Collective behaviour of herring during spawning

Abstract

The thesis focuses on the collective behaviour of Norwegian spring spawning herring (NSS-herring; Clupea harengus L.) in the specific ecological context of reproduction. The major part of the work is based on field research using quantitative echo sounders and sonar in combination with biological sampling. In addition, an individual based schooling model is applied to investigate underlying mechanisms of the collective formations and dynamics. All studies are founded in a classical mechanistic ecological approach interpreting behaviours as optimal evolutionary strategies given an individual’s physiological and cognitive constraints, internal state and the specific ecological context. Two of the works are case-studies from historical spawning grounds of NSSherring well-known to fishermen and scientists; the shallow (30-40 m) banks off Karmøy (Paper 1) receive a small proportion of the spawning stock, whereas deep (80-250 m) areas off Møre (Paper 2) comprise the main spawning grounds. In both works the areas are covered through repeated acoustic surveys in combination with sampling of herring and predators throughout the 24-h-cycle over several days. Fisheries scientists apply similar acoustic surveying when making abundance estimates of herring that serve as basis for advices of catch quotas. A major challenge during such surveying is the fact that herring tend to avoid the approaching research vessel, rendering them unavailable to detection. How this potential source of error operates during spawning is given the focus in one of the works (Paper 3). In this study we use a simple experimental design where a standard research vessel passes a presumably neutral stationary vessel that records the herring reaction. In the fourth work we apply a rule-based simulation model with high temporal resolution able to recreate life-like collective behaviours (Paper 4). The model is used to explore how collective dynamics and formations are affected when varying the amount of herring present and their degree of motivational synchronisation. In the final work (Paper 5) we investigate 11 years of acoustic data from herring spawning surveys along the Norwegian coast. We frequently find evidence of vertical hourglass formations, and based on knowledge gained from previous studies and the simulation model we formulate and test out a concrete hypothesis; are these formations founded in conflicting individual motivation for spawning? An overall evaluation of our field results strongly supports that herring behaviour during spawning reflects a compromise between survival and reproduction. We find that herring spawn in highly synchronous waves where individuals aggregate (Paper 2) and move together (Paper 1) in their quest for a successful spawning. This strategy is likely to be advantageous for survival since a synchronous emergence in high numbers will decrease an individual’s risk of being targeted by a predator (dilution effect). A similar advantage is gained through spawning in the dark hours when predators are less active. Shoals at night-time were observed to have a stable loose packing density, whereas a variable, but generally high packing density at daytime corresponds well with a presumed higher frequency of predator attacks, since it is known that shoals become more densely packed under attack. At the deep spawning grounds, large proportions of the herring became scattered in pelagic layers at night-time. Interestingly, these layers were positioned according to the thermocline in a way that indicates active use of the vertical temperature profile to fine-tune maturation rate in the days prior to spawning. The shallow spawning grounds had a different dynamic and such layers were not observed. In these areas, the vast majority of herring were in layers at the bottom at night-time. In this situation, herring did not react to the research vessel (Paper 3), probably because the focus on the spawning activity increases their reaction threshold. At both spawning grounds (Papers 1 and 2), there were substantial vertical dynamics that corresponded well with the results from the schooling model. In the model simulations, shoal height is dependent on the difference in individual motivation for spawning. With strong conflicts in motivation, shoals tend to split in the same way as we observed during spawning in the field (Paper 2), whereas with intermediate conflicts shoals maintain cohesion but become vertically extended like we observed during pre-spawning. A special case of the latter is the vertical hourglass formation that emerges in the model assorted according to maturation state. Similar assortment was observed in the hourglass formations in the wild, where spawners dominated in the lower parts relative to pre- and post-spawners, strongly indicating that motivational conflict is the driving force of the observed formation. Whether such formations are purely emergent from decisions made from local stimuli or implying communication between the two shoal parts is an interesting question open to future investigations.