The evolution of correlations between behavioural and morphological defence in Alaskan threespine stickleback fish (Gasterosteus aculeatus): evidence for trait compensation and co-specialization Craig A. Marshall 1 and Matthew A. Wund 2 1 Department of Biology, Colorado State University, Fort Collins, Colorado, USA and 2 Department of Biology, The College of New Jersey, Ewing, New Jersey, USA ABSTRACT Background: Prey often make behavioural and morphological adaptations to avoid predation, and these alternative defence mechanisms may either compensate for, or reinforce, one another. Objective: We examined correlations between anti-predator behaviour and morphology in threespine stickleback fish (Gasterosteus aculeatus) in four environments distinguished by their predation histories: marine/anadromous populations co-existing with native predatory fish (representative of the ancestral environment), and three freshwater environments with native, introduced, and no predatory fish, respectively. To determine whether morphological and behavioural defences reinforce one another (trait co-specialization) or whether they represent alternative strategies for defence (trait compensation), we related morphology of laboratory- reared stickleback to the intensity of their responses to a simulated predator attack, which had previously been assessed (Wund et al., 2015). Methods: Eight aspects of stickleback morphology were measured, all of which loaded heav- ily and positively on a single principal component that accounted for 61.3% of the variation in traits. Pearson correlation was used to determine whether PC1 was associated with the intensity of anti-predator response. Results: A weak negative correlation was observed between anti-predator behaviour and morphology overall. However, considering each predation environment separately revealed that populations from marine and freshwater environments containing native predatory fish dis- played trait co-specialization (positive correlation) between armour and behaviour, while those from environments with recently introduced predatory trout displayed trait compensation (negative correlation). No correlation was observed in populations lacking these predators. Not all populations within the ‘introduced predatory fish’ category showed the same pattern of relationship, however, indicating that additional factors mediate the co-evolutionary dynamics of anti-predator behaviour and morphology. Results were similar whether we considered size- standardized or unstandardized morphological traits. Correspondence: C.A. Marshall, Department of Biology, Colorado State University, Fort Collins, CO 80523, USA. email: cam13@colostate.edu Consult the copyright statement on the inside front cover for non-commercial copying policies. Evolutionary Ecology Research, 2017, 18: 305–322 © 2017 Craig A. Marshall