Abstract
Whale-watching activities are known to induce behavioural changes that are compatible with anti-predatory responses. However anti-predatory responses can vary depending on the species, population and/or site-specific environmental features. This variability makes it challenging to evaluate whale-watching disturbance and multiple metrics should therefore be used for any impact study. In this study we used meta-analyses to assess the consistency of anti-predatory responses among several studies on whale-watching disturbance. Changes in swimming speed, activity budget, inter-breath intervals, and deviation and directness index were used as proxies to measure whale-watching disturbance. We also assessed if these changes were due to extrinsic factors (the presence of whale watching regulations), intrinsic factors (species and animal body size) or environmental factors (habitat type). Our results showed that changes in the activity budget and directness and deviation index were the most consistently reliable metrics in capturing cetacean responses to whale-watching disturbance. Animals were more likely to travel and less likely to rest and forage (Q7=81.2, p< 0.0001, k=8; Q5=23.2, p=0.0003, k= 7; Q7=49.5, p< 0.0001, k=8) as a consequence of whale-watching presence. Body size also had an effect on resting activity with smaller animals being less likely to rest in the presence of boats (Q1=4.1, p=0.04, k=7). Our results showed a generalized tendency of cetaceans to increase their path sinuosity (deviation index) and decrease path linearity (directness index) in response to whale-watching disturbance (Q= 24.6, p= < 0.0001, τ2=0.09, k=5; Q= 8.94; p= 0.03, τ2=0.05, k=5). Path linearity and predictability also increased in corridor type habitats (SMD=1.55; p=0.004, k=5). We found that cetaceans overall showed similar anti-predatory responses to whale-watching activities. These behavioural changes reflect greater direct energy expenditure and fewer opportunities for energy recovery respectively and may cause long-term consequences to population fitness. We conclude with highlighting the validity of a meta-analytical approach in studying anthropogenic disturbances.