Genome-wide association study of durations in zones of an aviary for laying hens

TL;DR

This study uses genetic data to identify genes linked to how laying hens spend time in different areas of an aviary, aiming to improve commercial egg production.

Contribution

The study identifies candidate genes associated with aviary zone usage in laying hens, offering new insights into the genetic basis of their behavior.

Findings

13 significant SNPs were identified across different periods in the five aviary zones.

Significant SNPs were mapped to 64 genes, 22 of which were previously annotated and related to feeding, movement, and immune function.

Behavioral traits in aviaries may provide insight into hens' health and performance, though further validation is needed.

Abstract

Identifying how birds interact with their environment and selecting birds that are more suitable for specific housing systems could be beneficial for commercial egg production. Previous research has indicated heritability in behavior traits such as duration spent in different zones of an aviary barn, however, the underlying genetic mechanisms affecting aviary usage remain unknown. The objectives of this study were to further explore durations in different zones of an aviary through identifying potential candidate genes associated with these behaviors, to provide insight into the biological processes involved with laying hen behavior and to offer potential candidate genes for selection. Using tracking data and genotypes from 1098 white line pure line crosses (Hendrix Genetics), a genome wide association study (GWAS) was performed. From tracking data, durations spent in each of five designated zones of a semi-commercial aviary (wintergarden, litter, lower tier, nestbox tier, and upper tier) across days 4 to 290 in a production barn were investigated. Hens were genotyped with a proprietary 60K SNP array. After data filtering, 252,154 days of records were used, where phenotypes were recorded as average time spent in a specific zone over the days 4-290, 4-50, 51-100, 101-150, 151-200, and 251-250 of production. Analysis was carried out using a univariate mixed linear model with number of visits to the zone and pen number fitted as fixed effects, the top two principal components fitted as covariates, and the polygenic effects fitted as random. Of the duration in the five zones, 13 significant SNPs were identified across the different periods using chromosome-wise Bonferroni corrections (α = 0.05). The significant SNPs were mapped to within 100,000 base pairs of 64 genes, 22 of which were previously annotated, with several related to feeding behavior, movement, and immune function. The findings of this study suggest that behavioral traits within an aviary could be used to provide important insight into hens’ health and performance within their housing system, but further validation is needed.

Figures

Side view of aviary housing system, showing the wintergarden (1), litter (2), lower tier (3), nest tier (4), and upper tier (5) zones.Fig 1: dummy alt text

Manhattan plots for average duration in the wintergarden for days 24-290 (A), 24-50 (B), 51-100 (C), 101-150 (D), 151-200 (E), 201-250 (F), and 250-290 (G) of production, with the significance threshold set to chromosome-wise Bonferroni correction. Significant SNPs are indicated in blue. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).Fig 2: dummy alt textFig. 3Manhattan plots for average duration in the litter for days 4-290 (A),4-50 (B), 51-100 (C), 101-150 (D), 151-200 (E), 201-250 (F), and 250-290 (G) of prod

Full text

Introduction

In 2012, the European Union (EU) prohibited the use of non-enriched cage housing systems for laying hens, leaving enriched cages or alternative systems as the approved options. As a result, approximately 62% of laying hens in the EU are housed in cage-free systems as of 2024 (European Commission, 2025). A common cage-free housing system is an aviary, in which birds are provided a muti-tier structure with stacked platforms, allowing for greater use of space than single-tier or floor housing systems (Sosnówka-Czajka et al., 2021). The aviary system allows birds to exhibit natural behaviors often associated with improved welfare, such as perching, dust bathing, foraging, and nesting (Weeks and Nicol, 2006). However, cage-free housing systems generally use litter which can expose birds to higher levels of bacteria and fungi (Rodenburg et al., 2005), and birds in cage-free systems tend to experience more bone fractures compared to birds housed in caged systems (Wilkins et al., 2011). Individual birds’ response to these enrichments and potential challenges associated with aviary systems are often difficult to determine due to the large group size. With advances in sensory technology, the ability to track birds throughout cage free housing systems has become possible (Siegford et al., 2016). Sensor technologies allow for high resolution tracking and behavioral monitoring, which can provide novel phenotypes for individuals in large group housing and a more nuanced view into how they use the different features of the housing system.

Rufener et al. (2018) demonstrated the ability to collect individual behavior metrics within a commercial aviary system, highlighting the individuality of birds in large groups. Individual behavior patterns have been identified in hens and found to be repeatable throughout the production period (Montalcini et al., 2023a). In combination with the behavioral differences observed between hens of different hybrids (Odén et al., 2002), these findings suggest that genetic factors contribute to the different behaviors in the birds. Behavior is often an expression of an animal’s ability to cope with challenges in an effort to maintain homeostasis (Koolhaas et al., 1997), which could be used as an indicator of the bird’s suitability to the environment. As housing systems change, breeders often adapt selection goals to match the environment (Craig, 1982), and the use of behavior traits for selection could be advantageous to the shift to cage free housing. Heritability of behavior traits, such as laying rate in the nest (0.13 (0.02)), mean laying duration (0.54 (0.06)), and percentage of nests used (0.24 (0.04)) have been estimated (Becot et al., 2021). Another behavior trait of interest is spatial usage, which refers to the proportion of time spent by an individual at a particular location, and is often heterogeneous (Matthiopoulos, 2003). An investigation of spatial usage using durations spent in the different areas of an aviary resulted in heritability estimates ranging from 0.05 (0.01) to 0.28 (0.03) depending on the area (Makanjuola et al., 2024). If selection for behavioral traits, such as zone duration within an aviary, is of interest for breeders, identification of causative genes associated would be essential, due to the difficult nature and high cost of collecting phenotypes.

Due to the increased availability of high-density single nucleotide polymorphism (SNP) arrays, the use of genome wide association studies (GWAS) has been used to identify genetic associations between economically important phenotypes in livestock species and SNPs. In laying hens specifically, GWAS has been used to identify quantitative trait loci (QTL) for various production, health, physiological, reproductive, and other exterior traits, including conformation and behavior traits (Wolc et al., 2014, 2019; Raymond et al., 2018). There are relatively few studies that have identified QTLs associated with behavior traits in hens, and little is known about potential genes related to the differences in the bird’s behavior within housing systems. Therefore, the objectives of this study were to identify potential genes associated with durations spent in the different zones of an aviary, to provide insight into potential biological processes involved in different behavior traits of laying hens and to offer potential candidate genes for selection.

Materials and methods

Ethical statement

The experiment was approved by the Veterinary Office of the Canton of Bern (BE4/2021) and met all cantonal and federal regulations for the treatment of study animals.

Animals and management

The current analysis extended from the investigation by Makanjuola et al. (2024) and used the same animals. Hens were provided by Hendrix Genetics and were parental crosses of pure line matings from lines for the Dekalb white hens that were hatched in June 2021 and reared at the Aviforum facility (Zolikofen, Switzerland). For the study, 1124 pullets from 25 sires (based on available chicks and extremes of a breeding index incorporating feathering and survivability) were equipped with leg bands containing passive radio frequency identification (RFID) transponders (125 kHz) at approximately 16 weeks of age and then transferred to the on-site production barn at 18 weeks of age (WOA). The production barn was equipped with a Bolegg Terrace aviary divided into 20 pens (7 m length, 2.3 m width). Hens were housed in five of the pens, where they were stratified by sire and rearing pen into groups of 225 animals. The housing system was divided into five zones (Fig. 1); the upper, nest, and lower tiers, as well as the litter area, and from 21 WOA, a wintergarden area. Each tier was connected by a diagonal ramp, allowing the hens to transition without being required to jump or fly. Feedlines ran across the upper and lower tiers, and nipple drinkers were located on the upper and lower tiers, as well as in the wintergarden (drinkers in the wintergarden were non-functional during the winter months). Management procedures (e.g., feeding, vaccination, lighting program, etc.) were based on the standard guidelines for the Dekalb White hybrid. The maximum duration of light hens were provided was 14 h (from 03:00 to 17:00); access to the wintergarden was between 10:00 and 16:00.Fig. 1. Side view of aviary housing system, showing the wintergarden (1), litter (2), lower tier (3), nest tier (4), and upper tier (5) zones.Fig 1: dummy alt text

Tracking system

The housing system was equipped with thirty-two 12-field SPEED antennas (75 × 35 cm) from Gantner Pigeon Systems GmbH that functioned with the hens’ RFID transponders. The antennas were located on the edges of all tiers for both sides of the aviary, as well as in two rows on the littered floor, and on both sides of the pop-hole leading to the wintergarden (Fig. 1). The system registered the hens’ location as they traveled between the different zones (i.e., the upper, nest, and lower tiers; litter area, wintergarden), and the durations spent in each of the defined zones. Due to the design of the current tracking system, hen to hen interactions were unable to be detected beyond movement associations (Perinot et al., 2025), and were not considered for the current analysis. The tracking system was validated previously by Gebhardt-Henrich et al. (2023). Hens’ transponders were registered whenever they came within 10 cm of an antenna, with each record containing a time stamp, antenna ID, and the hen’s unique RFID transponder code. During each daily period of registration, if the hen was not registered for 15 seconds, it was considered a non-detection (ND), and unless it was next registered in the same zone as prior to the ND, the record was removed (Perinot et al., 2025).

The activity monitoring of hens began four days after their arrival in the production barn to avoid the RFID system interfering with another study in different pens of the barn. Records were collected until day 290 of production (approximately 59 WOA), with 68 days of records being removed due to unusual disturbances (e.g., health assessments, other irregular activity in the barn). Hens were tracked during hours of light; thus, records were restricted to less than 54,000 seconds. The resulting data file contained 266,554 total records, referred to as hen-days, representing daily records for individual hens. Hen-days without at least half a day (27,000 seconds) of records were removed, filtering out 10,767 hen-days. Additionally, 14 hens that died during the study, and four hens that lost their RFID transponders were removed, as well as two hens which had less than 10 days of sufficient records. Only hens that had genetic information were included in the analysis, leaving 1098 hens with 252,154 hen-days. Phenotypes were analyzed as average duration in different zones of the aviary over the entire period from days 4 to 290 in production (18-59 WOA). Additionally, zone durations were averaged in segments of approximately fifty-days to identify genetic factors impacting behavior at different points throughout the hen’s production, while including sufficient days to produce a meaningful average. Although exploratory, it is believed that there is continual behavioral variation over time due to hens adapting to their environment (Montalcini et al., 2023a), as well as changes in production and health status that are important to consider. The roughly fifty-day segments were: 4 to 50, 51 to 100, 101 to 150, 151 to 200, 201 to 250, and 251 to 290 of production. Investigation into time spent in each tier allows for the exploration of how the individual hens utilize the different features of the housing system, where each location provides the hens with access to varied resources. The upper tier provides the hens with access to feed and water, as well as a place to perch. The nest tier contains perches, and nestboxes where the hens can lay their eggs or a dark place to avoid other hens. The lower tier provides access to feed and water. The litter area allows the hens to exhibit natural behaviors like dustbathing and scratching, and the wintergarden area provides the hens with water, litter for the hens to dustbathe and scratch in, perches, and exposes them to the natural elements of being outside.

Genotype data

At 30 WOA all hens had blood samples collected from the brachial wing vein from which samples underwent genotyping using a proprietary 60K SNP panel (Illumina Inc. 60K). Genotypic quality control was performed with all non-autosomal SNP and uncharacterized SNP markers removed. Additionally, SNP markers with a call rate less than 0.95, minor allele frequency (MAF) less than 0.05, and a difference of more than 0.15 between observed and expected heterozygosity frequency were removed. After quality control, a total of 27,045 SNP markers were retained for further analyses.

Statistical analysis

To account for population stratification a principal component (PC) analysis was conducted using PLINK 1.9 software (Chang et al., 2015) with the indep-pairwise function. The pruning window size was 25 markers, the step size was five markers, and the r^2^ threshold of 0.2. The remaining 2058 independent markers were used to derive the top two PCs, accounting for 14.4% and 10.1% of the variation. Using ASReml 4.2 (Gilmour et al., 2021), univariate mixed models for average durations in each of the different zones across the various time periods were done with the following equation:

$$[eqn]$$

where y is the vector of phenotypes, b is the vector of covariates and fixed effects for the top 2 PCs, number of visits to the zone, and pen that the hen was housed in, a is the vector of random additive genetic effect, e is the random error term, X, and Z are incidence matrices for the fixed effects and random genetic effect, respectively.

Investigation into the association of SNPs and duration in different aviary zones used the following univariate linear mixed model using GCTA software by Yang et al. (2011):

$$[eqn]$$

where y is a n x 1 vector of phenotypic values for n individuals, W is a n x c matrix of covariates for the top 2 PCs, and the fixed effects of number of visits to the zone, and the pen the hen was housed in. $[eqn]$ is a c x 1 vector of the corresponding coefficients, x is a n x 1 vector of marker genotypes at the locus being tested, $[eqn]$ is the effect size of the marker, u is a n x 1 vector of random polygenic effects with a covariance structure as $[eqn]$ where K is the genetic relationship matrix derived from SNP markers and $[eqn]$ is the polygenic additive variance, and e is a n x 1 vector of random residuals with $[eqn]$ where I is an n x n identity matrix, and $[eqn]$ is the residual variance component. Inclusion of the dominance genetic matrix to account for the proportion of phenotypic variance attributed to the dominance genetic variation was considered, but when included, did not have a statistically significant impact on the results, and thus was removed from the analysis. The following equation, using the estimated size effect of the SNP markers, combined with the allele frequencies, estimated the genetic variance explained by each SNP:

$$[eqn]$$

The use of quantile-quantile (Q-Q) plots was done to assess population stratification in combination with the genomic inflation factor (λ). Genomic inflation factors were calculated by dividing the observed median of the Chi-squared statistic for the p-values obtained from the GWAS by the expected median Chi-squared statistic of 0.455 (1 df test). To determine significant SNPs from the GWAS, chromosome-wise Bonferroni correction (α = 0.05) was used as a threshold. Q-Q and Manhattan plots were generated using the package qqman (Turner, 2018) in R studio (version 4.3.1) where the negative logarithms 10 for the p-values (-log_10_p) of each SNP were displayed in the Manhattan plots.

Assignment of significant SNPs to genes

The Gallus Gallus genome assembly bGalGal1.pat.whiteleghornlayer.GRCg7w_WZ was used to assign significant SNPs to genes (Dyer et al., 2025). When investigating chickens, linkage disequilibrium (LD) has been reported to range from 25,000-150,000 base pairs (bp) (Rao et al., 2008; Qanbari et al., 2010). For the purpose of this study, SNPs were assigned to genes if they were located within the genomic sequence of an annotated gene, or within 100,000 bp of a gene, both up and downstream. The used distance is expected to capture proximal regulatory regions and other functional sites that may lie in close proximity to the gene, such as promoter regions.

Results and discussion

A summary of the data for average duration in each of the zones of the aviary is shown in Table 1, where large variation in time spent in the different zones by hens can be seen within and across the different time periods investigated. Differences in the average durations between this and the related study by Makanjuola et al. (2024) were found, which can be accounted for by the altered data filtering methods, particularly the inclusion of non-detections. On average, hens spent the most time in the litter area and the least in the wintergarden, which is to be expected as the litter area is the largest zone, and use of the wintergarden was restricted to certain periods of the day and only days where temperature was above freezing. Wintergarden durations had the largest coefficient of variation, showing the most variability in the time hens would spend in the outside area. Furthermore, 42 of the hens did not use the wintergarden area at all (average duration across the entire period of study under one minute).Table 1. Descriptive statistics for hens’ duration (in seconds) in the wintergarden, litter, lower tier, nest tier, and upper tier of an aviary barn. Durations were looked at as average over the entire period of study (days 4–290), as well as in 50-day segments during the production period.Table 1: dummy alt textDuration in secondsZoneTime periodMeanStandard deviationMinimumMaximumCoefficient of variationWintergardenDays 24–29017711791011,3091.01Days 24–5015661526010,1170.97Days 51-10012521458099471.16Days 101–1509641204010,0021.25Days 151–20012041540099661.28Days 201–25025782788016,0561.08Days 250–29028463105016,4651.09LitterDays 4–29018,9735508101932,5370.29Days 4–5015,255721714934,7490.47Days 51-10019,430631455535,6930.32Days 101–15019,6846403102038,5140.33Days 151–20020,300633754734,6760.31Days 201–25019,604615129535,8470.31Days 250–29019,384632725634,1510.33Lower tierDays 4–2906980337510118,9060.48Days 4–50406329081220,8950.72Days 51-100713541536327,6390.58Days 101–150754041834423,0960.55Days 151–200776741555024,6060.53Days 201–250772239737225,4290.51Days 250–290757438905128,4830.51Nest tierDays 4–2904146183361018,6190.44Days 4–503753224853630,5820.60Days 51-1003530172681517,5420.49Days 101–1504032225045822,6960.56Days 151–2004426267635629,9950.60Days 201–250459826524631,4620.58Days 250–2904577244655339,9440.53Upper tierDays 4–29010,394667330040,6480.64Days 4–5012,20979424440,6000.65Days 51-10010,04678593544,0130.78Days 101–15010,891788616942,5210.72Days 151–20010,481752219342,7340.72Days 201–2509549707217841,8920.74Days 250–2909480735113543,1750.78

The Q-Q plots for the seven time periods of each of the five different zone duration traits N are provided in the supplementary material (Figs. 7-11) and show a strong relationship between the observed and expected values. The genomic inflation factor ranged from 0.92 to 1.02 for the different duration traits, suggesting that lambda is less than or equal to one, demonstrating that inclusion of the top two PCs accounts for the population stratification (Hinrichs et al., 2009). Variance and heritability estimates for the duration phenotypes are listed in Table 2, where heritability estimates showed variation across the different periods of investigation. Heritability estimates for zone durations ranged widely: 0.10-0.23 (0.04-0.05) for the wintergarden, 0.33-0.46 (0.05-0.05) for the litter, 0.20-0.32 (0.05-0.05) for the lower tier, 0.16-0.41 (0.04-0.06) for the nest tier, and 0.29-0.39 (0.05-0.05) for the upper tier. When compared to previous parameter estimations for durations in zones of an aviary by Makanjuola et al. (2024), the current estimates are higher. The change in heritabilities can partially be explained by the changes in the data filtering of the traits. Additionally, the increase in the proportion of additive genetic variance can be attributed to the use of one mean duration compared to repeated records, which generally result in lower additive genetic variance and higher phenotypic variance (Åkesson et al., 2008). Although the use of trait means reduces the power of the study, the investigation into the segments of approximately 50 days allowed for the identification of genetic factors that impact behavior at different points throughout production.Table 2. Estimated genetic parameters for hens’ duration in the wintergarden, litter, lower tier, nest tier, and upper tier of an aviary barn. Durations were looked at as average over the entire period of study (days 4–290), as well as in 50-day segments during the production period.Table 2: dummy alt textZoneTime periodGenetic variancePhenotypic varianceHeritability (%)SEWintergardenDays 24–2902.35E+051.00E+0623.424.95Days 24–508.65E+047.28E+0511.894.10Days 51-1005.66E+045.59E+0510.133.99Days 101–1504.62E+043.97E+0511.634.07Days 151–2001.12E+058.02E+0513.974.43Days 201–2506.56E+053.22E+0620.394.83Days 250–2908.53E+054.05E+0621.074.85LitterDays 4–2909.50E+062.08E+0745.604.97Days 4–508.64E+062.60E+0733.185.27Days 51-1001.18E+072.98E+0739.665.15Days 101–1501.24E+072.86E+0743.405.02Days 151–2001.12E+072.94E+0738.215.06Days 201–2501.02E+072.66E+0738.445.11Days 250–2909.43E+062.60E+0736.295.05Lower tierDays 4–2902.39E+067.40E+0632.365.30Days 4–507.20E+053.58E+0620.094.74Days 51-1002.27E+061.05E+0721.655.29Days 101–1502.89E+061.17E+0724.645.22Days 151–2003.37E+061.20E+0728.135.26Days 201–2503.06E+061.13E+0727.165.17Days 250–2902.78E+061.07E+0726.045.16Nest tierDays 4–2906.49E+052.03E+0631.915.11Days 4–505.33E+053.34E+0615.974.45Days 51-1008.98E+052.19E+0641.085.13Days 101–1501.12E+063.49E+0632.195.19Days 151–2001.06E+064.67E+0622.624.97Days 201–2501.00E+064.55E+0622.035.50Days 250–2906.45E+053.87E+0616.654.66Upper tierDays 4–2901.58E+074.05E+0738.905.20Days 4–501.86E+076.09E+0730.485.18Days 51-1001.62E+075.51E+0729.375.12Days 101–1502.07E+075.97E+0734.705.22Days 151–2001.98E+075.22E+0737.955.22Days 201–2501.65E+074.56E+0736.165.22Days 250–2901.88E+075.02E+0737.435.14

As shown in the Manhattan plots (Fig. 2, Fig. 3, Fig. 4, Fig. 5, Fig. 6), 13 SNPs were found to be above the chromosome-wise Bonferroni threshold across the five zones investigated (Table 3). The significant SNPs identified were across different timeframes for different zones, highlighting the variation in factors affecting behavior. The identified SNPs were found for average durations across the entire period of study for the upper tier, nest tier, and lower tier. When the durations were examined in the individual 50-day periods, significant SNPs were found for days 24-50 and 251-290 in the wintergarden, days 51-100 in the litter, days 4-50, 51-100, 101-150, 151-200, and 201-250 in the lower tier, and days 101-150 in the upper tier. The 13 unique SNPs identified for duration in the five zones have been mapped to within 100,000 bp of 64 genes (Table 4), of which 22 are annotated and characterized.Fig. 2. Manhattan plots for average duration in the wintergarden for days 24-290 (A), 24-50 (B), 51-100 (C), 101-150 (D), 151-200 (E), 201-250 (F), and 250-290 (G) of production, with the significance threshold set to chromosome-wise Bonferroni correction. Significant SNPs are indicated in blue. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).Fig 2: dummy alt textFig. 3Manhattan plots for average duration in the litter for days 4-290 (A),4-50 (B), 51-100 (C), 101-150 (D), 151-200 (E), 201-250 (F), and 250-290 (G) of production, with the significance threshold set to chromosome-wise Bonferroni correction. Significant SNPs are indicated in blue. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).Fig 3: dummy alt textFig. 4Manhattan plots for average duration in the lower tier for days 4-290 (A), 4-50 (B), 51-100 (C), 101-150 (D), 151-200 (E), 201-250 (F), and 250-290 (G) of production, with the significance threshold set to chromosome-wise Bonferroni correction. Significant SNPs are indicated in blue. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).Fig 4: dummy alt textFig. 5Manhattan plots for average duration in the nest tier for days 4-290 (A),4-50 (B), 51-100 (C), 101-150 (D), 151-200 (E), 201-250 (F), and 250-290 (G) of production, with the significance threshold set to chromosome-wise Bonferroni correction. Significant SNPs are indicated in blue. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).Fig 5: dummy alt textFig. 6Manhattan plots for average duration in the upper tier for days 4-290 (A),4-50 (B), 51-100 (C), 101-150 (D), 151-200 (E), 201-250 (F), and 250-290 (G) of production, with the significance threshold set to chromosome-wise Bonferroni correction. Significant SNPs are indicated in blue. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).Fig 6: dummy alt textTable 3List of SNPs below chromosome wise Bonferroni threshold for laying hens zone duration in an aviary, detected by GWAS.Table 3: dummy alt textTraitChromosomeSNP nameLocationFreqEstimated effectStandard errorP valueVariance% of additive varianceLitter duration days 51–1001HGC005350137,353,1910.31,726.24369.362.96E-061251559.9110.61Lower tier duration days 51–1001Gga_rs13771160111,396,5870.34811.35182.588.83E-06295440.0213.02Lower tier duration days 51–1001Gga_rs13923156111,416,0360.34811.35182.588.83E-06295440.0213.02Lower tier duration days 51–1001HGC004361111,380,1420.34807.59183.471.07E-05292708.0812.90Lower tier duration days 101–1501GGaluGA038117110,487,5090.44772.52175.481.07E-05294096.7110.18Nest tier duration days 4–2903HGC016855100,798,0780.26449.29105.352.00E-0577676.3111.97Wintergarden duration days 24-507HGC02607731,669,3940.07317.5475.362.51E-0513128.2815.18Lower tier duration days 4–507HGC_rs1359955930,888,7340.27490.5113.861.65E-0594840.6813.17Lower tier duration days 4–507HGC02603330,886,2510.28464.88113.053.92E-0587136.9312.10Upper tier duration days 4–29011Gga_rs156084806,847,0800.34-1377.87340.995.33E-05852058.355.39Upper tier duration days 101–15011Gga_rs156084806,847,0800.34-1682.64414.084.83E-051270677.288.04Wintergarden duration days 251-29017HGC0367327,595,1240.29493.37125.858.84E-05100237.8711.75Lower tier duration days 4–29027HGC0435391,547,3180.42540.45131.483.95E-05142304.405.95Lower tier duration days 101–15027HGC0435391,547,3180.42678.481664.37E-05224275.277.76Lower tier duration days 151–20027HGC0435391,547,3180.42649.79168.051.10E-04205709.026.10Lower tier duration days 201–25027HGC0435391,547,3180.42673.66163.063.60E-05221100.037.23Lower tier duration days 201–25027HGC0435381,544,5340.22-784.26206.831.50E-04211089.886.90Table 4Genes associated (within 100,000 base pairs) with the significant SNPs identified for zone durations in an aviary housing system in laying hens during production, and the SNPs location relative to the gene (U = upstream and D = downstream).Table 4: dummy alt textSNPChromosomeLocation (bp)Distance from geneGene nameHGC0053501137,353,191Within intronTFDP311,196UTMCO340,508DATP4B42,589UDCUN1D267,321DTMEM255B70,552UADPRHL191,316ULOC121109116Gga_rs13771160111,396,587Within intronMAOB29,655DNDP54,004UMAOA76,349DLOC112531615Gga_rs139231561111,416,036Within intronMAOB34,555UMAOA49,104DNDP82,377ULOC11253105095,798DLOC112531615HGC0043611111,380,14210,816UMAOB13,210DNDP59,904DLOC11253161570,449UMAOAGGaluGA0381171110,487,509Within intronPWP28,645DTRAPPC109,891UGATD3A22,359UVTCN1L49,737ULOC10174950350,798DLOC12441765356,576UICOSLG65,700DAGPAT3HGC0168553100,798,07856,183ULOC121110199HGC026077731,669,394within intronLRP1BHGC_rs13599559730,888,734within intronLOC11253276010,160USPOPL65,138UNXPH296,182DHNMTHGC026033730,886,251within intronLOC11253276012,643USPOPL67,621UNXPH293,699DHNMTGga_rs15608480116,847,080within intronZNF42354,068ULOC107054281HGC036732177,595,1241,221DLOC1211070898,420ULOC12441819313,797DCOL5A150,822UOLFM179,129ULOC121107061HGC043539271,547,3181,877ULOC1211121673,764DLOC1211121525,242DLOC1244182396,347ULOC1211121777,359DLOC12111217011,775DLOC12111216611,831ULOC12111217612,776ULOC12111214817,719ULOC12111215119,105DLOC12111214921,447ULOC12111217821,844DLOC12111218325,333DLOC12441823626,650DLOC12111216332,539ULOC12111217334,154DLOC12111213237,818DLOC12111213340,517DLOC12111217942,019ULOC12111218948,610DLOC12111214357,914ULOC12111216465,449DLOC12111213574,092ULOC12111219075,148ULOC12111215079,094DLOC12441824280,248ULOC12111217583,277ULOC12111218784,935DLOC12111216894,586DLOC12111218497,274ULOC121112182HGC043538271,544,534980DLOC1211121522,458DLOC1244182394,575DLOC1211121704,661ULOC1211121678,991DLOC1211121669,131ULOC12111217714,615ULOC12111217615,560ULOC12111214816,321DLOC12111214919,060DLOC12111218320,503ULOC12111215122,549DLOC12441823623,866DLOC12111216324,231ULOC12111217831,370DLOC12111213235,034DLOC12111213335,323ULOC12111217337,733DLOC12111217944,803ULOC12111218945,826DLOC12111214360,698ULOC12111216462,665DLOC12111213576,310DLOC12441824276,876ULOC12111219077,932ULOC12111215082,151DLOC12111216883,032ULOC12111217586,061ULOC12111218791,802DLOC12111218499,343DLOC121112129

Genetic variations of complex traits, such as behavior, are often the result of small effects from many loci (Sella and Barton, 2019), however, the current study found relatively large percentages of the additive genetic variance explained by the significant SNPs (Table 3). The high percentage of variance explained by the identified SNPs can be related to the inflation of additive variance from the use of phenotypic means for the traits, as an increase in the proportion of additive genetic variance often is associated with an increase in the variance explained by each SNP. Additionally, since the mixed linear model tests each SNP individually, the effect of SNPs in linkage disequilibrium can be included, increasing the estimate for variance explained by the SNP.

The SNP identified on chromosome 11 for upper tier duration during days 4-290 and 101-150 is located within an intron region of ZNF423. Zinc finger protein 423 (ZNF423) is a transcription factor that has been linked to controlling preadipocyte differentiation related to adipogenesis (Gupta et al., 2010; Matsubara et al., 2013). In chickens ZNF423 expression is connected to adipocyte differentiation and the development of adipose tissue, suggesting it as an important factor in chicken adipogenesis (Matsubara et al., 2013). Previous GWAS have identified ZNF423 to relate to different feed efficiency and performance traits for crossbred beef cattle, showing strong evidence for association with residual feed efficiency (Abo-Ismail et al., 2014). In laying hens, less feed efficient hens spend more time feed pecking (Braastad and Katle, 1989), which is a potential justification for the increase in duration in the upper tier, as it is one of the zones that provides the hens with feed. One possible reason that the SNP was identified for only the 101- 150-day period could be related to the hens’ production cycle. When body mass in laying hens has been investigated across different periods of production, there is an increase in mass from early to peak lay, and then remains relatively constant into late production (Van Goor et al., 2020), of which days 101-150 of production align with peak lay.

The top two SNPs identified for duration in the lower tier during days 51-100 are located within intron regions of the gene MAOB, and upstream of the gene MAOA. Both MAOB and MAOA are amine oxidase genes¸ which are part of several metabolic pathways, predominantly being involved in the metabolism of multiple different amino acids (Kanehisa et al., 2025). In poultry studies, MAOA has been found to be related to increased feather damage (Biscarini et al., 2010), and an upregulation of the monoamine oxidase gene has been found in high feather pecking line birds (Wysocki et al., 2013). Feather pecking is often related to feeding behavior in hens (Jensen et al., 2005; Kjaer and Bessei, 2013; Rodenburg et al., 2013), which is of interest for durations in the lower tier, as it is one of the two zones in the barn that provides the hens with feed. In humans and mice, MAOA and MAOB are part of the dopaminergic synapse pathway (Kanehisa et al., 2025), an important driver in motivated behaviors and reward processes (Spruijt et al., 2001; Moe et al., 2014). In chickens the dopaminergic synapse has been found to relate to appetitive behavior (Moe et al., 2014), supporting the idea that the additional time in the lower tier is linked to feeding. Tracking data has previously identified increases in the number of hens on the lower tier aligning with the arrival of fresh feed, supporting that lower tier usage relates to feeding behavior (Montalcini et al., 2023b). Although the current behavior metrics investigated are unable to determine the real time spent feeding, the more time spent on the lower tier equates with increased access to feed, highlighting the need to further explore the relationship between behavior and feed efficiency.

Additionally, the hens’ durations spent in the lower tier were found to be associated with SNPs located on chromosome 27, where 30 uncharacterized genes were within 100 kbp upstream and downstream of the identified SNP markers. Although information regarding the genes in the region is limited, a previous study investigating feeding behavior found a QTL related to feeding behavior and social interaction in a similar region of chromosome 27 (Schütz et al., 2002). The trait found to be associated with chromosome 27 was from a principal component analysis, where the higher scoring birds prioritized free feeding without social interaction, supporting the idea that hens are spending more time in the lower tier due to feeding. Furthering the connection between feeding behavior and the usage of specific zones, the SNP identified for lower tier duration from days 101-150 was located within an intron region of the gene PWP2 on chromosome 1. Periodic tryptophan protein 2 (PWP2) is part of the ribosome biogenesis pathway (Kanehisa et al., 2025), an important process influencing cell and organismal biology, including maintaining cellular homeostasis (Ni and Buszczak, 2023). Ribosome biogenesis is an asset for cellular protein production, a driver for cellular growth (Lempiäinen and Shore, 2009), and strongly activated during skeletal muscle growth (Chaillou et al., 2014). In poultry species, the PWP2 gene has been identified in GWAS for wing weight (Kanlisi et al., 2024) and bone mass (Han et al., 2024), both being traits that can be influenced by feed usage.

Durations in the wintergarden were found to be associated with SNPs during days 24-50 and days 251-290, located on chromosome 7 and 17, respectively. The SNP located on chromosome 7 is within an intron region of the gene low-density lipoprotein receptor-related protein 1B (LRP1B), which is related to muscle fiber development in chickens (Lv et al., 2019). The gene LRP1B has additionally been connected to abdominal fat content in broiler chickens in a GWAS (Zhang et al., 2012), and differential expression of LRP1B has been linked to proximal femoral head separation in broilers (De Oliveira Peixoto et al., 2019). The SNP on chromosome 17 is downstream of the gene collagen alpha-1(V) chain precursor (COL5A1), a component of the cytoskeleton in muscle cells pathway (Kanehisa et al., 2025). Upregulation of COL5A1 has been found in chickens with high intramuscular fat content (Cui et al., 2023), and in humans it is related to quadricep muscle-tendon stiffness in humans (Kirk et al., 2016). Both genes identified for durations in the wintergarden suggest a relationship with the mobility of the hens and the time they spend in the wintergarden, which is supported by the positive correlation between wintergarden presence and vertical travel distance within an aviary by Montalcini et al. (2023a).

The SNP identified to be associated with duration in the litter during days 51-100 of production was found within an intron region of the transcription factor Dp-3 (TFDP3). In humans, TFDP3 has been linked to several different cancers, as it is considered a master regulator of apoptosis (Tovar et al., 2015). Apoptosis is essential for maintaining homeostasis of the immune system (Hildeman et al., 2007), which is of interest for the litter, as it represents an area where the hens face more potential immune challenges. Access to litter increases birds exposure to infectious diseases, ectoparasitic, and has greater incidence of bacterial infections (Meyer-Kühling et al., 2007; Fossum et al., 2009). A difference in microbiota populations has been observed in hens relative to extremes of visits to the litter area (Cazals et al., 2026), highlighting the differences in bacterial load on birds spending more time in the litter. TFDP3 has been previously related to immune challenges in poultry species, and has been suggested to be part of the gene network for non-adaptive chickens for heat stress (Kuijpers, 2022). The potential connection between immune activity and duration in the litter suggests that the difference in time spent in the litter could be a useful indicator for the hen’s ability to cope with immune challenges.

Identifying potential genomic regions attributed to differences in behavior, particularly durations spent in specific regions of an aviary over periods of production, highlights a number of potential factors impacting hens’ ability to respond to changing stressors throughout production. The variation across different periods of time is in line with Scheiner and Lyman's (1991) pleiotropic model that suggests differences in expression of casual genes across different environments; however, this also means the traits would have a weak response to selection, as their expression is dependent on environment. Even with a potential weak response to selection, there is concern over selecting against abnormal behavior since the behavior may be beneficial to the animal’s ability to cope with its environment (Canario et al., 2013). However, the findings of this study suggest that behavior can be used as an indicator of performance within an environment, and how selection for different performance metrics might impact the behaviors investigated.

Limitations

Although the current study helps to provide insight into the genetic mechanisms impacting laying hens’ behavior in an aviary housing system, it is important to note that there are certain limitations that may affect the interpretation of the findings. One major limitation is the studies’ statistical power to detect associations due to the investigation of only 1098 hens of the same genetic line, which were all housed together, as well as the complex nature of behavior traits. Given the power of the study, chromosome-wise Bonferroni comparisons were used, increasing the power to detect true associations, but additionally allowing for more false positives. Another limitation of the current study was the use of phenotypic means, which may have resulted in higher additive variances for the traits. The use of phenotypic means was done to identify factors impacting behavior at different points in production, but they can affect the interpretation of the findings.

Conclusion

In this study, a genome wide association study was performed for duration in different zones of an aviary, investigated over days 4-290, 4-50, 51-100, 101-150, 151-200, 201-250, and 251-290 of the production period. The findings of this study identified significant SNPs associated with durations in each of the five zones investigated across the different periods. Significant SNPs were mapped to 64 genes, of which some have been related to feeding behavior, movement, and immune function, suggesting that these behavior traits could be used to indicate a hen’s performance within an aviary system. These results are novel, and provide beneficial insight into the genetic architecture of behavior of hens, however, further investigation into the genes identified, as well as replication and validation with a larger independent population is needed before being considered for selection.

Funding

This publication has emanated from research conducted with the financial support of the Federal Food Safety and Veterinary Office on behalf of the Swiss government (2.23.04; Aramis Project Nr.). We also acknowledge funding support from Coefficient Giving (formerly Open Philanthropy).

CRediT authorship contribution statement

Kyle Hoeksema: Writing – original draft, Visualization, Investigation, Formal analysis. Christine F. Baes: Writing – review & editing, Supervision, Conceptualization. Sabine G. Gebhardt-Henrich: Writing – review & editing, Resources, Data curation. Matthew B. Petelle: Writing – review & editing, Resources, Data curation. Michael J. Toscano: Writing – review & editing, Supervision, Funding acquisition. Bayode O. Makanjuola: Writing – review & editing, Methodology, Conceptualization.

Disclosures

The authors of this study declare that they have no known conflict of interest or competing financial interests or personal relationships that influence the work reported in the present study.