Information can explain the dynamics of group order in animal collective behaviour

Abstract

Animal groups vary in their collective order (or state), forming disordered swarms to highly polarized groups. One explanation for this variation is that individuals face differential benefits or costs depending on the group's order, but empirical evidence for this is lacking. Here we show that in three-spined sticklebacks (Gasterosteus aculeatus), fish that are first to respond to an ephemeral food source do so faster when shoals are in a disordered, swarmlike state. This is because individuals' visual fields collectively cover more of their environment, meaning private information is more readily available in disordered groups. Once social information becomes available, however, the arrival times of subsequent group members to the food are faster in more ordered, polarized groups. Our data further suggest that first responding individuals (those that benefit from group disorder) maintain larger differences in heading angle to their nearest neighbours when shoaling, thereby explaining how conflict over whether private or social information is favoured can drive dynamic changes in collective behaviour.

Emergence of variation between groups through time in fish shoal collective motion

“…This is consistent with a weakening of attraction and responsiveness to near neighbours. Functionally, these temporal changes are associated with a reduction in social information sensing and anti-predator benefits (Ioannou et al, 2012; MacGregor et al, 2020; Partridge, 1982), and suggest that the groups were acclimatising over time to the low risk, predator free context. Maintaining social responsiveness during collective motion by forming fast, cohesive and highly polarised groups may be cognitively demanding and energetically costly (Di Santo et al, 2017).…”

“…Because individuals may seek refuge in heterogenous environments, no artifitial enrichment was provided in the arena to encourage collective movement. There was a small amount of variation in the length of each trial (25 ± 2.2, mean ± sd trial length in minutes for 84 trials) due to the experimental protocol (see MacGregor et al 2020 for further details). Trials were filmed from above with a Panasonic HC-VX980 video camera in 4K (3840 × 2178 pixels) and a temporal resolution of 25 frames per second.…”

“…At the start of a trial, all eight fish were netted into the centre of the arena and allowed 2 minutes to acclimatise. For the part of the experiment analysed and reported in MacGregor et al (2020), a red-tipped pipette that delivered a single food item (a defrosted bloodworm) was presented to each group on six occasions during each trial. The time interval between presentations was approximately 4 minutes, which allowed the group to resume normal swimming behaviour.…”

“…their speed, direction, and spatial organisation) differ fundamentally from non-living systems in that they emerge from interactions between complex and biologically-motivated individuals and are shaped by natural selection (Sumpter, 2006). Although there are costs to animal collective behaviour as it can result in competition for food (MacGregor et al, 2020) and exploitation by predators and parasites (Bauer et al, 2015; Rode et al, 2013), the independent evolution and diversity of collective behaviours across animal taxa suggest that benefits from improved predator avoidance (Ioannou et al, 2019) and use of resources (Sasaki et al, 2013) often outweigh the costs.…”

“…The groups were tested over multiple days up to twelve times over four weeks. Each trial lasted approximately 25 minutes and group behaviour was analysed during five 2.5-minute time intervals that were separated by the presentation of a food item (see Materials and Methods, and MacGregor et al, 2020 for analyses of responses to the food item presentations). We used high resolution video tracking to examine changes over time in six traits of collective motion that are known to be functionally important: polarisation (where higher levels of polarisation can improve social information transfer and reduce the risk of predation due to confusion effects: Attanasi et al, 2014; Ioannou et al, 2012), the speed of the group centroid (where group speed can affect responses to predators and encounter rates with food: Wood and Ackland, 2007), convex hull area as a measure of spacing between individuals (where less cohesive groups are at greater risk of predation: Duffield and Ioannou, 2017), the rate of transitions in state between polarised and unpolarised swarm-like states (Calovi et al 2015; Tunstrøm et al, 2013), the maximum cross-correlation in speed (which is a measure of information transfer: Katz et al, 2011), and the rate of switches in leadership over time (where individuals at the front of the group have greater access to foraging rewards at a cost of greater risk of predation, and increased switching is observed in shoals from high predation habitats: Burns et al, 2012; Herbert-Read et al, 2019; Ioannou et al, 2019; Reebs, 2000).…”