See discussions, stats, and author profiles for this publication at: http://www.researchgate.

net/publication/226085771

Brown, J.S. Patch use as an indicator of habitat

preference, predation risk, and competition.
Behav. Ecol. Sociobiol.

ARTICLE in BEHAVIORAL ECOLOGY AND SOCIOBIOLOGY · DECEMBER 1987

Impact Factor: 2.35 · DOI: 10.1007/BF00395696

CITATIONS READS

680 280

1 AUTHOR:

Joel S. Brown
University of Illinois at Chicago
216 PUBLICATIONS 8,571 CITATIONS

SEE PROFILE

Available from: Joel S. Brown

Retrieved on: 12 October 2015
 Behavioral Ecology
Behav Ecol Sociobiol (1988) 22:37 47 and Sociobiology
9 Springer-Verlag 1988

Patch use as an indicator of habitat preference,

predation risk, and competition
Joel S. Brown*
Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ 85721, USA

Received January 28, 1987 / Accepted September 25, 1987

Summary. A technique for using patch giving up can be used to investigate the properties of commu-
densities to investigate habitat preferences, preda- nities. A recent example of this includes Rosen-
tion risk, and interspecific competitive relation- zweig's (1979 et seq.) theory of habitat selection
ships is theoretically analyzed and empirically in- where the behavior of individuals behaving opti-
vestigated. Giving up densities, the density of re- mally is used to predict the outcome of intra- and
sources within a patch at which an individual interspecific interactions (Rosenzweig 1981, 1985;
ceases foraging, provide considerably more infor- Pimm and Rosenzweig 1981; Pimm et al. 1985).
mation than simply the amount of resources har- Other examples include short-term apparent com-
vested. The giving up density of a forager, which petition (Holt and Kotler 1987), moose foraging
is behaving optimally, should correspond to a har- dynamics (Belovsky 1978 et seq.), and resource
vest rate that just balances the metabolic costs of theory (Tilman 1982, 1985).
foraging, the predation cost of foraging, and the Testing models of population interactions
missed opportunity cost of not engaging in alterna- based upon the foraging behavior of individuals
tive activities. In addition, changes in giving up often requires measuring habitat or patch use, hab-
densities in response to climatic factors, predation itat preferences, and the rejection or acceptance
risk, and missed opportunities can be used to test of patches. In this paper, I present a measuring
the model and to examine the consistency of the technique to accomplish these goals. The technique
foragers' behavior. The technique was applied to uses artificial or manipulated resource patches to
a community of four Arizonan granivorous ro- measure a forager's giving up density. The tech-
dents (Perognathus amplus, Dipodomys merriami, nique is applicable to communities of active for-
Ammospermophilus harrisii, and Spermophilus tere- agers seeking comparatively immobile prey.
ticaudus). Aluminum trays filled with 3 grams of I begin by discussing the theory which provides
millet seeds mixed into 3 liters of sifted soil pro- an interpretation of patch giving up densities. The
vided resource patches. The seeds remaining fol- theory is an extension of Charnov's (1976) mar-
lowing a night or day of foraging were used to ginal value theorem and it uses what, in economics,
determine the giving up density, and footprints in is called the "marginal rate of substitution" (see
the sifted sand indicated the identity of the forager. Russell and Wilkinson 1979) to include the effects
Giving up densities consistently differed in re- of predation risk and alternative activities on patch
sponse to forager species, microhabitat (bush ver- use. I then give some preliminary results from ap-
sus open), date, and station. The data also provide plying this method to a community of four desert
useful information regarding the relative foraging granivorous rodent species: Arizona pocket mouse
efficiencies and microhabitat preferences of the (Perognathus amplus), Merriam's kangaroo rat
coexisting rodent species. (Dipodomys merriami), round-tailed ground squir-
rel (Spermophilus tereticaudus), and Harris's ante-
lope ground squirrel (Ammospermophilus harrisii).
Introduction
As first suggested by MacArthur and Pianka Optimal patch use
(~ 966) and Emlen (1966), optimal foraging theory
Theoretical treatments of optimal habitat use are
* Current address: Department of Biological Sciences, Universi- applicable to systems where foragers are able to
ty of Illinois at Chicago, Chicago, IL 60680, USA identify and direct their foraging efforts to subsets
38

of the environment (i.e. patches) that on average all alternative activites, a. I assume that the realiza-
yield higher harvest rates or benefits than the envi- tion of fitness is discrete and occurs following a
ronment at large. Once a habitat patch is located, fixed amount of time T. If the individual is preyed
a forager must decide whether to accept the oppor- upon during this interval its fitness is zero. The
tunity to harvest the patch (Rosenzweig 1974), and goal of the forager is to divide its time between
it must decide how much time or effort to devote foraging and alternative activities so as to maxi-
to those patches it accepts for harvesting (Charnov mize fitness.
1976). By varying the foraging time allotted to each Net energy gain, e, and the probability of sur-
patch, a forager can vary the total, average, mar- viving predation, p, are functions of time spent
ginal or net reward from each patch. foraging. Assume that time spent foraging is di-
In the simplest case of habitat utilization, an vided over m resource patches. Resource patches
individual can only forage and has no alternative may vary with regard to initial resource density
activities. Assume that resources are distributed in and predation risk. Harvest rate is assumed to be
discrete patches and that foragers deplete the re- an increasing function of patch resource density.
sources as they harvest a patch. However, assume As a forager devotes harvest time to the patch,
also, that the distribution of resources among its resources are depleted and the harvest rate de-
patches remains fixed. In other words, although clines. Let t I = ( 6 , " , tin) denote the vector of times
the resources of a given patch are depleted by a allotted to each of the m resource patches. Assume
forager while it is in the patch, the resources of that there are s alternative activities that contribute
the entire environment are not. Under these as- to the other inputs into fitness, a. Let t " = ( t m + b ,
sumptions, the marginal value theorem (Charnov tm+s) be the vector of times devoted to each alter-
1976) states that the rate of energy gain to a forager native activity. Note that alternative activities may
is maximized, and by assumption its fitness is also also expend energy and incur predation risk. For
maximized, when the forager's quitting harvest accounting purposes, in what follows, I will include
rate in a patch equals its average harvest. these in the vector of inputs into fitness from alter-
The above assumptions, which lead to the mar- native activities. Let
ginal value theorem, are restrictive. More realisti-
cally, foraging in patches will affect not only ener- G [e (ts), p (t~), a (ta)] = p (t y) .f[e (ts), a (ta)]
gy gain, but also other aspects of fitness such as
denote the expected per capita rate of growth (fit-
predation risk. Empirical work suggests that for-
ness) subject to the constraint that the sum of the
agers balance the benefits of energetic reward and
tj's and t~'s equals T where j = 1,..., m and i = m +
the cost of predation when making foraging de- 1,..., m + s. Futhermore, 0 < p (t s) _<1 is the proba-
cisions (Milinski and Heller 1978; Sih 1980; Grubb
bility of surviving to realize fitness as a function
and Greenwald 1982; Werner etal. 1983; Lima
of times exposed to predators while foraging, and
et al. 1985). Furthermore, most organisms can en-
F[e(ti), a(t")]>0 is the expected per capita rate
gage in other fitness-determining activities besides
of growth given the individual survives predation
foraging such as territorial defense, mate finding,
while foraging.
resting, dormancy, grooming, and nest mainte- I assume that energy enhances fitness (i.e. ~ G~
nance. Finally, foraging activity of individuals may ~e> 0) and that spending time foraging decreases
not only depress the resources of a given patch the probability of surviving predation to realize
but also, at least for a time, depress the resources
fitness (i.e. ~p/~tj<O where j = 1,..., m). If patch
of the entire environment. For example, following j has a lower risk of predation than patch k then
sunrise, hummingbirds and nectarivorous insects
~p/~ t; > ~p/~tk.
often deplete their nectar resources (Feinsinger The optimal values of t~ and ta can be found
1976; Schaffer et al. 1979; Brown et al. 1981). through the technique of Lagrange multipliers
In what follows, I relax these restrictive as- (Chiang 1974). I will consider those patches and
sumptions (1) by assuming that foraging activity alternative activities such that the optimal values
affects both net energy gain and predation risk; for tj and ti are greater than zero for j = 1,..., m,
(2) by permitting the foragers to engage in alterna- and i = m + 1,...m+s. Setting the necessary condi-
tive activities; and (3) by permitting the depletion tion for the optimal tj equal to the necessary condi-
of resources over the entire environment. tion for the optimal ti yields for all j = 1,.--,m and
Let fitness, denoted by G, be a function of net i = m + l,...,m+s:
energy gain from foraging, e, the probability of
surviving predation while foraging, p, and a vector [~ G (')/~ el [~e (.)/8 tj] + [~ G (-)/~p] [~p (")/~tj]
of inputs into fitness from engaging in any and = [~ G (')/~ a] [ea (.)/eh]
 39

where (.) is shorthand for the functions' variables.

Rearranging yields:
~e(.)/~tj
_ IS a (.)/ 8 a] [8 a (.)/8 ti] P
[~ a (')/8 e]
[~ a(')/~p] [~p(.)/~ tj] (1)
[8 a ( - ) / 8 e]

Note that the derivatives of G and a with respect e

to a and ti respectively are vector valued and the Fig. 1. The relationship between net energy and predation risk,
inner product of these two vectors is a scalar. and fitness. The x-axis is net energy gained from foraging (e);
The left side of equation (1) is the quitting net and the y-axis is the probability of surviving predation while
foraging (p). Each isopleth gives the combinations of energy
harvest rate of an individual in patch j (quitting and predation risk that yield the same fitness. The fitness of
harvest rate minus the energetic cost of foraging). R2 is greater than that of R1. The isopleths are concave to
The quitting harvest rate which satisfies (1) is gen- the origin because of diminishing returns to fitness from in-
eral. It applies to foragers which can engage in creasing energy and because individuals which have acquired
abundant reserves of energy have more potential fitness to lose
alternative behaviors and incur a risk of predation from predation than individuals which have not. The negative
while foraging. The first term on the right side and reciprocal of the slope of any point along an isopleth is
of (1) is the change in the amount of alternative the marginal rate of substitution of energy for predation risk,
activities per unit time weighted (multiplied as the MRSev (Caraco 1979)
inner product of two vectors) by the ratio of the
marginal fitness of alternative activites, ~ G (-)/8 a,
and the marginal fitness of energy, ~ G(.)/8 e. The
second term is the change in predation risk in- by p(e), when substituted into the fitness function
curred per unit time weighted by the ratio of the must satisfy:
marginal fitness of surviving predation, ~ G (.)/~p, G [e, p (e), a ] = R (2)
and the marginal fitness of energy.
where R is some constant. The negative of the slope
of this isopleth gives the rate at which an individual
The cost of predation and missed opportunity costs should be willing to exchange energy into safety
from predation risk (i.e. the MRS of predation
When comparing such disparate inputs as energy, risk for energy). The MRS is the negative of the
predation risk, nest maintenance, and mate selec- slope of the isopleth. To find the slope of the iso-
tion, what is the appropriate common currency? pleth take the derivative of expression (2) with re-
Since energy, predation risk, and alternative activi- spect to net energy gain, e, and rearrange to give:
ties jointly contribute to this fitness, each can be
8p (e)/8 e = - [8 G (-)/8 e]/[~ G (.)/Sp] (3)
considered as a surrogate for fitness (see Cheverton
et al. 1985). To see how one input can be converted A similar argument can be made to show that
into the currency of the other, consider net energy the slope of fitness isopleths in the state space of
gain and predation risk. We ask the question, how net energy gain and alternative activities is given
much additional energetic gain would it take to by:
get the individual forager to accept a higher risk
of predation? This exchange rate is called the mar- 8 a (e)/~ e = - [8 G (')/~e]/[~ G (-)/8 a] (4)
ginal rate of substitution, MRS, of energy for pre- where a (e) defines the isopleth that gives the com-
dation risk (see Caraco 1979 for the use of MRS binations of net energetic gain and alternative ac-
in ecology). Figure I is an example of fitness iso- tivities such that fitness is held constant.
pleths in the state space of net energy gain and The negatives of expressions (3) and (4) are
probability of surviving predation while foraging. the MRS's of predation risk and alternative activi-
Each isopleth gives the combinations of energy and ties for energy respectively. The negatives and re-
probability of surviving predation such that fitness ciprocals of (3) and (4) are the MRS's of energy
is held constant. An isopleth can be expressed as for predation risk and alternative activities (Car-
a function, p (e), which relates the net energy gain aco 1979). These exchange rates go from units of
to the value of surviving predation which will hold predation risk and alternative activities into units
fitness constant. The points of the isopleth, given of energy, MRSep and MRSea respectively:
40

MRSep [8 G ( ")/ap]/[~ G (')/8 e] = F/p (8 F/~ e)

= (5) of foraging over remaining dormant. Third, if two
MRSea = [8 G ( ) / 8 a]/[8 a (.)/8 e] (6) patches share the same energetic cost of foraging
and the same harvest rate function, then any differ-
By assumption, MRSep is always positive and ences in quitting harvest rates should reflect differ-
MRSea is vector valued giving the MRS's of energy ences in patch specific risks of predation. Fourth,
for the other inputs. if two patches differ in energetic costs and in risk
Note that in general these MRS's are complex of predation in such a way that one is safer but
functions of energy gain, predation risk, and alter- energetically more costly than the other, then the
native activities. For instance, it may often be rea- forager's relative quitting harvest rates in the two
sonable to assume that there are diminishing re- patches may reflect its state. If the forager is in
turns to fitness from increasing net energy gain a state of high energy gain then it may have a
(i.e. the marginal fitness of energy may decline with lower quitting harvest rate in the less risky patch.
net energy gain). On the other hand, the marginal The converse may occur if the forager is in a state
fitness of surviving predation increases with net of low energy gain (see McNamara and Houston
energy gain; 82 G/~p ~ e = ~ F/O e > 0 (if the individ- 1986 for how an individual's state influences be-
ual has more energy it has more to lose from being havior).
eaten). Thus, as the individual acquires more ener-
gy the MRS~p increases.
Measuring patch use
We can now reconsider expression (1) which
gave the quitting net harvest rate of an optimally Researchers have used four techniques for measur-
foraging individual in any arbitrary patchj. Substi- ing patch use. These include measuring giving up
tuting the MRS's given by expressions (5) and (6) times (Krebs et al. 1974; Hubbard and Cook 1978;
into (1) gives: Townsend and Hildrew 1980), measuring total
time spent in a patch (Cowie 1977; Hartling and
e/~ t~= - MRSep [~p (-)/~ tj] + MRSea [8 a (')/8 ti] Plowright 1979), quitting harvest rates (Pyke 1978,
(7) 1980; Milinski 1979; Hodges 1981), and the giving
The first term of the right hand side of expres- up density of resources (Whitham 1977; Hodges
sion (7) is the cost of predation expressed in units and Wolf 1981). These authors were interested in
of energy (P) and the second term is the value determining foraging decision rules and seeing
of alternative activities expressed in units of energy whether models of optimal foraging provide good
(MOC). The net quitting harvest rate can be split characterizations of foraging behavior [see Krebs
between the rate of harvest, H, and the energetic et al. (1983) and Pyke (1984) and references there-
cost of foraging, C. Thus, the quitting harvest rate in].
satisfies: Here, I will advocate: 1) the use of controlled
field experiments, 2) the use of giving up densities
H=C+P+MOC as a measure of patch use, and 3) the use of forag-
A forager should leave each resource patch ing behavior to investigate predation risk, habitat
when the patch harvest rate is no longer greater preferences, and interspecific competitive relation-
than the sum of the energetic, predation, and ships.
missed opportunity costs of foraging. Controlled field experiments using artificial or
The above result comes from applying an opti- manipulated resource patches offer two important
mization criterion to a general model of patch ex- advantages. First, the foragers remain in their nat-
ploitation. For illustration, the Appendix considers ural environment and are faced with familiar alter-
a specific example of a fitness function for an or- native activities, competitive interactions, and pre-
ganism which can either forage in one of two dation risks. Second, the use of artificial patches
patches or remain dormant. From the example in permits the controlled manipulation of one or sev-
the appendix (or from careful consideration of the eral variables of interest while the available set of
general formulation), a number of deductions can alternative activities is held constant.
be made. First, if patches share the same risk of If harvest rates are a function of patch type
predation, cost of foraging and harvest rate func- and resource density then giving up densities
tion, then they should be foraged to the same quit- (GUD's) provide an estimate of quitting harvest
ting harvest rate. Second, if fitness is only a func- rates. In many circumstances it is easier to measure
tion of net energy gain and there is no predation G U D ' s rather than quitting harvest rates. Measur-
risk, then each patch should be harvested until har- ing GUD's requires assessing the remaining den-
vest rate just compensates for the additional cost sity of resources following use by one or several
 41

foragers. On the other hand, measuring the quit- same missed opportunity cost in each patch. Har-
ting harvest rate requires an organism whose en- vest rate can be controlled by ensuring that the
counter and capture of prey can be observed and structure of the artificial patches is the same with
timed. It is particularly difficult to measure the regards to factors that affect harvest rate, such as
instantaneous harvest rate of a forager as it leaves substrate and resource type. The energetic cost can
a patch. be controlled by maintaining constant climatic fac-
To validate the model, all four components of tors across patches. Conversely, different combina-
the model can be tested. To test for the effect of tions of abiotic factors such as temperature, hu-
harvest rate (H) on G U D ' s requires holding C, midity, or wind can be used to test their effects
P, and M O C constant. Adjacent manipulated re- on foraging costs. Differences in the cost of preda-
source patches in the same microhabitat should tion can be measured if patches are manipulated
not differ in C, P, or MOC. Thus, G U D ' s in adja- in such a way that energetic costs, harvest rates,
cent patches which differ in substrate should reflect and missed opportunity costs remain constant.
differences in harvest rates. The patch with the
'slower' substrate should have the correspondingly
higher G U D which just equalizes the quitting har- Methods
vest rates in the two patches. I applied the approach of the previous section to a community
To test for the effect of energetic costs (C) on of desert granivorous rodents in September 1983. The study
G U D ' s requires holding H, P, and M O C constant. site is located near Tucson, Arizona (see Brown 1986). The
For endotherms, the metabolic costs of foraging dominant perennial plant species in descending order of
should be influenced by temperature when ambient groundcover are creosote (Larrea tridentata), desert zinnia (Zin-
nia sp.), mouse ears (Coldenia canescens), and mesquite (Prosop-
temperatures are below the forager's thermal neu- sis juliflora). Total vegetation ground cover by perennials is
tral zone. Adjacent patches with the same substrate about 20%. There are four granivorous rodent species (in pa-
should not differ in H, P, or MOC. If one patch rentheses : numbers and mean weights of individuals trapped):
is rendered colder then it should have the higher round-tailed ground squirrel, Sperrnophilus tereticaudus
(18/121 g); Merriam's kangaroo rat, Dipodomys merriarni
G U D which just compensates the forager for its (16/37 g); Arizona pocket mouse, Perognathus amplus
higher energetic cost of foraging in that patch. (55/12g); and antelope ground squirrel, Ammospermophilus
Similar tests can be used to test for the effects harrisii, (12/104 g).
of P and M O C on GUD's. Increasing P (e.g. by Aluminum trays, measuring 45 cm on a side and 2.5 cm
manipulating predator densities or cues of preda- deep, filled with 3 g of unhusked-millet seed mixed into 3 l
of soil provided resource patches. A total of 60 trays were
tion) should increase GUD's. Similarly, increasing divided over two grids. Each grid was laid out as a seven by
the M O C (e.g. by providing alternative resources seven small mammal trap grid with stations 25 m apart. The
or foraging opportunities) should increase GUD's. thirty seed trays on each grid were divided into pairs and as-
By manipulating patches in ways which are known signed to 15 stations picked at random from the 49 trap sta-
tions. Live trapping and seedtrays were not run simultaneously.
to increase predation risk, harvest rate, foraging
At each station with seedtrays, one tray was placed directly
costs, or missed opportunity costs, the researcher under the canopy of a creosote bush and the other was placed
can test whether the forager's behavior is consis- 2 4 m away from the first in the open microhabitat. Although
tent with that required to maximize G (-). shadowed by the canopy, the surface of trays under shrubs
The model can also be used to allow the forager was unobstructed by branches or leaves up to a height of no
less than 20 cm. The trays in the open were placed on bare
to reveal its preferences and assessments of the en- ground at least 2 m from the nearest creosote bush.
vironment. The researcher assumes that the for- For seven mornings (9, 10, 11, 14, 15, 17, and 18 Sep-
ager's G U D is a truthful revelation of its harvest tember) and afternoons (9, 10, 11, 13, 14, 15, and 17 Sept.),
rate and foraging costs. Temporal or spatial differ- I collected foraging data from the seed trays (rain prevented
ences in the G U D ' s of a forager reflect the effects consecutive mornings and afternoons). The morning data col-
lection was at sunrise, a period following the cessation of activi-
of different habitats on missed opportunity costs, ty by pocket mice and kangaroo rats and preceding the initia-
harvest rates, foraging costs, and predation risk. tion of activity by ground squirrels. The afternoon data collec-
The researcher can investigate species and habitat tion at sunset followed squirrel activity and preceded that of
specific differences in any one of these costs by pocket mice and kangaroo rats. Thus, pocket mice and kan-
garoo rats had one entire night to forage the trays and squirrels
holding other costs constant among habitats. had one entire day. (The rodents were very familiar with these
Missed opportunity cost can be controlled for seed trays from similar work done in the latter part of August.)
by ensuring that several patches are available to Data collection from trays consisted of noting any foot-
the same forager. If these patches are within a rela- prints in the sired soil, sifting the soil to recover the remaining
tively short distance of each other, then while for- seeds, and recharging the trays with millet. The distinctiveness
of footprints permitted identification of the forager down to
aging in either patch the forager has the same set species and sometimes to the exact individual based upon toe-
of alternative activities and thus, experiences the clips. The squirrels could be distinguished by the presence
42

Table 1. The mean G U D ' s (in grams) for the pocket mouse Table 2. The results of two one-way ANOVA's (separate analy-
(P.a.), kangaroo rat (D.m.), and squirrel species (S.t. and A.h.) ses for the bush and open microhabitats) showing the differ-
in the bush and open microhabitat. Bush/Open preference is ences between species in GUD's. The group variables are the
determined by a sign test comparing the number of times the four species and G U D is the dependent variable (log transform
bush tray at a station had the lower G U D (number preceding of grams of millet). The column and row headings are the four
the comma) to the number of times the open tray had the species (abbreviated scientific names). Entries are the F of im-
lower G U D (number following the comma). A " B " or " O " provements for each pairwise comparison. Entries above and
indicates whether the bush or open had the significantly lower below the diagonal are for the bush and open microhabitats
GUD respectively. The error mean sum of squares for the bush and
open analyses are 0.646 and 0.602 respectively. Because the
Species Bush Sample Open Sample Bush/open six pairwise comparisons within each analysis are not orthogo-
GUD size GUD size preference nal, the error rate of each test was adjusted according to the
Dunn-Sidak method (Sokal and Rohlf 1981)
P. amplus 0.439 121 0.591 103 51,35 B*
D. merriami 0.952 51 0.610 86 14,52 O*** Species P.a. D.m. S.t. A.h.
A. harrisii 0.994 39 1.591 32 36,3 B***
S. tereti- 0.974 97 1.571 89 81,14 B*** P.a. - 33.17"** 52.82*** 30.38***
caudus D.m. 0.08 - 0.03 0.06
S.t. 75.73*** 64.95*** - 0.02
* P < 0 . 0 5 , *** P < 0 . 0 0 1 A.h. 39.71 *** 35.54*** 0.01 -

*** P<0.001

(round-tailed squirrel) and absence (Harris's antelope squirrel) G U D ' s are reported as the backtransformation of
of tail drags. Even when no or a few seeds were removed, the logarithmic means).
footprints provided p r o o f of encounter. In some cases the foot-
One-way A N O V A ' s were used to examine dif-
prints of two rodent species were visible. In this case the forag-
ing was attributed to both species. I attributed the giving up ferences between species in each microhabitat (Ta-
density of a tray to a single rodent species if its were the only ble 2). The G U D ' s of the two squirrel species did
detectable tracks even though it may not have been the only not differ in either the bush or open microhabitat.
species present in the tray that day or night. In the case of In the bush microhabitat, pocket mice had a lower
squirrels, forages ascribed to Harris's antelope squirrels may
also include activity by the other squirrel species.
G U D than either kangaroo rats or squirrels. In
The seeds recovered from seedtrays were cleansed of debris the bush microhabitat there was no significant dif-
and weighed to measure the giving up density. I recorded an ference between the G U D ' s of kangaroo rats and
encounter and rejection if the tray had footprints and if fewer the two squirrel species. In the open microhabitat
than 0.2 g of seeds were removed. The 0.2 threshold is approxi- both pocket mice and kangaroo rats had lower
mately the number of seeds which can be harvested by simply
collecting seeds on the surface without digging or intensive G U D ' s than squirrels. In the open microhabitat
searching. Once discovered by a forager, a seedtray was invari- there was no significant difference between the
ably visited on following nights or days. G U D ' s of pocket mice and kangaroo rats. In sum-
mary, independent of microhabitat pocket mice
had the lowest and squirrels had the highest
Results
GUD's.
The data are the giving up densities (grams of mil- A sign test comparing G U D ' s between trays
let; millet has a mean weight of 0.0066 g per seed) at a station was used to test for differences between
and they are specified by date, station, microhabi- microhabitats. The round-tailed and antelope
tat, and species. The results are summarized in Ta- ground squirrels, and the pocket mice had lower
ble 1 as the mean G U D ' s by rodent species and G U D ' s in the bush trays than in the open. Kan-
microhabitat. Mean G U D ' s are for trays foraged garoo rats had a lower G U D in the open microha-
by a single species. The following analyses were bitat (Table i).
performed: i) differences between species in For each species, a Spearman Rank test was
G U D ' s within each microhabitat, ii) differences in used to test for a correlation between G U D ' s in
G U D ' s between microhabitats for each species, iii) the open and bush trays at a station. All species;
correlation between bush and open G U D ' s at a squirrels (the two squirrel species were combined
station for each species, iv) propensity for each for this analysis), kangaroo rats, and pocket mice;
species to reject the opportunity to forage a tray, had a positive correlation between the G U D ' s of
and v) the effect of date and site on G U D ' s in pairs of trays (r~=0.576, P<0.001 ; r~=0.682, P <
each microhabitat. Because the distributions of 0.001; rs=0.619, P < 0 . 0 0 1 respectively). Because
G U D ' s are skewed to the right, analyses were per- of possible differences in G U D ' s between days, I
formed on logarithmically transformed data (mean ran this correlation separately for each day. For
 43

Table 3. The results of six random-block ANOVA's show the Discussion

effect of date subset of 9 18 Sept. and microhabitat (bush and
open) on GUD's (in grams, log transform) of each species (the
two squirrel species A. h./S. t., have been combined for this
One assumption of optimal foraging theory is the
analysis). The columns include F of improvement, degrees of simple axiom that an individual should engage in
freedom (dr), and mean sum of squares, MS, of the error term an activity until the benefits and the costs are
equal. Identifying the components and fitness
Species Date Station Error values of these costs and benefits presents a signifi-
F (df) F (df) MS (df)
cant challenge. The seemingly insurmountable
P. amplus 12.00"** (5) 2.50* (8) 0.400(40) complexity of ecological systems has led to the crit-
(Bush) icism that models of optimal foraging ignore other
P. amplus 21.76"** (3) 0.35 (5) 0.292 (15) relevant activities besides foraging, and ignore
(Open) other relevant inputs into fitness besides energy
D. merriami 0.48 (2) 1.37 (4) 0.268 (8)
(Bush) (but see Tilman 1982; M c N a m a r a and Houston
D. merriami 0.32 (4) 1.66 (5) 0.211 (20) 1986).
(Open) Here, I present a simple cost-benefit analysis
A. h./S. t. 10.26"** (4) 4.66*** (16) 0.222 (64) of foraging that incorporates any number of alter-
(Bush)
A. h./S. t. 3.40* (3) 5.32*** (12) 0.195 (48)
native activities and permits any number of inputs
(Open) into fitness. By considering the marginal rates of
substitution of different inputs into fitness, all in-
* P<O.05, ** P<O.01, *** P<O.O01 puts can be expressed as a single currency. Here,
I advocate using the forager's giving up density
in resource patches to gain insights into the ener-
getic cost, predation cost, and missed opportunity
each species, all seven days yielded a positive corre- cost of foraging. When harvest rate is an increasing
lation. At a station, the G U D in the bush tray function of patch resource density, G U D is an esti-
is a good predictor of the G U D in the open mate of the harvest rate which balances the costs
tray. of foraging. Laboratory or field measurements of
A G-test of heterogeneity was used to examine patch harvest rates permits conversion of G U D ' s
differences between pocket mice and kangaroo rats into harvest rates.
in their propensity to reject the opportunity to for- Without sacrificing environmental complexity,
age in one of the trays of a pair. A double forage it is possible to gain insights into species interac-
occurs when both trays at a station are foraged tions, habitat preferences, patch use versus rejec-
by the species and a single forage occurs when tion, and species-specific foraging differences. The
one tray is foraged but the other tray is rejected. data collected for desert rodents provide an exam-
Kangaroo rats (46 double and 19 single forages) ple. In the experiment, except for microhabitat,
were more likely than pocket mice (85 double and trays at a station were identical and the same indi-
3 single forages) to forage at only one tray of a vidual had the opportunity to forage in both trays
pair (Gh----20.79 with William's correction, P < (travel time between trays was on the order of sec-
0.001). Of the 19 trays in which kangaroo rats re- onds). Thus, at a station the cost of foraging, C,
jected the opportunity to forage, 18 were in the and the missed opportunity cost, MOC, should
bush microhabitat. have been the same for both trays. Any differences
Two-way random-block ANOVA's were used in G U D ' s between the trays were therefore the re-
to test for the effect of station (space) and date sult of differences between microhabitats in preda-
(time) on G U D ' s in each microhabitat. To permit tion risk; kangaroo rats perceived greater preda-
consideration of each species and microhabitat, six tion risk in the bush microhabitat; squirrels and
separate analyses were performed (Table 3). For pocket mice perceived greater predation risk in the
each species, as many dates and stations as possible open.
were used while still maintaining a value in each While many models of habitat selection predict
cell. Both station and date affected the G U D ' s of the acceptence and rejection of resource patches
squirrels and pocket mice in the bush microhabitat. (Fretwell and Lucas 1970; Rosenzweig 1974,
In the open microhabitat, station and date affected 1979), tests of optimal habitat selection have mea-
the squirrels' GUD's, but only date affected the sured the proportional use of patch types (Heinrich
pocket-mice's GUD's. Neither station nor date af- 1979; Lewis 1980; Mittlebach 1981; Baharav and
fected the kangaroo-rats' G U D ' s in either micro- Rosenzweig 1985; Pimm et al. 1985). The experi-
habitat. ment of this paper demonstrates that GUD's pro-
44

vide information both on the proportional use of in relatively long term differences in foraging costs
patch types and on the acceptance and rejection between stations. Similarly, on a daily and nightly
of resource patches. It was possible to show that basis, there should be temporal variation in climate
kangaroo rats were much more likely to reject the and predation risk. Thus, in both microhabitats
opportunity to forage a tray than were pocket there should frequently be consistent differences
mice. The approach is suitable for testing models in G U D ' s between stations (spatially) and between
of habitat selection which make predictions con- days or nights (temporally). For squirrels and
cerning both patch acceptance and use (Brown and pocket mice (but not for kangaroo rats), the results
Rosenzweig 1986). indicate significant effects of space and time on
The ability to distinguish between the foraging GUD's.
bouts of the different species based on footprints
makes the technique suitable for investigating dif-
Problems with the approach
ferences among species. GUD's, besides being a
measure of harvest rate, are also a measure of for- The use of manipulated resource patches presents
aging efficiency (Tilman 1982; Vance 1985; Brown at least four problems: i) the patches are not natu-
1988). The results showed that pocket mice were ral, ii) the resource may not be appropriate, iii)
the most efficient foragers in both microhabitats. the foragers may become satiated, and iv) the trays
If the conditions pertaining to September were may be visited by more than one forager. The first
maintained throughout the year and if foraging two problems concern the realism of the foraging
efficiency on seeds was the sole determinant of situation and the last two concern possible distor-
competitive ability then pocket mice would out- tions of the results by the foragers.
compete both the squirrels and the kangaroo rats. How accurately the artificial patches mimic
natural resource patches depends upon the ques-
tions under investigation. The type of resource and
Is the forager's behavior consistent the substrate used in the patch will influence both
with fitness maximization ? the species that participate in the experiment and
Up to this point in the discussion it has been as- their GUD's. If the questions of interest concern
sumed that the foragers are behaving optimally differences between microhabitats or time periods
and that the G U D is a truthful revelation of the in predation risk or missed opportunity costs then
forager's perceived costs and benefits. Insofar as the realism of the substrate and resource are rela-
it is possible to control for the various costs of tively unimportant. However, when the significant
foraging it is possible to test directly whether the parameters involved in regulating habitat selection,
forager's behavior is consistent with fitness maxi- predation risk, and interspecific interactions in-
mization. Increasing the energetic cost, predation clude substrate and resource type then the choice
cost, and missed opportunity cost of foraging of experimental resource patch is important and
should result in higher GUD's. While these would will influence the results.
be worthwhile experiments, they were not per- The presence of artificial resource patches may
formed in the study reported here. However, infer- distort the normal behavior of foragers. If foragers
ences regarding foraging consistency with fitness devote time to the artificial patches, they will have
maximization can still be drawn from the data. less time for alternative activities and they may
At a station, independent of other differences, harvest more resources. The marginal fitness of
the missed opportunity cost should be the same alternative activities should increase as less time
for both the open and bush tray. As a result of is spent on alternative activities, and the marginal
individual differences (the footprints of a single fitness of energy should decrease as more resources
individual were rarely found at more than one seed are harvested. Together, these two effects should
tray station) and spatial variability between sta- increase the missed opportunity cost of foraging
tions in natural resource abundances and preda- above what it would be in the absence of artificial
tion risk, missed opportunity costs and the mar- patches. Any distortion is a concern for investiga-
ginal valuation of energy should vary between sta- tions where the G U D must reflect the animals,
tions. Thus, the GUD's in the open and bush mi- normal missed opportunity cost of foraging. For-
crohabitats at a station should be positively corre- tunately, it is possible to measure this distortion.
lated. The results indicate that all of the rodent The distortion in missed opportunity cost is
species exhibit a positive correlation (see Results). the result of the foragers taking time from alterna-
Spatial variation in natural resources, preda- tive activities to forage in the artificial patches.
tion risk, and edaphic characteristics should result As the initial density of resources in a patch in-
 45

creases, the time required to forage the tray to forager harvests resources. In all respects, m a n i p u -
the same giving up density also increases. Thus, lated resource patches with minimal or no search
increasing the a m o u n t o f resources in artificial time are less sensitive measures o f b e h a v i o r than
trays increases the u p w a r d distortion o f missed op- patches where the harvest rate is a decreasing func-
p o r t u n i t y costs. The m a g n i t u d e o f the distortion tion o f p a t c h resource density.
can be examined by measuring the change in G U D
that results f r o m increasing the resources available Acknowledgements. This work is in partial fulfillment of the
to a forager or g r o u p o f foragers. I f there is a requirements for a doctorate in Ecology and Evolutionary Biol-
ogy at the University of Arizona. The manuscript and the ideas
range o f initial resource availabilities that does n o t herein benefited greatly from comments by and discussions with
influence the G U D then over this range o f experi- Burr Kotler, William Mitchell, Jean Powlesland, Sandra Rode,
mental conditions the seed trays are n o t signifi- Thomas Valone, and three anonymous reviewers. I thank my
cantly distorting the missed o p p o r t u n i t y cost. Mit- committee, James Brown, James Cox, Robert Holt, Thomas
chell and B r o w n (ms.) showed that for up to 12 g Vincent, and David Vleck for their input and guidance. Special
thanks to my advisor Michael Rosenzweig. I am grateful to
o f millet per station, the G U D ' s o f k a n g a r o o rats Hubert Markl for his scientific and editorial comments. The
( D . m e r r i a m i ) remained constant. The experiment Graduate Program Improvement Fund from the Graduate Col-
reported here used only 6 g o f millet per station lege of the University of Arizona and the Mitrani Center for
a n d no distortion o f missed o p p o r t u n i t y costs Desert Ecology (Blaustein Institute, Ben-Gurion University,
Sede Boqer, Israel) provided financial assistance. This is publi-
should have occurred. cation #63 OFTHEM1TRANICENTER.
I f several individuals o f the same species forage
in the p a t c h then the researcher only measures the
G U D o f the last forager. Thus, m a n y foragers per Appendix
p a t c h m a y lead to a biased sampling o f the species' Consider a forager who alots a certain amount of time, T,
G U D . Individuals with low G U D ' s will be over- between foraging in one of two patches (tl and tz are times
represented in the data. spent foraging in each patch) or remaining dormant in its nest
(ta is time spent dormant). While foraging in patch 1 or 2 the
individual incurs a constant per-unit-time risk of predation
Comparison to other techniques (rl or r2), harvests energy, and expends a constant rate of ener-
gy (c~ or c2). While dormant the individual incurs negligible
Live-trapping is a frequently used m e t h o d for in- risk of predation, expends energy (ca), and acquires 'mainte-
vestigating habitat selection, p r e d a t i o n risk, a n d nance' (m). Maintenance can be thought of as an input into
competition. Because live-trapping is necessarily fitness resulting from the benefits of grooming, burrow mainte-
nance, or offspring care. I assume that the energetic cost of
an all or n o t h i n g m e a s u r e m e n t it is a less sensitive dormancy is less than the energetic cost of foraging; ca < cl, c2.
measure o f behavior t h a n G U D ' s . Live-trapping For many organisms, dormancy provides a valuable alternative
distorts the animal's o r d i n a r y activities; once activity either when the organism has a lot to lose from being
trapped, it is p r e v e n t e d f r o m engaging in other ac- preyed upon or when resource abundances and harvest rates
are low.
tivities. O n the o t h e r hand, G U D ' s in artificial
Let the fitness function be given by:
patches need not distort o r d i n a r y activity, at any
time the forager is free to engage in other activities. G(t~,t2,ta)=p(tl,t2) {[1 -- E X P ( - - m ( t d ) )][1 + fle(t~,tz,ta)]} ~
O t h e r researchers have used artificial resource where 0 < ~ < 1, and fl > 0 are constants (henceforth, when con-
patches to gain insights into m a m m a l i a n foraging venient, probability of surviving predation (p), maintenance
(m), and energy gain (e) will be written without their argu-
b e h a v i o r ( B r o w n 1971; M a r e s and Rosenzweig ments). The fitness function assumes that there are diminishing
1978; Frye a n d R o s e n z w e i g 1980; A b r a m s k y 1983; but positive returns to fitness from energy gain (c~<1), and
Vickery 1984). H o w e v e r , their previous use has n o t that the fitness value of energy is weighted by an efficiency
included e m b e d d i n g the resource in a substrate term, 1 - E X P ( - m ) , whose value increases with maintenance
performed by the individual.
[but see Schneider (1984) for G U D ' s in artificial
The population dynamics of individuals using a particular
patches a n d see W h i t h a m (1977), a n d H o d g e s and strategy of time allocation is: N(7) = G(-) N(0) (G(-) > 1 or
W o l f (1981) for G U D ' s in u n m a n i p u l a t e d patches]. G(-)<I determines whether population size is increasing or
W h e n the resources are free standing there is little decreasing). To insure that G(.) is greater than zero assume
or no search time a n d so harvest rate does n o t that G(.) is positive even if the individual harvests no resources;
i.e. 1-flTci>O for i=1,2.
decline with resource density. I f the missed o p p o r -
Let probability of surviving predation, maintenance, and
tunity cost o f foraging remains undistorted, the energy gain be:
C U D in the p a t c h is zero a n d p a t c h use should
p - EXP[--rltl-rzt2] (A.la)
be all or n o t h i n g (the entire p a t c h is treated as
a single resource). I f there is a giving up point m=yta (A.lb)
in the p a t c h it is because the missed o p p o r t u n i t y e = Hi(t1) + H2(t2) -- c1t I -- c2t 2 -- catd (A.lc)
cost o f foraging is being distorted u p w a r d s as the where Hi(h) is the cumulative harvest rate in patch i= 1,2.
46

To solve for the optimal patch quitting harvest rate, first Frye RJ, Rosenzweig ML (1980) Clump size selection: a field
calculate the MRSev and MRSem using expressions (5) and (6) : test with two species of Dipodomys. Oecologia (Berlin)
47:323 327
MRSe, : (1 +fle)/~flp
Grubb TC, Greenwald L (1982) Sparrows and a brushpile -
[1 + f l e ] E X P ( - m ) foraging responses to different combinations of predation
MRSem = fl[1 - E X P ( - m)] risk and energy cost. Anita Behav 30:637-640
Hartling LK, Plowright RC (1979) Foraging by bumblebees
(If tl*, t2*, td* >0, the optimal values for time spent in patch on patches of artificial flowers: a laboratory study. Can
1, tl*, and 2, t2*, must satisfy condition (7).) Take the deriva- J Zool 57:1866-1870
tives of (A.la-c) with respect to tl or td and substitute these Heinrich B (1979) Foraging strategies of caterpillars: Leaf dam-
and the MRS's into (7) to yield: age and possible predator avoidance. Oecologia (Berlin)
aHi(.) r~[l+fle] 7[l+fle]EXP(-m) 42:325-337
at, - c ~ + ~ + fl[1--EXP(-m)] ed (A.2) Hodges CM (1981) Optimal foraging in bumblebees-hunting
by expectation. Anim Behav 29:1166-1171
where i = 1,2. The optimal allocation of time can be obtained Hodges CM, Wolf LL (1981) Optimal foraging in bumblebees:
by solving three simultaneous equations. The first two gener- Why is nectar left behind in flowers? Behav Ecol Sociobiol
ated by (A.2) and the third generated by the constraint: tl + 9:41-44
t2 + ta = T. Holt RD, Kotler BP (1987) Short-term apparent competition.
The term on the left is the quitting harvest rate, H. The Am Nat 130:412-430
first term on the right is the energetic cost of foraging in the Hubbard SF, Cook RM (1978) Optimal foraging by parasitoid
patch, C. The second term is the cost of predation, P. And, wasps. J Anim Ecol 47 : 593-604
the iast two terms are the missed opportunity costs, MOC. Krebs JR, Ryan JC, Charnov EL (1974) Hunting by expecta-
The first component of MOC is the foregone benefit of addi- tion or optimal foraging? A study of patch use by chicka-
tional maintenance, and the second is the foregone cost of dor- dees. Anim Behav 22:953-964
mancy. Krebs JR, Stephens DW, Sutherland WJ (1983) Perspectives
in optimal foraging theory. In: Clark GA, Bush AH (eds)
Perspectives in ornithology. Cambridge Univ Press, New
References York
Abramsky Z (1983) Experiments on seed predation by rodents Lewis A R (1980) Patch use by grey squirrels and optimal forag-
and ants in the Israeli desert. Oecologia (Berlin) 57:328 332 ing. Ecology 61:1371-1379
Baharav D, Rosenzweig ML (1985) Optimal foraging in Dorcas Lima SL, Valone TJ, Caraco T (1985) Foraging-efficiency-pre-
gazelles. J Arid Environ 9:167-171 dation - risk trade-offs in the grey squirrel. Anita Behav
Belovsky G (1978) Diet optimization of a generalist herbivore, 33:155-165
the moose. Theor Pop Biol 14:105 134 MacArthur R, Pianka E (1966) On optimal use of a patchy
Brown JH ( 1971) Mechanisms of competitive exclusion between environment. Am Nat 100:603-609
two species of chipmunk. Ecology 52:305-311 Mares MA, Rosenzweig ML (1978) Granivory in North and
Brown JH, Kodric-Brown A, Whitham TG, Bond H W (1981) South American desert rodents. Ecology 49:235 241
Competition between hummingbirds and insects for the nec- McNamara JM, Houston AI (1986) The common currency for
tar of two species of shrubs. Southwest Nat 26:133-145 behavioral decisions. Am Nat 127:358-378
Brown JS (1986) Coexistence on a resource whose abundance Milinski M (1979) Evolutionarily stable feeding strategies in
varies: a test with desert rodents. Unpubl PhD Diss, Univ sticklebacks. Z Tierpsychol 51:36-40
Arizona, Tucson Milinski M, Heller R (1978) Influence of a predator on the
Brown JS (1988) The role of resource variability in structuring optimal foraging behaviour of sticklebacks (Gasterosteus
desert rodent communities. In: Morris D, Abramsky Z, Fox aculeatus L.). Nature 275:642-644
B (eds) Patterns in the structure of mammalian communi- Mittlebach G G (1981) Foraging efficiency and body size: a
ties. Texas Tech Univ Press, Lubbock (in press) study of optimal diet and habitat use by bluegills. Ecology
Brown JS, Rosenzweig ML (1986) Habitat selection in slowly 62:1370-1386
regenerating environments. J Theor Biol 123:151-171 Pimm SL, Rosenzweig ML (1981) Competitors and habitat use.
Caraco T (1979) Time budgeting and group size: a theory. Oikos 37 : 1-6
Ecology 60 : 611-617 Pimm SL, Rosenzweig ML, Mitchell W (1985) Competition
Charnov EL (1976) Optimal foraging, the marginal value theo- and food selection: field tests of a theory. Ecology
rem. Theor Pop Biol 9:129 136 66: 798-807
Cheverton J, Kacelnik A, Krebs JR (1985) Optimal foraging: Pyke GH (1978) Optimal foraging in hummingbirds: Testing
constraints and currencies. In: H611dobler B, Lindauer M the marginal value theorem. Am Zool 18:739-752
(eds) Experimental behavioral ecology. Sinauer, Sunder- Pyke GH (1980) Optimal foraging in bumblebees: Calculation
land, Mass of net rate of energy intake and optimal patch use. Theor
Chiang AC (1974) Fundamental methods of mathematical eco- Pop Biol 17:232-246
nomics, 2nd ed. McGraw-Hill, New York Pyke G H (1984) Optimal foraging theory: A critical review.
Cowie RJ (1977) Optimal foraging in the great tits (Parus ma- Ann Rev Ecol Syst 15:523-575
jor). Nature 268 : 137-139 Rosenzweig ML (1974) On the evolution of habitat selection.
Emlen JM (1966) The role of time and energy in food prefer- Proceedings of the ]st International Congress of Ecology,
ence. Am Nat 100:611-617 pp 401-404
Feinsinger P (1976) Organization of a tropical guild of nectari- Rosenzweig ML (1979) Optimal habitat selection in two-species
vorous birds. Ecol Monogr 46:257-291 competitive systems. Fortschr Zool 25:283-293
Fretwell SD, Lucas HL Jr (1970) On territorial behavior and Rosenzweig ML (1981) A theory of habitat selection. Ecology
other factors influencing habitat distribution in birds. I. the- 62:327-335
oretical development. Acta Biotheor 19:16-36 Rosenzweig ML (1985) Some theoretical aspects of habitat s e -
 47

lection. In: Cody ML (ed) Habitat selection in birds. Aca- Tilman D (1985) The resource-ratio hypothesis of plant succes-
demic Press, New York, pp 517 540 sion. Am Nat 125:827-852
Russell RR, Wilkinson M (1979) Microeconomics. A synthesis Townsend CR, Hildrew A G (1980) Foraging in a patchy envi-
of modern and neoclassical theory. Wiley, New York ronment by a predatory net-spinning caddis larva - a test
Schneider KJ (1984) Dominance, predation, and optimal forag- of optimal foraging theory. Oecologia (Berlin) 47:219-
ing in whitethroated sparrow flocks. Ecology 65:182(~1827 221
Sih A (1980) Optimal behavior: can foragers balance two con- Vance R R (1985) The stable coexistence of two competitors
flicting demands ? Science 210 : 1041-1043 for one resource. Am Nat 126:72-86
Schaffer WM, Jensen DB, Hobbs DE, Gurevitch J, Todd JR, Vickery WL (1984) Optimal diet models and rodent food con-
Schaffer MV (1979) Competition, foraging energetics, and sumption. Anita Behav 32:340-348
the cost of sociality in three species of bees. Ecology Werner EE, Gilliam JF, Hall DJ, Mittlebach G G (1983) An
60: 97(~987 experimental test of the effects of predation risk on habitat
Sokal RR, Rohlf FJ (1981) Biometry, 2nd ed. Freeman, San use in fish. Ecology 64:1540-I 548
Francisco, Calif Whitham TG (1977) Coevolution of foraging in Bombus and
Tilman D (1982) Resource competition and community struc- nectar dispensing in Chilopsis: A last dreg theory. Science
ture. Princeton Univ Press, Princeton 197:593-596