DISCUSSION
Using the new method of fatty acid-specific stable hydrogen isotope analysis, a clear compensatory mechanism based on n‑3 LC-PUFA biosynthesis and routing was demonstrated, potentially offering protection against neural impairment under a diet deprived of n‑3 LC‑PUFA (Lund et al., 2012). However, our results also clearly show that an n‑3 LC‑PUFA-deprived diet is suboptimal for brain development, for which there are three main indicators: 1) n‑3 LC‑PUFA-deprived trout needed to expend energy to biosynthesise DHA, shown by Δδ 2H depletion of DHA and high ARA concentration in most tissue/lipid types; 2) n‑3 LC-PUFA-deprived trout needed to route DHA to the brain at the expense of muscles, shown by DHA content differing by dietary treatment in muscle tissue but not brain; and 3) n‑3 LC‑PUFA-deprived trout had smaller brains. In short, even when the biochemical composition of brain PL was maintained through increased biosynthesis and allocation of DHA from muscle tissue, poor diets resulted in smaller brains, and trout with smaller brains performed less well in the behavioural trials.
Growth rates did not differ significantly between diet groups, which suggests that, as intended, there was no important difference in overall nutritional or energetic value between the two pellet formulae. However, it also shows, against our hypothesis, that groups raised on the low n‑3 LC-PUFA diet did not sacrifice somatic growth to fuel the energetic demand of fatty acid biosynthesis. Our study did not consider caloric intake, and it may be that ad libitum feeding allowed sufficient energy from the diet to make a sacrifice of somatic growth redundant.
Trout raised in complex habitats had smaller bodies than the trout raised in simple habitats, which appears to contradict previous findings of reduced competition via visual isolation associated with increased habitat complexity (cf. Sundbaum and Näslund, 1997; Koljonen et al., 2012). However, the size discrepancy may be explained by a decrease in aggressive, dominant strategies in the complex habitats, as territory size and resource monopolisation by dominants may be reduced (Höjesjö et al., 2004). The complex habitat tanks in this experiment were designed so that there were at least two hiding spots for each trout, minimising the effectiveness of dominant strategies; and the weekly partial exchange of tank inhabitants occasioned the regular collapse and re-establishment of dominance hierarchies. The lower density of trout in the complex habitats may also explain their smaller size: low stocking densities have been found in juvenile rainbow trout Oncorhynchus mykiss to induce chronic stress and lower feeding efficiency (Roy et al., 2021).
Sex also played a small role in size differentiation (i.e. somatic growth), which was surprising. Sex differences are normally unexpected until brown trout near maturity (Reyes-Gavilán et al., 1997), but the juveniles of the present experiment were much younger than that. Perhaps the unlimited food resources promoted growth which accentuated sex differences that ordinarily would not be apparent until later in ontology (cf. Riguad et al., 2023). In any case, the effect of sex upon any other aspect of the experiment proved minimal.
On cognitive performance, diet quality exerted its effect only indirectly. Its direct effects were on brain size and stimulation of n‑3 LC-PUFA biosynthesis and routing to brain polar lipids, which themselves played important roles in influencing cognitive performance. In contrast, the habitats in which trout were reared resulted in direct significant differences in cognitive performance between treatment groups. Although trout raised in complex habitats showed significantly better cognitive performance than those raised in simple habitats, as predicted, this was not because their brains were larger; nor did they show significantly different percentages of either EPA or DHA (except in muscle NL). Habitat complexity did not appear to stimulate biosynthesis of n‑3 LC‑PUFA, or their preferential allocation to the brain. We suggest that constant exposure to habitat complexity during ontogeny may continually reinforce interactions between existing neurons without requiring n‑3 LC‑PUFA for the formation of new neurons (vide Dorman et al., 2018).
Although nursery habitat played no role in encephalisation, counter to our prediction, the effect of diet followed the predicted pattern previously observed among wild brown trout: those trout with access to greater proportions of n‑3 LC‑PUFA in their diet had significantly larger brains than their lower dietary n‑3 LC‑PUFA counterparts (cf. Závorka et al., 2022b). However, save one exception, there was no difference in the relative size of any specific brain region, including optic tectum or telencephalon, between treatment groups, contrary to our prediction. We suspect that n‑3 LC‑PUFA routed to the brain was distributed proportionately to brain regions, but our study design, which analysed lipids in the whole brain, was unable to determine fatty acid contents of individual regions. The exception was the olfactory bulb, which showed an interaction effect of diet and habitat. Presumably there is an advantage in complex habitats to having heightened processing abilities of olfactory cues, although this may less important than other brain functions and, so, may be sacrificed when trout are subjected to n‑3 LC-PUFA scarcity. However, when raised in a simple habitat, it remains a mystery why trout fed a low n‑3 LC-PUFA diet should have larger olfactory bulbs than those fed the n‑3 LC-PUFA-enriched diet.
As predicted, trout with larger brains showed significantly better cognitive performance than smaller-brained trout in a task requiring ecologically important competences of recognition, memory and inference to de-escalate conflicts (Drew, 1993), in line with our prediction. However, no individual brain region had an effect on cognitive performance. Brain regions in teleost fishes are each involved in a variety of specific cognitive functions from learning and engagement in complex social tasks (telencephalon), through processing primary visual input (optic tectum), to spatial orientation and proprireception (cerebellum) (Kotrschal & Kotrschal, 2020). We suggest that all these functions may be needed together to confer the cognitive abilities required by the environments presented in this study. Therefore, total brain size was a better predictor of cognitive performance than any particular brain region.
This study found evidence that deprivation of dietary n‑3 LC‑PUFA stimulated trout to biosynthesise EPA and DHA from precursor fatty acids (such as ALA; vide Appendix Fig. A2). Converting short-chain to long-chain PUFA as a likely compensatory mechanism has been previously established in experimental rats (Rapoport and Igarashi, 2009; Rapoport et al., 2010) and humans (Barceló-Coblijn and Murphy, 2009; Domenichiello et al., 2015), reminiscent of patterns seen in the present study. Significant depletion of δ 2H in EPA and DHA in the trout raised on the low n-3 LC-PUFA diet compared to those raised on high n-3 LC-PUFA, without significant differences in the percentage of total brain polar lipids composed of those n‑3 LC‑PUFA, suggests compensation for deficiency in the diet.
The increased percentage of the n-6 LC-PUFA, ARA, across all tissue/lipid types amongst the trout fed low n-3 LC-PUFA may appear to be an overcompensation. Omega‑6 PUFA, and ARA in particular, are important for wound healing, inflammation, coagulation and osmoregulation (Castro et al., 2016), although they can also have negative effects by increasing risk of hyperinflammation (Layé, 2010). The abundance of ARA in n‑3 LC-PUFA-deprived trout is more likely to be merely a consequence of bioconversion. Neither ALA 18:3n‑3 nor LIN 18:2n‑6 can be synthesised de novo by vertebrates and must be obtained from food sources (Blondeau et al., 2015; Malcicka et al., 2018). However, they compete for the same elongases and desaturases to perform endogenous conversion to respective n‑3 and n‑6 LC‑PUFA (Sprecher, 2000). Although n‑3 fatty acids have been observed to be the preferred substrates for desaturase activity (Jeromson et al., 2025; Nakamura and Nara, 2004), this is not absolute; the conversion of n‑6 fatty acids has been seen, at least in zebrafish (Danio rerio ), to occur in a ratio to n‑3 of ca.  1:2.5 (Hastings et al., 2001). Therefore, the relative abundance of biosynthetic ARA found in trout raised on the LC-PUFA deprived diet provides additional evidence of compensatory biosynthesis of DHA (and EPA).
Furthermore, the distribution of fatty acids amongst various tissue/lipid samples, particularly the increasing percentages of longer chain PUFA in brain PL (n‑3) or muscle PL (n‑6), suggests routing priorities (Lacombe et al., 2018; vide Appendix Fig. A1). Faced with deficiency in dietary n‑3 LC-PUFA, trout routed available DHA (and EPA) away from muscle tissue, where it might promote hyperplasia and muscle fibre development (Wang et al., 2020), to brain polar lipids, where it might help maintain neural function, and hence cognition (Pilecky et al., 2021; Zavorka et al., 2023). Trout raised on a low n-3 LC-PUFA diet appear to have used a combination of biosynthesis and priority routing of LC-PUFA for active use in membranes (PL) at the expense of triacylglycerol storage (NL) to compensate for dietary lack. The compensation was, however, not complete: despite similar fatty acid composition in brain PL between the two diet treatment groups, the low n-3 LC-PUFA diet still resulted in smaller brains.
A limitation of the study was that all fish were fed ad libitum . With no curtailment of the amount of energy or precursor short-chain PUFA available, the compensatory effects of LC‑PUFA biosynthesis and routing under a suboptimal LC‑PUFA deprived diet in the present study are likely to be exaggerated. In nature, where ad libitum feeding is not observed, we expect differences in fatty acid profiles between natural diet groups to be more pronounced with attendant ramifications for brain morphology and cognitive ability (Závorka et al., 2022b).
The alternative diets upon which trout were raised proved to be the most important differentiator of treatment groups in this study. Diet exercised clear effects on the brain development, cognitive abilities, and LC‑PUFA biosynthesis and routing of brown trout, and effected a divergence in the fatty acid profiles of muscle tissue. The potential decrease in the production of n‑3 LC‑PUFA by primary producers due to climate change appears, based on the results of this study, to presage profound changes to the behavioural ecology of stream-dwelling fishes such as brown trout. Although the extent of these changes has yet to be determined, this study makes clear that a diet bereft of adequate n‑3 LC‑PUFA is suboptimal. In wild settings, where fish do not feed ad libitum , the effects of n‑3 LC‑PUFA deprivation are likely to be more severe. Furthermore, the complexity of nursery habitat also plays an essential role, independently of diet quality, in the development of cognitive skills. Therefore, further studies are needed, that integrate consideration of life history and diet in wild animals. Whether the divergence observed in this study is substantial enough to play a role in the development of the morphological and life-history variants observed in wild populations of brown trout deserves study (Závorka et al., 2022b).