, :3, f .v . . . 3;}, r;
.3 if”? hi. ( .

. d. H

J : ~ ..
an; x. w it Quasi. w... . V
um. $3 4.3%. , . . . V . V . . _
V . . V, . 3;... .
m.

w. Wm. ”swan m. V .. 1.“ V V V ‘ . . V . . V V hmc .
_ . “Ev? . V, V V A H V . . V V V ..V_;...w,n “
V, . V . H . . . V. . . . A . . , . u . {kWW-he
. . . Q

«A.
5,»:
u.» n da.

I\ 3‘ V
3:53,? , . . . £11. £3 «.3: . .
gm . .. ﬁx} , , V

, v . A n 4|—

Bun)“;
V : ._ V
i... ..

R.
.

in. 5 V V
’ ¢ . , V
’3 .4 ‘ ‘ 1 h v
.. 2, . 3......J4xn.

a. , . , . . . git
, . 4 . I

, , , Jaw...
, . . .‘7‘.
. Err: . .
many...

.5
I a... .Lwo‘.
z...

V _
... {:5
I".V\v.urx.r:1...r.

J.
‘

«51¢:th 4.x“: A fiuv...
.3... v . ,y r» ..
J. 3.33....rr :3! 1 .5: V
in... é 4.0... 52a...

 

 

. $21.83;: .. V .
. ‘ 7....

r

.2! £31. 5?qu 5.,

 

This is to certify that the

dissertation entitled

THE INTERACTION OF NEST PREDATION AND BROOD
PARASITISM: FACTORS AND CONSEQUENCES OF
RE-NESTING IN SONGBIRDS

presented by

Steven J. Vieira
has been accepted towards fulﬁllment

of the requirements for

Ph . D. degree in 7.001 ngy

WW

Major professor

Date 3 W 0?ng

MS U is an Afﬁrmative Action/Equal Opportunity Institution 042771

 

 

LEBHARY

Michigan State

University

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE PUF- q

DATE DUE

 

JUL 2 1 2004

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 chlRC/DateDuepGS-p. 15

THE INTERACTION OF NEST PREDATION AND BROOD PARASITISM:
FACTORS AND CONSEQUENCES OF RE-NESTING IN SONGBIRDS

BY

Steven J. Vieira

A DISSERTATION

Submitted to
Michigan State University
In partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Zoology

2001

ABSTRACT
THE INTERACTION OF NEST PREDATION AND BROOD PARASITISM:
FACTORS AND CONSEQUENCES OF RE-NESTING IN SONGBIRDS
By

Steven J. Vieira

Most research of the breeding biology of songbirds fails to explicitly account for
the inﬂuence of re-nesting after nest predation. Not only can re-nesting ameliorate the
effects of nest predation but it can also lead to an unusual effect where nest predation
enhances seasonal fecundity (relative to when nest predation is absent) if brood
parasitism is frequent. This is because when nest predation occurs on parasitized nests, it
acts as a rejection mechanism for cowbird acceptors. A dynamic quantitative model is
introduced that can simulate the nesting cycle. of many species of single-brooding
songbirds, and is used to explore this interaction between nest predation and brood
parasitism. Simulation results conﬁrm that a signiﬁcant interaction between nest
predation and brood parasitism can exist such that at high parasitism frequencies,
intermediate levels of nest predation led to small positive changes in seasonal fecundity.
This increase is enhanced when the cost of parasitism is severe, there is greater nest
predation on parasitized nests, and the risk of parasitism and predation declines through
the breeding season. In general, nest predation ranging from zero to as high as 60% does
not reduce seasonal fecundity when moderate to high parasitism (>60%) exists in the

breeding population.

A review of the literature and ﬁeld results of a monitored population of Yellow
Warbler conﬁrm that all the special conditions required to observe this beneﬁt to nest
predation could occur in a natural population. However, it remains unclear if all
necessary conditions would occur collectively in a single population. Moreover, repeated
nest predation on some individual bird’s nests decreases the ability of re-nests to recover
losses from nest predation. Field observations and ﬁeld experimentation comparing nest
success of actual nests with artiﬁcial nests suggests that some individuals, namely
younger parents, occupy lower quality territories that increase the chance of nest failure.
This leads to an overall greater failure rate for re-nests compared to initial nest attempts,
and eliminates the potential for a beneﬁt to nest predation regardless of other conditions.
Since many populations Show age-speciﬁc variation in nest success, it is concluded that a
beneﬁcial effect to nest predation is unlikely to occur often in natural populations.
Regardless, the complex interaction of nest predation and brood parasitism with various
other factors such as age and habitat quality has signiﬁcant management consequences.
Proper management strategies will depend greatly on an accurate assessment of these
interactions that will unquestionably vary inter- and intra-speciﬁcally both spatially and
temporally. Further use of advanced simulation modeling will greatly improve the ability
of managers and researchers to efﬁciently conserve avian populations, which in many
cases have been declining for several decades. To improve these models, future ﬁeld
research will have to focus more on seasonal fecundity instead of nesting success by

explicitly accounting for the re-nest attempts of individual birds.

ACKNOWLEDGEMENTS

This project was handed by the Department of Zoology, the Ecology, Evolution,
and Behavioral Biology (EEBB) program at Michigan State University, and the Wallace
Memorial Research Scholarship. I acknowledge the support and guidance of my
graduate committee: Drs. Donald Beaver, Thomas Getty, Kay Holekamp, and Scott
Winterstein. I especially thank my graduate advisor Dr. Beaver, for all his support and
encouragement throughout these years. Dr. J ianguo Lui’s instruction and guidance was
instrumental in the development of the original STELLA® model. I also thank Dr.
Joseph Grzybowski, Stephen Rothstein and two anonymous reviewers for their
constructive criticism of a previous manuscript concerning the model. I thank my peers
in the Zoology Department and EEBB program at MSU. I especially thank, Stacy Epp
and Mary Martin for their services in the ﬁeld and Natalie Dubois for her help with the
model. I am forever, indebted to several undergraduate students who volunteered to
assist me in the collection of ﬁeld data while never complaining about the early
mornings, the weather, or even the abundance of mosquitoes and poison sumac. These
people include, Audrey Calkins, Kevin Swanson, Kristin Hasselbring, Beth Apple, Lori
Polkowski, Derek Freridge, and Todd White. I thank the Michigan Department of
Natural Resources for permitting use of their land. Finally, a special thank you to my
wife Lori, for her patience, love, support, and sacriﬁces during these years. This
dissertation is dedicated to my parents who have always encouraged me in my interest
with biology and to my high school Biology teacher Roger Dubois who was my mentor

as a novice biologist exploring local wetlands.

iv

TABLE OF CONTENTS

LIST OF TABLES .................................................... vii
LIST OF FIGURES ................................................... x
CHAPTER 1- Introduction ............................................. 1
Brood parasitism ............................................... l
Nest predation ................................................. 4
The interaction. ................................................ 6
CHAPTER 2- Field Methods ............................................ 13
Field sites ..................................................... 13
Focal Species .................................................. 17
Nest Monitoring ................................................ l9
Vegetation Assessment. ......................................... 25
Color Banding ................................................. 28
Statistical Analysis .............................................. 30
CHAPTER 3- Field Results ............................................. 33
F ecundity and Life History ...................................... .. 33
Nest Placement. ............................................... 41
Parental age ................................................... 56
Parental Behavior ............................................... 64
Summary ..................................................... 67
CHAPTER 4- The Model .............................................. 68
Model structure ................................................ 69
Assumptions and limitations ...................................... 75
Validation .................................................... 78
Simulations experiments ......................................... 82
Overall pattern ........................................... 84
Declining parasitism trend. ................................. 87
Changes in nest predation patterns ........................... 88
Cost to parasitism ........................................ 92
Nest abandonment ........................................ 94
Season length and re-nesting frequency ....................... 99
Sensitivity Analysis ............................................. 103
Discussion. ................................................... 105
Biological signiﬁcance of ﬁndings ........................... 107
Conditional simulations .................................... 109
Differential predation. ............................... 110
Declining predation ................................. 1 12

Cost to parasitism .................................. 113

Nest abandonment. ................................. 114

Season length and re-nest frequency .................... 117

Comparison with existing models ............................ 118

Conclusions and Implications for Management and Research ...... 122

CHAPTER 5- Artiﬁcial nest experiment ................................... 125
Introduction ................................................... 125
Individual based-hypotheses ................................ 125

Predator behavior based hypotheses .......................... 129

Experimental design ............................................ 130
Results ....................................................... 133
Discussion .................................................... 140
Improvements and future research .................................. 144
CONCLUSIONS AND RECOMMENDATIONS ........................... 146
APPENDIX ......................................................... l 5 1
LITERATURE CITED ................................................ 153

vi

Table
2-1

2-2

2-3

3-3

3-6

LIST OF TABLES

Breeding Bird Survey abundance trends for Yellow Warbler and
Common Yellowthroat from 1966 to 1998 (Sauer et a]. 2000). Trends
were estimated using the route-regression method (Geissler and Sauer
1990).

Criteria for elusiveness rating of adult YWAR at the nest when the
observer was nearby .

Categories and number of nest failures for 347 monitored YWAR nests.

Categories of vegetation that were quantiﬁed in vegetation assessment
of a 2 meter radius around each nest.

Summary of the numbers of females, nests, and young ﬂedged by year
for all four sites combined . . . . . . . . . . .

Estimated cumulative young ﬂedged per female by nest attempt. First
nest attempts contributed the most to seasonal fecundity. Estimates of
young ﬂedged were determined according to the total number of young
ﬂedged up to that nest attempt. Values in parentheses are proportional
contributions to total fecundity .

Comparison of fecundity, predation, and parasitism by site, all years
and attempts combined. Letters denote signiﬁcant differences
according to Chi-square tests .

Comparison of marginal means for three measure of nest concealment
between parasitized and non-parasitized nests .

The distribution of nest plants in a random sample of non-use sample
sites compared to actual nest plants reveals that YWAR preferred
meadowsweet and rose, with meadowsweet being the most common
shrubby plant. See text for further discussion .

Nest plant preferences did not change signiﬁcantly between nest
attempts (Pearson’s X2 = 8.22, df = 8, p = 0.412). However, a shiﬂ
occurs toward a greater proportion of later nests placed in dogwoods .

Nests of adults, who had bred at least once before, ASY, were less
likely to be depredated, than ﬁrst time breeders, SY adults .

Comparisons of nest site characteristics by age of females Show that

vii

Page

18

22

23

27

33

34

38

42

49

53

56

3-10

4-7

4-8

4-9

nests of ASY females had more side cover, shorter tape distance, and
fewer trees and snags than SY females .

Comparisons of nest site characteristics by age of males Show that nests
of ASY males had more side cover, shorter tape distance, greater stem
density, and fewer snags than SY males .

Comparisons of three measures of parental behavior by nest stage
revealed that only parental elusiveness during the nest building was
linked to nest predation. See Chapter 3 for an explanation of each
variable. (A Bonferroni corrected critical value for signiﬁcance is p =
0.00556) .

List of model parameters necessary to run a simulation .

Estimates of parameter values (5: SD) used in model validation from
Michigan YWAR ﬁeld data pooled for all years .

Mean i SD of the number of young ﬂedged per successful YWAR nest
attempt (ft), by site, years combined. Estimates of clutch size were
applied to all three sites .

Weekly predation rates by site .

Summary of standard conditions for critical parameters used in
simulation experiments .

Summary of simulation experiments performed using PLEDGE. The
tested condition is the main parameter of interested. Values used for
this main parameter are provided. Alternate factors are other
parameter(s) that were also altered or differed from standard conditions.
A series of simulations for the set of main parameter values was
repeated for each level of an alternate factor .

Estimates of the minimum cost to parasitism for a positive change in
fecundity to exist at a 30% predation relative to 0% predation.
Estimates are given for two levels of parasitism and differential
predation. Values were calculated by ﬁnding the x-intersection from
regression equations derived in SYSTAT .

Approximate values of the cost of parasitism (c) necessary before an
average female will gain in fecundity from abandoning a parasitized
nest for 16 combinations of predation and parasitism rates .

The change in mean young ﬂedged per female between a season length
of 40 and 70 days for three levels of parasitism. .

viii

62

63

72

79

79

79

83

85

94

97

100

4-10

4-11

5-1

5-2

5-3

The top eight most inﬂuential parameters according to frequency range.
Parameter elasticity varied depending on the level of predation, d(t) and
parasitism, p(t). Therefore, parameters are listed by Speciﬁc levels for
d(t) and p(t) .

Key comparisons of parameterization and assumptions between three
existing models for predicting seasonal fecundity .

A summary of hypotheses to explain the greater nest predation rate on
re-nests including the aspects of the study used to evaluate them, and
predictions if supported. .

Means (:t SE) of tape distance and overhead cover were signiﬁcantly
different between dummy nests that were successful versus depredated.

In all cases successful nests were more concealed .

Matched pairs Wilcoxon’s signed rank tests found that only the

coverage of dead plant matter and abundance of large trees signiﬁcantly

changed between ﬁrst and second nest attempts within individuals.
Sample sizes for ﬁrst, second, and third attempts were 11 = 158, 76, and
25, respectively .

ix

106

119

126

138

141

Figure
1-1

2-1

2-2

3-1

3-2

3-3

3-5

LIST OF FIGURES

The nest of a single-brooding songbird can follow many pathways
depending on several internal and external factors as described here .

Map of lower Clinton and Shiawassee Counties in Michigan showing
the locations of the two Rose Lake State Game Area ﬁeld sites. Circled
numbers indicate the locations of the sites. 1) Site 1: R-S, and 2) Site 2:
R-R. .

Map of central Ingham County in Michigan showing the locations of
the two Dansville State Game Area field sites. Circled numbers
indicate the locations of the sites. 1) Site 1: D-W, and 2) Site 2: D-E. .

The frequency at which replacement nests are initiated following nest
failure is very high in the ﬁrst month of nesting, but declines rapidly in
the second month (mid-June). Numbers represent the total number of
nests that could potentially re-nest in an interval.

Figure. Mean clutch and brood size declined signiﬁcantly with nest
attempt.

Comparison of nest predation estimates between non-parasitized and
parasitized nests. Parasitized nests had a 50% greater risk of predation
(X2 = 12.06, p = 0.0005). Sample sizes are placed below standard error
bars. Sites and years are pooled. . . . . . . . .

Mayﬁeld estimates of nest predation were consistently higher for re-
nests than for initial nest attempts, though not Signiﬁcant within years.
Error bars denote standard error .

Three independent indices of nest concealment Show that successfully
ﬂedged nests were more concealed than depredated nests. Error bars
are 95% conﬁdence intervals .

All three indices of nest concealment reveal no signiﬁcant difference
between nest attempts, though a slight downward trend is evident in all.

Error bars are 95% conﬁdence intervals .

Successfully ﬂedged nests had more overhead cover than depredated
nests .

Nests placed in rose had signiﬁcantly lower nest predation than nests

Page

\1

14

15

20

36

39

40

43

45

46

3-9

3-11

3-12

placed in meadowsweet (spirea) or dogwoods (X2 = 4.48, p = 0.034).
Error bars denote standard error. Data are pooled for years, sites and
attempts .

Three comparisons of percent coverage of surrounding vegetation in a
2 meter radius around the nest: Fledged versus depredated nests, by
nest attempt, and non-parasitized versus parasitized nests. The percent
cover of rose between ﬂedged and parasitized nests was signiﬁcantly
different in a logistic regression analysis (t-ratio 2.612, p = 0.009, odds
1.022). A logistic regression with three possible outcomes for the
dependent variable (nest attempt) found the change in meadowsweet
(spirea) to be signiﬁcantly different (t ratio 2.366, p = 0.019) with
respect to nest attempts. A non-Signiﬁcant trend toward re-nesting in
patches with more dogwood is notable. Only the abundance of open
water was a signiﬁcant predictor of parasitism (t ratio= 2.002, p =
0.045, odds 1.019) .

Logistic regression analysis comparing differences in surrounding
vegetation characteristics of non-use samples found that territories of
depredated nests were composed of a greater proportion of grass and
sedge and had less rose (p = 0.048, and p = 0.040, respectively).
Territories of depredated nests also have more dogwood and less
meadowsweet (spirea), but means were not signiﬁcantly different .

Logistic regresSion analysis of differences in surrounding vegetation
coverage between active nest sites (use) and non-use plots suggests that
YWAR select nest sites with more meadowsweet (spirea) and less
herbaceous plants (*, p < 0.05; **, p < 0.01; or ***, p < 0.001). The
difference in coverage of rose was marginally insigniﬁcant (p > 0.108).

A dichotomous logistic regression of surrounding vegetation cover
between ﬁrst nest attempts and re-nests found the variation in coverage
of dogwood/willow to approach signiﬁcance (* p < 0.10) and found the
difference in percent cover of forbs, and dead plant matter to be
signiﬁcantly different (**, p < 0.05) .

Predated and parasitized nests had more trees and snags nearby and
later nests tend to be near more large trees and snags than earlier nest
attempts. Compared are number of trees by category: small tree-
smltree, large trees- lgtree, and snags, in a 10 m radius around the nest
for: ﬂedged versus predated nests, by nest attempt, and non-parasitized
versus parasitized nests. Kruskal-Wallis One-way Analysis of
Variance tests were used to test for signiﬁcant differences. Asterisks
signify level of signiﬁcance C“, p < 0.05; **, p < 0.01; or ***, p <
0001)

xi

49

50

52

52

54

55

3-13

3-15

3-16

3-17

4-3

SY males and females had signiﬁcantly higher Mayﬁeld estimates of
nest predation compared to their ASY counterparts (X2 = 7.119, p=
0.008). Error bars are 95% conﬁdence intervals; Sample sizes are
included.

The proportion of SY aged adults in re-nests increased regardless of
sex. However, the increase was not signiﬁcant in either case.
Pearson’s Chi-Square results: X2 = 1.97, df = l, p = 0.16 and X2 =
0.98, df = 1, p = 0.322 for females and males respectively. However,
an analysis of both sexes combined was marginally signiﬁcant (X2 =
3.455, p = 0.063) .

Nests of SY adults of both sexes were parasitized in greater proportions

than their ASY counterparts. However, Chi-square tests were not

SigILIﬁcant in either sex (males: X2 = 2.48, df = l, p = 0.115; females:
1.2,—,=..6df1p0262)

The percent cover of meadowsweet (spirea) was signiﬁcantly greater in
ASY females in comparisons of surrounding vegetation coverage by
age and sex. Signiﬁcance was determined using logistic regression (*
p = 0.05) .

Two-way contingency table analyses between age and nest plant
preference found no signiﬁcant difference for females and males (X2=
1.,073 df= 3, p= 0.784 and x2: 0.263, df= 3, p= 0.,967
respectively). . . . . . . . . . . .

The distribution of number of nests depredated by nest days revealed
that most YWAR nests were depredated during the ﬁrst 10 days of the
nest cycle regardless of attempt number . . . . . . . .

Model predictions of seasonal fecundity per site (open circles with
95% conﬁdence intervals, n = 10) slightly overestimate actual ﬁeld
data estimates measured at each site (dark circles) except DE. This is
signiﬁcantly different only at R and DW .

A surface plot of the mean number of young ﬂedged per female
revealed that predation and parasitism have an interaction. The rate of
decline in fecundity as either factor increases in ﬁ'equency varies with
respect to the alternate factor such that the decline in fecundity due to
predation is reduced as parasitism increases and vice-versa. A DWLS
smoothing operation was used to produce this plot using SYSTAT 8.0
(SPSSInc.1998). . . . . . . . . . . .

This ﬁgure compares fecundity estimates versus nest predation for a
population for which the frequency of parasitism does not change

xii

57

58

58

6O

61

66

80

86

4-5

4-6

4-8

4-9

4-10

throughout the season (solid points) with a one where parasitism
declines as the season progresses (line with 95% conﬁdence intervals,
n = 10). Only at very high frequencies of parasitism and predation do
these two population differ signiﬁcantly in fecundity .

Increasing the proportionate difference in predation (8) of parasitized
nests relative to non-parasitized nests generally results in signiﬁcant
but small linear increases in fecundity regardless of the parasitism, p(t),
or predation, d(t) rates. This trend reverses only when both rates are
high. Regression coefﬁcients (B) and signiﬁcance values are provided .

Shown here is the change in fecundity relative to zero nest predation at
two levels of parasitism and nest predation, d(t), when 8 = 0.75. The
change in fecundity is positive for all four combinations of predation
and parasitism. This change is greater at a low predation rate than at a
moderate level and is signiﬁcantly greater than zero only at 80%
parasitism. Error bars are 95% conﬁdence intervals (n = 10).

Positive mean changes in fecundity per female relative to zero
predation are enhanced by a declining predation rate and when 6 is
greater than zero. The magnitude of this change is greater at high
parasitism rates, p(t). Furthermore, the range of predation at which a
positive or no change in fecundity occurs is greater compared to
previous results where predation was held constant through time. Error
bars denote 95% conﬁdence intervals .

The relative change in total fecundity (between 0% and 30% predation)
is positively related to the percent cost of parasitism (c). The point
where regression lines (with 95% CI’s) cross the x-axis denotes the
level of cost to parasitism that must exist for a positive change in
fecundity to occur. This point declines with increased parasitism rate,
p(t) and differential predation (8). See Table 2-7 for x-intercept values.

The mean change in fecundity (between 0% and 30% predation) at
80% parasitism is not strongly inﬂuenced by the abandonment rate
except at very high abandonment. Results are given for two levels of
cost to parasitism (c), and show that although the cost can inﬂuence the
change in fecundity, this pattern is consistent with respect to the
abandonment frequency .

Nest abandonment positively changes fecundity relative to zero
abandonment when the parasitism rate, p(t), is lower and the cost to
parasitism, c, is higher, but has negative effects at high parasitism rates
and/or lower costs to parasitism. These results are for 50% predation. .

All else equal, increasing the length of the breeding season reduces the

xiii

89

9O

91

93

95

96

98

4-11

4-12

5-1

5-2

negative effect of nest predation, d(t), on the change in seasonal
fecundity relative to zero predation. Results are for 80% parasitism .

The relationship between nest predation and seasonal fecundity shifts
from curvilinear to a linear decline as the initial re-nest frequency, r(t),
declines. Results are shown for two levels of parasitism, p(t) .

Any positive change in seasonal fecundity at 30% predation relative to
zero predation quickly dissipates when the re-nest frequency declines.
The x-intersection indicates the minimum initial level of r(t) necessary
for a zero or positive change in fecundity to exist at d(t) = 0.3. A third-
order polynomial was ﬁtted to the data to generate the line from which
the x-intersection was calculated .

Mayﬁeld estimates of overall nest predation were nearly identical
between paired dummy and active nests (n = 81) (X2 = 0.0032, p =
0.96). Error bars denote standard error. . . . . . . .

Dummy nests on territories of successful active nests (good, n = 42)
had signiﬁcantly lower predation than dummy nests on territories of
active nests that failed (bad, n = 39) .

Nest predation rates of dummy nests introduced in the ﬁrst 25 days
(early) did not differ from dummy nests introduced in the latter half of
theseason. . . . . . . . . .

A linear regression analysis found no signiﬁcant increase in nest
predation in relation to nest abundance. Points represent the predation
rate for a single site by week (years pooled). The linear regression line
and equation are included .

YWAR nests in both species of dogwood are signiﬁcantly less
concealed than nests in meadowsweet (spirea) or rose. These nests also
have signiﬁcantly lower stem density surrounding the nest than
meadowsweet but not rose. Signiﬁcance was determined from Fisher’s
LSD, and is denoted by letters. Error bars are standard error .

Shown here is the total percent cover of all woody shrub vegetation for
non-use samples compared with age of female YWAR. ASY females
had marginally signiﬁcantly more shrubs on their territories than SY
females. . . . . . .

xiv

101

102

103

135

135

136

137

139

143

CHAPTER 1- INTRODUCTION

Brood parasitism by the Brown-headed Cowbird (Molothrus ater) and predation
on eggs and nestlings are widely implicated in the decline of several Neo-tropical migrant
bird species (Terborgh 1989, Askins et al. 1990, Martin 1992a, Robinson et al. 1995a).
Habitat fragmentation has been linked to increased frequencies of both mortality factors
(Gates and Gysel 1978, Brittingham and Temple 1983, Wilcove 1985, Robinson 1992).
Because of extensive fragmentation of the North American forest landscape, moderate or
high frequencies of nest predation and brood parasitism exist together in many songbird
populations. Most ﬁeld studies that have attempted to evaluate the impact of these two
mortalities have used nest success as an index to overall reproductive success (Pease and
Grzybowski 1995). Nest success does not account for replacement nests (re-nests) that
typically follow a nest failure. Nor do measures of nest success explicitly identify how
nest predation and brood parasitism interact. Only very recently have we begun to
examine how these two causes of nest mortality interact to inﬂuence seasonal fecundity
in songbirds (Pease and Grzybowski 1995, Schmidt and Whelan 1999). The way in
which songbirds respond to these nest mortalities can be quite different and can lead to
interesting interactive effects. The aim of this dissertation is to discuss the interaction of
nest predation and brood parasitism and the effect of this interaction on seasonal

fecundity for Neo-tropical migrant songbirds that typically ﬂedge one brood in a season.

Broodparasitism- Interspeciﬁc brood parasitism is when one species of bird, the parasite,
lays its eggs in the nest of another species, the host, who then raises the parasite’s young,

often at the expense of its own young. There are at least 85 known obligate brood

1

parasites in birds (May and Robinson 1985). Most brood parasites, such as those
belonging to the family Cuculidae (Cuculifonnes), are host speciﬁc, specializing on a
single host or small subset of species. However, several members of the family Icteridae
(Passeriforrnes) known as cowbirds will parasitize multiple hosts, of which the Brown-
headed cowbird is the champion.

The Brown-headed Cowbird is the principle brood parasite of small terrestrial
birds in North America. It is unusual among brood parasites in that it is known to
parasitize a large number of bird species. Some estimates are as high as 216 host species
(Robinson et al. 1995b). In general, brood parasitism by cowbirds has a three-fold
negative impact on the fecundity of the host (Rothstein 1990). First, female cowbirds
commonly remove a host egg when laying an egg. Second, cowbird eggs are often larger
than the host’s eggs and may hatch sooner thereby disrupting proper incubation of the
remaining eggs. This may lead to an increase in hatching failure in host clutches.
Finally, the cowbird hatchling has a voracious appetite and will often out-compete other
nestlings for food provided by parents. Many host nestlings starve to death in parasitized
nests. Unlike the cuckoos of Europe and Africa, cowbirds nestlings rarely eject host eggs
and young from nests (Robinson et al. 1995a).

Variation in the response and performance of the host does exist among bird
populations. Birds which have historically shared overlapping ranges with the Brown-
headed Cowbird have evolved adaptive responses to brood parasitism. These include
ejection of the cowbird egg, nest abandonment followed by a re-nest attempt, or building
a new nest over the parasitized nest (Robinson et al. 1995a; Rothstein 1990). However,

during the last two centuries the Brown-headed Cowbird has undergone a rapid range

expansion (Mayﬁeld 1965). This has led to many new populations of songbirds
becoming exposed to brood parasitism. Most of these songbird populations have not
evolved an adaptive response to brood parasitism (Brittingham and Temple 1983,
Robinson et al. 19953). Host species are categorized as either a rejector or acceptor
depending on how individuals respond to the presence of a cowbird egg in their clutch.
The majority of North American species are categorized as acceptors (Rothstein 1975).
There is also considerable variation in performance when parasitized among acceptor
species. Some species are capable of raising a cowbird nestling with minimal losses to
their brood (Wolf 1987, Eckerle and Breitwisch 1997) while others suffer severe
reductions in ﬂedgling success (Goldwasser et al. 1980, Walkinshaw 1983). Many small
Neo-tropical migrant songbirds experience elevated host young mortality from brood
parasitism. This is because the cowbird nestling is often larger than the parents and
occupies the majority of space in the nest. Consequently, many small hosts cannot
provide enough food for the aggressive cowbird nestling and their own brood, which is
oﬁen concealed by the cowbird nestling (for examples see Nolan 1978, Rothstein 1990,
Robinson et. al. 1995a).

Many Neo-tropical migrants must also deal with another problem. Because they
must allocate time to migrate long distances, most only have time to successfully raise a
single brood in a short breeding season. Birds that can raise only a single brood will not
lay an additional clutch after they have ﬂedged young. If a cowbird successfully ﬂedges
from the nest, the host parents typically behave as if they have produced a successful
clutch, even if none of their own young survived. Clearly, the potential cost in terms of

seasonal fecundity for parasitized individuals can be great for Neo-tropical migrant

songbirds. It is for this reason that cowbird parasitism has been implicated in the decline
of many migrant songbirds (Brittingham and Temple 1983). It is members of this group
of birds that are the subject of the remainder of this dissertation.

The frequency at which nests become parasitized varies greatly with host species,
region, habitat, and landscape. For most host species, brood parasitism occurs in less
than 50% of nest attempts. However, brood parasitism rates have been regularly
observed to be quite high. For example, bird populations nesting in southern Illinois have
consistently experienced parasitism rates as high as 80% of all active nests (Robinson et.
al. 1995b). Instances of even higher parasitism frequencies can be found in the literature,
and in ﬁagmented forests, it is not uncommon to ﬁnd frequencies of brood parasitism
greater than 50% for many preferred hosts (Robinson et al. 1995a). With natural .
frequencies this high, there is a potential for large reductions in seasonal productivity in
some populations. The ultimate cost of brood parasitism on a population will depend on
the life history of the species, how frequent brood parasitism occurs in the population,

and as we will see, the interaction with nest predation.

Nest predation- In contrast to brood parasitism, nest predation typically results in
loss of the entire brood. Nest predation is the removal or destruction of the contents of a
nest by another organism. Nearly every breeding bird population experiences some
degree of nest predation. In North America many organisms are known to depredate the
nests of birds including many mammals such as raccoons, skunks and several rodents,
reptiles such as snakes, and other birds, especially corvids (Corvidae: crows and jays).

When a nest becomes depredated many birds will quickly respond by constructing a

replacement nest and laying another clutch. This is the case even for those birds that will
only raise a single brood. In addition, if this re-nest is also depredated some birds may
attempt another clutch and so on. The signiﬁcance of this is that birds are able to recover
most of the loss in fecundity from nest predation.

The total reproductive cost to nest predation will depend on how often a host will
re-nest aﬁer failure. The potential to re-nest varies among populations, though ﬁeld
observations are surprisingly few in the literature (Martin and Li 1992). Nolan (1978)
observed one female Prairie Warbler (Dendroica discolor) attempt eight replacement
nests in a single season. Yet, in another study of Tree Swallows (T achycineta bicolor),
fewer than 50% attempted a single replacement nest, and of those that attempted
replacements, only 12% made a third attempt (Rooneem and Robertson 1997). The
probability that a bird will re-nest depends on a number of factors including the length of
time available in the breeding season and the availability of resources to accommodate
additional egg production. Nolan (1978) observed that the percentage of ﬁrst nesting
attempts that were replaced was constant and high through June (average of 89%) but
rapidly declined to zero by mid-July; only one female attempted a new nest after July 1.
This result suggests that there are trade-offs between producing a successful brood and
other needs, such as preparing for the fall migration, molting, or even adult survival. A
short-term beneﬁt in fecundity by re-nesting may impose a higher cost in an individual’s
lifetime fecundity by increasing adult mortality (Linden and Moller 1989). It is also
plausible that survival of young ﬁ'om later nests is lower, making it less advantageous to
attempt additional nests.

There is usually a reduction in clutch size from one nest attempt to the next

(Ricklefs 1973, Nolan 1978, Rooneem and Robertson 1997). Yellow Warblers
(Dendroica petechia) breeding in Michigan lay about one less egg per nest attempt. The
magnitude of this clutch reduction is likely to vary between species and region. The size
of the brood at ﬂedging is also likely to be smaller by the same amount. The greater the
proportionate reduction in brood size the greater the cost of nest predation. If we ignore
any potential long-term costs to re-nesting such as lower ﬂedgling survival of young from
re-nests or lower adults survival, then the reduction in clutch size is the only immediate

cost to nest predation in terms of seasonal fecundity for a single-brooding songbird.

The Interaction- This re-nest response can have important consequences when
considering how nest predation and brood parasitism interact to inﬂuence seasonal
fecundity. In most North American songbird populations, brood parasitism and nest
predation can occur in the same nesting event. The typical events that can occur during a
breeding season are summarized in Figure 1-1. If a bird's nest is parasitized, the bird can
either care for the clutch, which could result in raising fewer young than normal, or
terminate this nest attempt and start a new clutch. As Figure 1-1 depicts, nest predation is
a plausible means of escaping brood parasitism for those songbirds that have not evolved
a rejection mechanism. If re-nesting following nest predation ﬂedges more young on
average than would have ﬂedged from the parasitized nest had it not been depredated
then nest predation can be interpreted as enhancing fecundity for the host. It is important
to recognize that this is only a relative beneﬁt and is contingent upon the failed nest being

parasitized and having a low potential for ﬂedging many host young. Simply stated, nest

 

Parasitism
frequency

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Start
01'
Renest
Renest Non-parasitized
Frequency nest cycle *
Parasitized
nest cycle
. Y t
«'4 N
Re-nest T- Y
N .....
Predation
Abandonment
C Stop nesting ) Fledge
Reduced Brood
Fledge
Full brood

Brood reduction for re-nests

 

Figure. 1-1. The nest of a single-brooding songbird can follow many pathways depending
on several internal and external factors as described here.

predation, in this case, is the lesser of two evils. I use the term beneﬁt to describe the
result for lack of a better term.

When a bird is unlikely to ﬂedge any host young when parasitized, it is obvious
that nest predation can improve their chance of successfully ﬂedging young by means of
re-nesting. It is much less obvious if nest predation can have this result in an entire
population, since non-parasitized nests will also be depredated and re-nests can be
parasitized or depredated. Two conditions are necessary for a beneﬁt to nest predation to
be possible in a songbird population. First there must be a signiﬁcant level of brood
parasitism since only these nests stand to gain from nest predation. In addition, re-nests
must ﬂedge more young on average than parasitized nests.

Even if these conditions are met, whether an actual beneﬁt to nest predation is
likely to occur greatly depends on the success of re-nest attempts. A re-nest is also
susceptible to nest predation and parasitism. If repeated nest predation occurs, the bird
could run out of time and energy to re-nest and fail to raise any young. There are several
plausible means by which this may occur. In some cases the frequency of nest predation
may simply increase later in the season (Beaver 197 5), resulting in greater nest predation
on re-nests. Sonerund and F j eld (1987) found evidence that some avian nest predators
may memorize the locations of previously discovered nests and repeatedly return to the
same area, leaving re-nests more susceptible to nest predation. It is also possible for re-
nests to be less successful relative to the average original nest attempt simply because
those birds that re-nests are inherently less capable of rearing a successﬁrl nest.

However, there are also reasons to expect lower nest predation for re-nests. There

is evidence to suggest that songbirds improve nesting success with experience (Saether

1990). Therefore, it is reasonable to expect that with each failed attempt birds gain
experience to improve their chance of success in a subsequent nest attempt. Some also
suggest that nest predation is density-dependent (Martin 1993b). Since some nests will
succeed, the number of active re-nests is likely to be lower than the original number of
nests. This temporal change in nest density should correspond to lower nest predation for
re-nests if is density-dependency exists. While support varies between studies, nest
concealment is believed to be an important characteristic of a successful nest. All else
being equal, re-nests should be more concealed since vegetation in most habitats
continues to grow during the season. For example, I commonly observed exposed small
shrubs become completely concealed by fast growing tall grasses.

Re-nests that are initiated while cowbirds are still actively laying eggs remain at
risk to brood parasitism. Most Brown-headed Cowbirds have an egg-laying season of
approximately 10 weeks with most eggs laid in May and early June; very few cowbirds
continuing to lay eggs into July (Nolan 1978, Wolf 1987, Brown 1994, Robinson et al.
19953 and references therein). However, for many songbirds nearly all nesting occurs
during this same period. Therefore, when the frequency of brood parasitism is high the
risk of re-nests becoming parasitized is also high. However, it is still possible for the risk
to brood parasitism to decline. Egg-laying by cowbirds has been shown in many studies
to peak early in the breeding season (early June) followed by a subsequent drop in
frequency as the season progresses. For example, Hofslund (1957) found that most
parasitism was concentrated in earlier nests of the Common Yellowthroat (Geothlypis
trichas) and most re-nest attempts were free of cowbird eggs. If this trend is common we

can expect that re-nests should experience lower brood parasitism rates.

Differences in nest predation frequency between non-parasitized and parasitized
nests are also important to the outcome of a predation-parasitism interaction on a
songbird population. There will be a greater relative cost to nest predation regardless of
the success of re-nests, if nest predation is more likely to occur on non-parasitized nests
than parasitized nests. Likely reasons for this include temporal or spatial differences in
the risk to nest predation and brood parasitism, or nest predation by cowbirds. For
example, while both nest predation and brood parasitism are widely known to increase in
frequency in fragmented landscapes in eastern North America, the same is not true in the
west. Tewksbury et al. (1998) have shown that western avian communities in fragmented
habitat have greater rates of parasitism, but nest predation occurs more often in un-
fragmented habitat. This suggests that some western species have lower nest predation
on parasitized nests relative to non-parasitized nests.

At a smaller scale, increased predation on non-parasitized nests may be the direct
result of cowbird behavior. Arcese et al. (1996) have hypothesized that adult cowbirds
will depredate non-parasitized nests that have completed egg-laying in order to be able to
parasitized the re-nest attempt. Also, Robinson (1992) found that nests placed on or near
the ground had lower brood parasitism frequencies, while Ricklefs (1973) suggests that
ground nests have the highest risk to predation. Thus, cowbirds may select host nests that
are more likely to avoid nest predation (Tewksbury, pers com.). If these patterns are
real, nest predation is likely to occur more often for non-parasitized nests. However,
considerable disagreement exists in whether such trends are common. Several studies
have shown that lower nests are actually more often parasitized (Newman 1970, Smith

1981, DellaSalla 1985, Briskie et al. 1990). And while direct evidence for Arcese et al.’s

10

hypothesis exists (Elliott 1999), no supporting evidence has been found beyond the
unique conditions of their study area, which has few natural nest predators.

The relative frequency of nest predation on non-parasitized and parasitized nests
is likely to depend on the predator community. Cowbirds are known to most often locate
nests by monitoring adult movements (Norman and Robertson 1975). On the other hand,
a diverse community of predators is likely to locate nests through a variety of means
including adult activity, nest exposure, systematic searching, or opportunity (luck). Nest
predators may use sight, smell, or sound to locate nests. If the predator community
consists largely of diurnal nest predators that locate nests in a similar manner to
cowbirds, then nest predation might occur more frequently on parasitized nests. This is
because the same factors that made it possible for a cowbird to locate a nest are likely to
make it easier for a predator to discover the nest. For example, Martin (1992b) suggests
that the number of visits an adult makes to a nest can increase the risk to nest predation.
This behavior can also increase the risk to brood parasitism (Norman and Robertson
1975). Furthermore, many songbirds must make more visits to a nest when feeding a
cowbird nestling. Subsequently, nest predation may be expected to be higher on
parasitized nests. Moreover, cowbird nestlings tend to beg loudly, and could conceivably
increase the potential for predators to locate the nest (Dearbom 1999).

How nest predation and brood parasitism ultimately affect seasonal fecundity will
depend on a large number of factors. Many of these deal with the success of subsequent
nest attempts. Few studies have explicitly accounted for re-nests and factors inﬂuencing
their success. Consequently, many of the factors that may inﬂuence re-nest success and

the resulting interaction between nest predation and brood parasitism are largely

ll

unstudied. Recent modeling work has begun to explore this interaction and suggests that
a beneﬁt to nest predation is plausible (Pease and Grzybowski 1995, Schmidt and Whelan
1999). However, these papers did not explicitly investigate the potential beneﬁt to nest
predation from this interaction.

The objective of this dissertation is to begin to explore these issues using
ﬁeldwork and simulation modeling to understand how re-nests can inﬂuence seasonal
fecundity and what inﬂuences the success of these re-nests. I will ﬁrst discuss results
from a nesting study of the Yellow Warbler focusing on the role of re-nesting and factors
affecting the success rate of these nests (Chapters 2 and 3). I will then introduce a
Simulation model designed to predict seasonal fecundity for any single-brooding songbird
under a variety of conditions (Chapter 4). This chapter will also discuss modeling results
speciﬁc to the relative beneﬁt to nest predation, and contributions of certain parameters to .
seasonal fecundity. Finally, in Chapter 5, I will examine one mechanism observed in the
ﬁeld that may prevent a beneﬁt to nest predation from occurring in real populations,

increased failure of re-nests.

12

CHAPTER 2- FIELD METHODS

A nest monitoring study was performed to explore how nest predation and brood
parasitism interact in a natural population. This empirical study focused on the role of re-
nests and factors affecting nesting success that may contribute to how nest predation and
brood parasitism interact. This would improve our understanding of whether such
interactions are likely to occur in real populations. Another goal of this aspect of the
study was to collect data to evaluate the performance of the simulation model described
in Chapter 4. This chapter describes details of the ﬁeld study sites, the focal species, and

methodology used to monitor and analyze data.

Field sites- Four ﬁeld sites in south-central Michigan were used in this study. Two sites
are within the Rose Lake Wildlife Research Area located at the southern end of
Shiawassee and Clinton counties, approximately 19 km northeast of Lansing (Figure 2-
1). Two other sites are located in the Dansville State Game Area in central Ingham
County approximately 4 km south of Dansville (Figure 2-2). The Michigan Department
of Natural Resources (DNR) manages both land areas. The DNR granted access to all
sites for purposes of this study. Field sites range in size from approximately 6-12
hectares. Sites were selected according to accessibility, similarities in habitat and
vegetation structure, and abundance of breeding birds as determined through observations
made in early May 1997. Site boundaries in most instances are clearly deﬁned close-
canopy woodlands except that a small creek bounds one side of a site at Rose Lake.

Site 1 in Rose Lake is located along Stoll road 3 km west of Upton Rd (42° 48'

00" N and 84° 21' 50" W). This site, subsequently referred to as R-S, was monitored in

13

Imam ”N 35 AN 28 .mi ; 85 2 .83 05 mo 25:82 2: 88:05 E385: v0.35 .35 Bow
u8< 0:30 88w 8:5 30m 95 05 mo 25:82 05 9:32? Sewing: E 8:580 3.0.3335 was :955 526. MO 922 .TN 05!.»

 

    
 

nil.

 

1}"
5:830 3. 38m

 

 

 

      

 

 

Upton Rd

 

3. =3m N

O

Peacock Rd

  

 

 

"3.3.. 255
83m 3:5 98m

Shiawassee Co.
Clinton Co.

/‘
:'
pa uosqou

 

 

 

 

,. . 3. .56 mm

 

 

 

 

14

 

 

Dansville State
0 Game Area:

rm 131306

  

 

 

 
 

I
4
\
. '- u '
. .
9
l
-.- g
f- |
’ P-I'
L'-
JJIH‘EL9.
'7.\.
-:r: .
I;- r ,
I “'r’ .
'Q' I ‘5' I
"’63 v. "-53 .5”; .0 4 l .

 

M ‘
'0
K
, 5
S
C
.9
B s;
a: r
t. 3
2.!
g.
“I... \b? we.

 

 

 

 

Figure 2-2. Map of central Ingham County in Michigan showing the locations of the two

Dansville State Game Area ﬁeld sites. Circled numbers indicate the locations of the sites.
1) Site 1: D-W, and 2) Site 2: D-E.

15

1997 and 1998. Site R-S was not monitored in 1999 and 2000 due to poor breeding
densities of warblers and insufﬁcient ﬁeld assistants. Site 2 in Rose Lake, referenced as
R-R, is located on the east side of Robson road, 0.5 km north of Stoll road (42° 48' 10" N
and 84° 24' 35" W). This site was monitored in 1998 through 2000. Site 1 in Dansville
is located at the northwest comer of the intersection of Williamston and Ewers roads (42°
30' 45" N and 84° 18' 00" W). This site referenced as D-W was monitored 1997 through
2000. The second site in Dansville is located on the south side of Ewers road 300 meters
west of Howard road (42° 30' 45" N and 84° 16' 50" W). This site will be referred to as
DE and was monitored 1998 through 2000.

All ﬁeld sites consist of similar wetland habitat characterized by a mosaic of
shrubs and open marsh. Common shrubs on all sites include meadowsweet (Spiraea
latrfolia), red-osier dogwood (Camus stolom'fera), gray dogwood (Camus drummondi)
several species of willow (Salix spp.), and rose (Rosa spp.). These shrubs are typically
organized as dense patches of varying size surrounded by patches of mixed sedges
(Juncus spp.), grasses and occasionally cattails (T ypha spp.). Trees are generally
sporadic and include American Elm (Ulmus americana), Black Willow (Salix nigra),
Silver Maple (Acer saccharinum), Red Maple (Acer rubrum), Box Elder (Acer negundo),
Cottonwood (Populus deltoides), and Tamarack (Larix Iaricina). With the exception of
site R-S, these sites had very dense populations of the focal species, Yellow Warbler,
ranging from 2-3 breeding pairs per hectare. Other common breeding birds included
Common Yellowthroat (Geothlypis trichas), Swamp Sparrow (Melospiza georgiana),

Song Sparrow (Melospiza melodia), and Red-winged Blackbird (A gelaius phoeniceus).

l6

Focal Species- The primary objective of the ﬁeld study was to monitor nesting activity of
a songbird population in order to assess the consequences of re-nesting behavior on
seasonal fecundity. To be able to evaluate the interactive impact of nest predation and
brood parasitism this study population had to meet some basic conditions. The most
important of these conditions is that the population must be a common cowbird host that
is subject to signiﬁcant losses to fecundity when parasitized. Two species nest in
sufﬁcient densities on the ﬁeld sites and met these conditions, the Yellow Warbler
(Y WAR) and Common Yellowthroat (COYE). Both are members of the family
‘ Parulidae. Common Yellowthroat nests proved extremely challenging to locate since
they tend to be placed in dense grass or sedge and adults were very secretive in
approaching and leaving their nests. Despite considerable effort trying to locate their
nests, only a handful was monitored each year. Alternatively, Yellow Warbler nests were
readily located, thus nearly all the data discussed in this thesis concerns this species. ‘
Yellow Warblers prefer shrubby wetlands but will nest in a variety of habitats
(Brewer et a1. 1991). While small declines in total population numbers have been
observed in some regions, YWAR remain abundant throughout most of North America
and are especially increasing in abundance in southern Michigan (Table 2-1). They are
generally considered to be a leading cowbird host (Ehrlich et al. 1988). YWAR are
known to occasionally recognize and reject nests that have been parasitized, by building a
replacement nest directly over the parasitized nest (Burgham and Picman 1989).
However, they tend only to reject nests that are parasitized early in the egg-laying stage
and when there is sufﬁcient time to re-nest. Burgham and Picman (1989) discovered that

parasitism still accounted for 41% of all egg losses in their study population despite the

17

rejection behavior. Therefore, this species was deemed a suitable candidate for observing

the interaction between nest predation and brood parasitism.

Table 2-1. Breeding Bird Survey abundance trends for Yellow Warbler and Common
Yellowthroat from 1966 to 1998 (Sauer et al. 2000). Trends were estimated using the
route-regression method (Geissler and Sauer 1990).

 

 

Species Region Mean count Trend Variance P-value
YWAR Michigan 6.05 2.89 1.15 0.008
YWAR Continental 4.33 0.68 0.0319 0.0053
COYE Michigan 12.11 1.19 0.339 0.0446
COYE Continental 7.72 -0.3 1 0.0204 0.0292

 

YWAR consistently return to lower Michigan during the last week of April (pers.
obs, Brewer et al. 1991). Upon arrival, males quickly establish territories. Females
arrive a few days after males, and select mates. Males aggressively defend their territory
boundaries from con-speciﬁcs. However, different species will regularly occupy
overlapping boundaries and do not exhibit territorial aggression toward each other.
Skirrnishes between con-speciﬁcs are commonly observed along territory boundaries.
Observers used these skirmishes to delineate the boundaries of territories (Kendeigh
1944). Territory size, while not directly measured in this study, was estimated to range
from approximately 500 to 1000 m2. Adults remain on these territories until all nesting is
complete, and do not change territories between nest attempts. YWAR are monogamous
with pairs remaining together for the entire breeding season (Ehrlich et al. 1988, Brewer
et al. 1991).

Depending on the local weather conditions courtship can last from a few days to

18

two weeks. During an unusually cool spring in 1997, most YWAR did not begin nesting
until the third week of May despite arriving near May 1. During mild springs in 1999
and 2000, YWAR nest construction was observed in the ﬁrst week of May. Females
construct nests from mostly plant matter including ﬁne grasses, old milkweed stem ﬁbers,
and cotton-like plant ﬁbers. In these sites, YWAR place nests in small shrubs, usually
about one meter high (99.5 i 2.72 cm, mean :1: se, 11 = 257). Meadowsweet was the
preferred nest plant though rose and dogwood were frequently used. Typical clutch size
is 5 eggs, though clutch sizes of 4 are not uncommon (Ehrlich et al. 1988). Only one
YWAR laid a clutch larger than 5 eggs (6 eggs). Females incubate the eggs, but both
parents feed the young. Incubation typically lasts 11 days (Ehrlich et al. 1988). Hatching
is synchronous and young typically ﬂedge in 9 days (Ehrlich et al. 1988).

Upon ﬂedging, the parents will continue to care for the ﬂedglings for at least one
week and remain on or near the original territory. Since ﬂedging was rarely directly
observed, this behavior Was used in most cases to conﬁrm that a nest had indeed ﬂedged.
YWAR do not attempt a second brood after ﬂedging young, however, those whose nests
failed, re-nested. While nearly all individuals attempted re-nests in May and early June,
the frequency of re-nest attempts began to decline quickly during the third week of June
(Figure 2-3). Most individuals would attempt nests up to 3 times in a season if time

permitted. Rarely, four nest attempts were made (11 = 3).

Nest monitoring- Nest searching and monitoring techniques follow the guidelines offered
by Martin and Geuple (1993). These guidelines call for careful observations of adult

activity to locate active nest sites. To be sure we located all nesting YWAR, we

19

systematically searched each study site locating all active territories mainly by observing
singing males. Generally, territories saturated each site with the exception of a few large
grassy areas void of suitable nesting substrate for YWAR. Nests were most often located
through carefully observing the female until her nest was located. Female YWAR
regularly visit the nest site during nest construction. Consequently, most nests were

located during this stage.

0.8 ;V# VV V VV V _. VV i L.» V7 .,.

  
 

Re-nest Frequency
.0 .o
«b O)
i
i
i
i
i
i
i

 

-___ .. -
l
i

0 “To “ 2630 40 7‘ 50 * 60 70
Daysfromﬁrstegg

Figure 2-3. The frequency at which replacement nests are initiated following nest failure
is very high in the ﬁrst month of nesting, but declines rapidly in the second month (mid-
June). Numbers represent the total number of nests that could potentially re-nest in an
interval.

Upon locating a nest, a ﬂag was tied to nearby vegetation. Marking the location
of nests with ﬂags placed near the nest has been shown to increase predation (Picozzi
1975). To minimize this, ﬂags were placed more than 5 meters from the nest and were

most commonly 8-10 meters away. The location of the nest was recorded on a nest card.

Recorded information included, approximate nest location on the site, a compass bearing

20

from the ﬂag, estimated distance from the ﬂag to nest (m), nest plant species, estimated
nest height (cm), and a written description or drawing of the nest location from the ﬂag.

Monitoring active nests may potentially inﬂuence nest success through observer
disturbance (Gotrnark 1992). The following efforts were made to minimize any biases
caused by observer disturbances at the nest. In order to prevent abandonment, nests were
generally not approached during construction unless adult behavior suggested that the
nest was complete and egg-laying had begun. When the nest was complete, the nest
contents were checked every 2-4 days by using a mirror attached to a 2-meter pole. This
prevented any trampling of vegetation directly to the nest. In certain situations direct
observations of the nest contents were made when the mirror could not provide a clear
view of nest contents. After a complete clutch was laid the nest contents were not
checked again until the estimated hatch-date, unless adult behavior suggested that the
nest had been depredated. The nest contents were checked once more 2-3 days prior to
ﬂedging to conﬁrm the number of young. The number of young ﬂedged was assumed to
equal the number of young present during this last check. Occasionally, the exact
number of young could not be counted during the last check. In these cases, the total
number ﬂedged was assumed to equal the total number of hatched eggs. Nests were
avoided when potential avian nest predators, including cowbirds, were present. When
nest contents were not visually inspected, 15-minute observations were made ﬁ'om a
distance of 10-15 m to determine that the nest was still active based on adult activity.

Several behavioral measures were recorded during these lS-minute observations.
These measures were tested for correlation to parasitism or predation risk. When

approaching the nest, an adult usually lands in nearby vegetation and travels the

21

remaining distance through the vegetation. This distance varies depending on either adult
behavior or surrounding vegetation. The distance between the nest and where an adult
drops into the vegetation when approaching the nest (drop distance) was recorded during
nest construction and feeding of young. This distance was visually estimated in half-
meter increments. Finally, the relative elusiveness of the adult when approaching the
nest was recorded. Elusiveness was rated on a scale of l to 4 (Table 2-2). These
measures, along with additional observations, were recorded on the nest card during each
visit. Behavioral measurements were recorded for active nests monitored in 1998-2000;
only nest activity was recorded in 1997. For purposes of data analysis of nest survival, a

nest was assumed active if at least one egg was laid.

Table 2-2. Criteria for elusiveness ratingof adult YWAR at the nest when the observer
was nearby.

Rating Criteria

 

1 A bird that ﬂies directly or nearly so to the nest without regard for the presence
of the observer.

2 A bird that drops into the vegetation before approaching the nest but otherwise
does not hesitate.

3 A bird that is hesitant to approach the nest and is often difﬁcult to follow to
nest.

4 A highly elusive bird- rarely seen approach nest or will not go to nest when the
observer is present.

 

A nest was parasitized if a cowbird egg(s) was observed in the nest. Cowbirds
will lay eggs in a host nest during the ﬁnal days of nest construction through the egg-
laying stage (Robinson et al. 1995a). Nests that failed during egg-laying, where it was

not known if it had been parasitized between checks, were excluded from analyses of

22

parasitism rate and comparisons concerning parasitized nests.

If a nest failed before ﬂedging young, the type of failure was record. Table 2-3
summaries the categories of nest failure observed. Nests were assumed to fail due to nest
predation when the contents of the nest were missing and/or the nest was dislodged or
torn. A nest was assumed to have failed due to abandonment if nest construction was not
completed. Occasionally, a female would abandon the nest after laying one or two eggs.
In these cases, abandonment was assumed the cause of failure since no other eggs were
expected in the nest at the time the nest was checked. Failure due to burial of cowbird
eggs was obvious. However, in a few instances egg burial was not discovered until

nesting was complete and the egg(s) were found in the nest lining.

Table 2-3. Categories and number of nest failures for 347 monitored YWAR nests.

 

 

 

Failure type j. N Percent of failed nests
Predated 155 83.8
Abandoned V 15 8.1

Buried 8 4.3
Weather/adult mortality 3 1.6
Unknown 4 2.2

Total 185 100

 

In only two cases was weather thought to have caused failure. In both, the eggs or
young had fallen from the nest, which was apparently dislodged from the plant by wind
or heavy rain. In another unusual case, a female was found dead on her nest with no
noticeable harm. While the cause of her death is unknown, 3 severe storm did pass

through the area the previous night.

23

Only a Blue Jay (Cyanocitta cristata) was actually observed depredating a nest,
though Crows (Corvus brachyrhynchos) and Common Grackles (Quiscalus quiscula)
were also seen in isolated instances ﬂying by with eggs or nestling (species not known) in
their bill. Also, both northern water snakes (Nerodia sipedon) and common garter snakes
(ﬂamnophis spp.) were seen in close proximity to recently depredated nests with bulges
in the body suggesting eggs or nestlings. Other potential nest predators seen or known to
inhabit the area include the raccoon (Procyon lotor), skunk (Mephitis mephitz's), white-
footed mouse (Peromyscus spp.), eastern chipmunk (T amias striatus), gray (Sciurus
carolinensis) and red squirrel (T amiasciurus hudsonicus), Cooper’s Hawk (Accipiter
cooperii), Great-homed Owl (Bubo virginianus), and Eastern Massasauga (Sistrurus
catenatus). All sites were also inhabited by two other species of bird, the Gray Catbird
(Dumetella carolinensis) and House Wren (T roglodytes aedon), which are not generally
recognized as nest predators but are known to destroy nests (Spooner et al. 1996, Sealy
1989, Bellesisles and Picman 1986). Since the identity of nest predators cannot be
accurately determine without direct observation much of the discussion conceming the
predator community will be speculative.

Every effort was made to locate any re-nest attempt after failure. Female YWAR
will commonly pull nest material from the previous nest to construct a replacement nest.
The replacement nest is always built somewhere within the current territory.
Construction of the re-nest typically begins only a day or two after failure of the previous
attempt (pers. obs.). This information along with efforts to uniquely color band adults
(see below) allowed for a conﬁdent determination of the identity of re-nest attempts. A

re-nest was assumed to exist if the adult remained active on the territory for more than 10

24

days after failure of a prior nest. Most re-nests were found quickly during construction.
Only a few re-nest attempts were not eventually located. In these rare instances, activity
of the adults was still recorded to determine if a nest was eventually successful. Only
when no adult activity was observed during several days of at last a half-hour of

observation was nesting activity assumed terminated for the season.

Vegetation Assessment- When an active nest was completed, additional measurements of
the nest site were made. These consisted mostly with assessing the concealment and
vegetation around the nest and are summarized below. Measurements were performed
promptly after either the nest failed or young ﬂedged. By making these measurements
soon after the nest was complete, the nest site conditions at the time the event occurred
were more precisely ascertained. However, ﬂedglings may remain in the proximity of
the nest and females often use nest material ﬁ‘om a failed nest to construct her .
replacement nest. Therefore, to avoid disturbing the birds these measurements were
occasionally not performed immediately upon nest completion. Also, nest measurements
were not made on buried nests until the subsequent nest attempt was complete. These
measurements were recorded on the nest card.

Nest concealment was assessed two ways. One method followed a procedure
commonly used in the literature and described by Martin and Conway (1995). This
method has the observer subjectively determine the percentage of the nest that is
concealed by vegetation. This is made ﬁ'om one meter away at nest level from the four
cardinal directions and overhead. In many instances the overhead percent cover could

not be viewed from one meter above due to nest height. Therefore, overhead cover was

25

estimated at eye-level. This generally was not a problem since most often vegetation did
not exceed more than a meter above the nest and was usually below eye level. Percent
cover was recorded in 10% intervals. Nests had to be at least a certain percent cover to
be rated at that percentage level. For example, if a nest was actually 47% concealed it
was rated as 40% covered. This approach was used to minimize observer bias. A second
measure of concealment followed the method used by Cresswell ( 1997). Here, a 3-inch
piece of bright-pink ﬂagging tape was placed on the rim of the nest. Using a 30-meter
measuring tape, the distance (m) at which the colored tape was no longer visible was
recorded at the four cardinal directions. For both methods, a mean percent cover or
distance was calculated from the four cardinal measurements.

The height of the nest and nest plant (cm) were measured. The number of plant
stems intersecting a plane through the nest rim in a 25 cm radius around the nest center
was recorded as a stem density. This stem density was also used as an index to nest
concealment in analyses. The percent cover of eight major plant categories (Table 2-4)
found within a 2-meter radius around the nest was recorded. Percent cover was recorded
in 10% intervals in the same manner as nest concealment. The total percent cover of all
plant categories could not exceed 100%.

The number of snags and trees near a nest may inﬂuence vulnerability to nest
predation or parasitism since it would provide perching locations for avian predators and
cowbirds (Clotfelter 1998, Anderson and Storer 1976). Therefore, the number of snags
and trees located within 10 meters of a nest were recorded. While trees greater that 10
meters away may also contribute to nest fate this distance was selected since in most

cases trees within 10 m were still in the breeding territory boundary. A snag was any

26

dead tree that stood at least 4 meters high. The number of live trees was recorded for two
size classes: <15 and >15 cm DBH. Trees were any non-shrubby woody plant greater
than 4 meters high. These nest site measurements were made in 1998 through 2000.
This data was used to compare differences in vegetation structure in proximity to the nest

to estimates of nest survival, and to determine if nest site preferences change for re-nests.

Table 2-4. Categories of vegetation that were quantiﬁed in vegetation assessment of a 2
meter radius around each nest.

 

Plant category Description

 

Grass/sedge All species of grasses and sedges.

Dog/willow All species of willow and two species of dogwood- gray and red-osier
dogwoods. Generally the two dogwoods species dominated this
category.

Spirea This is meadowsweet, the dominated low shrub preferred by YWAR’S
for nesting.

Forb This category included any herbaceous plants excluding grasses and
sedges, but including ferns.

Rose This category included multiﬂoral rose shrubs.

Other This category included all other species of woody plants not in previous
categories.

Dead This category included any dead plant matter such as dried grass and

sedge from the previous year or dead shrubs.

Water This category included any standing water covering the ground

 

In the ﬁnal year of the study, non-use nest sites for all active nests were selected
and measured for vegetation at the same time as active nests. The following procedure
was used to select these non-use nest sites. A random direction centered at the active nest
was selected by tossing a stick into the air. The recorder would walk ten paces in the

direction the stick pointed when it landed. If this placed the non-use site outside the

27

known boundary of the territory then the opposite direction was taken. At this point, the
height of the nearest plant was measured. Percent cover of surrounding vegetation within
a 2-meter radius centered on the measured plant was recorded in the same manner as for
active nests. Also in the same manner, trees and snags were counted in a 10 m radius.
These non-use sites were used to evaluate the selectivity of nest sites by birds and for

assessing the quality of nesting territories.

Color Banding- In addition to monitoring nest activity, efforts to capture and band
breeding adults were made in 1998-00. Nylon mist-nets measuring 5-meters long and 2-
meters high with approximately 3-cm mesh were used to capture adults. In 1998, 12-
meter nets were also used as part of a separate study (Epp 1999). Adult males were
drawn to the mist-net by playing a recording of the advertisement song near the net. A
dummy bird carved from styro-foam and painted with appropriate markings was often
placed in nearby vegetation to lure males. Females almost never respond to these
recordings or dummy birds. Instead, females were captured by strategically placing the
mist-net in common ﬂight paths between vegetation. In some cases this meant placing
the net close to the nest though rarely closer than 5 meters, and nets were not placed in
locations that required signiﬁcant trampling of vegetation to be set up. Nets were
constantly monitored and birds removed immediately upon capture to avoid injury.
Upon capturing a focal bird, the bird was banded. A numbered US. Fish and
Wildlife Service (USFWS) aluminum band was place on either the left or right tarsus
depending on the year. Unique combination of two or three plastic color bands were also

place on either tarsi. Color bands may be a single solid color or two colors split

28

horizontally. Most often one band was placed on the same tarsus as the USFWS band
and the other was place on the opposite tarsus. No more than two bands were placed on a
single tarsus. Both adults of a breeding pair were given the same band combination.

The sex and age were recorded before a bird was released. Sex was easily
determined from plumage differences. Male YWARs are bright golden-yellow and have
striking chestnut streaks down the breast that are absent or faded in duller females. Age
was determined using criteria described by Pyle (1997). These include comparing the
contrast in color and feather-wear between the primary and greater covert feathers of the
wing. The condition and shape of the tail feathers (rectrices) as well as length of the
wing cord were also used. As recommended by Pyle (1997) and the Bird Banding
Laboratory guidelines (Gustafson et a1. 1997), age was categorized as second year (SY),
after second year (ASY), or unknown (UNK) when feather condition was inconclusive.
SY birds are in their second calendar year and are breeding for the ﬁrst time, while ASY
birds are assumed to have bred at least once before in a prior year. ASY birds have
brighter less contrasting color between the primary and greater coverts of the wing,
generally have longer wing cords, and have less-wom rectrices with more rounded tips.
Age was used in comparative analyses of nest survival and nest attempts.

Both YWAR and COYE were banded and measured when captured. All other
species that accidentally ﬂew into the mist-net were immediately released. Banding was
done under appropriate state and federal permits (Federal Permit No. 09633-P).
Permission to handle live animals was also obtained ﬁom the Michigan State University
Committee on Animal Use and Care (Application # 02/98-023-00). Most mist-netting

effort was focused on capturing YWAR’S. All COYE were captured accidentally during

29

attempts to capture an adult YWAR. The color bands served to identify birds and aid in
conﬁrming the identity of replacement nests. Attempts to color band at least one member
of each breeding pair were made, but not all birds were banded, especially in 1998.

Some banded individuals returned in subsequent years increasing the proportion of
banded birds in 1999 and 2000. Although not all breeding pairs were banded the identity
of re-nests were conﬁdently determined in nearly all cases through several criteria

mention earlier.

Statistical Analysis- Nest survival was estimated using the Mayﬁeld method (I-Iensler and
Nichols 1981, Mayﬁeld 1975, 1961). This commonly applied method uses the total
number of nest failure events and nest exposure days to estimate the probability of failure
as a daily rate or for the span of the nest cycle. Nest exposure days are the total number
of days each nest was vulnerable to nest predation. I assumed that a nest was exposed to
failure beginning on the ﬁrst day of egg-laying. This method attempts to eliminate any
bias resulting from the discovery of nests after egg-laying is complete since some failures
in these cases may have gone undetected. However, most nests were located during
either nest construction or egg-laying. If a nest was discovered during egg-laying, the
total exposure days were calculated back to the day the ﬁrst egg was laid under the
assumption that one egg is laid per day (McMaster et al. 1999).

End dates of exposure were determined as follows. I assumed that nests were
vulnerable to failure until ﬂedging. If a nest ﬂedged young between checks, the mid-
point between checks was assumed the ﬂedge date unless the hatch date was known. In

the latter case, I assumed that ﬂedging occurred 9 days following the hatch date (Ehrlich

30

et al. 1988). Usually nests were checked every other day when ﬂedging was anticipated
and the exact day of ﬂedgling could be easily determined. When a nest was depredated,
the failure date was, again, assumed to have occurred at the mid-point between the
present and previous check date. A nest was considered successﬁrl if it ﬂedged at least
one young, including cowbirds. These follow guidelines recommended by Mayﬁeld
(1975 and 1961). Survival or failure (predation) rate estimates and variances were
generated using the program MICROMORT (v 1.3). Tests of signiﬁcance between
estimates of nest survival or failure rate were made by means of the Chi-Square test with
the program CONTRAST (v 4.1) (Hines and Sauer 1989, Sauer and Williams 1989).
Comparisons included survival for each nest attempt, each age class, and other
comparisons as noted in Chapters 4 and 5. Unless otherwise noted, estimates of nest
success or nest predation are presented for the span of the nest cycle.

The probability that a brrd would re-nest following a prior nest failure at a given
time (re-nest frequency) was estimated as follows. I ﬁrst determined the number of birds
whose nests failed during each of several 5-day intervals. I then counted the number of
these birds that built‘a replacement nest. The re-nest frequency was the proportion of
birds that re-nested out of those that failed during that interval. A bird was assumed to
have re-nested if at least one egg was laid in a replacement nest. Birds in which it was
unclear if they re-nested were excluded from this analysis.

Behavior data were summarized in the following manner. Drop distances and
elusiveness rating were sub-categorized according to the nest stage at which it was
measured (Building, Egg, or Nestling). When more than one measure of a single variable

was taken during a particular nest stage, the mean was entered into the analysis. Visits

31

were averaged per lS-minute periods. The number of visits by adults was recorded and
sub-divided by nest stage as above. Not all measures were taken for all nests. For
example, drop distance may have only been recorded during the building stage and not
the egg or nestling stages or elusiveness rating was recorded only during the nestling
stage. Because of this, traditional multivariate analysis such as logistic regression could
not be used since incomplete cases would be excluded, thereby eliminating most nests,
especially those that were depredated. Therefore, statistical analyses were made on each
variable independently using a Kruskal-Wallis AN OVA for multi-way comparisons and
Mann-Whittney U test for two-way comparisons.

A logistic regression analysis was used to determine which surrounding
vegetation variables best distinguished various dichotomous dependent variables such as
nest success (ﬂedged versus predated) or age (ASY versus SY). A stepwise regression in
the forward direction was implemented to ﬁnd the best-ﬁt model. Variables were
included or removed from logistic regression models using a criterion of P < 0.15.

When appropriate, the Kruskal-Wallis Analysis of Variance test was made on data
that were not normally distributed. When parametric tests such as AN OVA revealed the
same conclusions as those of non-parametric tests, the parametric tests results were
presented. Normality of data were veriﬁed by examining residual plots. Other statistical
tests such as contingency tables were used as noted. All tests were two-tailed and
conducted at the a. = 0.05 signiﬁcance level. Statistical analyses were carried out using
SYSTAT v.8.0 (SPSS 1998). Means are presented with standard error (SE) unless

otherwise noted.

32

CHAPTER 3- ANALYSIS OF FIELD OBSERVATIONS

F ecundity and Life History- Nests of 201 female YWAR were monitored during the four-
year study (Table 3-1). A total of 347 nests were discovered and monitored. We
conﬁdently located each nest attempt for all but 14 females (7 %). This was determined
based on observations of activity on territories and the tracking of banded birds. Most
females attempted two additional nests following nest failure of the ﬁrst nest, but in three
cases a fourth nest was also made. A total of 501 young ﬂedged from these nests. Thus

fecundity was estimated to be 2.49 young per female per season.

Table 3-1. Summary of the numbers of females, nests, and young ﬂedged by year for all
four sites combined.

 

No. Young Fecundity per Fecundity per

 

 

Year No. females No. nests Fledged female nest
1997 28 52 51 1.82 0.98
1998 51 86 . 123 2.41 1.43
1999 55 89 161 2.93 1.81
2000 67 120 166 2.48 1 . 38
Total 201 347 501 2.49 1.44

 

To evaluate the performance of the population, I estimated the fecundity
necessary to sustain a population. This requires estimates of adult and juvenile mortality,
which were not measured in this study. Instead, I selected estimates of 0.50 and 0.25 for
adult and juvenile over-winter survival as conservative estimates (Ricklefs and Bloom
1977, Ricklefs 1973, Lack 1954). Using these values and assuming equal sex ratios, a

33

population would need to produce 2.0 young per female to sustain a population. With the
exception of 1997, the average young produced per female was 2.0 or greater (Table 3-1).
Therefore, assuming that the estimates of adult mortality are accurate, this warbler
population appeared to be sustainable and likely growing.

Fecundity per nest lies well below this threshold suggesting that nesting only once
in a season will not sustain a population (Table 3-1). In fact, the estimate of seasonal
productivity is approximately 40% greater than that estimated per nest. This difference
indicates the contribution of re-nests to productivity. To directly identify the contribution
of each nest attempt to seasonal fecundity, I determined the cumulative number of young
ﬂedged per attempt. From this, I determined fecundity for only ﬁrst nests attempts,

fecundity for ﬁrst and second attempts combined, and so on (Table 3-2).

Table 3-2. Estimated cumulative young ﬂedged per female by nest attempt. First nest
attempts contributed the most to seasonal fecundity. Estimates of young ﬂedged were
determined according to the total number of young ﬂedged up to that nest attempt.
Values in parentheses are proportional contributions to total fecundity.

 

 

 

Year Number of nest attempts
l 2 3" _
1997 1.25 1.79 1.82
(0.69) (0.29) (0.02)
1998 1.51 2.18 2.41
(0.63) (0.28) (0.10)
1999 1.85 2.64 2.93
(0.63) (0.27) (0.10)
2000 1.67 2.31 2.48
(0.67) (0.26) (0.07)
All years 1.62 2.29 2.49
(0.65) (0.27) (0.08)

 

a. Includes 3 fourth nest attempts producing a total of 7 young.

34

In no year did this population produce a sustainable number of young per female
by attempting only one nest. However, attempting one additional nest after failure did
result in a sustainable productivity rate, again assuming that the mortality rates estimated
above are appropriate for this population. At least 90 percent of all young produced came
from the ﬁrst two nest attempts (Table 3-2). Birds that attempt a third nest may do so as
an insurance policy against a bad year when unexpected high adult or juvenile mortality
occurs, or may be trying to maximize their individual fecundity.

Mean brood size (number of young ﬂedged) for successful nests declines for each
attempt (Figure 3-1, Table 4-3). A two-factor ANOVA (year and attempt, with sites
pooled) revealed no signiﬁcant effect for year but the decline in brood size by attempts
was signiﬁcant (F = 12.745, df = 116, p < 0.001). Pair-wise comparisons using Fisher’s
least signiﬁcant difference (LSD) revealed a non-signiﬁcant decline in brood size
between ﬁrst and second attempts (p = 0.125) but a highly signiﬁcant decline in brood
size from second to third broods (p = 0.001). Moreover, mean clutch size (number of
eggs laid) was signiﬁcantly reduced in each subsequent nest attempt (Figure 3-1, F =
34.93, df= 107, p < 0.001).

Mean brood size per nest attempt was slightly lower than clutch size in the same
attempt (Figure 3-1). This is expected since not all eggs that are laid successﬁrlly ﬂedge.
No nestlings were ever found dead in a nest suggesting death by starvation was rare,
while several instances of un-hatched eggs were observed. Thus, the difference between
mean clutch size and brood size likely reﬂects hatch failure. Therefore, the observed
reduction in broods by nest attempt is believed entirely the result of reductions in clutch

size and not losses after laying.

35

01

4.5 l I o clutch (1) i
‘ . brood (2)
G) i
.N 4 ~
0)
3.5 3 :
'5 i
c 2.5 .
.9. i V
2 2 .1 i
O i i
n(2) = 76 47 15
1 VLV - H- -V -_ I _ _ __ V
1 2 3

Figure 3-1. Mean clutch and brood size declined signiﬁcantly with nest attempt.

Factors affecting clutch size is a richly studied topic in avian biology, and many
hypotheses have been proposed to explain Clutch size patterns (Murphy and Haukioja
1986). Most hypotheses interpret clutch size as a trade-off between maximizing
fecundity and a variety of costs, which depend on the hypothesis. Examples include food
availability, predation, or lifetime survival. The decline in clutch size with successive
nesting attempts observed in this population suggests that there is a physiological cost to
re-nesting, which limits the number of eggs a female can continue to produce. Similar
reductions in clutch size have been observed in other YWAR populations (Mitra 1999).
However, another study of this species found no signiﬁcant reduction in brood size
within the season (Goossen and Sealy 1982). This suggests that local variation in
external resources, namely food, and deterioration of weather conditions, also contribute

to this decline by limiting egg production (Smith and Andersen. 1982, Crick et al. 1993).

36

Brown-headed Cowbird parasitism rates on YWAR nests were relatively low and
ranged from 0-33% between sites and years with an overall rate of 12.6 % (n = 310) for
all years (Table 3-3). In contrast, parasitism on COYE was 41.2% (n = 17) over the same
time period. In addition, more than half of accidentally discovered Swamp Sparrow and
Song Sparrow nests had at least one cowbird egg and many of these had multiple cowbird
eggs. This suggests that YWAR were not a preferred host despite references in the
literature as such (Erhlich et al. 1988).

Reasons for the relatively low parasitism rate for YWAR despite frequent
cowbird activity are unclear. Parasitism rates of YWAR have been measured at or near
50% in other localities in and near southern Michigan (Batts 1961, Berger 1951,
DellaSala 1985, also see McGeen 1972) and as high as 80% in western populations
(Tewksbury, pers. mm). In addition, Freeman et al. (1990) found that Brown-headed
Cowbirds are better at ﬁnding nests in marsh habitat with a high density of trees along the
perimeter, which characterized all four sites used in this study.

Parasitism varied signiﬁcantly between sites and was lowest at DW and highest at
DE and RS (X2 = 14.748, df = 3, p = 0.002). The difference in parasitism between DE
and DW is particularly notable since these sites are located only 2 km apart. It is well
documented that cowbirds are more frequent in fragmented landscapes near agriculture
(Robinson et al. 1995b). The higher parasitism frequencies at RS, RR, and DE are likely
due to their close proximity to agricultural lands and human habitation as compared to
DW, which is more isolated (per. obs.). Individual differences in host preference by
female cowbirds may also contribute to the variation between each site, and to the

unexpectedly low parasitism rate observed for YWAR (Alderson, et al. 1999).

37

Table 3-3. Comparison of fecundity, predation, and parasitism by site, all years and
attempts combined. Letters denote signiﬁcant differences according to Chi-square tests.

 

 

 

Site No. Females Fecundity ' Predation Parasitism
DE 51 3.14 0.394 a 0.247 3
DW 69 2.35 0.645 b 0.063 b
RS 16 2.00 0.543 “b 0.280 “
RR 65 2.26 0.709 b 0.190 a
All sites 201 2.41 0.546 0.126

 

1. Fecundity is calculated as the number of young ﬂedged/ number of breeding females.

Nest predation rates between sites and years were variable and ranged ﬁ'om as
low as 14.4% to as high as 78.3%. Overall, an average of 54.6% of nests were lost to
predation. Predation rates were not signiﬁcantly different between years (X2 = 4.197, p =
0.241). However, predation did vary by site (X2 = 19.056, p = 0.0003). This difference
was mainly due to the signiﬁcantly lower predation rate observed on DE compared to
other sites (Table 3-3). A site-by-site comparison of overall predation and parasitism
rates reveals that with the exception of DE increased parasitism is negatively correlated
with fecundity (Table 3-3). Site DE had the highest mean productivity per female despite
having the second highest parasitism rate, because of the low predation rate. No apparent
relationship between predation and fecundity is evident from these sites. For example,
while the lowest predation rate is associated with the highest fecundity, the second lowest
predation rate is associated with the lowest fecundity.

Parasitized nests were 50% more likely to be depredated than non-parasitized

nests (Figure 3-2). More frequent predation of parasitized nests seems likely due to

38

several factors. Nests that were depredated or parasitized shared similar vegetative
characteristics that link the two (see below). For example, nests placed near snags were
more likely to be parasitized but were also more likely to be depredated. Also, cowbird
nestling begging is loud and frequent (Dearbom 1998), and this may increase the risk of

predation for parasitized nests (Briskie et al.1999, Leech and Leonard. 1997).

.09
‘10)
i—o—i

51

'0)

Nest predation rate
0 O O O
A or
g m

 

0..
N00

.f__._1. __ ..
l

.O
A

 

Figure 3-2. Comparison of nest predation estimates between non-parasitized and
parasitized nests. Parasitized nests had a 50% greater risk of predation (X2 = 12.06, p =
0.0005). Sample sizes are placed below standard error bars. Sites and years are pooled.
Variation exists in the literature whether parasitized nests are more or less likely
to be depredated. For example, Strausberger and Ashley (1997) found no difference in
nest success for six common hosts, while Payne and Payne (1998) and Weatherhead
(1989) found greater predation on parasitized nests in the Indigo Bunting (Passerina
cyanea) and Yellow Warbler, respectively. Thus, the correlation of nest predation and

parasitism apparently varies depending on habitat and host species.

Those parasitized nests that did successfully ﬂedge young averaged 1.56 young

39

per nest. However, in two cases cowbird eggs failed to hatch in the nest. When these
two nests are removed from the sample, the mean number ﬂedged per nests drops to 1.11
young. Given that the overall mean brood size for successful non-parasitized nests was
4.11 young per nest (Table 4-3), successful parasitism reduced brood size by 73%.

Nest predation for re-nests was consistently higher compared to initial nest
attempts (Figure 3-3). In general, predation on re-nests was 8-10 percentage points
higher than initial nest attempts, which was not a signiﬁcant difference for any given year
but the overall trend was signiﬁcant (paired t-test p = 0.004). Experimentation to

determine the cause of this pattern is the subject of Chapter 5.

   

0.9. - _ _ .j_. _
0.8 1 DIMal ,
.0
S 0.5
,3; 0.4
‘17:
2 0.34
0.2 i 4
0.1 %~ *1— _

 

 

 

 

 

 

 

1997 1998 1999 2000 all

Figure 3-3. Mayﬁeld estimates of nest predation were consistently higher for re-nests
than for initial nest attempts, though not signiﬁcant within years. Error bars denote
standard error.

The potential for nest predation to enhance fecundity could exist only if the mean

fecundity from parasitized nests which were not depredated was less than the mean

fecundity of females who’s parasitized nests were depredated. The difference between

40

these two estimates is a measure of the beneﬁcial effect of nest predation on parasitized
nests, excluding any cost to nest predation on non-parasitized nests. Successful
parasitized nests, including nests where cowbird eggs failed to hatch, averaged 1.56 :E
0.32 (3: SE) young ﬂedged, while females that escaped parasitism through nest predation
averaged 1.53 :t 0.40. This suggests that this population could not gain in fecundity since
females of depredated nests that contain cowbird eggs did not experience improved
fecundity from subsequent re-nests. Failure to improve fecundity relative to successful
parasitized nests probably resulted from increased predation on re-nests (Figure 3-3).

The total cost to predation can be evaluated by comparing the mean fecundity of
successful ﬁrst nest attempts to the total seasonal fecundity estimate for all nest attempts.
The mean fecundity per female of successful ﬁrst nests was 3.880 :1: 0.112. This value is
the expected fecundity if no predation was present while including parasitism.
Comparing this value to the seasonal fecundity value estimated for the entire population
(2.46 young/female) suggests that predation reduced fecundity by approximately 36% in
this population. If we discard those nests that were parasitized, the mean fecundity per
nest without predation is 4.11 young (Table 4-3). Comparing this value to the 3.88 young
ﬂedged per nest with parasitism suggests that parasitism reduced fecundity by only about
6%. This low reduction is undoubtedly due to the low overall frequency of parasitism

observed in the population since the individual cost per nest was more than 70%.

Nest Placement- Nest concealment was measured three ways as described in Chapter 2.
These were tape distance, coverage, and stem density. Poorly concealed nests should

have greater tape distances, lower percent side cover, and lower stem densities around the

41

nest. Therefore, tape distance is inversely related to the latter two measures. An
ANOVA found that tape distance had a signiﬁcant effect on nest fate (F = 6.252, p <
0.001). Pair-wise comparisons using Fisher’s LSD revealed that tape distance was
signiﬁcantly greater for depredated nests compared to successful nests (Figure 3-4).
Interestingly, mean tape distance for abandoned nests was similar to predated nests while
buried and successful nests were similar. These patterns were consistent for side cover
and stem density (Figure 3-4), suggesting that nest concealment may inﬂuence a bird’s
decision to abandon or bury a parasitized nest. However, sample sizes for abandoned and
buried nests were too small to explore this further.

Parasitized nests have signiﬁcantly less side cover and lower stem density
compared to non-parasitized nests but they do not differ in tape distance (Table 3-4).
These results suggest that proximate concealment of the nest inﬂuences vulnerability to
parasitism but long distant concealment does not. Female cowbirds will search for nests
by moving through low vegetation and can ﬁnd nests by ﬂushing adults ﬁ'om them
(Strausberger 1998). Thus, the disparity in results of the three measures of concealment

may reﬂect cowbird nest searching behavior.

Table 3-4. Comparison of marginal means for three measure of nest concealment
between parasitized and non-parasitized nests.

 

 

 

Measure of Non parasitized nests Parasitized nests P-value

concealment (LSD)
Mean N Mean N

Tape distance 4.21 139 4.55 35 0.433

Side cover 65.3 191 56.1 44 0.024

Stern density 24.2 190 20.8 44 0.047

 

42

100

 

 

 

 

 

 

 

 

 

 

 

 

90 ~-
31? 80
§ 70 i)
8 i
8
a 60. .- i
.4 1
n=9 6 114 121
40 V , .
abnd bury ﬂdg pred
7
6 -
e 5— i i
8
E 4 V
S 3 §
§
,_ 2 -
1
n=5 6 88 86
0 . V .
abnd bury ﬂdg pred
4O .VVVCVV VVVVVA- . - - . . 2
f 35 ..
8 I i
3 30 . -- w
E i 9 i
L”. 25 . ,
*3 l l. i
3 20. ,_ § .
E l
9 15 4 i
U) . __ l
1 n=9 6 113 121 z
10 WWW LIV. . VIII- -__._V.i
abnd bury ﬂdg pred

Figure 3—4. Three independent indices of nest conceahnent show that successfully
ﬂedged nests were more concealed than depredated nests. Error bars are 95% conﬁdence
intervals.

43

Parasitism events occur at the beginning of a nest attempt while concealment
measurements were made after nest termination. Consequently, results may not reﬂect
nest site conditions at the moment of parasitism. However, while Burhans and
Thompson (1998) found nest concealment to vary with time, this pattern was not
different between parasitized and non-parasitized nests. In other words, they found that
all nests became less concealed with time. Because the change was consistent with nest
type, comparisons using data collected after nest termination should still reﬂect real
differences between them. Clearly, both nest predation and brood parasitism are
inﬂuenced by nest concealment.

Results indicate no signiﬁcant trend in nest concealment with respect to nest
attempt. However trends were toward shorter tape distances, lower side cover, and less
stem density with later nest attempts (Figure 3-5). The trend for tape distance is counter
to that of the latter two measures. That is, shorter tape distance reﬂects greater
concealment while lower side cover and stem density suggest less concealment. One
plausible explanation for this result: Annual vegetation growth, especially of tall grasses
away from the shrub patch where the nest is located could reduce tape distance measures
in later nests. However, the proximate measures (side cover, stem density) trends toward
lower concealment suggesting that adults select less concealed nest sites in later attempts,
or that re-nests consist of a greater proportion of adults that selected less concealed nest
sites. Again, this trend is not signiﬁcant.

An ANOVA of overhead nest cover revealed a signiﬁcant effect on nest fate (F =
4.678, p = 0.003) but not between nest attempts (F = 0.543, p = 0.653,) or parasitism (F =

1.93, p = 0.147). A contrast comparing overhead cover between successful and

44

f" P 9"
cams-orator

 

Tape Distance (m)

1..

n=111 56 15

1"
no:

Attern pt 1 attempt 2 attempt 3

70. 7 ,7
65.

60.

r—Tr—l

 

Side Cover (%)
01
0'1

; n=153 72 22

Attempt 1 attempt 2 attempt 3
28 VV 7777,, V
26 .

24 i

22

20 -. >
18 i
16 ..
14 V;

12
; n=153 71 22
10 . 7,

Mom pt 1 attempt 2 attempt 3

 

Stem density (stems/0.196 m2)

Figure 3-5. All three indices of nest concealment reveal no signiﬁcant difference
between nest attempts, though a slight downward trend is evident in all. Error bars are
95% conﬁdence intervals.

45

depredated nests found successful nests had more overhead cover (F = 5.596, df = 234, p
= 0.019). This pattern (Figure 3-6) was similar to that of all three measures of side cover
(Figure 3-4), such that ﬂedged and buried nests had the highest overhead coverage while

abandoned and predated nests had the lowest.

100

Overhead cover (%)

 

n=9 6 114 121

abnd bury ﬂdg ‘ pred
Figure 3-6. Successfully ﬂedged nests had more overhead cover than depredated nests.

These results suggest that overall nest coverage in vegetation does inﬂuence nest
success in this population. A review by Burhans and Thompson (1998) found few
studies showing a relationship between nest concealment and either predation or
parasitism. However, Martin (1992a) found in his review that more concealed nests were
less prone to predation. Burhans and Thompson (1998) suggest that the inﬂuence of nest
conceahnent on nest success seems to depend on nest predators. For example, Clark and
Nudds (1991) found conceahnent to be important when avian nest predators were
prevalent, but not mammalian predators. Best (1978) and Wray and Whitrnore (1979)
found no relationship between conceahnent and snake predation. Based on the condition

46

of depredated nests and eye-witness accounts of predation, it appears that most nest
predation was caused by avian predators such as the Blue Jay. Figure 3-6 supports this
since overhead cover would seemingly only affect aerial predators.

In an analysis of nest height the best-ﬁt model was an ANCOVA with plant
height as a covariate. No signiﬁcant difference in nest height with respect to nest fate,
parasitism, or nest attempt was evident for all years and sites combined. There was a
slight trend toward lower mean nest height in depredated nests (96.1 cm d: 2.26 versus
98.7 cm i 2.40) and parasitized nests (92.8 cm i 2.75 versus 97.8 cm i 1.92) compared
to successful and non-parasitized nests, respectﬁrlly. A trend toward placing re-nests
slightly higher in nest plants compared to earlier attempts was also evident, but again, this
was not signiﬁcant. However, little variation in nest height exists in this population
limiting the ability to detect a pattern. It seems that nest height has little inﬂuence on nest
success or parasitism. This pattern is consistent with many studies (Smith and Arcese
1994, Best 1978, Payne and Payne 1998), though Dellasalla (1985) found parasitized
YWAR nests to be lower.

The most common nest substrate was meadowsweet, followed by both species of
dogwoods and then rose (Table 3-5). Nests placed in rose were signiﬁcantly less likely to
become depredated (Figure 3-7). A logistic regression analysis of proximate surrounding
vegetation characteristics revealed that only the proportion of rose was signiﬁcantly
different between successful and failed nests (t-ratio = 2.612, p = 0.009, odds 1.022,
Figure 3-8). F urtherrnore, non-use samples taken from territories of depredated nests

were composed of a greater proportion of grass and sedge, and had less rose (Figure 3-9).

47

Table 3-5. The distribution of nest plants in a random sample of non-use sample sites
compared to actual nest plants reveals that YWAR preferred meadowsweet and rose, with
meadowsweet being the most common shrubby plant. See text for further discussion.

 

 

 

 

Nest plant Non-Use Relative Chi-Square

sample Difference (%) P-value

Plant N % N %

Meadowsweet 121 47.6 35 32.1 +326 0014

Gray Dogwood 53 20.9 21 19.3 +7.7 0579

Red Dogwood 38 15.0 19 17.4 -13.8 0797

Rose 36 14.2 8 7.3 +48.6 0.086

Other 6 2.4 26 23.9 -89.9 '< 0.001

 

Clearly rose was a deterrent to nest predators in this community. The dense and
thorny characteristics of rose may prevent some larger predatory species, such as crows
or jays, from reaching nests placed in or near rose patches. Also notable is a non-
signiﬁcant increase in the proportion of dogwood associated with depredated nests
(Figures 3-8, and 3-9). Both species of dogwood shrubs tend to be less dense and thus
provide less coverage around the nest.

The relatively low proportion of nests placed in rose is a reﬂection of the overall
low frequency of rose on the sites. Relative differences provided in Table 3-5 are an
index of nest plant preference by YWAR with respect to abundance of that particular
plant species in the habitat as determined from non-use sampling. Positive values mean
that this nest plant was preferred. Rose had the greatest positive difference showing that
YWAR preferred to nest in rose but the low frequency of this substrate limited the overall
proportion. Meadowsweet was also a preferred nest plant species. These preferences are

also reﬂected in surrounding vegetation coverage between active nest plots (use) and

48

non-use plots. YWAR placed nests near signiﬁcantly more meadowsweet and less grass
and sedge, forbs, and dead plant matter (Figure 3-10). They also preferred to nests near
rose patches, but this was not signiﬁcant. Preference by YWAR to place nests in rose

and other low shrubs is consistent with other studies (Mitra 1999, Erhlich et al. 1988).

0.8- VV V, 7

0.7 .

0.6 f i

0.5

i

0.3 V-

Predation rate

0.2.

0.1 .
| n=121 91 36
0 V, V__ V V V

spirea dogwd rose

Figure 3-7. Nests placed in rose had signiﬁcantly lower nest predation than nests placed
in meadowsweet (spirea) or dogwoods (X2 = 4.48, p = 0.034). Error bars denote standard
error. Data are pooled for years, sites and attempts.

A two-way contingency table shows that nest plant preference did not change
signiﬁcantly with respect to nest attempt (Table 3-6). However, re-nests seemed more
likely to be placed in dogwoods and less likely in meadowsweet. This pattern is also
reﬂected in surrounding vegetation data (Figure 3-8). Logistic regression analysis of the
various classiﬁcations of surrounding vegetation cover by nest attempt (3 levels) found
the decline in meadowsweet coverage to be signiﬁcant between the three attempts (t ratio

= 2.339, p = 0.019). A more traditional logistic regression analysis coding the dependent

variable dichotomously (ﬁrst versus re-nests) found the percentage of dogwood (t ratio =

49

Figure 3-8. Three comparisons of percent coverage of surrounding vegetation in a 2
meter radius around the nest: F ledged versus depredated nests, by nest attempt, and non-
parasitized versus parasitized nests. The percent cover of rose between ﬂedged and
parasitized nests was signiﬁcantly different in a logistic regression analysis (t-ratio 2.612,
p = 0.009, odds 1.022). A logistic regression with three possible outcomes for the
dependent variable (nest attempt) found the change in meadowsweet (spirea) to be
signiﬁcantly different (t ratio 2.366, p = 0.019) with respect to nest attempts. A non-
sigrriﬁcant trend toward re-nesting in patches with more dogwood is notable. Only the
abundance of open water was a signiﬁcant predictor of parasitism (t ratio= 2.002, p =
0.045, odds 1.019).

50

Surrounding coverage (%)
10 15 20 25 30 35

O
01

 

grass/sedge [F j

- l
dog/wallow FT . . ‘V V. ., 4- . J

spirea F, , m

row

 

 

 

 

 

Ji—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[ 1 "’
r086 E. 93.,
other
dead “I,”
. gﬂdg(n= 116)
water 1:: pred (n = 122)
grass/sedge 1r... - . q‘
'dog/willow < - - e l 1
7
spuea .
i
forb
I
dead DAttempt 1 (n = 154) g
. . Attempt 2 (n = 73)
water - gAttempt 3 (n = 26) 3
.. 7.1.,__7 .1 _7777 _ 7 7 _
graSSISBdge - .. . " “177: -;‘-:V‘.;l.:E
dog/willow \ __ . .
Spirea : ... . . .—-. V I

 

 

D non-para (n = 197)
9 para (ti = 43)

 

51

Surrounding Coverage (%)

 

 

 

 

 

 

 

 

 

 

 

 

 

0 5 1O 15 20 25 30 35
grs/sedge ,L ,V7_ . ’V r-V———a——4
dog/WHO L .- 51?? _I—', -_.‘ . '9.“ six” «ﬁjﬁ—‘1
. f r—-1——4
Spirea r7 . 5.3—...
i. 1
fort) L Li". 7' I .11.. .;:; ‘I—v-i—H

rose F—ﬁ ’
7.
I:

other m:
1 a ﬂdg (n = 48)

d... .-«, . m =49

 

 

 

 

 

Figure 3-9. Logistic regression analysis comparing differences in surrounding vegetation
characteristics of non-use samples found that territories of depredated nests were
composed of a greater proportion of grass and sedge and had less rose (p = 0.048, and p =
0.040, respectively). Territories of depredated nests also have more dogwood and less
meadowsweet (spirea), but means were not signiﬁcantly different.

Surrounding coverage (%)
O 5 10 15 20 25 30 35

 

 

 

 

 

 

 

 

 

 

 

other [:1 use (n = 255) ,
J |
dead - .3] 7 .. a nonuse (n = 106) ‘

 

 

Figure 3-10. Logistic regression analysis of differences in surrounding vegetation
coverage between active nest sites (use) and non-use plots suggests that YWAR select
nest sites with more meadowsweet (spirea) and less herbaceous plants (*, p < 0.05; **, p
< 0.01; or ***, p < 0.001). The difference in coverage of rose was marginally
insigniﬁcant (p > 0.108).

52

—1.842, p= 0.065), dead plants (t ratio = 3.366, p = 0.018), and forbs (t ratio = —2.088, p =
0.037) to be signiﬁcant or nearly so (Figure 3-11). Similarly, Mann-Whitney U tests of
each vegetation variable versus nest attempt found dogwood (p = 0.046), forbs (p =
0.054), and dead (p = 0.071) to be signiﬁcant or approach signiﬁcance. As plant growth
progressed during the season one would expect the percentage of dead plant coverage to
diminish and the percentage of annual forbs to increase as observed. Therefore, only the

increase in cover of dogwood/willow seems biologically important to nest placement.

Table 3-6. Nest plant preferences did not change signiﬁcantly between nest attempts
(Pearson’s X2 = 8.22, df = 8, p = 0.412). However, a shift occurs toward a greater
proportion of later nests placed in dogwoods.

 

 

 

 

 

Attempt l Attempt 2 Attempt 3
Plant N % N % N %
Meadowsweet 79 51.0 33 45.2 9 39.1
Gray Dogwood 31 20.0 17 23.3 5 21.7
Red Dogwood 18 11.6 12 16.4 9 39.1
Rose 23 14.8 9 12.3 4 17.4
Other 4 2.6 2 2.7 0 0

 

Finally, a logistic regression found only water to be a signiﬁcant predictor of
parasitism (t ratio = 2.002, p = 0.045, odds 1.019, Figure 3-8). This relationship is likely
due to correlation with density of snags. Most tall snags were in or near areas of standing
water. In some habitats, Brown-headed Cowbirds are known to utilize snags as
observation perches while searching for nests (Clotfelter 1998, Anderson and Storer

1976). Figure 3-12 shows that parasitized nests were associated with a signiﬁcantly

53

greater abundance of trees and snags. This pattern is consistent with other studies except
that it has been suggested that cowbirds tend only to uses such perches in habitats that
lack a closed canopy similar to the shrub swamps of this study (Brittingham and Temple
1996). Figure 3-12 also shows that depredated nests were associated with a signiﬁcantly
greater number of snags and small trees and that re-nests were associated with more
snags. It would appear that the abundance of elevated exposed perches near nests has

some inﬂuence on nest fate, suggesting that avian predators are common.

 

 

 

 

 

 

 

 

 

 

Surrounding coverage (%)

0 5 10 15 20 25 3O 35

' _ 777. 717; 77 77. 7- 7

grass/sedge _. V7 ,7. a» I

i r

' a: i

' i

Spirea L": .' 1...:-'..;;.;;‘i‘;.~ M 2

 

 

o
r
n .1 I
I7 . >4

 

 

_. D Attempt 1 (n = 154)
water i , .All re-nests (n = 99) [

 

Figure 3-11. A dichotomous logistic regression of surrounding vegetation cover between
ﬁrst nest attempts and re-nests found the variation in coverage of dogwood/willow to
approach signiﬁcance (* p < 0.10) and found the difference in percent cover of forbs, and
dead plant matter to be signiﬁcantly different (**, p < 0.05).

54

Dﬂdg (n =114)
‘ apred(n=122)

   

   
  

3 ,
i .. ..
2 i
l
i
1 ‘ l
‘ l
o l , 7,
lgtree snag
4 7 7 V, , 7V, 7 -7
anon (n = 193) m
1 para (n=44)
3 i
(I)
i i
C: 2
o l
o. .
z
1 i
i
0 -
4 4
i [jatmp1 (n = 155) ,
.atmp2(n=72) ’
3 j Datmp3(n=26) i
2 l
) i
1 ‘ i
l i
0

Figure 3-12. Predated and parasitized nests had more trees and snags nearby and later
nests tend to be near more large trees and snags than earlier nest attempts. Compared are
number of trees by category: small tree- smltree, large trees- lgtree, and snags, in a 10 m
radius around the nest for: ﬂedged versus predated nests, by nest attempt, and non-
parasitized versus parasitized nests. Kruskal-Wallis One-way Analysis of Variance tests
were used to test for signiﬁcant differences. Asterisks signify level of signiﬁcance (*, p
< 0.05; **, p < 0.01; or ***, p < 0.001).

55

Parental age- Nests with both adults aged as second year (SY) were twice as likely to
become depredated as nests where both adults were after-second year (ASY) (Table 3-7).
In nests where only one adult was aged, ASY adults were also more successful than SY

adults, but this difference was not signiﬁcant (Table 3-7).

Table 3-7. Nests of adults, who had bred at least once before, ASY, were less likely to be
depredated, than ﬁrst time breeders, SY adults.

 

 

 

Parents Age N Predation rate X2 p-value
ASY 30 0.402 8.173 0.0043
Male and female
SY 10 0.825
ASY 92 0.538 1.949 0.1627
At least one‘il
SY 31 0.680

 

a. These are cases in which only one adult was sexed.

Contrasts of age and sex reveal that both SY males and females were signiﬁcantly
more likely to be depredated than their ASY counterparts (Figure 3-13, X2 = 7.119, p =
0.008). These results suggest that older adults are more successful at avoiding nest
predation. Consequently, re-nests were composed of a greater proportion of SY adults
than original nest attempts (Figure 3-14). The greater nest failure rate of re-nests (Figure
3-3) is therefore partly explained by the greater proportion of inexperienced SY adults
that were re-nesting due to prior nest failure. See Chapter 5 for further discussion of this.
Brood and clutch sizes were not signiﬁcantly different between adult ages.

Nests of SY adults were parasitized more ﬁ'equently than ASY adults (Figure 3-
15). However, Chi-square tests were not signiﬁcant in either sex (males: X2 = 2.48, df =

1, p = 0.115; females: X2 = 1.26, df = l, p = 0.262). Difference in males would approach

56

signiﬁcance in a one tailed test. An analysis of sample size suggests that this sample is
insufﬁcient to detect a signiﬁcant difference (more than 400 nests would be needed to be
80% conﬁdent of detecting a difference). I have found only three other studies that
compared host age to risk of parasitism (Smith and Arcese 1994, Payne and Payne 1998,
Nolan 1978). Of these, only Smith and Arcese (1994) found a signiﬁcant relationship. In
contrast to my data, they found a greater risk of parasitism associated with older adults.
Assuming cowbirds cannot discriminate hosts age, it is not likely that age is a direct
determinate of parasitism risk but indirectly inﬂuences parasitism through factors that are

correlated with age such as nest site characteristics.

 

0.8 .

0-7 i { 39 i 50
0.6

O

E: 051 i 134 {56

E

g 0.4 .7

g i

93 0.3 .

o.
0.24 osy .
0.1 OASY:

O . - 777. ....-_-_ ._

Male Female

Figure 3-13. SY males and females had signiﬁcantly higher Mayﬁeld estimates of nest
predation compared to their ASY counterparts (X2 = 7.119, p= 0.008). Error bars are
95% conﬁdence intervals; Sample sizes are included.

Logistic regression analysis of surrounding vegetation composition by age of
nesting adults found that ASY females placed nests in patches with signiﬁcantly more

meadowsweet (t- ratio = 1.960, p = 0.050, odds 1.02, Figure 3-16). No other category of

57

0.7 7V, 7 7 ,V,

06 D1st

renests

0.5

SY proportion

   

Male Female

Figure 3-14. The proportion of SY aged adults in re-nests increased regardless of sex.
However, the increase was not signiﬁcant in either case. Pearson’s Chi—Square results:
X2 = 1.97, df= 1, p = 0.16 and X2 = 0.98, df= 1, p = 0.322 for females and males
respectively. However, an analysis of both sexes combined was marginally signiﬁcant
(X = 3.455, p = 0.063).

0.5 7777,
0.4 . sv
0.3

0.2

% parasitized

0.1 .

0.0

_77

   

Figure 3-15. Nests of SY adults of both sexes were parasitized in greater proportions
than their ASY counterparts. However, Chi-square tests were not signiﬁcant in either sex
(males: X2 = 2.48, df= 1, p = 0.115; females: X2 = 1.26, df= l, p = 0.262).

58

vegetation was signiﬁcantly different for female age, though nests of SY females tended
to have more dogwood. No signiﬁcant trend in surrounding vegetation in relation to
male age was detected. Results of two-way contingency tables showed no signiﬁcant
difference in nest plant preference between ASY and SY males or females (Figure 3-17).
However, nests of ASY females were more likely to be placed in meadowsweet while
nests of SY females were more likely to be placed in gray dogwood.

Nests of ASY females were signiﬁcantly more concealed from the side and had
shorter tape distances (Table 3-8). Likewise nests of ASY males had signiﬁcantly more
side conceahnent, shorter tape distances and higher stem density (Table 3-9). Nests of
both SY males and females had signiﬁcantly more snags but only nests of SY females
had signiﬁcantly greater number of small and large trees, though males were only
marginally insigniﬁcant (Table 3-8 and Table 3-9).

Given that nest success was related to nest conceahnent (Figure 3-4) and the
number of nearby trees and snags (Figure 3-12), it seems evident that variation in
territory quality or nest site selection between SY and ASY adults explains the difference
in nest success observed with age (Figure 3-13). SY adults occupied territories that offer
less concealment of the nest and had more trees and snags or were less skilled at selecting
optimal nest site locations. It appears, then, that territory quality is a signiﬁcant
determinate of nest success and subsequent seasonal fecundity. Territories with a greater
proportion of preferred nest plants, and fewer trees and snags performed better. It is also
apparent that ASY adults occupied better quality territories. For example, nests of ASY
adults had more meadowsweet (Figure 3-16) and thus more options for nest placement,

and had fewer trees and snags (Tables 3-8 and 3-9). Therefore, shifts in nest placement

59

Coverage (%)
0 1O 20 30

 

 

--r
dOQWd/ 1 E l
.hf. .‘u. .v. 45:3.“ “J-Lfiﬁifxff :_§:::J.f.. iii-i 1.4“... A..-:§n'4:'\"“‘ - “in“. 1’:'~.;Kr,-.‘ £7 .'.‘L‘.;i‘r:r"_" . iii]
—«

 

 

 

 

 

spirea

 

“('3 ‘.. ~ ,, '» --\ _. ,‘.‘ » ' ‘
.... 7-.-... ..
orb WW1}!
..I...
.
.

 

 

Females

E
dead I'm Dasy (n = 53)
0-

 

 

 

 

cars/sedge .J
‘1
' 1
dOQWd/WIIO E ;-:‘: “i6 ; V ;-L 5‘ : 31711-5534. 1:;-.~.;:".r:- ’::- ‘1‘;-E~§-t~f‘T-“-.‘"T-T:"3 : 7:3
. ‘1
spirea
forb

rose .5

 

 

 

 

 

‘“ 1
other ,I Males

dead Uasy (n = 125)

water sy (n = 35)

 

Figure 3-16. The percent cover of meadowsweet (spirea) was signiﬁcantly greater in
ASY females in comparisons of surrounding vegetation coverage by age and sex.
Signiﬁcance was determined using logistic regression (* p = 0.05).

60

 

 

 

spirea

a
.2
i:
8. o 8 f—
2
Ga

 

 

16

 

spirea

 
 

13.14%
‘--;j a"?
.1 . ..-‘.
5g. .;
‘7 '"xlij‘ﬁ
“:y‘: “7.:
v‘ .Jﬂ
5.55.2,
;'{
3" *I»‘é
3.-
ﬂ“ .
7. ii'
a u
13 J “‘“r‘
,. .;

QWdOQ

9

QWdOQ

'F‘eﬁeues‘
Cl ASY
5 SY

 

reddog rose
Males
l
4
reddog rose

Figure 3-17. Two-way contingency table analyses between age and nest plant preference
found no signiﬁcant difference for females and males (X2 = 1.073, df = 3, p = 0.784 and
x2 = 0.263, df= 3, p = 0.967, respectively).

61

 

 

 

 

 

e86 :H 93 H a: 3 36 H 36 mm eweem
N80 8H 2.0 H 8.0 me e; H 86. mm EBA see
25.0 t .e wed H 02 a. mac H e: 9 so 2v $2...
33 Red O? H EN 9. m? H «MN m €83 82m
80.0 83 :3 H 8H 3 Rd H SH 3 853% £5
32 23 3m H 2K 2. on... H ”S mm :38 eeeego
”So on: New H men 3 e3 H 58 mm :38 eEm
2%; Nx mm H :82 2 mm H :82 2
”233-335 .3 >m<
mac—«Eon

 

.mo_mE£ >m 55 mwmem EB moo: $30.“ 98 .8566 38

Steam £960 25 0.88 was 838$ >m< mo 88: :2: 305 8388 mo own 3 3558820 86 ”we: we mnemtmmﬁov .w-m 033.

62

 

Edﬁmﬁ;

 

 

 

 

30.0 :H 3.0 H was en m2 $25
2 .o R.“ moo H e: 9.. 86 H Se «2 EBA we;
weed MS as H N3 em 86 H 3; m2 8e 2v 82H
80.0 0.2 a: H 0.8 em 33 H 3A ms 92% see.
80.0 Noe e2 H 8H 3. 25 H 8e 2: 892% 25
RS SN e3 H wee 8 m? H N: ms 525 32:26
sod 8.: mg H new 2 SN H eee HS Hose eem
2%; Nx mm H :82 2 mm H 5e: z
$334835 3 >m<
832

 

. .838 >m 52.: mwmcm .530.“ v5 .bmmcoc 82m 83on .8566
093 5:27. £960 25 808 can 838 >m< mo 38: 35 26% $38 no own .3 83388523 87. “won me $533800 .o-m 2an

63

of re-nests observed for nest plant preference (Table 3-6), vegetation composition
(Figures 3-8, and 3-10), and snags (Figure 3-12), appear to be the product of the greater

proportion of SY adults (Figure 3-14) and not changes in individual nest site selection.

Parental Behavior- A comparison of behavior data between depredated and ﬂedged nests
was analyzed using Mann-Whitney U tests. Table 3-10 summarizes results of this
analysis. Only behavior during the nest building stage appeared to be a signiﬁcant
determinate of nest success. Birds were more likely to succeed if they dropped further
away from the nest when approaching, made fewer visits per 15-minute interval, and
were scored more elusive. In general, successful adults were more elusive. However, if
experiment-wise error is corrected using the Bonferroni correct alpha-value, only
elusiveness during nest construction would be signiﬁcantly associated with nest success.

A majority of nests failed within the ﬁrst ten days of activity (Figure 3-18).
Nolan (1978) observed snakes, chipmunks, and Blue Jays revisiting nests from which
they previously consumed contents, and even observed predators revisiting nests found
prior to egg-laying. This behavior has also been experimentally documented (Sonerund
and F jeld 1987), suggesting that some predators, especially avian predators, may ﬁrst
locate nests using parental activity during nest construction and return later to consume
eggs.

No differences in parental behavior were detected between parasitized and non-
parasitized nests or between nest attempts using Mann-Whitney U tests and Kruskal
Wallis tests, respectively. ASY males visited nests more often (1.98 d: 0.177, n = 35) and

dropped into the vegetation closer to the nest (2.31 d: 0.30 meters (m), n = 25) during the

64

 

 

 

 

$0.0 2m R3 H 3% a wad H 2H 3 262m

$3 a: 23 H 9: m. on; H N3 8 33> .e2

:3 we. 83 H 8H N 3.8 H SH 9. a5 85 mesmez

32.0 83 mm; H EH E 08.0 H 3N i 252m

83 men 33 H m? 2 mm; H on; x. 22> .2

3.3 m. 5. 02.0 H e3 8 83 H com 3 e5 85 mwm

god 38m 32 H e3 3 :55 H m2 ee 355

~86 38 :3 H an S 83 H e? we 33$ 52

H80 82 83 H a: ee 33 H e? 3 e5 85 weﬁam
832 D :52 mm H 72m: 2 mm H z<m2 z esecH> emem

33 Z 62¢on 38 Z Romeo—m

 

.6286 u e e 8:85?
8m 2:? 302.8 382.50 Eotomcom $ .0322; :08 .3 532298 5 co.“ m SEED oom 623on so: 8 68?: 33 memo—:5
“mo: 05 macaw $25383 3:23 Eco 35 @2828 09% “we: 3 8333 3:83 we 8.5308 8:: mo mnemthEoU .on 033.

65

nestling stage than SY males (1.13 i 0.08, n = 8 and 3.67 2‘: 0.76 m, n = 6) (X2 = 6.18, df
= l, p = 0.013 and 3.89, df = l, p = 0.049 for visits and drop distance respectively).
Males that occupied higher quality territories may ﬁnd sufﬁcient food more frequently
and thus visited the nest more oﬁen, as observed. No behavior variables were
signiﬁcantly different among female age. These results indicated that behavioral
differences and parental age were not strongly associated with each other. Therefore, age
speciﬁc variation in nest success (Figure 3-13) appears not to be inﬂuenced by parental

behavior but more likely by variation in territory quality (See chapter 5).

 

4o ..

 

 

 

 

\ A‘lTEMPT
I, \ —— 1
/ ‘ \ _ _ 2
0 1 l 1 ----- ~l~ \ 3
O 5 10 15 20 25
Nestdays

Figure 3-18. The distribution of number of nests depredated by nest days revealed that
most YWAR nests were depredated during the ﬁrst 10 days of the nest cycle regardless
of attempt number.

Summary- Assuming the estimates of annual survival are appropriate, this YWAR
population appeared to producing a sufﬁcient number of young to be sustainable. This

would not be the case had adults re-nested less frequently since re-nests contributed about

66

40% to seasonal fecundity. Overall nest predation was similar to that of most songbird
populations, but the frequency of parasitism was unexpectedly low for Yellow Warblers.
Those nests that were parasitized were 50% more likely to become depredated than non-
parasitized nests; however, re-nests had consistently higher nest predation. Moreover, re-
nests of depredated parasitized nests ﬂedged equal or slightly fewer young than that of
successful parasitized nests. Consequently, this population exhibited no potential to
beneﬁt from nest predation.

Nests placed in vegetation with more concealment did better than nests with less
concealment. YWAR preferred to nest in rose or meadowsweet, and nests placed in rose
or near rose patches were more likely to succeed, while nests placed near snags and trees
were more likely to fail or become parasitized. Re-nests tended to have more snags
nearby than ﬁrst attempts. A trend toward a greater proportion of re-nests placed in or
near dogwood species, which were associated with lower nest success, was evident.

Nest success also differed by age and re-nests consisted of a greater proportion of
less successful SY adults. Both age classes also differed signiﬁcantly in the composition
of surround vegetation in proximity to nests. ASY adults placed nests near more
meadowsweet, and had fewer snags, and possibly trees. ASY nests were also more
concealed. Only parental behavior during the nest building stage signiﬁcantly inﬂuenced
nest predation. Male and female age did not differ with respect to behavior during nest

building, indicating parental behavior did not inﬂuence nest success between ages.

67

CHAPTER 4 - THE MODEL

The Yellow Warbler ﬁeld study provided many insights into nesting success and
the impact of nest predation and brood parasitism. However, due to the overall low
parasitism frequency and low nest success of re-nests, evidence of a relative beneﬁt to
nest predation was not found. The aim of this chapter is to address whether there exists
any hypothetical conditions for which some level of nest predation might be better than
no nest predation in a songbird population. To address this requires an examination of
how various levels of nest predation affect seasonal fecundity across a range of
parasitism levels and costs. Realistically, this cannot be done in a natural system. To do
so would require manipulation of the frequency of nest predation and brood parasitism
levels on various treatment populations, yet neither can be reasonably controlled in a
natural breeding population. Therefore, simulation modeling was deemed the only
appropriate means of exploring the interaction between these two phenomena.

There are two known quantitative models that predict avian productivity while
independently accounting for the effects of predation and parasitism (Pease and
Grzybowski 1995, Schmidt and Whelan 1999). Pease and Grzybowski’s (P&G) model
(1995) is a powerﬁil, mathematically complex model that will generate fecundity
estimates for a wide variety of conditions. Schmidt and Whelan (S&W) describe a
mathematically simpler model consisting of a series of probability equations for all
possible fates of a nesting female. Both explicitly and independently model nest
predation and brood parasitism while allowing for re-nesting, and parameters can be
readily manipulated. Therefore, either could be used to explore the relative effects of

nest predation and brood parasitism. The S&W model was published following the end

68

of this study and therefore was not available. Attempts were made to utilize the P&G
model but failed mainly due to the complexity of their model and the fact that a number
of parameters critical to the model could not be adequately estimated. Instead, I elected
to design my own simulation model that would permit manipulation of parameters such
as the frequency of nest predation or parasitism freely and would generate estimates of
seasonal fecundity for a hypothetical population given essential life history information.
In this chapter, I present a dynamic model entitled FLEDGE which can simulate
seasonal nesting cycles of single-brooding songbird populations. FLEDGE explicitly
models the effects of nest predation and brood parasitism separately, and predicts the
total number of ﬂedglings produced by a female population for a single season. It is not
speciﬁcally designed for predicting population grth across several years. Here, the
model is used speciﬁcally to illustrate the interaction of nest predation and brood
parasitism and to determine what conditions are necessary for nest predation to induce a
relative increase in seasonal fecundity. The speciﬁc goals are outlined as follows: 1.
Describe a quantitative population-based model that predicts seasonal fecundity given
various conditions including brood parasitism and nest predation. 2. Test model
performance against real ﬁeld data. 3. Assess the relative impact of various levels of both
brood parasitism and nest predation on seasonal fecundity using model simulations. 4.
Evaluate the relative effects of selected conditional parameters on predicted seasonal

fecundity with respect to the predation-parasitism interaction.

Model structure. - This simulation model was created using STELLA II Research®

version 5.1.1 soﬁware (High Performance Systems 1997, Constanza et al. 1998).

69

STELLA is a graphically based program for creating dynamic mathematical models. A
model is ﬁrst constructed graphically in a form similar to a ﬂow diagram. Then
mathematical or logical relationships between ﬂows and linked parameters are deﬁned in
a second layer. When a model simulation is executed, Stella calculates essential
variables according to a time step process, recalculating variables at each time step as
deﬁned by the user. For example, consider a basic model of population growth. Stella
would recalculate the size of the population for each deﬁned time step (day, week, year,
etc.) by taking the population size at the beginning of the time step and applying the birth
and death rates, which can be ﬁxed, time dependent, or density dependent according to
model design. The model can be run for any number of time intervals and the result
might be a plotted curve of population growth during that interval. A user can run a
model only within the software program.

FLEDGE follows the basic structure described below and diagrammed in Figure
1-1. The parameters necessary to run the model are listed in Table 4-1. Initially a
population of females, N, is uniformly divided so that they begin the nesting cycle over a
5 day period. As each set of females enters the nesting cycle, they are exposed to a brood
parasitism frequency, p(t). This frequency is equivalent to the probability that a nest is
parasitized and is commonly referred to by many authors as a rate. The parasitism
ﬁequency divides the population into those that are parasitized and those that are not.
These two groups then continue temporally under a daily time step interval through the
nest cycle: egg laying, incubation, hatching, and ﬂedging. During each daily time step, a
small proportion of nesting females are removed from this nest cycle according to a nest

predation rate, d(t), which is converted to a daily survival rate (3) using the following

70

equation,

S = (e+t,~+t,,)(1_d(t) , (1)

where e is the clutch size, ti is the length of the incubation period in days, and tn is the
length in days from hatching to ﬂedging. The sum of these three parameters is the total
number of days a nest is exposed to predation assuming one egg is laid per day. Again,
the predation rate or frequency is equivalent to the probability that a nest is depredated.
The daily failure rate is 1-s, and the number of females that fail (Nd) and are removed
from the nest cycle during any given day is calculated as follows: Nd = N -(1-s), where N
is the number of females still actively nesting. To prevent rounding errors from
propagating through the model, fractions of females are permitted.

The predation rate for parasitized nests, dp(t), is adjusted by a measure of the
difference in predation between non-parasitized and parasitized nests. This parameter, 6,
is the proportionate increase (or decrease) in predation rate for parasitized nests. Thus,

nest predation on parasitized nests is calculated using the following equation,

dpﬂ) = d(t) (1+ 5)- (2)

When predation, d(t), is high or 5 is large dp(t) can become a frequency value greater than
one. This is equivalent to nest predation being greater than 100%, which is illogical. To
correct for this, the model truncates d,,(t) to the value of 1.0 when this occurs.
Abandonment is assumed to only occur during egg-laying for parasitized nests.
The nest abandonment rate (a) is incorporated into the failure rate of parasitized nests

during egg-laying, dp(t)¢, according to the following equation,

71

Table 4-1. List of model parameters necessary to run a simulation.

 

 

Parameter Symbol Format Stochastic Time
Dependent

Population size N Number of breeding females. No No

Predation rate d(t) Proportion of nests that fail No Yes
due to predation (non-
parasitized nests).

Parasitism rate p(t) Proportion of nests that are No Yes
parasitized.

Abandonment a Proportion of parasitized No No

rate nests that are abandoned.

Re-nest r(t) Proportion of failed females No Yes

frequency that re-nest following failure.

Parasitism cost c Fractional reduction of brood No No
size for parasitized nests.

Predation 8 Fractional difference in No No

difference predation rate for parasitized
nests.

Clutch Size e; Mean number of eggs laid in Yes No
the ith nest attempt.

Brood size f,- Mean number of young that Yes No
ﬂedge from successful non-
parasitized nests for attempt i.

Nest building tb Mean number of days spent Yes No

time building a nest.

Incubation time t, Mean number of days from Yes No
laying of last egg to hatching.

Nestling time tn Mean number of days from Yes No
hatching to ﬂedging.

 

72

dptt). = dpﬂ) + a[1- dp(t)], (3)

which takes the contribution of abandonment, a, relative to the predation rate, d,,(t), and
adds this to the total failure rate of parasitized nests during the egg-laying stage. Thus,
the contribution of abandonment to total nest failure declines as nest predation increases.

Females that fail (Nd) can re-nest based on the number of previous attempts and a
re-nesting frequency, r(t), which is dependant upon the time in the nesting season the
failure occurred. The number that re-nest, N,, is determined from the calculation, Nr = Nd

r(t). A maximum of ﬁve nest attempts is allowed. Those that do not re-nest are

removed ﬁ'om the model. Females that re-nest repeat the nest cycle and are exposed to
parasitism and nest predation probabilities as before. Each female that successfully
completes a nest cycle ﬂedges a mean number of young (if) or brood size as deﬁned by
the user. Since it is possible for each nest attempt to have different clutch size each nest
attempt has an independently deﬁned mean brood size per female.

In F LEDGE, the mean number of young that ﬂedge from parasitized nests (IQ) is
calculated by adjusting the initial brood size 0”,) by a fractional cost to parasitism (c)

using the following equation:

ﬁa=f1 '(1-0)- (4)

This value, j; is used for all nest attempts unless the number of young that ﬂedge from
later non-parasitized nests (ﬂ) is a smaller value. In which case, the model assumes that
there is no difference between the number that ﬂedge from parasitized and non-

parasitized nests for those nest attempts. After completion of the breeding season, the

73

total number of ﬂedglings of all nesting females is tabulated.

Each nesting attempt has the same structure as the original one with two
exceptions. The number of females entering an additional nest attempt is determined by
the number that fail in the previous nest attempt, and birds that fail in the ﬁfth nest
attempt do not re-nest. The user deﬁnes several parameters to run the model (Table 4-1).
Life-history parameters can be modeled either as stochastic or deterministic values.
During each time step of a simulation, Stella software algorithms select values for each
stochastic variable. Values are selected from a normal distribution given the mean and
standard deviation. Therefore, FLEDGE can ﬁmction either as a partially stochastic (not
all parameters are stochastic) or deterministic model depending on how parameters are
deﬁned by the user.

F LEDGE is not limited to constant probabilities of nest failure d(t), parasitism
p(t), or re-nesting frequency r(t). The user enters these probabilities as graphical
ﬁmctions versus time. The user plots a relationship between each of these parameters
and day of the season, resulting in a plotted curve (for example see Figure 2-3). STELLA
software automatically uses the value lying on this curve corresponding to a given day in
the simulation. Consequently, changes in parasitism or nest predation during different
times in the breeding season can be simulated. This is a particularly powerful feature of
this model that gives the user great freedom when describing these parameters. For
example, nest predation can be deﬁned as a constant probability (a horizontal line) or as a
declining probability (a downward sloping curve). In addition, daily estimates of these
parameters are not necessary. Instead, one can estimate each parameter (if using ﬁeld

data) for a longer time interval such as by week. The median day in each interval is used

74

to plot a curve through these points. STELLA automatically selects the appropriate daily

value lying on this curve between the points.

Assumptions and limitations - FLEDGE assumes a maximum of ﬁve separate nest
attempts. This limit may underestimate the maximum for some species (e. g., Nolan
1978) but the effect on the estimate of fecundity is likely small. For example, with a
predation rate of 0.5, only little more than 6% of a population would attempt a ﬁfth nest
and only half of these would ﬂedge young. After adding the effects of parasitism,
abandonment, the cost to parasitism, and the reduction in brood size, the increase in
fecundity contributed to ﬁfth attempts will be minuscule. Some populations may attempt
a maximum of less than ﬁve nests, as was the case for the Yellow Warbler. For such
populations, the user can switch off later nest attempts to limit the total number a
population will make.

FLEDGE is appropriate for only those species that ﬂedge a single brood in a
season. Successful females do not re-enter the nest cycle. However, general life-history
parameters such as brood size and the length of nest stage are user deﬁned. Thus,

F LEDGE can be applied to several species, especially many of the small Neo-tropical
migrants breeding in northern regions where more than one successﬁrl brood is rare.

Nests are assumed to be susceptible to predation from the beginning of egg-laying
through the ﬁnal day of the nestling stage, since this is the time period when a nest has
consumable contents. Nest predation is not inﬂuenced by nest stage such that nestlings
are no more likely to be depredated than eggs. There is some evidence to suggest that

begging by nestlings may attract predators (Briskie, et al. 1999). However, this stage of

75

the nest cycle had the lowest predation rate in my data (Figure 3-18). Therefore, I
decided not to distinguish differences in predation between nest stages.

Only parasitized nests can be abandoned, and the frequency of nest abandonment
is assumed to remain constant throughout the entire breeding season. A host will,
therefore, abandon their ﬁnal nest attempt even if doing so means zero fecundity. In
other words, the model assumes that females have no knowledge of future nesting
opportunities. If females can account for this we would expect to observe lower
probabilities of abandonment for late nests. Graham (1988) found no such pattern for ﬁve
host species. Moreover, in the current study there were at least two instances when
yellow warblers abandoned their ﬁnal nest attempt apparently due to parasitism.

The model does not explicitly include hatch failure, nestling mortality, or partial
predation, where only a portion of the nest contents is consumed. The user accounts for
such partial reductions in brood size by appropriately reducing the deﬁned average
number of ﬂedglings per female (ﬂ) compared to the initial clutch size (ei). Multiple-
parasitism, where more than one cowbird egg is laid in a single nest is not explicitly
modeled. Instead, the overall cost to parasitism including singular and multiple events is
incorporated into the fractional cost to parasitism value (c).

A host is likely able to sufﬁciently nurture a predetermined number of young
when parasitized, regardless of the initial brood size. For example, a warbler may only
successfully raise two young when a cowbird is also present in a nest, and this will be the
case if she had laid six or three eggs. Therefore, the number of young a host can ﬂedge
from parasitized nests (1;) is assumed to be the same for all nest attempts even though a

clutch reduction is likely to occur. However, it is possible that the clutch size or average

76

brood size for a late nest attempt, ﬂ, is smaller than the brood size for parasitized nests.
In this case, as stated earlier, the model automatically selects the lower value between 12-
and ﬁ, as the appropriate brood size.

As previously discussed, a variety of mechanisms can hypothetically cause
differences in predation risk between nests that are parasitized and those that are not.
These mechanisms are not likely to change or disappear within a breeding season.
Therefore, the relative difference in predation between parasitized and non-parasitized
nests (6) is assumed to remain constant throughout the breeding season. However, the
predation rate of non-parasitized nests, d(t), can change with time and is used to calculate
the parasitized nest predation rate, dp(t). Consequently, changes in these rates can occur
but the relative difference between them cannot.

Females that re-nest are assumed to begin construction of the new nest the day
after failure of the previous nest. Nolan (1978) has shown that most Prairie Warblers
begin construction of re-nests between one and two days following failure. This pattern
was also observed in the Michigan YWAR population (pers. obs.). Therefore, this
assumption appears realistic. However, if this is not the case in a particular population
under study, it can be corrected indirectly by extending the time spent building the nest
(tb). One egg is laid each day such that the egg-laying stage lasts as many days as the size
of the clutch. The initial number of females has no effect on other parameters in the
model such that effects of population density cannot be simulated. Moreover, FLEDGE
excludes events occurring before or after the nesting cycle and thus, does not deal with
nest site availability or post-ﬂedgling survival. Finally, there is no provision for adult

mortality, and all females are assumed equal in their experience and nesting ability.

77

Validation. — Before FLEDGE can be utilized to explore the effects of nest predation and
brood parasitism, model performance should be evaluated against real data to determine
whether the model realistically describes seasonal fecundity given a realistic set of
parameters. To do this, model estimates of total young ﬂedged were compared to the
actual number of young ﬂedged using YWAR nesting data calculated from three ﬁeld
sites (RR, DE, DW, see Chapter 2). All the necessary parameters listed in Table 4-1 were
calculated from the ﬁeld data and are summarized in Table 4—2, 4—3 and Figure 2-3.

In order to appropriately validate the performance of a model, data used to
compare model results must be independent from any data used to build the model (Grant
et al. 1997). In this case, FLEDGE was designed according to general mathematical
relationships focused on the probability events occur, such as nest predation. While ﬁeld
observations were used to understand the important parameters, no ﬁeld data were
incorporated into the permanent structure of the model or to deﬁne these mathematical
relationships. Therefore, use of my own ﬁeld data was deemed an appropriate validation
of the model performance. Model estimates were compared to ﬁeld estimated from each
site and for all sites combined, with years pooled.

Estimates for entering the graphical inputs of nest predation and re-nesting
frequency were calculated from weekly intervals. Weekly nest predation rates were
estimated by counting the total number of nest exposure days and the total number of
nests that failed due to predation within a given week. From this, weekly Mayﬁeld
estimates of nest survival or predation were calculated (Table 4-4, Mayﬁeld 1961,
Hensler and Nichols, 1981). Weekly re-nest frequencies were calculated by determining

the ratio of nests that eventually re-nested versus the number of nests that failed in any

78

Table 4-2. Estimates of parameter values (1 SD) used in model validation from Michigan
YWAR ﬁeld data pooled for all years.

 

 

Parameter symbol RR DE DW all 3 sites
No. Females N 65 5 l 69 185

F ledge/attempt f Table 4-3 Table 4—3 Table 4-3 Table 4-3
Clutch size/attempt e Table 4-3 Table 4-3 Table 4-3 Table 4-3
Parasitism‘ p(t) 0. 19 0.247 0.063 0. 168
Predation d(t) Table 4-4 Table 4-4 Table 4-4 Table 4-4
Predation difference 8 0.371 1.974 0.405 0.61
Parasitism cost c 0.524 0.625 1.0 0.619
Re-nest Frequency r(t) Fig. 2-3 Fig. 2-3 Fig. 2-3 Fig. 2-3
Abandonment a 0.27 0.27 0.27 0.27
Building time tb 4.0 :1: 0.5 4.0 d: 0.5 4.0 :I: 0.5 4.0 d: 0.5
Incubation time t, 11.0 at 0.2 11.0 :1: 0.2 11.0 :1: 0.2 11.0 i 0.2
Nestling time t,, 9.0 i 0.1 9.0 :1: 0.1 9.0 i 0.1 9.0 i 0.1

 

a. Parasitism was held constant through time.

Table 4-3. Mean :t SD of the number of young ﬂedged per successful YWAR nest
attempt (ﬂ), by site, years combined. Estimates of clutch size were applied to all three
sites.

 

 

 

 

 

 

Nest Attempt RR DE DW All 3 sites Clutch size
1 4.20 :1: 0.56 4.00 :t 0.60 4.19 d: 0.63 4.11 i: 0.60 4.4 :h 0.30
2 3.83 i 0.58 3.44 :1: 1.33 3.50 d: 0.67 3.53 :t 0.85 4.0 :t 0.50
3' 2.89 i 0.75 2.50 :t 0.71 2.25 i 0.96 2.67 t 0.80 3.0 :1: 0.50
a. Includes 2 successful fourth nest attempts from RR
Table 44. Weekly predation rates by site.
Site Median Day
3.5 10.5 17.5 25.5 31.5 38.5 49.5
DE 0.0 0.230 0.350 0.198 0.493 0.263 0.795
DW 0.0 0.689 0.433 0.727 0.540 0.705 0.515
R 0.791 0.623 0.643 0.722 0.719 0.535 0.242
All sites 0.419 0.529 0.488 0.603 0.590 0.599 0.488

 

79

given week. The model inputs for re-nest frequency, abandonment rate, and lengths of
time for each stage of the nest cycle were the same for all sites since small samples sizes
prevented estimates of these parameters within single sites. Stochastic parameters for
mean brood size and nest stage lengths were used to predict seasonal fecundity. Ten
simulation trials were performed for each site and for the combination of sites.

From each set of trials, a mean and 95% conﬁdence interval were calculated and
compared to the actual number of ﬂedglings counted in the ﬁeld. Based on observations
in the ﬁeld, 1 determined that all nesting attempts, except approximately 7% (14 of 201
females), were located during the ﬁeld study. Therefore, it is likely that some ﬂedglings
were not counted. In an attempt to correct for this I multiplied this 7% of females by the
average seasonal fecundity value per female associated with each site (Table 3-1). The
result is an estimate of the average number of ﬂedged young that went undetected. This
number was then added to the total number of ﬂedglings actually observed. Figure 4-1

plots these corrected estimates against the model estimates.

4.0 , -Efn ..
3.5 .1

3.0 l i

2.5 a o o
2.0 .1 l

l .
1.5 . .

F ecundity per femal

1,0 _ _ - _—
RR DE DW all3

Figure 4-1. Model predictions of seasonal fecundity per site (open circles with 95%
conﬁdence intervals, n = 10) slightly overestimate actual ﬁeld data estimates measured at
each site (dark circles) except DE. This is signiﬁcantly different only at R and DW.

80

F ecundity measurements for DE and for all sites combined lie within the 95%
conﬁdence limits of model results (Figure 4-1). However, FLEDGE seems to
overestimate the total young ﬂedged especially at R and DW. While it is possible that
additional undetected ﬂedglings may account for some of this difference, other conditions
may also inﬂuence this result. For example, the same re-nest frequency function was
used for all sites. However, in 1999, a total of seven females from RR and DW
terminated nesting prematurely, possibly due to a heat wave that occurred the same week
their prior nests had failed. The re-nest frequency for these two sites is lower for that
week than was used. The lower frequency of birds re-nesting during that week would
have undoubtedly lowered the total fecundity predicted by the model. Similarly, it is also
likely that the abandonment rate, also estimated from all sites, was signiﬁcantly different
for these speciﬁc sites. Finally, estimates of weekly nest predation do have some error
associated with them since the number of active nests in any given week can be small.
Any error in these estimates may be reﬂected in these results.

As an additional test of model reliability I compared model estimates to Nolan’s
(1978) Prairie Warbler data. This is the only data set available that provided time interval
estimates of parasitism (Nolan 1978: Table 126) and re-nest frequency (Nolan 1978:
Table 149). Using this data set, I was able to ﬁnd estimates for all parameters except 8,
which I assumed to be equal to zero. I also assumed that nest predation remained
constant. FLEDGE estimated fecundity as 1.99 young per female. This is lower than the
2.2 young per female estimated by Nolan but within the 95% conﬁdence interval (1.9 —
2.5). However, the Prairie Warbler populations made more nest attempts, up to 7

attempts, and also had second broods. Therefore, we should expect FLEDGE to

81

underestimate seasonal fecundity. According to Nolan’s observations, 2% of females
attempted more than 5 nests (Nolan 1978: Table 143). Therefore, the discrepancies in
estimated fecundity is not likely caused by the failure to include these late nests in the
model. I can estimate the contribution of second broods as follows. According to Nolan
(pg. 418) only 7% of second broods were successful. Given an average brood size of
3.18 for all successﬁrl nests (Nolan 1978: Table 142), second broods would contribute an
additional 0.22 young per female. Adding this to the estimate made by FLEDGE results
in an adjusted result of 2.21 young per female, which is nearly identical to that observed
by Nolan. Based on these results FLEDGE appears to accurately describe seasonal

fecundity for natural populations.

Simulation experiments. — Several simulation experiments were performed to explore
circumstances affecting the interaction between nest predation and brood parasitism. All
simulations determined the number of young that successfully ﬂedge during the season
for a population of 100 females. Some simulation experiments used a stochastic model
with the mean and standard deviations for clutch size, young ﬂedged per female, and nest
cycle lengths, provided in Tables 4-2 and 4-3 for the combination of sites. The purpose
of using stochastic simulations was to gain an appreciation for the natural range of
variability one might expect to observe. Except for the analysis of length of season, up to
four nest attempts were allowed, with fecundity values for the fourth nest attempt equal
to the third (Table 4-3). Other standard conditions are summarized in Table 4-5. In

stochastic simulations, ten iterations for each combination of parameters were made.

82

Table 4-5. Summary of standard conditions for critical parameters used in simulation
experiments.

 

 

Parameter description symbol Standard Value(s)
No. of Females , N 100

No. of nest attempts n 4

Cost to parasitism c 0.75
Differential predation 8 0.0
Abandonment a 0. 10
Re-nest Frequency r(t) Figure 2-3
Clutch size e,- Table 4-3
Brood size 1? Table 4-3
Nest stage lengths te, ti, and tn Table 4-2
Simulation time -- 100 days

 

STELLA requires a speciﬁc length of time be deﬁned for which calculations will
be made. This time limit is not a model parameter but a speciﬁcation of STELLA
programming. This time was set at 100 days for most simulations though nearly all
nesting was complete after 70 days according to the standard re-nest frequency. The 100-
day limit was selected only for precautionary measures. As long as this time is greater
than the time necessary for all nests to complete, it will not affect predictions made
during a simulation.

Several sets of simulations were performed to examine speciﬁc manipulations of
parameters on predicted seasonal fecundity. Table 4-6 summarizes these experimental
simulations. Below are descriptions of each followed by their results. Means, 95%
conﬁdence intervals, linear regressions, and other statistical analyses, as described below,
were performed using SYSTAT 8.0 (SPSS Inc. 1998). A two-tailed signiﬁcance level of

5% was used in all cases.

83

Overall pattern- The ﬁrst series of simulations evaluate the overall impact of
various combinations of nest predation and brood parasitism on seasonal fecundity.
Here, estimates of seasonal fecundity were made while the rates of nest predation and
brood parasitism were systematically altered at 10 percent intervals. These simulations
used the standard conditions, and both nest predation and brood parasitism did not change
temporally through the season. Stochastic variables were not use in this set of
simulations (i.e. only mean values were entered) resulting in a set of deterministic
estimates.

Figure 4-2 summarizes these results in a Surface plot. It is evident from this
ﬁgure that both nest predation and brood parasitism negatively affect seasonal fecundity.
However, the relative impact of one is closely dependent on the magnitude of the other.
‘To assess the relative impact of nest predation or brood parasitism on seasonal fecundity
one must compare the slope of the surface at the intersection of speciﬁc frequencies of
each. Whichever axis has the steeper slope has the greater inﬂuence, or is more limiting,
on fecundity. This varies depending on the particular combination of predation and
parasitism ﬁequencies. As a general rule, the slope is steepest for nest predation rates
greater than 70%. Below this level, the slope for parasitism is similar or steeper than that
for predation. The point where this switch occurs differs relative to the parasitism
frequency. At lower parasitism, predation has a steeper slope at frequencies greater than
approximately 40-50%, but at higher parasitism, this shifts to higher predation rates.
Overall, brood parasitism has a greater negative impact on fecundity. For example, given
these conditions, a bird with 60% predation and 30% parasitism will ﬂedge on average

2.28 young while a bird with reciprocal conditions (30% predation and 60% parasitism)

84

6.65 m 38.6 05 marge o 9 a; Bob an 5552 .3 35:6 33 588 .5 553 .o
dozens—axe 5m :8. com .o .5 53> some 66 .0558.“ 5 855665 05 9a 9 end one o n 36 :6 .55 203 6:25:55 .9
.65. 6036.: 5:252.“ 665-2 5.“ 5 .5535 can 85885 .5 555 55:05 5 836.» 258% 5% < x5595 cow .6

666 5 666 n 86
oz 66 5 666 666 6 n 86‘ 666 5 666 666 666 n or 525 . 556666 5.8.66

666 5 666 .6 u 86

oz 666 5 666 66.6 6N6 .6 u 86 55 66 .5 66 66 66 . 55686 5663
6.2 5 #6 666 .26 .6 u o
666 5 666 666 666 u 86

85

 

oz 26 5 666 666 .26 u 86 66 65 666 666 666 666 .6 u 6 565565.66.
666 5 666 u 86
5> 66 5 6 n 6 66 65 666 666 666 6N6 .6 u 6 .5888 2 8o
66 5 6." 6
566 666 5 666 u 86 666 666 666. 666 .26 n 86 . 86666.5 65555
666 5 666 n 86
55> 666 .5 666 .86 u 86 656 666 666 .86 .6 n 6 866665 5655.65
566 . 666 5 666 666 . 666 6 n 86 666 666 666 666 6N6 n 86 . 585656 6:555
62 2.5.55 26 5 666-6 u 86 56:35 26 5 666-6 u 86 556856 .2, 85626
6:25:55 acme—50% 655$ 82:82 6636..» 558656” 53650 coach

 

556.6 255:6. =6 .5 :32 some 5.“ @8658 663 52? 56856.85 :68 .5 How 2: 5.6 6536356 .5 6256 < 65:55”. 25:56
58.6 505.55 5 85:6 026 053 :2: $3225.85 5:5 86 6556.6 8658166 56255 26 5586.85 £65 £5 5.“ wow: 6023/
586825 .5 5586.85 :55 2: 62 5358 @862 25. .mODmAm mama: 35.68.65 855550 5:63:56 .5 55m .04. 2an

 

 

Figure 4-2. A surface plot of the mean number of young ﬂedged per female revealed that
predation and parasitism have an interaction. The rate of decline in fecundity as either
factor increases in frequency varies with respect to the alternate factor such that the
decline in fecundity due to predation is reduced as parasitism increases and vice-versa.
A DWLS smoothing operation was used to produce this plot using SYSTAT 8.0 (SPSS
Inc. 1998).

86

will ﬂedge 2.10 young. Also, a bird with 80% predation and 0% parasitism does better
than a bird with 0% predation and 80% parasitism (1.85 and 1.65 young per females,
respectively).

In addition, the ﬂattening of the fecundity surface at high parasitism and low
predation along the predation axis indicates that nest predation has minimal effect on
fecundity across a widening range of rates as parasitism increases (Figure 4-2). In other
words, fecundity becomes less sensitive to nest predation as brood parasitism increases.
For example, the average number of young ﬂedged per female is practically equal
whether she is exposed to 0% nest predation or 40% predation when parasitism is greater
than 70%. Similarly, at high predation the effect of parasitism is also dampened, though
a slight negative slope along the parasitism axis persists across the entire range of
predation. Finally, no increase in fecundity ever occurs as nest predation increases from
zero. Naturally, these results depend on the cost of parasitism, which in this case is a 75
% reduction in brood size. In any case, given these conditions, nest predation and
parasitism interact antagonistically such that one reduces the effect of the other on
ﬂedgling production. Conditions exist at which increasing nest predation does not reduce
seasonal fecundity. Similar conditions do not occur for brood parasitism.

Declining parasitism trend- The next set of simulations examined a declining
brood parasitism frequency that begins the season at a maximum and declines eventually
to zero as the season progresses. In this scenario, re-nest attempts will have a reduced
probability of becoming parasitized. In many populations, parasitism by cowbirds is
most intense early in the season and diminishes as the season progresses often

terminating by early July (Scott 1963, Nolan 1978, Robinson et al. 1995a). Therefore,

87

this declining function of parasitism reﬂects a pattern commonly expected in natural
systems. The input values used for each declining parasitism function are provided in the
Appendix (Table A—l). The slope of decline is not constant between different initial
frequencies of parasitism. If this were the case at low parasitism, zero parasitism would
quickly be reached in the season. Even when parasitism is low, it is not likely that the
declining frequency of parasitism events would completely vanish early in the season.
Instead, I deﬁned the decline in parasitism such that the rates of decline are a constant
percentage relative to the initial frequencies (Table A- 1 ). All other parameters were held
at standard conditions. The impact of a declining parasitism frequency was tested across
a range of constant predation rates.

Figure 4-3 summarizes these results, which are similar to the overall pattern using
constant parasitism shown in Figure 4-2. As before, at high levels of parasitism, nest
predation, at low frequencies, does not signiﬁcantly reduce fecundity. This trend is only
slightly different from that seen in Figure 4-2. The main difference being that there is a
slightly expanded range of predation at which no effect to nest predation exists. In
general estimates of fecundity with constant parasitism lie within 95% conﬁdence
intervals, except at high parasitism and nest predation. The latter cases are probably due
in part to the greater change in absolute parasitism frequency for higher parasitism as a
result of the percent of decline used (Table A-l). Since predictions generally were not
signiﬁcantly different from those using a constant parasitism, this more realistic declining
parasitism trend was used in all remaining simulations.

Changes in nest predation patterns- Simulations were also performed to explore

changes in seasonal fecundity for different patterns in nest predation. One series of

88

simulations tested the effect of a difference in predation between parasitized and non-
parasitized nests such that parasitized nests had a 25, 50 or 75% greater risk to nest
predation (6 = 0.25, 0.50 or 0.75). Fecundity estimates were generated for nest predation

at 20, 40, and 60% and at two l'evelsof parasitism (60% and 80%).

 

5 ,
. ____ t =00
45. ....... %§=04
‘t=08

 

 

 

151* ------ 5 ------ P ~~~~~~
1 f“‘~ ’
o5. °
0 .
o 20 4o 60 80

% Predation

Figure 4-3. This ﬁgure compares fecundity estimates versus nest predation for a
population for which the frequency of parasitism does not change throughout the season
(solid points) with a one where parasitism declines as the season progresses (line with
95% conﬁdence intervals, 11 = 10). Only at very high frequencies of parasitism and
predation do these two population differ signiﬁcantly in fecundity.

Figure 4-4 summarizes the results of this analysis. With the exception of when
parasitism is 80% and predation is 60%, increasing predation on parasitized nests results
in a signiﬁcant linear increase in fecundity. The change is small, generally less than 0.15
young per female between 5 = O and 0.75. The other effect of increasing predation on
parasitized nests is that a small beneﬁt to nest predation occurs (Figure 4-5). A positive

change in fecundity relative to zero predation occurs at both 20 and 40% predation

89

602.63 8.6 828, oogocmemﬁ 98 2: 856508 commmoamom .nmE 2.6 6088 Son 5:3 baa 68:52
95: $5. .688 Ev .coumcoa .8 .36“ .8629“wa 06 mo $.23me 56:58.“ 5 68.885 30:: =me :5 «gouacwa E 6:32
xzmbcuw $8: comEmﬁaméo: 8 0358 $60: wonmzmﬁa mo at 5:363 E oocobbmv oumcomtomoa 05 mﬁmmouoﬁ 416 853m

A8 566655 3555665

 

 

 

 
 

 

 

 

 

 

 

 

 

 

 

 

 

no he 0.0 no v.0 nd «.0 to o.o._. ad 5.0 $6 0.0 v.0 n6 No to 06.. ad Bo 0.0 md to nd «.0 to 0.0.6
- - - < u < q C - u q _ a - — C - q n u i u d C
m .l m L JW.F
_. -oN . . , row
. Sodva .. - Sodun .. . Keane - .
663.; 2 «251. 6... $31. 3
in h - - - - - CF - n - b - p p C.” - p _ h h l- - C.” %
.556 u 86 68.. n 86 .58 u 86 o
.556 u 85 .556 ... 85 .556 u 35 m
m..
no No wd m.o «.5 n6 No ﬁo 0.0. m6 Bo od 00 v.0 n6 Nd to 0.0. no No m6 Do Ya m.o No ﬁo o.o. rm
. _ _ _ _ _ _ c r . . _ . . _ _ c w _ . _ ii 4|< c F
- . 6 «Song - , - Sodua - _
1 . 2 58.61. 3 23; 6.

 
 
 

 

 

 

 

 

 

 

 

 

. 90.0 n a -3 . .
R3 "6 , I.
p P n l- n — Li C.” F . n - - . . C.” n! n - - u . _ C.”
.556 n 86 .59. u 86 .58 u 8

.58 n 85 :56 u 85 .556 u 85

90

(Figure 4-5). While the magnitude of this positive change is similar in these stochastic
results, the change is not signiﬁcant at 60% parasitism but is signiﬁcant at 80%
parasitism. This result conﬁrms that intermediate nest predation can cause a small

positive change in seasonal fecundity under intense parasitism.

 

 

 

 

 

0.3 . _ . __ __ _
gd(t)= 0.20 i
b d(t) = 0.40
'g 0.2 v ' E
3
0
If 1
0 1 ' .
.5 . I‘ F 3
2’ L 1
2 O ,. m__-._ _f .
0 l .— i
.01 L _ _ *- __ __ -- _ - _. -_._J
60% 80%
Percent Parasitism

Figure 4-5. Shown here is the change in fecundity relative to zero nest predation at two
levels of parasitism and nest predation, d(t), when 6 = 0.75. The change in fecundity is
positive for all four combinations of predation and parasitism. This change is greater at a
low predation rate than at a moderate level and is signiﬁcantly greater than zero only at
80% parasitism. Error bars are 95% conﬁdence intervals (n = 10).

In another series of simulations, the frequency of nest predation was gradually
decreased over the breeding period to half the initial frequency, simulating a reduced risk
to nest predation for re-nest attempts. Values used in the graphical input of this declining
predation trend are provided in Table A-2 of the Appendix. Similar to the declining
parasitism functions, the rates of decline in predation are a consistent percentage relative
to the initial frequencies. The effect of decreasing nest predation was tested over a wide

ﬁ'equency range of predation, at 60% and 80% parasitism, and 8 = O and 0.50.

91

Figure 4-6 summarizes the results of this analysis by plotting the change in
fecundity relative to zero predation. At 6 = 0 and 60% parasitism, there is no signiﬁcant
change in fecundity at predation frequencies of at least 30%. At 5 = 0 and 80% predation
there is a slight non-signiﬁcant positive change in fecundity at 30% predation and the
zero effect exists at 50% predation. If a differential predation effect is included, 8 = 0.5,
there is no signiﬁcant difference in fecundity for predation as high as 50% when
parasitism is 60%. However, at 80% parasitism, a signiﬁcant increase in fecundity exists
at both 30% and 50% predation. Moreover, there is no difference in fecundity at a
predation frequency of 70%. In other Words, the mean number of young ﬂedged per
female does not differ whether nest predation is zero or 70%, when initially 80% of nests
are parasitized, parasitized nests have a 50% greater risk of failure, and re-nests have
greater nest success. Increased success for re-nests is an important and powerful
condition in determining the outcome of the predation-parasitism interaction.

Cost to parasitism- Previous simulations determined that a positive change in
fecundity or non-negative “effect to nest predation can exist at moderately low predation
rates when parasitism is frequent. These results were based on a 75 % reduction in brood
size (or cost to parasitism) for parasitized nests. The next series of simulations attempts
to examine what effect the cost to parasitism has on these results. Here, I determine the
level of cost to parasitism that must exist for it to pay on average to have a parasitized
nest depredated. This was done by systematically altering the parasitism cost value (c).
For each value, mean estimates of total young ﬂedged were generated for 0% and 30%
constant nest predation. For the conditions simulated, 30% nest predation corresponds

closely to the maximum positive change in seasonal fecundity as discovered from

92

p(t) = 0.60

 

(13
(12 ._ .. ﬂ 6
OJ

-031.5"
412..
413 a- u -_ “-6... -__- _ __-
414 l

 

 

 

 

 

 

 

10 3O 50 7O

 

Change in Fecundity

(13

 

 

0.3 .1 .. _ WW---— .- ..
0.4

 

 

 

10 30 50 7o
Declining Predation (%)

Figure 4-6. Positive mean changes in fecundity per female relative to zero predation are
enhanced by a declining predation rate and when 5 is greater than zero. The magnitude
of this change is greater at high parasitism rates, p(t). Furthermore, the range of
predation at which a positive or no change in fecundity occurs is greater compared .to
previous results where predation was held constant through time. Error bars denote 95%

conﬁdence intervals.

93

previous simulations. This was repeated for four treatments, 60 and 80% declining
parasitism and 6 = 0 and 0.5. The frequency of nest predation was held constant through

the season and all other conditions were set at standard levels.

Table 4-7. Estimates of the minimum cost to parasitism for a positive change in
fecundity to exist at a 30% predation relative to 0% predation. Estimates are given for
two levels of parasitism and differential predation. Values were calculated by ﬁnding the
x-intersection from regression equations derived in SYSTAT.

 

Differential predation value (6)

 

 

Parasitism 0 0.5
60% 108% 79%
80% 79% 69%

 

The difference in total young ﬂedged between 0% and 30% nest predation was
calculated. These differences were plotted against the corresponding cost to parasitism
value (Figure 4-7). The intersection of the line at the x-axis (cost to parasitism level)
indicates the minimum cost to parasitism that must exist for predation to lead to an
increase in fecundity for a population under the identiﬁed conditions. The values at this
intersection were calculated from a linear regression analysis for all four treatments and
are provided in Table 4-7. Results indicate that a relatively large cost to parasitism must
exist for nest predation to increase fecundity. A higher cost is necessary when the
frequency of parasitism drops and when parasitized nests do not have greater predation.

Nest Abandonment— Another set of deterministic simulations explored theeffect
of nest abandonment. Here, the level of abandonment was altered at 20% intervals while

nest predation was held constant at four levels (Table 4-6). The frequency of brood

94

Change in total F ecundity (100 females)

Figure 4-7. The relative change in total fecundity (between 0% and 30% predation) is
positively related to the percent cost of parasitism (c). The point where regression lines
(with 95% Cl’s) cross the x-axis denotes the level of cost to parasitism that must exist for
a positive change in fecundity to occur. This point declines with increased parasitism

p(t) = 0.60

 

 

 

 

 

 

 

5 = 0.0
50 I I I I
o ‘5 J“
81 R Q % 0,—iI-J
‘ ° 3 f;Q.J—-"{—;
afﬁx"?! ° 4
riff” 8
1
~50 ti- . s _.
1»
100 l l l l
0 20 40 50 30 100
mo=oso
5 = 0.0
50 I I T T
1
G o 0/‘1
o
91 -.

 

 

 

J. x..- 0
1%» .612,” .
.I/K

rav-

‘50 4;— q
1P 0
4

100 l L l I

D 20 40 60 80 100

 

 

 

 

 

 

 

 

 

 

% Cost to parasitism (c)

p( t) = 0.60
5 = 0.5
50 T I T I
4‘
o 1’
8
.3 ’1’! F
A A / ALI-’- 4:-
" o , {Ted}!
8 1.5:, '5'“ g t
.3 -{'_--:-_..--""
1 $1133)”
axis:- 8
(VI/(:65; o
‘50 I—,.-" d
I!” o
1| 0
400 l l l l
0 20 60 30 100
“0:030
8 = 0.5
50 I I I I ‘
o 1
e /'
.«41;
.4671
8 i443" ‘
: 3-5¢“‘°
9 /-6-i§€:§/l
o _/{,
/{:f;/
w- / 1'
.50 tP/»’/. o .1
V .
1
-lDD l l l L
0 20 40 60 80 100

rate, p(t) and differential predation (8). See Table 2-7 for x-intercept values.

95

parasitism was set at 20, 40, 60, or 80% and each combination was simulated at ﬁve
levels of cost to parasitism (c). The set resulted in 480 individual simulations for each
combination of parasitism, predation, parasitism cost, and abandonment rate. All other
parameters were held at standard conditions.

Several interesting results where apparent from these simulations. First, most
abandonment frequencies only slightly affect the small positive change in fecundity when
parasitism is high (Figure 4-8). Abandonment sharply inﬂuences the change in fecundity.
only when abandonment approaches 100%. Near this level, the positive change in

fecundity due to predation disappears.

 

 

 

 

 

0.1
0
Q
E -0.1 .
§
u. 0.2 .
.5
8: 0.3 .
C
2
o -0.4 -
-o.5
-O.6

 

0% 20% 40% 60% 80% 100%

Abandonment rate

Figure 4-8. The mean change in fecundity (between 0% and 30% predation) at 80%
parasitism is not strongly inﬂuenced by the abandonment rate except at very high
abandonment. Results are given for two levels of cost to parasitism (c), and show that
although the cost can inﬂuence the change in fecundity, this pattern is consistent with
respect to the abandonment frequency.

The second result is that abandonment is not always proﬁtable when parasitized

(Figure 4-9). When there is a 50% cost to parasitism it does not pay to always abandon

96

parasitized nests regardless of whether the frequency of parasitism is 40% or 80%.
However, there is a clear advantage to always abandoning when the cost to parasitism is
75% and parasitism is 40% but not when parasitism is 80% (Figure 4-9). One would
expect that as the cost and intensity of abandonment increased, the advantage of nest
abandonment as a means to alleviate the effects of brood parasitism would increase.
Modeling results contradict this conventional perception of the adaptive nature of nest
abandonment. The reason for this contradiction is due to the probability of success for
the subsequent re-nest following abandonment of a parasitized nest.

The speciﬁc threshold abandonment rate that will result in greater fecundity
depends on three conditions: the cost to parasitism, and both the frequency of predation
and parasitism. It pays to abandon a parasitized nest only if the subsequent re-nest has a
higher probability of success. Thus abandonment is more likely to be proﬁtable when the
frequency of nest predation and parasitism are low and the cost to parasitism is high.
Table 4-8 provides example threshold values of the cost of parasitism above which

abandonment will increase seasonal fecundity.

Table 4-8. Approximate values of the cost of parasitism (c) necessary before an average
female will gain in fecundity from abandoning a parasitized nest for 16 combinations of
predation and parasitism rates.

 

 

 

Predation (%)
Parasitism (%) 10 30 50 70
20 0.23 0.34 0.46 0.62
40 0.30 0.41 0.54 0.68
60 0.40 0.51 0.63 0.76
80 0.55 0.65 0.75 0.85

 

97

t =o.4
03 p()

0.2 -
0.1 .

o P s-.- 3’;_ ,1
-0.1 -
-o.2 -
-0.3 -
-o.4 -
-0.5 .
-0.6 -
07

20% 40% 60% 80% 100%

 

7 Q . C = 0.75
’ -5_c=050

 

 

 

0.3 T .‘ ---_.__--_ P027. Q_:§__--__4

0.2 _.‘ l
0.1 l

0 + .- -2 __ .5411: _. 8_;~:__:1~_ 1;%:._.:: __.__V_E_;::;_'-;- _5.

-0.1 l \

Change in Fecundity

-0 2 l \

-0.3 - \
0 4 i
-0.5 ] \\

-0.6 . "‘5
-0.7 L. ----_.__.. .__ ._ ______l

20% 40% 60% 80% 100%

Abandonment rate

Figure 4-9. Nest abandonment positively changes fecundity relative to zero abandonment
when the parasitism rate, p(t), is lower and the cost to parasitism, c, is higher, but has
negative effects at high parasitism rates and/or lower costs to parasitism. These results are
for 50% predation.

98

Numbers in Table 4-8 were estimated through minute manipulation of the value
of c until fecundity was equal at zero and 100% abandonment. Below this value
abandonment is not adaptive. For example, in a population with 50% predation and 40%
parasitism it will pay to abandon an average parasitized nest only if the reduction in
brood size (c) is greater than 54%. One important caveat is that that there is a range
below this threshold value where intermediate frequencies of abandonment result in
slight increases in fecundity such as seen for c = 0.50 at 40% parasitism in Figure 4-9.
This. is because at intermediate levels abandonment acts much like an increased predation
rate for parasitized nests, allowing some birds to escape parasitism. Furthermore
abandonment strongly affects changes in fecundity only at ﬁ'equencies close to 100%.

Season length and Re-nest Frequency- Two ﬁnal series of deterministic
simulations explored the effect of re-nesting frequency on seasonal fecundity. One tested *
the length of the breeding season while the other tested the level of re—nesting frequency.
To modify the length of the season the re-nest frequency was altered such that no more
nest attempts were made after the given ﬁnal day of the season. Five days prior to the
last day r(t) was 1.0 (or all failed nests were replaced) and linearly declined to zero on the
ﬁnal day. Fecundity estimates were generated for four breeding season lengths at 20%
intervals of predation and three levels of parasitism (Table 4-6). All other conditions
were held at standard levels except that ﬁfth nest attempts were allowed, with an average
brood size equal to that of third attempts for all sites (Table 4-3). To accommodate the
extended breeding season length, the total run time was extended ﬁom 100 days to 120
days to be sure that all nests were complete by the end of a simulation.

To test the level of re-nesting frequency, the initial value of r(t) was altered but

99

the length of time when r(t) declines to zero remained similar to that seen in Figure 2-3.
Four initial re-nest frequencies were tested: r(t) = 0.3, 0.5, 0.7, or 1.0. Exact values used
for plotting r(t) in this second series are provided in Appendix A (Table A-3). Results
were generated for four predation rates and two parasitism frequencies (Table 4-6). All
other parameters were held at standard conditions.

Table 4-9 indicates that the change in mean young ﬂedged between a 40 and 70
day season does not differ more than 0.03 young per female for all three parasitism rates.
Therefore, the ﬁequency of parasitism does not appear to inﬂuence the relationship
between length of season and fecundity. This result is based on an abandonment rate of
0.10. Altering the abandonment rate does change the estimates of fecundity. However,
the difference in fecundity did not vary from the pattern in Table 4-9 since increasing
abandonment causes a slight increase in fecundity at both season lengths (data not
shown). Also observed ﬁ'om this table is that the change in fecundity increases with
greater predation frequencies, suggesting that the length of season is only important when

nest predation is frequent.

Table 4-9. The change in mean young ﬂedged per female between a season length of 40
and 70 days for three levels of parasitism.

 

 

Predation %
Parasitism % 20 40 60 80
0 0.05 0.21 0.40 0.44
40 0.06 0.21 0.39 0.41
80 0.05 0.21 0.38 0.41

 

100

The length of season also plays a small but notable role in whether intermediate
nest predation can lead to positive changes in fecundity. A small positive change in
fecundity occurs if the length of the season exceeds 50 and 62 days for 20% and 40%
predation, respectively, for standard conditions at 80% parasitism (Figure 4-10). This
plot suggests that even the impact of 60% predation on fecundity would approach zero at
greater season lengths. In general, it appears that increasing the length of the breeding

season reduces the negative effect of nest predation on seasonal fecundity.

0.1

 

 

 

   

+ d(t) = 0.20
_._ d(t) = 0.40
+ d(t) = 0.60

Change in Fecundity

   

 

 

 

40 45 50 55 60 65 70
Season length (days)

Figure 4-10. All else equal, increasing the length of the breeding season reduces the
negative effect of nest predation, d(t), on the change in seasonal fecundity relative to zero
predation. Results are for 80% parasitism.

The relationship between nest predation and seasonal fecundity shifts from a
curvilinear to a linear pattern as the initial re-nest frequency declines (Figure 4-1 1). This
pattern remains consistent at different parasitism levels. The effect of predation becomes

increasingly more uniform across the entire frequency range when re-nesting becomes

less frequent (Figure 4—11). Since the positive change in fecundity is a result of this

101

3____- ____- _ _ a H
2 i -
1 _ _ - z _
—-a—'(t)=0-3
—e—"(t)=0-7
+r(t)=1.0
0 ; - _ ._ - ,5 _
E o 02 0.4 06 08
'U
C
3
0
OJ
u. 3 - p(t)=0.8 -

 

 

0.4 0.6 0.8

Predation, d(t)

Figure 4-11. The relationship between nest predation and seasonal fecundity shiﬁs from
curvilinear to a linear decline as the initial re-nest frequency, r(t), declines. Results are
shown for two levels of parasitism, p(t).

102

curvilinear pattern, at lower predation frequencies we expect that the potential for a
positive effect to nest predation to dissipate quickly as the initial re-nest frequency drops.
This is conﬁrmed in Figure 4-12. The initial re-nesting frequency must be higher than

r(t) = 0.967 in order for a population to gain ﬁ‘om predation at d(t) = 0.3.

Change in fecundity

 

.04 i..____---_ ----______ _ _ _ -
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Re-nest Frequency, r(t)

Figure 4-12. Any positive change in seasonal fecundity at 30% predation relative to zero
predation quickly dissipates when the re-nest frequency declines. The x-intersection
indicates the minimum initial level of r(t) necessary for a zero or positive change in
fecundity to exist at d(t) = 0.3. A third-order polynomial was ﬁtted to the data to generate
the line from which the x-intersection was calculated.

Sensitivity analysis- It seems warranted to evaluate which of the above parameters have
the greatest inﬂuence on seasonal fecundity. To do this requires the calculation of an
index of the relative effect of each parameter on fecundity that is comparable for all
parameters. This was done through the calculation of an elasticity index deﬁned in the

equation:

103

Where, E1 is the elasticity index, f. and f; are the two estimates of fecundity, and m and
n; are the corresponding values of the parameter for which elasticity is being calculated.
The procedure is to run the model using the initial value of the parameter (m) and to ﬁnd.
the estimate of fecundity (f1). This is repeated for the second value (n2) to ﬁnd f2.
Therefore, the elasticity index is a measure of the relative change in fecundity relative to
the change in the parameter in question. Parameters with higher magnitude EI values
have a greater impact on the estimates of fecundity than those with lower values.

This analysis was performed on the following parameters, predation d(t),
parasitism (declining) p(t), declining predation dd(t), cost to parasitism c, differential
predation 8, abandonment a, and initial re-nest frequency r(t). All parameters are
deﬁned in the model as frequencies that can range from zero to one]. Also, as observed
in previous results, most parameters vary in their relative impact on fecundity depending
‘on their value. Therefore, elasticity of each parameter was calculated for three segments
that divided the frequency range into thirds. The lower third was a comparison of the
fecundity estimates for parameter values of 0.1 and 0.3, the middle third comparison was
between 0.4 and 0.6, and the upper third was between 0.7 and 0.9. Most parameters also
vary with respect to the level of predation and parasitism. Therefore EI values were
calculated for c, 8, a, and r(t) for d(t) = 0.3, and 0.7 and p(t) = 0.4 and 0.8. EI values for

predation (including declining predation) and parasitism were calculated at p(t) = 0.4 and

 

' The exception is 8, which can include all real numbers, but realistically would range between 0 and l for
most situations where parasitized nest failed more oﬁen.

104

0.8 and d(t) = 0.3 and 0.7, respectively.

Table 4-10 summarizes the top eight most inﬂuential parameters for each
frequency range. Negative EI values indicate that increasing parameter values will
decrease fecundity. As expected, the inﬂuence of a parameter generally increases with
increasing frequencies of that parameter. In the lower and middle third frequencies
ranges, results show that fecundity is most sensitive to the frequency of re-nesting and the
effect of parasitism, both through the frequency p(t) and cost (c) Re-nest frequency is
more inﬂuential at higher predation and parasitism. F ecundity is more sensitive to
parasitism frequency at low predation rates and most sensitive to the cost of parasitism at
high parasitism and lower predation rates. Both of these patterns show that predation
can, under certain circumstances, reduce the overall impact of parasitism. Similarly,
predation is more inﬂuential at low parasitism rates and is less inﬂuential when predation
rates decline. Nest predation is the most inﬂuential parameter in the upper frequency
range. Regardless, the inﬂuence of parasitism remains strong at high frequencies.
Differential predation and abandonment were consistently the least inﬂuential parameters
in predicting fecundity. This analysis supports the observation that fecundity is more
limited by parasitism at low and intermediate frequencies, and is more limited by

predation at high frequencies.

Discussion
Acting alone nest predation does not yield a positive change in fecundity under standard
testing conditions. However, nest predation, in spite of lowering daily nest survival can

increase fecundity when speciﬁc conditions are met. These conditions include parasitism

105

 

 

6.5" 555

N555- 95 60565055 38:80
5.5 n 555
6.5 u 555

55.5 95 $550325 5605-35
5.5 u 555

med- 35 865266565
6.5" 555

555.5- 35 .8656865
5.5 u 85
6.5 u 55

53. T 5 .86566555 8 5600
m5 n 35
6.5 n 555

m5 _ . 5 - o .Eﬁzmﬁa 8 5650
6.5 u 555.

365.5- 95 525650.55
«.5 n 555

$5.7 85 523585

_m 8553

w5md-

mmmd

mmmd-

52.5-

mmmd-

oovd

256.?

£36..
E

”O " Ova
85.. 50:30.55

5.5 u 555
6.5 u 85
35 $582585 6605-35

5.5 u 85
35 5555585

5.5 u 85
35 56556565

6.5" 555
85 .8626865

5.5 u 85
5.5 n 35
35 $553585 5625-35

5.5 u 85
6.5 u 85
5 .8656855 9 5600
6.5 u 555
6.5 u 35
5 556266565 3 5600

0:522

owSﬁ 55555525

wmod

wmod-

mood-

N555-

555.5-

555.5

555.5-

555.5
E

6.5 u 555
6.5 n 85
or $555585 562735

6.5 u 555
5.5 u 85
5 .5696865 9 6650

5.5 u 85
w.O H Cvﬂ
o €655,855 8 5650

5.5 u 555
35 .86566565

65" 85
35 #:6266565

5.5 u 85
66.5 n 85
$5 $555585 6625-35

6.5 u 555
w.o .I.I CvQ
5 .86266565 8 6600
5.5 n 85
6.5 u 555
35 $283525 5605-35
.8264

 

.35 5:6 35 50.5 6_o>o_ 2.55856 .3 586: 85 652085565 .2882: .35 .8656855 555 95 5255525 .55 _26—
oﬁ :o $555505 52.56.» b53565? 58586.35 .omcﬁ 5555525 8 Max:586 6525:8565 _gcoscﬁ 5658 Ems 59 5:5. .536 0355.

106

generally greater than 60%, a cost to parasitism greater than 75% reduction in brood size,
greater predation on parasitized nests, and lower predation and parasitism for re-nests.

This positive change in fecundity is more likely to be observed when the breeding
season is longer with a high frequency of re-nesting and when nest abandonment is
infrequent. Most importantly, the positive change exists only when predation is generally
lower than 40%. Not all of these conditions must be met simultaneously for a positive
change in fecundity to exist. The two most essential conditions are frequent and costly
parasitism. Given just these two conditions, a small non-signiﬁcant positive change in
fecundity can occur at or below 40% predation if the probability that a re-nest will be
parasitized is lower (Figure 4-3). Other conditions only enhance this change and in some
cases make it statistically signiﬁcant (Figure 4-5).

A more striking result of these simulation experiments is that both parasitism and
predation act in a complimentary way to reduce the effect of the other as each increases.
This antagonistic effect is more evident for predation. There appears to be no change in
fecundity for predation frequencies between zero and 30% when parasitism rates are
approximate 70% or greater (Figure 4-2). The range of predation frequencies at which
this occurs expands as the above conditions are met. Even predation frequencies as high

as 70% can have little inﬂuence on fecundity (Figure 4-6).

Biological signiﬁcance of ﬁndings- The positive effect of nest predation is not likely to be
biologically signiﬁcant in most situations since the change in fecundity tends to be
relatively small. One exception is when that small positive change is sufﬁcient to elevate

productivity over the sustainability threshold. For example, if annual survival rates are

107

0.70 for adults and 0.35 for juveniles, a self-sustaining population with a 1:1 sex ratio
would require 1.7 ﬂedglings per female per year. For the conditions observed in Figure
4-6, fecundity increases from 1.63 to 1.78 young per female. Assuming the estimates of
adult and juvenile survival are realistic, this small increase effectively switches the
population from unsustainable productivity without immigration to a self-sustaining
population. However, this conclusion is highly speculative because it is based on a
number of assumptions and conditions being met and ignores possible difference in
ﬂedgling survival.

On the other hand, the effect of a reduced or no change in fecundity due to
predation at high parasitism is likely to have signiﬁcant biological consequences. This
antagonistic relationshipultimate leads to weaker selection against both mortalities.
Parasitism effectively reduces any selective pressure against nest predation if the overall
frequency of predation is low. If cowbird abundance declines in the future, naturally
higher predation may exist in the population due to the previous lack of selection to
reduce predation of nests. Alternatively, nest predation, by reducing the impact of
parasitism, has a similar effect of reducing selection for a host to evolve a defense against
parasitism. This may contribute to the absence of cowbird rejection observed in many
hosts despite severe costs to parasitism (Rothstein and Robinson 1994).

Current thought is that nest predation has the greatest impact on avian
productivity (G&P, S&W, but see Woodworth 1999). While this may still hold true for
many bird populations, this is clearly not the case for those populations suffering severe
parasitism while only experiencing moderate predation rates. Since observed frequencies

of predation in local populations are generally near 50% (Ricklefs 1969), we can expect

108

parasitism when present and costly to be the limiting factor on productivity. Regardless,
since the intensity of nest predation and parasitism is not ﬁxed spatially or temporally,
the relative impact of both will continually shift in populations. Moreover, results
consistently show that both mortalities have a strong inﬂuence on seasonal fecundity.
Therefore, proper management to recover an avian population must simultaneously
reduce both nest predation and brood parasitism to effectively improve reproductive
success. The best strategy to do so is to manage the habitat and landscape (Donovan et

al. 1995a, Martin 1992a, Askins et al.1990).

Conditional simulations- Clearly, the interaction of nest predation and brood. parasitism is
inﬂuenced by an array of secondary conditions. Whether these conditions are likely to be
observed in natural populations is of obvious concern. For example, as stated before
most host species exhibit differences iii level of cost to parasitism. Some, such as the
Northern Cardinal can successfully raise a brood including a cowbird that is not
signiﬁcantly different in size to non-parasitized nests (Eckerle and Breitwich 1997).
Other species like the Kirtland’s Warbler (Dendroica kirtlandii) (Walkinshaw 1983),
Prairie Warbler (Nolan 1978), or certain ﬂycatchers (Sedgewick and Knopf 1988,
Walkinshaw 1961) lose most or all young when parasitized. Therefore, the consequences
brood parasitism will have on productivity will be species-speciﬁc.

The frequency of parasitism and predation are also highly variable in natural
populations. Local frequencies for brood parasitism vary anywhere from zero to nearly
100% (Elliott 1978, Peck and James 1987, Robinson 1992, Robinson et al. 1995a and b).

Frequencies of parasitism at or above 50% are a common occurrence for many cowbird

109

hosts nesting in forest fragments and are typically highest along ﬁeld-forest edges (Gates
and Gysel 1978, Brittingham and Temple 1983, Robinson et al. 1995b, Donovan et al.
1995a). Every studied population of passerine experiences some level of nest predation.
The frequency varies greatly depending on habitat, location, and an array of other
conditions (Martin 1993a). Martin (1993b) has observed frequencies as high as 80%, but
more commonly the frequency of predation ranges between 40 and 60% for open—cup
nests (Ricklefs 1969).

Given this, a population with a high frequency of parasitism and moderately low
predation may be common. However, east of the Rocky Mountains, the frequency of
nest predation and parasitism are both positively correlated with habitat fragmentation
(Gates and Gysel 11978, Wilcove.1985). ”In this region, one is more likely to ﬁnd elevated
nest predation in the same population that has high parasitism. Such a relationship is not
conducive to the occurrence of a positive change in fecundity at intermediate levels of
nest predation. Alternatively, current research on western populations suggests that
predation and parasitism may be negatively correlated in naturally fragmented habitats
common to this region (Tewksbury et al. 1998). Therefore, is seems plausible that high
parasitism and low predation is likely to exist locally in some western regions. Finding
those conditions that lead to a null or positive effect to predation in natural populations
may prove challenging in eastern populations but perhaps less difﬁcult in western
populations. Ultimately, other auxiliary conditions must also have the potential to exist
to increase the probability that this interaction does occur naturally. The following
discussion addresses these speciﬁc secondary conditions.

Diﬁferential predation- How the overall frequency of nest predation is partitioned

110

among nests seems to play a limited role in the consequences of nest predation and
parasitism. While fecundity is relatively insensitive to this parameter, increased
predation of parasitized nests does produce a positive change in fecundity that would not
have existed if predation rates between nests were equal (Figure 4-5). As one might
expect, if we increase predation on parasitized nests relative to non-parasitized nests we
observe a gradual increase in seasonal fecundity (Figure 4-4). This is because predation
is occurring on those nests that can beneﬁt from nest failure.

The reverse trend, where increasing predation on parasitized nests results in
depressed fecundity, is expected only at high rates of parasitism and predation. This
occurred at 80% parasitism and 60% predation (Figure 44). Under these conditions,
parasitized nests make a greater contribution to fecundity despite their low brood size
because there are so few successful non-parasitized nests. Increasing the failure rate of
parasitized nests at this point will naturally result in a decline in fecundity. For any given
level of cost to parasitism, there is a threshold frequency for predation and parasitism,
where greater predation on parasitized nests will no longer increase seasonal fecundity.
As the cost to parasitism increases, this threshold will shift to higher frequencies.
Therefore, for those species that are among the most concern due to parasitism (i.e. suffer
severe consequences to parasitism) this reverse trend is not likely to occur oﬁen.

Are parasitized nests more likely to be depredated? Increased predation of
parasitized nests exists in the mid-Michigan YWAR population (Figure 3-2). However,
as discussed in Chapter 1, this pattern is not consistent in the literature and varies greatly
from one study to another. In one classic study, parasitized nests of Song Sparrows failed

more often (Nice 1937). On the other hand, a recent study of the Song Sparrow found

111

greater nest predation for non-parasitized nests (Arcese et al. 1996). Variation in the
pattern of nest predation between nests is widespread, and suggests that whether
parasitized nests are depredated more frequently depends on a variety of factors that may
include the composition of the predator community, behavior of the local cowbird
population, or habitat characteristics. Therefore, this condition is likely to exist locally
but vary both spatially and temporally.

Declining predation- Greater nest success for re-nests means a reduction in the
overall cost to predation. Since there is less cost from predation to overcome, it can be
expected that this pattern would increase the likelihood of a positive effect to nest
predation. Simulation results conﬁrm this expectation. Slowly reducing predation to half
its initial rate has one main result. The rate of decline (or slope) in fecundity as the initial
predation rate increases is slower (Figure 4-6). This is also supported by the reduction in
- sensitivity of fecundity when predation declines (Table 4-10: upper range). Ultimately,
this results in a wider frequency range where predation has no effect on fecundity.

Results from the YWAR study seem to suggest that lower nest predation for re-
nests does not exist (Figure 3-3). According to model simulations, the cost of parasitism
would need to exceed 80%, and there would need to be no change in nest predation
between attempts before a detectible increase in fecundity would occur at intermediate
nest predation rates in the studied YWAR population. If overall predation was half that
measured in the ﬁeld, or about 27 %, parasitism would need to exceed 90% for no effect
of predation to exist across a broad range.

However, simulations found that no level of nest predation would lead to a

positive change in fecundity, even at extraordinary high parasitism, when predation of re-

112

nests compared to ﬁrsts nests are similar to that observed in the YWAR population. This
is the case even when the cost to parasitism was 100%. It is clear from this observation
that the success of re-nests is instrumental in the manifestation of a positive change in
fecundity. Increased predation on re-nests appears consistent across years (Figure 3-3),
suggesting an inherent causal mechanism within the population. Efforts to identify the
reason for this pattern are the focus of the following chapter (Chapter 5).

Cost to parasitism- The success of parasitized nests is the central component of
this system. Fecundity is strongly inﬂuenced by the cost of parasitism (Table 4—10).
Ultimately, the effect brood parasitism has on seasonal fecundity depends on the impact
raising a cowbird has on the host. If this impact is small then the cost of parasitism is
small and nest predation will have similar costs for parasitized and non-parasitized nests.
Results of these simulations show that the cost of parasitism must by high. In fact, if
nests have a constant equal probability of failure then the cost must be nearly 80% of the
initial brood size (Table 4-7). In other words, if a host can typically ﬂedge 5 young when
not parasitized, they must ﬂedge only one young when parasitized before predation will
increase fecundity for the entire population.

Though this cost seems high this is not uncommon. The Prairie Warbler
population studied by Nolan (1978) would meet this condition, as would several other
species including the Least Flycatcher (Empidonax minimus) (Briskie and Sealy 1987),
Black-capped Vireo (Vireo atricapillus) (Gryzybowski et. al. 1986), Solitary Vireo
(Vireo solitarius) (Marvil and Cruz 1989), Willow Flycatcher (Empidonax traillii)
(Sedgewick and Knopf 1988, Walkinshaw 1961), and Bell’s vireo (Vireo bellii)

(Goldwasser et. al. 1980) (also see, Graham 1988). In fact, many populations suffer

113

severe costs to parasitism, especially when multiple-parasitism is frequent (Robinson et
al. 1995a). This threshold level of cost drops as other conditions are met such as
increased predation on parasitized nests (Table 4-7). Therefore, the group of species for
which this interaction could be relevant might be limited to those having high parasitism
costs or populations where multiple-parasitism per nest is frequent, but certainly does not
exclude all populations. In some cases, species of most concern, such as the Kirtland’s
Warbler, apparently experienced sharp population declines partly because of the costly
effect of parasitism, making the interaction of predation and parasitism very relevant to
their management. Current management practices do not explicitly take into account the
potential importance of this interaction (see below).

Nest abandonment- It was initially expected that nest abandonment would
eliminate the positive change in fecundity resulting from predation, since fewer
parasitized nests would survive the abandonment stage to be exposed to predation that
may beneﬁt them. However, this was found not to be the case for most intermediate
frequencies of abandonment (Figure 4-8). In fact, fecrmdity was not greatly inﬂuenced
by abandomnent rate. The reason for this is not entirely clear, but one reason may be that
abandonment only occurs during a short window of time (during the egg-laying stage),
while the time frame during which predation will occur includes this stage plus the much
longer incubation and nestling stages.

Abandonment becomes effective only when sufﬁciently high so as to greatly
reduce the number of parasitized nests that survive egg-laying. Beyond this point
abandonment begins to have a much greater effect on estimates of fecundity. This

threshold point lies greater than 80-90% abandonment under conditions simulated here.

114

Prior to reaching this threshold, abandonment has only a minor effect in reducing the
beneﬁt to predation as reﬂected by its consistently low sensitivity. Therefore, the fact
that many populations of small songbirds show some level of abandonment (Rothstein
1975) will not eliminate them as candidates for observing a predation interaction. Only
those species that show a very strong afﬁnity to abandoning parasitized nests (rejectors)
will never gain in fecundity at intermediate nest predation.

Another unexpected, and perhaps more interesting, result of the abandonment
simulations is the conclusion that nest abandonment is not always an adaptive strategy to
cope with brood parasitism. This directly contradicts results presented using Schmidt and
Whalen’s model (1999). They found abandonment was not necessarily adaptive for
double-brooding songbirds but was always adaptive in single brooding populations. Nest
abandonment is frequently cited in the literature as an adaptation to brood parasitism and
is widespread in hosts (reviewed in Rothstein 1990). However, it is rarely an ubiquitous
response to parasitism and for some populations is quite rare despite an apparently high
potential for selection imposed by severe costs to parasitism (Rothstein 1990, Rothstein
and Robinson 1994).

Several authors have pondered why hosts vary in their ability to discriminate and
reject parasitic eggs (reviewed in Winfree 1999). Two general hypotheses, equilibrium
and time lag, have been proposed to explain the maintenance of this variability. The time
lag hypothesis suggests that abandonment or other forms of ejection have not become
common because the population has not had sufﬁcient time for genetic variants to appear
and increase in the population (Rothstein 1990). As I have already stated, the

antagonistic relationship between predation and parasitism may contribute to this lag time

115

by reducing selectivity. The equilibrium hypothesis suggests that rejection can be non-
adaptive if rejection incurs greater cost than acceptance. The most obvious cost to
abandonment is the cost of losing young that would have ﬂedged if the host accepted the
parasitized nest. However, costs can also be measured by the potential for success of re-
nests following abandonment. F or example, in one host-parasite system, the parasitic
bird is known to retaliate against those hosts that reject parasitized nests by depredating
subsequent nests (Soler et al. 1999, Robert et al. 1999). Recognition errors where hosts
abandon non-parasitized nests may also result in equilibrium (Rothstein 1982).

Results presented here suggest another plausible mechanism to support the
equilibrium hypothesis. Abandonment is not adaptive when the ﬁequency of predation
and parasitism is high, even when the cost to parasitism is also high (Figure 4-8).
Abandonment is not beneﬁcial because the risk of failure for re-nests remains high. In
order for nest abandonment to be adaptive, the potential reproductive success of the re-
nest must be greater than that of the parasitized nest. As already stated, the frequency of
parasitism and predation are both positively correlated with habitat fragmentation.
Therefore, conditions where abandonment is likely to be non-adaptive can be expected to
be common in ﬁagmented habitats especially in localities east of the Rockies. However,
the relationship between predation and parasitism is reversed in western populations
(Tewksbury et al. 1998). Therefore, abandonment strategies will vary depending on
region for those species with broad distributions. For these songbirds, mixing would lead
to an overall equilibrium frequency to abandonment, while locally sharp differences in
strategy may arise. Naturally, all this depends on the cost of parasitism. Abandonment

will be adaptive if the cost to parasitism is sufﬁciently high as to incur a greater cost than

116

the lower reproductive success rate of re-nests.

Season length and re-nestingfrequency— Nest predation has less of an impact on
fecundity in a longer breeding season (Figure 4-10). This result is consistent with others
(Grzybowski and Pease, in press). If a songbird has more opportunities to nest following
failure a re-nest is eventually likely to succeed. Longer breeding seasons provide more
opportunity for re-nests to ameliorate the negative effect of nest predation. However, the
relative contribution of length of the season on seasonal fecundity is not large. A more
than 40% increase in the length of the season increases fecundity per female by 0.41-0.44
young (Table 4-9). Also, if we were to extend the breeding season of the YWAR
population by 20 days, through appropriately adjusting the observed re-nest frequency
(Figure 2-3), the resulting increase in fecundity would be between 0.11- 0.22 young per
female for the range of predation observed in the ﬁeld. These are small changes in
fecundity relative to those resulting ﬁom equivalent percent changes in nest- predation
and brood parasitism. Season length is likely to play a greater role in seasonal fecundity
in multiple brooding species, since those that were successful will have more opportunity
to start second broods.

Alternatively, seasonal fecundity is clearly very sensitive to the frequency at
which failed attempts are replaced (Table 4-10). The likelihood that a nest will
eventually succeed is greater when failed nests are replaced more frequently. Moreover,
higher re-nesting frequencies increase the curvilinear relationship between predation and
fecundity (Figure 4-11). Grzybowski and Pease (in press) found a similar result and
suggest that this curvilinear relationship is because increasing predation does not produce

proportionate decreases in seasonal fecundity. The reason being that when predation is

117

more frequent it occurs, on average, earlier in the nesting cycle thereby utilizing less of
the breeding season. Therefore, more re-nest attempts can be made. Increasing the
frequency of re-nesting ampliﬁes this relationship. A high frequency of re-nesting seems
to be an adaptation to ameliorate the negative effects of nest predation.

An obvious question is if re-nesting can reduce the effect of predation why do
birds eventually stop re-nesting? The answer involves additional long-term costs to re-
nesting. The survival of ﬂedglings from re-nests may be lower if food supply diminishes
in the latter part of the breeding season. These young also have less time to prepare for
the southern migration and thus may have lower over-wintering survival compared to
ﬂedglings from original nests. Moreover, there is a well-recognized trade off between
fecundity and survival (Williams 1966, Linden and Moller 1989). In general, greater
fecundity comes at the cost of lower adult survival. This relationship should lead to an
optimal breeding season length that limits the number of nest attempts. Birds must also
cope with other necessary stages of their annual cycle such as the need to molt feathers
and/or prepare for long migrations. Extending the breeding season would shorten or

eliminate the time necessary for these activities.

Comparison with existing models- FLEDGE differs in several ways from two similar
existing models (Table 4-11). Both existing models are superior to F LEDGE in that they
can model species that ﬂedge more than one brood in a season. On the other hand,
FLEDGE has notable advantages over these nesting models. The most striking is that
FLEDGE does not assume that brood size for all nest attempts remains constant. This

assumption is inconsistent with Nolan’s (1978) data and the YWAR data (Table 4-3). It

118

6:235:68 583% 35: 6352: 8.55:7. 555.5 .0655 .3 525mm 352: gun—33% 22: 25 555 5558 253 38:55:80 .5

525558 65:55:55.8 :02

8%

55:06.35 5:556:00
6558555 50 55:5: 65:55

558555
565— 55:35 Ho: :55 55 5:556:50

6558555
:5 .85 5556 5555 5:56:50

6565: :5 :05 6:556:50

85.8585 65 505.55

5:385 6wwo 5:338 55 558:5:
8 w:m5:ooo5 65mm 5055 565.555 :50

8%

.5583
5555:2555: E 555555 5555555
55599 6558855 55:82::

:5 :06556 50 55:2 35:5

558555 565— 55:35 :5,
36555555

55 3555 :5 $5555 55>
:55 525.» :5 5:556:55 5:585:55.

6558555 :5 85 5.5.56 5053 5:56:50

36556555 2553:: 85
56: .555 :56 :5 6565: :5 :55 5:556:00

6:556:55 5055:5558 :5 5:555

52559: 5556:5555 8 Z

02
5:5 656:?
55:25 55:55:35 65 55:55.5

£w:2 :06556 65:55
65:6: 55 5558:: 655:5

558855
565. 85:35 :5» .85: 5:556:50

558555 .65 5556 5055 5355.53

6565: :5 :55 5:556:00

5:5 656.8.» 533 50:55:56.5

5:628:55 252:2

5505 53:59

3:65:65

55855535

:550552 5820

86556855 8 5600

 

5:556:00 Ho_n5:5> 5 w:€o_5 3 55:56.9 .555:5> 85: 865565.55
6:555:55 365582555: 5 5:559 65:5 63:2, 6258 68:252.“
5:556:50 535t5> @5525 .3 555.550 5555255» 85.— 555585
523 5 525556 36355566 5 385 5055.5 552558

 

5555:5555 5:85.56 3505585 :65 2552: $56555 ooh: 59555 6555:5665 5:5 55358585655 50 6:855:86 55M .5 75 5355.

119

also differs from both in how it models re-nesting behavior. Pease and Grzybowski
(1995) allow re-nesting to occur for a ﬁxed length of time, while Schmidt and Whelan
(1999) limit the number of re-nests a bird will attempt. Using F LEDGE, the number of
re-nests a female can make and the length of time during which re-nests will be made are
both incorporated into the probability of re-nesting. Furthermore, unlike the S&W
model, FLEDGE does not assume predation and parasitism rates must be constant.
Finally, FLEDGE takes advantage of STELLA’s® graphical input devices to greatly
simplify data entry, manipulation, and mathematics (Constanza et al. 1998). Parameter
values can be deﬁned with the click of a mouse and many different simulations can be
performed in a matter of minutes. Neither of the other models has this feature.

While the P&G model is superior in ability to model a broad set of situations, this
advantage is at the cost of simplicity, thus limiting those who can quickly understand and
utilize it. Furthermore, Pease and Grzybowski (1995) freely admit that there are rarely
sufﬁcient data available to parameterize their general model, noting that even Nolan’s
(1978) outstanding data set is inadequate. To rectify this problem, they propose a simpler
model that assumes constant predation and parasitism, and assumes that the number of
young that ﬂedge from parasitized nests is constant regardless of number of cowbird eggs
laid, among other restrictions. FLEDGE inherently does not require any assumption
about the variability of nest predation and parasitism with regards to time, and all
necessary data to parameterize the model for the YWAR were attained from existing data
and from a low cost four year study. Thus FLEDGE offers a compromise between the
overly simple S&W model and complexity of the P&G model.

Both Schmidt and Whelan (1999) and Grzybowski and Pease (in press) have used

120

their models to evaluate the relative effect of predation and parasitism on seasonal
fecundity. In S&W’s analysis, they concluded that nest predation has a greater impact on
fecundity than parasitism. This conclusion seems contrary to the ﬁndings presented here
which suggest that fecundity is more sensitive to parasitism over a larger range of
frequencies. However, my simulations examined a slightly different set of conditions
than that of S&W. Most notably, my simulations dealt with birds that suffer at least 75%
reduction in fecundity when parasitized. Although less apparent, a similar pattern of
increased fecundity within a limited range of nest predation exists in S&W’s results when
the cost of parasitism was severe. They also found that the effect of nest predation was
diminished when brood loss was large in parasitized nests as shown by the vertical lines
of the isopleths (Schmidt and Whelan 1999: ﬁgure 2). They conclude that nest predation
is more important because the range at which this occurs is small.

Grzybowski and Pease (in press) also conclude that predation generally inﬂuences
seasonal fecundity to a greater degree than parasitism, but they are quick to point out that
parasitism becomes increasingly more important as the cost of parasitism increases.
Their analysis did not explore the possibility of a beneﬁt to nest predation. However, all
three models consistently found that the effect of predation is reduced with increasing
parasitism. Perhaps the main reason why predation is more important for a greater range
of parameter space in S&W and P&G’s results lies in the most signiﬁcant difference
between these models and F LEDGE. Both of these models presented results for double-
brooding songbirds. Since second broods occur later in the season when parasitism is
likely to be infrequent, nest predation incurs the greatest cost on these nests. Moreover,

re-nesting contributes an additional cost to multiple-brooding birds in that it shortens time

121

available to attempt second broods. This likely results in a ﬁnding that predation is more
inﬂuential. This is supported by S&W’s results for single-brood populations, where
parasitism makes a greater contribution to fecundity.

Variation in our conclusions is reﬂective of the importance of knowing the
conditions that exist in the population of concern. Which factor is more limiting will
depend not only on the frequencies observed in the population but also an array of other
conditions and life-history characteristics. As Grzybowski and Pease (in press) point out, i
the complexity of the interaction between brood parasitism and nest predation greatly
complicates the interpretability of nesting data alone. All available information needs to
be considered when trying to assess the factors inﬂuencing population growth in -
songbirds. The inﬂuence of brood parasitism and the resulting interaction with nest
predation is dependent upon the performance of the host when parasitized as well as the
female's ability to successfully re-nest following failure from nest predation. The
interactive effects of nest predation and brood parasitism will be species-speciﬁc and no
generalized conclusions for all species can be convincingly made. For example,
Woodworth (1999), using the model introduced by Pease and Grzybowski (1995), found
brood parasitism to be more important than predation to population viability for a tropical

endemic songbird.

Conclusions and Implications for Management and Research. - Results presented here
suggest that when brood parasitism exists and the cost to the host is sufﬁciently high,
seasonal fecundity is inﬂuenced more by brood parasitism than nest predation in single-

brooding songbirds. Moreover, moderate to low predation rates have little or no

122

detrimental effect on fecundity in populations experiencing frequent and costly
parasitism. These results support Brittingham and Temple’s (1983) conclusion that brood
parasitism and not nest predation is a leading cause in the decline of several species in
temperate North America.

Efforts should be made to minimize cowbird abundance in managed populations.
Combined with habitat management, this approach has proven effective in the past in
some populations (Walkinshaw 1983). However, this is not always the case (Robinson et
a1. 1993). Stutchbury (1997) found cowbird removal to have no detectible effect on .
fecundity of the Hooded Warbler despite dramatically decreasing the parasitism rate on
nests. Modeling results presented here or by others (Schmidt and Whelan 1999,
Grzybowski and Pease, in press) help to explain why the same management strategy can
give such divergent results. As one mortality factor is reduced, the other becomes more
limiting on seasonal fecundity. Moreover, seasonal fecundity becomes more sensitive to
nest predation when it exceeds 60% no matter what the level of parasitism. These are
results shared by all models.

Therefore, to reiterate the conclusion by others, the best approach to promoting
population grth is to reduce both causes of mortality. This is best accomplished by
preserving larger tracts of high quality breeding habitat (Martin 1992a, Donovan et al.
19953). When this is not a reasonable strategy, the use of models such as FLEDGE or
P&G can evaluate which management options will result in the greatest increase in
seasonal fecundity (for example see Liu et al. 1995).

Several questions remain unanswered that have direct implications for the

observed interaction between nest predation and brood parasitism. F LEDGE was not

123

designed to address lifetime fecundity, thus it is not clear what cost nest predation has
over the lifetime of a host. Although re-nesting may improve short-term seasonal
fecundity, it is plausible that there is a greater cost in lifetime fecundity (Linden and
Moller 1989). For example, some young that ﬂedge later in the season have been shown
to have lower over-winter survival (Lack 1954, Williams, 1966). Lower survival of
young from re-nests may cancel out any increase in fecundity due to nest predation.
However, there may be additional unaccounted costs to raising a cowbird as a result of . .
the extra effort parents must make to feed the young. Parasitism may shorten an adult’s
lifespan and host young that ﬂedge may have lower post-ﬂedgling survival compared to
young ﬂedged from non—parasitized nests. As discussed, differences in nest survival '
between parasitized and non-parasitized nests also remain unclear. Furthermore, I found
few studies that clearly identify the probability of re-nesting following nest failure and
how this changes as the season progresses (but see, Rooneem and Robertson 1997). To
better understand the impact of nest predation and parasitism on real populations, more

data are necessary on the frequency and consequences of re-nesting.

124

CHAPTER 5- ARTIFICIAL NEST EXPERIMENT

The frequency and success of re-nests can inﬂuence the outcome of predation-
parasitism interactions. As noted in Chapter 4, model simulations suggest that no level of
cost or frequency of parasitism will result in a positive change in fecundity at
intermediate predation, if re-nests have lower nest success. While YWAR do re-nest
frequently these re-nests are not as successful as ﬁrst nest attempts (Figure 3-3). Even so
re-nests make a signiﬁcant contribution to seasonal fecundity and thus are important to
sustain a population (Table 3-2). Therefore, identifying the mechanism(s) causing the
observed increased failure for re-nest was of great interest. Understanding the
mechanisms and differences by which re-nests survive could provide insight into
management of certain avian populations and could provide insight into the evolution of
nest site speciﬁcity.

Several hypotheses can explain the pattern of reduced success of re-nest attempts.
These can be divided into two general themes: those concerned with variation within or
between individual birds (Individual-based hypotheses) and those concerned with
variation in external factors beyond the inﬂuence of the breeding adults, such as attributes
of nest predators. These are not necessarily mutually exclusive since adult behavior can
be modiﬁed by external factors, and the actions and choices of individual birds may
inﬂuence some external factors, namely predator activity. These hypotheses are
summarized below and in Table 5-1.

Individual based hypotheses- A re-nest occurs because a prior nest attempt failed.
Attributes associated with nesting adults may very well increase the likelihood of failure.

It would seem then that these very same attributes would reduce success for any

125

 

568562 8558555

 

 

55685 >656 6855525 55685 565: 56553; .6565: 5585555 .8 5558::
5885 .8550 855525 56582 .8 65585 6:256, 25: 855525 5652 5585:5555 5:685 5652
.865: 58:85
.5555 :5 656555 558855 5:5 5>555 .85 585 823
855525 565: E 85558 5585855. 855525 565: .8 558825 55.8555 5555 85552.5
.585 :5 5558 5555 8 5625 .8555 5583 5565:
5885 5:5 5>555 5285 .8 685.5 855525 685>5 855525 8 8855. 885555 82 855525 ”65655855: 55655285525
85.885855 85555m5> 8 58> 656558555
:55 6558555 565: .8 686555855 5655-555 6858256558 85555wo> 8585558 5652
.6565: 5>555 556655556 665— .8 62.85.85
8 6655556 565: 5532 5>55 6565: 588:0 6565: 5>555 .6> 5880
25:55:58
8566555568 5:5 556655556 85.555 5558 5,858.0.
55.555 85555555, 5858886 .8 65855856 6858256558 85555w5> ”555555 058255 805
5655556 565: 555525 5>55 6565: 58830 6655536 565: 5>555 .6> 5880
2535858 856658368 5:5 885555 058255
55665836 853555 25.58 885555 88255 6858586558 55885550 ”555535 058255 855 ”65655855: 55655 05555555
653555.» 5586558
55585536 .5 855825 8 5.3655 558588550 6555:2275 656558550 852

 

55585536 .5 6852525 5:5 .855 25258
8 556: 5586 5.5 .8 6555565 25 855258 6565:-2 :o 555: 855525 565: :252w 25 8555555 8 656558555 .8 5:58:86 < ._-m 535.0.

126

subsequent nest attempt. Consequently, re-nests show greater failure because they are
composed of a greater proportion of nests by these inferior breeding adults. Thus the ﬁrst
hypothesis, which I will call the poor parental quality hypothesis, states: Re-nests, which
include only those birds that have already failed, are inherently composed of inferior
birds with a lower potential for nest success. The main assumption of this hypothesis is
that those qualities that contribute to lower nest success in some individuals do not vary
within a season. For example, birds do not “learn” from previous failure. Two main
qualities of breeding adults are plausible attributes related to nest success: parental
behavior and quality of the defended territory. These are not necessarily mutually
exclusive since adult behavior affects selection and defense of territories.

Parental behavior has been documented as a factor in nest success. The ﬁ'equency
of visits made to the nest (Martin 1992b), their age and hence experience (Saether 1990),
or how they behave in the presence of potential predators (McClean et al. 1986,
Cresswell 1997, Hatch 1997) are linked to nest success. Other attributes associated with
behavior such as nest placement and how well they conceal the nest, or how elusively
they approach their nest, can also contribute to nest success. In addition to parental
behavior, characteristics of the territory such as the diversity of vegetation structure
(Martin 1988a, Marzluff 1988) or the abundance of snags (Anderson and Storer) are also
known to inﬂuence nest success. Moreover, territories with large shrub patches may
contain more potential nest sites for a visually searching predator to investigate thus
reducing the risk of a single nest becoming discovered (Martin and Roper 1988). All
these attributes are likely to vary between individuals (Martin 1986) and thus all nest

attempts by one individual will have a different probability of success than those of

127

another individual.

An alternative individual-based hypothesis states that sub-optimal nest placement
of re-nests caused their lower nest success (nest-placement hypothesis). It is likely that
natural selection has acted to have birds choose nest locations that have the greatest
potential to minimize discovery by predators (Martin 1988b). YWAR have been shown
to select nest sites based on characteristics of the nest patch, preferring horizontally larger
shrub patches (Knopf and Sedgwick 1992). Thus it seems plausible that ﬁrst nests are
placed at a location within the territory that maximizes the chance of success.

Subsequent re-nests are less successfully because the better nest sites were already used.

For example, female YWAR may place her ﬁrst nest in the largest patch of
meadowsweet or rose that is furthest away from snags. However, if by chance this nest
fails, it is unlikely that she will re-nest near this location (pers obs.). In fact, those studies
that have examined future nest placement have found that re-nests following predation
tend to be placed spatially distant from the previous nest (Harvey et al. 197 9, Marzluff
1988, Howlett and Stutchbury 1997). This may occur because, once a nest is found,
predators might re-visit the general area around the nest (Martin 1992a, Sonerund and
Fjeld 1987). Thus a re-nest placed in the sarrre general area would have a greater risk of
becoming discovered. A bird may sacriﬁce quality of nest site by placing the re-nest in a
more remote location from the original nest. Because of the behavior of predators, the re-
nest has a better probability of success than a second nest placed in the higher quality nest
site. Abundance and dispersion of suitable nest patches within a territory will contribute
to this. For example, a greater discrepancy in nest success might be observed between

ﬁrst nests and re-nests if preferred nest plant patches are small or few and far between,

128

leaving re-nests to be placed in distinctly less suitable sites. Thus habitat and territory
quality are likely linked to this hypothesis as well.

Predator behavior based hypotheses- One alternative to individual based
hypotheses states that temporal changes in predator behavior or abundance results in
increased nest predation for re-nests. As state above, some predators may re-visit
previously discovered nest sites (Sonerund and F jeld 1987). All else being equal, this
behavior would seemingly increase the probability of failure for subsequent nest attempts
within the general area of the previous nest (the predator return hypothesis). It is also
plausible that predator nest searching behavior varies temporally and that re-nests occupy
a greater proportion of the total available nests during the peak foraging rate of dominate
nest predators (Peak activity hypothesis). Moreover, some predator communities show
density dependent nest predation (Martin 1993 a and b). A greater proportion of all
available nests (including other species) may be re-nests during the period of highest
active nest density (density-dependent hypothesis).

According to foraging theory most vertebrate predators that actively search for
speciﬁc prey should show density dependence (Martin 19933). As the number of nests in
a given area of all similar species increases, predators learn to associate a reward with
bird activity. Therefore we would expect to observe a relationship between nest
predation and the number of active nests if nest predators are actively searching out nests.
YWAR arrive and begin breeding earlier than many other migrants. Therefore, initial
nest attempts may be made during a period of lower nest density. It is also plausible that
some nest predators simply increase their intensity of foraging on nests later in the season

when more replacement nests are active. For example, circumstantial evidence suggests

129

that Blue Jays are a common nest predator. These birds begin breeding earlier in the
season than most Neo-tropical migrants (Brewer et al. 1991). The stage when most Blue
Jay nests are ﬂedging and young jays are foraging for the ﬁrst time may overlap with the
period of time when re-nests dominate the pool of active nests.

In this chapter I describe and present results from a ﬁeld experiment designed to
aid in discriminating between these alternate hypotheses. Since most of these hypotheses
are not mutually exclusive, a myriad of secondary measures were collected to support or
refute certain hypotheses noted above. These include both behavioral assessments and
territory quality assessments as described in Chapters 2 and 3. Many of these
comparisons have been presented in Chapter 3 but more speciﬁc contrasts, as described

in Table 5-1, are presented below.

Experimental design- Previously used nests of YWAR and American Goldﬁnches
(Carduelis tristis) (which have similar nests) were collected during the non-breeding
season for use as dummy nests (U SFWS collection permit MB008925-0). During the
breeding seasons of 1999 and 2000, active nesting tenitories were randomly selected and
a dummy nest, with at least two infertile Zebra ﬁnch eggs as bait, was placed in a random
location within these territories. Territory boundaries were determined through male
activity. Active nesting territories had a nest with at least one egg and frequent adult
activity determined ﬁ'om nest monitoring.

Dummy nests were placed in locations with qualitatively similar vegetation and
conceahnent as active nests and were not placed closer than 5 meters from active nests.

For example, if the active nest was placed in meadowsweet I tried to place the dummy

130

nest in another meadowsweet patch within the territory. If no other suitable
meadowsweet patches were available I selected another preferred nest plant such as gray
dogwood or rose depending on availability. In this manner, all factors that may inﬂuence
nest success were controlled except for the inﬂuence of parental activity. Dummy and
natural nests were monitored every 2 to 4 days for signs of predation. If an active nest
was depredated and a re-nest was built, the dummy nest was relocated to a new site
within the territory. Dummy nests were not replaced if depredated unless the active nest
also became depredated and a re-nest followed. Before a dummy nest was removed the
tape distance, side conceahnent, and overhead cover were recorded. The latter two
measures were not recorded during 1999.

In addition to the Zebra Finch eggs, one or two clay eggs shaped and painted to
resemble YWAR eggs were added to dummy nests. Since actual predators consuming
nests can differ from those observed to be present (F enski-Crawford and Niemi 1997),
nest damage (Best and Stauffer 1980) and markings on clay eggs were used to identify
predators (Haskell 1995). The following criteria were used to classify nest predators. If
the clay eggs remained, I examined them for any markings and indentations. Tooth
marks indicated a mammalian predator. I also classiﬁed a predator as mammalian if the
nest was dislodged from the support branches or tom or if fresh tracks were observed
nearby. I assumed avian predators if markings on clay eggs suggest the edges of a bill or
puncture by a bill or if all eggs were missing with no damage to the nest. I assumed
snake predation if only the zebra ﬁnch eggs were missing and no markings were observed
on the clay eggs. Based on these criteria, I determined that 50% (n =13) of dummy nests

were depredated by birds, 35% (n = 9) by mammals, and 15% (n = 4) by snakes.

131

Artiﬁcial nests are commonly used in studies of avian nest predation but have
some inherent problems associated with them (Wilson et al. 1998). This experiment
includes several modiﬁcations from traditional artiﬁcial nest studies to reduce these
problems. First, artiﬁcial nests were actually previous-year or in some cases earlier nest
attempts of YWAR. Some goldﬁnch nests were also used simply because they were
often difﬁcult to distinguish from YWAR nests during off-season collection. Using the
same or very similar nests will eliminate any bias in predation resulting from the size,
shape or color of the nest that may inﬂuence detection. One potential problem is that the
scent of these nests differed. Goldﬁnches will defecate on the rim of nests (Watt and
Dimberio 1990). Although visual evidence of this was absent in weathered nests,
goldﬁnch nests may be more detectable than other nests. However, these nests
represented only a small proportion of those collected since goldﬁnches were far less
abundant on the sites. Thus any affect this might have was likely small.

An additional improvement from traditional artiﬁcial nest studies was the use of
zebra ﬁnch eggs, the dimensions of which closely match those of YWAR eggs.
Consequently, potential predators that may be able to consume one egg size and not
another are not excluded from either set of nests (Haskell 1995, Lindell 2000). Many
artiﬁcial nest experiments also place dummy nests on the ground. However, some
studies reveal differences in nest predation between ground nests and elevated nests
(reviewed in Martin 1992a). In an attempt to match all conditions of nest placement with
that of real nests, I placed dummy nests in elevated positions in preferred nest plants
(Table 4-5). Dummy nests were sewn in to branches using a needle and dark thread, in

order to secure nests into the shrubs. Lastly, dummy and active nests were monitored

132

simultaneously for direct comparisons.

Zebra ﬁnch eggs were provided by the avian research lab on the campus of
Michigan State University supervised by Dr. Juli Wade. The supply of eggs proved to be
the limiting factor in sample size. Dr. Wade’s lab supplied about 100 eggs in each
season. With the inevitable breakage of some eggs during placement, only about 40
artiﬁcial nests were used in each year.

By comparing nest predation rates of natural versus dummy nests, I was able to
determine which factors seem likely to inﬂuence nest success. For example, greater nest
predation on natural nests would suggest that parental activity inﬂuenced a predator’s
ability to locate them. If this pattern persisted for additional nest attempts, one could
conclude that certain birds have poor nesting skills. Increased predation of dummy nests
that are on territories of less successful birds would suggest that poor territory quality
affects seasonal nesting success. Finally, any temporal change in intensity of nest
predation on dummy nests could reveal the impact of changes in predator foraging rate
during the breeding season. Speciﬁc predictions from experimentation and data collected

are outlined in Table 5-1.

Results- Overall nest predation rates did not differ signiﬁcantly between dummy and
active nests (Figure 5-1). Parental behavior was associated with nest fate during nest
construction but the relationship was weak and practically absent when corrected for
experiment-wise error (Table 3-10). These results indicate that adult behavior at the nest
did not inﬂuence the probability that a predator would discover a nest. However, dummy

nests on territories for which the active nest also was depredated had a much greater

133

failure rate (Figure 5-2). This result supports the poor parental quality hypothesis. Given
that dummy nests lacked parental activity it seems likely that characteristics of the
occupied territory that both the active and dummy nest shared caused greater nest failure.
This could have been speciﬁc nest placement or overall territory quality.

Alternatively, it is also possible that greater predation on dummy nests in some
territories was caused by predator behavior. If predators do in fact return to the location
of a previously discovered nest,.other nests in that immediate area would seemingly have
a greater chance of becoming discovered. Unfortunately, this experimental design cannot
discriminate this from the effect of territory quality. To do so would have necessitated
simulated nest predation by humans and testing for differences in nest success between
re-nests following human predation and natural predation events. I elected not to
increase predation artiﬁcially to avoid undue mortality in the population.

For those territories in which nest predation occurred on both the active nest and
dummy nest, the mean number of days between predation of each nest was four. No
indication of skew in the data existed. Of the 20 nests for which this occurred, the
distribution was almost evenly divided by the mean. Predation events occurred less than
4 days apart in ten cases and eight occurred more than 4 days apart. This relatively
uniform distribution suggests that nest predation was random. This is likely biased
toward a lower mean time period since many dummy nests were removed within a week
after active nests were destroyed in order to accommodate a reénest attempt. If predators
do return to the location of a previously discovered nest, it is not known for how long of a
time interval. One would assume that they would do so most ﬁequently during the same

day. A greater time interval between predation events leads to a greater likelihood that

134

0.9 -1
0.8 .
0.7 ;
0.6 i.
0.5 . i i
0.4 -
0.3 .
0.2 ;

0.1 -55-- -5..- -5. -___.._ ~ .
active dummy

Nest predation rate

Figure 5-1. Mayﬁeld estimates of overall nest predation were nearly identical between
paired dummy and active nests (n = 81) (X2 = 0.0032, p = 0.96). Error bars denote
standard error.

0.9 5 5 5 ,5. ., ,5, I
0.8

0.7 .1 f
0.6 r

0.5 J

0.4 . i
0.3

1
0.2 .1

0.1 ___ ____ e _f _ g
900d bad

 

 

Nest predation rate

Figure 5-2. Dummy nests on territories of successful active nests (good, u = 42) had
signiﬁcantly lower predation than dummy nests on territories of active nests that failed
(bad, 11 = 39).

135

each event was independent. An average of four days between predation events appears
sufﬁciently long enough to suggest the predation events were independent. However,
there is no way of knowing this for certain with the data available.

There is no evidence to suggest that abundance or frequency of nest predation
varies within the season. A comparison of predation rates on dummy nests placed during
the ﬁrst half of the season to those placed during the second half of the season found no
difference (Figure 5-3). There is also no evidence to suggest that nest predation rates
ﬂuctuate with nest abundance (Figure 5-4). These results suggest that increased predator
activity or density dependent predation does not exist in this habitat and thus cannot
explain the greater predation on re-nests. While in the ﬁeld, no noticeable temporal
change in abundance of suspected predators such as Blue Jays or crows were detected. I
regret not making point count and scent station surveys of potential predators during the

season, but time and human resources available were limited.

0.8 5 55555 55.5 , e 5 , 55 ,
0.7 I ~

312i i i

0.4
0.3
0.2
0.1

__l___

Nest predation rate

_.._ .L_5' _ __l_5___L_

Figure 5-3. Nest predation rates of dummy nests introduced in the ﬁrst 25 days (early)
did not differ from dummy nests introduced in the latter half of the season.

136

y = 0.5023x 4 39.519

 

Number of Nests
8

1o 9 .

55555 . ‘ ,. 5 . .

10% 30% 50°o 70% 90%
Predation

55, l 55‘? 7

Figure 5-4. A linear regression analysis found no signiﬁcant increase in nest predation in
relation to nest abundance. Points represent the predation rate for a single site by week
(years pooled). The linear regression line and equation are included.

Like active nests, nest concealment of dummy nests inﬂuenced nest success. An '
AN OVA found depredated dummy nest had signiﬁcantly longer tape distances (Table 5-
2). Side cover was not signiﬁcantly different between successful and depredated dummy
nests, but mean side cover of depredated nests was lower (Table 5-2). Similarities with
active nest concealment indicate that predators do seem to be opportunistic (especially
avian predators). In other words, predators do not appear to actively search out nests but
depredate nests when accidentally discovered.

Tape distance can be an indicator of territory quality. Dummy nests on territories
of depredated active nests had signiﬁcantly longer tape distances than those on successful
territories (F = 4.345, df = 81, p = 0.040). Dogwood species tend to be less dense with
respect to branches and foliage and thus do not obscure nests as well (Figure 5-5). Thus,

the plant composition of a tenitory can inﬂuence tape distance and hence overall

137

concealment of nests. Territories with an abundance of meadowsweet, rose, and other
dense shrubs tended to be more successful because these shrubs obscure the view of the

nest better than dogwoods or other more open shrubs such as poison sumac.

Table 5-2. Means (2% SE) of tape distance and overhead cover were signiﬁcantly different
between dummy nests that were successful versus depredated. In all cases successful
nests were more concealed.

 

 

 

Dummy Nests
Successful Predated DF F-stat p-value
Tape distance (m) 2.62 :1: 0.13 3.64 i 0.21 82 18.255 < 0.001
Side cover (%) 64.34 :t 2.91 59.55 i 2.42 36 1.607 0.213
Overhead cover (%) 75.26 :t 3.19 62.16 :t 4.26 36 6.077 0.019

 

Patterns of vegetation presented in Chapter 3 suggest that ﬁrst nest attempts were
placed at locations within a territory with a greater potential for success. For example, 5
ﬁrst nest attempts were located in a greater proportion of meadowsweet and rose (Figure
3-8), both of which also show tendencies for greater nest success, especially rose (Figures
3-7, and 3-8). First nest attempts were also surrounded by fewer trees and snags, which,
again, were associated with nest predation (Figure 3-12). And while not signiﬁcant, re-
nests were increasingly less concealed than previous attempts (Figure 3-5). These
observations lend support to the nest placement hypothesis. However, previous analyses
examined changes in variables between individuals and not within individuals. These
variations could also be caused by a change in the sample of individuals between nest

attempts. For example, the mean increase in number of snags around re-nests could

138

Side Concealment (%)
01
01
1—0—1
1—0-1

45* a b b

40 55555555, ,55555555,___5555
spirea grydog reddog rose

i

b b b

15,5_- 5_-_555555_5 555-5555- 5__ 5

spirea grydog reddog rose

Stem density (stens per 0.196 n?)

Figure 5-5. YWAR nests in both species of dogwood are signiﬁcantly less concealed
than nests in meadowsweet (spirea) or rose. These nests also have signiﬁcantly lower
stem density surrounding the nest than meadowsweet but not rose. Signiﬁcance was
determined ﬁom Fisher’s LSD, and is denoted by letters. Error bars are standard error.

139

be caused by the removal of those successful birds (which do not re-nest) whom had
fewer snags surrounding their nests, thus shiﬁing the mean estimate.

If the nest placement hypothesis were valid, we would expect to observe
signiﬁcant changes in nest placement attributes between nest attempts within individual
birds. Matched pairs Wilcoxon’s signed rank tests (non-parametric equivalent to paired
t-test) found only the percent coverage of dead plant matter and large trees to be
signiﬁcantly different between ﬁrst and second nest attempts (Table 5-3). Coverage of
dead material diminished as the season progressed, since fresh annual grth tended to
replace these areas. The mean difference in number of large trees was small (0.41 trees),
which suggests that the signiﬁcant result for this variable is a product of directional
change and not magnitude of change. No other surrounding vegetation classiﬁcation
varied signiﬁcantly between nest attempts within individuals including the abundance of
snags. Moreover, paired t-tests of nest height or stem density around the nest found no
signiﬁcant difference between attempts. These results indicate that within individual nest

placement variation does not exist with respect to surrounding vegetation coverage.

Discussion- Results overwhelmingly support the poor parental quality hypothesis.
Dummy nests placed in territories of depredated active nests were more likely to become
depredated than those on successful territories (Figure 5-2). This pattern suggests that
characteristic associated with either the behavior of the parents or the territory they
defend increases the risk of nest predation within the territory. The lack of strong
relationships between behavioral attributes and nest success (Table 3-10) and the lack of

a difference in overall nest predation between dummy and active nests (Figure 5-1),

140

Table 5-3. Matched pairs Wilcoxon’s signed rank tests found that only the coverage of
dead plant matter and abundance of large trees signiﬁcantly changed between ﬁrst and
second nest attempts within individuals. Sample sizes for ﬁrst, second, and third attempts
were n = 158, 76, and 25, respectively.

 

Attempt 1 vs. Attempt 2 Attempt 2 vs. Attempt 3

 

 

 

Elle 4582:3201} Wilcoxon Z“ p-value Wilcoxon Z p-value
Grass/sedge -0.410 0.681 0.655 0.513
Dogwood/willow 0.827 0.408 0.230 0.818
Meadowsweet -0.592 0.554 -0.874 0.382
F orb -l .231 0.218 -0.177 0.860
Rose 1.492 0.136 -0. 141 0.888
Other -0.392 0.695 0.068 0.946
Dead -2099 0.036‘ -0277 0.782
Water -1 .734 0.083 1.363 0.173
Snags 0.596 0.551 -0.329 0.742
Small trees -1.224 0.221 0.844 ‘ 0.399
Large trees 2.597 0.009” 0.913 0.361

 

a. Wilcoxon’s Z = Sum of signed ranks divided by square root (sum of squared ranks). Positive 2 values are increases
in coverage or abundance. '

suggest that adult behavior is not a proximal factor in nest success. Instead, attributes of
occupied territories or nest site selection appear more likely to inﬂuence nest success.
For example, territories with more snags were more likely to be discovered by cowbirds
and nest predators (Figure 3-12). Moreover, territories with more rose overwhelmingly
suffered less nest predation (Figures 3-7, 3-8, 3-9). While not signiﬁcant, territories with
less meadowsweet or more dogwood and willow patches had greater nest predation
(Figures 3-8, 3-9). With the exception of a small increase in large trees near re-nests, no
evidence was found to suggest that changes in nest site selection (Table 5-3), or predator
foraging (Figure 5-3) contributed to increased predation for later nests.

141

Since evidence suggests that avian nest predators are common, it seems plausible
that snags and isolated trees serve as perches for potential predators. These avian
predators likely include corvids, grackles, and possibly raptors, all of which are much
larger than YWAR and locate nests using visual cues. Their greater body size appears to
prevent them from ﬁnding or getting to nests placed in or near rose. Moreover, nests in
dogwood are less concealed than those in rose or meadowsweet (Figure 5-5) making it
easier for visually oriented avian predators to locate. Nests on territories with these
attributes have a greater probability of becoming discovered by predators. Re-nests by
the same individuals have an equally high risk of failure, because they are on the same
territory with the same vegetation characteristics.

It remains possible that predatory behavior may contribute to increased predation
of re-nests and of dummy nests on territories of failed nests, since some predators may
repeatedly return to the territory of a previously discovered nest. However, the existence
of other signiﬁcant relationships that can explain the increased predation of re-nests (see
below) and the apparent temporal independence of predation events between dummy and
active nest on the same territory appear to refute this.

Age did not appear to inﬂuence fecundity directly with respect to egg production
or ability to feed young. This is consistent with other YWAR studies (Lozano and
Lemon 1999), though not in other species (Saether 1990, Forslund and Part 1995).
Alternatively, age of the parents was linked to nest predation (Table 3-7, Figure 3-13).
Moreover, while parental behavior at the nest did not differ by age, some habitat
attributes were associated with parental age. Speciﬁcally, SY adults were more likely to

occupy tenitories with more snags and live trees (Tables 3-8 and 3-9), and less

142

meadowsweet (Figure 3-15). The association of trees and snags to nest predation has
already been discussed. Meadowsweet was clearly a preferred nest plant (Table 3-5,
Figure 3-10). When I pooled non-use vegetation coverage data for all categories of
woody shrubs, I found that ASY females have more shrubs on their territories than SY
females (Figure 5-6). Most of these shrubs were preferred nest plants. A greater
abundance of preferrednest substrate has been shown to reduce the risk of nest predation,

by increasing the predator’s search area (Martin 1993b).

70 l 5 5 5 5-5 55 .5
. p=0.067
a: 60 ;
§’ 1
_> 50 ". l
o .
o
'3 40 l I
5 1 r
(I) . I
or so 1
‘ n=22 19
20 5 55 5 5-5 5 5 5:.
asy sy

Figure 5-6. Shown here is the total percent cover of all woody shrub vegetation for non-
use samples compared with age of female YWAR. ASY females had marginally
signiﬁcantly more shrubs on their territories than SY females.

These results suggest that older adults did better than younger adults because they
were able to obtain and defend the best territories within a habitat. Consequently, SY
adults suffered greater nest predation because they occupied lower quality habitat. These
birds represented a greater proportion of re-nests (Figure 3-14) and their re-nests

inherently had a high risk of predation (Figure 3-3) due to their association with lower

quality habitat.

143

This pattern is consistent with the constraint hypothesis for explaining age-
speciﬁc variation in reproductive success. This hypothesis states that age-speciﬁc
difference in reproductive success exists because younger individuals are unable to
reproduce as well as older ones because they are less capable of competing for the best
breeding territories (Curio 1983, Forslund and Part 1995). Evidence to support this
hypothesis have been documented in the Great Tit (Harvey et al 1979), Black-headed
Grosbeak (Hill 1988), American Redstart (Lemon et al. 1996) and even other populations
of Yellow Warbler (Lozano and Lemon 1999). Younger birds, due to their inexperience,
unfamiliarity with the habitat, and/or inability to arrive as soon as older adults are left

with the lower quality sites within the habitat.

Improvements and ﬁxture research- Several suggestions can be made to improve
ﬁrture studies attempting to identify the array of factors inﬂuencing seasonal fecundity.
Complete vegetation assessment of dummy nests was not performed in this study. To do
so would have permitted more detailed comparisons of vegetation characteristics that
inﬂuence nest success. In general, ﬁeld studies should not ignore detailed vegetation
assessment. In addition to measurements made in this study I suggest including more
non-use sampling (especially within territories), measurements of vegetation
stratiﬁcation, and measuring territory size as an alternative index to territory quality.
Additional measurements should include more detailed behavioral assessment, including
longer observation periods and other behaviors such as vocalization. Arrival dates of
individuals should be documented when possible. This would help determine if those

that arrived ﬁrst selected the best territories.

144

Additional questions, which were derived from this study, should also be explored
in the future. For example, in 1999, there was a period during early June when no re-
nesting occurred following failure, an event not observed in any other year. Apparently,
high ambient temperatures during this time period contributed to this since only those
nests that failed then did not re-nest. Those that failed the following week in cooler
weather did re-nest. Given the important role re-nesting has on fecundity, will global
climate change have a dramatic impact on some songbird populations? It was also
discovered that nest abandonment does not necessarily improve fecundity, and nests that
were abandoned varied in concealment relative to nest that were not. The decision to
abandon or not may depend on the parents assessment of the potential for success of the .
current nest attempt versus future nest attempts. Answering such questions will improve
our understanding of how nest predation and brood parasitism interact to inﬂuence

fecundity, and thus improve our ability to preserve and protect bird populations. '

145

CONCLUSIONS AND RECOMMENDATIONS

Nest predation and brood parasitism clearly do not interact in such a way as to
compound their effect on seasonal fecundity. Moreover, the combined effect of the two
is less than the sum of the effects of each of _the parts. For example, a population exposed
to a moderate to high parasitism rate may perform equally well reproductively at either
20% or 50% nest predation frequency, while a population not exposed to parasitism may
show a decline in fecundity between‘the same two nest predation ﬁequencies. Likewise,
a population exposed to moderate to high nest predation will be less affected by a 50%
parasitism rate than a population exposed to a low predation rate. These two phenomena
interact antagonistically to reduce the. effect of the other on seasonal fecundity.

This antagonistic interaction results ﬁ'om the ﬁmdamental difference in a host
bird’s response to each phenomenon, namely the re—nest response. Single-brooding
songbirds are able to recover much of the loss in fecundity ﬁom nest predation through
re-nesting, but cannot recover losses from parasitism unless the parasitized nest is
depredated or otherwise abandoned. For those species that do not ﬁequently abandon '
parasitized nests, nest predation is the most common means of escaping parasitism. Thus
nest predation can recover some of the potential loss due to parasitism.

Moreover, under speciﬁc conditions, increasing nest predation may actually lead
to slight increases in seasonal fecundity in a population. This special case can exist when
brood parasitism frequency is high, generally greater than 70%, and costly, generally
greater than 80% reduction in brood size. This effect of nest predation is most likely to
occur in populations that have greater nest predation on parasitized nests and have greater

nest survival for re-nests. In theory, it is possible for all these conditions to occur

146

simultaneously in a single population but is likely uncommon. While not observed in this
study population, locally high parasitism rates of Yellow Warbler and other small
migrant songbirds are common, and many of these species frequently exhibit severe
reductions in brood size when parasitized. In addition, nest predation can be higher for
parasitized nests, as was the' case in this study. However, greater nest success for re-nests
does not appear to exist, and in fact'was observed to be just the opposite. Lower nest
success for re-nests can by itself preclude nest predation from increasing fecundity in any
parameter space.

The lack of studies that clearly differentiate re-nests ﬁ'om initial attempts prevents
an assessment of how frequent this pattern exists in other natural populations, but greater
predation on later nests has been documented in other species (Beaver 1975). Field
observations and experimentation with the Yellow Warbler seem to suggest that greater
nest predation on re-nests resulted from a shift in composition of breeding adults. Only
those birds that failed in their ﬁrst nest attempt would re-nest. This changed-the
proportion of individuals in re-nests. Characteristics of these individuals, namely age,
and quality of the territories they occupy, led to a greater risk of nest failure. Thus re-
nests did not necessarily suffer an increased predation rate relative to ﬁrst nest attempts,
but instead suffered greater predation because there were composed of a greater
proportion of parents that were inherently inferior.

The relationship of habitat quality with nest success and the association of
parental age with territory quality are well documented in birds (Hinde 1956, Brown
1969, Curio 1983). Therefore it would appear reasonable to expect increased predation

on re-nests to be frequent in single-brooding songbirds. Greater nest predation on re-

147

nests can prevent the manifestation of a population wide increase in fecundity at
intermediate nest predation. Thus no level of nest predation is likely to increase seasonal
fecundity in populations. If this effect does occur, it is likely rare and inconsequential to
most bird populations, especially those with a large proportion of young adults that would
lead to greater re-nest predation.

Regardless, while nest predation may not increase seasonal fecundity when re-
nests have lower success, nest predation still does not greatly reduce seasonal fecundity
when parasitism is frequent. Thus the interaction of these two phenomena can still have
important consequences to a population. Those populations experiencing moderate to
high parasitism will not be affected by an increase in nest predation through a range
below 60%. Moreover, the beneﬁt of nest predation can still exist at the individual level.
For example, in at least three cases yellow warblers ﬂedged three young following
predation of their parasitized nest. This was greater than the average brood size of '
successful parasitized nests, suggesting these birds gained from nest predation. However,
the vast majority of parasitized nests that were depredated never successfully ﬂedged
young because their re-nests were repeatedly depredated.

The consequences of nest predation and brood parasitism on nesting success in
birds have been a signiﬁcant focus of theory (Pease and Grzybowski 1995, Schmidt and
Whalen 1999). However, the interaction of nest predation and brood parasitism is largely
unexplored in ﬁeld studies. The main reason for this appears to be the lack of focus in
nest success studies on differentiating nest attempts. This appears to be due to the great
effort involved in doing so. In concurrence with other authors working on this problem

(Pease and Grzybowski, in press), I recommend that more emphasis be placed on

148

assessing the impact of nest predation and brood parasitism on seasonal fecundity instead
of on individual nest success. This will require careful nest searching and monitoring to
properly identify each nest attempt made by individual birds. Tracking each nest attempt
is easier in populations that exhibit strong territoriality, remain on the same territory
during all breeding activity, and in habitats where territory boundaries are clearly
identiﬁable. Extensive color banding efforts also help to identify and track individuals
and their nest attempts. While challenging this is not impossible and can be done with
limited resources as demonstrated in this study. By utilizing voluntary labor, this
research was done with an annual budget under $1,000 and with three or fewer volunteers
each year.

As demonstrated here, a myriad of secondary factors inﬂuence how nest predation
and brood parasitism interact. Variations in these conditions clearly exist both spatially
and temporally. Management practice would beneﬁt ﬁom a better understanding of the
interaction of nest predation and brood parasitism as well has an understanding of
temporal changes in nest success and the factors involved. Clearly, as others have stated,
the best management practice is to preserve large areas of high quality habitat, since this
greatly reduces the risk to both nest predation and parasitism simultaneously. However,
this option is rarely available and is costly. Therefore more hands-on management
practices are often used, at least for short-term recovery while habitat is restored. For
example, cowbird removal has been used to improve reproductive success in some
songbirds (Robinson et a1. 1993). However, the results of this management practice have
been mixed. While cowbird removal has helped the Kirtland’s Warbler (Walkinshaw

1983) it had no affect on the Hooded Warbler despite sharply reducing parasitism events

149

(Stutchbury 1997). The decision to manage cowbird abundance as a means of improving
nesting success must depend on other factors in the target population, especially nest
predation. Other factors to consider include age structure, and habitat quality, as well as
life history traits such as the frequency of re-nesting, and whether they double-brood.
Because of the complexity of the interactions involved, simulation modeling will
be an increasingly powerful tool for managers. Simulation models that allow for easy
manipulation of parameters, such as in F LEDGE, will help identify those parameters that
will result in the greatest change in seasonal fecundity. In this way, managers can focus ~
on speciﬁc aspects of the target population and more efﬁciently use available resources.
In general, managers will have to take into account the fact that nest predation and brood
parasitism reduce the effect of the other. As one factor is diminished the effect of the
other is likely to increase. In addition, general management practices will likely beneﬁt
from focusing on improving habitat quality. This may initially cost more but will have
long-term positive results and savings, since more quality habitat, even locally, will likely
improve reproductive success, and will require minimal hands-on management to

maintain.

150

APPENDIX

Table A-1. Each column in the body of this table includes the parasitism frequency
values used for plotting graphical input of a declining brood parasitism probability.
Frequencies down a column change by the same given percent. '

 

 

 

Initial Percent Parasitism Percent
Day 20% 40% 60% 80% of decline

0 0.200 0.400 0.600 0.800 --

10 0.200 0.400 0.600 0.800 0%

20 0.184 0.368 0.552 0.736 8%

30 0.160 0.320 0.480 0.640 20%
40 0.130 0.260 0.390 0.520 35%
50 0.080 0.160 0.240 0.320 60%
60 0.030 0.060 0.090 0.120 85%
70 0.00 0.00 0.00 0.00 100%
80 0.00 0.00 0.00 0.00 100%
90 0.00 0.00 0.00 0.00 100%

 

Table A-2. Each column in the body of this table includes frequency values used to plot a
decreasing nest predation function for four initial predation levels. Frequencies down a
column change by the same given percent.

 

 

 

Initial Percent Predation Percent _

Day 10% 30% 50% 70% of decline

0 0.100 0.300 0.500 0.700 -

10 0.100 0.300 0.500 0.700 0%

20 0.095 0.285 0.475 0.665 5%

30 0.085 0.255 0.425 0.595 15%
40 0.066 0.198 0.330 0.462 33%

50 0.055 0.165 0.275 0.385 45%

60 0.050 0.150 0.250 0.350 50%

70 0.050 0.150 0.250 0.350 50%

80 0.050 0.150 0.250 0.350 50%

90 0.050 0.150 0.250 0.350 50%
100 0.050 0.150 0.250 0.350 50%

 

151

Table A-3. Each column in the body of this table includes the values used to plot the re-
nest frequency function, r(t), used in simulations of altering the probability of re-nesting.

 

 

 

Initial Re-nest frequency Percent
Day 30% 50% 70% 100% of decline
0 0.30 0.50 0.70 1.0 -
10 - 0.30 0.50 0.70 1.0 0%
20 0.30 0.50 0.70 1.0 0%
30 0.30 0.50 0.70 1.0 0%
40 0.30 0.50 0.70 1.0 0%
50 0.24 _ 0.40 0.56 0.80 20%
60 0.0 0.0 0.0 0.0 100%
70 0.0 0.0 0.0 0.0 100%
80 0.0 0.0 0.0 0.0 100%
90 0.0 0.0 0.0 0.0 100%
100 0.0 0.0 0.0 0.0 100%

 

152

LITERATURE CITED

Alderson, G. W., H. L. Gibbs, and S. G. Sealy. 1999. Determining the reproductive
behaviour of individual brown-headed cowbirds using microsatellite DNA
markers. Animal Behaviour 58:895-905.

Anderson, W. L., and R. W. Storer. 1976. Factors inﬂuencing Kirtland’s Warbler nesting
success. Jack-Pine Warbler 54:105:115.

Arcese, P., J. N. M. Smith, and M. I. Hatch. 1996. Nest predation by cowbirds and its
consequences for passerine demography. Proceedings of the National Academy of
Science, USA 93:4608-5611.

Askins, R. R., J. F. Lynch, and R. Greenberg. 1990. Population declines in migratory
birds in eastern North America. Current Ornithology 7: 1-57.

Batts, H. L. 1961. Nesting success of birds on a farm in southern Michigan. Jack-Pine
Warbler. 39:72-93.

Beaver, D.L. 1975. Breeding success of early and late-nesting Red-winged Blackbirds in
a Michigan oldﬁeld. Jack Pine Warbler 53:86-94.

Bellesisles, J. C., and J. Picman 1986. House Wren nest-destroying behavior. Condor
88:190-193.

Berger, A. J. 1951. The cowbird and certain host species in Michigan. Wilson Bulletin.
63:26-34.

Best, L. B. 1978. Field Sparrow reproductive success and nesting ecology. Auk. 95:9-22.

Best, L. B., and D. F. Stauffer. 1980. Factors affecting nesting success in riparian bird
communities. Condor 82: 149-1 58.

Brewer, R., G. A. McPeek, and R. J. Adams, Jr. 1991. The Atlas of Breeding Birds of
Michigan. Michigan State University Press, East Lansing. '

Briskie, J. V., and S. G. Sealy. 1987. Responses of Least Flycatchers to experimental
inter- and intraspeciﬁc brood parasitism. Condor 89:899-901.

Briskie, J .V., S.G. Sealy, and K.A. Hobson. 1990. Differential parasitism of Least
F lycatchers and Yellow Warblers by the Brown-headed Cowbird. Behavioral
Ecology and Sociobiology 27:403-410.

Briskie, J. V., P. R. Martin, and T. E. Martin. 1999. Nest predation and the evolution of

nestling begging calls. Proceedings of the Royal Society of London, Series B:
Biological Sciences 266:2153-2159

153

Brittingham, M. C., and S. A. Temple. 1983. Have cowbirds caused forest songbirds to
decline? BioScience 33:31-35.

Brittingham M. C., and S. A.Temple. 1996. Vegetation around parasitized and non-
parasitized nests within deciduous forest. Journal of Field Ornithology 67:406-
41 3.

Brown, B. T., 1994. Rates of brood parasitism by brown-headed cowbirds on riparian
passerines in Arizona. Journal of Field Ornithology 65: 160-168.

Brown, J. L. 1969. Territorial behavior and population regulation in birds: a review and
re-evaluation. Wilson Bulletin 81: 293-329.

Burgham, M. C. ‘J., and J. Picman. 1989. Effect of Brown-headed Cowbirds on the
evolution of Yellow Warbler anti-parasite strategies. Animal Behavior 38:298-
308.

Burhans, D. E., and F. R. Thompson III. 1998. Effects of time and nest-site
characteristics on concealment of songbird nests. Condor 100:663-672.

Clark, R. G., and T. 'D. Nudds. 1991. Habitat patch size and duck nesting success: the
crucial experiments have not been performed. Wildlife Society Bulletin 19:534-
543.

Clotfelter, E. D. 1998. What cues do brown-headed cowbirds use to locate red-winged
blackbird host nests? Animal Behaviour 55:1181-1189.

Constanza, R., D. Duplisea, U. Kautsky. 1998. Ecological Modeling and economic
systems with STELLA. Ecological Modelling 110: l -4.

Cresswell, W. 1997. Nest predation: the relative effects of nest characteristics, clutch size
and parental behavior. Animal Behavior 53:93-103.

Crick, H. Q. P., D. W. Gibbons, and R. D. Magrath. 1993. Seasonal changes in clutch
size in British birds. Journal of Animal Ecology. 62:263-273.

Curio, E. 1983. Why do young birds reproduce less well? Ibis 125:400-404.

Dearbom, D. C. 1998. Begging behavior and food acquisition by brown-headed cowbird
nestlings. Behavioral Ecology and Sociobiology 43:259-270.

Dearbom, D. C. 1999. Brown-headed Cowbird nestling vocalizations and risk of nest
predation. Auk 116:448-457.

DellaSala, D. A. 1985. The yellow warbler in southeastern Michigan: factors affecting its
productivity. Jack Pine Warbler 63:52-60.

154

Donovan, T. M., F. R. Thompson 111, J. F aaborg, and J. R. Probst. 1995a. Reproductive
success of migratory birds in habitat sources and sinks. Conservation Biology
9:1380-1395.

Eckerle, K. P., and R. Breitwich. 1997. Reproductive success of the Northern Cardinal, a
large host of Brown-headed Cowbirds. Condor 99:169-178.

Ehrlich, P. R., D. S. Dobkin, and D. Wheye. 1988. The birders handbook: A ﬁeld guide
to the natural history of North American birds. Simon & Schuster Inc., New York.

Elliott, P. F. 1978. Cowbird parasitism in the Kansas tallgrass prairie. Auk 95: 161-167.

Elliott, P. F. 1999. Killing of host nestlings by the Brown-headed Cowbird. Journal of 1'
Field Ornithology. 70:55-57. ‘

l

1

Epp, S. 1995. Productivity and survivorship of the Yellow Warbler in mid-Michigan. A r‘_
Thesis. Michigan State University. ~

Fenske-Crawford, T. J. and G. J. Niemi. 1997. Predation of artiﬁcial ground nests at two
type of edges in a forest-dominated landscape. Condor 99: 14-24.

Forslund, P., and T. Part. 1995. Age and reproduction in birds-hypotheses and tests.
Trends in Ecology and Evolution 10:374-378.

Freeman, S., D. F. Gori, and S. Rohwer. 1990. Red-wing Blackbirds and Brown-headed
Cowbirds: some aspects of a host-parasite relationship. Condor 92:336-340.

Gates, J. E., and L. W. Gysel. 1978. Avian nest dispersion and ﬂedgling success in ﬁeld
forest ecotones. Ecology 59:871-883.

Geissler, P. H. and J. R. Sauer. 1990. Topics in route-regression analysis. Pages 54-57 in
J. R. Sauer and S. Droege, editors. Survey Designs and Statistical Methods for the
estimation of Avian Population Trends. U. S. Fish and Wildlife Service,
Biological Report 90(1).

Goldwasser, S., D. Gaines and R. S. Wilbur. 1980. The Least Bell’s Vireo in California: a
de facto endangered race. American Birds 34:742-745.

Goossen, J. P., and S. G. Sealy 1982. Production of young in a dense nesting population
of yellow warblers, Dendroica petechia, in Manitoba (Canada). Canadian Field
Naturalist 96: 1 89-199.

 

Gotrnark, F. 1992. The effect of investigator disturbance on nesting birds. Current
Ornithology. 9:63-104.

155

ml

Graham, D. S. 1988. Responses of ﬁve host species to cowbird parasitism. Condor
90:588-591.

Grant, W. E., E. K. Pedersen, and S. L. Marin. 1997. Model Evaluation. Pages 65-83 in
Ecology and Natural Resource Management: Systems Analysis and Simulation.
John Wiley & Sons, New York.

Grzybowski, J. A., R. B. Clapp, and J. T. Marshall, Jr. 1986. History and population
status of the Black-Capped Vireo in Oklahoma. American Birds 40:1151-1161.

Grzybowski, J. A. and C. M. Pease, in press. Comparing the relative effects of brood
parasitism and nest predation on seasonal fecundity in passerine birds. In Ecology
and management of Cowbirds. (J .N.M. Smith, T. Cook, S.K. Robinson, S.I.
Rothstein, and S.G. Sealy, Eds.) University of Texas Press, Austin, TX.

Gustafson, M. E., J. Hildenbrand, and L. Metras. 1997. The North American Bird
Banding Manual (Electronic Version). Version 1.0, US. Fish and Wildlife
Service.

Harvey, P. H., P. J. Greenwood, and C. M. Perrins. '1 979. Breeding area ﬁdelity of the
Great Tits (Parus major). Journal of Animal Ecology 48:305-313.

Haskell, D. G. 1995. Forest fragmentation and nest predation: are experiments with
Japanese Quail eggs misleading? Auk 112:767-770.

Hatch, M. I. 1997. Variation in song sparrow nest defense: Individual consistency and
relationship to nest success. Condor 99:282-289.

Hensler, G. L., and J. D. Nichols. 1981. The Mayﬁeld method of estimating nesting
success: a model, estimators and simulation results. Wilson Bulletin 93:42-53.

Hill, G. E. 1988. Age, plumage brightness, territory quality, and reproductive success in
the black-headed Grosbeak. Condor 90:379-388.

Hinde, R. A. 1956. The biological signiﬁcance of the territories of birds. Ibis 98:340-369.
Hines, J. E., and J. R. Sauer. 1989. Program CONTRAST - a general program for the
analysis of several survival or recovery rate estimates. Fish and Wildlife

Technical Report 24: 1-7.

High Performance Systems, Inc. 1997. STELLA 11 Technical Documentation. Hanover,
New Hampshire.

Hofslund, P. B. 1957. Cowbird parasitism of the Northern Yellowthroat. Auk 74:42-48.

156

Howlett, J. S., and B. J. M. Stutchbury. 1997. Within-season dispersal, nest-site
modiﬁcation, and predation in renesting hooded warblers. Wilson Bulletin
109:643-649.

Kendeigh, S. C. 1944. Measurement of bird populations. Ecological Monographs 14:67-
106.

Knopf, F. L., and J. A. Sedgwick. 1992. An experimental study of nest-site selection by
Yellow Warblers. Condor 94:734-742. '

Lack, D. 1954. The natural regulation of animal numbers. Oxford University Press.
London. .

Leech, S. M., and M. L. Leonard. 1997. Begging and the risk of predation 1n nestling
birds. Behavioral Ecology 8: 644-646.

Lemon, R. E., S. Perreault, and G. A. Lozano. 1996. Breeding dispersions and site
ﬁdelity of American Redstarts (Setophaga ruticilla). Canadian Journal of
Zoology. 74: 2238-2247.

Lindell, C. 2000. Egg type inﬂuences predation rates in artiﬁcial nest experiments.
Journal of Field Ornithology. 7:16-21.

Linden, M., and A. P. Moller. 1989. Cost of reproduction and covariation of life history
traits in birds. Trends in Ecology and Evolution 4:367-371.

Liu, J ., J. B. Dunning, Jr., and H. R. Pulliarn. 1995. Potential effects of a forest
management plan on Bachrnan’s Sparrow (Aimophila aestivalis): Linking a
spatially explicit model with GIS. Conservation Biology 9:62-75.

Lozano, G. A., R. E. Lemon. 1999. Effects of prior residence and age on breeding
performance in yellow warblers. Wilson Bulletin 111:381-388.

Martin, T. E. 1986. Competition in breeding birds: On the importance of considering
processes at the level of the individual. Current Ornithology 4: 1 81-210.

Martin, T. E. 1988a. Habitat and area effects on forest bird assemblages: Is nest predation
an inﬂuence? Ecology 69:74-84.

Martin, T. E. 1988b. Nest placement: Implications for selected life-history traits, with
special reference to clutch size. American Naturalist 132:900-910.

Martin, T. E. 1992a. Breeding productivity considerations: What are the appropriate
habitat features for management? Pages 455-473 in Ecology and Conservation of
Neotropical Migrant Landbirds (J .M Hagan and DW. Johnston, Eds.).
Smithsonian Institute Press, Washington DC.

157

Martin, T. E. 1992b. Interaction of nest predation and food limitations in reproductive
strategies. Current Ornithology 9:163-197.

Martin, T. E. 1993a. Nest predation among vegetation layers and habitat types: revising
the dogmas. American Naturalist 141:897-913.

Martin, T. E. 1993b. Nest predation and nest sites: new perspectives on old patterns.
BioScience 43:523-532.

Martin T. E., and C. J. Conway. 1995 . BBIRD ﬁeld protocol. Montana Cooperative
Wildlife Research Unit. University of Montana.

Martin TE, and GR. Geupel. 1993. Nest-monitoring plots: Methods for locating nests
and monitoring success. Journal of Field Ornithology 64:507-519.

Martin, T. E., and P. Li. 1992. Life history traits of cavity- versus open nesting birds.
Ecology 73:579-592. '

Martin, T. E., and J. R. Roper. 1988. Nest Predation and nest-site selection of a western
population of the Hermit Thrush. Condor 90:51-5 7.

Marvil, R. E., and A. Cruz 1989. Impact of Brown-headed Cowbird parasitism on the
reproductive success of the Solitary Vireo. Auk 106:476-480. .

Marzluff, J. M. 1988. Do Pinyon Jays alter nest placement based on prior experience?
Animal Behavior 36: 1-10.

May, R.M., and SK. Robinson. 1985. Population dynamics of avian brood parasitism.
American Naturalist 126:475-494.

Mayﬁeld, H. 1961. Nesting success calculated from exposure. Wilson Bulletin 73:255-
261.

Mayﬁeld, H. F. 1965. The Brown-headed Cowbird with old and new hosts. Living Bird
4:13-28.

Mayﬁeld, H. F. 1975. Suggestions for calculating nesting success. Wilson Bulletin
87:456-466.

McClean, I. G., J. N. M. Smith, and K. G. Stewart. 1986. Mobbing behavior, nest
exposure, and breeding success in the American robin (T urdus migratorius).
Behaviour 96:171-186.

McGeen, D.S. 1972. Cowbird-host relationships. Auk. 89:360-380.

158

McMaster, D. G., S. G. Sealy, S. A. Gill, and D. L. Neudorf. 1999. Timing of egg laying
in yellow warblers. Auk 116:236-240.

Mitra, S. S. 1999. Ecology and behavior of yellow warblers breeding 1n Rhode Island' 5
Great Swamp. Northeastern Naturalist. 6: 249- 262.

Murphy, E. C., and E. Haukioja. 1986. Clutch size in nidicolous birds. Current
Ornithology 4:141 :180.

Newman, GA. 1970. Cowbird parasitism and nesting success of Lark Sparrows in
southern Oklahoma. Wilson Bulletin 82.304-309.

Nice, M.M. 1937 . Studies in the life history of the Song Sparrow. Part 1. Trans Linn '
Society, New York.

Nolan, V., Jr. 1978.. The ecology and behavior of the Prairie Warbler Dendroica
discolor. Ornithological Monographs 26.

Norman, R. F. and R. J. Robertson. 1975. Nest Searching behavior 1n the Brown-headed
Cowbird. Auk 92. 610-611. ‘

Payne R. B., and L. L. Payne 1998. Brood parasitism by cowbirds: Risks and effects on
reprodrictive success and survival 1n indigo buntings. Behavioral Ecology. 9. 64-
73.

Pease, C. M., and J. A. Grzybowski. 1995. Assessing the consequences of brood
parasitism and nest predation on seasonal fecundity in passerine birds. Auk
112:343-363.

Peck, G. K., and R. D. James. 1987. Breeding birds of Ontario Nidology and distribution,
V012: passerines. Royal Ontario Museum, Toronto, Canada.

Picozzi, N. 1975. Crow Predation on marked nests. Journal of Wildlife Management
39: 1 5 1-1 55.

Pyle, P. 1997. Identiﬁcation Guide to North American Birds. Part I. Braun-Brumﬁeld,
Ann Arbor.

Ricklefs, RE. 1969. An analysis of nesting mortality in birds. Srrrithsonian Contributions
to Zoology 9: 1-48.

Ricklefs, RE. 1973. F ecundity, mortality, and avian demography. Pages 366-435. in
Breeding Biology of Birds (D. S. F amer, Ed). National Academy of Sciences,
Washington DC.

Ricklefs, R. E., and G. Bloom. 1977. Components of Avian Breeding Productivity. Auk

159

94:86:96.

Robert, M., G. Sorci, A. P. Moller, M. E. Hochberg, A. Pomiankowski, and M. Pagel.
1999. Retaliatory cuckoos and the evolution of host resistance to brood parasites.
Animal Behaviour 58:817-824. ,

Robinson SK. 1992. Population dynamics’of breeding Neotropical migrants in a
fragmented Illinois landscape. Pages 408-418. in Ecology and conservation of
Neotropical migrant landbirds (J .M. Hagan and DW. Johnston, Eds).
Smithsonian Institute Press, Washington DC.

Robinson S. K., J. A. Grzybowski, S. I. Rothstein, L. J. Petit, M. C. Brittingham, and F.
R. Thompson. 1993. Management implications of cowbird parasitism on
Neotropical migrant songbirds. Pages 93-102 in Status and management of
Neotropical migrant birds (D. M. F inch and P. W. Stangel, Eds). General
technical report RM-229. US. Department of Agriculture, Washington DC.

Robinson S. K., S. I. Rothstein, M. C. Brittingham, L. J. Petit, and J. A. Grzybowski.
1995a. Ecology and behavior of cowbirds and their impact on host populations.
Pages 428-460 in Ecology and Management of Neotropical Migratory Birds (T. E.
Martin and D. M. Finch, Eds. ). Oxford University Press, New York.

Robinson, S. K., F. R Thompson, T. M. Donovan, D. R. Whitehead. and J. F aaborg.
1995b. Regional forest fragmentation and the nesting success of migratory birds.
Science 267: 1987-1990.

Rooneem, T. M., and R. J. Robertson. 1997. The potential to lay replacement clutches by
Tree Swallows. Condor 99:228-231.

Rothstein, S. I. 1975. An experimental and teleonomic investigation of avian brood
parasitism. Condor 77:250-271. ‘

Rothstein, S. I. 1982. Successes and failures in avian egg recognition with comments on
the utility of optimality reasoning. American Zoologist 22:547-560.

Rothstein, S. I. 1990. A model system for coevolution: avian brood parasitism. Annual
Review of Ecology and Systematics 21:481-508.

Rothstein, S. I., and S. K. Robinson. 1994. Conservation and coevolutionary implications
of brood parasitism by cowbirds. Trends in Ecology and Evolution 9: 162-164.

Saether, B. E. 1990. Age-speciﬁc variation in reproductive performance of birds. Current
Ornithology. 7:251-283.

Sauer, J. R., J. E. Hines, 1. Thomas, J. Fallon, and G. Gough. 2000. The North American
Breeding Bird Survey, Results and Analysis 1966 - 1999. Version 98.1, USGS

160

Patuxent Wildlife Research Center, Laurel, MD.

Sauer, J. R., and B. K. Williams. 1989. Generalized procedures for testing hypotheses
about survival or recovery rates. Journal of Wildlife Management 53:137-142.

Schmidt, K. A., and C. J. Whelan. 1999. The relative impacts of nest predation and brood
parasitism on seasonal fecundity in songbirds. Conservation Biology 13:46-57. .

Scott, D. M. 1963. Changes in the reproductive activity of the Brown-headed Cowbird
within the breeding season. Wilson Bulletin 75: 123-239.

Sealy S. G. 1989. Incidental egg dumping by the House Wren 1n a Yellow Warbler nest.
Wilson Bulletin 101 24.91-493

Sedgwick, J. A., and F. L.-Knopf. 1.988. A high. incidence of Brown-headed Cowbird
parasitism of Willow F lycatchers. Condor 90:253-256.

Smith, J. N. M. 1981. Cowbird parasitism, host ﬁtness, and age of the host female in an
island Song Sparrow population. Condor 83:152-161.

Smith, J. N. M., and P. Arcese 1994. Brown-headed Cowbirds and an island population
of Song sparrows: A 16-year study. Condor 96:916-934.

Smith, K. G., and D. C. Andersen. 1982. Food, predation and reproductive ecology of the
dark-eyed junco (Junco hyemalis meamsi) in northern Utah. Auk 99:650-661.

Soler, J. J ., G. Sorci, M. Soler, and A. P. Moeller. 1999. Change in host rejection ' ,
behavior mediated by the predatory behavior of its brood parasite. Behavioral
Ecology 10:275-280.

Sonerund, G.A., and RE. F jeld. 1987. Longterrn memory in egg predators: an experiment
with a Hooded Crow. Orrris Scandinavica 18:323-325.

Spooner A., S. Pribil, and J. Picman. 1996. Why do Gray Catbirds destroy eggs in nests
of other birds? Experimental tests of alternative hypotheses. Canadian Journal of
Zoology 74:1688-1695.

SPSS Inc. 1998. SYSTAT Standard Version 8.0.

Strausberger, B. M. 1998. Evident nest-searching behavior of female Brown-headed
Cowbirds while attended by males. Wilson Bulletin 1 10:133-136.

Strausberger, B. M., and M. V. Ashley. 1997. Community-wide patterns of parasitism of
a host "generalist" brood-parasitic cowbird. Oecologia. 112:254—262.

Stutchbury, B. J. M. 1997. Effect of female cowbird removal on reproductive success of

161

:33

 

 

Hooded Warblers. Wilson Bulletin 109174-8 1.

Terborgh, J. 1989. Where have all the birds gone? Princeton University Press, Princeton,
NJ. '

Tewksbury, J. J ., S. J. Hejl, and T. E. Martin. 1998. Breeding productivity does not
decline with increasing fragmentation in a western landscape. Ecology 79:2890-
2903.

Walkinshaw, L. H. 1961. The effect of parasitism by the Brown-headed Cowbird on
Empidonax ﬂycatchers in Michigan. Auk 78:266-268.

Walkinshaw, L. H. 1983. Kirtland’s Warbler. Cranbrook Institute of Science, Bloomﬁeld
Hills, MI.

Watt, D. J ., and A. M. Dimberio. 1990. Structure of successful nests of the American
goldﬁnch (Carduelis tristis). Journal of Field Ornithology 61: 413-418.

Weatherhead, P.J. 1989. Sex ratios, host-speciﬁc reproductive success, and impact of
Brown-headed Cowbirds. Auk 106:358-366. -

Wilcove, D. S. 1985. Nest predation in forest tracts and the decline of migratory
songbirds. Ecology 66:1211-1214.

Williams, G. C. 1966. Natural selection, the cost of reproduction and a reﬁnement of
Lack’s Principle. American Naturalist 103:501-516.

Wilson, G. R., M. C. Brittingham, and L. J. Goodrich. 1998. How well do artiﬁcial nests
estimate success of real nests? Condor 100:357-364.

Winfree, R. 1999. Cuckoos, cowbirds and the persistence of brood parasitism. Trends in
Ecology and Evolution 14:338-342.

Wolf, L. 1987. Host-parasite interactions of Brown-headed Cowbirds and the Dark-eyed
Junco of Virginia. Wilson Bulletin 99:338-350.

Woodworth, B. L. 1999. Modeling population dynamics of a songbird exposed to
parasitism and predation and evaluating management options. Conservation

Biology 13:67-76.

Wray, T., II, and R. C. Whitrnore. 1979. Effects of vegetation on nesting success of
Vesper Sparrows. Auk 96:802-805.

162

111111111111111111111111111