Larvae used in all three experiments in this study were obtained from pond-cultured northern pike broodstock in Fishery Nove Hrady Ltd. [15 females and 45 males with body weights (BWs) between 0.8 and 1.5 kg], environmentally stimulated by increasing water temperature and sunshine under ambient outdoor conditions. Females and males were separated and maintained in two earthen ponds with a total area of 500 m² and littoral vegetation covering approximately 100 m² of pond area. No hormones were injected. Females were checked for ovulation at weekly intervals, when water temperature fluctuated during the day from 6°C to 12°C. Females were captured by hand and with fishing nets in the littoral area of the pond and were tested for egg ovulation. When ovulation was detected, females were moved near the hatchery for egg stripping and fertilization, according to the method of Bondarenko et al. (2015b). For egg fertilization of each female, testicular sperm collected from three killed males was used, according to the method of Kristan et al. (2020). Fresh fertilized eggs were desticked and incubated under controlled condition with an optimal water temperature of 6–10°C (Bondarenko et al., 2015a). Eggs from each female were incubated separately in 10-L Zug jars. For each experiment, fertilization and incubation of eggs and larvae from five females were performed continuously at one 18-day interval and two 14-day intervals.

Larvae obtained at 11 days post-hatching (DPH) were tested by osmotic shock based on Policar et al. (2010) prior to the transport and stock into each experiment, and only high-quality larvae at 12 DPH presenting survival rates >85% after 90 min of shock (Bondarenko et al., 2015a) were transported in plastic bags (30 L) filled with 2/3 oxygen and 1/3 water from the Fishery Nove Hrady Ltd. hatchery to the Experimental Fish Facility of the Faculty of Fisheries and Protection of Waters, University of South Bohemia (FFPW USB), Vodňany, Czech Republic. After transportation, the larvae (12 DPH) were counted and stocked in experimental tanks for acclimatization over a day. The same procedure was repeated for each experiment. Significantly smaller larvae were obtained and stocked at the same age [BW = 8.9–9.0 mg and total length (TL) = 10.8 mm] at the beginning of the first and third experiments, representing the beginning and the end of northern pike reproductive season compared to larvae stocked for the second experiment performed in the middle of the reproductive season (BW = 11.5–12.5 and TL = 10.8 mm).

At the beginning of each experiment, the larvae were stocked and cultured in 16 cylindrical plastic tanks with an identical volume of 180 L of the same RAS in the FFPW USB. This system comprised culture tanks and water outlets connected with a mechanical parabolic screen filter Ultra Sieve 200 with a maximum water flow of 15,000 L. After the mechanical filtration, water in the RAS continued to the biological filter designed for this study, which comprised two parts: (1) a submerged filter with a total volume of 550 L filled with 50% water and 50% polyurethane foam filter mats (PU PPI30), and (2) a bed-moving filter with a total volume of 550 L filled with 50% water and 50% filtration elements (Random Media BT 10, Ratz Ltd., Leinfelden-Echterdingen, Germany). An EL-S 250 W air blower (Secoh Shanghai Mec. Ltd., Shanghai, China) was used as an air source in both biofilters through air stones and plastic grates installed on the bottom of filters. After biological filtration, the treated water was sterilized with UV light (UV lamp EVO 110, Evolution Aqua Ltd., Wigan, UK) and pumped into a 1,000-L retention tank; from this tank, the water was gravitated back to the fish tanks.

Water temperature (°C) and oxygen saturation (%) were determined during the three experiments using a portable YSI ProODO oximeter (YSI Inc., Yellow Springs, OH, USA) in each tank twice daily (7:30 a.m. and 6:30 p.m.). The pH values were determined using a WTW 3310 pH (WTW GmbH, Weilheim, Germany) meter daily at 8:00 a.m. by submerging it in the biological filter of the RAS system. Total ammonia and nitrite concentrations were measured using a colorimetric reference kit according to Pěnka et al. (2021) once daily. The average water quality parameters for the three experiments are summarized in Table 1 .

An Otohime dry starter (Marubeni Nisshin Feed, Tokyo, Japan), which is frequently and successfully used in larval pikeperch (Sander lucioperca), was used as a high-quality starter (Ljubobratovic et al., 2017; Kucera et al., submitted). Depending on the larval stage, different Otohime sizes were used: B1 (250–360 µm) for days 1–5, B2 (360–650 µm) for days 6–10, and C1 (580–840 µm) for days 11–18; the nutrient content was as follows: 55.8%–55.1% protein, 14.3%–14.9% fat, 2.8%–2.9% fiber, 15.0%–16.4% ash, 2.5%–2.9% calcium, and 2.2%–2.4% phosphorus, vitamin A 10,000 IU kg⁻¹, vitamin D3 2,000 IU kg⁻¹, vitamin E 1,250 mg kg⁻¹, and Cu 7.0 mg kg⁻¹, (Marubeni Nisshin feed Co. Ltd., Japan), which was applied by hand at 15-min intervals during the light regime of the dry feeding period of all three experiments.

Data analysis for all three experiments was performed using one-way analysis of variance (ANOVA) in Statistica version 13 for Windows (StatSoft CR s.r.o., Prague, Czech Republic). The homogeneity of variance and normality of distribution were evaluated using Levene’s test. Post hoc comparisons of means were conducted using the Tukey HSD method when ANOVA detected significant differences among groups, with statistical significance set at p< 0.05. When the data did not meet the parametric assumptions, the Kruskal–Wallis test was used for global analyses and Dunn’s test was used for pairwise comparisons. All data are presented as the mean ± standard deviation.

The effects of the first exogenous feeding regime on the growth performance of pike larvae and early juveniles are shown in Table 3 . Fish fed live feed had significantly higher growth rates than those fed a dry diet (p< 0.05). Extending the feeding period to live feeding resulted in a significantly higher growth rate (24.6% d⁻¹), variability (13.5%), and heterogeneity (299.5% d⁻¹), whereas the opposite was observed for fish fed the dry starter. A similar trend was observed for FC values, where the group offered only the dry starter exhibited the lowest FC value. However, the highest survival rate (49%) and lowest cannibalism rate (21%) were recorded in fish fed solely dry starter (p< 0.05). BW was significantly affected by the weaning time over time ( Figure 2 ). A detailed comparison of the size and condition of the northern pike larvae and juveniles for 18 days is presented in Table 4 .

Groups exposed to complete light (L24:D0) and 16 h of light (L16:D8) exhibited significantly higher weight gain and SGR (%) than the other groups ( Table 5 ). The highest survival rate was observed in fish reared with complete light (68.5%) and under L8:D4:L8:D4 light regimes (61.4%), whereas the highest mortality was in fish reared under L4:D8:L4:D8 light regimes. Cannibalism was significantly affected by photoperiod. The cannibalism rate was significantly higher in fish reared under 4 h light and 4 h dark (L4:D8:L4:D8), and the lowest rate was observed in fish reared with complete light and 16 h light (L16:D8; L8:D4:L8:D4).

The results showed significant effects of stocking density on the growth, survival, and cannibalism rates of pikeperch juveniles (p< 0.05; Table 6 ). Fish with densities of 20 and 40 ind L⁻¹ had significantly higher BWs (205.4–218.6 mg), TL (30.5–31.6 mm), SGR (24.1–24.5% d⁻¹), and SR (58.5%–61.1%) than those in the other groups. The survival rates at the end of the experiment were 72% and 53% for larvae at densities of 20 ind L⁻¹ and 10 ind L⁻¹, respectively. These values were significantly different (p< 0.05). Fish at densities of 10 ind L⁻¹ and 80 ind L⁻¹ showed significantly higher cannibalism rates than those in the other groups. The highest FC value (0.7) and the lowest cannibalism rate (8.5%) were found in groups H40 and H20, respectively. Other parameters such as CV and DR were unaffected by the initial stocking larval density.

Weaning time is crucial for larval rearing. Fish farms are interested in introducing inert diets as early as possible into larval cultures to achieve full replacement of live diets. In the present study, weaning time significantly affected the growth of pike larvae. Upon completion of the experiment, fish fed live feed showed significantly higher growth rates compared to fish fed solely a starter diet, indicating that an artificial diet is currently unsuitable for first feeding larvae. This result is in agreement with previous studies on Vimba vimba (Hamáčková et al., 2009), Clarias gariepinus (Appelbaum and Mcgeer, 1998), C. carpio (Szlamińska and Przybył, 1986), Pleteobagrus fulvidraco (Wang et al., 2005), Clarias batrachus (Giri et al., 2002), and O. mykiss (Akbary et al., 2010). Moreover, there were significantly higher numbers of larvae in fish fed live feed than in those fed artificial feed, 3 days after initiating feeding. Although there was significantly lower growth in larvae fed only the starter diet, the survival rates in this group were the highest. The difference in survival rate observed between fish fed the starter diet alone and those fed live feed was associated with the cannibalism rate. Surprisingly, northern pike displayed the lowest cannibalism rate in fish fed only a starter diet (21%), whereas the highest cannibalism rate (49.5%) was seen in group C under early weaning (day 3), with a survival rate of 13%. This is possibly because larvae fed a starter diet grew slower without developing large differences in size, allowing for higher survival rates. This study indicates that northern pike larvae require at least 9 days of live feed feeding before switching to co-feeding and then to artificial diets.

Photoperiod is an important factor that influences the growth and survival of fish larvae (Fielder et al., 2002; Villamizar et al., 2011). Similar results have been reported for fish larvae, in which the growth rates improved with increasing light periods. The growth of Sparus aurata (Tandler and Helps, 1985) and Siganus guttatus (Duray and Kohno, 1988) larvae was significantly higher under the 24-h light regime. In contrast, increased photoperiod decelerated growth and reduced survival in Dicentrarchus labrax and Archosargus rhomboidalis, possibly due to higher energy expenditure (Barahona-Fernandes, 1979; Dowd and Houde, 1980).

In the present study, the growth of pike larvae in the complete light regime was significantly different from that of larvae reared in complete darkness. The highest survival rate was observed in fish reared under complete light (68.5%) and the L8:D4:L8:D4 light regime (61.4%). Under 24 h of the dark regime, pike larvae mortality reached 100%. The same trend has been observed in previous studies on fish larvae, as demonstrated by decreased growth performance and complete mortality (Fielder et al., 2002; Villamizar et al., 2009; Shi et al., 2010). A possible explanation for mortality under complete darkness is that larvae did not establish first feeding and required light to start feeding. The better growth and survival of pike larvae during longer light periods in the present study is possibly because, during the long light phases, sufficient time was available to establish a robust rhythmicity, and larvae learned quicker how to catch prey (Barlow et al., 1995; Fielder et al., 2002). Cannibalism was significantly affected by photoperiod. Longer light periods also had a positive impact on the cannibalism rate. The current study revealed a low cannibalism rate in larvae reared under complete light and 16-h light groups (L16:D8; L8:D4:L8:D4). The low cannibalism rate observed in the present study was related to the high survival rate of pike larvae.

The optimal density varies among species and depends on the developmental stage of the fish (Baldwin, 2010). Our results demonstrated that stocking density significantly affected pike larvae (p< 0.05; Table 6 ). Larvae reared at low and high stocking densities had significantly lower BWs, SGR, and survival rates than those reared at medium densities. The negative correlation between growth and high stocking density in pikes is consistent with previous studies on fish larvae (Sharma and Chakrabarti, 2003; Bolasina et al., 2006) and can be explained by several factors. An increase in stocking density results in a lower amount of food being consumed by the fish (UEMATSU and Ogawa, 1975); deteriorated water quality (Kebus et al., 1992) and higher cortisol levels (Barnett and Pankhurst, 1998; Montero et al., 1999; Bolasina et al., 2006) lead to decreased growth rates. In addition, a high stocking density results in a greater probability of encounter between a cannibal and potential prey (Baras et al., 2003), allowing a few large larvae to dominate the area and become cannibals. The results of cannibalism in the present study clearly indicate a density-dependent cannibalism rate in pike larvae. Fish at densities of 10 and 80 ind L⁻¹ showed significantly higher cannibalism rates than those in the other groups. This study revealed that a decrease in stocking density resulted in a high cannibalism rate. Similarly, lower growth rates at lower densities were observed in the juvenile African catfish, C. gariepinus (Hecht and Uys, 1997), and according to the authors’ observations, fish at lower densities most likely maintain the territory rather than looking for food.