Predator-prey interactions are vital to the stability of many ecosystems. Yet, few studies have considered how they are mediated due to substantial challenges in quantifying behavior over appropriate temporal and spatial scales. Here, we employ high-resolution sonar imaging to track the motion and interactions among predatory fish and their schooling prey in a natural environment. In particular, we address the relationship between predator attack behavior and the capacity for prey to respond both directly and through collective propagation of changes in velocity by group members. To do so, we investigated a large number of attacks and estimated per capita risk during attack and its relation to the size, shape, and internal structure of prey groups. Predators were found to frequently form coordinated hunting groups, with up to five individuals attacking in line formation. Attacks were associated with increased fragmentation and irregularities in the spatial structure of prey groups, features that inhibit collective information transfer among prey. Prey group fragmentation, likely facilitated by predator line formation, increased (estimated) per capita risk of prey, provided prey schools were maintained below a threshold size of approximately 2 m(2). Our results highlight the importance of collective behavior to the strategies employed by both predators and prey under conditions of considerable informational constraints.

Fish stock productivity, and thereby sensitivity to harvesting, depends on physical (e.g. ocean climate) and biological (e.g. prey availability, competition and predation) processes in the ecosystem. The combined impacts of such ecosystem processes and fisheries have lead to stock collapses across the world. While traditional fisheries management focuses on harvest rates and stock biomass, incorporating the impacts of such ecosystem processes are one of the main pillars of the ecosystem approach to fisheries management (EAFM). Although EAFM has been formally adopted widely since the 1990s, little is currently known to what extent ecosystem drivers of fish stock productivity are actually implemented in fisheries management. Based on worldwide review of more than 1200 marine fish stocks, we found that such ecosystem drivers were implemented in the tactical management of only 24 stocks. Most of these cases were in the North Atlantic and north-east Pacific, where the scientific support is strong. However, the diversity of ecosystem drivers implemented, and in the approaches taken, suggests that implementation is largely a bottom-up process driven by a few dedicated experts. Our results demonstrate that tactical fisheries management is still predominantly single-species oriented taking little account of ecosystem processes, implicitly ignoring that fish stock production is dependent on the physical and biological conditions of the ecosystem. Thus, while the ecosystem approach is highlighted in policy, key aspects of it tend yet not to be implemented in actual fisheries management.

The mesopelagic community is important for downward oceanic carbon transportation and is a potential food source for humans. Estimates of global mesopelagic fish biomass vary substantially (between 1 and 20 Gt). Here, we develop a global mesopelagic fish biomass model using daytime 38 kHz acoustic backscatter from deep scattering layers. Model backscatter arises predominantly from fish and siphonophores but the relative proportions of siphonophores and fish, and several of the parameters in the model, are uncertain. We use simulations to estimate biomass and the variance of biomass determined across three different scenarios; S1, where all fish have gas-filled swimbladders, and S2 and S3, where a proportion of fish do not. Our estimates of biomass ranged from 1.8 to 16 Gt (25–75% quartile ranges), and median values of S1 to S3 were 3.8, 4.6, and 8.3 Gt, respectively. A sensitivity analysis shows that for any given quantity of fish backscatter, the fish swimbladder volume, its size distribution and its aspect ratio are the parameters that cause most variation (i.e. lead to greatest uncertainty) in the biomass estimate. Determination of these parameters should be prioritized in future studies, as should determining the proportion of backscatter due to siphonophores.

Oceans constitute over 70% of the earth's surface, and the marine environment and ecosystems are central to many global challenges. Not only are the oceans an important source of food and other resources, but they also play a important roles in the earth's climate and provide crucial ecosystem services. To monitor the environment and ensure sustainable exploitation of marine resources, extensive data collection and analysis efforts form the backbone of management programmes on global, regional, or national levels. Technological advances in sensor technology, autonomous platforms, and information and communications technology now allow marine scientists to collect data in larger volumes than ever before. But our capacity for data analysis has not progressed comparably, and the growing discrepancy is becoming a major bottleneck for effective use of the available data, as well as an obstacle to scaling up data collection further. Recent years have seen rapid advances in the fields of artificial intelligence and machine learning, and in particular, so-called deep learning systems are now able to solve complex tasks that previously required human expertise. This technology is directly applicable to many important data analysis problems and it will provide tools that are needed to solve many complex challenges in marine science and resource management. Here we give a brief review of recent developments in deep learning, and highlight the many opportunities and challenges for effective adoption of this technology across the marine sciences.

Acoustic-trawl surveys are an important tool for marine stock management and environmental monitoring of marine life. Correctly assigning the acoustic signal to species or species groups is a challenge, and recently trawl camera systems have been developed to support interpretation of acoustic data. Examining images from known positions in the trawl track provides high resolution ground truth for the presence of species. Here, we develop and deploy a deep learning neural network to automate the classification of species present in images from the Deep Vision trawl camera system. To remedy the scarcity of training data, we developed a novel training regime based on realistic simulation of Deep Vision images. We achieved a classification accuracy of 94% for blue whiting, Atlantic herring, and Atlantic mackerel, showing that automatic species classification is a viable and efficient approach, and further that using synthetic data can effectively mitigate the all too common lack of training data.

Behavior of herring (Clupea harengus) is stimulated by two ocean-going research vessels; respectively designed with and without regard to radiated-noise-standards. Both vessels generate a reaction pattern, but, contrary to expectations, the reaction initiated by the silent vessel is stronger and more prolonged than the one initiated by the conventional vessel. The recommendations from the scientific community on noise-reduced designs were motivated by the expectation of minimizing bias on survey results caused by vessel-induced fish behavior. In conclusion, the candidate stimuli for vessel avoidance remain obscure. Noise reduction might be necessary but is insufficient to obtain stealth vessel assets during surveys.

No abstract

De Robertis, A. and Handegard, N. O. 2013. Fish avoidance of research vessels and the efficacy of noise-reduced vessels: a review. – ICES Journal of Marine Science, 70:34–45. It has long been recognized that fish can avoid approaching vessels and that these behaviours can bias fishery surveys. Underwater noise is considered the primary stimulus, and standards for research vessel noise have been established to minimize fish reactions. We review the literature on fish reactions to vessels appearing since these recommendations were made, focusing on acoustic surveys, and compare how fish react to noise-reduced and conventional vessels. Reactions to approaching vessels are variable and difficult to predict. However, the behaviour can bias acoustic abundance measurements, and should be considered when performing acoustic surveys. The few comparisons of acoustic abundance measurements from noise-reduced and conventional vessels are contradictory, but demonstrate that the sound pressure level, on which the noise-reduction criterion is based, is insufficient to explain how fish react to survey vessels. Further research is needed to identify the stimuli fish perceive from approaching vessels and the factors affecting whether fish perceiving these stimuli will react before further recommendations to reduce vessel-avoidance reactions can be made. In the interim, measurement of the biases introduced by fish avoidance reactions during surveys, and timing of surveys when fish are in a less reactive state, may reduce errors introduced by vessel avoidance.