All the experiment was conducted according to the European directive 2010/63/UE. Four lines were purchased from the European Zebrafish Resource Center (EZRC) (Karlsruhe, Germany): AB (WT embryo #1175), SJD (WT embryo #30,607), TU (WT embryo #1173) and WIK (WT embryo #1171). The embryos were transferred the same day to the Aquaculture Experimental Platform (registration number for animal experimentation C5454718) belonging to the L2A laboratory at the University of Lorraine (Nancy, France). For each line, embryos arrived at the same age (2 days post-fertilisation, dpf) in 4 groups, each one coming from several batches derived from an unknown number of parents with unknown degrees of genetic relationship. The groups remained unchanged throughout the experiment, setting 4 replicates per line that we considered as independent. The embryos were kept in embryo medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl, 0.33 mM MgSO4, 7.2 < pH < 7.4) in Petri dishes placed in an incubator at 28.5 °C until hatching (72 h post-fertilisation, hpf). After hatching, the larvae were placed in 8 L aquariums (17 cm × 28 cm × 18 cm, L × W × H) with recirculating water (a renewal rate of 10% per day with osmosis water) at 27 ± 1 °C, with a density of 6 fish/L. A 10-h dark and 14-h light photoperiod was applied, with 30 min of dawn and dusk and 300 lx of light intensity at the surface water. Over the course of the experiment, the temperature, photoperiod and light intensity were controlled automatically by central technical management. Three times a week, the pH (7.5 ± 0.5), conductivity (850 ± 50 µS/cm) and ammonia and nitrite nitrogen levels were measured and remained below 0.1 mg/L. Mortality was monitored daily but remained anecdotal and equivalent between lines, which did not significantly affect density.

The dry food Gemma Micro ZF (Skretting, Fontaine-lès-Vervins, France; proteins 59%; lipids 14%; fibre 0,2%; minerals 14%; phosphorus 1,3%; calcium 1,5%; sodium 0,7%; vitamin A 23000 UI/kg; vitamin D3 2800 UI/kg; vitamin C 1000 mg/kg; vitamin E 400 mg/kg) was distributed ad libitum in different granulometry depending on the age of the fish. ZF75 was used 3 times per day between 5 and 15 dpf. ZF150 was used twice a day between 15 and 90 dpf, followed by ZF300 once per day until the end of the experiment. From 5 dpf until the end of the experiment, Artemia nauplii (EG > 225) was given as a supplement once per day.

Growth control experiments were performed every 20 days from 10 to 110 dpf. A final checkpoint was added at 150 dpf (Fig. 1).

Experimental timeline. The “S” abbreviation is used for “sampling”. Time is expressed in days post-fertilisation (dpf)

Three sampling times were performed to measure hormone levels (Fig. 1): (1) S1 (52 dpf), which consisted of collecting the whole body to estimate basal cortisol levels and the head to estimate serotonin levels; and (2) S2 (100 dpf) and S3 (157 dpf), which consisted of female whole-body sampling to estimate 17α,20β dihydroxyprogesterone (DHP) levels corresponding to the maturation-inducing steroid in most female fish species. At S2, the brain was sampled to assess its weight in both males and females.

At the timepoint S1, fish are considered as juvenile, following Parichy et al. definition (i.e. “state at which most adult characteristics have been acquired in the absence of sexual maturity” associated with complete squamation and “complete loss of the larval fin fold”) [24]. Fish are considered as adult if they have produced viable gametes [24] (commonly around 3-month-old, i.e. around 90 dpf) [5]. Consequently, at S2 and S3 samplings, fish are defined as “young adult” and “more advanced adults”, respectively.

To determine whether the females were mature, reproduction tests were performed from 86 to 168 dpf. These tests are designated “puberty assessments” in the Fig. 1.

Finally, both male and female reproductive performance were tested (Fig. 1). Female performance was tested concomitantly with the puberty assessment and consisted of absolute fecundity and fertilisation rate measurements. Male performance was tested by sperm analysis at 367 dpf.

Four fish per replicate, for a total of 16 fish per line, were anaesthetized with 50 mg/L (10 dpf), 60 mg/L (30 dpf to 70 dpf) or 150 mg/L (from 90 to 150 dpf) of ethyl 3-aminobenzoate methanesulfonate solution (tricaine, Sigma) and NaHCO3 (at the same concentration as tricaine). The fish were placed under a binocular loupe (SZX7, Olympus), and pictures were taken with a camera (Cam SC50, Olympus) using specific software (CellSens CS-EN-V3, Olympus). Depending on the size of the individual, a × 0.5 or × 1.25 lens was used combined with a zoom in the range of 0.5 to 2. At 10 dpf, the pictures were taken dorsally, while they were taken laterally after 30 dpf. All the measurements were conducted using Parichy et al. [24] as a reference for the definition of the measured traits (standard length (SL), total length (TL), snout-vent length (SVL), snout-operculum length (SOL), height at the nape (HAN), height at the anterior of the anal fin (HAA), and eye diameter (ED)). MesurimPro software was used to perform the various measurements.

The fish were weighed at each sampling time (S1 to S3). After being euthanised by an overdose of anaesthetic (300 mg/L + 300 mg/L NaHCO3), the fish were wiped on absorbent paper to remove external water and weighed with a precision of ± 1 mg (Practum313-1S, Sartorius). The number of individuals weighed at 52 dpf was 32/line and varied from 10 to 24 at 100 dpf and from 5 to 15 at 157 dpf, depending on the line and sex.

At 100 dpf, after the total weight was recorded, the whole brain without the spinal cord was carefully removed and weighed. Brain weight was normalised to body weight. The number of individuals used varied between 14 and 16, depending on the line.

Additionally, global morphology analyses were performed using 150 dpf images. The software tpsDig (v2.32) was used to locate 26 landmarks (supplementary data S2), and MorphoJ (v1.07) was used for the procrustean analysis.

Immediately after death, the whole body or the head (depending on the sample taken) was wiped on absorbent paper to remove external water, weighed and placed in a tube with 10 µL of 1X phosphate-buffered saline (PBS) for 1 mg of tissue and zirconium oxide beads (ZrOB10, Next Advance). The samples were homogenised and centrifuged (5000 × g, 5 min, 4 °C), and the supernatants were collected and stored at − 20 °C until enzyme-linked immunosorbent assay (ELISA, detailed information in Table 1).

To determine when the fish were mature, 8 randomly chosen pairs for each line were subjected to reproduction conditions twice per week. Before dusk, one male and one female were placed in 1 L breeding tanks with a separator between them and with a grid preventing the parents from having access to the eggs. At dawn, the separator was removed. Two hours after removal of the separator, the presence (or absence) of eggs was observed.

The first trial took place at 86 dpf, and the last one occurred at 168 dpf. For a given line, the maturity onset was defined by the occurrence of the first spawning. Reproduction was considered successful when the female had spawned, regardless of whether the eggs were fertilised. After each reproductive event, the eggs were collected using an 80 µm sieve and rinsed with embryo medium (see section “Experimental animals and rearing conditions”). Each spawn was put into a Petri dish with 35 mL of embryo medium and placed in an incubator at 28.5 °C. Six hours post-fertilisation, unfertilised eggs were counted and removed using a binocular loupe. Fertilised eggs were counted to determine the absolute fecundity (i.e., total number of eggs per female = fertilised + unfertilised eggs + empty chorion) and fertilisation rate (i.e., (number of fertilised eggs/total number of eggs) × 100). The spawns were returned to 28.5 °C in an incubator and monitored every day. Dead embryos (those whose heartbeat stopped) were removed every day until hatching.

A sperm analysis was performed at 367 dpf. After anaesthesia (150 mg/L tricaine, Sigma + 150 mg/L NaHCO3), the fish were gently blotted dry to remove water excess from the area around the vent and stripped to collect the sperm (n = 4 or 8, depending on the line). The volume collected varied from 0.5 to 2 µL depending on the individual. Immediately after collection, the sperm were placed in 24 µL of immobilizing solution (1 L of distilled water: 8 g of NaCl, 0.4 g of KCl, 0.16 g of CaCl2, 2H2O, 0.2 g of MgSO4, 7H2O, 0.06 g of Na2HPO4, 0.06 g of KH2PO4, 0.35 g of NaHCO3, and 1 g of C6H12O6, pH = 7.5) and kept on ice until analysis (4 h after stripping at most).

Analyses were performed using computer-assisted sperm analysis (SCA® Veterinary, Motility and Concentration, Microptic) and a microscope (Ci-L, Nikon) with a × 10 phase contrast lens and heating plate (PE120, Linkam). The sperm were activated with tap water at a ratio of 4:16, and 4 µL was immediately placed into the counting chamber (MOT-20–6 SCA®, Microptic) on a heating plate at 28 °C. The samples were automatically read 7 to 11 s after activation, depending on the sample.

Statistical analyses were performed with R (v. 4.2.0). Normality and homogeneity of variance were tested using the Shapiro‒Wilk and Levene tests, respectively. When both assumptions were met, ANOVA was performed. If normality and/or homogeneity were not met, a Kruskal‒Wallis test was performed. If only homogeneity was absent, a Welch’s ANOVA was performed. If necessary, a post hoc test was conducted with a Bonferroni correction (Tukey test following ANOVA and Dunn test following the Kruskal‒Wallis test).