Lecture Notes II - Foraging Behavior
Natural selection may favor 'efficient' foragers, and efficient
may mean that
What is maximized? Let's assume for now that an 'optimal'
foragers is attempting to maximize energy intake. If so, what decisions
must a forager or predator make?
individuals maximize energy intake or intake of some nutrient, or
minimize fluctuations in energy intake, or
maximize energy intake during certain periods . . .
Any food item has both a cost (time & energy) & a benefit
(net food value). The relative value of each of these determines how much
'profit' a particular item represents. In other words:
What types of food to eat?
Where & how long to search for food?
What type of search path to use?
'Profit' = net food value divided by time required to obtain & handle
the food item, and
Efficient foragers should select most profitable prey!
For example, let's look at two studies:
"Preferred length of Neomysis integer eaten by different sizes of
15-spined sticklebacks" (J. Exp. Mar. Biol. Ecol. 25:151-158)
Given that foragers may want to maximize 'profit', what should
they do when less than optimal prey are encountered?
CLASSICAL MODEL OF PREY CHOICE (also in text):
Predator - searching for 2 types of prey (1 & 2) that require search
times of S1 & S2
Prey - 2 types that yield E1 & E2
units of 'reward' (e.g., energy) AND take h1 &
h2 seconds to handle
So, their profitabilities = E1/h1
What should a predator's strategy be to maximize energy intake/unit
time? Should a predator take only PREY TYPE 1 & always ignore PREY
TYPE 2? Should a predator always take both?
Let PREY TYPE 1 be more profitable than PREY TYPE 2:
Depends on S1 (Search time for Type 1 prey)
E2/h2 > E1/S1
+ h1 (which would be true if S1
is large), then take both BUT if
E2/h2 < E1/S1
+ h1 (which would be true if S1
is small), then take only Type 1 prey
Assumptions of this model:
Prey value is measurable net energy or some other comparable single dimension
Handling time is fixed
Handling & searching cannot be done at the same time
Prey are recognized instantaneously (with no errors)
Prey are encountered sequentially & randomly
Energetic costs of handling are the same for different prey
Predators wish to maximize rate of energy (or some other measure of value)
Predictions of model:
Most profitable prey should never be ignored
Less profitable prey should be ignored according to preceding equation
Exclusion of less profitable prey should be all-or-nothing (depending on
direction of inequality)
Exclusion of less profitable prey does not depend on S2
(when Type 1 prey are sufficiently abundant using time to handle Type 2
prey is not profitable)
Test of the model (Krebs et al. 1977):
Great Tits (Parus major)
captive birds presented with 2 different-sized pieces of mealworms (Large
= Type 1 prey & Small = Type 2 prey) on a moving belt.
‘Search time' for large prey (Type 1 prey) varied by varying distance between
successive large prey on the moving belt (to cross the 'threshold' of the
Results ---> Demonstrated absence of 'all-or-nothing' response,
i.e., partial preferences
Extra payoff for specializing (prey/sec) on large
worms rather than taking both (From Fig. 3.6b; Krebs and Davies 1993)
Runs of 'bad luck'
Need to sample the environment
Other factors may influence behavior, such as the need to avoid predators,
find mates, & so on
Exploitation of patches
Prey availability within a patch decreases as a result of the predator's
foraging activity because of:
Depletion of prey
Evasive action by prey
As a result, to maximize the rate of gain of a resource, predators should
follow the 'marginal value theorem':
To maximize gain (e.g., energy) per unit time, a predator should leave
at the point (maximum net gain) that gives the greatest gain or food intake
per unit time (steepest slope of the line). The line is not as steep (which
means less gain or intake per unit time) when the predator leaves too early
(or too late).
If travel time increases (e.g., patches further apart), the optimal
time to stay in a patch also increases:
If patches vary in quality, each patch should be exploited until the
gain rate within the patch drops to the average for the environment:
Parallel lines represent equal rates of energy gain
Assumptions of the Marginal
1 - Each patch type is recognized instantaneously
2 - Travel time between patches is known by the predator
3 - Gain curve is smooth, continuous, & decelerating
4 - Travel time between & searching within a patch have equal energy
Predictions of the Marginal Value Theorem:
1 - If travel time & the gain curve are known, then Topt
can be predicted
2 - If there is more than one patch type in an environment, all should
be reduced to the same gain rate
Tests of the Marginal Value Theorem:
Great Tits foraging in large aviary for pieces of mealworm hidden in
sawdust-filled plastic cups on the 'branches' of artificial trees (Cowie
When average feeding rate in the aviary 'environment' was high, the birds
should spend less time at each patch (plastic cup)
If travel time between 'patches' increases, then birds should spend more
time at each 'patch'
TO TEST PREDICTION 1 ---> Six Great Tits foraged individually in aviary
for 2 10-min trials each:
Each point = different aviary environment
TO TEST PREDICTION 2 ---> Travel time was increased by placing a top on
each cup, which took about 15 sec to remove (so, 'travel time' increased
from about 5 sec to about 20 sec):
Tests of the MVT often show qualitative but not quantitative support.
Other factors important besides energy (e.g., predation risk)
'Rich' patches = increased competition with others (= smaller loads?)
Animals may need to 'sample' the environment
Foraging for young located in a nest or den: Central Place Foraging
Assume a diminishing energy gain with increasing load size (e.g., a
bird with a beakful of insects) or increasing time in patch:
The optimal load size (energy delivered to young/unit time) may be influenced
by the time it takes to travel between a 'central place' (nest or den)
& a feeding area
Test of Central Place Foraging model using House Martins (Bryant and Turner
Cumulative food gain within patches for an average
House Martin (open circles). The straight lines show maximum overall rates
of food gain within bouts for foraging near the nest (Tn),
at the mean observed distance from the nest (Tm),
& far from the nest (Td).
The Bryant and Turner (1982) model yields a different value for optimal
bolus size (Bopt) for each travel
Martin (Delichon urbica)
Optimal Load Size
Observed Load Size
Effect on foraging behavior when a predator requires as particular
feed in forests (deciduous leaves with lots of energy but little sodium)
& small lakes (aquatic plants rich in sodium but less energy)
Nutrient quality - more important to herbivores because plants often lack
essential nutrients & careful selection is required for a balanced
Example = Moose diet & need for sodium & energy
- What is the optimal mixture?
The diet of moose is constrained by the need for sodium
& energy: the daily requirements
are shown by the lines above (and moose must eat a mixture
of plants which lies in the space
above the two lines). The third constraint is the size
of a moose's rumen (see line above).
Aquatic plants are bulkier than terrestrial plants, so
fewer grams can be fitted into the rumen.
The moose diet was found to lie at the point inside the
triangle that maximizes daily energy
intake (indicated by the star) (Belovsky 1978, as cited
by Krebs and Davies 1993; Figure from
Krebs and Davies 1993, p.71)
Photo source: http://www.primatesofpanama.org/howler.htm
Howlers - feed on fruits, flowers, & leaves of trees (96 species
present in study area)
% of total
feeding time (B)
% of total
trees present (A)
B/A x 100
Howlers obtain most food from relatively rare trees
Howlers maximize intake of protein, water, & most amino acids
Howlers minimize intake of crude fiber & plant secondary compounds
(e.g., alkaloids & tannins)
Stochastic foraging models
Assume that travel times & patch quality vary in an unpredictable
way. Possible consequences:
Risk of doing badly or well in patches:
Animals might respond to the risk of doing badly or well in a particular
patch or with a particular prey
Animals might have a strategy for assessing patch quality & acquiring
information by sampling
So, Juncos switched from 'risk-averse' to 'risk-prone' as deprivation increased.
(Junco phaeonotus; Caraco et al. 1980) - offered a choice between
2 feeding sites in an aviary (with the other site no longer
available after a choice is made):
Site 1 = constant reward (e.g., 4 mealworms)
Site 2 = unpredictable reward (e.g., 0 or 8 mealworms but a mean of 4 mealworms)
AFTER 1 HOUR STARVATION, juncos preferred the predictable site BUT AFTER
4 HOURS STARVATION, juncos preferred the unpredictable site.
Expected energy budget rule: Be risk prone if the daily energy budget
is negative & be risk averse if daily energy budget is positive.
After 1 hour of deprivation:
Constant site - provided enough food to meet energy requirements
Variable site - 50% chance of providing more than enough & 50% chance
of not providing enough
After 4 hours of deprivation:
Constant site - NO chance of providing enough food
Variable site - 50% chance of providing enough food
Effect of 'information' on foraging behavior
Lima, S.L. 1984. Downy Woodpecker
foraging behavior: efficient sampling in simple stochastic environments.
A forager acquiring information about a patch should, on average, stay
longer than predicted by the marginal value theorem because extra time
may reveal that the patch is better than the 'current' estimate. In other
words, it may pay to 'sacrifice' the maximum intake rate to gain extra
Do foragers acquire & use 'information'?
Study in small woodlot in southeastern Michigan
37-cm length of sapling ash tree, 2.5-3 cm in diameter
24 holes drilled (0.5 cm deep) & arranged into 6 groups of 4 holes
each. Groups were placed at 4-cm intervals & covered with a strip of
may or may not be a single sunflower seed in any given hole of a patch
60 'patches' (about 4 m apart) in environment
Downy Woodpeckers - given experience in patch use by providing them with
60 totally full patches (24 seeds in 24 holes) for 14 days
Downies - then exposed to 3 'environments':
'Full' = 36 patches empty & 24 patches full
'Half-full' = 36 patches empty & 24 patches half-full (12 seeds
in 24 holes)
'Quarter-full' = 30 patches empty & 30 patches quarter-full
(6 seeds in 24 holes)
Day 1 ---> Downies expected only totally full patches
& opened almost all holes in empty patches (although no seeds were
Days 2 - 8 ---> Downies learned that some patches had
seeds & some did not
Days 9 - 11 ---> Downies opened 1.7 holes/empty patch
Day 12 ---> 1st day of half-full environment
Day 13 ---> Downies opened 5 holes/empty patch
Day 23 ---> 1st day of quarter-full environment
Days 29 - 31 ---> Downies opened 6.3 holes/empty patch
Increased uncertainty that patch was empty or had seeds led to increased
sampling by woodpeckers before giving a patch up as empty
In patches with seeds, the woodpeckers' strategy was to open every hole
(efficient but suboptimal . . . why??)
Downies distinguished between patches with seeds & empty patches, &
gathered enough information to exploit different environments in a reasonably
Downies responded rather quickly to environmental changes
Foraging & conflicting demands
Behavior of foraging animals may be influenced by need to watch for
predators, search for mates, defend territories, defend nest sites, and
so on (e.g., Martindale 1982):
While adult Gila Woodpeckers forage, nestlings are vulnerable to intruders
& predators (e.g., other Gila Woodpeckers or Common Flickers attempting
to take over the nest site). How does the need for nest defense alter foraging
N & F = 2 food patches identical except for distance from
nest (N = near & F = far)
TN & TF = travel
BN & BF = benefit
curves (proportional to rate at which food can be delivered to the nest)
CH & CL = cost
functions (proportional to probability that a successful attack on the
tN & tF = foraging
times that optimize delivery rate
tNL & tFL= foraging
times that maximize net benefit (survival of young) at low 'attack' rate
('far' patch affected more than 'near' patch)
tNH = optimal time in near patch at high
attack rate (far patch no longer confers positive net benefit)
Increased risk of predation ---> Defending forager should move closer to
the nest, leave patches sooner, & deliver smaller loads
Males behaved as expected but females didn't
Influence of a predator on the optimal foraging behaviour of sticklebacks
(Milinski and Heller 1978):
From Milinksi and Heller 1978, as cited by Krebs and
(Figure from Krebs and Davies 1993, p. 69)
Flocks of finches generally move in a way that minimizes path crossing
(to reduce chances of returning to areas that have already been 'depleted')
Animals may increase amount of 'turning' in their paths following encounters
with prey (possible adaptation for exploiting a patchily-distributed food
resource) (Pyke 1978).
Animals may avoid patches and areas within patches already visited (Zach
and Falls 1978):
Ovenbirds exhibit increasingly tortuous search paths & reduced speed
in areas with greater prey densities.
Age-specific foraging behavior (e.g., Wunderle
Each of these hypotheses is likely to be correct depending on the species.
After young become independent of parents, recognition & selection
can differ with age because:
Foraging sites - many (but not all) studies have revealed age- related
differences in habitat or patch use.
Dominant adults displace juveniles from certain foraging sites
Inability of juveniles to accurately evaluate patch differences
Diet differences arise from lack of juvenile proficiency for searching,
capturing, or handling prey
Different nutritional requirements
Search methods & patterns - quantification of searching methods is
difficult & few field studies have compared adults & juveniles
Food recognition & selection - How do juveniles recognize prey?:
Juveniles may learn from parents
Juveniles may learn from personal experience
Juvenile learning is mostly independent of personal experience & is
primarily genetically based
Prey capture - capture techniques commonly proceed from simple movements
to more complex movements requiring coordination & skill. Generally,
the greater the skill needed to capture prey, the less successful are juveniles
in comparison with adults.
Juveniles are less proficient than adults
Juveniles are less skilled at some other aspect of foraging (e.g., capturing
or handling) & compensate by selecting a different diet
Juveniles have different nutrient requirements
Adult dominance influences juvenile diet selection by preventing juveniles
from selecting the most profitable prey types
Food handling time & technique:
Handling time is the time required for a forager to eat a captured food
item, and can represent a substantial portion of foraging time in some
Juveniles in a variety of species & with a variety of diet types (piscivores,
insectivores, mollusc eaters, scavengers, carnivores, & seed eaters)
have been found to be less adept at handling food items
lack of food handling proficiency can have major implications for survival
because it can increase the likelihood that food items will be stolen &
also increase the time & energy expended in acquiring food
improvement in juvenile handling time & ability can largely be attributed
to LEARNING & to MATURATION.
Belovsky, G.E. 1978. Diet optimization in a generalist
herbivore: the moose. Theor. Pop. Biol. 14:105-134.
Bryant, D.M. and A.K. Turner. 1982. Central place foraging
by swallows: the question of load size. Anim. Behav. 30:845-856.
Caraco, T., S. Martindale, and T.S. Whitham. 1980. An
empirical demonstration of risk-sensitive foraging preferences. Anim. Behav.
Cody, M.L. 1971. Finch flocks in the Mohave Desert. Theor.
Pop. Biol. 2:142-158.
Cowie, R.J. 1977. Optimal foraging in Great Tits, Parus
major. Nature 268:137-139.
Davies, N. B. 1977. Prey selection and social behaviour
in wagtails (Aves: Motacillidae). Journal of Animal Ecology 46:37-57.
Glander, K.E. 1981. Feeding patterns in mantled howling
monkeys. Pp. 231-257 in A.C. Kamil and T.D. Sargent, ed., Foraging
behavior: ecological, ethological and psychological approaches. Garland
STPM Press, New York.
Krebs, J.R. and N.B. Davies. 1993. An introduction to
behavioural ecology, third ed. Blackwell Scientific Publications, London.
Krebs, J. R., J. T. Erichsen, M. I. Webber & E. L.
Charnov. 1977. Optimal prey selection in the great tit (Parus major).
Anim. Behav. 25: 30-38.
Martindale, S. 1982. Nest
defense and central place foraging: a model and experiment. Behav. Ecol.
Milinski, M. & R. Heller. 1978. Influence of a predator
on the optimal foraging behaviour of sticklebacks. Nature 275:642-644.
Pyke, G.H. 1978. Are animals efficient harvesters? Anim.
Wunderlee, J. M., Jr. 1991. Age-specific foraging proficiency
in birds. Curr. Ornithol. 8: 273-324.
Zach, R. & J. B. Falls. 1978. Prey selection by captive
ovenbirds (Aves: Parulidae). J. Anim. Ecol. 47: 929-943.
More lecture notes:
- Optimal Foraging Simulation
Optimal Foraging Models
Background to Optimal Foraging
foraging experiments on captive Steller sea lions: a feasibility study
theory; Foraging Strategies
use in cranes: a field test of optimal foraging predictions
to Behavioral Ecology syllabus