Abstract
<title>Abstract</title> <p>Individual variation in spatial learning ability can affect fitness, yet the environmental contexts enabling cognitive advantages to enhance growth remain unclear. We tested whether the growth benefit of high spatial learning ability in cyprinid fish depends on environmental predictability and is mediated by stress–growth axis coordination. Juvenile crucian carp were screened for spatial learning to identify high- and low-spatial-learning phenotypes. Fish were assigned to three feed-restricted treatments: a non-spatial small-group competition (LS), a fixed-landmark maze (FM) with constant food location, and a dynamic-landmark maze (DM) where food–landmark associations changed every three days. High- and low-spatial-learning individuals were mixed 1:1 in each treatment. Growth, serum and hepatopancreatic cortisol, IGF-1, hepatic antioxidant enzymes, and brain morphology were measured over 30 days. High-spatial-learning fish showed significantly higher specific growth rate and weight gain rate exclusively in the DM treatment, accompanied by lower cortisol, higher hepatopancreatic IGF-1, and reduced antioxidant enzyme activities. In the FM treatment, high-spatial-learning fish lacked a growth advantage and had lower weight gain rate than low-spatial-learning fish in the later phase. No differences were observed in the LS treatment. Brain morphology did not differ between groups; however, high-spatial-learning individuals maintained tightly coupled stress, growth, and neural axes, whereas low-spatial-learning individuals displayed systemic disconnection. These findings demonstrate that the adaptive value of high spatial learning ability is gated by environmental predictability and relies on an intrinsically integrated physiological phenotype. This study provides mechanistic insights into how cognition shapes life-history traits through whole-organism physiological integration.</p>