Discussion
Here we demonstrate how butterflyfishes respond to the loss and subsequent recovery of their primary food source by adjusting their aggression levels in line with resource availability. We found that this was unlikely to be caused by a reshuffling of the butterflyfish community, or a selective removal of more aggressive individuals, but more plausibly linked to energy conservation during an acute disturbance event. Once coral cover recovered, aggression returned to pre-disturbance levels, demonstrating the importance of behavioural plasticity to maintain species persistence during environmental disturbances. As predicted by models of economic defendability, the drop in the probability of aggressive interactions occurring on Iriomote following bleaching, closely matched the availability of the primary food resource for many butterflyfish species; staghorn Acropora corals. Similarly, as coral cover recovered, in years following the bleaching event, so did aggressive interactions. While we do not have sufficient replication to statistically analyse aggression at the species level across the four years, visual examination of trends suggest the majority of species closely followed the these trends (Supporting information). Engaging in territorial and aggressive interactions carries a significant energetic cost [26], both in terms of energy directly allocated to the behaviour [27], and opportunity cost due to reduced feeding [28]. Thus, a decline in aggressive interactions is likely to lead to lower energy consumption during an ongoing disturbance event. Reductions in energy-intensive behaviours during times of low resource availability may be a successful evolutionary strategy to conserve energy, and species with higher levels of behavioural plasticity tend to have a higher survival during disturbance events [29,30]. As environmental disturbances become more frequent and more severe globally, it is increasingly important to understand how these processes play out in the long-term.
Altered behaviour is invariably linked to a trade-off between energy expenditure (maintenance, growth, reproduction) and energy acquisition, and populations are unlikely to persist on either extreme of this trade-off spectrum. In species or individuals with a high degree of behavioural variability or capacity to adapt their behaviour, we would expect behaviour to always be close to optimum for a given resource availability. For example, urban nesting birds that experienced an increase in availability of nesting and foraging sites during Covid-19 lockdowns were less territorially aggressive [31]. Similarly, male black-throated blue warblers (Setophaga caerulescens ) that were given a supplemental diet reduced their singing rates, representative of territorial behaviour, in favour of behaviours more directly linked to increased reproductive success, such as mate guarding [32]. In contrast, we would expect strong selective pressure against individuals with limited behavioural flexibility, such that the population and community composition is altered in affected ecosystems [33,34].
The variation in extent and frequency of disturbance will also interact with a species’ life history strategy, potentially affecting some species disproportionally. For species with a relatively short lifespan, a severe disturbance event can have a large impact on population dynamics and leave populations with little time to recover between events. However, shorter generation times facilitate more rapid evolutionary adaptation through natural selection. Butterflyfish can live over a decade, with mean maximum age for common species spanning 5-10 years [35], and attain sexual maturity within the first two years of life [36]. The demography of this family of reef fishes suggest they are likely to experience multiple disturbance events throughout their life cycle. Studies of animal behaviour are often limited in spatial and temporal replication making it difficult to tease apart whether it is behavioural plasticity at the level of an individual or within populations, or adaptative selection across generations, that is the underlying mechanism of altered behaviour. By monitoring the abundance and species composition of a community over time we can disentangle whether behavioural changes can be plausibly explained by the selective removal of individuals with maladaptive behavioural traits.
In this study, we found weak evidence of a reshuffling of the butterflyfish community post bleaching, however this pattern was largely driven by rare species. There was no significant shift in the relative proportions of the four most common species in the population, nor an overall change in the abundance of butterflyfishes. We predicted that the degree of feeding preference specialisation would affect how species responded to coral mortality, as highly specialised species are more constrained by resource availability than generalists [37]. Butterflyfishes display a range of dietary preferences, from highly specialised obligate coral feeders that feed exclusively on a small subset of Acropora species (e.g. C. trifascialis ), to facultative coral feeders whose diet consists of over 80% corals (e.g.C. citrinellus ), through to invertivores (e.g. C. vagabundus ). However, the relative abundance of these three feeding guilds (facultative, obligate coral feeders, and invertivores) did not differ significantly between the four surveyed years. Combined, these results suggest that a plasticity in behaviour, rather than selective removal of aggressive individuals, species or feeding guilds, has allowed this butterflyfish population to adapt to shifting resource availability and weather the storm of a substantial disturbance event.
Our results follow the classic pattern of the economic defendability model, where individuals must balance the advantages (e.g. energy gained) and disadvantages (e.g. energy lost) associated with competing for the resource to maximise fitness. For example, individuals risk a net loss of fitness if engaging in aggression over an ephemeral or highly mobile resource that is difficult or impossible to defend effectively [11]. Under this model, aggression is predicted to be highest at intermediate resource availability, where resources are limited yet abundant enough to warrant energy investment into aggressive interactions. It follows then that aggression is predicted to be low at both extremes of the resource availability spectrum where resources are either too abundant to require investment in aggressive interactions, or so scarce that such investment is associated with reduced fitness[38]. While Brown’s model of economic defendability has been supported in a multitude of systems at varying fixed levels of resource availability (e.g salmonids[39], herbivorous damselfishes[40] and convict cyclids[41]), our current understanding is largely based on such static snapshots in time. This is the first time the model has been tested in the context of environmental change, using repeated sampling in the field during an ongoing disturbance event.
While a temporary behavioural shift to conserve energy may have allowed this population to weather the storm of a large-scale disturbance event, these findings do not suggest reef fishes are safe from anthropogenic climate change. Only species with adaptable behaviour can access this short-term survival strategy. For example, birds with a capacity to alter migration routes can avoid wintering sites that have suffered habitat loss, while their less flexible counterparts will suffer the consequences of reduced resource availability [42]. Second, a behavioural response to habitat loss hinges on the reliability of cues to assess habitat quality, with incorrect cues leading to maladaptive behaviours (e.g. caught in ecological traps) [43]. Indigo buntings (Passerina cyanea ) for example, prefer forest edge habitat due to a low abundance of predators in naturally created edges. However, the birds are equally attracted to anthropogenically created edges that contain increased numbers of nest predators, and suffer lower reproductive success as a result [44,45]. Finally, the frequency of disturbance events is likely to increase, reducing the capacity for systems to recover fully between events, limiting the effectiveness of behavioural shifts [46].
Habitat loss remains one of the most pervasive threats against global biodiversity, reshuffling ecosystems and communities and causing a worldwide biodiversity crisis. Here we have demonstrated that a community of butterflyfishes alters their behaviour in line with resource availability by engaging in fewer aggressive interactions when resources are low, and returning to pre-disturbance behaviours as resources recover. The most plausible explanation to understand this change in behaviour is a reduction in energy expenditure during environmental change, to maintain optimal responses following the economic defendability model. While good news for this family of reef fishes benefiting from relatively rapid resource recovery, our findings highlight the vulnerability of individuals, and species, with low behavioural flexibility. Further, behavioural flexibility may not safeguard species experiencing environmental change where the conditions, magnitude of disturbance, or other variables are not as favourable to recovery. Our findings further emphasise the importance of understanding behavioural responses to habitat loss to predict how ecosystems may adapt to future environmental conditions.