Introduction
To understand predator impacts on prey and, in particular,
non-consumptive effects of predators on prey (Preisser et al. 2005;
Peacor et al. 2020), a key issue is to elucidate factors that explain
variation in prey responses to perceived predation risk (Lima 1998;
Brown & Kotler 2004; Creel & Christianson 2008; Wirsing et al. 2020).
The overall prey response is commonly split into two main stages: escape
when an attack occurs or appears imminent, and subsequent avoidance
often by staying in or near refuge (Lima & Dill 1990; Heithaus et al.
2009). Although this basic framework has long been part of standard
predator-prey behavioral ecology, most studies have either quantified
escape responses (e.g., flight initiation distances (FIDs); e.g.,
Stankowich & Blumstein 2005; Moller et al. 2016), or refuge use (e.g.,
time spent in refuge; Cooper & Sherbrooke 2015). Surprisingly, few have
examined both for the same individuals, particularly in the field. Here,
we repeatedly quantified the responses of individually marked,
free-ranging animals to the approach of a potentially threatening human.
We present what we believe is the first study to test for consistent
individual differences in multiple stages of the overall prey response
to risk: in initial escape tendencies, in behavior during the escape,
and in latency to resume activity after spending time in refuge. In
particular, our study is, to our knowledge, the first to examine the
fundamental point of individual-level correlations among these aspects
of the antipredator response. Under field conditions, even with natural
variation in context (e.g., distance to refuge, vegetation height,
presence of conspecifics), are the same individuals that are more
fearful than others in the sense of fleeing more readily also more
fearful in the sense of staying longer in refuge before resuming
activity? This correlation across stages of the prey response, if it
exists, provides an underlying mechanism for a tradeoff involving not
just the cost versus benefit of early versus late escape, or early
versus late emergence from refuge, but for a tradeoff based on variation
in fearfulness expressed across stages. Before describing our field
study on ground squirrel antipredator responses, we next provide a bit
more detail on the general conceptual overview.
When prey first detect a potential predator, they can, but often do not,
immediately initiate an escape attempt. Instead, prey often monitor the
predator’s behavior, and only initiate escape when the predator
approaches more closely. A standard metric for assessing fear is thus an
animal’s flight initiation distance (FID), the distance at which an
individual flees from an approaching intruder. Theory predicts that
because active escape from predators has costs (e.g., energy and lost
opportunities), prey should typically not initiate escape as soon as
they detect predators but should instead optimize their FID by fleeing
only when predators have come close enough that the costs of not fleeing
are higher than the costs of escaping (Ydenberg & Dill 1986). A large
literature (e.g., Stankowich & Blumstein 2005; Moller et al. 2016;
Morelli et al. 2019) shows that FIDs can depend on characteristics of
the predator (e.g., greater FID if predators are perceived to be more
dangerous), the prey (e.g., the prey’s state, escape ability, or
behavioral type), the social context (e.g., presence of conspecifics),
and the ecological context (e.g., availability and distance of refuge).
With regard to prey traits, the current interest in animal personalities
(Sih et al. 2004a,b; Reale et al. 2007) suggests a need to measure
consistent individual differences in FIDs; however, to date, relatively
few studies have quantified the repeatability of FIDs in nature (but see
Carette et al. 2009; Moller & Tryjanowski 2014; Cabrera et al. 2017).
After animals flee from a predator, they have further decisions to make
including whether to run into shelter and if so, when to emerge. Rather
than run all the way to shelter, animals sometimes flee and then ‘stop
and look’ to apparently re-assess the danger. The distance that they
flee before they ‘stop and look’ can be used as an additional measure of
fearfulness (i.e., more fearful animals likely have a larger ‘stop and
look’ distance). If prey flee to shelter, then a key decision is when to
emerge to resume activity (Sih 1992; Cooper & Frederick 2007). More
fearful animals likely have a longer latency to resume activity. While
FIDs have been measured in many species (Bonenfont & Kramer 1996;
Stankowich & Blumstein 2005; Engelhardt & Weladji 2011; Petelle et al.
2013; Moller & Tryjanowski 2014; Uchida et al. 2015), fewer studies
have explored post-FID responses (but see Bonenefont & Kramer 1996;
Cooper & Sherbrooke 2015; Tätte et al. 2018; Breck et al. 2019) and, to
our knowledge, no studies have quantified consistent individual
differences (repeatability) of post-FID responses – either in isolation
or in relation to other components of the antipredator response.
If FIDs, ‘stop and look’ distances, and latency to resume activity all
reflect differences among individuals in underlying fear, then
consistent individual differences in these should be positively
correlated. These correlations are ecologically important; for example,
the core idea that more fearful animals suffer greater opportunity costs
(e.g., greater reductions in feeding rate) from avoiding predators
hinges not just on them escaping more readily to shelter, but crucially,
on them hiding, often for long periods, before resuming activity. It is
thus striking that, to our knowledge, no previous studies have tested
the hypothesis that larger FIDs are positively correlated with longer
latencies to resume activity. Ideally, analyses of multi-stage prey
responses to predators should test for effects of both individual
differences in behavioral tendencies and multiple aspects of the context
(ecological and social) on each stage of the overall response; however,
as far as we know, no previous studies have attempted to do that.
We studied the responses of focal animals to approaching humans. With
the global expansion of human presence, animal responses to human
activity can have important effects on individual and species success
(Strasser and Heath 2013; Arroyo et al. 2017). How well animals cope may
depend on a variety of factors including their behavior and/or their
past experience with human disturbance (Sih et al. 2011, 2012; Lapiedra
et al. 2017). In many cases, animals respond to humans as predators,
actively avoiding areas of human activity (Oriol-Cotterill et al. 2015;
Clinchy et al. 2016; Suraci et al. 2019). In other cases, however,
repeated exposure to humans leads to habituation (Stankowich &
Blumstein 2005; Geffroy et al. 2015; Blumstein 2016; Uchida & Blumstein
2021). The reduced fear of humans can be associated with a general
increase in boldness, exploration or aggressiveness as often seen in
animals in urban environments (Moller 2008; Rodriquez-Prieto et al.
2008; Miranda et al. 2013; Uchida et al. 2015; Breck et al. 2019).
However, whilst behavioral adjustments in animals inhabiting urban
environments are well-documented, less is known about how human
activities shape behavior or behavioral variation in animals residing in
natural areas, such as reserves or parks that are comparatively
insulated from urban disturbance (Gonson et al. 2016; Watson et al.
2016; Corisini et al. 2019).
Here, we examined how variation in rates of human activity shape
risk-sensitive behavior throughout multiple steps of the antipredator
response in a free-living mammal, the California ground squirrel
(Otospermophilus beecheyi ). Ground squirrels are ecosystem
engineers, a major prey species in the California grasslands (Smith et
al. 2016) and display a suite of behavioral responses to threats (Owings
& Ledger 1980; Hanson et al. 1997; Putman et al 2015; Ayon et al. 2017)
including human approach (Hammond et al. 2019). While ground squirrels
are often deemed pests by humans, they are generally not directly killed
by humans. This species therefore offers an interesting opportunity to
examine how animals exposed to varying levels of human activity adapt
their behavior in the presence of humans. Specifically, we repeatedly
recorded both the squirrels’ flight initiation distances (Ydenberg &
Dill 1986, Bjorvik et al 2014; Uchida et al. 2015), and their post-FID
behaviors as discussed above (Fig. 1). Thus, our study is unique in
allowing us to both (1) explore how human activity influences each
decision of a squirrel’s antipredator response (i.e., when to flee and
whether and how long to shelter) and (2) examine the covariation between
different components of the antipredator response. We predicted that
human activity would influence multiple components of a squirrel’s
risk-sensitivity and anti-predator response. We further predicted that
if aspects of the antipredator response are correlated, then an
individual’s FID response should also correspond to their
risk-sensitivity across other contexts, such as their willingness to
enter a trap across multiple potential trapping sessions. Finally, we
also examined other factors that might contribute to risk-sensitive
decision-making including age and sex of the focal individual, the
surrounding microhabitat features, and conspecific presence.