Article Summaries

� Author for correspondence ([email protected]).

Proc. R. Soc. Lond. B (2004) 271, 2473–2479 2473 doi:10.1098/rspb.2004.2913

Received 25 July 2004

Accepted 24 August 2004

Published online 23 November 2004

Balancing food and predator pressure induces chronic stress in songbirds

Michael Clinchy 1� , Liana Zanette

1 , Rudy Boonstra

2 , John C. Wingfield

3 and

James N. M. Smith 4

1 Department of Biology, University of Western Ontario, London, Ontario N6A 5B7, Canada

2 Centre for the Neurobiology of Stress, University of Toronto, Toronto, Ontario M1C 1A4, Canada

3 Department of Biology, University of Washington, Seattle, WA 98195-1800, USA

4 Department of Zoology, University of British Columbia, Vancouver, British Columbia V6T 1Z4, Canada

The never-ending tension between finding food and avoiding predators may be the most universal natural

stressor wild animals experience. The ‘chronic stress’ hypothesis predicts: (i) an animal’s stress profile will

be a simultaneous function of food and predator pressures given the aforesaid tension; and (ii) these insepar-

able effects on physiology will produce inseparable effects on demography because of the resulting adverse

health effects. This hypothesis was originally proposed to explain synergistic (inseparable) food and predator

effects on demography in snowshoe hares (Lepus americanus). We conducted a 2 � 2, manipulative food addition plus natural predator reduction experiment on song sparrows (Melospiza melodia) that was, to our

knowledge, the first to demonstrate comparable synergistic effects in a bird: added food and lower predator

pressure in combination produced an increase in annual reproductive success almost double that expected

from an additive model. Here we report the predicted simultaneous food and predator effects on measures of

chronic stress in the context of the same experiment: birds at unfed, high predator pressure (HPP) sites had

the highest stress levels; those at either unfed or HPP sites showed intermediate levels; and fed birds at low

predator pressure sites had the lowest stress levels.

Keywords: chronic stress; food supplementation; predator pressure; synergistic effects; Melospiza melodia

1. INTRODUCTION For sound logistical reasons population-scale experiments

on terrestrial vertebrates have focused on one limiting factor

at a time and most experiments on birds have focused on

food effects (Newton 1998). Virtually all these experiments

have been conducted in environments, or under circum-

stances, where predator pressure is low (Zanette et al.

2003). We conducted a bifactorial experiment to test

whether birds responded equally to food addition in high

predator pressure (HPP) and low predator pressure (LPP)

environments. Our results showed that food and predators

did not operate in an additive way, but instead had an inter-

active (or ‘synergistic’) effect on annual reproductive

success (Zanette et al. 2003). The combined effect of added

food and LPP produced an increase in annual reproductive

success almost twice that expected from an additive model.

To our knowledge, this is the first experimental study to

show such synergistic effects in birds.

The first experimental study to demonstrate synergistic

food and predator effects on the demography of any terres-

trial vertebrate was published by Krebs et al. (1995). This

study involved a bifactorial experiment on snowshoe hares.

The combined effect of adding food and removing

predators produced densities 1.9 times greater than that

expected from an additive model. The hare results were

completely unexpected and the ‘chronic stress’ hypothesis

(Boonstra et al. 1998) was proposed to explain the

mechanism responsible (Krebs et al. 2001). The ‘chronic

stress’ hypothesis predicts that: (i) an animal’s stress profile

will be a simultaneous function of both food and predator

pressures as a consequence of the never-ending tension

between finding food and avoiding predators; and (ii) the

inseparable effects of food and predators on physiology will

result in inseparable effects on demography owing to the

long-term adverse health effects of chronic stress. The

snowshoe hare study ended before the ‘chronic stress’

hypothesis could be tested experimentally (Krebs et al.

2001).

The ‘chronic stress’ hypothesis is an extension of the

‘predator-sensitive foraging’ hypothesis (Hik 1995;

Boonstra et al. 1998; Krebs et al. 2001). The results from

the hare study were consistent with predictions from mod-

els (McNamara & Houston 1987; Abrams 1993) suggest-

ing nonlinear (e.g. synergistic) changes in demography may

be commonplace as a consequence of linear changes in the

individual’s anti-predator and foraging behaviour. The

difficulty in testing these models lies in the fact that behav-

iour is fleeting data collected at a given point in time that

cannot be inferred to have lasting consequences. Sceptics

can argue that the animal’s behaviour over the hour or two

during which observations were conducted may be unrep-

resentative of what it is doing the rest of the time. The

physiological stress effects proposed by the ‘chronic stress’

hypothesis provide the missing link between these short-

term behavioural and longer-term demographic processes,

the stress effects being the lasting ‘imprint’ of ‘predator-

sensitive foraging’ (Boonstra et al. 1998). Results from

literally hundreds of behavioural studies suggest that the

constant tension between finding food and avoiding

# 2004 The Royal Society

2474 M. Clinchy and others Chronic food and predator stress

predators afflicts animals in virtually every vertebrate taxon

(reviewed in Lima 1998). Inseparable food and predator

effects on demography ought then to be the norm in both

birds and mammals if the ‘chronic stress’ hypothesis is

correct. Having demonstrated inseparable (synergistic)

food and predator effects on the demography of song

sparrows (Zanette et al. 2003) paralleling those shown in

snowshoe hares (Krebs et al. 1995), we undertook to test

whether each individual’s stress profile was a simultaneous

function of food and predators, as predicted by the ‘chronic

stress’ hypothesis.

2. METHODS (a) Experimental design

Our study was conducted in the context of the same experiment

described in Zanette et al. (2003). We monitored 91 song sparrow

territories for the entire breeding season at 16 study sites near

Victoria, British Columbia, Canada in 2002. Song sparrows in

this area are resident and multi-brooded. Breeding begins in late

March and ends in late July. Individuals can rear up to four broods

of 1–4 young per year. We conducted a standard 2 (fed or unfed)

by 2 (HPP or LPP) experiment. Added food, consisting of high fat

and high protein (45%) pellets (Purina Mills Aquamax Grower

400) and millet was provided ad libitum throughout the breeding

season (1 March onwards) to all of the territories at half of the

sites. HPP sites (three fed plus three unfed) were located on the

Vancouver Island (32 137 km 2 ) ‘mainland’, while LPP sites

(five fed plus five unfed) were less than 20 km distant, on several

small (less than 200 ha) coastal islands. There were no significant

differences between HPP and LPP sites in either nesting

density (nearest neighbour distances: HPP ¼ 52:7 ^ 4:7 m; LPP ¼ 62:5 ^ 5:3 (mean^s:e:m:); t89 ¼ 1:38, p ¼ 0:179) or microsatellite heterozygosity (L. Zanette, unpublished data) and

no significant difference in extra-pair paternity rates between fed

birds at the HPP and LPP sites (L. Zanette, unpublished data).

HPP sites supported a greater diversity and abundance of poten-

tial predators (Zanette et al. 2003) and song sparrows at HPP sites

demonstrated significantly higher nest predation (68% versus

55%, HPP–LPP), higher brood parasitism rates (40% versus 9%),

lower survival from fledging to independence (53% versus 82%)

and lower adult breeding season survival (84% versus 92%)

(Zanette et al. 2003; L. Zanette, unpublished data).

Proc. R. Soc. Lond. B (2004)

(b) Measures of chronic stress

We tested for stress effects at five separate scales: hormonal,

energetic, haematological, immunological and reproductive. We

evaluated two measures at each scale. The authors of the ‘chronic

stress’ hypothesis (Boonstra et al. 1998) predicted changes in all of

these measures and in most cases they also predicted the expected

direction of change (table 1). Changes in the same direction are

predicted in response to food shortage (# food) or increased pred- ator pressure (" predators), because either is potentially stressful. The critical prediction of the ‘chronic stress’ hypothesis is that

both food and predators simultaneously affect stress levels.

The most direct measure of chronic stress, among those we eval-

uated, is the maximum concentration of the principal stress hor-

mone, corticosterone, recorded in response to a standard stressor.

The standard stressor used in most songbird studies, known as the

‘capture stress protocol’ (Wingfield et al. 1995), involves restrain-

ing the animal for a set period of time post-capture. Extensive prior

research on song sparrows has demonstrated that corticosterone

levels are maximal in blood collected 30 min post-capture (Wing-

field et al. 1995). Baseline corticosterone, established by collecting

blood less than 3 min post-capture (Scheuerlein et al. 2001), refers

to concentrations in animals going about their daily routine. Base-

line corticosterone may be biased if stressful events (unknown to

the experimenter and unassociated with capture) occur immedi-

ately prior to capture. Maximum corticosterone is less likely to be

biased by immediate events and is therefore a truer measure of

chronic stress because levels are generally thought to be contingent

on long-term enlargement of the adrenals (Boonstra et al. 1998).

The authors of the ‘chronic stress’ hypothesis predicted that

chronically stressed animals should have a greater ability to mobilize

energy for immediate muscle use at the expense of ‘maintenance’ or

reproduction (Boonstra et al. 1998). In birds, free fatty acids

(FFAs) power the flight muscles and glucose powers the leg mus-

cles (Butler & Bishop 2000). Accordingly, both FFA and glucose

levels should be higher in chronically stressed birds (table 1).

Anaemia (inverse of packed cell volume (PCV)) may be expec-

ted in response to chronic stress as a result of red blood cell (RBC)

loss attributable to ulcers, high blood pressure or poor initial cell

formation (Campbell 1988). Polychromasia is an index of the

proportion of RBCs that are immature, reflecting RBC regener-

ation (Campbell 1988).

The authors of the ‘chronic stress’ hypothesis predicted changes

in the composition of white blood cells (WBCs) reflecting changes

in immune function in response to chronic stress (Boonstra et al.

Table 1. Food and predator effects on measures of chronic stress. (Up and down arrows signify the expected direction of change in each measure as predicted by the ‘chronic stress’ hypothesis, and that observed, in response to food shortage (# food) and/or increased predator pressure (" predators). Statistical results are main effects from two-way ANOVAs comparing fed versus unfed birds at HPP and LPP sites.)

change

# food

" predators

scale

measure

expected

observed

d.f.

F

p

F

p

hormonal m

aximum corticosterone

"

"

1,40

5.22

0.014

7.76

0.004

baseline corticosterone

"

"

1,40

9.12

0.002

5.92

0.010

energetic

FFAs

"

"

1,35

4.58

0.020

5.08

0.015

glucose

"

"

1,29

13.55

0.001

0.72

n.s.

haematological

anaemia

"

"

1,41

2.99

0.046

4.13

0.025

polychromasia

"

"

1,42

6.94

0.006

0.77

n.s.

immunological

basophils

—

"

1,42

0.28

n.s.

6.74

0.013

H : L ratio

"

—

1,42

0.59

n.s.

0.40

n.s.

reproductive

nestling brood size

#

#

1,65

5.35

0.012

0.22

n.s.

nestling FA

"

"

1,43

0.14

n.s.

5.59

0.009

Chronic food and predator stress M. Clinchy and others 2475

1998). They did not predict which WBCs would predominate

(table 1). In birds, a higher heterophil to lymphocyte (H : L) ratio

is commonly used as an index of general stress based on poultry

studies showing elevated H : L ratios in response to exogenous

corticosterone (Carsia & Harvey 2000; see also McFarlane &

Curtis 1989).

Chronic stress may affect reproductive rates by reducing either

the quantity or quality of offspring produced. The authors of the

‘chronic stress’ hypothesis proposed using fluctuating asymmetry

(FA) to judge offspring quality (Boonstra et al. 1998). FA is a

measure of deviation from symmetry in bilateral organisms

gauged by differences in paired appendages (e.g. arms or legs

(Palmer 1994)). Greater stress is expected to result in more

frequent developmental anomalies and greater FA (table 1).

(c) Sampling

We captured and collected blood from fathers with 6 day old

nestlings at 46 territories (10–13 per treatment). We conducted the

capture stress protocol on fathers to avoid any potentially adverse

effects on nestlings that might result from restraining the mother.

Birds were captured using mist-nets. Up to 150 ml of blood was collected from the brachial vein less than 3 min from the time the

animal hit the net. Blood from this first (baseline) bleed was used

for all physiological (i.e. excluding reproductive) measures except

maximal corticosterone concentration (table 1). The latter was

measured in blood from a subsequent bleed conducted 30 min

post-capture (Wingfield et al. 1995). Birds were held in cloth bags

in the intervening period. All animals were bled at 10 min post-cap-

ture and eight were bled at 60 min post-capture to verify (Wingfield

et al. 1995) that corticosterone levels increased from 10 to 30 min

(paired t42 ¼ �9:43, p < 0:001) and did not increase further after more than 30min (paired t7 ¼ 1:79, p ¼ 0:117). Glucose was measured and blood smears were prepared within 2 min of the first

bleed. All remaining blood was stored on ice for transport to the

laboratory. All samples were centrifuged, measured for PCV, and

plasma was extracted and frozen at �20 vC, within 8 h. As repro- ductive measures (table 1) of potential stress effects at this stage of

the breeding cycle we tallied brood size and measured the right and

left tarsus lengths of all nestlings to compare levels of FA. Tarsus

lengths were each measured twice, to the nearest 0.01 mm, and

analyses were restricted to measurements made by a single

observer (Palmer 1994).

(d) Laboratory analyses

Radioimmunoassays of corticosterone concentrations were con-

ducted using 5–20ml of plasma following extraction in dichloro- methane (Wingfield et al. 1992). Plasma FFA concentrations were

determined using the NEFA C test kit (Wako Chemicals, Neuss,

Germany) (Johnson & Peters 1993). Glucose was measured using

the ONE TOUCH Ultra (LifeScan Canada Ltd, Vancouver,

Canada). Smears were stained using Wright’s stain and poly-

chromasia and WBC differentials were evaluated by trained techni-

cians at the Animal Health Laboratory, Ontario Veterinary

College, University of Guelph (Guelph, Canada). Basophils con-

stituted 11:9^1:0% (mean ^ s:e:) of all WBCs. While basophil

counts are typically lower in poultry, higher counts are found in

other species (King & McLelland 1984) and may be the norm in

New World sparrows (Ruiz et al. 2002). Eosinophils and mono-

cytes each constituted less than 5% of all WBCs

(3:7 ^ 0:6% and 1:3 ^ 0:3%, respectively). Eosinophils and

monocytes are difficult to distinguish from heterophils and

lymphocytes, respectively (Campbell 1988). Separate analyses

of the H : L ratio were conducted either pooling ((het: þ eos:):

Proc. R. Soc. Lond. B (2004)

(lym: þ mono:); table 1) or discriminating (het::lym: only) difficult to distinguish WBCs. There was no effect of either pooling or dis-

criminating on the outcome of analyses.

(e) Data analyses

We conducted two-way ANOVAs of all the measures in table 1

comparing fed versus unfed birds at HPP and LPP sites. Only

main effects are reported because only one of the interaction terms

(figure 1f) approached significance (p > 0:10 in all other cases).

Results in table 1 are one-tailed where a priori predictions exist.

Prior to analysis, all data except |R–L| tarsus length were

Box–Cox transformed (Krebs 1999) and tested for normality and

homogeneity of variances. Relevant regressions (stepwise, where

multiple independent variables) were conducted of time (seconds)

from capture to bleed, time (minutes) of day, date and/or brood

size, versus each measure. There was no significant correlation

between baseline corticosterone and time from capture to the first

bleed (r2 ¼ 0:05, t43 ¼ 1:46, p ¼ 0:151), nor were any other regressions significant except in the case of FA. Degrees of free-

dom for the physiological measures in table 1 vary because of

occasional sample losses. Analyses of brood size and FA include

all available data from the larger demographic study. Where there

was more than one day 6 brood per territory the average day 6

brood size for the territory was used. Meaningful FA (Palmer

1994; Palmer & Strobeck 2003) was indicated by a significant

‘sides � individuals’ interaction (F48,98 ¼ 3:92, p < 0:001; repeat- ability [ME5] rA ¼ 0:056), using data from one randomly-selec- ted nestling per brood. Average FA per brood showed a negative

correlation (r2 ¼ 0:28, t46 ¼ �4:27, p < 0:001) with date and a positive correlation (r2 ¼ 0:45, t46 ¼ 6:12, p < 0:001) with brood size. Consequently, date and brood size were included as

covariates when analysing effects on FA.

3. RESULTS Both food and predators significantly affected the indivi-

dual’s stress hormone profile, as predicted by the ‘chronic

stress’ hypothesis. Both maximum (figure 1a) and baseline

(figure 1b) corticosterone levels varied significantly with

both food and predators (table 1). The direction of change

was also as predicted (table 1), being greater at unfed than

at fed sites and greater at HPP sites than at LPP sites

(figure 1a,b).

Both food and predators also significantly affected the

individual’s energetic profile. FFA levels varied significantly

withbothfoodandpredators(table1),showingthesamepat-

tern as corticosterone levels (figure 1c). Food alone

affected glucose levels (table 1; fed ¼ 361:1 ^ 5:6 mg dl�1; unfed ¼ 393:8^6:4Þ:

Both food and predators significantly affected the

individual’s haematological profile, in accordance with the

‘chronicstress’hypothesis.Anaemiavariedsignificantlywith

both food and predators (table 1), showing the same

pattern as corticosterone and FFA levels (figure 1d).

Food alone affected polychromasia (table 1; fed ¼ 5:5 ^ 0:5%; unfed ¼ 7:5 ^ 0:6).

Predator pressure alone affected our immunological

measures. Basophils made up a significantly greater pro-

portion (table 1) of WBCs at HPP sites (14:5 ^ 1:5%) than at LPP sites (9:4 ^ 1:3%). Neither food nor predators affected the H : L ratio (table 1). There was no significant

correlation between the H : L ratio and either maximum

(r2< 0:01, t42 ¼ 0:40, p ¼ 0:688) or baseline (r2 ¼ 0:03, t43 ¼ 1:16, p < 0:254) corticosterone concentrations.

2476 M. Clinchy and others Chronic food and predator stress

Food and predators affected the quantity and quality,

respectively, of day 6 nestlings. Broods at unfed sites were

significantly smaller than those at fed sites (figure 1e; table 1)

and average FA in nestling tarsus lengths per brood was sig-

nificantly greater at HPP sites than at LPP sites (figure 1f;

table 1). The direction of change in both cases was as pre-

dicted by the ‘chronic stress’ hypothesis (table 1).

4. DISCUSSION Chronic stress in response to both food and predators was

evident at the hormonal, energetic, hematological and

reproductive scales (table 1). Song sparrows at the unfed,

HPP sites showed the highest stress levels (figure 1a–d),

birds subject to either the unfed or HPP treatments showed

intermediate stress levels, and fed birds at the LPP sites

showed the lowest stress levels (figure 1a–d). Thus, the

stress profile of parental male song sparrows on day 6 of the

nestling period clearly appears to be a simultaneous

function of food and predators, as predicted by the ‘chronic

stress’ hypothesis.

Proc. R. Soc. Lond. B (2004)

We are not aware of any prior field study showing effects

on corticosterone in response to food addition in songbirds

despite the numerous laboratory studies on corticosterone

effects on songbird foraging (Wingfield & Silverin 2002).

Similarly, while several songbird studies have shown acute

activation of the stress axis in response to the experimental

presentation of a predator (Silverin 1998; Canoine et al.

2002; Cockrem & Silverin 2002), only one has tested

whether predators can induce chronic activation of the

stress axis. Scheuerlein et al. (2001) reported that tropical

stonechats (Saxicola torquata axillaris) with predatory fiscal

shrikes (Lanius collaris) in their territories had significantly

higher baseline corticosterone than those without shrikes.

In contrast to the strong effect of predators on maximum

corticosterone that we observed (figure 1a), shrikes did not

affect maximum corticosterone in the stonechats. The

simplest explanation of this apparent difference is that our

HPP versus LPP treatment involved a suite of predators

rather than just one.

While food effects on the energetic profile of songbirds

have been reported previously (Jenni & Jenni-Eiermann

1996; Totzke et al. 1998) we are not aware of any other

study examining predator effects. The finding that food

affected both FFAs (figure 1c) and glucose (table 1) is con-

sistent with the fact that song sparrow foraging involves both

flying and hopping, entailing the use of FFAs and glucose

respectively. The finding that predator pressure affected

FFAs alone (table 1) is consistent with predator evasion rely-

ing primarily on FFA-dependent flight muscles rather than

glucose-dependent leg muscles (Butler & Bishop 2000).

As with energetics, food effects on the haematological

profile of songbirds are well documented (Hoi-Leitner

et al. 2001; Cucco et al. 2002) while predator effects have

been neglected. Because both polychromasia and anaemia

were greater at unfed sites (table 1) RBC regeneration

(polychromasia) may have been compensating for RBC

loss (anaemia), whereas the effect of predator pressure on

anaemia alone (table 1) suggests RBC regeneration was not

compensating for RBC loss at HPP sites and the haemato-

logical profile of birds at these sites was actually worse than

that indicated by anaemia alone (figure 1d).

While there has been much research on whether the

immune response affects vulnerability to predators in song-

birds (Moller & Erritzoe 2000), the significant effect of

predator pressure on basophils reported in table 1 is, to our

knowledge, the first demonstration of predator effects on

the immune response. Like most avian WBCs the function

of basophils is largely unknown (Campbell 1988). At

present, therefore, there is no way to judge the importance

of this result. Despite its common usage, the adequacy of

the H : L ratio as an indicator of general stress has rarely

been evaluated in wild birds (Ots et al. 1998). Given the

lack of correlation between the H : L ratio and corticoster-

one in song sparrows we suggest caution in the interpret-

ation of the H : L ratio as an indicator of general stress in

species for which this has not been corroborated.

Increased brood size in response to experimental food

addition (figure 1e) has been shown before in song spar-

rows (Arcese & Smith 1988). While predator pressure

alone significantly affected FA (table 1) there was an obvi-

ous trend ðp ¼ 0:079Þ towards an interactive effect of food and predators (figure 1f). FA increases with brood size

(see x 2e) suggesting the strain of producing more offspring

125

110

95

80

65

2.1

1.9

1.7

1.5

1.3

3.2

3.0

2.8

2.6

2.4

14

12

10

8

6

54

53

52

51

50

0.18

0.15

0.12

0.09

0.06

fed unfed fed unfed

(a) (b)

(c) (d )

(e) ( f )

m ax

. co

rt . (n

g m

l– 1 )

F F

A s

(m m

o l

l– 1 )

b ro

o d s

iz e

b as

e. c

o rt

. (n

g m

l– 1 )

an ae

m ia

( %

p la

sm a)

F A

Figure 1. Measures of chronic stress in song sparrows on day 6 of the nestling period at fed and unfed sites subject to high (closed circles) and low (open circles) predator pressure. Values are means^s:e:m: (a) Maximum plasma corticosterone concentration; (b) baseline plasma corticosterone concentration; (c) plasma FFA concentration; (d) anaemia (percentage plasma in hematocrit); (e) average nestling brood size; and ( f ) average FA in nestling tarsus lengths, per brood.

Chronic food and predator stress M. Clinchy and others 2477

results in poorer quality offspring. Because food addition

increases brood size it may also increase FA. The results in

figure 1f are, however, corrected for brood size. Food

addition also significantly increases the number of broods

per season (Arcese & Smith 1988; L. Zanette, unpublished

data). Both food and predators may affect FA if the strain

of producing more broods in response to food addition is

exacerbated at HPP and attenuated at LPP sites (figure 1f).

There is a growing literature on the effects of early

nutritional stress on neuronal development and resultant

adult learning disabilities in birds (Nowicki et al. 1998;

Kitaysky et al. 2003). Higher FA at HPP sites (table 1) may

result from predator effects on parental foraging. Mothers

spend significantly more time guarding the nest on day 6 of

the nestling period at HPP sites (L. Zanette, unpublished

data), which presumably limits their ability to find food for

themselves and their young. If predator-induced

nutritional stress affects FA this could also affect neuronal

development and adult learning ability. Song learning sig-

nificantly affects pairing success in song sparrows (Nowicki

et al. 2002). Thus, both food- (Nowicki et al. 1998) and

predator-induced stress on parents may affect the pairing

success of their sons.

Our measures at different scales reflect different points on

the ‘stress axis’ (see Boonstra et al. 1998, fig. 1; Romero

2004, fig. 1). Elevated corticosterone should have cascading

effects on energetics, haematology, immunology and repro-

duction. Eight of our 10 measures are repeated measures of

the same individual. If head length, arm length and leg

length were consistently greater in group A than B, we

would conclude individuals in group A were larger, even if

the differences in each measure were slight. Correcting for

multiple comparisons would clearly be inappropriate. Simi-

larly, evidence of chronic stress is clearly more obvious the

more points on the ‘stress axis’ are affected. Measures at dif-

ferent scales represent multiple tests of the same hypothesis.

In this case, a Fisher’s combined probabilities test (Sokal &

Rohlf 1995) is the most correct statistical procedure to fol-

low. Such a test is, however, manifestly redundant, given

the dramatically significant results at so many different

scales (table 1).

We based our selection of HPP and LPP sites on known

differences in predation rates on nearby island and main-

land song sparrow populations (Arcese et al. 1992; Smith

et al. 1996; Rogers et al. 1997) and the general observation

that predator pressure is lower on islands (Palkovacs

2003). Given our a priori selection of sites likely to differ in

predator pressure, the fact that those sites do differ in pred-

ator pressure (see x 2a) and the absence of obvious alter- natives (see x 2a), we think it entirely reasonable to ascribe the observed differences in stress levels (open versus closed

circles in figure 1) to differences in predator pressure. Simi-

larly, we can reject alternative explanations of our results as

being attributable to among treatment differences in the

point in the reproductive cycle sampled or differences asso-

ciated with circadian or seasonal fluctuations in hormone

levels because we controlled for the point in the repro-

ductive cycle sampled in our experimental design by sam-

pling all males on the same day (day 6) during the nestling

period (see x 2c), and potential biases associated with time of day or date were found to be non-significant in our stat-

istical analyses (see x 2e).

Proc. R. Soc. Lond. B (2004)

A simple example illustrates how nonlinear changes in

demography result from linear changes in behaviour. In

theory, doubling time spent vigilant will halve the indivi-

dual’s time spent foraging, resulting in a decreased prob-

ability of death owing to predation and increased

probability of death owing to starvation (McNamara &

Houston 1987; Abrams 1993). Total mortality will remain

unchanged only in the peculiar event that predation and

starvation are linearly proportional. If a unit increase in

time spent vigilant decreases predation faster than the unit

decrease in time spent foraging increases starvation, total

mortality will decrease (McNamara & Houston 1987;

Abrams 1993). In this case demography (total mortality) is

not a simple linear function of linear changes in behaviour.

Similarly, there is no reason to expect that nonlinear (e.g.

synergistic) changes in demography can only result from

nonlinear (interactive) physiological effects. Indeed,

Romero (2004) has recently argued that linear physiologi-

cal responses such as we observed may result from non-

linear physiological processes. The focus of the ‘chronic

stress’ hypothesis, like the ‘predator-sensitive foraging’

hypothesis, is the inseparable link between food and

predators at the individual scale rather than the linearity or

nonlinearity of the resulting physiological or behavioural

phenomena. Synergistic food and predator effects on

demography demonstrate an inseparable link at the popu-

lation scale. Showing inseparable food and predator links at

both the individual and population scales is the critical step

in testing the ‘chronic stress’ hypothesis.

At the scale of the individual, chronic stress is normally

associated with pathology. In our study, euthanizing indivi-

duals to look for tissue damage would mean sacrificing our

ability to simultaneously test for demographic effects.

Larger sample sizes than ours would be necessary to permit

subsampling for pathology. While chronic stress may

induce pathology it is not necessarily maladaptive. Rather,

the organism may be ‘making the best of a bad job’.

Elevated corticosteroid levels are very adaptive when

behaviour must be redirected from resting to running, for

example, when face to face with a predator. ‘Allostasis’,

defined as maintaining stability through change, has been

proposed as an alternative to ‘stress’, to avoid the connota-

tions of maladaptation associated with the latter term

(McEwen & Wingfield 2003). ‘Allostatic overload’ in turn

refers to situations where the organism’s ability to maintain

internal stability is exceeded. Allostatic overload may result

from either acute situations where energy demand exceeds

supply (type 1), or chronic situations where no clear alter-

native behavioural response can alleviate the threat (type

2). The pathological consequences of the latter can be seen

in the illnesses that often afflict animals in captivity. We

suggest wild animals face an analogous challenge in being

unable to escape the necessity of finding food in an

environment where the threat of death is always imminent.

Physiological and behavioural compromises must be made.

The truly maladaptive options are either not to find food or

to ignore the predators.

In this study we have, in a sense, been working backwards.

After first demonstrating inseparable (synergistic) food and

predator effects on demography (Zanette et al. 2003), we

have now verified the presence of the individual-level

mechanism predicted by the ‘chronic stress’ hypothesis to

be responsible. The next step is to determine which

2478 M. Clinchy and others Chronic food and predator stress

demographic parameters are most affected. What we find

most compelling about our data is how well the results

concerning song sparrows correspond with predictions orig-

inally derived from work on snowshoe hares (table 1). Given

the numerous differences between these species, this speaks

volumes for the probable generality of these phenomena.

The underlying behavioural link, being the never-ending

tension between finding food and avoiding predators, is

acknowledged as being nearly universal (Lima 1998). There

is no reason to expect the stress axis differs dramatically

among organisms (Sapolsky et al. 2000). Thus, there is every

reason to expect that the ‘chronic stress’ hypothesis should

apply to the majority of vertebrates. Because the individual-

level mechanism is very likely nearly universal, the resulting

demographic effects should be as well. Consequently, we

suggest future demographic studies begin by assuming food

and predator effects are inseparable. Indeed, this may be

critical for species protection. Conservation efforts aimed at

either food or predators are often disappointing (Zanette

2000; Zanette et al. 2000, 2003). Simultaneously targeting

both may not only be the key, but could also provide

disproportionate benefits per dollar spent given the more

than additive demographic responses shown in both song

sparrows and snowshoe hares.

We thank B. Clinchy, C. de Ruyck, L. Erckman, A. Duncan, J. Malt, I. K. Barker, D. Smith, T. Sperry, B. S. McEwen and two anonymous reviewers for assistance; and BC Parks, the Saanich Municipality and the owners of Tortoise and Domville islands for access. Funding was provided by the Natural Sciences and Engineering Research Council of Canada and the US National Science Foundation.

REFERENCES Abrams, P. A. 1993 Why predation rate should not be pro-

portional to predator density. Ecology 74, 726–733. Arcese, P. & Smith, J. N. M. 1988 Effects of population

density and supplemental food on reproduction in song

sparrows. J. Anim. Ecol. 57, 119–136. Arcese, P., Smith, J. N. M., Hochachka, W. M., Rogers, C. M.

& Ludwig, D. 1992 Stability, regulation, and the determi-

nation of abundance in an insular song sparrow population.

Ecology 73, 805–822. Boonstra, R., Hik, D., Singleton, G. R. & Tinnikov, A. 1998

The impact of predator-induced stress on the snowshoe

hare cycle. Ecol. Monogr. 68, 371–394. Butler,P.J.&Bishop,C.M.2000Flight.InSturkie’savianphysi-

ology, 5th edn (ed. G. C. Whittow), pp. 391–435. London:

Academic. Campbell, T. W. 1988 Avian hematology and cytology. Ames,

IA: Iowa State University Press. Canoine, V., Hayden, T. J., Rowe, K. & Goymann, W. 2002

The stress response of European stonechats depends on the

type of stressor. Behaviour 139, 1303–1311. Carsia, R. V. & Harvey, S. 2000 Adrenals. In Sturkie’s avian

physiology, 5th edn (ed. G. C. Whittow), pp. 489–537.

London: Academic. Cockrem, J. F. & Silverin, B. 2002 Sight of a predator can

stimulate a corticosterone response in the great tit (Parus

major). Gen. Comp. Endocrinol. 125, 248–255. Cucco, M., Ottonelli, R., Raviola, M. & Malacarne, G. 2002

Variations of body mass and immune function in response

to food unpredictability in magpies. Acta Oecologica 23,

271–276.

Proc. R. Soc. Lond. B (2004)

Hik, D. 1995 Does risk of predation influence population

dynamics? Evidence from the cyclic decline of snowshoe

hares. Wildl. Res. 22, 115–129. Hoi-Leitner, M., Romero-Pujante, M., Hoi, H. & Pavlova, A.

2001 Food availability and immune capacity in serin

(Serinus serinus) nestlings. Behav. Ecol. Sociobiol. 49, 333–

339. Jenni,L.&Jenni-Eiermann,S.1996Metabolicresponsestodiurnal

feeding patterns during the postbreeding, moulting and

migratoryperiodsinpasserinebirds.Funct.Ecol.10,73–80. Johnson, M. M. & Peters, J. P. 1993 An improved method to

quantify nonesterified fatty acids in bovine plasma. J. Anim.

Sci. 71, 753–756. King, A. S. & McLelland, J. 1984 Birds: their structure and

function, 2nd edn. London: Baillière Tindall. Kitaysky, A. S., Kitaiskaia, E. V., Piatt, J. F. & Wingfield, J. C.

2003 Benefits and costs of increased levels of corticosterone

in seabird chicks. Horm. Behav. 43, 140–149. Krebs, C. J. 1999 Ecological methodology, 2nd edn. Menlo

Park, CA: Benjamin Cummings. Krebs, C. J., Boutin, S., Boonstra, R., Sinclair, A. R. E.,

Smith, J. N. M., Dale, M. R. T., Martin, K. & Turkington,

R. 1995 Impact of food and predation on the snowshoe hare

cycle. Science 269, 1112–1115. Krebs, C. J., Boonstra, R., Boutin, S. & Sinclair, A. R. E. 2001

Conclusions and future directions. In Ecosystem dynamics of

the boreal forest (ed. C. J. Krebs, S. Boutin & R. Boonstra),

pp. 491–501. Oxford University Press. Lima, S. L. 1998 Nonlethal effects in the ecology of predator–

prey interactions. Bioscience 48, 25–34. McEwen, B. S. & Wingfield, J. C. 2003 The concept of

allostasis in biology and biomedicine. Horm. Behav. 43,

2–15. McFarlane, J. M. & Curtis, S. E. 1989 Multiple concurrent

stressors in chicks. 3. Effects on plasma corticosterone

and the heterophil:lymphocyte ratio. Poultry Sci. 68, 522–

527. McNamara,J.M.&Houston,A.I.1987Starvationandpredation

asfactorslimitingpopulationsize.Ecology 68,1515–1519. Moller, A. P. & Erritzoe, J. 2000 Predation against birds with

low immunocompetence. Oecologia (Berlin) 122, 500–504. Newton, I. 1998 Population limitation in birds. London: Aca-

demic. Nowicki, S., Peters, S. & Podos, J. 1998 Song learning, early

nutrition and sexual selection in songbirds. Am. Zool. 38,

179–190. Nowicki, S., Searcy, W. A. & Peters, S. 2002 Quality of song

learning affects female response to male bird song. Proc. R.

Soc. Lond. B 269, 1949–1954. (doi:10.1098/

rspb.2002.2124) Ots, I., Murugami, A. & Horak, P. 1998 Haematological

health state indices of reproducing great tits: methodology

and sources of natural variation. Funct. Ecol. 12, 700–707. Palkovacs, E. P. 2003 Explaining adaptive shifts in body size

on islands: a life-history approach. Oikos 103, 37–44. Palmer, A. R. 1994 Fluctuating asymmetry analyses: a primer.

In Developmental instability: its origins and evolutionary impli-

cations (ed. T. A. Markow), pp. 335–364. Dordrecht, The

Netherlands: Kluwer. Palmer, A. R. & Strobeck, C. 2003 Fluctuating asymmetry

analyses revisited. In Developmental instability: causes and

consequences (ed. M. Polak), pp. 279–319. Oxford Univer-

sity Press. Rogers, C. M., Taitt, M. J., Smith, J. N. M. & Jongeian, G. 1997

Nest predation and cowbird parasitism create a demographic

sink in wetland-breeding song sparrows. Condor 99, 622–633. Romero, L. M. 2004 Physiological stress in ecology: lessons

from biomedical research. Trends Ecol. Evol. 19, 249–255.

Chronic food and predator stress M. Clinchy and others 2479

Ruiz, G., Rosenmann, M., Novoa, F. F. & Sabat, P. 2002 Hema- tological parameters and stress index in rufous-collared spar- rows dwelling in urban environments. Condor 104, 162–166.

Sapolsky, R. M., Romero, L. M. & Munck, A. U. 2000 How do glucocorticoids influence stress responses? Integrating permissive, suppressive, stimulatory, and preparative actions. Endocrine Rev. 21, 55–89.

Scheuerlein, A., Van’t Hof, T. J. & Gwinner, E. 2001 Predators as stressors? Physiological and reproductive consequences of predation risk in tropical stonechats (Saxicola torquata axillaris). Proc. R. Soc. Lond. B 268, 1575–1582. (doi:10.1098/rspb.2001.1691)

Silverin, B. 1998 Behavioural and hormonal responses of the pied flycatcher to environmental stressors. Anim. Behav. 55, 1411–1420.

Smith, J. N. M., Taitt, M. J., Rogers, C. M., Arcese, P., Keller, L. F., Cassidy, A. L. E. V. & Hochachka, W. M. 1996 A metapopulation approach to the population biology of the song sparrow, Melospiza melodia. Ibis 138, 120–128.

Sokal, R. R. & Rohlf, F. J. 1995 Biometry, 3rd edn. New York: Freeman.

Totzke, U., Hübinger, A. & Bairlein, F. 1998 Glucose utiliza- tion rate and pancreatic hormone response to oral glucose loads are influenced by the migratory condition and fasting in the garden warbler (Sylvia borin). J. Endocrin. 158, 191–196.

Proc. R. Soc. Lond. B (2004)

Wingfield, J. C. & Silverin, B. 2002 Ecophysiological studies of hormone-behaviour relations in birds. Hormones Brain Behav. 2, 587–647.

Wingfield, J. C., O’Reilly, K. M. & Astheimer, L. B. 1995 Modulation of the adrenocortical responses to acute stress in arctic birds: a possible ecological basis. Am. Zool. 35, 285–294.

Wingfield, J. C., Vleck, C. M. & Moore, M. C. 1992 Seasonal changes in the adrenocortical response to stress in birds of the Sonoran Desert. J. Exp. Zool. 264, 419–428.

Zanette, L. 2000 Fragment size and the demography of an area-sensitive songbird. J. Anim. Ecol. 69, 458–470.

Zanette, L., Doyle, P. & Trémont, S. M. 2000 Food shortage in small fragments: evidence from an area-sensitive passer- ine. Ecology 81, 1654–1666.

Zanette, L., Smith, J. N. M., Van Oort, H. & Clinchy, M. 2003 Synergistic effects of food and predators on annual reproductive success in song sparrows. Proc. R. Soc. Lond. B 270, 799–803. (doi:10.1098/rspb.2002.2299)

As this paper exceeds the maximum length normally permitted, the

authors have agreed to contribute to production costs.