PRODUCTION PERFORMANCE OF TAMBAQUI JUVENILES SUBJECTED TO SHORT FEED-DEPRIVATION AND REFEEDING CYCLES

This study proposes to investigate whether short restricted-feeding cycles during the pre-fattening phase of tambaqui (Colossoma macropomum) juveniles affect their production performance and yield in a semi-intensive system and if this practice can be introduced as an alternative to reduce production costs in the pre-fattening phase of this species. The fish (8.07 ± 0.07 g) were subjected to short restricted-feeding periods under controlled laboratory conditions (experimental boxes) for 60 days (Phase I). Three feeding strategies were tested, namely: dairy feeding (Control); five days of feeding followed by two days of deprivation (5F/2D); and two days of feeding followed by four days of deprivation (2F/4D). Subsequently, the fish from each treatment (feeding regimes) evaluated in Phase 1 were transferred to an excavated pond with low water exchange and no supplemental aeration (semi-intensive system) where they were fed continuously for 65 days (Phase II). At the end of both phases, performance, metabolic and hematological parameters and the centesimal composition of the filets were analyzed and an economic assessment was undertaken based on the total operating cost (TOC) of production in the different feeding regimes. The restricted-feeding cycles 2F/4D and 5F/2D (Phase I) affected the performance of the fish, which showed lower daily weight gains and specific growth rates (SGR), resulting in lower viscerosomatic index (VSI) and morphometric measurements. However, apparent feed conversion did not differ across the treatments. The feed-deprived fish used triglycerides as an energy source, maintaining their blood glucose levels close to those of the continuously fed group (control). When they started to be feed daily for 65 days (Phase II), the metabolic (triglycerides) and hematological parameters (hematocrit and hemoglobin) equaled those of control group. Nevertheless, their final weight and morphometric measurements were lower than those of non-restricted feed, indicating a partial compensatory growth in tambaqui juveniles. Short restricted-feeding cycles applied in the pre-fattening phase for 60 days negatively affected performance and led to metabolic and hematological alterations in tambaqui (Colossoma macropomum) juveniles. Although these short restricted-feeding cycles resulted in less expenditure on labor and feeding, the fish under those conditions did not develop equally to those fed daily, culminating in less biomass produced during the fattening period in a semi-intensive system for 65 days. Ultimately, this led to a higher total operating cost per gram of produced fish, demonstrating the economic infeasibility of this practice for tambaqui juvenile production.


INTRODUCTION
Brazil produced approximately 691,700 t of fish in the year 2017, which represents a growth of 8% compared with the previous year. The production of native species accounted for 43.7% of the national total, led mainly by tambaqui (Colossoma macropomum) farming (Peixe BR, 2018). This species is characterized by its hardiness, rapid growth and good acceptance in the Brazilian market (Pedroza Filho et al., 2016). Because of these valuable characteristics, it is largely cultivated nationwide.
In view of the expansion of fish production, one of the main aspects of management to be evaluated is feeding, because of its representativeness in costs (Barros and Martins, 2012;Sabaini et al., 2015;Brabo et al., 2017), followed by man labor and purchase of fingerlings. The significance of those expenses varies according to the rearing phase and the species in question (Sanches et al., 2013;Brabo et al., 2015;Costa et al., 2016). In this regard, restrictive feeding regimes may have a direct impact on the costs of fish farming in that they allow for a reduction of the total volume of feed consumed in every production cycle as well as a reduction of man labor and use of equipment/materials.
This compensatory growth is typically characterized by hyperphagia, better feed conversion and elevated specific growth rate (Won and Borski, 2013). However, several factors such as species, reproductive state, age, temperature, among others, can affect this growth's compensation level (Navarro and Gutiérrez, 1995).
On these bases, daily feed supply in fish farming started to be questioned. The use of short restricted-feeding cycles could maximize the rates of growth and feed utilization for muscle formation, implicating economic benefits to the producer and to the environment. Although it has been documented that restrictive-feeding programs have positive effects on fish farming, little information exists on its effect on profitability, especially in terms of reduction of operating costs through savings on feeding itself and on the man labor required for feeding.
In this scenario, studies investigating which parameters are influenced during restrictive feeding regimes in fish rearing and their economic impact on the activity are necessary to establish better strategies for their application. The present study was thus carried out to determine whether short restricted-feeding cycles during the pre-fattening phase of tambaqui (Colossoma macropomum) juveniles affect their production performance and yield in a semi-intensive system and if this practice can be adopted as an alternative to reduce production costs during the pre-fattening phase of this species.

Experimental design
The study took place in the Fish Farming Section at the Federal University of Mato Grosso (UFMT), located in Santo Antônio de Leverger -MT, Brazil (15° 51' 56" S and 56° 04' 36" W), in the period of April to August 2017. The experimental procedures were previously approved by the Ethics Committee on Animal Use (CEUF/UFMT;approval no. 23108.186540/2016-44).

3/9
Brazil) containing 420 g kg -1 crude protein and 70 g kg -1 ether extract, with a pellet size of 2 to 3 mm.
In Phase I, which was developed under laboratory conditions for 60 days, no natural food was available to the animals, meaning they were fed only with the feed. A total of 240 tambaqui (Colossoma macropomum) juveniles with an initial weight of 8.07 ± 0.07 g (mean ± standard error) was allocated at random to 12 polyethylene cages with 150-L capacity (20 fish per cage) in an open system with recirculation of water collected by a semi-artesian well (exchange rate of 4 L min -1 ) and constant aeration. The photoperiod was controlled by an automatic illumination system with 12 h of light and 12 h of darkness. Fish were fed ad libitum for 60 days, twice daily (09h00 and 15h00), according to the following treatments: daily feed supply (Control); five days of feeding followed by two days of deprivation (5F/2D); and two days of feeding followed by four days of deprivation (2F/4D).
On the 61st day, after a feed-deprivation period of 24 h, the fish (n = 20 per treatment) were anesthetized with eugenol (10 mg L -1 ), following the methodology of Inoue and Moraes (2007) and weighed. Subsequently, the following variables were measured using a fish measuring device (ictiometer) and a caliper rule: total length (TL), standard length (SL), head size (HS), body height (BH) and body width (BW). Afterwards, the animals had their blood drawn by caudal puncture to evaluate their metabolic and hematological responses and were killed (by sectioning the branchial arch) for removal of offal and filetting.
Next, the remaining fish were marked with a microchip (inserted in the back musculature near the head) and transported inside plastic bags with oxygen and water with common salt (3 g NaCl L -1 ) to an excavated pond without water exchange or supplemental aeration. The pond was subdivided with screens into three 133-m 2 units, each of which housed fish that were subjected to the same feeding regime in Phase 1 (60 fish/unit), initiating Phase 2 of the experiment, where the treatments consisted of the fish from each of the feeding regimes in Phase 1. In this second phase, which lasted 65 days, the fish were fed daily (DF) with a commercial extruded feed ad libitum twice per day (09h00 and 15h00). At the end, the fish from each treatment were captured with a trawl and anesthetized for biometric measurements (weight and body measurements), followed by blood collection. Subsequently, the fish were killed for removal of offal and filleting, in accordance with the procedures described in Phase I.

Blood analyses
Blood was harvested with and without anticoagulant (ethylene diamine tetra acetic acid -EDTA). Hemoglobin (hemoglobin cyanide method, using a Labmax Flex  analyzer and Labtest  commercial kits) and hematocrit (centrifugation in a Spin100 microcentrifuge at 12,000 rpm for 5 min) were determined in the total collected blood. Afterwards, the blood was centrifuged for 10 min at 3000 rpm (Serological Centrifuge Model 80-2B) for separation of serum and plasma. In the blood plasma, glucose was measured by the enzymatic glucose oxidase method and Trinder's reagent (Labtest  commercial kit) and triglycerides were determined by Trinder's reagent enzyme method (Labtest  commercial kit), both using a Labmax Flex  analyzer.

Centesimal analysis
The filets obtained from the fish were used for triplicate analysis of moisture, crude protein, lipid and ash contents according to the Official Physical-Chemical Analytical Methods for the Control of Fish and Derivatives Brasil, 2011).

Economic indicators
As an economic indicator, the total operating cost (TOC) was calculated by following the method proposed by Martin et al. (1994), considering the sum of the effective operating costrepresented by items such as man labor, feeding, purchase of the fingerlings, general inputs, occasional expenses and maintenance of machinery and improvements-and the depreciation values. For the economic analysis of producing tambaqui subjected to short restricted-feeding cycles, we calculated the average TOC (R$ g -1 ), which was determined as TOC (R$) divided by the obtained final biomass (g).

Statistical analysis
Phase 1 of the study was conducted as a completely randomized design with three treatments (control and two feeding regimes) and four replicates. In Phase II, fish originating from the same feeding regime in Phase 1 constituted one treatment, totaling three treatments with 60 replicates, since each fish individualized with a microchip was considered a replicate. The completely randomized design was also adopted in Phase II, since all replicates 4/9 were subjected to the same non-perceptible variations. Data were tested for normality and homogeneity of variance and subjected to analysis of variance (ANOVA), and when the F values indicated significant differences, means were compared by Tukey's test (5%). Results were expressed as means ± standard error.

Production performance
The fish subjected to restricted-feeding periods (Phase I) had a lower daily weight gain (DWG) than control group (Figure 1). These effects consequently influenced the final weight of the animals and their viscerosomatic index (VSI), which were significantly lower (Table 1). Restricted feeding in this phase also led to a lower specific growth rate (SGR) (Figure 1A), while apparent feed conversion (AFC) did not differ across the treatments. The condition factor (K) was significantly lower in the fish subjected to the more intense deprivation period (2F/4D) (Table 1).
When the tambaqui were fed continuously for 65 days (Phase II), DWG did not differ across the treatments ( Figure 1B), but those subjected to the more intense feed restriction (2F/4D+DF) showed a significantly higher SGR ( Figure 1B) and a significantly lower AFC (Table 1). In spite of their higher SGR, final weight was significantly lower when compared with that of the other treatments (Table 1).
Short restricted-feeding cycles (Phase I) affected the morphometric measurements of the juveniles, whose values were lower than those of control group (Table 2). This difference remained even after continuous daily feeding for 65 days (Phase II) ( Table 2).

Blood parameters
Short restricted-feeding cycles (Phase I) did not affect blood glucose levels in the fish. The fish that underwent the longer period of feed restriction (2F/4D) showed lower a hematocrit percentage and a lower hemoglobin concentration (Table 3). The feed-deprived fish exhibited lower triglyceride values than those of control group (Table 3). After being fed continuously for 65 days (Phase II), the blood parameters glucose, triglycerides, hematocrit and hemoglobin did not differ across the treatments (Table 3).

Centesimal composition
The tambaqui juveniles subjected to short restricted-feeding cycles (Phase I) exhibited a lower concentration of lipids in their muscle (Table 4), although no significant differences were seen in moisture, ash and protein across the treatments. This difference in lipid concentration in the filet in relation to control treatment remained even after the fish were fed continuously for 65 days (Phase II) (Table 4).

Economic analysis
During Phases I and II, general expenses and depreciation were the equal for all treatments. Differences were only found for feeding and man-labor expenses in Phase I, in which the regimes of restricted feeding for 2 days (5F/2D) and 4 days (2F/4D) provided savings in feeding (33.0% and 76.2%, respectively) and labor (33.3% and 66.7%, respectively) costs. Consequently, the TOC of production in the respective treatments were lower (by 17.9% and 37.5%, respectively) than in the daily-feeding strategy (control treatment) (Table 5). However, the feed-deprived fish (5F/2D and 2F/4D) exhibited a lower weight gain and consequently lower final biomass production (by 30.9% and 69.2%, respectively) than those fed continuously (control) ( Table 5). In Phase II, the only expense that differed across the treatments was feeding, but, as in Phase I, TOC was reduced in the treatments involving feed restriction (Table 5).

Figure 1.
Daily weight gain (DWG) and specific growth rate (SGR) of tambaqui (C. macropomum) juveniles subjected to short restricted-feeding cycles for 60 days (A -Phase I) and continuous feeding during rearing in a semi-intensive system for 65 days (B -Phase II). Different lowercase letters indicate significant differences for DWG and different uppercase letters denote differences for SGR, according to Tukey's test (5%). Control: daily feeding (DF); 5F/2D = five days of feeding/two days of fasting; and 2F/4D = two days of feeding/four days of fasting'. Mean ± standard error. Different letters in the same row indicate significant differences by Tukey's test (5%). K factor = condition factor; VSI = viscerosomatic index; AFC = apparent feed conversion; Control = daily feeding (DF); 5F/2D = five days of feeding/two days of deprivation; and 2F/4D = two days of feeding/four days of deprivation. 8.57 ± 0.12 a 7.97 ± 0.11 b 7.03 ± 0.14 c Head (cm) 5.55 ± 0.09 a 5.17 ± 0.08 b 4.53 ± 0.10 c Mean ± standard error followed by different letters in the same row indicate significant differences by Tukey's test (5%). Control = daily feeding (DF); 5F/2D = five days of feeding/two days of deprivation; and 2F/4D = two days of feeding/four days of deprivation. Mean ± standard error followed by different letters in the same row indicate significant differences by Tukey's test (5%). Control = daily feeding (DF); 5F/2D = five days of feeding/two days of deprivation; and 2F/4D = two days of feeding/four days of deprivation.
Despite the reduction observed in the total operating costs of both phases, the TOC per gram of juvenile produced during short periods of restricted feeding followed by refeeding became 8% (R$0.013) and 25.9% (R$0.016) higher than that of control group (R$0.012), due to the lower biomass produced (Table 5).

DISCUSSION
The restricted-feeding cycles (Phase I) did not interfere with the efficiency of utilization of the supplied feed for transformation into live weight, since no significant difference was found in apparent feed conversion in the restricted and continuously fed fish (control). Nevertheless, the results indicate that performance was affected by the offer of feed, with restriction leading the fish to consume less, which resulted in lower DWG, final weight, VSI, morphometric measurements and SGR. Similar findings were reported by Takahashi et al. (2011), who observed lower weight gain, feed intake and SGR in pacu (Piaractus mesopotamicus) (initial weight of 36 g) subjected to short periods of restricted feeding (3 days of feeding / 3 days of fasting) for 36 days. Rodriguez and Landines (2011), in turn, did not find significant differences in the productive parameters of pirapitinga (Piaractus brachypomus) (initial weight of 54 g) subjected to short deprivation periods (3 days of feeding / 2 days of fasting) for eight weeks.
The metabolic response in fasting may be influenced by several factors such as species, age, photoperiod, energy reserves, reproductive period, temperature, among others (Navarro and Gutiérrez, 1995). The difference in the above-mentioned results may be related to the fact that these are different species in distinct development phases, which lead to differences in energy reserves at the beginning of feed restriction. Table 4. Centesimal composition of the filet of tambaqui (C. macropomum) juveniles subjected to short restricted-feeding cycles for 60 days (Phase I) and continuous feeding during rearing in a semi-intensive system for 65 days (Phase II). *Percentage on a wet basis. Mean ± standard error followed by different letters in the same row indicate significant differences by Tukey's test (5%). Control = daily feeding (DF); 5F/2D = five days of feeding/two days of deprivation; and 2F/4D = two days of feeding/four days of deprivation. Table 5. Production cost of tambaqui (C. macropomum) juveniles subjected to short restricted-feeding cycles for 60 days (Phase I) and continuous feeding during rearing in a semi-intensive system for 65 days (Phase II). 0.012 0.013 0.016 *Inputs, maintenance and fees. Purchase of fingerlings considered only in Phase I. Control = daily feeding (DF); 5F/2D = five days of feeding/two days of deprivation; and 2F/4D = two days of feeding/four days of deprivation.

7/9
Ali et al. (2003) stated that when fish are subjected to feed deprivation and have a decline in growth, its compensation after the normalization of feeding may be total, when they attain the same weight at the same age as those fed continuously; or partial, when they show a rapid growth rate and better feed conversion, but do not reach the same weight as those that were not restricted at the same age. Short restriction cycles lead to partial compensatory growth in pacu and pirapitinga, since the final weight of restricted animals was lower than that of the continuously fed fish (Rodriguez and Landines, 2011;Takahashi et al., 2011). This phenomenon was also observed in the present study in the tambaqui juveniles subjected to restricted-feeding periods, after feeding was normalized (daily feeding for 65 days).
The magnitude and duration of compensatory growth depend on the duration of restricted feeding. Longer restriction periods may result in higher growth rates, leading to total compensatory growth (Navarro and Gutiérrez, 1995). However, long deprivation periods may cause damage to the animals and compromise their growth. Ituassú et al. (2004) observed total compensatory growth when tambaqui juveniles were restricted for longer periods (14 days of deprivation), but when restriction was extended for more than 21 days, the fish showed partial growth.
In the present study, tambaqui juveniles subjected to restricted-feeding cycles showed a lower condition factor (K), and those subjected to the more intense restriction regime (2F/4D) exhibited hematological changes such as decreased hematocrit and hemoglobin concentration. In spite of those alterations, the blood glucose level in the restricted fish did not differ from that recorded in the animals fed continuously (control), which means that the fish under restriction used triglycerides as an energy source, maintaining their blood glucose values constant. The present results agree with the reports of Pérez-Jiménez et al. (2007), who observed a reduction in the plasma triglyceride levels of Dicentrarchus labrax subjected to restricted feeding.
According to Navarro and Gutiérrez (1995), when fish are subjected to feed-deprivation periods, their blood glucose concentrations are kept at a stable level, which is majorly due to the use of hepatic glycogen, and the mobilization of intramuscular lipids in some species is initiated as soon as fasting starts. As stated by those authors, proteolysis occurs when the reserves of easily accessed energy such as liver glycogen and lipids have been used.
In the current study, no proteolysis was observed in the tambaqui juveniles under short restriction cycles. However, these fish showed a lower concentration of lipids in the muscle (filet), demonstrating that there was lipid mobilization at the muscle level to be used as a source of energy during feed deprivation, as described in other fish species (Takahashi et al., 2011;Urbinati et al., 2014). Additionally, in the feed-restricted fish, the energy from the feed was used for the maintenance of metabolism rather than for muscle deposition. This finding corroborates Navarro and Gutiérrez (1995), who reported that lipogenic activity seems to be reduced during the entire feed-deprivation period.
After being fed daily for 65 days (Phase II), the tambaqui juveniles that had undergone short restriction cycles showed similar metabolic and hematological parameters and condition factor to the fish fed continuously (control), indicating recovery of these parameters after feeding was reestablished. As for the productive parameters, after being fed daily (DF), the fish subjected to restriction periods in Phase I showed a higher specific growth rate, which suggests compensatory growth. However, this growth was partial, since neither the final weight nor the morphometric measurements equaled those of the unrestricted fish.
When fed continuously, the fish which had been subjected to the more intense feeding restriction (2F/4D) showed the lowest apparent feed conversion (AFC), indicating their greater ability to convert the ingested feed. This is likely due to the sensitivity of biochemical mechanisms which is exacerbated during feed deprivation (Jobling, 2010). The fact that natural feed was available in the rearing environment may explain the observed AFC lower than 1.0 for the fish in that treatment group, since tambaqui has large filtering capacity despite being an omnivorous species.
During growth, a larger portion of the acquired volume in the fish is white muscle, which implies accumulation of myofibrillar proteins (Mommsen, 2001). Considering that the protein synthesis requires ATP, the growth process requires use of energy (Bombardelli et al., 2004). Even after feeding was reestablished, the feed-restricted fish showed a lower lipid concentration in the filet than those fed continuously (control), suggesting utilization of the energy from the feed for the maintenance of metabolism and somatic growth rather than for lipid deposition in the muscle. Results obtained in laboratory conditions with short restricted-feeding cycles indicate that tambaqui juveniles have the capacity to mobilize their energy stock to go through fasting periods, but this physiological adaptation was not sufficient to elicit total compensatory growth.
The economic analysis of Phase I indicated that the operating expenses were lower in the regimes of feed deprivation for 2 and 4 days (5F/2D and 2F/4D). In Phase II, when all fish started to be fed daily, a 36.6% lower feeding cost was obtained in the group that was fasted for 4 days in comparison with control group, which was due to the significantly lower produced biomass and AFC. However, the average total operating cost (R$ g -1 ) of this treatment was 25.9% lower than that of control, suggesting lesser participation of feeding in the production costs during this phase. This can be verified in other studies (Sanches et al., 2013;Brabo et al., 2015) in which the authors reported that, unlike the fattening stage, in the fingerling production stage, feeding is not the greatest expense. Rather, man labor and the purchase of juveniles are the costliest items, which contrasts with the present findings obtained during the pre-fattening stage.
Although feed deprivation allowed for decreased spending on man labor and feeding, the fish under such conditions did not develop equally to those fed continuously, resulting in less biomass produced. This explains the higher total operating cost per gram (R$ g -1 ) of fish produced in those treatments (8.0% for 5F/2D and 25.9% for 2F/4D) in relation to control. In this way, the feeding management protocols in which the fish were fasted for a period resulted in a reduction of operating costs. Nevertheless, production in live weight was also lower, which means that despite the decreased expenses, the average total operating cost per gram of produced fish increases due to the lower productivity. Therefore, this practice is not indicated for tambaqui (C. macropomum) during this development phase. 8/9 CONCLUSION Tambaqui (Colossoma macropomum) juveniles subjected to short restricted-feeding cycles for 60 days exhibited metabolic and hematological alterations and lower productive performance and used endogenous energy reserves to maintain their blood glucose level.
Continuously feeding for 65 days prompted partial compensatory growth and reestablishment of the metabolic and hematological parameters of tambaqui juveniles subjected to restricted-feeding cycles, though no muscle lipid deposition occurred. However, restricted feeding for short periods followed by refeeding did not prove to be economically viable for rearing tambaqui (Colossoma macropomum) fingerlings.