ÐÓ°ÉÔ­´´

VIGILANCE in animals is a frustrating thing to study. Feeding behaviour,
for example, is far more straightforward: there is little problem in telling
when an animal is grazing, say, or consuming prey. Aggressive behaviour
is similarly clear cut: threats and attacks are what you look for, and they
are easy to identify. These behaviours are also a response to some observable
stimulus. But how do we know when an animal is being vigilant? How do we
know when it is adopting some posture or movement that increases the likelihood
of it detecting something in the environment? The stimulus the animal is
looking for is not around while the behaviour is going on, and, strictly
speaking, the idea that one posture is more vigilant than another can only
be tested once the animal has detected some event, after the vigilant behaviour
is over.

Biologists have usually tackled these problems in a rather indirect
way: they take a behaviour that looks, on the face of it, to be a good candidate
for vigilance and then see if it varies as one would expect according to
some hypothesis about what the animal is looking for. Geese are good subjects
for the study of vigilance because they interrupt bouts of grazing to stand
with their heads up and this markedly improves their view of their surroundings.
When Ian Inglis, of the Ministry of Agriculture, Fisheries and Food, and
I studied pink-footed geese breeding in Iceland, we noticed that this potentially
vigilant behaviour was in fact composed of two kinds: ‘head-up’, the posture
adopted by walking birds, and ‘extreme head-up’, in which the neck and body
are stretched further upwards. The ‘extreme head-up’ seemed to be the more
likely candidate for vigilance, as birds generally adopted this posture
when standing, possibly in order to look around and to see further over
vegetation. This circumstantial evidence was strengthened when we found
that parental geese adopted the extreme head-up posture – but not the head-up
posture – more often than birds without young and devoted less time to it
as their goslings developed. So extreme head-up was surely a vigilant posture
designed to protect the young. Other evidence persuaded us that the birds
watched out for snowy owls, skuas and arctic foxes, which would make an
easy meal of a young gosling if given the chance.

There are now many studies showing how vigilance for predators varies
adaptively, in ways fitting to the environment, but we also need to answer
the question of where an animal is looking. If we knew more about this we
would be in a better position to say what it was looking for. It is easy
to know roughly where another person is looking, presumably because we learn
the relationship between subtly different eye positions and the direction
and distance of the fixation of gaze. We can even infer the same information
from a painting, in which the eyes may be conveyed by a few brushstrokes.
Better still, it is possible to pinpoint precisely where human subjects
are gazing – provided they are willing to be confined in the laboratory
with an infrared light bounced off their cornea and a light sensor strapped
to their spectacles by which to pick it up. This technique, developed by
David Noton and Lawrence Stark at the University of California at Berkeley,
can show the precise features of a scene that attract visual attention,
as well as the way in which eye movements shift the gaze from one point
of interest to another.

It would be illuminating to have the same insight into the visual worlds
of other species, but the technical problems are obviously great. They are
not insurmountable, however – at least not for birds which, unlike mammals,
can move their eyes only a little within the skull. So movements of the
head are good indicators of eye movements, and far easier to monitor. Mark
Friedman, working at the University of Edinburgh, discovered how Barbary
doves use their eyes when approaching a seed on the ground by filming them
and painstakingly analysing, frame by frame, the position of the dove’s
head in relation to the seed. He discovered that the doves use two distinct
areas of the retina when performing this task. As they approach a seed from
a distance, birds turn their heads from side to side in order to fixate
it, first with one eye’s centrally placed fovea (the area of maximum acuity)
and then the other’s. But just before pecking the seed, they face it head
on, to fixate it in binocular vision with the temporal regions of the two
retinas. This binocular view provides the doves with more accurate information
about the exact position of the seed immediately before they eat it. Most
of the time, however, the birds use only one eye to view seeds on the ground,
so they might use the other fovea – which points upwards – to monitor the
surroundings for other important information, such as the approach of a
predator.

Falcons, swifts and hummingbirds have a fovea in both monocular and
binocular visual fields, and they presumably use the binocular field to
fixate prey during a chase or, in the case of the hummingbird, a flower
while hovering above the nectary. But we know little about how these and
other birds use their eyes for different tasks in nature. Why, for example,
do some water birds, and birds of open plains, have a long horizontal fovea,
rather than the more usual circular area? If, as some researchers suggest,
these birds use their fovea to fixate the horizon as a point of reference,
what do they use this reference for? Rebecca Torrance has suggested to me
that a long stretch of horizon could be fixated on the fovea so that the
bird can readily detect anything crossing it.

Keeping watch for predators is such an important function of vigilance
because of the high cost of failure; missing the object in question can
mean the difference between life and death. The most direct evidence that
vigilance is directed towards the detection of danger is that individuals
become more vigilant when the risk of predation increases. Tom Caraco and
his colleagues at the University of Arizona demonstrated this by showing
that flocks of finches spent more time scanning when a tame Harris’ hawk
flew over them. Individual finches in larger flocks also spent less time
being vigilant. Mark Elgar of the University of New South Wales has recently
reviewed 32 studies in birds and 20 in mammals demonstrating this phenomenon,
in species from ostriches to pronghorn antelopes.

Why should individuals in larger groups devote less time to vigilance
than those in smaller ones? There are two obvious explanations. First, the
larger the group, the more likely it is that some other individual will
detect the predator, so diminishing the benefit of each individual’s vigilance.
Secondly, an individual’s risk of being captured decreases in a larger group,
simply because the predator has a wider choice of alternative prey. (This
holds true even when one allows for the greater visibility of larger groups.)
But as is often the case, the same phenomenon can have more than one explanation,
and there are at least two other possibilities.

One explanation rests on two facts: individuals on the edge of groups
are more vigilant because they are at greater risk of being captured by
ground predators on the periphery, and a peripheral band of animals forms
a greater proportion of the whole group as the group becomes smaller. So
indi viduals on average are more vigilant in a smaller group. But it could
be that vigilance is directed not at predators but at other members of the
species in order to obtain information about feeding opportunities. John
Krebs of the University of Oxford found, while working at the University
of British Columbia, that great blue herons searching for fish in mudflat
pools looked up more often when they were in smaller flocks than in larger
ones. When he analysed the data in greater depth, he found that this relationship
was solely a conse quence of the fact that birds in smaller flocks found
fish at a slower rate. Krebs suggested that when feeding conditions were
poor, individuals looked up to see any other herons that they might follow
to better feeding pools. Feeding suc cess was greater in larger flocks of
herons simply because birds were less inclined to leave a good feeding site,
so flocks built up to a larger size in areas rich in prey.

Taking cover

One way of escaping capture when a predator threatens is to dive for
cover. So when animals move further from a safe refuge we might expect them
to be more vigilant, as they would need to be warned of danger earlier to
reach cover before the predator reaches them. This prediction has been borne
out in studies of the hoary marmot and in house sparrows, blue and willow
tits and yellow-eyed juncos (a kind of bunting). More recently, however,
Steven Lima of Simon Fraser University in Burnaby, British Columbia, has
found the opposite effect in flocks of house sparrows: when they are further
from cover, they are less vigilant. Lima points out that vigilance should
increase further from cover only if it enhances the animal’s safety more
than it does when the animal is close to cover. This is not necessarily
the case; an alter native strategy for the bird away from cover is to get
on with feeding so that it is satiated more quickly and can return to cover
sooner.

Far from acting as refuges, some objects in the environment represent
a hazard to prey, because they obscure a predator’s approach. Neil Metcalfe
of the University of Glasgow found that turnstones and purple sandpipers
feeding on rocky shores took account of this hazard in deciding how vigilant
to be; when an individual’s view was obstructed by nearby boulders it was
more vigilant, in order to spot any dogs or falcons approaching along the
shore.

Recent research has suggested that animals try to vary the pattern of
their bouts of vigilance in time so as to minimise their chances of being
eaten. The time an animal spends being vigilant is made up of bouts of watchfulness
separated by ‘inter-scan intervals’. These intervals are the times when
the danger is greatest, and the longer the interval the more likely it is
that a predator will be able to reach its unfortunate prey before it lifts
its head. The ostriches studied by Brian Bertram in Tsavo National Park,
Kenya, appear to vary at random the lengths of their inter-scan intervals,
so that an individual is equally likely to look up at any instant. An approaching
predator then has minimal information about when the prey is next likely
to look up and consequently has difficulty planning its approach. The situation
is very different, however, if the predator has seen the prey put its head
down.

A predator stalking Bertram’s ostriches, for example, could potentially
learn that 40 per cent of a solitary bird’s inter-scan intervals are longer
than 10 seconds so that, assuming the ostrich is blind to the predator’s
approach while feeding, it had a 40 per cent chance of pouncing on the ostrich
unseen if it could reach it within 10 seconds of it putting its head down.

How often predators actually take advantage of such information is largely
unknown, but Clare Fitzgibbon of the University of Cambridge has recently
found that cheetahs in the Serengeti in East Africa certainly do so. Faced
with two equidistant Thomson’s gazelles at the beginning of a concealed
stalking approach, a cheetah tends to end up chasing the one that has spent
less time being vigilant during the stalk. Such animals are less likely
to be vigilant when the cheetah initiates its chase at the end of the stalk,
are slower to flee and easier to catch. So individual gazelles cannot afford
to take too much advantage of the vigilance of others in the group, because
if they selfishly do less than their fair share of keeping watch they are
likely to pay with their lives.

Prey living in groups would do best to share out their vigilant periods
so as to ensure that at least some individuals are always vigilant, rather
than synchronising the times when they are watchful and so oscillating between
periods of maximum security and mass recklessness. The lesson is, do not
put all your vigilant eggs in one basket. But studies of flocks of house
sparrows and herds of pronghorn antelopes revealed that individuals lifted
their heads independently of each other. It may be that the time taken to
monitor one’s neighbours in order to achieve the necessary behavioural organisation
is not worth the benefit it bestows, and that randomly related bouts of
vigilance provide adequate protection, particularly in large groups.

In stark contrast to this random distribution of vigilance in groups
comes recent evidence for the existence of sentinels – individuals that
watch for trouble while the rest of the group feeds in security. Most claims
for such sentinels have been anecdotal, such as Thomas Bewick’s fanciful
description two centuries ago of the greylag goose ‘who, placed on some
eminence, with outstretched neck surveys every thing that moves within the
circle of the centre on which he takes his stand’. But for the scrub jays
of North America and the meerkats of the Kalahari Desert, sentinels are
a real phenomenon and form an integral part of a cooperative social system.
Kevin McGowan and Glen Woolfenden of the University of South Florida studied
family groups of the scrub jay and found that individuals take turns to
perch above the oak scrub, scanning continuously. Because this activity
correlates seasonally with the incidence of aerial predators, but not with
that of other jays’ territorial intrusions, and because the sentinel is
generally the bird that raises the alarm, the behaviour clearly functions
to warn others of danger.

The meerkats studied by David Macdonald of the University of Oxford
are the sentinels par excellence of the animal world. This small mongoose-like
carnivore forages through the desert in tightly knit social groups, with
individuals pausing momentarily to scan the barren landscape for predators
such as eagles, bat-eared foxes, wildcats and snakes. But when the group
forages in a restricted area, sentinel duty becomes a far more onerous task,
with a single individual sometimes perching precariously in a tree for an
hour or more while the rest of the group is feeding. There are consistent
differences in this behaviour between individuals, with a few animals taking
on the bulk of the vigilance burden.

It may be clear by now that ethologists have rather sidestepped the
thorny issue of whether one behaviour pattern really is more vigilant than
another. Rather than seek direct evidence for this – a difficult task –
they have instead looked for adaptive variability in what seem like vigilant
behaviours and, finding it, have implicitly taken this as evidence that
the behaviour is vigilant. Alternatively, researchers have simply assumed
that the behaviour in question is vigilant, and have then sought its function.
Although the adaptive story that emerges from many studies is a rich and
convincing one, it is worrying that we still have so little evidence that
a posture we assume to be vigilant really does increase the probability
that an animal will detect a predator.

The difficulty lies in measuring detection. This is normally done by
measuring the animal’s response and then assuming that detection and response
can be equated. But can they? Surely an animal might sometimes detect something
but choose to do nothing about it, or detect now but respond later. This
problem has beset studies attempting to test the idea that animals can detect
a predator’s approach earlier if they are gregarious. In a typical study,
researchers fly a model aerial predator towards birds in flocks of different
sizes and record the time that the first bird flies up from the ground in
each flock. Sometimes, the results show no relationship between flock size
and the timing of the first response. How do we explain this, given that
we would expect the first detection to occur earlier in a larger group,
everything else being equal? But everything else is not equal, because once
a predator has been detected, an individual in a larger group is safer –
it has less chance of being singled out for pursuit. In a small group, then,
an individual might flee at the very first sight of the predator’s approach,
before it was even sure how dangerous the predator was, or even before it
was certain that what it saw was really a predator. In larger groups, individuals
could afford to wait just a little longer while observing the approaching
object, before deciding that it really should be avoided. So the effect
of group size on the delay between detection and response may be exactly
opposite to that expected for the time to the first detection. Because what
we are measuring in this experiment – the time elapsed to the first response
– is really these two things added together, we cannot predict the effect
of group size in any straightforward way.

In an experiment with red-billed weaver birds, I found that adaptive
decision-making was indeed going on in the bird’s head, in the instant between
detecting a stimulus and deciding what to do about it. When birds responded,
very rapidly, to a new and so potentially dangerous stimulus – in this case,
a low-intensity 80-millisecond light flash from an unpredictable direction
– they either took flight, made movements showing an intention to flight
(wing flicking and crouching) or just turned the head, presumably to locate
the position of the light. The interesting finding was that as I increased
the size of the flock from 1 to 32, individuals were more likely to orient
themselves to the light source and less likely to flee or make movements
showing they intended to flee. In other words, birds in the smaller groups
could not afford the luxury of seeking more information about the stimulus:
they fled, just in case it was malevolent.

Although such findings are a problem for any researcher who assumes
that detection can be equated with response, they show, more positively,
the power that animals have for adaptive decision-making, even in the control
of what might seem like reflex responses. They also remind us that animals
can sometimes make mistakes. The problem for the individual that flees as
soon as it detects an uncertain event in its environment is that the event
may turn out to be innocuous, in which case it has wasted time and energy,
and perhaps lost contact with its companions. The distant soaring bird with
the finger-like tips to its wings might turn out to be a crow, and not a
hawk, for example.

These false alarms are more likely in a larger group, simply because
there are more animals around to make them. But the range of responsiveness
is also likely to be greater, so there are likely to be more flighty individuals
than in smaller groups. This would not matter to other birds if they all
made up their own minds about frightening stimuli, but one advantage of
living in a group is to take advantage of the detections made by others.
If the whole group blindly follows any individual that flees, there are
likely to be many group false alarms, and this is another reason why animals
in larger groups would be expected to be less vigilant, and less responsive
to uncertainty.

We know very little about how animals modify their responsiveness to
predators in an adaptive way, but my guess is that there is much to be discovered
here. The other area that calls out for further study is the extent to which
predators monitor the vigilance and responsiveness of their prey, and use
the information gained in planning their attacks. Fitzgibbon’s observations
of cheetahs and gazelles show that predators are as interested in these
things as biologists are, although perhaps for more practical reasons.

Dr John Lazarus lectures in the Department of Psychology at the University
of Newcastle upon Tyne.