The main purpose of this research was to assess the feeding behaviour and ability of Nile tilapia and gilthead seabream to self-select their preferred diets. Fish choose feeds and regulate their feeding behaviour based on homeostatic and hedonic mechanisms. The homeostatic pathway maintains normal energy balance homeostasis and it takes over in response to nutritional demands and metabolic needs. The hedonic control is related to the brain reward system, where fish mainly sense the orosensory, pellet quality and palatable aspects of feed, being independent on nutritional requirements (Kulczykowska and Sánchez Vázquez, 2010; Volkoff, 2019). Thus, fish can choose which feed items to ingest mainly based on size, palatability and nutritional properties (Raubenheimer et al., 2012). All feeds offered had the same size, thus fish choice was only based on the other two parameters. For this reason, fish tested with self-feeders and diet encapsulation, which enabled the isolation of palatability and nutritional parameters, allowed for a better understanding of the effects of “nutritional wisdom”.
Fish feeding is modified by both Pavlovian (to find the feed) and operant learning (to catch and manipulate the feed) behaviour (Millot et al., 2014). Operant learning is the process that is associated between a behavioural action and its outcomes. If fish are rewarded enough times, fish will learn the relationship and will increase the probability and frequency of repeating the same action. In a population, there may be only one dominant fish that is curious enough to pull the trigger, but if rewarded, this information may be socially transmitted and become more common in all individuals (Millot et al., 2014). On the initial days of the self-feeding experiment tilapia showed a preference for one feeder, which changed after some time after assessing the content of the other feeders, demonstrating their exploratory and learning behaviour, as shown by Figueiredo et al. (2023). A similar situation was recorded using European seabass, that when fish were fed with the standard diet they exhibited a preference for one of the self-feeders (Aranda et al., 2000). Although tilapia kept in groups developed strong social hierarchies that could affect feeding behaviour (Toguyeni et al., 1997), this was not observed as fish did not show an aggressive behaviour and still choose a feed on both types of experiment and regardless being kept alone or in groups. Similarly, Fortes-Silva et al. (2012) showed that tilapia were not aggressive and a exhibited a similar pattern of diet selection whether maintained isolated or in groups, when offered diets with a balanced or unbalanced composition of essential amino acids. Indeed, in our self-feeders experiment, tilapia were not isolated as in diet encapsulation, meaning that social learning may not play a significant role in the feeding and behavioural mechanism as suggested by Vivas et al. (2006). Using self-feeders, Nile tilapia chose diet ORG2 with an intake 0.75% of fish weight/day. Similarly to other studies, it was possible to observe that tilapia consumed most diet pellets and that the feed waste (less than 2% of the total given feed) that remained in the tanks was residual. For example, Fortes-Silva et al. (2012) reported a negligible food waste of 1% with tilapia. To compare the intake, Pratiwy et al. (2017) tested the growth performance of Nile tilapia reared under self-feeding systems and showed feed intake values of around 1.85%/body weight. The choice for diet ORG2 was presumably based on tilapia nutritional needs (post-ingestive) coupled with feed organoleptic characteristics. Likewise, other studies with European seabass (Dicentrarchus labrax) and tilapia reported a similar behaviour (Fortes-Silva et al., 2016, 2012; Rubio et al., 2006). It is important to note since fish required almost three weeks to exhibit a preference and no feed was predominantly chosen from the beginning of the study, this can reflect the less clear taste differences between feeds. In a study by Carlberg et al. (2015), Arctic charr (Salvelinus alpinus) took 9 days, while Fortes et al. (2010) noted that tilapia clearly preferred since the beginning of the experiment diets containing phytase with an intake and that after switching feeds, the pattern was re-established only after 3 days. In the present study, after feed was switched between feeders, tilapia also resumed, re-established and sustained the previous pattern of selection of diet ORG2, while maintaining a constant consumption of other diets, meaning that the fish established levels of consumption for each feeds. However, once again they took some time (9 days), pointing out the effect of the minor differences between the diets. These findings are in accordance with Fortes et al. (2010), who reported that a diet with 1500 IU kg− 1 phytase was preferred throughout the trial, even after switching feeders. The same author showed that tilapia retained and maintained the intake of specific levels of protein, fat and carbohydrate when feeds were switched over (Fortes-Silva and Sánchez-Vázquez, 2012). Fish are able to identify and evaluate distinct amino acid profiles between diets (Fortes-Silva et al., 2012). All feeds were formulated to contain the minimum requirements of every essential amino acid (EAA). However, diet ORG1 presented the lowest levels of methionine. Although methionine was near the lowest requirement in ORG1, it still was enough to fulfil the species physiological state for normal growth (NRC, 2012). Overall, the self-feeder experiment with tilapia resulted in a preference for diet ORG2, which was influenced by a combination of several factors including learning-reward behaviour, nutritional requirements, as well as the orosensory properties of the diets.
To discard the effects of olfactory factors on the selection of the feeds, Nile tilapia were fed encapsulated diets in a second experiment. Tilapia preferred diet C (with spirulina and quinoa), after 11 days and after colour rotation, and presented an intake of 0.51% of their body weight. While individual differences between fish of the same species can exist, in our case the preference was unanimous for all fish. The same species with a final body weight of around 53g required 15 days to distinguish between two vegetable oil blends at 30ºC and after switching capsules’ colours only 3 days, with an average intake of 1.12% between all treatments (de Almeida et al., 2021). In another study, the feed intake of tilapia with around 80g was 1.36% of the animal’s biomass and required 7, 11 and 23 days to differentiate diets with distinct protein levels (0, 25 and 42%) and 6, 10 and 4 days after inverting the content of the capsules (Costa et al., 2022). Capsule colour used in this study did not affect our results. In other studies, using tilapia, by Fortes-Silva et al. (2011) and de Almeida et al. (2021) coloration also did not affect fish preference and only served as a “reference” for the fish to identify which diet was associated with each colour. Therefore, in the present study it was clear that by randomizing capsules’ colours for each individual and after changing the colour paired to each diet, tilapia continued to choose capsules with diet C, meaning that the content, rather than the capsule colour, was the main factor considered for their preference. Tilapia is able to address the content based in a wide range of physiological processes. All capsules presented the same chemosensory properties at oropharyngeal level, meaning that palatability, texture, flavour and odour associated with traditional pelleted diets were negligible for the fish (Rubio et al., 2006). Moreover, it is important to note that neither the capsules or pellets leached nutrients into the water, which could act as attractants to fish (Busti et al., 2022). Therefore, fish had to evaluate the quality and nutritional composition of the content of the capsules and learn to associate this information with the capsules' colour (Almaida-Págan et al., 2006; Rubio et al., 2003). Accordingly, tilapia opted for the diet C, which after evaluation, was probably more able to satisfy their physiological state, based on post-ingestive and/or post-absorption processes as suggested by Fortes-Silva et al. (2016) and Rubio et al. (2003). However, it is important to note that in the present study, besides the constituting ingredients, the formulations of the three experimental diets, mainly protein and lipid, were different, thus they could have affected fish feeding behaviour. For the experimental size (around 300g) of Nile tilapia used, the recommend percentage of these macronutrients on the feeds is between 32–36% for proteins and 8–12% for lipids (NRC, 2012). Our results were in accordance with Pereira-da-Silva et al. (2004) that registered a crude protein feed intake of 24% by Nile tilapia, when given the possibility to self-select between distinct protein dietary levels. Although diet A (mainly with casein and dextrin) and B (rich in spirulina) had more similar nutritional profiles to these values than diet C (mixture of spirulina and quinoa), the latter was still preferred. Since fish had at their disposal other feeds, they also consumed them in a different proportion, possibly to create a balanced diet and compensate the lack of some essential nutrients from diet C (Simpson and Raubenheimer, 2001). A similar situation was found by Costa et al. (2022), who reported that when tilapia of 80g were offered two rations, with different crude protein and amino acids levels, it showed a significant preference for the consumption of one of the feeds, while also eating at lower levels the other one. In the present study, the intake of diet C, which had the lowest protein content (23.93%) increased, could mean that this defect was mild for the fish, perhaps due to their size, so they increased its intake, and possibly also consumed other feeds available, to reach their level of requirement, especially in the case of protein (Henry, 1985). This scenario was also confirmed by Fortes et al., (2011), who noted that when protein was restricted, tilapia increased its intake by consuming more capsules with this nutrient to maintain their energy intake (Fortes-Silva et al., 2011). Similarly, Almaida-Págan et al. (2006) and García-Meilán et al. (2013) showed that when given a diet with lower protein to sharpsnout and gilthead seabream, respectively, the fish increased its intake. Indeed, studies have shown that fish have the ability to regulate its consumption and defend a given nutritional intake target (Almaida-Págan et al., 2006; Brännäs and Strand, 2015; Fortes-Silva et al., 2016; Rubio et al., 2003). Nile tilapia exhibited a consistent preference for diet C driven by post-ingestive and post-absorption processes, even in the absence of olfactory factors present in the self-feeder experiment, reflecting that they are able to identify and select a feed with spirulina and quinoa as functional ingredients.
Conversely to tilapia, gilthead seabream did not show a consistent preference for a particular feed using self-feeders. The experiment was performed with rapidly growing juveniles at high water temperature in summer (since it was a flow-through aquaculture system), a scenario which could have made that the homeostatic system, associated with high energetic demands, override the hedonic regulation of feeding behaviour, not allowing seabream to discriminate diets efficiently (Puchol et al., 2022). Another possible explanation for the lack of diet discrimination is because all three experimental diets were nutritionally similar to the previously fed commercial feed, meaning that fish were familiar with it and did not notice enough differences (Pulido-Rodriguez et al., 2021). In our study, gilthead seabream wasted and rejected more feed than tilapia, while the intake was at an average of 1.21%/body weight. In addition, there was a much higher variation on the daily intake of feeds compared with tilapia, which could be related with the more curious and aggressive behaviour of seabream towards feed, especially when limited and defendable as in self-feeders (Puchol et al., 2022). There are few studies available regarding seabream using self-feeders with similar methodology to compare results. Nevertheless, in a previous investigation with seabream, it was shown that fish with 254g could select a diet with distinct oxidation levels of dietary lipids after 10 days with a preference of 82% and 7 days after switching feeders with an average intake of 1.57%/body weight (Montoya et al., 2011). Similarly, to what occurred with tilapia, it is possible that on the initial days seabream were preferring a specific feeder on each tank that, by coincident, contained diet ORG1. A specific feeder was also selected on the initial experimental days with European seabass and gilthead seabream by Aranda et al. (2000) and Montoya et al. (2011), respectively. The preference for a particular feeder can then indirectly affect fish’ choice based on pre-ingestive parameters (Montoya et al., 2011). Therefore, it was necessary to change the positions of the feeders to assert dietary preferences and avoid any preference for a specific position as it was noted by Puchol et al. (2022). Indeed, after changing the position of the feeders, seabream decreased their intake for ORG1 and never achieved a clear preference for any of the given feeds, suggesting that both pre- and post-ingestive signals were involved in diet selection (Montoya et al., 2011). Montoya et. al. (2011) observed two selection patterns after changing the position of the feeders: some fish groups resumed their selection for a specific diet, while the other groups did not show a clear preference for any diet until they were subjected to a 3-week fasting period, after which they shortly resumed their dietary preferences. In the present study, seabream were also fasted, for a duration of 10 days in order to present them with a challenge, aiming to define a feeding pattern, as the physiological state of fish caused by oxidative stress due to fasting would reinforce their selection behaviour (Montoya et al., 2011). However, seabream were not observed to demand more feed and define a pattern, as the compensatory bite activity increase was not enough to attain a sufficient feed intake level to reach a preference. Conversely, European seabass fasted for 6 and 15 days, increased demand for diets to recover metabolic status with hyperphagia, mainly for protein used as energy source (Aranda et al., 2001; Vivas et al., 2003). Since seabream were unable to exhibit a preference for any feed, it was not necessary to conduct a capsule experiment.
The feed intake and growth rates obtained in all our studies were in general lower compared to other performance experiments, as it was expected. It should be noted that fish sizes differ among experiments, which in turn directly affects their intake requirements and growth rates. Moreover, the diets were not formulated with the goal of optimizing fish growth but rather to study fish behaviour. Indeed, experiments on dietary selection do not necessarily correlate the most selected diet with optimal performance (Fortes-Silva et al., 2012). Even in growth experiments, although a diet is formulated to provide maximum growth, when given the opportunity, fish might not prefer that formula (de la Higuera, 2001). When using capsules, tilapia “feel” that it is a “stranger” method of feeding, which can lead to a decrease in the intake. It is important to note that although tilapia were isolated in the diet encapsulation experiment, it should not represent a factor of stress that would have affected feed intake (Fortes-Silva et al., 2012). Despite it has been seen that animals in a self-feeding scheme can perform well (in terms of weight gain, dietary intake, etc), in some species, these method can impair growth and decrease feeding efficiency (Gélineau et al., 1998; Montoya et al., 2011; Santos et al., 2019; Tidwell et al., 1991). These differences are related to the adaptation of self-feeders by fish, where some animals in the same group assimilate the self-feeding system more than others (Ferrari et al., 2014; Tidwell et al., 1991). Nevertheless, the lower performance indicators obtained in our experiments were not a concern, especially as no mortality occurred and fish still gained weight.