While various marine predators form associations, the most commonly studied are those between subsurface predators and seabirds, with gulls, shearwaters or terns frequently co-occurring with dolphins, billfish or tuna. However, the mechanisms underlying these associations remain poorly understood. Three hypotheses have been proposed to explain the prevalence of these associations: (1) subsurface predators herd prey to the surface and make prey accessible to birds, (2) subsurface predators damage prey close to the surface and thereby provide food scraps to birds, and (3) attacks of underwater predators lower the cohesion of prey groups and thereby their collective defences making the prey easier to be captured by birds. Using drone footage, we investigated the interaction between Indo-Pacific sailfish (Istiophorus platypterus) and terns (Onychoprion sp.) preying on schooling fish off the eastern coast of the Malaysian peninsula. Through spatio-temporal analysis of the hunting behaviour of the two predatory species and direct measures of prey cohesion we showed that terns attacked when school cohesion was low, and that this decrease in cohesion was frequently caused by sailfish attacks. Therefore, we propose that sailfish created a by-product benefit for the bird species, lending support to the hypothesis that lowering cohesion can facilitate associations between subsurface predators and seabirds.

In this paper, a delayed diffusive predator-prey model with schooling behaviour and Allee effect is investigated. The existence and local stability of equilibria of model without time delay and diffusion are given. Regarding the conversion rate as bifurcation parameter, Hopf bifurcation of diffusive system without time delay is obtained. In addition, the local stability of the coexistent equilibrium and existence of Hopf bifurcation of system with time delay are discussed. Moreover, the properties of Hopf bifurcation are studied based on the centre manifold and normal form theory for partial functional differential equations. Finally, some numerical simulations are also carried out to confirm our theoretical results.

Many piscivorous fish species are depleted and/or threatened around the world. Marine Protected Areas (MPAs) are tools for conservation and fisheries management, though there is still controversy regarding the best design for increasing their ecological effectiveness. Here, on the basis of a weighted meta-analytical approach, we have assessed the effect of 32 MPAs, distributed worldwide, on the biomass and density of piscivorous fishes. We analysed the MPA features and the biological, commercial and ecological characteristics of fishes that may affect the response of species to protection. We found a positive effect on the biomass and density of piscivores inside MPAs. This effect was stronger for the biomass of medium-sized fishes (in relation to the maximum size reported for the species) and the density of large and gregarious species. The size of the no-take zone had a significant negative impact on both response variables and differed according to the level of enforcement, with smaller no-take zones having higher levels of enforcement. Thus, MPAs help to protect piscivorous fish species, with smaller, but well enforced reserves being more effective for the protection of the local populations of piscivorous fishes throughout the world.

Animals are exposed to variable and rapidly changing environmental flow conditions, such as wind in terrestrial habitats and currents in aquatic systems. For fishes, previous work suggests that individuals exhibit flow-induced changes in aerobic swimming performance. Yet, no one has examined whether similar plasticity is found in fast-start escape responses, which are modulated by anaerobic swimming performance, sensory stimuli and neural control. In this study, we used fish from wild schools of the tropical damselfish Chromis viridis from shallow reefs surrounding Lizard Island in the Great Barrier Reef, Australia. The flow regime at each site was measured to ascertain differences in mean water flow speed and its temporal variability. Swimming and escape behaviour in fish schools were video-recorded in a laminar-flow swim tunnel. Though each school\'s swimming behaviour (i.e. alignment and cohesion) was not associated with local flow conditions, traits linked with fast-start performance (particularly turning rate and the distance travelled with the response) were significantly greater in individuals from high-flow habitats. This stronger performance may occur due to a number of mechanisms, such as an i n s itu training effect or greater selection pressure for faster performance phenotypes in areas with high flow speed.This article has an associated First Person interview with the first author of the paper.

Food searching behaviour in a group of individually tagged 1-5 kg Atlantic cod Gadus morhua was studied in a set of three experiments in a sea cage with two underwater platforms, where restricted amounts of food was delivered several times per day during an acoustic training period. It took c. 1 week to train 20 naïve cod to associate low frequency (250 Hz) sound with food, whereas the training time was reduced to less than 2 days when 19 naïve G. morhua were accompanied with one trained fish. The fish formed a school that cruised between the platforms in search of food. Usually, there was one leader in the school, a fish that swam faster, arrived first at the platforms and visited the platforms more frequently than other members of the school. The leader spent more energy on swimming but also received more food and grew faster than the rest of the fish. At the start of the experiments, the leaders were not larger than the average fish but always among the leanest ones in the group. The study reveals how social learning can facilitate the acoustic training in adult G. morhua, information that may be useful in finding ways to aggregate valuable fish species for environmentally friendly fishing and ranching.

In many species, there are antipredator benefits of grouping with conspecifics. For example, animals often aggregate to better avoid potential predators (the \'avoidance hypothesis\'). Animals also often group together in direct response to predators to facilitate escape (the \'escape hypothesis\'). The avoidance hypothesis predicts that animals with previous experience with predation risk will aggregate more than animals without experience with predation risk. In contrast, the escape hypothesis predicts that immediate exposure to predation risk causes animals to aggregate. We simultaneously tested these two nonexclusive hypotheses in threespine sticklebacks, Gasterosteus aculeatus. Schooling behaviour (time spent schooling, latency to school and time schooling in the middle of the school) was quantified with a mobile model school. Fish that had been chased by a model predator in the past schooled more, started schooling faster and spent a marginally greater proportion of time schooling in the middle of the school than fish that had not been chased. In contrast, there was no difference in the schooling behaviour of fish that were immediately exposed to either a model pike or a control, stick stimulus. A second experiment confirmed that fish perceived the model pike and stick differently: fish froze more often in the presence of the model pike, oriented to it more often and spent less time with the model pike than they did with the stick. These results provide strong support for the avoidance hypothesis.