IntroductionAvoiding predation is a multistep process, and selectionshould act at every step (Lima and Dill 1990). From theprey’s perspective, protecting oneself from predation mayinclude having certain morphological traits, being able torecognise predators, possessing a variety of individualbehavioural strategies to reduce predation risk, engaging ingroup defenses, and making decisions about when and howto flee (Lima and Dill 1990; Caro 2005). While most studiesfocus on a single aspect of antipredator behaviour at a time(Caro 2005), these multiple stages of antipredator defencemay be influenced by certain ubiquitous factors and, if so,they may not be independent of each other. Knowledge of theinfluence of these factors on predation risk and its subse-quent effect on the expression of antipredator behaviour islimited, but identifying the effect of such factors is importantbecause it may highlight constraints on adaptive responses.Habitat structure is a ubiquitous factor that may affect timeallocation and escape decisions because habitat influencesvisibility and thus can influence overall predation risk.Habitat structure and the resulting visibility thus may influ-ence multiple stages of antipredator behaviour.Habitat structure can be considered obstructive or protec-tive (Finne et al. 2000; Whittingham and Evans 2004). If thehabitat structure is obstructive, the individual’s ability todetect predators is limited by the visibility. For instance,Arnez and Leger (1997) found that peripheral visibilityinfluenced foraging bout length of squirrels. If habitat struc-ture is considered protective, the individual is able to seekrefuge from the predator based on the amount of surround-ing foliage, or may be less conspicuous when in cover(Cooper 2003). Lima (1992) suggested that birds behaved inways consistent with the hypothesis that they assessed lowerrisk in trees with larger trunks and less visibility. Regardlessof whether cover is protective or obstructive, foliage densityhas been shown to influence time allocated to vigilance inbirds (Metcalfe 1984; Whittingham et al. 2004), and flightdecisions in lizards (Martin and López 1995), mammals(Blumstein et al. 2004b) and birds (Fernández-Juricic et al.2002, 2004).Flight-initiation distance is the distance from a predator atwhich an individual initiates its escape, and alert distance isthe distance from a predator that an individual orientstowards the approaching predator. Flight-initiation distance(Stankowich and Blumstein 2005), and the relationshipbetween alert distance and flight-initiation distance(Blumstein et al. 2005b), are often used to quantify warinessin animals and to identify factors that influence wariness.Australian Journal of Zoology, 2006, 54, 363–36710.1071/ZO05080 0004-959X/06/050363© CSIRO 2006Jennifer S. BoyerA, Laura L. HassA, Matthew H. LurieAand Daniel T. BlumsteinA,BADepartment of Ecology and Evolutionary Biology, University of California, 621 Young Drive South, Los Angeles, CA 90095-1606, USA.BCorresponding author. Email: [email protected]. Most studies of antipredator behaviour have focused on a single behaviour at a time, but ubiquitousfactors may simultaneously influence a variety of activities. Habitat structure influences visibility, which influencesboth the ability of prey to detect and respond to their predators. We studied how habitat visibility influenced timeallocation and escape decisions of crimson rosellas (Playcercus elegans). We examined the effect of visibility ontime allocated to looking, locomotion and foraging. We measured escape decisions by experimentally approachingrosellas until they fled. We measured visibility by standing 12 m away from a 1-m2white sheet containing 36 pointsand counting the number of points that were visible. As measured, visibility influenced time allocated to bothforaging and locomotion. However, there was no effect of visibility on flight-initiation distance or two other relatedmeasures of escape (a measure of the latency to detect an approaching threat or the latency to flee once the threatwas detected). Therefore, crimson rosellas modified their behaviour as a function of visibility, but this did notinfluence their decision to flee from an approaching human. We infer that they are sensitive to variation in visibility,but that this does not influence their overall perception of risk as we measured it. These two stages of antipredatorbehaviour may thus be largely independent; ubiquitous factors need not have ubiquitous effects on different aspectsof antipredator behaviour.Effect of visibility on time allocation and escape decisionsin crimson rosellaswww.publish.csiro.au/journals/ajzCSIRO PUBLISHINGJ. S. Boyer et al.364 Australian Journal of ZoologyOur study focused on time allocation and escape decisions ofcrimson rosellas in habitats with varying degrees of visibil-ity. We measured habitat visibility by standing 12 m awayfrom a 1-m2white sheet containing 36 points and countingthe number of points that were visible. We analysed timeallocated to looking, foraging, and locomotion using 2-minfocal animal samples (Martin and Bateson 1993). Escapedecisions were measured by experimentally approachingrosellas until they fled. We assumed that if rosellas perceivedcover as obstructing their field of view and their ability todetect approaching predators, (1) rosellas should be morevigilant and forage less, and (2) that if this is the case, theymay be slower to detect an approaching predator and henceflight-initiation distance would decrease.MethodsStudy site and target speciesBetween 20 and 31 October 2005, we studied the crimson rosella, amid-sized parrot, in and around Booderee National Park, AustralianCapital Territory. This species was commonly found foraging on theground or in tea trees (Leptospermum sp.). During the breeding season,we found them typically in groups of 1–3 individuals and usuallystudied solitary subjects. Additionally, birds were found perched in orflying through Eucalyptus spp. Importantly, the rosellas were found inareas of varied visibility. Most of the park is characterised as eucalyptforest, dominated by E. pilularis and E. botryoides, much of which wasrecovering from a widespread bushfire during 2003. We capitalised onthis