Animals and experimental treatments
The experiment was carried out in Huayu Poultry Breeding farm, which is located in Handan, Hebei province, China. Hy-Line Brown parent-stock pullets (n = 1440) and cockerels (n = 160) were obtained from a commercial farm and transferred into the experimental house at the age of 16 wk and randomly distributed into 32 identical natural mating colony cages (2.40 × 1.20 × 0.71 m, length × width × height) with 5 males and 45 females in each cage, which was less than the minimum guidelines for housing density provided by Europe and North America. All double-sided experimental cages were arranged in 4 rows. Each experimental cage with a floor area of approximately 576 cm2 per bird, equipped with commercial feed (9.6 cm trough/bird) and drinking facilities (6.25 birds/nipple). However, the environment of the experimental cage was without any enrichment protocol, except for claw abrasive devices (Figure 1). The feed is evenly distributed in the trough and automatically distributed 4 times a day at 07:00, 11:00, 15:00, and 19:00 to ensure birds had permanent ad libitum access to feed. All birds were provided the same standard diet, containing (g/kg; calculated) 178 CP, 4.2 Met, 8.5 Lys, 38.2 Ca, 6.5Pt and 11.4 MJ ME/kg. Eggs and manure were collected once a day through egg conveyor belts and manure belts, respectively. Average air temperature and relative humidity were maintained between 16°C and 23°C, 50% and 80%, respectively, during the experiment. All birds were subjected to the same standard management regime throughout the experiment.
Eight treatments were compared in a factorial arrangement, including 4 LED light colours and each at two light intensities, and giving 4 replicate cages for each light treatment. As shown in Figure 2, the 4 LED light colours were i) red LED light (RL), at a peak wavelength (λp) of 660 nm and a dominant wavelength (λd) of 641 nm, half band width (△λd) of 20 nm; ii) yellow-orange LED light (YO), λp = 616 nm, λd = 600 nm, △λd = 38 nm; iii) blue-green LED light (BG), λp = 445 nm, λd = 479 nm, △λd = 21 nm; iv) white LED light (WL), λp = 449 nm, λd = 491 nm, △λd = 23 nm. All LED light lamps (Huazhaohong Optoelectronic Technology Co. Ltd., Wuxi, China) were attached to the two sides of the cage celling. Randomized block design was adopted for this experiment. These 4 rows were designated as 4 blocks. For each block, the distribution of 8 treatments were arranged according to the random number generator, as shown in Figure 3. Voltage for red, yellow-orange, blue-green, and white LED was adjusted based on the relative spectral sensitivity curve indicated by Prescott and Wathes , to ensure that the 4 lightings appeared iso-illuminant to hens. Light intensity was measured at the level of birds’ heads using a precision luminometer (SRI-PL-6000, Shang Ze Photoelectric Co. Ltd., Taiwan, China) with a resolution of 0.01 lux according to human spectral sensitivity. The two light intensity were 25 lux (HLI) and 10 lux (LLI) respectively. Adjacent experimental cages were separated by shading curtains to avoid light pollution form different light colours. During the experiment period, the lighting rhythm was adjusted based on the different age phase, with a starter 8-h light at the age of 16 and 17 wk, 10-h light at the age of 18 wk, and then increased stepwisely each week to reach 16-h light at the age of 30 wk.
Direct observations were conducted on 4 consecutive days during the age of 30, 40, and 50 wks, respectively, by 2 trained people from an elevated seat in the corridor that allowed a panoramic view of the experimental cages. Each observer observed the 2 adjacent cages simultaneously in the morning and another two in the afternoon. Observation principles were brought into correspondence with each other before observation and the inter-observer agreement was frequently assessed by Kappa value during date collection to ensure the reliability of the results of behavioural observation. Treatments, replicates and time of day (morning and afternoon) of the observation were randomized as a Latin square design over the 4 days in each week. All observers observed an equal number of cages from each treatment (Table 1). Data of mating behaviour were collected between 08:00 and 12:00, 14:00 and 20:00 on the measurement days. The observers sat in position and allowed 10 min for the birds to habituate to the observer’s presence and settle down. Descriptions of behavioural elements recorded were shown in Table 2. Frequencies of all components of mating behaviour for all male-female in each experimental cage were recorded on check sheets over 10 h.
Given that 5 males were confined with females in a very limited space, aggressions might be frequently observed between birds. Therefore, we also recorded feather pecking behaviour, aggressive behaviour, plumage condition, and mortality from cannibalism in another published research using the same animals as the present experiment .
Integrity of mating behaviours
Evaluation for integrity was conducted after behavioural observation of mating behaviour at the age of 30, 40, and 50 wks. According to the field observation and literature research, we divided the mating behaviour into 5 components: courtship display, mounting, treading, cloacal contact, and ejaculation. Courtship display involved any one of the 3 display behaviours. Although it is not always possible to see the actual cloacal contact, there are some good indicators that this has taken place. For example, the cock always stops treading and makes a definite backwards and downwards thrust with his pelvis. Ejaculation was generally accompanied by the rooster's wing flapping and the feathers of the male were raised and the entire body shaken, and the hen nearly always gave a very characteristic high intensity feather shake immediately after the cock dismounts. The integrity of mating behaviour was evaluated according to the completion of the above 5 stages in a single mating process, a score from 0 to 4 was assigned. The details about how the scoring was done is given in Table 3.
Determination of fertility
On weeks when mating behaviour was determined, all eggs weighted between 45 g and 78 g, excluded uncracked and deformed eggs from each experimental cage are suitable for setting were incubated. The incubator temperature was set at 38.3°C and the relatively humidity was maintained at 68%. They were candled at 14 d on transfer to hatcher. Eggs which were apparently infertile and eggs failing to hatch were opened to determine true fertility.
All statistical analyses were performed using linear mixed models parameterized with SPSS (IBM SPSS Statistics 22.0, Armonk, USA). Cage was used as the experimental unit and individual sample data within each of the replicate units (i.e., individual cage) were averaged before analysis, and the residuals were tested for normality and heterogeneity of variance. When data were not normally distributed and could not be solved by transformation, non-parametric statistics were used for analysis. The data were analysed with the fixed effects of light colour, light intensity, weeks of age, 2-factor and 3-factor interactions and the random effect of cage. The model equation was as follows:
Yijkl = µ + LCi + LIj + WKk + CAl + LC*LIij + LC*WKik + LI*WKjk + LC* LI *WKijk + εijkl,
where Yijkl = traits we have investigated; µ = model constant; LCi = effect of light color (i =1 to 4); LIj = effect of light intensity (j =1 to 2); WKk = effect of weeks of age (k =1 to 3); CAl = effect of cage (l =1 to 4); LC*LIij = effect of interaction between light colour and light intensity; LC*WKik = effect of interaction between light colour and weeks of age; LI*WKjk = effect of interaction between light intensity and weeks of age; LC* LI *WKijk =effect of three-way interaction of light colour, light intensity, and weeks of age.
Effects in the statistical model were tested simultaneously and the effects were removed from the original model when they were not significant. When the effect was statistically different (P < 0.05), further analysis was performed. Frequencies of mating behaviour and scores for integrity of mating showed non-normal distributions that were not suitable for transformation, so the Mann-Whitney U test was applied for Post Hoc group comparisons. Mean comparisons were evaluated on fertility by Duncan’s Multiple Range test. Statistical significance was determined at P < 0.05 unless otherwise stated. In addition, a Spearman rank-order correlation was performed to determine correlations between mating behaviour and fertility.