A physiological cost to behavioural tolerance

•We simultaneously measured heart rate and behavioural of incubating Masked Lapwings Vanellus miles.•A distinct Physiological-Initiation Distance (PID) occurred.•PID precedes Flight-Initiation Distance (FID) but does not necessarily Alert Distance (AD).•Short FIDs are generally interpreted as a sign of tolerance.•Short FIDs are associated with greater physiological costs through longer durations of elevated HR. Few studies of animal escape behaviour simultaneously investigate behavioural and physiological responses. Differences between these response types, however, have consequences for the way in which habituation or tolerance is interpreted – behavioural habituation may incur physiological costs. We simultaneously measured heart rate (HR) and behavioural responses during standardised approaches to incubating Masked Lapwings Vanellus miles, an urban-frequenting ground-nesting bird. We describe the existence of a distinct Physiological-Initiation Distance (PID) that precedes Flight-Initiation Distance (FID) but does not necessarily precede Alert Distance (AD). Two distinct response types occurred: ‘startle’, where a behavioural or physiological response coincided with the appearance of a person (always the investigator; 75.9 % of 58 birds) and ‘non-startle’ responses, where a behavioural or physiological response occurred after the appearance of, and commencement of the approach by, the person (24.1 % or 14 birds). For birds which were not startled, the interval between the initial heart rate increase and heart rate peak increased with clutch age. For birds which were startled, longer durations of post-peak HR elevation were associated with shorter FIDs and older clutches. Thus, reduced FIDs (generally interpreted as a sign of habituation or tolerance) are associated with greater physiological costs through longer durations of elevated HR. Additionally, the existence of, often long and undetectable, PIDs suggests: 1) that behavioural measures of response underestimate responses in general, and 2) that the methodological assumption when collecting FIDs, that starting distances exceed response distances, may often be incorrect yet are visually undetectable. Further studies of a variety of taxa are warranted to determine the associations between behavioural and physiological responses, and should these associations prove reliable, they would ideally generate general predictions of PID from readily measurable behavioural metrics (FID or AD), thus enabling pre