678	0
Low-Energy Broiler Feed	Family	45DW	13449234.51	152	88481.806	2.391	0
		FCR	2.84	152	0.019	1.911	0
		WG	8767937.489	152	57683.799	2.131	0
		MWT	66001.078	152	434.218	2.44	0
		FI	10676560.48	151	70705.699	2.073	0
	Batch	45DW	10856875.06	2	5428437.531	146.717	0
		FCR	0.046	2	0.023	2.373	0.093
		WG	10960458.65	2	5480229.323	202.443	0
		MWT	40460.06	2	20230.03	113.665	0
		FI	22053803.45	2	11026901.73	323.363	0
	Family*Batch	45DW	10071282.53	266	37861.964	1.023	0.388
		FCR	2.432	261	0.009	0.953	0.693
		WG	6972670.848	266	26213.048	0.968	0.63
		MWT	48737.913	266	183.225	1.029	0.362
		FI	10505148.24	265	39642.069	1.163	0.041
High-Energy Broiler Feed	Family	45DW	7365600.742	97	75934.028	1.745	0
		FCR	1.68	112	0.015	2.383	0
		WG	6153292.909	112	54940.115	1.797	0
		MWT	41667.146	113	368.736	1.725	0
		FI	11153532.91	112	99585.115	2.16	0
	Batch	45DW	708977.238	1	708977.238	16.292	0
		FCR	0.036	1	0.036	5.664	0.017
		WG	1423319.476	1	1423319.476	46.552	0
		MWT	1354.088	1	1354.088	6.335	0.012
		FI	2400418.342	1	2400418.342	52.061	0
	Family*Batch	45DW	3536320.562	76	46530.534	1.069	0.324
		FCR	0.857	110	0.008	1.238	0.051
		WG	3883629.809	109	35629.631	1.165	0.122
		MWT	24427.946	110	222.072	1.039	0.375
		FI	5696494.734	109	52261.42	1.133	0.169

Under the breeder feed environment, the family effect had a significant impact on 45-day weight (45DW), feed conversion ratio (FCR), weight gain (WG), mid-term metabolic weight (MWT), and feed intake (FI) (all P-values < 0.0001). Under the low-energy broiler feed environment, the batch effect had a significant impact on all traits except FCR; the family effect had a significant impact on 45DW, FCR, WG, and MWT (P-values < 0.0001); and the family × batch interaction effect showed significance on FI (P = 0.0410). Under the high-energy broiler feed environment, the family effect had a significant impact on all traits; the batch effect had a significant impact on 45DW, FCR, WG, and MWT (P-values < 0.0001); and the family × batch interaction effect had no significant impact on any traits. Overall, the family effect on growth traits was relatively consistent across different environments, while the batch effect and family × batch interaction effect showed varying degrees of impact under different environments. This indicates that feeding environments and management measures play an important role in the expression of growth traits in broilers. The stable and significant family effect suggests that there is abundant genetic variation for growth traits in the population, which is conducive to genetic selection. Meanwhile, the inconsistent batch and interaction effects across environments highlight the necessity of including environmental and management factors in statistical models to avoid biased estimation of genetic parameters.

In the multi-environment analysis of variance for growth traits of white-feathered broilers, the environment effect, family effect, and family × environment interaction effect all showed extremely significant impacts on 45-day weight (45DW), feed conversion ratio (FCR), weight gain (WG), mid-term metabolic weight (MWT), and feed intake (FI) (P < 0.0001). This indicates that different feeding environments have an extremely significant impact on the growth traits of broilers, and the performance of the genetic background of families also varies significantly under different environments. In particular, the family × environment interaction effect had a significant impact on each trait (P < 0.05), indicating that the growth trait performance of families differs under different environments, reflecting the varying adaptability of families to the environment and the existence of genotype-environment (G × E) interaction effects for growth traits. The extremely significant family × environment interaction further confirms that genotype-environment interaction cannot be ignored in broiler breeding. For practical production and breeding, it is necessary to implement targeted selection or environmental-specific breeding strategies according to different feeding conditions, so as to fully exert genetic potential and improve breeding efficiency. Table 5

Table 5.

Multi-environment combined analysis of variance for growth and feed efficiency traits of half-sib families of white-feathered broilers.

Source	Trait	Sum of Squares	Degrees of Freedom	Mean Square	F-Value	P-Value
Environment	45DW	65881586.3100	2	32940793.1500	946.7510	0.0001
	FCR	28.2120	2	14.1060	1735.9750	0.0001
	WG	62560091.9000	2	31280045.9500	1243.2380	0.0001
	MWT	203100.3080	2	101550.1540	578.8200	0.0001
	FI	51308487.2600	2	25654243.6300	741.4480	0.0001
Family	45DW	15177463.7600	159	95455.7470	2.7430	0.0001
	FCR	3.5350	159	0.0220	2.7360	0.0001
	WG	10193025.4000	158	64512.8190	2.5640	0.0001
	MWT	77619.1550	158	491.2600	2.8000	0.0001
	FI	15133680.6800	158	95782.7890	2.7680	0.0001
Batch*Environment	45DW	11565852.3000	3	3855284.1000	110.8050	0.0001
	FCR	0.0820	3	0.0270	3.3660	0.0180
	WG	12383778.1200	3	4127926.0410	164.0660	0.0001
	MWT	41814.1480	3	13938.0490	79.4450	0.0001
	FI	24454221.7900	3	8151407.2650	235.5880	0.0001
Family*Environment	45DW	7549871.4710	178	42415.0080	1.2190	0.0260
	FCR	2.0620	192	0.0110	1.3220	0.0020
	WG	5767663.6420	193	29884.2680	1.1880	0.0400
	MWT	41626.2630	191	217.9390	1.2420	0.0140
	FI	8357940.9940	189	44221.9100	1.2780	0.0070
Family*Environment*Batch	45DW	13607603.1000	342	39788.3130	1.1440	0.0380
	FCR	3.2890	371	0.0090	1.0910	0.1160
	WG	10856300.6600	375	28950.1350	1.1510	0.0260
	MWT	73165.8590	376	194.5900	1.1090	0.0770
	FI	16201642.9700	374	43319.9010	1.2520	0.0010
Error	45DW	235204187.7000	6760	34793.5190
	FCR	57.2130	7041	0.0080
	WG	180222052.8000	7163	25160.1360
	MWT	1258807.0680	7175	175.4430
	FI	246595728.2000	7127	34600.2140

Heritability analysis of white-feathered broilers

The Table 6 presents the genetic parameter estimates for five broiler growth traits under three feeding environments. The low-energy feed group exhibited the highest heritability for 45-day weight (0.4307±0.0460), metabolic weight (0.4439±0.0450), and daily gain (0.4096±0.0460), with G × E interaction heritability for these three traits (0.1023±0.0186, 0.1075±0.0190, 0.1100±0.0200) also higher than the other two groups. The high-energy feed group performed best in feed intake (0.4158±0.0603), followed by daily gain (0.3833±0.0576) and feed conversion ratio (0.3593±0.0677). The breeder feed group showed relatively higher heritability only for feed conversion ratio (0.3658±0.0933), with the lowest heritability for the remaining four traits, yet its G × E interaction effect was most pronounced for feed conversion ratio (0.2027±0.0265).

Table 6.

Estimation of genetic parameters in white-feathered broilers.

Trait	Environment	Heritability	SE of Heritability	Sire × Feeding Environment Interaction				Residual
				Variance Component	SE of Variance	Heritability	SE of Heritability	Variance Component	SE of Variance
Breeder Feed	45DW	0.2248	0.1336	173.9128	358.0352	0.0107	0.0220	12182.6005	1604.4068
	FCR	0.3658	0.0933	0.0011	0.0002	0.2027	0.0265	0.0023	0.0003
	MWT	0.3947	0.1333	1.4047	2.5792	0.0139	0.0253	59.7900	9.6110
	WG	0.2287	0.1114	0.0957	97201819758.2184	0.0000	8600670.8187	6326.3483	911.3644
	FI	0.3789	0.1212	0.0000	262.6641	0.0000	0.0226	7147.9986	991.7740
Low-Energy Broiler Feed	45DW	0.4307	0.0460	3049.8110	580.6270	0.1023	0.0186	13922.7750	904.6323
	FCR	0.3730	0.0430	0.0013	0.0002	0.1798	0.0285	0.0032	0.0002
	MWT	0.4439	0.0450	15.2417	2.8303	0.1075	0.0190	63.5800	4.1977
	WG	0.4096	0.0460	2358.5563	452.5397	0.1100	0.0200	10298.9869	645.9696
	FI	0.3621	0.0426	7297.6889	1258.1723	0.2392	0.0363	12163.4412	789.3737
High-Energy Broiler Feed	45DW	0.4034	0.0666	5874.9354	1838.6521	0.1781	0.0522	13801.9686	1406.3369
	FCR	0.3593	0.0677	0.0006	0.0002	0.1118	0.0289	0.0028	0.0002
	MWT	0.4067	0.0570	29.0028	7.5726	0.1827	0.0444	65.1789	5.7502
	WG	0.3833	0.0576	5004.6853	1232.3529	0.2107	0.0476	9643.3068	862.2244
	FI	0.4158	0.0603	6233.0690	1592.3121	0.1676	0.0401	15500.0405	1448.4280
All Environments	45DW	0.4142	0.0362	4549.2705	959.2931	0.1426	0.0284	14135.7122	676.1510
	FCR	0.3847	0.0389	0.0012	0.0002	0.1425	0.0283	0.0038	0.0002
	MWT	0.4264	0.0334	19.3662	3.8566	0.1358	0.0256	62.4475	2.7487
	WG	0.3963	0.0368	3728.9095	788.7332	0.1546	0.0306	10832.4996	511.8374
	FI	0.4031	0.0366	3620.9114	752.2894	0.1190	0.0236	14539.3754	661.2979

The pooled multi-environment analysis revealed that heritability estimates for all five traits fell between the extremes observed in single environments, with generally reduced standard errors. However, the across-environment G × E interaction heritability (0.1190-0.1546) are lower than those in certain single environments, such as 45-day weight, metabolic weight, and feed intake in the low-energy group, and feed conversion ratio in the breeder feed group. This "averaging" approach may obscure genetic characteristic differences across feeding environments: the breeder feed group showed relatively constrained genetic potential due to lower nutritional levels, while the low-energy feed group demonstrated higher genetic sensitivity and environmental interaction intensity for growth traits. Therefore, although multi-environment analysis improves estimation precision, it may not fully capture genetic potential under specific nutritional conditions, suggesting that selection strategies should account for actual feed condition variations, and differentiated genetic evaluation methods tailored to specific feed types may be more appropriate.

Analysis of genotype-environment (G × E) interaction effects in white-feathered broilers

Table 7 shows the genetic correlations and G × E interaction effects of five core growth traits (45-day weight [45DW], feed conversion ratio [FCR], weight gain [WG], mid-term metabolic weight [MWT], and feed intake [FI]) of broilers under three feed environment combinations (breeder feed × low-energy broiler feed, breeder feed × high-energy broiler feed, and low-energy broiler feed × high-energy broiler feed).

Table 7.

Analysis of genotype-environment (G × E) interaction effects in white-feathered broilers.

	Environment
Trait	Breeder Feed *Low-Energy Broiler Feed	Breeder Feed *High-Energy Broiler Feed	Low-Energy Broiler Feed* High-Energy Broiler Feed
	Genetic Correlation (SE)	Genetic Correlation (SE)	Genetic Correlation (SE)
45DW	0.8973 (0.3113)##	0.5259 (0.3821)*	0.9633 (0.1356)##
FCR	0.8471 (0.3066)##	0.7288 (0.3757)*	0.7803 (0.2674)##*
WG	0.6265 (0.3024)#*	0.3668 (0.3821)*	0.7807 (0.1701)##*
MWT	0.8138 (0.2038)##	0.7157 (0.2549)##*	0.9121 (0.1046)##
FI	0.1942 (0.3683)*	0.5084 (0.4224)*	0.5799 (0.1701)##*

Under different feed environment combinations, the genetic correlations of the five growth traits of broilers showed obvious differences. The genetic correlations between high-energy broiler feed and low-energy broiler feed were the most significant overall: the genetic correlation coefficients of 45DW (0.9633) and MWT (0.9121) were the highest and reached an extremely significant level, while FCR, WG, and FI also showed significant or extremely significant genetic correlations. The genetic correlations between breeder feed and low-energy broiler feed were also prominent: the genetic correlation coefficients of 45DW (0.8973), FCR (0.8471), and MWT (0.8138) all exceeded 0.8 and reached an extremely significant level, and WG reached a significant level. In contrast, the overall genetic correlations between breeder feed and high-energy broiler feed were slightly lower: only MWT (0.7157) reached an extremely significant level, while the other traits reached a significant level, and their correlation coefficients were generally lower than those of the previous two feed combinations. The higher genetic correlation between low-energy and high-energy broiler feed environments suggests that the genetic basis of growth traits is relatively similar between these two commercial feeding environments, whereas the lower genetic correlation between breeder feed and another feeds indicates a greater difference in genetic regulation mechanisms under different nutritional levels.

G × E interaction effects were widely present in the combinations of different feed environments and broiler growth traits. All five detected growth traits showed G × E effects in at least one feed combination. Among them, FI exhibited the most stable G × E effect, which was present in all three comparisons (breeder feed vs. low-energy broiler feed, breeder feed vs. high-energy broiler feed, and high-energy broiler feed vs. low-energy broiler feed). 45DW, FCR, WG, and MWT showed G × E effects in two feed combinations, indicating that the genetic expression of these traits is significantly regulated by changes in feed environments, and different traits have different responses to feed environments. The widespread existence of G × E interaction suggests that direct selection in a single environment may lead to biased prediction of genetic performance in other environments. Therefore, the development of environment-specific breeding programs or multi-environment joint selection is necessary to improve the accuracy and adaptability of broiler genetic selection.

Discussion

Phenotype quality control and description

In this study, the growth trait data records of white-feathered broiler half-sib families were systematically sorted and strictly quality-controlled. Most of the phenotypic data passed the quality control, indicating that few families had outliers. In addition, after quality control of the data from each family, the data retention rate under the breeder feed environment was significantly lower than that under the other two feed environments (Table 2). In this study, the step with the largest amount of data screened out was Step (3) (removing data with abnormal phenotypic distribution in each family), which indicated that the growth traits of most families failed to meet the quality control requirements. Taking the growth trait phenotypes under the breeder feed environment as an example, abnormal data records may be due to the small number of individuals in this environment and the insufficient number of individuals in certain half-sib families under this environment. The quality control process in this study referred to previous studies on dairy cow reproductive traits (Liu et al., 2017; Shi et al., 2021). Although the research objects are different, previous studies have clearly pointed out that as a key link to ensure the accuracy of trait genetic evaluation, the collation of original data itself poses certain challenges.

Before conducting genetic evaluation of growth traits, it is still necessary to perform detailed and rigorous quality control on the original data to avoid deviations in evaluation results caused by data recording problems.

This study evaluated the growth traits of white-feathered broilers raised under three feeding environments with different energy and protein levels. The values of different traits under different energy and protein level environments were generally close, but there were differences due to the variations in energy and protein levels. In this study, significant differences were observed in feed intake (FEEDW), feed conversion ratio (FCR), and 45-day weight (45DW) among the three dietary treatments. Specifically, birds fed Low-Energy Broiler Feed exhibited the highest FEEDW and 45DW, while those fed High-Energy Broiler Feed showed the lowest FCR. Although the feed intake of white-feathered broilers under the low-energy broiler feed environment was higher than that under the breeder feed environment, it is speculated that this result may be related to the higher weight gain of white-feathered broilers in this feeding environment. An imbalance in the ratio of dietary energy to protein can inhibit the growth and development of broilers. Adjusting this ratio or adding feed additives can improve the utilization rate of nutrients, thereby enhancing broiler growth performance (Khwatenge et al., 2020). There is an interaction between dietary energy and protein levels, which can significantly affect the average daily gain, average daily feed intake, and feed conversion ratio of broiler chicks (Abdel-Hafeez et al., 2016; Loyiso et al., 2023). In the other study, moderately increasing dietary energy density improves feed conversion and sustains energy intake under heat stress, matching it with optimal protein levels supports broiler growth while controlling costs and avoiding the negative effects of excessive protein (Yunana et al., 2019). On the other hand, the 45-day weight (45DW) of white-feathered broilers under the low-energy broiler feed environment was higher than that under the high-energy broiler feed environment. Yu et al. (2024) found that increasing the crude protein content in the diet can significantly improve the feed conversion rate and crude protein utilization rate of broilers. Long-term exposure to an environment with imbalanced energy and protein (such as excessive energy and insufficient protein) may still increase the intestinal burden of broilers and indirectly reduce the uniformity of group growth. In addition, mid-term metabolic weight (MWT), as a key indicator reflecting the physiological state of the organism, may be more sensitive to energy and protein levels than intuitive traits such as body weight and feed intake. These findings confirm that the nutritional environment dominated by dietary energy and protein levels is an important exogenous factor affecting broiler growth and development, and the differential responses of growth traits to different environments reflect the complexity of nutrient regulation mechanisms, which provides a theoretical reference for the rational matching of feed formulas and precision feeding in actual production; meanwhile, the significant environmental effects revealed also remind us that in broiler genetic selection and breeding, the interaction between genetic background and nutritional environment must be fully considered to achieve the coordinated improvement of genetic potential and production efficiency. Future studies should conduct more refined physiological indicator monitoring to further explore the in-depth impact of such environmental differences on broiler growth performance.

Influencing factors of growth traits in white-feathered broilers

Through single-environment and multi-environment analysis of variance, this study found that the family effect had an extremely significant impact on all growth traits of broilers in both single and multi-environment analyses. Under the three single feeding environments, the F-values of the family effect on each trait varied but showed consistent significance; in the multi-environment combined analysis, the F-values of the family effect on traits ranged from 2.564 to 2.800, still maintaining extreme significance. This indicates that family (genetic background) is a stable genetic factor affecting traits, with the potential to improve traits through family selection.

The regulatory role of environment (feed type) on traits is prominent. In the multi-environment analysis, the main effect of environment on all traits was extremely significant, and the F-values were much higher than those of other effects; in single-environment analysis, although there was no statistical analysis of the main environmental effect, the batch effect was extremely significant for most traits. Combining the results of single-environment and multi-environment analysis of variance, the batch effect varied with the environment and was only significant for some traits in low-energy and high-energy broiler feed environments. This suggests that batch differences mainly stem from non-nutritional environmental fluctuations, such as temperature, humidity, light, feeding methods, and density (Nielsen, 2012). In summary, the environment (especially feed type) is a core external factor affecting traits.

The heritability estimates for all traits in white-feathered broilers ranged from 0.2248 to 0.4439, with 45-day body weight ranging from 0.2248 to 0.4307 and weight gain from 0.2287 to 0.4096. Aggrey (2002) reported corresponding trait heritability of 0.45 to 0.51; this discrepancy is primarily related to data structure, as this study only included data from male chickens for feed conversion ratio determination, and factors such as population size and the single-generation design may also have contributed to the lower estimates. Mid-term metabolic weight ranged from 0.3947 to 0.4439, feed conversion ratio from 0.3593 to 0.3730, and feed intake from 0.3621 to 0.4158. The heritability estimates for FI and FCR were close to but lower than those reported by Aggrey et al. (2010), Sell-Kubiak et al. (2017) (0.45 to 0.49), possibly due to greater environmental influences during the feed conversion ratio measurement process. G × E interaction heritability ranged from 0.0107 to 0.2392. The low-energy feed group showed the highest heritability for 45-day body weight, metabolic weight, and daily gain, with G × E interaction heritability for these three traits also higher than the other two groups; additionally, feed intake showed the highest G × E interaction heritability in this group, indicating that this feed environment provides better stability and potential for genetic expression of growth traits. The high-energy feed group performed best in feed intake heritability, followed by daily gain and feed conversion ratio. The breeder feed group showed relatively higher heritability only for feed conversion ratio, with the lowest heritability for the remaining four traits, and extremely weak G × E interaction heritability for 45-day body weight and metabolic weight, suggesting constrained genetic potential under this nutritional level.

The pooled multi-environment analysis revealed that heritability estimates for all five traits fell within the range observed in single environments, with generally reduced standard errors, indicating that cross-environment analysis improved parameter estimation precision. However, the across-environment G × E interaction heritability was significantly lower than single-environment levels for 45-day body weight, metabolic weight, and feed intake in the low-energy group, and feed conversion ratio in the breeder feed group. Although this "averaging" approach yields robust overall estimates, it obscures genetic characteristic differences across feed environments and may not fully capture genetic potential under specific nutritional conditions. These results suggest that feed energy and protein levels not only significantly affect heritability levels of broiler growth traits, but more critically, alter the intensity of G × E interaction effects, emphasizing the importance of considering feed environment specificity in genetic evaluation for broiler breeding.

The family × environment interaction effect clearly existed in the multi-environment analysis, with a significant impact on all traits, among which FCR and FI showed extreme significance; in single-environment analysis, there was no direct statistical analysis of the interaction effect, but some family × batch interaction effects were significant, indirectly indicating the potential for genotype-environment interaction in white-feathered broilers. This indicates that the genetic effect of families is environment-dependent, and the trait expression of the same genetic background varies under different feed environments. This is a key finding of this study, suggesting that the growth traits of white-feathered broilers are jointly regulated by three factors: genes, environment, and gene-environment interaction. In practice, it is necessary to first optimize the feed environment to exert the advantages of external regulation, then tap the genetic potential through family selection, and at the same time pay attention to the interaction between genetics and environment, and carry out breeding matching the target environment, so as to efficiently improve the growth performance of broilers. Collectively, these findings highlight that ignoring genotype-environment interaction may lead to inaccurate prediction of breeding values and unsatisfactory selection progress; thus, establishing environment-specific genetic evaluation systems and implementing targeted selection strategies according to different nutritional conditions in future broiler genetic improvement programs will help fully release the genetic potential of populations and enhance the stability and efficiency of commercial production.

Genotype-environment (G × E) interaction of growth traits in white-feathered broilers

In the variance components analysis, the genotype-environment (G × E) interaction effect had an extremely significant impact on all growth traits. On the one hand, feeding environments with different energy and protein levels directly changed broiler growth performance, confirming the regulatory role of nutritional supply on growth traits; on the other hand, the significant G × E effect indicates that different families have varying adaptability to the environment. The results of the two-trait model also showed that G × E effects are widely present, revealing the important regulatory role of feed environments on the genetic expression of broiler growth traits. Combining the results of variance components analysis and the two-trait model, it can be concluded that the G × E interaction effect not only exists in the regulation of growth traits in white-feathered broilers but also has a significant impact on growth traits, becoming one of the key variables affecting broiler growth performance. In previous studies, de Kinderen et al. (2023) found that the G × E interaction effect of body weight traits in five chicken breeds was extremely significant across all tested strain and region combinations; in another study (de Kinderen et al., 2020), they reported the presence of G × E interaction for egg production in five dual-purpose chicken breeds, with significant differences in the effect of this interaction among different regions. Erdem and Savaş (2021) found that changes in the rearing environment led to variations in the response of laying hen genotypes to parasite infestation during the growth stage.

The G × E effects of growth traits detected in this study have important guiding significance for the practical production of the white-feathered broiler industry. In the genetic breeding process, excellent families cannot be selected solely based on trait performance in a single environment. Instead, it is necessary to consider the characteristics of energy and protein supply in the feed of the target breeding area and prioritize the selection of families with stronger adaptability to specific environments to ensure the stable expression of genetic potential. In actual breeding practices, it is necessary to targetedly optimize the energy and protein ratio in the feed according to the genetic background of the broiler families raised. Through precise matching of genetics and environment, the impact of adverse interaction effects can be minimized, and ultimately the efficient improvement of broiler growth performance can be achieved. These findings underscore that rational utilization of G × E interaction is conducive to enhancing the accuracy and efficiency of genetic evaluation, and provide a scientific basis for developing environment-specific breeding strategies and precision feeding systems in commercial white-feathered broiler production.

Conclusions

This study focused on white-feathered broilers reared under different breeding environments, conducting genetic analysis and G × E effect detection on core growth traits and feed efficiency traits such as weight gain and feed conversion ratio. The results showed that abnormal phenotypic values of traits can be eliminated through precise quality control. The genetic performance of the same growth trait exhibited heterogeneity under different breeding environments, and G × E effects were widely present in the growth traits and feed efficiency traits of white-feathered broilers across different breeding environments. These findings contribute to our understanding of the complex genetic architecture underlying broiler growth traits and highlight the critical importance of accounting for environmental heterogeneity in genetic improvement programs. The identification of significant G × E interactions suggests that genetic selection strategies optimized for specific environments may not universally translate to improved performance across all production systems. This has profound implications for breeding program design, indicating that environment-specific breeding objectives or the incorporation of G × E effects into selection indices may be necessary to maximize genetic progress. Furthermore, our results underscore the need for comprehensive environmental characterization in commercial breeding operations to ensure that selection decisions are made within the appropriate environmental context. Future research should explore the molecular mechanisms driving these G × E interactions and develop predictive models that can forecast trait performance across diverse environmental conditions, ultimately facilitating more robust and resilient genetic improvement strategies for the global poultry industry.

Ethical statement

All experimental procedures with broilers were performed according to the Guidelines for Experimental Animals established by the Ministry of Science and Technology (Beijing, China). Ethical approval on animal survival was given by the animal welfare and ethics committee of the Institute of Animal Sciences (IAS), Chinese Academy of Agricultural Sciences (CAAS, Beijing, China) with the following reference number: IAS2022-37.

CRediT authorship contribution statement

Wenlong Zhao: Writing – original draft. Fan Ying: Resources, Data curation. Dan Zhu: Validation, Data curation. Jianhui LI: Validation, Supervision. Jie Wen: Project administration, Funding acquisition. Guiping Zhao: Writing – review & editing, Investigation, Funding acquisition. Bingxing An: Writing – review & editing, Writing – original draft, Project administration.

Disclosures

The authors declare no conflict of interest.