Aquaculture provides more than half of the world’s fish (FAO, 2016) but better feeding strategies are needed in order to keep it sustainable. One option is to use natural feed additives instead of artificial antibiotics, which cannot be used as growth promoters in the European Union (EC, 2003) since they promote microbial resistance. Arthrospira platensis is a natural product that can improve the quality of fish production by providing a rich source of vitamin B ₁₂ and β-carotene (20 times more than carrots), essential amino acids, fatty acids (3 and 6), minerals (James et al., 2006), protein (approx. 65% dry weight; Phang et al., 2000), as well as anti-oxidant properties. For all of these reasons, it is commonly used as a natural supplement in animal feeds, including for fish (McCarty, 2007). Less is known about how A. platensis can improve the condition of weaker fish or their response to stress.

Several studies have considered the effects of A. platensis supplementation (less than 10% of diet) in fish. Red swordtail ( Xiphophorus helleri) given 8% A. platensis increased feed consumption, body weight, length, gonad weight and number of offspring (James et al., 2006). In three-spot gourami ( Trichopodus trichopterus), Khanzadeh et al. (2016) found that 5% A. platensis (replacing fishmeal) improved feed intake, feed conversion ratio (FCR) and the gonadosomatic index. Rainbow trout ( Oncorhynchus mykiss) fed with 10 % A. platensis (a natural pigment source) deposit more carotenoids in muscle tissue (Teimouri et al., 2013). Adding A. platensis to feed for guppies ( Poecilia reticulata) improved their tolerance to a toxin (methyl red), manifested by a noticeable reduction in the cytotoxic effects on red blood cells (Sharma et al., 2005). In Oreochromis niloticus, adding 0.5, 1 or 2% A. platensis to feed improves its immune system (Abdel-Tawwab & Ahmad, 2009), has antioxidant effects and promotes growth (Takeuchi et al., 2002), but less is known about its possible effect on behaviour under stressful situations.

Regarding possible behavioural effects, A. platensis contains many different nutrients, including tryptophan and B6, which decrease psychological distress in humans (Shor-Posner et al., 1994). Fatty acids 3, also present in A. platensis, can inhibit adrenal activation elicited by a mental stress in dogs (Delarue et al., 2003). Thus, supplementing O. niloticus diet with A. platensis could change their stress responsiveness. In their natural habitat, O. niloticus normally consumes algae (including cyanobacteria) and some invertebrates (Ibrahem et al., 2013).

In the past decade, there have been many studies on fish welfare (Huntingford et al., 2006), and several methods have been proposed to measure stress, including non-invasive behavioural methods to measure stress coping styles using a two-choice test (Laursen et al., 2011). Barreto & Volpato (2011) found that O. niloticus can be sorted into two opposing coping styles under stress, namely proactive (dominant) or reactive (passive). It is assumed that the development of a coping style depends on the genetic makeup of the fish and its previous experience (Sørensen et al., 2013), but less is known about possible effects of nutrition. In this study we set out to verify whether a 1% A. platensis supplementation affected growth and survival as well as the behavioural response to stress using the two-choice test.

At the end of the trial (50 d), we performed a two-choice test based on Laursen et al. (2011), built using two circular tanks that were identical to the grow-out tanks (120 L) and attached to one another via a closable gate, 16 cm in diameter (Fig. 1). Tank A was tightly covered with black plastic (shaded) and Tank B was illuminated (natural light). Twelve two choice tests were carried out, using six tanks per treatment, discarding the tanks with less than 20 remaining fry (for A. platensis fry, where all tanks had over 20 fish, the six tanks were chosen randomly). For each test, all the fish in a tank were placed in Tank A at the same time and left to acclimate for 30 min. Then the door to the second tank (Tank B) was opened, and Tank A was bubbled with nitrogen gas to decrease oxygen levels (at a rate of 1 ppm per 10 min). Tank B was oxygenated to maintain normal oxygen levels (approx. 7 ppm) and a counter-current water flow was set up in each tank to avoid water flowing between tanks. Oxygen levels were monitored in both tanks every 5 min. Each test was recorded with a video camera to avoid the effects of any human disturbance. We noted the times when any fish left Tank A. The test ended when the first five fish had left Tank A or after one hour had passed. Then we measured the length of the fish in Tank B.

All statistical tests were carried out using R.commander (R Core Team, 2013). According to a Shapiro/Wilk test, the data on growth and survival were normally distributed and were analyzed using a one-way ANOVA. The data from the two-choice tests did not follow a normal distribution and were analyzed using a Kruskal-Wallis test. The level of significance was p‹0.05 in both cases.

The main goal of fish nutrition is to provide a balanced mixture of ingredients to support vital functions at an acceptable cost. Following this line of thought, at the moment it would not be cost-effective to use A. platensis (≈40 €/kg) as a protein substitute, but it may be useful as a nutritional supplement since it improves feed efficiency, carcass quality, and physiological response to stress in several species of fish (Takechi et al., 2002; Abdel-Tawwab & Ahmad, 2009; Velasquez et al., 2016). In the current study, adding 1% A. platensis to feed did not affect production indices (FCR, growth) compared to controls. This contradicts previous studies such as James et al. (2006), using Xiphophorus helleri (8% supplementation) and Abdel-Tawwab & Ahmad (2009) (0.5% to 1% supplementation) who observed improved growth. However, Takeuchi et al. (2002) found that the specific growth rate and feed efficiency of Arthrospira sp.-fed Orechromis sp. were lower than that of controls. Our findings of a 9% higher survival rate are in accordance with other studies that report 6% higher survival at 1% Arthrospira sp. supplementation after a bacterial challenge (Ibrahem et al., 2013). However, Abdel-Tawwab & Ahmad (2009), and Ungsethaphand et al. (2010) did not observe differences in survival among fish fed with Arthrospira sp. compared to controls. Our study may be different, however, due to the size of the fish. When the trial began, tilapias were in the fry stage (around 2-3 cm body length), when they begin to change from being more omnivorous to carnivorous, at about 6-7 cm. After that they change back again to more phytoplanktivorous lter feeding (Ibrahem et al., 2013). The change to carnivorous feeding could increase hierarchical stress and cannibalism, reducing the survival of weaker fry (Berrios & Snow, 1983). In this scenario, the high vitamin content of A. platensis, among other possibilities, could have played a role, helping to promote growth and decrease mortality.

Regarding the two-choice test, our results suggest that more of the fry consuming A. platensis supplemented feed adopted a proactive coping style, since more of them remained in a stressful environment despite the possibility of escape. As suggested by Barreto & Volpato (2011) the fish that remain in the "stressful" tank have a lower reaction to stress and can be classified as proactive. Although several recent studies have assessed coping styles in fish ( e.g., Ferrari et al., 2016), few have considered the effects of nutritional supplementation. Some authors have shown that A. platensis affects plasma cortisol and glucose levels. In rainbow trout ( Oncorhynchus mykiss) cortisol and glucose significantly decreased with increasing levels of inclusion of A. platensis at 0 to 10% (Yeganeh et al., 2015). In great sturgeon ( Huso huso), Adel et al. (2016) found that blood glucose increased at a higher inclusion of A. platensis (10%). Furthermore, according to Delarue et al. (2003), adding 3 fatty acids to fish diets can inhibit adrenal activation elicited by psychological stress. Arthrospira sp. has a high level of unsaturated long chain fatty acids such as 3 and 6, which may have provided a similar effect in O. niloticus. In this sense, Anzola (2013) found a decrease in psychological stress with higher tryptophan intake in dogs, and Shor-Posner et al. (1994) found the same effect in humans with tryptophan and vitamin B6 supplementation. A. platensis contains tryptophan (1.5% DM) and vitamin B6 (0.8% DM). Taking into account that the tryptophan requirement of O. niloticus is 1% of the diet, with the inclusion of 1% A. platensis, it was increased to 1.15%. The vitamin B ₆ requirements in juvenile Oreochromis sp. are 15 to 16 ppm of the diet (Shiau & Hsieh, 1997), but in the current study that was increased to 80 ppm (500% more than estimated control levels) by adding 1% of A. platensis.

In fish, body length is a strong predictor of individual positions in a hierarchy (Ward et al., 2006). Large fish are presumably more proactive, dominating other individuals, taking prime feeding sites and aggressively excluding smaller subordinate competitors (Webster & Hixon, 2000). In our study, although it appeared that smaller fish left Tank A first, we did not find significant differences with regard to fish size. However, the control fish left Tank A in groups, as compared to the fish fed A. platensis who left individually, which supports the idea that when fish which feel threatened they seek the safety of others. Finally, although it was not our aim to measure resistance to hypoxia, the results also suggest that SPR fish were more tolerant to low oxygen levels. This is in accordance with other studies in humans, where Arthrospira sp. supplementation increases cardiovascular capacity (Kalafati et al., 2010). Moreover, it has been found that Arthrospira sp. increase hematocrit levels of O. niloticus (Ibrahem et al., 2013), which improves the efficiency of oxygen consumption. This finding is important for the aquaculture industry since oxygen levels are often a limiting factor for production (Abdel-Tawwab et al., 2015).

In summary, supplementing O. niloticus fry with A. platensis strain PCC 9108 increased survival and the proportion of individuals that adopted a proactive coping style, based on the results from a two-choice test. The results also suggest that A. platensis increased resistance to hypoxia, which is important for intensive aquaculture production.