Ecology, 91(1), 2010, pp. 233–241� 2010 by the Ecological Society of America

Foraging behavior of an urban bird species:molt gaps, distance to shelter, and predation risk

IDO TSURIM,1,3 BURT P. KOTLER,2 AMIR GILAD,1 SHIRA ELAZARY,1 AND ZVIKA ABRAMSKY1

1Department of Life Sciences, Ben-Gurion University of the Negev, P.O. Box 653, Beer-Sheva 84105 Israel2Jacob Blaustein Institute for Desert Research, Mitrani Department of Desert Ecology, Ben-Gurion University of the Negev,

Sede Boqer Campus, Midreshet Ben-Gurion 84990 Israel

Abstract. Predation cost (Pc) is often regarded as a pivotal component determiningforaging behavior. We hypothesized that variations in two of its major constituents, predationrisk (l) and the marginal value of energy (]Fs/]e, where Fs is the survivor’s fitness and erepresents the amount of acquired energy), will translate into variations in patch use behaviorof ground-foraging birds. We studied patch use behavior of House Sparrows (Passerdomesticus), as affected by the proximity to shelter, in two large outdoor aviaries. Proximity toshelter should affect l. We manipulated the birds’ flight performance by clipping primaryflight feathers from their wings to increase l, but the clipping may also increase ]Fs/]e. To helpdistinguish between the birds’ response to these confounding effects, we further augmentedfood in the aviaries to reduce ]Fs/]e.

Patch use, as measured by giving-up densities (GUD, the amount of food left behind in aresource patch following exploitation) was affected by distance from shelter only slightly andmainly when the birds were feather-clipped and food was not augmented. Food augmentationhad a homogenizing effect on foraging costs by increasing GUDs and washing out the effectsof distance and feather clipping.

We argue that l increases with distance from shelter but that, for the highly urban HouseSparrow, this increase is only slight. Feather clipping then increased l further to the point atwhich patch use discernibly decreased with distance from shelter. Our experimentalmanipulation of feather clipping also acted to increase ]Fs/]e and resulted in an overalllowering of GUDs. The seed augmentation counteracted the effect of feather clipping on]Fs/]e, allowing the birds to reduce their foraging efforts and washing out the qualitative effectof l with respect to distance from shelter.

Key words: Beer-Sheva, Israel; cost of missed opportunities (MOC); feather molt; giving-up densities(GUD); House Sparrow; marginal value of energy; optimal foraging; Passer domesticus; predation cost;predation risk.

INTRODUCTION

Coping with predation risk is an important compo-

nent of foraging behavior (McNamara and Houston

1987, Brown 1988, Kotler and Brown 1988, Lima and

Dill 1990, Lima 1998, Brown and Kotler 2004). In

ground foragers, and particularly in birds, successful

predatory attacks usually occur while the victim is

feeding away from shelter (Kenward 1978, Lima 1993,

Witter et al. 1994). The safety of shelter reduces the risk

of being preyed upon. Hence, predation risk associated

with foraging may vary with the proximity of shelter,

creating a gradient of predation risk from shelters (Lima

and Dill 1990, Todd and Cowie 1990, Watts 1990,

Brown et al. 1992b, Kotler 1997, Oyugi and Brown 2003,

Tsurim 2005). It follows that the speed with which a

forager can respond to the appearance of risk and reach

shelter may determine the level of predation risk it faces

in a certain foraging environment and the time it can

afford to forage there. The proximity of shelter and the

bird’s takeoff and flight abilities should thus play

important roles in determining the level of predation

risk a forager experiences. The loss of flight feathers (as

occurs during feather molting) increases a bird’s wing

loading and should thus have a negative effect on

takeoff speed and aerobatic performance of the bird

(Swaddle and Witter 1997, Hedenstrom and Sunada

1999, Swaddle et al. 1999). This may impair the bird’s

ability to escape predatory attacks and consequently

increase the level of predation risk experienced by the

bird (e.g., Hedenstrom and Rosen 2001).

Optimal patch use theory predicts that a forager

should cease foraging in a food patch when food harvest

rate in the patch (H ) just balances the sum of costs from

the various inputs to fitness, converted to the common

currency of energy (Brown 1988). The cost of foraging is

generally divided into three components: (1) The energy

costs of additional thermoregulation away from the

roost and searching for and handling food (C ). (2) The

Manuscript received 2 October 2008; revised 20 April 2009;accepted 23 April 2009. Corresponding Editor: T. J. Valone.

3 E-mail: tsurim@bgu.ac.il

233

costs incurred from the risk of predation (Pc). (3) The

cost of missed opportunities (MOC). Hence the forager

is expected to quit foraging when H ¼ C þ Pc þMOC.

Brown (1988, 1992) proposed that the cost of predation

is comprised of the product of two components: risk of

predation (l) and the marginal rate of substitution of

energy and survivorship (Fs/(]Fs/]e), where Fs represents

the survivor’s fitness and e represents the amount of

energy acquired, i.e., the fitness of those that survive the

risk of predation long enough to reproduce); ]Fs/]e is

the marginal value of energy. Hence, Pc ¼ l 3

Fs/(]Fs/]e). Interestingly, on short temporal scales,

during which Fs can change but slowly, what matters

most is the ratio between l and ]Fs/]e.

Previously (Tsurim 2005), the relationship between

predation risk, proximity to shelter, wing molt gaps, and

patch use behavior of House and Spanish Sparrows

(Passer domesticus and P. hispaniolensis) was investi-

gated. We reduced wing surface area to manipulate l by

clipping four out of 15 flight feathers, simulating natural

wing molt gaps. Birds subjected to this treatment were

expected to experience increased levels of l. Both species

utilized seed patches situated farther from shelter less

thoroughly, with Spanish Sparrows showing a steeper

response to distance from shelter than House Sparrows.

We argued that the decrease in patch utilization results

from increased foraging costs due to predation risk with

increased distance from shelter. We also suggested that

the differences between the species in the intensity of

this response may be attributed to the different

evolutionary and ecological history of the two species:

House Sparrows live in urban environments and

consequently experience more intense food competition

and have higher marginal value of food (]Fs/]e) and

thus lower overall Pc than Spanish Sparrows (Brown

1992, Shochat 2004, Shochat et al. 2004). Surprisingly,

however, reduction of wing surface area had no

detectable effect on seed patch utilization of either

species. We suggested that the wing manipulation may

have not been extreme enough and that the birds may

have been able to compensate for the wing surface

reduction or may not have been affected strongly

enough to be detected by our measuring technique.

Similarly, perhaps flight performance does not affect

risk of predation in resource patches as envisaged.

However, an alternative explanation is that the wing

manipulation did affect flight performance as intended,

with the feather clipping increasing l, but at the same

time also increasing the birds’ overall energy expendi-

ture and thus ]Fs/]e. If so, then each would have

affected the amount of food left behind in resource

patches, but in opposite directions.

The goal of this study was to investigate the roles of land ]Fs/]e in determining patch use behavior of ground-

foraging birds, exemplified by House Sparrows, with

respect to distance from shelter. We used seed trays to

measure foraging efficiency of House Sparrows with

respect to distance from shelter in large outdoor aviaries.

In order to manipulate l, we clipped flight feathers to

reduce wing surface area; in order to manipulate ]Fs/]e,

we augmented the seeds available to sparrows. If the

feather clipping also increases the birds’ energy expen-

diture by making it more energetically expensive to fly,

then when food is limited, as is the case in our

experiment, ]Fs/]e should increase. Hence, birds with

clipped feathers should experience a lower ratio

of survivor’s fitness to marginal value of energy,

Fs/(]Fs/]e), than birds with unclipped feathers. This

ratio helps comprise the foraging cost of predation (Pc),

so birds should thus experience a reduction in Pc even if

l has not changed or has even increased. Additionally,

feather clipping may help reduce the birds’ long-term

prospects, i.e., reduce the survivor’s fitness (Fs), by

reducing the ability to accumulate energy for reproduc-

tion. This would further contribute to a reduction in Pc.

Nonetheless, Fs is slow to change, and we believe that in

the context of our short-term experiments this effect is

negligible. Decreasing ]Fs/]e, e.g., by augmenting food

levels, should thus tease apart the effects of feather

clipping caused by the change in l from those caused by

the change in ]Fs/]e.

To quantify sparrow patch use and foraging behavior,

we used the giving-up density method (GUD; Brown

1988). The GUD is the density of food resources in the

patch that corresponds to the quitting harvest rate (H )

in the patch and reflects the individual’s overall foraging

costs in the patch (Brown 1988). The GUD is expected

to rise with an overall increase in the costs of foraging.

Under controlled experimental conditions, the GUD can

be used to assess the cost of predation (Kotler et al.

1994).

We predicted that: (1) Seed patch utilization will

decrease, and thus GUD increase, with increased

distance from shelter due to higher l farther from

shelter (Fig. 1). (2) Clipping flight feathers will hamper

flight performance and hence increase both l and the

energy requirements of the birds. Greater energy

requirements should mean greater marginal value of

energy for the forager, i.e., higher ]Fs/]e. The increase in

l and ]Fs/]e may counteract one another, resulting in

little or no net change in Pc, and thus in seed patch

utilization, i.e., no change in GUDs (Fig. 1). (3) Seed

augmentation will decrease ]Fs/]e, thus increasing Pc

and amplifying the role of l. Consequently, overall

GUDs are expected to be higher and the slope of the

relationship between distance from shelter and GUD is

expected to be steeper when food is augmented. Feather

clipping, through its effect on l, will interact with the

seed augmentation to further increase GUDs and

steepen the distance effect (Fig. 1). A statistical

interaction between feather clipping and seed augmen-

tation would support the suggestion that past results are

explicable through the effect of feather clipping on both

l and ]Fs/]e; no interaction would support the

suggestion that l is not affected by feather clipping.

IDO TSURIM ET AL.234 Ecology, Vol. 91, No. 1

METHODS

We used the GUD method to tease apart the roles of

predation risk and the marginal value of energy as partsof the cost of predation in determining patch use

behavior of House Sparrows. We measured patch usebehavior of House Sparrows in two large outdooraviaries by simultaneously placing seed trays as foraging

patches at various distances from shelter. Under theexperimental settings, differences in GUDs among seedtrays reflect differences in the costs of foraging resulting

from variations in l (Brown 1988, Kotler et al. 1994).We attempted to manipulate l by varying the proximity

of foraging patches (seed trays) to shelter and byclipping flight feathers to reduce wing surface area andthereby reduce the flight performance of the birds. The

feather clipping was intended to simulate wing moltgaps. We manipulated ]Fs/]e by augmenting seed food,

thus reducing ]Fs/]e of food in the aviaries.We conducted the experiments in two large outdoor

aviaries at the Zoological Gardens, Beer-Sheva, Israel

(348440 E, 318150 N) during October and November2007. The aviaries were 9 3 9 3 4 m cages, constructedfrom metal frames covered with 40% shade cloth (1-cm

mesh). The seed trays were 40 3 55 cm sheets of denseartificial grass, 2.5 cm thick, upon which we scattered

unhusked millet seeds (see Plate 1). The seeds settled onand among the blades of the artificial grass to create arange of accessibilities and encounter rates to foragers

exploiting the resource patch. We placed a large brushpile approximately 1 m tall and 1 m across at the base inone corner of each aviary to provide perching opportu-

nities and shelter for the birds. Seed trays were placed ina staggered array at 1-m intervals from the margin of the

shelter out to a distance of 4 m, always keeping thedistance between the seed trays and the aviary wallslonger than the distance to the shelter. From time to

time, foraging on the seed trays was interrupted for afew seconds to several minutes when the birds fled into

the shrubs, presumably for refuge. Occasionally, birds

flew to and perched on one of the aviary walls. The birds

would then resume foraging on the seed trays until the

next interruption.

We manipulated wing surface area of individual birds

by clipping the three inner primary and the three

outermost secondary feathers (Fig. 2) from both wings

FIG. 1. A graphical illustration of our predictions for the change in patch use (giving-up density, GUD; y-axis) by HouseSparrows (Passer domesticus) as a function of distance from shelter (x-axis) and feather clipping. With intact wings (solid line),GUDs are predicted to increase with distance from shelter and even more so when food is augmented. Food augmentation will alsoresult in overall higher GUDs, as indicated by the higher intercept of the response line with the GUD axis. Feather-clipped birds(broken line) are predicted to have marginal net change in the response of GUDs to distance from shelter, when food is notaugmented, while a steeper response is predicted when food is augmented. We conducted the experiments in two large outdooraviaries at the Zoological Gardens, Beer-Sheva, Israel.

FIG. 2. We reduced the wing surface area of HouseSparrows by clipping the three innermost primary and thethree outermost secondary feathers. Panel A illustrates theintact wing; panel B illustrates the wing with clipped flightfeathers.

January 2010 235PREDATION RISK AND MOLT GAPS IN BIRDS

(see Swaddle and Witter 1997, Swaddle et al. 1999, Lind

2001), thus simulating wing molt gaps. Feathers were cut

near their bases rather than plucked to avoid stimulating

immediate regrowth of the feathers and the associated

costs (e.g., Lindstrom et al. 1993). In so doing, we

reduced the wing surface area and increased the wing

load. This presumably impaired takeoff ability and

maneuverability and hence the ability of the birds to

contend with predator attacks (Swaddle and Witter

1997, Swaddle et al. 1999, Hedenstrom and Rosen 2001,

Lind 2001, Senar et al. 2002, Williams and Swaddle

2003).

We manipulated ]Fs/]e by augmenting the aviaries

with seeds that required minimum time, effort, and

energy to harvest (i.e., minimum costs). This was done

by placing a plate, containing 2 g/bird of unhusked

millet seeds, on the ground just below the shelter, in

addition to the seeds in the seed trays.

Protocol

(1) After capturing wild birds from within a radius of

300 m of the aviaries, we immediately introduced them

to the two aviaries in groups of five to six individuals

each for a 2-d acclimation period. During this acclima-

tion period, the aviaries were set exactly as they would

be during the experiments, only with seeds supplied ad

libitum on the seed trays. (2) We measured the GUDs in

the trays during two consecutive days following the

acclimation period. We placed 10 g of millet seeds in

each seed tray. The seeds were placed in the trays 2 h

after sunrise. Following patch exploitation by the birds

(see Plate 1), the remaining seeds were retrieved 7 h later,

;2 h before sunset. We then placed in the aviary a plate

with a free supplement of seeds and left it there until the

beginning of the next day’s session. Drinking water was

available at all times. (3) We repeated step 2, with the

addition of the seed augmentation during the day. (4) At

the end of the fourth day of data collection, we clipped

the six flight feathers from each wing of each individual

and placed the birds back into the aviary. (5) We gave

the birds 1 d for acclimation. (6) We then repeated steps

2 and 3. (7) At the end of step 6, the birds were released

at their capture locations.

Ideally, each session (replicate) lasted 11 d. However,

two out of the seven sessions took 13 d because we could

not conduct measurements during rainy or very windy

days. On the few occasions of a bird’s death or escape

(total of five birds during three different sessions), we

replaced it with another bird.

To ensure independence of observations, we averaged

the GUD data over the two measuring days of each

treatment at each distance in each replicate (session). We

used repeated-measures ANOVA to analyze the data.

For this analysis, we used feather clipping, food

augmentation, and distance from shelter as three levels

of within-subject factors (hereafter, ‘‘wing,’’ ‘‘augmen-

tation,’’ and ‘‘distance,’’ respectively).

RESULTS

As predicted, augmentation overall had a significantly

positive effect on GUDs (Fig. 3, Table 1), indicating that

GUDs increased with a decrease in ]Fs/]e. Wing did not

have an overall effect on seed patch utilization (Table 1).

While we expected an overall positive effect of distance

PLATE 1. House Sparrows exploiting millet seeds from a seed tray made of artificial turf (‘‘Astro-Turf’’), used for measuringpatch-use behavior under varying predation risks. Photo credit: I. Tsurim.

IDO TSURIM ET AL.236 Ecology, Vol. 91, No. 1

on GUDs, it had only a marginally significant effect

(Fig. 3, Table 1). However, the interactions augmenta-

tion 3 distance and wing 3 distance were statistically

significant or nearly so (Table 1). These indicate that

GUDs indeed varied with distance from shelter, but that

this variation is complex and may possibly be affected

by the flight ability of the birds and the marginal value

of food. Accordingly, the relationship between GUDs

and distance from shelter was more prominent when the

birds were feather clipped and when seeds were not

augmented (Fig. 3, Table 1). The significant interaction

(Table 1) augmentation 3 wing further supports this

conclusion, indicating that the effect of feather clipping

on GUDs differ under different augmentation treat-

ments. Notably, seed augmentation washed out the

effect of feather clipping on the variation in GUDs with

FIG. 3. The effect of distance from shelter, food augmentation, and flight feather clipping (simulated wing molt gaps) on giving-up densities (GUDs; mean 6 SE) of House Sparrows. Table 1 shows results of the repeated-measures ANOVA, with distance fromshelter, food augmentation, and feather clipping as three levels of within-subject factors.

TABLE 1. Repeated-measures ANOVA of the effects of distance from shelter, food augmentation,and flight feather clipping (simulated wing molt gaps) on giving-up densities (GUDs) of HouseSparrows (Passer domesticus), as three levels of within-subject factors.

Effect SS df MS F P

Intercept 4104.3 1 4104.3 1748.8 ,0.001Error 14.1 6 2.3

Wing 2.6 1 2.6 2.3 0.18Error 7.0 6 1.2

Augmentation 27.1 1 27.1 36.1 ,0.001Error 4.5 6 0.8

Distance 4.8 4 1.2 2.3 0.09Error 12.7 24 0.5

Wing 3 augmentation 5.1 1 5.1 7.2 0.036Error 4.2 6 0.7

Wing 3 distance 6.3 4 1.6 2.4 0.08Error 15.8 24 0.7

Augmentation 3 distance 4.8 4 1.2 3.6 0.02Error 8.1 24 0.3

Wing 3 augmentation 3 distance 2.8 4 0.7 1.2 0.33Error 13.8 24 0.6

Note: We conducted the experiments in two large outdoor aviaries at the Zoological Gardens,Beer-Sheva, Israel.

January 2010 237PREDATION RISK AND MOLT GAPS IN BIRDS

respect to distance from shelter (Fig. 3), suggestive of a

compensatory effect of seed augmentation on the birds’

seed patch utilization. Evidently, however, when food

was not augmented, overall GUDs were lower for

feather-clipped birds than for birds with unclipped

feathers (Fig. 3). These results contradict our predictions

in that augmentation indeed increased the overall

GUDs, yet reduced the slope of the relationship between

distance from shelter and GUD (distance), rather than

increasing it.

DISCUSSION

In the context of our experiments, the House

Sparrows displayed high quitting harvest rates, as

indicated by the relatively high overall giving-up

densities (;4–6 g of initial amount of 10 g seeds per

seed tray; Fig. 3). They thus seem to behave as cream-

skimmers, taking the more accessible seeds from the

trays and bothering little with seeds that are harder to

extract. As predicted, seed patch utilization decreased

even more (GUDs increased) when food in the aviaries

was augmented, indicating that seed augmentation

indeed reduced the marginal value of food (]Fs/]e). In

contrast to our predictions and the results reported by

Tsurim (2005), seed patch utilization was only slightly

affected by distance from shelter. The influence of

distance was strongest when the birds were feather-

clipped and food was not augmented. Additionally,

while we expected GUDs of feather-clipped birds to be

overall higher than GUDs of birds with intact wings,

they were actually lower. Furthermore, food augmenta-

tion had a lessening, rather than amplifying, effect on

foraging costs of feather-clipped birds, and thus on the

variation in patch utilization and GUDs with respect to

distance from shelter. These results strongly implicate

the interaction of predation risk and food availability in

determining patch use.

Predation risk is commonly regarded as pivotal in

determining foraging behavior (Brown 1988, Kotler and

Brown 1988, Lima and Dill 1990, Kotler et al. 1994,

Lima 1998), including also in urban ecosystems (Sorace

2002, Woods et al. 2003). Yet, with their wings intact,

the House Sparrows in this study were little affected by

variation in Pc with respect to distance from shelter; they

seemed to perceive all microhabitats as equally risky.

Nonetheless, we argue that l does indeed increase with

distance from shelter, but that for the highly urban

House Sparrow the consequent variation in Pc might be

slight and thus difficult to detect (Tsurim 2005, Tsurim

et al. 2008). Indeed, Shochat (2004) and Shochat et al.

(2004) argue that the ecological conditions in urban

environments lead to overmatching between bird pop-

ulation densities and actual food availability, resulting in

intense competition for food at the individual level (Sol

et al. 1998, Marzluff 2001). The consequent increase in

the marginal value of food (]Fs/]e) is then argued to

reduce Pc even if l has not changed or has even

increased, reducing the impact of predation on bird

foraging behavior. Similar behaviors were also recorded

for urban squirrels (Bowers and Breland 1996).

We suggest that the more extensive feather clipping

(six vs. four flight feathers, respectively) in this

experiment, relative to that of Tsurim (2005) increased

wing loading and with it l, resulting in a now detectable

increase in GUD with distance from shelter that

contradicted our second prediction. Impaired escape

abilities due to increased wing-loading and reduced

flight and takeoff performance has previously been

recorded in other species. Robins (Erithacus rubecula),

gaining extra body mass due to migratory fat deposition,

experienced reduced takeoff performance (Lind et al.

1999). Experimental flight feather loss in European

Starlings (Sturnus vulgaris; Swaddle and Witter 1997,

Swaddle et al. 1999) and Tree Sparrows (Passer

montanus; Lind 2001) reduced the birds’ takeoff and

flight performance and impaired their escape ability.

Similarly, reduced body mass in Great Tits (Parus

major) was related to increased predation risk due to

elevated predator abundance (Gosler et al. 1995) or

simulated wing molt gaps (Senar et al. 2002). However,

Van der Veen and Lindstrom (2000) noted that daily

increases in body mass, ranging from 7% to 8%, did not

affect escape performance of Yellowhammers (Emberiza

citrinella) and Greenfinches (Carduelis chloris).

Our results are also suggestive of an important effect

of the simulated molt gaps (feather clipping) on foraging

costs. We predicted that increased l due to feather

clipping will be balanced by a simultaneous increase in

]Fs/]e, resulting in a marginal net change in foraging

costs and GUDs (Tsurim 2005). However, the more

extensive feather clipping performed in the present study

seemed to have tipped the balance between l and ]Fs/]e

in favor of ]Fs/]e, thus driving overall foraging costs,

and consequently GUDs, down. Notably, the effect of

feather clipping on GUDs was more prominent in

patches nearest to shelter, indicative of the important

role of l in determining patch use behavior. A somewhat

similar result was obtained in simulated wing molt gaps

in European Starlings (Swaddle and Witter 1997,

Swaddle et al. 1999). It was shown that the consequent

reductions in flight and takeoff performance as well as

changes in wing movement trajectories are energetically

costly. Additionally, Lind (2001), studying molt gaps in

Tree Sparrows, indicated that the escape ability of

sparrows undergoing slow, natural molt was not

affected by the molt gaps, while experimental increase

of the molt gaps resulted in impaired escape ability.

While l may vary locally, e.g., with respect to distance

from shelter, ]Fs/]e is a property of the whole foraging

environment (Olsson and Molokwu 2007). We suggest

that feather clipping not only increased l, but also the

birds’ energy expenditure in flight. This overall increase

in energetic expenditures should then elevate ]Fs/]e. This

has two effects. The first is to lower the marginal rate of

substitution of energy for survivorship, i.e., Fs/(]Fs/]e),

which somewhat mitigates the effect of higher l on Pc.

IDO TSURIM ET AL.238 Ecology, Vol. 91, No. 1

The second is to lower MOC, which is inversely related

to ]Fs/]e (Brown 1992). The overall effect is a decrease

in GUDs, regardless of distance from shelter.

Hence, for the same value of l, feather-clipped birds

should have lower foraging costs and GUDs than birds

with unclipped feathers. Under the influence of l, thiselevated patch utilization is only likely to be more

intense where l is lowest. At seed trays nearest to shelter

(0–1 m from the shrub), l is probably similar for birds

with clipped and unclipped feathers. Consequently,

foraging costs of birds with clipped feathers are lower

than those of birds with unclipped feathers and their

GUDs are lower (Fig. 3). Predation risk then increases

more with distance from shelter for birds with clipped

feathers, hence the steeper increase in GUDs with

distance from shelter of clipped birds, evident in Fig.

3, but lower, rather than higher, overall GUDs.

Feather clipping may also reduce survivor’s fitness

(Fs), contributing further to reduction in Pc and

consequently in overall foraging costs. However, Fs is

slow to react to changes such as those produced in this

study, and we think that such changes in Fs were likely

negligible under the short time frame of this study. Also,

food supplements were given at the end of each

experimental day regardless of treatment, and feathers

were cut rather than plucked so that feather clipping

provided little direct physiological indications to the

bird that events with potential long-term consequences

had occurred.

Other factors may contribute to explaining our

results. For example, travel costs are undoubtedly

affected by feather clipping. Travel costs can directly

affect foraging decisions because of central place

considerations (Charnov 1976, Orians and Pearson

1979, Schoener 1979, van Gils and Tijsen 2007). While

this may be the case, feather clipping may still affect Pc

(through increasing l), too. Both effects would result in

increasing GUDs. However, assuming no predation

risk, a positive correlation between GUD and distance

from the central place is only predicted when the cost of

traveling back to the central place (in our case the

shelter) increases with the amount of food harvested in

the patch (Olsson et al. 2008). Currently, we are unable

to confidently decompose the effects of the two factors,

and specifically, whether the harvested seeds significant-

ly affect the cost of flying the short distance to the

shelter (1–4 m), within our aviaries. However, unlike

large foragers (such as swans and geese) that make a

single daily trip between the roost and the foraging area,

small foragers, such as sparrows, make multiple short

feeding trips per day from a central place to nearby

foraging patches. These multiple trips likely increase the

time exposed to an approaching predator with increas-

ing distance to the central place (van Gils and Tijsen

2007). The sparrows in our experiments repeatedly fly to

shelter, indicating that predation risk indeed affects their

foraging behavior, regardless of other foraging costs.

We think that the additional cost in time and energy of

flying the extra 1 m between seed trays is probably

negligible with regards to classic central place consider-

ations, irrespective of predation risk. Additionally, the

sparrows’ foraging behavior in the aviaries is not classic

central place foraging, as they often visit more than one

tray between consecutive flights to shelter. This proba-

bly makes foraging considerations in the patches closer

to those of classic optimal patch use than central place

foraging and makes travel costs to the central place less

influential. Still, central place considerations predict that

when average travel costs differ between environments,

GUDs should negatively correlate with travel costs

(Olsson et al. 2008). This is in accordance with our

results of GUDs of feather-clipped vs. unclipped birds,

when food was not augmented (Fig. 3).

Seed augmentation and the consequent reduction in

]Fs/]e possibly compensated for the extra energy

expenditure caused by feather clipping, allowing the

birds to reduce their foraging efforts. This was likely

more prominent in the most depleted patches nearest to

shelter, consequently washing out the qualitative effect

of l with respect to distance from shelter. This effect

may be especially prominent when GUDs are high to

begin with, so that potential further increases in GUDs

are minor and hard to detect, i.e., birds with intact wings

(Fig. 3). Therefore, when food was augmented, GUDs

for neither birds with clipped nor unclipped feathers

varied with distance from shelter. This result contradicts

with our third prediction. Seed augmentation also

resulted in decreased patch use (elevated GUDs) by

two gerbil species (Gerbillus andersoni allenbyi and G.

pyramidum; Kotler 1997). However, unlike our findings

with House Sparrows, the gerbils showed a steeper

increase in GUDs with respect to distance from shelter

when food was augmented. Perhaps an even heavier seed

augmentation for the House Sparrows would have

driven ]Fs/]e even lower than we achieved in the present

study, making the relative role of l in determining Pc

bigger and resulting in increased GUDs with distance

from shelter and higher overall GUDs, as expected in

our third prediction.

The effects of feather clipping and food augmentation

on ]Fs/]e and thus on foraging costs are probably

exacerbated, owing to the role of ]Fs/]e in also

determining MOC. Like Pc, MOC is inversely related

to ]Fs/]e and is thus affected by variations in ]Fs/]e in a

similar manner. Thus ]Fs/]e plays a double role in

determining overall foraging costs. Consequently, it is

likely to have more prominent effects on foraging costs

than l when the two vary simultaneously (Olsson and

Molokwu 2007). Changes in MOC may also come about

through variation in the marginal value of time (U;

Brown 1992), which is positively related to MOC and is

predicted to increase in rich environments, as the value

of missed opportunities is higher (Olsson and Molokwu

2007). However, if foraging activities are not con-

strained by time, then variation in U will probably have

only negligible effect on MOC. These may possibly

January 2010 239PREDATION RISK AND MOLT GAPS IN BIRDS

further obscure local variations in l, e.g., when food is

augmented or when moving between foraging environ-

ments (Olsson and Molokwu 2007).

Our results suggest that the major effect of seed

augmentation in our study was to increase MOC. Since

MOC is a property of the whole environment, increased

foraging costs due to increased MOC following seed

augmentation should be homogeneous and simulta-

neous in all seed trays (i.e., with no variation with

distance to shelter). Hence, if the dominant effect of seed

augmentation was to increase MOC rather than Pc, then

changes in GUDs should not vary within the environ-

ment. We do not think that variations in U following

seed augmentation in the aviaries had a measurable

effect on the sparrows’ patch use behavior because time

does not seem to have been in short supply for the birds

in our present experiment. Hence, the main effect of seed

augmentation on MOC was likely through variations in

]Fs/]e. This is in accord with our results (Fig. 3) and

with predictions based on considerations of central-

place foraging when comparing patch use in food-rich

vs. food-poor environments (Olsson et al. 2008).

Many studies explored the role of predation risk and

its associated costs on foraging behavior. This was most

often achieved by manipulating the foraging environ-

ment. In our experiment we manipulated forager

performance through direct manipulation of its mor-

phology, expecting that impairing flight capabilities will

increase perceived risk of predation. Rarely can one

manipulate the performance abilities of the study

organism to isolate the effect of environmental variables

such as predation risk on its behavior. In contrast,

habitat manipulations are easier to perform and are

more readily executed. Rosenzweig (1973) and Thomp-

son (1982) manipulated the spatial distribution of

predation risk in the habitat for desert rodents by

changing the structural characteristics of the habitat

through redistributing shelters from predators. Rodent

foraging behavior in the manipulated areas was biased

towards the safer habitats and microhabitats, presum-

ably reflecting gradients of predation risk around

shelters, much as seen in our sparrows. Abramsky and

co-workers (e.g., Abramsky et al. 1996), studying gerbils

(G. a. allenbyi and G. pyramidum), and Caraco et al.

(1980), studying Yellow-eyed Juncos (Junco phaeonotus),

used trained birds of prey to study foraging behavior

around shelters. Both concluded that foragers biased

their foraging efforts towards safer microhabitats,

nearer to shelter, when encounter rates with the

predators increased. Accordingly, Brown et al. (1992a)

reported increasing GUDs by a psammophilic gerbil

(G. a. allenbyi ) when foraging farther into the nearby

rocky substrate, presumably due to increased predation

risk due to unsuitable escape substrate or scarcity of

sheltering vegetation. Our experiment is unique among

these studies in that the effect of predation risk on

foraging behavior was studied by manipulating the

forager’s characteristics rather than the foraging envi-

ronment. Our results support previous findings on the

effect of predation risk on habitat and patch use

behavior, but also indicate a more important role for

the interaction between predation risk and food

availability than commonly regarded. More than that,

our study of House Sparrows allows us a glimpse into an

organism with a long association with urban environ-

ments. Intense competition for resources along with

potentially lower risk of predation found in these

environments may lead to adaptations that include the

mild behavioral responses to the risk of predation that

we observed. Such species may be constrained in their

ability to respond to variation in predation risk,

resulting in paler responses to variation in predation

risk than those of nonurban species, even under similar

conditions.

ACKNOWLEDGMENTS

We thank Joel Brown and Ofer Ovadia for valuable adviceand discussions at various stages of this study. We also thankHaim Sivan, Shlomo Hadad, Tomer Zada, and the Beer-ShevaZoological Gardens for allowing the use of their facilities. Thisis publication number 650 of the Mitrani Department of DesertEcology. The experiments performed in this study comply withthe current Israeli laws and were conducted under the requiredpermits from the Israeli Nature and Parks Authority.

LITERATURE CITED

Abramsky, Z., E. Strauss, A. Subach, B. P. Kotler, and A.Reichman. 1996. The effect of Barn Owl (Tyto alba) on theactivity and microhabitat selection of Gerbillus allenbyi andG. pyramidum. Oecologia 105:313–319.

Bowers, M. A., and B. Breland. 1996. Foraging of graysquirrels on an urban–rural gradient: use of the GUD toassess anthropogenic impact. Ecological Applications 6:1135–1142.

Brown, J. S. 1988. Patch use as an indicator of habitatpreference, predation risk, and competition. BehavioralEcology and Sociobiology 22:37–47.

Brown, J. S. 1992. Patch use under predation risk. 1. Modelsand predictions. Annales Zoologici Fennici 29:301–309.

Brown, J. S., Y. Arel, Z. Abramsky, and B. P. Kotler. 1992a.Patch use by gerbils (Gerbillus allenbyi) in sandy and rockyhabitats. Journal of Mammalogy 73:821–829.

Brown, J. S., and B. P. Kotler. 2004. Hazardous duty pay andthe foraging cost of predation. Ecology Letters 7:999–1014.

Brown, J. S., R. A. Morgan, and B. D. Dow. 1992b. Patch useunder predation risk. 2. A test with fox squirrels, Sciurusniger. Annales Zoologici Fennici 29:311–318.

Caraco, T., S. Martindale, and R. H. Pulliam. 1980. Avianflocking in the presence of a predator. Nature 285:400–401.

Charnov, E. L. 1976. Optimal foraging, marginal valuetheorem. Theoretical Population Biology 9:129–136.

Gosler, A. G., J. J. D. Greenwood, and C. Perrins. 1995.Predation risk and the cost of being fat. Nature 377:621–623.

Hedenstrom, A., and M. Rosen. 2001. Predator versus prey: onaerial hunting and escape strategies in birds. BehavioralEcology 12:150–156.

Hedenstrom, A., and S. Sunada. 1999. On the aerodynamics ofmoult gaps in birds. Journal of Experimental Biology 202:67–76.

Kenward, R. E. 1978. Hawks and doves: factors affectingsuccess and selection in Goshawk attacks on Woodpigeon.Journal of Animal Ecology 47:449–460.

Kotler, B. P. 1997. Patch use by gerbils in a risky environment:manipulating food and safety to test four models. Oikos 78:274–282.

IDO TSURIM ET AL.240 Ecology, Vol. 91, No. 1

Kotler, P. B., and J. S. Brown. 1988. Environmentalheterogeneity and the coexistence of desert rodents. AnnualReview of Ecology and Systematics 19:281–307.

Kotler, P. B., J. S. Brown, and W. A. Mitchell. 1994. The role ofpredation in shaping the behavior, morphology and commu-nity organization of desert rodents. Australian Journal ofZoology 42:449–466.

Lima, S. L. 1993. Ecologial and evolutionary perspectives onescape from predatory attack: a survey of North Americanbirds. Wilson Bulletin 105:1–47.

Lima, S. L. 1998. Nonlethal effects in the ecology of predator–prey interactions: What are the ecological effects of anti-predator decision-making? BioScience 48:25–34.

Lima, S. L., and L. M. Dill. 1990. Behavioral decisions madeunder the risk of predation: a review and prospectus.Canadian Journal of Zoology 68:619–640.

Lind, J. 2001. Escape flight in moulting tree sparrows (Passermontanus). Functional Ecology 15:29–35.

Lind, J., T. Fransson, S. Jakobsson, and C. Kullberg. 1999.Reduced take-off ability in robins (Erithacus rubecula) due tomigratory fuel load. Behavioral Ecology and Sociobiology46:65–70.

Lindstrom, A., G. H. Visser, and S. Daan. 1993. The energeticcost of feather synthesis is proportional to basal metabolic-rate. Physiological Zoology 66:490–510.

Marzluff, J. M. 2001. Worldwide urbanization and its effects onbirds. Pages 19–38 in J. M. Marzluff, R. Bowman, and R.Donnelly, editors. Avian ecology and conservation in anurbanizing world. Kluwer, Boston, Massachusetts, USA.

McNamara, J. M., and A. I. Houston. 1987. Starvation andpredation as factors limiting population size. Ecology 68:1515–1519.

Olsson, O., J. S. Brown, and K. L. Helf. 2008. A guide tocentral place effects in foraging. Theoretical PopulationBiology 74:22–33.

Olsson, O., and M. N. Molokwu. 2007. On the missedopportunity cost, GUD, and estimating environmentalquality. Israel Journal of Ecology and Evolution 53:263–279.

Orians, G. H., and N. E. Pearson. 1979. On the theory ofcentral place foraging. Pages 154–177 in D. J. Horn, R. D.Mitchell, and G. R. Stairs, editors. Analysis of ecologicalsystems. Ohio State University Press, Columbus, Ohio, USA.

Oyugi, J. O., and J. S. Brown. 2003. Giving-up densities andhabitat preferences of European Starlings and AmericanRobins. Condor 105:130–135.

Rosenzweig, M. L. 1973. Habitat selection experiments withpair of coexisting heteromyid rodent species. Ecology 54:111–117.

Schoener, T. W. 1979. Generality of the size–distance relationin models of optimal feeding. American Naturalist 114:902–914.

Senar, J. C., J. Domenech, and F. Uribe. 2002. Great tits (Parusmajor) reduce body mass in response to wing area reduction:a field experiment. Behavioral Ecology 13:725–727.

Shochat, E. 2004. Credit or debit? Resource input changespopulation dynamics of city-slicker birds. Oikos 106:622–626.

Shochat, E., S. B. Lerman, M. Katti, and D. B. Lewis. 2004.Linking optimal foraging behavior to bird communitystructure in an urban-desert landscape: field experimentswith artificial food patches. American Naturalist 164:232–243.

Sol, D., D. M. Santos, J. Garcia, and M. Cuadrado. 1998.Competition for food in urban pigeons: the cost of beingjuvenile. Condor 100:298–304.

Sorace, A. 2002. High density of bird and pest species in urbanhabitats and the role of predator abundance. Ornis Fennica79:60–71.

Swaddle, J. P., E. V. Williams, and J. M. V. Rayner. 1999. Theeffect of flight feather moult on escape take-off performancein starlings. Journal of Avian Biology 30:351–358.

Swaddle, J. P., and M. S. Witter. 1997. The effects of moult onthe flight performance, body mass and behavior of EuropeanStarlings (Sturnus vulgaris): an experimental approach.Canadian Journal of Zoology 75:1135–1146.

Thompson, S. D. 1982. Structure and species composition ofdesert Heteromyid rodent species assemblages: habitatmanipulation. Ecology 63:1313–1321.

Todd, I. A., and R. J. Cowie. 1990. Measuring risk of predationin an energy currency: field experiments with foraging bluetits (Parus caeruleus). Animal Behavior 40:112–117.

Tsurim, I. 2005. Diversity and composition of wintering birdassemblies in a semi-desert scrubland. Dissertation. Ben-Gurion University of the Negev, Beer Sheva, Israel.

Tsurim, I., Z. Abramsky, and B. P. Kotler. 2008. Foragingbehavior of urban birds: Are human commensals lesssensitive to predation risk than their nonurban counterparts?Condor 110:772–776.

Van der Veen, I. T., and K. M. Lindstrom. 2000. Escape flightsof yellowhammers and greenfinches: more than just physics.Animal Behaviour 59:593–601.

van Gils, J. A., and W. Tijsen. 2007. Short-term foraging costsand long-term fueling rates in central-place foraging swansrevealed by giving-up exploitation times. American Natural-ist 169:609–620.

Watts, B. D. 1990. Cover use and predator-related mortality inSong and Savanna Sparrows. Auk 107:775–778.

Williams, E. V., and J. P. Swaddle. 2003. Moult, flightperformance and wingbeat kinematics during take-off inEuropean starlings Sturnus vulgaris. Journal of AvianBiology 34:371–378.

Witter, M. S., I. C. Cuthill, and R. H. C. Bosner. 1994.Experimental investigation of mass-dependent predation riskin the European Starling, Sturnus vulgaris. Animal Behaviour48:201–222.

Woods, M., R. A. McDonald, and S. Harris. 2003. Predation ofwildlife by domestic cats Felis catus in Great Britain.Mammal Review 33:174–188.

January 2010 241PREDATION RISK AND MOLT GAPS IN BIRDS