Modeling the Effects of Water Temperature on Growth Rates, Gastric Evacuation and the Return of Appetite in Juvenile Nile Tilapia, Oreochromis niloticus L

Optimized aquafeeds have long been a major concern of the sustainable aquaculture development. Not only should the feed composition meet the nutritional requirements of the fish, it should also be reasonably managed (feed ration and feeding frequency) to enhance the feed utilisation efficiency, growth performance and decrease the amount of wastes. At present there is no detailed information on how rearing temperature impacts gastric evacuation rate, return of appetite (RA) and daily feed ration among tilapias, considered as one of the leading fish species for worldwide aquaculture production. The objective of this study was to develope mathematical models to estimate maximum daily feed intake for Nile tilapia in relation to feeding frequency and water temperature. Growth was measured in 480 fish (initial body mass 4.30±0.02 g) fed in slight excess, following their exposure to four thermal treatment (22, 26, 30 and 34 °C) (four replications per treatment, 21-days rearing period, growth monitoring at 5-days intervals). Gastric evacuation and return of appetite measurement were made by radiographic technique. A growth model was developed using a stepwise multiple-regression analysis against fish body mass and water temperature (r = 0.939, df = 15) as follow: SGR (%M/day) = -70.606 + 98.433 Log T° – 33.762 (Log T°) 0.153 Log M (Log T°). The gastric emptying was described by an exponential function, which was found to be inversely related to the RA. The instantaneous evacuation rates (Re) determined by linearizing the data were strongly affected by rearing temperature (Q10 = 0.047) in Nile tilapia. The RA following a satiation meal was also significantly dependent on rearing temperature. Based on these data, the maximum daily feed consumption was estimated in relation to feeding schedule for juvenile tilapia reared at different temperatures. Considering that unsuitable use of feed adversely impacts on the farm revenue and profit, these results contribute to improve feed management strategies.


Introduction
In poikilotherms like fishes, rearing water temperature is one of the most important factors affecting growth performance and other physiological parameters (e.g., digestion, absorption, catabolism, anabolism and excretion). For several of fish species there is detailed information on how temperature affects growth, feed intake, digestibility and feeding behaviour (e.g., Azaza et al., 2008Azaza et al., , 2010aHandeland et al., 2008;El-Asely et al., in press). Generally, he metabolic processes of fish are sensitive to changes in environmental temperature and if water temperatures depart from the optimum, fish eventually lose appetite, reduce feed intake and growth decreases.
For intensive rearing of a fish species to be successful, it is essential that aquafeed be appropriate and sustainable, as feeds represent the main contributor in the production costs, may present from 50 to 60% of total variable expenses (Azaza et al., 2015). It should improve feed efficiency, promote fish growth and lowest feed wastes. There have been increasing concerns over the sustainability of fish feed, in particular as regards the substitution of costly ingredients. These approaches are prerequisite, but they may not enough, since several other parameters (Houlihan et al., 2001). In fact, adequate feeding strategy is essential not only in terms of improvement of fish growth but also in terms of reduction of feed cost and environmental impacts.
In fish farming practice, it is important to have information on the rate at which eaten feed is processed in the digestive tract. This is controlled by the storage capacity, rate of digestion, and time taken to the return of appetite (Houlihan et al., 2001). The processes of gastric digestion and evacuation are complex, depending upon the interrelationship of many factors such as temperature, fish mass, meal size, dietary composition, feeding frequency and feed particle size (Bromley, 1994;Pääkkönen & Marjomaki, 1997;Koed, 2001;Azaza et al., 2010b). Of these factors, water temperature is perhaps the predominant factor affecting the feed conversion efficiency and growth of the fish (Azaza et al., 2008). It has been demonstrated that hunger in fish is, particularly, determined by the amount of feed in stomach tract (i.e., available space in the stomach), at the same time, the return of appetite is strictely related to the rate of gastric emptying (Lee et al., 2000). Determining the gastrointestinal evacuation rate can allow one to predict the return of appetite under a given set of conditions and thus to develop efficient feeding strategies for farming operation to be porofitable. Therefore, estimation of the gastric evacuation rate (GER) is a prerequisite for modeling of daily feed consumption in fish (Azaza et al., 2010b).
little information is available on the GER of O. niloticus, these studies have investigated GER in relation to feeding frequency (Riche et al., 2004) and feed particle size (Azaza et al., 2010b). At present, insuffisant information on how rearing temperature affects GER, return of appetite and daily feed ration among tilapias. Consequently, the present study aimed at developing general mathematical models for growth, evacuation and return of appetite in O. niloticus based on water temperature, using radiographic tool. Through this study was to demonstrate the useful of this latter technique to elaborate prediction models which are crucial for planning fish production and refining the feeding schedule in intensive culture operations. This which may assist in the development of intensive culture strategies of this species.

Fish and Experimental Rearing Conditions
O. niloticus fingerlings of Maryut strain of both sexes, weighing 4.30±0.02 g fish -1 (n = 480, mean±SE), were selected from a large population maintained at 28-30 °C at the fish-culture research station of the Tunisian National Institute of Marine Sciences and Technologies (INSTM). Fish were sedated with tricaine methanesulphonate (50 mg L -1 ), weighed individually to produce 16 groups of 30 fish each, in which mean mass and size heterogeneity (expressed by coefficient of variation of fish mass, CV < 8%) were as similar as possible. Thereafter, the groups of fish were randomly allocated to the different thermal treatments (22, 26, 30 and 34 °C) and tested in quadruplicate for 3 weeks. According to Azaza et al. (2008) this period allowed reliable recordings of growth of juvenile O. niloticus reared at 22 °C.
The four thermal regimes were evaluated in four indoor water recirculating systems in the rearing facilities of the Aquaculture Research Station, previously described in Azaza et al. (2008Azaza et al. ( , 2010b. Each system comprised four 85 L rearing aquaria (35 × 80 × 45) cm 3 , a 180-L reservoir tank for filtration. In order to maintain the temperature of the water, a 2-kW thermostatic immersion heater was used. A 20-L biofilter with UV-sterilization lamp (Oase, model Filtoclear UWC 9/11W; www.pondsupplies.com.au) was also used for water treatment.
Faecal matter removal was accomplished by siphoning out the sedimented feacal matter and uneaten feed (if any). Also, a submerged filtration (Rena, Filstar) was installed in each aquarium to improve filtration process. A light/dark cycle of 12:12 h (08:00-20:00, light period) was maintained, by an automatic timer (Time switch, CHNT), throughout the experiment with four fluorescent lamps suspended over the aquaria and provided about 800 lux (Digital Lux Meter, Digital Instrument LX-101) light intensity at the surface of the water. Before the start of the experiment, fish were kept for one week to acclimatize them to the rearing environment. During this period water temperature was maintained circa 28°C in all aquaria, and all dead or apparently stressed fish were substituted. Following acclimation, a 21-day thermal treatment trial was conducted. Experimental target water temperatures were attained progressively by heating or cooling water at 1 °C h -1 . The experiment was conducted during winter, when the air temperature inside the rearing facility was cool enough to enable to maintain easily the lowest experimental temperature (i.e., 22 °C). In all aquaria, water flow was adjusted to 1.5-2.5 L min -1 to provide oxygen and remove excess nitrogenous wastes. Air stones were placed within each aquarium, to maintain adequate dissolved oxygen levels, if necessary.
The temperature of each aquarium was recorded using temperature loggers (Tidbit® v2 Temp Logger, Onset, USA). For water quality monitoring, dissolved oxygen and pH levels were recorded automatically (every hour) with a digital thermo-oxymeter surveillance system (WTW, MIQ/C184, www.memecosales.com; accuracy of 0.1 °C and 0.1 mg O 2 L -1 ). Total ammonium and nitrite were measured on the days of fish measurement (days 6, 11, 16 and 21) by standard methods (APHA, 1995).
Fish were fed with formulated feed (350 g/kg crude protein, 99 g/kg fat and 17.40 kJ g -1 gross energy) which stimulate good growth performance (Azaza et al., 2015). The feed was prepared as described by Azaza et al. (2009). Fish were hand-fed to apparent satiety four times daily during the week (0800, 1100, 1400 and 1700 h). Pellets were distributed slowly, allowing all fish to eat without feed wastage. Apparent satiation was considered achieved when the fish would no longer accept the offered feed after a period of active feeding. On days 6, 11, 16 and 21, all fish in each aquarium were individually weighed. Fish were captured with a dipnet, sedated with MS-222, 100 ppm, (tricaine methanesulfonate) in order to reduce stress and to improve the accuracy of the weighing, and then weighted individually to the nearest 0.01 g and returned to their aquarium. On the weighing days, feeding fish was suspended during the morning, and resumed in the early afternoon, in order to enable all fish to recover from handling stress.

Gastric Evacuation and the Return of Appetite
For the gastric evacuation measurement, the feed was similar except that X-ray-dense lead glass beads (Ballotini, type H, 450-600 μm diam, DLO, Braine-L'Alleud, Belgium) were mixed (1.5%, M M -1 ) with the ingredients before compressing them into pellets using a meat grinder. Marked feed was prepared as described in Azaza et al. (2010bAzaza et al. ( , 2013. Finally, the moist pellets were sun-dried and conserved in sealed containers at -20 °C until use. Thirty-five samples of marked feed of known masses (0.05-1.5 g) were X-rayed to determine the relationship between the masses of particles feed and the number of Ballotini (Nb); (feed mass (g) = 0.0073 Nb + 0.0037; with r 2 = 0.974; n = 35. This enabled the amount of feed in the gastro-intestinal tracts of fish to be back-calculated from X-ray plates.
The stomach evacuation experiments began at the end of the growth trials because its protocol was potentially more stressing for the fish and required a 2-day period of feed deprivation, which would have impacted on fish growth. Fish were deprived of feed for 48 h to ensure complete gut emptying. On day 23, the fish in all aquaria were fed to satiation with Ballotini marked feed. Labeled feed was presented to the fish in the same manner as the standard diet. Just after the last pellets were delivered (time zero), and every 2 hours thereafter, six fish were randomly sampled in each aquarium, anesthetized with tricaine methanesulfonate (100 ppm), X-rayed (see below), weighed (nearest 0.01 g), then tagged to avoid the fish being sampled again, before at least 6 hours. The decrease in the amount of marked feed in the gastrointestinal tract over time was used to estimate the gastric evacuation rate (GER). After a deprivation period of 36 h to ascertain that evacuation was complete in all groups, a satiation meal with unlabeled feed was offered to all aquaria. Then the return of appetite was measured every 2 hours (0-18 h) after the satiation meal, by offering a labeled diet (same as for the gut emptying experiment) until apparent satiation. Six fish from each aquarium were randomly sampled for X-ray photographs (same protocol as above). The increases of the amount of ingested marked feed with increasing deprivation time were served to estimate the return of appetite (RA).

X-Ray Protocol
The radiographs were determined using the X-ray protocol described in Azaza et al. (2010b) and Azaza et al. (2013). The radiographs were taken using GE AMX 110 X-ray machine and Kodak Ma film (Kodack, X-OMAT MA), the exposure time was 2 s at 2 kv. An example of a typical radiograph is given in figure 1 to show the number of Ballotini. Stomach content was estimated by counting the radio-opaque Ballotini which easily identified on X-ray photographs. Following the development of X-ray plates, two independent observers counted the number of radio-opaque beads in the gastrointestinal tract of each fish. Counting of glass bead is facilitated by viewing developed plates on a light box. The gastrointestinal tract contents were calculated in terms of percent body mass following the relationship between feed mass and the number of glass beads.

Water
No critical ammonia-N recorded w
The relati illustrated temperatur T°, r 2 = 0.9 showed th the 34 ºC g after 18 ho

Discuss
A high sta mass and w any set of decreases be mainta consumpti oxygen co feed intake L -1 . In the above 5.11 org Regression equ a satiation me  Vol. 12, No. 8; argued whether feed quality effect or whether changing the energy content and protein levels influenced the results. The influence of feed quality was probably minimal since in the diet used in this study, dietary protein and energy content were enough to support maximum growth for O. niloticus in the range size tested in this study.
A multitude of models have been developed to describe gastric emptying in fish (Bromley, 1994). The developed regression analysis of the gastric evacuation time in the present study follows an exponential regression function. Previous studies corroborhate that the exponential model is the most appropriate to describe the evacuation of small digestible feed particles (Jobling, 1987). This relationship has also been demonstrated in O. niloticus (Riche et al., 2004;Azaza et al., 2010b), Oreochromis mossambicus (De Silva & Owoyemi, 1983). Enzyme activity follows an exponential function and is obviously affected by temperature. Therefore, it is possible that the exponential regression may be more adequate for O. niloticus fed on particle feed and reared in warmer temperatures (Brodeur, 1984).
GERs (about 0.089% h -1 ) and GETs (approximately 20-22 h) for 14 g-fish reared at 30 °C are slower than this reported by Riche et al. (2004) which show that O. niloticus (183 g) take circa 18 h to empty their stomach at 28 °C. Some of the difficulty in comparing studies can be attributed to the different experimental conditions or analytical procedures involved. Jobling et al. (2012) reviewed some of the factors that are thought to influence these rates. It concludes that feed type may often affect the evacuation rate through the efficiency of its digestibility. On the other hand, feed particle size is an important factor that affects emptying in catfish (Hossain et al., 2000) and O. niloticus (Azaza et al., 2010b). As observed in other fish species such as Pikeperch (Koed, 2001) and Atlantic salmon (Handeland et al., 2008), the evacuation rate of juvenile O. niloticus was strongly temperature-dependent. The influence of temperature on gastric evacuation was examined by plotting the instantaneous evacuation rates (Re) against the experimental temperatures and the results showed that evacuation rates increased with increasing temperature and varied significantly among test temperatures (F 3,63 = 5.68, P < 0.001). The Re increased from 0.047 to 0.104% h -1 between 22 and 34 °C. A Van't Hoff's Q 10 value of Re over the temperature range of 22-34 °C was 0.0475, which indicates that the metabolic rate of O. niloticus (6-14 g) nearly double for a 10 °C increase of temperature. Pääkkönen and Marjomäki (1997) calculate a Q 10 value of 2.89 for burbot Lota lota, which is circa six-fold of the value obtained here for juvenile O. niloticus. This indicates that the burbot is more sensitive to water temperature than O. niloticus and confirms the eurythermic specificity of this latter species. This parameter seems to be depending on fish species due to the difference in metabolic process and growth caracteristics (Brett & Groves, 1979). The relationship of Re vs rearing temperature follow a logarithmic pattern (GER = 0.128·Log(T°) -0.345; R 2 = 0.981, n = 5). Based on this relationship (equation), the the temperature for O. niloticus at which Re is zero (the regression line passes through zero) was estimated, thus theoretically fish feeding is stopped. This analysis suggests that the temperature threshold of ceased voluntary feed intake is 18.5 °C for O. niloticus of masses between 6 and 14 g.
In salmonids, previous studies show that GER increases exponentially as temperature increases (e.g., Sweka et al., 2004), and that GER depends primarily on rearing temperature and feed intake act as a secondary factor (e.g., Kawaguchi et al., 2007;Azaza et al., 2010b). It has been shown that water temperature may affect feed utilization efficiency and growth performance by affecting the gastrointestinal transit time (Dos Santos & Jobling, 1995). Fish submitted to higher water temperature (34 °C in the present study) evacuated more quickly, resulting from a temperature dependent increase in gastric peristalsis. In this situation, feed is attending the intestine without engouh gastric digestion thus decreasing significantly the enough time for digestion, possibly negatively affecting nutrient absorption efficiency (Jobling et al., 2012).
In aquaculture, much attention should be accorded to the return of appetite after a satiation meal (Hossain et al., 1998;Riche et al., 2004). During a meal, an animal achieved satiaty, when internal physiological signals act as negative feedback, which at a certain threshold will instruct the animal to cease eating (Langhans, 1999). In the time after this satiation meal (deprivation time) when the ingested feed is processed (digested and assimilated), the animal becomes hungry again (return of appetite). At this moment, fish re-fed to full the new empty space in their stomach (Jobling et al., 2012). Results of the present study show that the return of appetite in juvenile O. niloticus after a single satiation meal closely follows the gastric evacuation curve and when fish were offered feed, the fish adjust their intake, so that stomach fulness is maintained at near maximum fulness, i.e., gastric evacuation and return of appetite were inversely shaped. This relationship between GER and feed intake (i.e., return of appetite) has been demonstrated in O. niloticus (Riche et al., 2004;Azaza et al., 2010b) and is common to other fish species (Hossain et al., 1998).
stomach fulness is maintained. This implies that the two physiological processes (i.e., evacuation and return of appetite) occur at a comparable rate only in a range of temperatures around the optimal temperature of the species (28-32 °C). Lee et al. (2000) demonstrated that feeding intervals or feeding frequency is significantly correlated with gastric evacuation time, and the return of appetite is closely related to the rate of gastric emptying. In fish culture, making feed available at an appropriate rate and as soon as appetite has returned, can maximize feed intake and not obviously improve utilization feed efficiency. Besides, a suitable feeding strategy is essential not only in terms of improvement of fish growth but also in terms of reduction of feed cost and environmental problems. In practice, appetite combined with the daily number of meals, often named "the applied feeding frequency" or feeding schedule will allow an optimal feeding strategy to be predicted under a given set of rearing conditions and diets specificities. In this study, two parameters were used to develop appropriate feeding strategies in relation to water temperature and feeding frequency. Observations on the maximum daily feed intake in O. niloticus suggest a strong temperature dependent regulation of feed intake through the return of appetite. In fact, there was doubling of maximum feed intake of fish reared at 34 °C compared to those reared at 22 °C. For example, this analysis shows that, for a temperature of 30 °C, fish (circa 10 g) fed every hour for 12 hours consume 23% of their body mass; however, they consume only 14.6% of their mass if the meals were distributed every 6 hours. In the absence of parallel growth data, it is not possible to specify whether the growth is proportional to the feed ration, which is plausible, but probably not with a slope of 1.00, because it is evident that when GER increases, digestibility and feed conversion efficiency decrease. From a practical point of view, it is important to assess if the observed variations in maximum feed intake related to feeding frequency influence the feed conversion efficiency.

Conclusion
The results of the present study allow aquaculturists to select and adapt the suitable feed management based on the rearing temperature. This allows avoiding either under-feeding, which inhibits growth and enhances growth dispersion (Azaza et al., 2013), or over-feeding, which increases feed wastage and reduces feed conversion efficiency.