In ovo green light remodels the hepatic transcriptome and shapes sex-dependent growth and stress behaviour in meat-type ducks

Abstract

Light during incubation can program poultry growth and stress resilience, but the duck eggshell transmits only a small, variable fraction of external light, so conventional incubation lighting is ineffective and inconsistent. We developed an in ovo lighting (IOL) system that implants a light-emitting diode into the air cell to deliver a defined photic dose, and used it to optimise incubation lighting for Cherry Valley meat ducks and to define its mechanism and post-hatch consequences. We quantified eggshell transmittance, screened light colours and photoperiods for effects on hatchability, incubation duration and embryonic growth, profiled the embryonic day (ED) 24 hepatic transcriptome by RNA-sequencing, and assessed post-hatch sex-specific growth, behaviour, antioxidant status and stress hormones in a 2 × 2 design. Transmittance was low (3–6%), wavelength-selective (red > white > green) and intensity-independent. IOL did not impair hatchability but shortened incubation by 5.7–8 h and widened the hatch window. Green light most increased ED24 body weight (39.6 vs 32.5 g; P < 0.05) and metatarsal length. Daily durations ≥18 h were embryolethal, whereas 12 h produced the largest hepatic response, down-regulating extracellular-matrix and collagen genes and up-regulating CYP7A1, PPAR and steroid-hormone pathways, indicating a shift from structural construction to energy mobilisation. Post-hatch, green light raised 14-d male body weight, improved female leg developmental stability and antioxidant capacity, but green-light females showed reduced exploration and the highest corticosterone, serotonin and interleukin-1β. Rhythmic in ovo green light is a precise, safe intervention that accelerates incubation and growth through hepatic metabolic reprogramming, but its photoperiod must be calibrated to limit a sex-specific stress cost.

Graphic abstract

Introduction

Duck meat is one of the fastest-growing segments of global poultry production, and intensification of the sector has driven a rapid transition away from traditional open-water ponds and extensive floor systems towards high-density, multi-tier cage rearing. While caged systems improve biosecurity, land-use efficiency and carcass hygiene, they remove water access, restrict locomotion and raise stocking density, thereby intensifying fear responses, abnormal behaviour and physiological stress in meat ducks (Jones, 1996; Rodenburg and Turner, 2012). Because ducks are a precocial, water-adapted species, the welfare penalty of confinement may be especially pronounced. Improving the capacity of ducklings to cope with the cage environment, ideally through a low-cost intervention that can be integrated into existing hatchery practice, is therefore an increasingly important target for the modern meat-duck industry.

The incubation period is an increasingly attractive target for such strategies. In meat ducks the 28-day incubation accounts for a large fraction of the pre-slaughter lifespan, so conditions experienced in ovo represent a substantial and tractable window of early life (De Oliveira et al., 2008). During incubation the embryo develops in a largely closed, controllable environment, and environmental cues acting during this period can exert lasting, epigenetically mediated effects on the phenotype — a phenomenon broadly described as developmental programming (Ho and Burggren, 2010; Liu et al., 2021). Among incubation cues, light is particularly compelling because it entrains circadian organisation, influences melatonin rhythms and shapes the developing hypothalamic–pituitary–adrenal (HPA) axis, the principal neuroendocrine system governing the lifelong stress response (Cassone, 2014; Cain and Cidlowski, 2015; Faluhelyi and Csernus, 2007). In broiler chickens, light during incubation — especially monochromatic green light — accelerates embryonic and post-hatch growth, stimulates satellite-cell proliferation and reduces fearfulness (Archer et al., 2009; Rozenboim et al., 2004; Halevy et al., 2006; Archer and Mench, 2013), suggesting that photic programming could improve both productivity and welfare.

Translating this concept to ducks is complicated by a basic physical constraint. The avian eggshell and its membranes attenuate incident light strongly, and the degree of attenuation depends on wavelength and on shell properties such as thickness, pigmentation and porosity (Shafey and Al-Mohsen, 2002; Cooper et al., 2011). Only a small percentage of incident light reaches the embryonic compartment in duck eggs, so conventional external incubation lighting delivers an uncertain and highly variable dose. This is a major reason why the literature on light during incubation is inconsistent: differences in shell characteristics and in the spectral and intensity parameters actually experienced by the embryo confound comparisons between studies (Shafey and Al-Mohsen, 2002; Özkan et al., 2012).

To overcome this limitation we adopted an in ovo lighting (IOL) approach, implanting a miniature light source into the air cell so that defined photic stimulation reaches the embryonic compartment independently of shell optics. The air-cell route is used routinely for in ovo nutrient and vaccine delivery, and such interventions are compatible with normal development and hatching (Uni and Ferket, 2004; Williams and Hopkins, 2011), providing a sound basis for an implanted-light model. By decoupling the delivered dose from shell variability, IOL allows spectral quality, intensity and photoperiod to be controlled with precision unattainable by external illumination.

Using this platform, we conducted a systematic series of experiments in Cherry Valley ducks to: (i) quantify eggshell transmittance across wavelengths, intensities and embryonic stages, and thereby evaluate the rationale and safety of IOL; (ii) screen light colours and photoperiods for effects on hatchability, incubation duration, hatch synchrony and embryonic growth; (iii) dissect the hepatic transcriptomic mechanism through which the selected regime acts; and (iv) assess the long-term, sex-specific behavioural and physiological consequences after hatching, including light-colour preference, fear behaviour, growth, developmental stability, antioxidant status and HPA-axis-related stress responses.

Materials and methods

Animals and ethics statement

The experimental procedures were approved by the Research Committee of Jiangsu Academy of Agricultural Sciences and conducted in adherence to the Regulations for the Administration of Affairs Concerning Experimental Animals (Decree No. 63 of the Jiangsu Academy of Agricultural Science on 8 July 2014). Fertile Cherry Valley duck eggs and the resulting ducklings were obtained from a commercial breeder flock. All experiments used independent batches of eggs from the same breeder flock; no eggs or birds were shared between experiments. The five experiments are described sequentially below: Experiment 1 — eggshell transmittance (Section 2.2); Experiment 2 — implantation safety and hatchability (Section 2.5); Experiment 3 — light-colour screening (Section 2.6); Experiment 4 — photoperiod screening and hepatic transcriptome (Section 2.7); Experiment 5 — post-hatch sex-specific effects (Section 2.8). Sections 2.3 and 2.4 describe the In ovo lighting procedure and growth measurements common to Experiments 3–5. Unless otherwise stated, eggs were incubated under standard commercial conditions, candled on embryonic day (ED) 9 to remove infertile and dead-germ eggs, and statistical significance was set at P < 0.05.

Eggshell light transmittance (Experiment 1)

Transmittance was measured in a light-tight dark box divided into upper and lower chambers by a partition bearing a small aperture on a support. An intensity-adjustable LED source was placed below the shell, and a photometer recorded the light reaching a position as close as possible to the embryonic compartment; transmittance was calculated as the ratio of post-shell to pre-shell intensity (× 100%). Two hundred fertile eggs were set for incubation. At each of ED0, ED4, ED10, ED14 and ED20, 15 viable eggs were randomly selected for transmittance measurement. Each egg was measured under all three colours— red (peak 625–627.5 nm), green (520–526 nm) and white (dominant 455–457.5 nm) — at each of six incident intensities (50, 100, 200, 500,1000 and 1500 lux), yielding 18 measurements per egg in a within-egg repeated-measures design (n = 15 eggs per age × 3 colours × 6intensities). A different set of 15 eggs was sampled at each age (destructive sampling). Egg water loss was determined gravimetrically at the same ages.

In ovo lighting procedure

On ED9, the air-cell shell of each viable egg was disinfected with 70% ethanol, and a hole approximately 3 mm in diameter was drilled over the centre of the air cell with a micro-drill. A 3-mm LED diode (24 V, 20 mA) of the required colour was inserted so that its emitting surface faced the embryonic compartment, and the aperture was sealed with a medical-grade hot-melt adhesive. Eggs were returned to the incubator and photoperiods were controlled by programmable timers. Illumination was applied from ED10 to ED24 at 150 lux; from ED25 eggs were transferred to hatchers (with the diode retained and the power lead severed) and allowed to hatch under darkness up to ED28.

Embryonic and hatchling growth measurements

At ED17, ED24 and at hatch (D1), 12 viable embryos or ducklings per group were sampled. Body weight, heart weight and liver weight were measured with an electronic balance, and body length and metatarsal length with digital callipers. Relative body, heart and liver weights were expressed relative to egg weight or body weight as appropriate.

Light form and hatchability (Experiment 2)

To verify that the physical implantation procedure itself did not impair embryo viability, a dedicated safety trial was conducted. Three hundred fertile eggs were allocated by egg weight to three groups (n = 100 eggs per group, arranged in six sub-replicates of 16–17 eggs): (i) conventional dark incubation (no drilling or implantation), (ii) external lighting (an LED strip mounted above the egg tray), and (iii) In ovo lighting (green LED, 520–526 nm, implanted into the air cell as described in Section2.3). Groups (ii) and (iii) were illuminated at 150 lux on a 12L:12D cycle from ED10 to ED24. Because this experiment was designed solely to assess implantation safety, hatchability at ED28 was the only endpoint recorded. No embryonic growth, behavioural or physiological measurements were made.

Light-colour screening (Experiment 3)

An independent batch of 400 fertile eggs from the same commercial breeder flock (Jiangsu Yike Food Group Co., Ltd.) was used. Eggs were evenly allocated to four in-ovo lighting groups — white-light control, red, green and blue — with 100 eggs per group in six replicates of 16–17 eggs, illuminated at 150 lux for 12 h/d (12L:12D) from ED10 to ED24. At each of ED17, ED24 and D1, 12 viable embryos or ducklings per group were randomly sampled for growth measurements (Section 2.4). In a separate, larger hatching trial (1,440 eggs from the same breeder flock), eggs were allocated to six groups: conventional incubation [Nor], in-ovo dark [Dark], and in-ovo white, red and green light (eight replicates of 30 eggs per group; Table 1), and hatchability, incubation duration and the hatch window were recorded, the latter from video monitoring of the time of emergence of each individual.

Table 1.

Group setting of the hatching-performance trial (Experiment 3).

Group	Lighting form	Photoperiod	Light intensity
Conventional incubation (Nor)	Conventional	12L:12D	None
In-ovo dark (Dark)	IOL	12L:12D	None
In-ovo white (White)	IOL	12L:12D	150 lux
In-ovo red (Red)	IOL	12L:12D	150 lux
In-ovo green (Green)	IOL	12L:12D	150 lux

IOL, in ovo lighting.

Green-light photoperiod screening and hepatic transcriptome (Experiment 4)

A preliminary dose-finding trial was first conducted to determine the safe photoperiod range. Four hundred fertile eggs (from the same breeder flock) were allocated equally to five groups (n = 80 eggs per group, arranged in six replicates of 6–7 eggs each) receiving 0, 6, 12, 18 or 24 h/d of in-ovo green light (150 lux, ED10–ED24). Daily durations ≥18 h were embryolethal, with embryos dying during early incubation; the 6 h and12 h durations were fully compatible with normal development and were therefore selected for the main trial.

In the main trial, fertile eggs were allocated to a dark control (CON,0 h/d), a short photoperiod (STL, 6 h/d, 07:00–13:00) and a long photoperiod (LTL, 12 h/d, 07:00–19:00), each with eight replicates of12–13 eggs, using green LEDs at 150 lux from ED10 to ED24. Embryonic growth was recorded at ED17 and ED24, and incubation duration was monitored by video.

For transcriptomics, at ED24 four livers per group (4 biological replicates × 3 groups = 12 samples) were snap-frozen in liquid nitrogen. Total RNA was extracted, libraries were prepared and sequenced on an Illumina platform, clean reads were aligned with HISAT2, gene expression was quantified as transcripts per kilobase million (TPM), and differentially expressed genes (DEGs) were identified with DESeq2(|log2 fold-change| ≥ 1, adjusted P < 0.05). Functional interpretation used GO and KEGG enrichment and gene-set enrichment analysis (GSEA), and six DEGs were validated by quantitative real-time PCR with results fully consistent with sequencing.

Post-hatch sex-specific effects (Experiment 5)

A total of 320 ducklings were obtained from the main trial of Experiment3. The ducklings originated from the in-ovo dark control (0 h/d) and in-ovo green-light (12L:12D, 150 lux, ED10–ED24). After hatching, ducklings were sexed by cloacal examination and assigned in a 2 × 2 (incubation treatment × sex) factorial arrangement to four groups: dark males (DM; males incubated in darkness), dark females (DF), green males (GM; males incubated under green light) and green females (GF), with 80 birds per group housed in four replicates of 20. All birds were reared to 14 days under a uniform white-light environment (23L:1D,30 lux) with ad libitum access to feed and water. A light-colour preference test was conducted on day 5 using a cross-shaped foraging–social maze (FSM) constructed of 10-mm basswood boards and wrapped externally with black blackout cloth. The maze comprised four identical rectangular compartments (90 cm L × 60 cm W × 90 cm H each) illuminated in red, blue, green and white light, respectively, connected to a central zone (60 × 60 × 90 cm) in a cross configuration. LED strips were mounted to provide 150 lux at bird-head height in each arm. Each duckling (n = 8 per group, 2 randomly selected per cage) was placed individually in the central zone for a 5-min acclimation period (all lights on at equal intensity), after which the test phase began. Behaviour was video-recorded for 10 min, and the first 5 min were scored for the number of zones entered and the time spent and time spent moving in each colour zone. An open-field test (OFT) was conducted on day 6 in a 125 × 125 × 60 cm arena constructed of 10-mm basswood boards with an open top; the exterior was wrapped in black blackout cloth to eliminate visual distractions. The arena was placed in a quiet, uniformly lit room (30 lux white light). The floor was divided into 25 equal squares (25 × 25 cm) by black lines; the inner 9 squares were defined as the central zone. Each duckling (n = 8per group, 2 randomly selected per cage) was placed individually in the centre of the arena, and behaviour was video-recorded. The following variables were scored during the first 5 min: latency to the first movement (s), locomotion (number of grid lines crossed), number of grid crossings, exploration time (s; defined as time spent walking and actively investigating the arena), and total time in the central zone (s). The arena was cleaned with 70% ethanol between consecutive trials.

On day 14, six birds per group (selected to represent the group mean body weight) were euthanised by jugular exsanguination. Blood was collected into EDTA-coated tubes and centrifuged (3,000 × g, 15 min, 4°C); plasma was stored at −80°C until analysis.

Plasma total protein, glucose and total cholesterol were measured using commercial biochemical assay kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China; Cat. Nos. A045-4, F006-1-1 and A111-1-1, respectively) following the manufacturer's protocols.

Antioxidant markers — malondialdehyde (MDA; Cat. No. A003-1), total superoxide dismutase (T-SOD; Cat. No. A001-1), catalase (CAT; Cat. No. A007-1-1) and total antioxidant capacity (T-AOC; Cat. No. A015-2-1) — were measured using colorimetric assay kits from the same manufacturer. Plasma corticosterone (Cort), serotonin (5-HT) and interleukin-1β (IL-1β) were quantified by double-antibody sandwich ELISA (Shanghai Xinfan Biological Technology Co., Ltd., Shanghai, China) following the manufacturer's standard operating procedures (microplate loading, incubation, washing, substrate colour development). Absorbance was read on a microplate reader, and concentrations were calculated from standard curves. The intra-assay coefficient of variation was <10% and the inter-assay coefficient of variation was <15% for all ELISA kits. Sensitivity: Cort, 0.1 ng/mL; 5-HT, 1.0 ng/mL; IL-1β, 1.0 pg/mL.

Statistical analysis

All statistical analyses were performed in GraphPad Prism 9.0 (GraphPad Software, San Diego, CA, USA). The statistical model for each experiment was as follows: Experiment 1 (Eggshell transmittance): Transmittance data were analysed by two-way ANOVA with light colour (red, green, white) and embryonic age (ED0, ED4, ED10, ED14, ED20) as fixed factors, followed by Tukey–Kramer multiple-comparison tests. Because each egg was measured under all three colours and six intensities, the egg was a unit of replication (n = 15 eggs per age). Intensity independence was assessed by one-way ANOVA within each colour × age combination.

Experiment 2 (Hatchability): Hatchability (%) was compared among the three groups by one-way ANOVA (Tukey's test) with the sub-replicate (n = 6 per group) as the experimental unit.

Experiment 3 (Colour screening): Embryonic and hatchling growth variables measured at ED17, ED24 and D1 were analysed by two-way ANOVA with light colour (white, red, green) and age as fixed factors. Because different embryos were destructively sampled at each age (n = 12 per group per age), the data are cross-sectional, not longitudinal, and no repeated-measures or nested model was required. Hatchability and incubation duration in the 1,440-egg trial were analysed by one-way ANOVA (Tukey's test) with the replicate tray (n = 8 per group) as the experimental unit.

Experiment 4 (Photoperiod): Embryonic growth was analysed as in Experiment 3 (two-way ANOVA, photoperiod × age; destructive sampling at each age). Incubation duration was compared by one-way ANOVA (Tukey's test). RNA-seq differential expression analysis used DESeq2 with Benjamini–Hochberg correction (4 biological replicates per group).

Experiment 5 (Post-hatch sex-specific effects): All response variables were analysed by two-way ANOVA with incubation treatment (dark vs green) and sex (male vs female) as fixed factors, followed by Tukey–Kramer tests. The experimental unit was the individual bird for behavioural variables (n = 8 per group) and the individual plasma sample for biochemistry, antioxidant and hormone assays (n = 6 per group).

For all experiments, normality was assessed by the Shapiro–Wilk test and homogeneity of variance by Levene's test. Non-normal variables (grid crossings and RFA in Experiment 5) were log-transformed or square-root-transformed before analysis; back-transformed means are presented where applicable. Results are expressed as mean ± SEM, and significance was set at P < 0.05. Within a row, means without a common superscript letter differ significantly (P < 0.05).

Results

Eggshell transmittance is low, wavelength- and age-dependent, and intensity-independent

Eggshell transmittance in Cherry Valley duck eggs was low throughout incubation, averaging approximately 3–6% of incident light, and was significantly affected by both colour and embryonic age, with a significant colour × age interaction (P < 0.01; Fig. 1-A). Red light penetrated the shell more readily than white light, which in turn exceeded green (red > white > green), and transmittance rose progressively with embryonic age in parallel with an age-dependent increase in egg water loss. In contrast, varying incident intensity from 50 to 1500 lux did not alter transmittance for any colour or age (Fig. 1-B–D), indicating that the shell behaves as a fixed-fraction, wavelength-selective attenuator. These data establish that external illumination cannot deliver a defined dose to the embryo and justify the in ovo approach.

Fig. 1.

Eggshell optics, hatchability and incubation kinetics.

(A) Eggshell light transmittance (%) of white, green and red light across embryonic day (ED0–ED20); two-way ANOVA effects of colour, age and their interaction are indicated. (B–D) Transmittance as a function of incident intensity (50–1500 lux) at each embryonic age for red (B), green (C) and white (D) light, showing no effect of intensity. (E) Hatchability (%) of the conventional (Nor), in-ovo dark (Dark), white, red and green groups (P > 0.05). (F) Incubation time (h) of the same groups; groups without a common letter differ (P < 0.05). (G) Cumulative hatch curves illustrating the wider hatch window of the in-ovo lit groups. Data are mean ± SEM.

In ovo lighting does not impair hatchability but shortens incubation and reduces hatch synchrony

Neither the implantation procedure nor any light form affected final hatchability: values did not differ significantly among the conventional, in-ovo dark, white, red and green groups (P > 0.05; Fig. 1-E), confirming that air-cell LED implantation is a safe model. Incubation lighting did, however, significantly shorten incubation duration (Fig. 1-F): all lit groups hatched earlier than the conventional control, with red light producing the shortest incubation and green light also significantly advancing emergence, shortening incubation by approximately 5.7–8 h. This acceleration was accompanied by reduced synchrony; the cumulative hatch curves show that all in-ovo treatments widened the hatch window relative to conventional incubation, because the developmental acceleration was not uniform across individuals (Fig. 1-G).

Green light is the optimal colour for embryonic and hatchling growth

Among colours, green light most consistently promoted embryonic growth (Table 2). At ED24, green light significantly increased absolute and relative body weight relative to the white control (e.g. body weight 39.59 ± 1.84 g for green vs 32.49 ± 2.14 g for white; P < 0.05) and significantly increased metatarsal length (2.47 ± 0.08 cm vs 2.22 ± 0.07 cm; P < 0.05), whereas relative heart and liver weights were unaffected. At hatch, body weight and length showed significant main effects of age and light colour without an interaction, and organ indices did not differ among colours (Table 3). Green light was therefore selected as the optimal colour for subsequent photoperiod work.

Table 2.

Effects of in ovo lighting with different light colours on embryonic development.

Item	Age	White	Red	Green
Body weight, g	ED17	15.13 ± 0.49	15.23 ± 0.35	15.64 ± 0.34
	ED24	32.49 ± 2.14b	34.51 ± 1.92ab	39.59 ± 1.84a
Relative body weight, %	ED17	19.02 ± 0.55	19.07 ± 0.67	19.75 ± 0.43
	ED24	41.93 ± 2.44b	46.03 ± 2.91ab	52.63 ± 1.93a
Body length, cm	ED17	11.43 ± 0.20	11.51 ± 0.23	11.61 ± 0.14
	ED24	15.20 ± 0.31	15.30 ± 0.24	16.02 ± 0.29
Metatarsal length, cm	ED17	1.22 ± 0.06	1.25 ± 0.05	1.25 ± 0.03
	ED24	2.22 ± 0.07b	2.34 ± 0.08ab	2.47 ± 0.08a
Relative heart weight, %	ED17	0.53 ± 0.01	0.66 ± 0.06	0.55 ± 0.04
	ED24	0.58 ± 0.03	0.57 ± 0.03	0.61 ± 0.06
Relative liver weight, %	ED17	1.28 ± 0.06	1.49 ± 0.06	1.22 ± 0.09
	ED24	1.59 ± 0.09	1.53 ± 0.08	1.49 ± 0.07

Within a row, means without a common superscript differ (P < 0.05); n = 12. ED, embryonic day.

Table 3.

Effects of in ovo lighting with different light colours on body development at hatch.

Item	White	Red	Green	P-age	P-colour	P-int
Body weight, g	49.37 ± 1.36	47.78 ± 1.24	49.47 ± 1.17	<0.01	<0.05	ns
Body length, cm	21.58 ± 0.16a	20.77 ± 0.22b	20.92 ± 0.23b	<0.01	<0.05	ns
Metatarsal length, cm	3.18 ± 0.04	2.94 ± 0.07	3.07 ± 0.06	<0.01	ns	ns
Relative heart weight, %	0.85 ± 0.03	0.83 ± 0.04	0.88 ± 0.05	<0.01	ns	ns
Relative liver weight, %	2.82 ± 0.13	2.88 ± 0.09	2.66 ± 0.11	ns	ns	ns

Within a row, means without a common superscript differ (P < 0.05); n = 12. ns, not significant. P-int, colour × age interaction.

Green light shortens incubation through a photoperiod-dependent effect

In the preliminary trial, daily green-light durations of ≥18 h were embryolethal (n = 12), with embryos dying early in incubation, whereas 6 h and 12 h were fully compatible with development; the safe durations of 6 h (STL) and 12 h (LTL) were therefore studied further. Both photoperiods significantly shortened incubation relative to the dark control (CON), and the effect was duration-dependent: the short photoperiod produced the greatest advance and the long photoperiod a smaller one (CON vs STL, P < 0.01; CON vs LTL, P < 0.05; STL vs LTL, P < 0.01; Fig. 2-A).

Fig. 2.

Photoperiod-dependent incubation kinetics and hepatic transcriptome.

(A) Incubation time (h) of the dark control (CON), short (STL, 6 h/d) and long (LTL, 12 h/d) green-light photoperiods; *P < 0.05, **P < 0.01. (B) Ranked volcano plot of differentially expressed genes in the LTL vs CON comparison, highlighting up-regulation of CYP7A1 and down-regulation of collagen (COL) and actin (ACTA1) genes; point size denotes |log2 fold-change| and colour denotes P-value. (C) KEGG enrichment of differentially expressed genes (Sankey plot linking genes to pathways, left; bubble plot of rich factor, gene number and −log10 P-value, right). Data are mean ± SEM.

Green light reprograms the hepatic transcriptome in a photoperiod-dependent manner

RNA sequencing of ED24 livers revealed photoperiod-dependent transcriptional reprogramming. Relative to CON, the long photoperiod produced the largest response (533 DEGs; 340 down-regulated, 193 up-regulated), the short photoperiod a more modest one (106 DEGs; 40 down, 66 up), and the LTL-versus-STL contrast 87 DEGs (61 down, 26 up); quantitative PCR confirmed the sequencing data. The most strongly down-regulated genes under the long photoperiod were structural and matrix genes, including collagens (COL1A1, COL1A2, COL4A4, COL11A2) and the muscle actin ACTA1, whereas CYP7A1, the rate-limiting enzyme of cholesterol-to-bile-acid conversion, was markedly up-regulated (Fig. 2-B). KEGG enrichment confirmed this pattern: the long photoperiod most strongly enriched ECM–receptor interaction and focal adhesion — driven by coordinated down-regulation of collagen genes — alongside complement and coagulation cascades, while engaging metabolic and biosynthetic pathways including steroid-hormone biosynthesis, PPAR signalling, biosynthesis of amino acids, protein digestion and absorption, and cytochrome-P450-mediated metabolism (Fig. 2-C). Together these data indicate that long-photoperiod green light shifts the embryonic liver away from structural construction and towards energy mobilisation and functional maturation.

In ovo green light does not alter the innate light-colour preference

In the light-preference test, in ovo green light did not change post-hatch colour choice and there was no sex difference (both P > 0.05), but ducklings showed a strong innate colour preference (P < 0.01): across all groups they spent the most time in, and entered most often, the green zone and avoided the red zone (Fig. 3-D–F). Because the proportion of time spent moving was lowest in the green zone while total time there was greatest, green light appeared to exert a calming effect. Thus the innate preference for green was independent of incubation treatment.

Fig. 3.

Sex-specific post-hatch growth, developmental stability and behaviour.

(A) Experimental design: ducklings from in-ovo dark and in-ovo green-light (12L:12D) incubation were sexed and reared to day 14 (DM, dark male; DF, dark female; GM, green male; GF, green female). (B) Body weight at D1 and D14; *P < 0.05 for GM at D14. (C) Combined relative fluctuating asymmetry (cm) by treatment and sex. (D) Light-preference (foraging–social maze) apparatus. (E) Distribution of movement among colour zones and (F) time spent in each colour zone, by group. (G) Open-field apparatus. (H–K) Open-field measures by treatment and sex: time in the centre (H), exploration time (I), grid crossings (J) and latency to first movement (K); *P < 0.05 for green females. Data are mean ± SEM.

Sex-specific effects on growth and developmental stability

Body weight increased with age and did not differ between sexes at hatch, and in ovo green light did not affect body weight on the day of hatch; by day 14, however, green light had significantly increased body weight in males (GM) relative to the other groups (P < 0.05), with no effect in females (Fig. 3-B). Relative fluctuating asymmetry was not increased by green light for any trait, indicating that the treatment did not add developmental stress; females showed higher mid-toe and combined asymmetry than males (P < 0.05), and a significant treatment × sex interaction for leg-length asymmetry reflected lower leg RFA in green-light females than in dark females (P < 0.05; Fig. 3-C). Green light therefore promoted male growth and improved female leg developmental stability.

Sex-specific effects on fear behaviour

In the open-field test, fear-related measures did not differ significantly between sexes overall, and in ovo green light did not change the latency to the first movement, the number of grid crossings or the time in the central zone (Table 4; Fig. 3-H–K). The salient effect was a significant reduction in arena exploration time by green-light females, which fell from 154.53 ± 15.76 s in DF to 94.84 ± 14.40 s in GF (P < 0.05), with no change in males; green-light females also tended to show the longest latency, fewest crossings and most time in the centre. These patterns indicate increased fearfulness specifically in green-light females.

Table 4.

Effects of in ovo rhythmic green light on open-field behaviour at day 6.

Item	Dark		Green
	DM	DF	GM	GF
Exploration time, s	139.50 ± 16.83ab	154.53 ± 15.76a	139.86 ± 12.52ab	94.84 ± 14.40b
Latency to first movement, s	1.71 ± 0.35	1.97 ± 0.46	1.39 ± 0.29	2.26 ± 0.49
Grid crossings, n	93.86 ± 28.26	113.50 ± 16.78	99.00 ± 16.57	64.57 ± 9.93
Total time in the centre, s	171.56 ± 26.37	170.42 ± 21.98	156.92 ± 19.58	224.80 ± 17.36

Within a row, means without a common superscript differ (P < 0.05); mean ± SEM. DM, in-ovo dark male; DF, in-ovo dark female; GM, in-ovo green-light male; GF, in-ovo green-light female.

Plasma biochemistry, antioxidant capacity and stress hormones

Plasma biochemistry showed sex effects but no treatment effect: total protein, glucose and total cholesterol were higher in females (P < 0.05–P < 0.01), and glucose was significantly higher in green-light females than green-light males, with no interaction (Table 5). In contrast, in ovo green light significantly enhanced antioxidant defence: plasma T-AOC and T-SOD activity were increased by green light (P < 0.01) and were also subject to a sex effect (P < 0.05), with green-light females showing the highest values; CAT showed a sex effect and MDA was unchanged, indicating that green light did not provoke oxidative stress (Table 6). For neuroendocrine and immune markers, green light significantly raised plasma Cort and 5-HT (P < 0.01), and green-light females had the highest concentrations of Cort, 5-HT and IL-1β; IL-1β showed a significant treatment × sex interaction, being lowest in green-light males and highest in green-light females (Table 7). Collectively, green light strengthened early antioxidant capacity in both sexes but elevated the stress and inflammatory state selectively in females.

Table 5.

Effects of in ovo rhythmic green light on plasma biochemistry at day 14.

Item	Dark		Green		P-value
	DM	DF	GM	GF	Sex	Trt	Int
TP, g/L	37.92 ± 0.62	44.06 ± 0.78	40.71 ± 2.03	44.51 ± 2.49	<0.01	ns	ns
GLU, mmol/L	14.93 ± 1.21ab	16.55 ± 1.15ab	13.43 ± 0.99b	18.91 ± 1.38a	<0.01	ns	ns
T-CHO, mmol/L	5.70 ± 0.31	5.84 ± 0.19	5.41 ± 0.24	6.25 ± 0.11	<0.05	ns	ns

Within a row, means without a common superscript differ (P < 0.05); mean ± SEM; n = 6. Trt, treatment; Int, treatment × sex interaction; ns, not significant.

Table 6.

Effects of in ovo rhythmic green light on plasma antioxidant indices at day 14.

Item	Dark		Green		P-value
	DM	DF	GM	GF	Sex	Trt	Int
MDA, µmol/L	5.13 ± 0.38	5.11 ± 0.09	4.40 ± 0.59	5.52 ± 0.37	ns	ns	ns
T-SOD, U/mL	35.43 ± 3.44b	45.73 ± 2.75b	50.82 ± 8.52ab	80.49 ± 5.26a	<0.05	<0.01	ns
CAT, U/mL	5.55 ± 1.11b	7.65 ± 1.36b	5.88 ± 1.46b	9.30 ± 1.14a	<0.05	ns	ns
T-AOC, mM	2.05 ± 0.07b	2.07 ± 0.07b	2.12 ± 0.07b	2.39 ± 0.07a	<0.05	<0.01	ns

Within a row, means without a common superscript differ (P < 0.05); mean ± SEM; n = 6. Trt, treatment; Int, treatment × sex interaction; ns, not significant.

Table 7.

Effects of in ovo rhythmic green light on plasma corticosterone, 5-HT and IL-1β at day 14.

Item	Dark		Green		P-value
	DM	DF	GM	GF	Sex	Trt	Int
IL-1β, pg/mL	27.46 ± 1.30bc	28.45 ± 1.43b	20.84 ± 1.47c	36.45 ± 2.42a	<0.01	ns	<0.01
5-HT, ng/mL	18.34 ± 0.93c	28.33 ± 1.40b	25.92 ± 2.23bc	40.63 ± 3.75a	<0.01	<0.01	ns
Cort, ng/mL	8.81 ± 0.63c	15.43 ± 1.40b	16.67 ± 1.40b	23.00 ± 1.97a	<0.01	<0.01	ns

Within a row, means without a common superscript differ (P < 0.05); mean ± SEM; n = 6. Trt, treatment; Int, treatment × sex interaction; ns, not significant.

Discussion

Using an in ovo lighting model that bypasses the optical barrier of the eggshell, we show that rhythmic green light during incubation reprograms the embryonic hepatic transcriptome in preparation for hatching and exerts sex-dependent effects on post-hatch growth, developmental stability and stress physiology. The optical behaviour of the shell both motivates the model and frames the interpretation of the biological responses.

The eggshell barrier justifies the in ovo lighting model

Our measurements confirm that the duck eggshell is a strong, wavelength-selective optical barrier transmitting only 3–6% of incident light, with red penetrating more efficiently than white or green and transmittance rising as the shell thins and loses water with age, while incident intensity has no effect (Shafey and Al-Mohsen, 2002; Cooper et al., 2011). Because the embryo receives only a small and variable fraction of any externally applied light, external incubation lighting cannot deliver a defined dose, which parsimoniously explains the inconsistent effects of incubation lighting reported across studies (Archer and Mench, 2013; Özkan et al., 2012). Implanting the source into the air cell removes shell optics from the dose equation, and the fact that implantation did not impair hatchability — consistent with the established safety of in ovo nutrient and vaccine delivery (Uni and Ferket, 2004; Williams and Hopkins, 2011) — confirms that IOL is a valid platform and that the responses we observed reflect light per se.

Green light is the optimal colour and accelerates the hatching trajectory

Green light produced the most consistent promotion of embryonic growth, increasing ED24 body weight and metatarsal length, in agreement with the well-documented growth-promoting action of monochromatic green light in poultry embryos (Rozenboim et al., 2004; Halevy et al., 2006). All in-ovo treatments shortened incubation without affecting hatchability, indicating accelerated development; however, the acceleration was not uniform across individuals and widened the hatch window. A longer hatch window is operationally important because the first-hatched chicks must wait in the hatcher, increasing the risk of dehydration and energy depletion (Bergoug et al., 2013; Moran, 2007). The shortened incubation is consistent with photic entrainment of embryonic melatonin rhythms and the regulation of developmental timing by the light–dark cycle (Cassone, 2014; Faluhelyi and Csernus, 2007). Thus green light offers a clear growth benefit that must be weighed against reduced emergence synchrony.

Long-photoperiod green light shifts hepatic metabolism from structure to function

The transcriptome provides a coherent mechanism for the growth response and reveals a duration-dependent reprogramming of the liver, the central organ of embryonic energy and nutrient metabolism. Under the long photoperiod, the most strongly enriched and most heavily down-regulated programme was structural: ECM–receptor interaction and focal adhesion, driven by coordinated down-regulation of collagen genes (COL1A1, COL1A2, COL4A4, COL11A2) and the actin gene ACTA1. In the late-term avian embryo the liver normally undergoes progressive functional maturation as the proportion of extracellular-matrix-producing hepatic stellate cells declines (Gordillo et al., 2015; Wang et al., 2025); our data suggest that long-photoperiod green light accelerates this transition, curtailing the energetically costly synthesis of structural and matrix proteins. In parallel, the long photoperiod engaged energy-yielding and biosynthetic pathways — PPAR signalling, steroid-hormone biosynthesis, biosynthesis of amino acids, and protein digestion and absorption — indicating a redirection of limited resources towards metabolic function. The marked up-regulation of CYP7A1, the rate-limiting enzyme of cholesterol-to-bile-acid conversion, points to enhanced mobilisation of yolk-derived cholesterol and lipid (Sun et al., 2023; Angove and Forder, 2023), a plausible contributor to the green-light growth advantage. We therefore interpret the long-photoperiod response as an adaptive reallocation in which the liver reduces structural investment and prioritises energy mobilisation and functional maturation in anticipation of the highly energy-demanding hatching process (De Oliveira et al., 2008; Moran, 2007).

Sexual dimorphism in the behavioural and physiological response

The long-term consequences of in ovo green light were strongly sex-dependent. On the beneficial side, green light increased 14-day body weight in males, improved leg developmental stability (lower fluctuating asymmetry) in females, and raised plasma T-AOC and T-SOD in both sexes without increasing MDA, indicating a benign, hormetic strengthening of antioxidant defence rather than oxidative damage (Surai et al., 2019). The unchanged innate preference for green light, together with the reduced movement and longer dwell time in the green zone, further indicates that green light is perceived as a calming, non-aversive stimulus (Li et al., 2019). On the adverse side, green-light females explored a novel arena less and showed the highest plasma corticosterone, serotonin and IL-1β, a constellation indicating heightened fear and an activated stress and inflammatory axis.

We propose two complementary, literature-supported mechanisms. First, an intrinsic sex difference in HPA-axis reactivity: sexually dimorphic programming of the axis is well established, with females frequently showing greater glucocorticoid responsiveness, and gonadal steroids modulate axis sensitivity from early development (Cain and Cidlowski, 2015; Goel et al., 2021). A photic stimulus during incubation may therefore be integrated differently by the developing female neuroendocrine system; the elevated serotonin in green-light females is consistent with the reciprocal coupling between serotonergic signalling and HPA activity (Lanfumey et al., 2008). Second, a “light-transition stress” at hatching: the embryos experienced a rhythmic green-light environment in ovo but were reared under a near-constant white-light regime, and the abrupt loss of the entraining rhythm represents a circadian and sensory mismatch that can itself act as a stressor (Cassone, 2014; Faluhelyi and Csernus, 2007). If females are inherently more sensitive to such mismatch, the two mechanisms would compound, producing the observed female-biased stress sensitivity. Importantly, this cost was expressed at the level of stress physiology and behaviour rather than as structural instability, since female developmental asymmetry was in fact improved by green light.

Trade-offs, limitations and practical implications

Our findings define a clear trade-off: rhythmic in ovo green light delivers accelerated embryonic growth, metabolic priming for hatching, shorter incubation, greater early antioxidant capacity, improved male growth and improved female leg stability, but at the cost of increased behavioural and neuroendocrine stress sensitivity in females. Incubation-lighting protocols therefore cannot be optimised on growth alone; the photoperiod and the post-hatch lighting environment should be calibrated to capture the benefits while minimising the welfare penalty — for example by limiting photoperiod, smoothing the post-hatch light transition to avoid an abrupt loss of rhythm, and considering sex-specific management. The widened hatch window is a further operational consideration. Several limitations should be acknowledged: the behavioural and physiological assessment was limited to the first two weeks, so the persistence of the female phenotype to market age is unknown; the mechanistic data derive from the liver transcriptome and peripheral hormones, so direct interrogation of HPA-axis ontogeny and central serotonergic and melatonergic signalling is needed; and the interaction of incubation light with temperature and humidity, together with optimisation of hatch synchrony, remains to be explored.

Conclusion

In ovo lighting is a safe and precise method that overcomes the strong optical barrier of the duck eggshell and enables controlled photic stimulation of the embryo. Using this model, we show that rhythmic green light during incubation reprograms the embryonic hepatic transcriptome — shifting the liver from structural construction towards energy mobilisation and functional maturation — thereby preparing the embryo for hatching, shortening incubation and promoting growth, with the strongest body-weight benefit in post-hatch males. Green light also strengthened early antioxidant capacity in both sexes and improved female leg developmental stability, without altering the innate preference for green light. These advantages are counterbalanced by a sex-specific cost: green-light females showed reduced exploratory behaviour and the highest stress-hormone concentrations, most plausibly arising from greater female HPA-axis reactivity compounded by a post-hatch light-transition stress. In ovo green light is therefore a promising welfare- and productivity-oriented tool whose photoperiod and post-hatch lighting context must be carefully calibrated, particularly to mitigate stress penalties in females.

Author contributions

Han Guofeng and Li Mingyang conceived and supervised the study. Li Mingyang performed the experiments, analysed the data and drafted the manuscript. Li Yongsheng and Ding Di contributed to data collection and analysis. Li Baoming, Bai Zongchun and Liu Hang contributed intellectual input to the study. All authors critically revised the manuscript, approved the final version, and agree to be accountable for all aspects of the work, including its accuracy and integrity.

Ethics declaration

This study was conducted in accordance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines. This study was approved by the Research Committee of Jiangsu Academy of Agricultural Sciences. (Approval No. Decree No. 63 of the Jiangsu Academy of Agricultural Science on 8 July 2014)

Ethics statement

The experimental procedures were approved by the Research Committee of Jiangsu Academy of Agricultural Sciences and conducted in adherence to the Regulations for the Administration of Affairs Concerning Experimental Animals (Decree No. 63 of the Jiangsu Academy of Agricultural Science on 8 July 2014).

Disclosures

The authors declare that they have no competing interests.