Main
Anthropogenic climate change is causing substantial loss of habitat in global ecosystems, with devastating effects on biodiversity [1,2]. Changes in behaviour are often the first response of individuals to disturbance [3–5], yet behavioural responses and their ecological consequences, especially in real-world settings, are poorly understood relative to direct lethal impacts [6]. When faced with acute disturbances, organisms can adopt temporary behavioural changes that increase their chances of survival. [7]. For instance, when resources are depleted, individuals can conserve energy by changing activity levels [8] or reducing territorial behaviour [9]. Despite the significant ramifications for population persistence, species coexistence and wider ecosystem function [10], the mechanisms that underlie these behavioural shifts are unclear. This gap in knowledge hinders our capacity to predict the resilience of populations to environmental change, and to understand the implications for population and community dynamics.
Economic models provide a useful lens for understanding behavioural flexibility [11]. Aggressive behaviour, which functions to defend territories and maintain access to resources, can be dynamically regulated according to the costs and benefits of defence [12]. This dynamic regulation predicts a hump-shaped relationship between aggression and resource availability, where aggression is highest at intermediate resource levels. Decreased aggression of reef fish in response to resource depletion offers partial support for this model [9], but the critical test lies in determining whether aggression returns to pre-disturbance levels following resource recovery. Resolving this question would improve predictions of population resilience to ongoing environmental changes.
An alternative explanation for behavioural change following disturbance posits differences in fixed behavioural phenotypes across individuals within a population. In this case, natural selection can lead to a loss of individuals with an uneconomic level of aggression, which directly impacts on the probability of aggressive encounters within the population. Under such a scenario, a return to pre-disturbance frequencies of behaviour relies on processes that occur over a substantially longer time period, i.e. an increase in the proportion of aggressive individuals either through immigration or reproduction. Reproduction is likely to be delayed or reduced when individuals are energetically compromised following resource depletion [13,14], and a return of aggressive behaviour could be further limited if such behaviours are heritable rather than learned. Thus, reverting back to the original state hinges on several variables, including extant behavioural diversity from which to replenish affected populations, dispersion rates, reproductive rates and the extent of variation in aggressive behaviour within the repertoires of individuals. If behaviour is fixed, the subsequent loss of behavioural diversity following disturbance could affect the potential of the population to adapt to future environmental changes [15]. Therefore, identifying which mechanism is in action can provide vital information on how quickly populations can recover behaviourally from disturbance, if they can recover at all, and the potential ramifications at the community and ecosystem scale.
Coral reef fishes offer an excellent model system with which to explore behavioural responses to disturbance and are particularly vulnerable to environmental change. In particular, butterflyfishes are highly reliant on live coral [16], have small home territories [17], are abundant and easy to identify [18], and their behaviours can reliably be assessed in the field [e.g. 9,14,19]. Butterflyfishes also display a range of territoriality, with a species’ levels of aggression positively correlated to their degree of feeding specialisation [20,21]. Here, we use an unprecedented dataset that tracks aggressive behaviour between 23 species of butterflyfishes on three reefs in Japan from 2016-2022. This includes surveys through a full cycle of disturbance and resource recovery, following a mass coral bleaching event in 2016. In the year after this disturbance, corals (the primary food source of butterflyfishes) were significantly depleted throughout the Indo-Pacific (18-65% loss,) leading to a three-fold reduction in aggressive territorial defence [9]. Here, we undertake the first test of the long-term impact of altered resource availability on aggressive behaviour in the field to ask whether the reduction in aggression (1) reverts to pre-disturbance levels, indicating behavioural flexibility, or (2) persists in the population indicating that behaviour is fixed. To ascertain the fit with the predictions of the economic model of defendability, we also tested (3) whether the main food source of the butterflyfishes (Acropora corals) recovered in abundance, and (4) if changes in aggression could be a by-product of potential changes in the relative abundance of Chaetodon species caused in response to the bleaching event.