gnu...“ 3?...

.3. 9.4

 

'3

,1
«22:14.

:31

1.6.
a #1 ﬁll.
315...:

 

...
I.
r

 

1.94:6:

.\

 

 

 

 

 

 

01¢ ...p..u.!: I
i .13.
1... tr.lI..Xx :9... 1
{(v1- .5XWWQJXI ulcwh‘.
51...]: 3:2. 1.. z .3?

. .

”£9.41? ”2. «r. mange... “Arsxruaawé a 5.

:1... . .12.

i. : 35:3 a. .d .51)!

,L ...V

 

THE bis

; on}

This is to certify that the

dissertation entitled

Proximate factors affecting intraclutch egg-size variation
in the Squacco Heron Ardeola ralloides

 

presented by

Grigorios Papakostas
has been accepted towards fulﬁllment

of the requirements for

Ph . D . degree in Zoology

QMOz/BQW

Major professor

 

 

Date ”I/IQI/OL

MS U is an Afﬁrmative Action/Equal Opportunity Institution 0-12771

 

. LIBRARY
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 DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIRC/DateDue.pGS-p. 1 5

 

PROXIMATE FACTORS AFFECTING
INTRACLUTCH EGG-SIZE VARIATION
IN THE SQUACCO HERON ARDEOLA RALLOIDES

By

Grigorios Papakostas

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Zoology

2002

ABSTRACT

PROXIMATE FACTORS AFFECTING INTRACLUTCH EGG-SIZE VARIATION
IN THE SQUACCO HERON ARDEOLA RALLOIDES

By

Grigorios Papakostas

I studied certain aspects of the breeding biology of the Squacco Heron (Ardeola
ralloides), a poorly known species, focusing on egg-size variation within clutches and
the proximate factors that may affect it. I found that, in most cases-this heron’s clutch
proﬁle (pattern of intraclutch egg-size variation) is arched, that is both ﬁrst and last eggs
are signiﬁcantly smaller than middle ones. I tested Parsons’s (1976) hypotheses that the
size of ﬁnal eggs is a) physiologically constrained due to hormonal changes associated
with the onset of female incubation, or b) nutritionally limited by shortages in female
resources for egg formation. I also examined the possibility that the onset of incubation
may restrict female feeding opportunities, and thus limit egg-resource availability and
last-egg size. I was unable to directly assess either female nutritional resources or the
timing of the onset of incubation, but I inferred the levels of these factors, respectively,
from a) the average egg volume of clutches (AvVol), and b) their hatch span (H-Span),
that is the time interval between the hatching of their ﬁrst and ﬁnal eggs. I also tested the
hypotheses that the relative size of ﬁrst eggs may be nutritionally constrained or limited
anatomically by an initial inelasticity of the oviduct.

Neither of the hypotheses concerning the size of ﬁrst eggs was supported: a) ﬁrst

eggs became relatively smaller in 4-egg clutches with higher AvVol (the opposite of what

the nutritional hypothesis predicted); b) they were not signiﬁcantly thinner than later
eggs, neither were their volume and width/length ratio positively correlated (as the
anatomical hypothesis predicted). There was no correlation between clutches’ H-Span
and AvVol values, which refutes the hypothesis of nutritional constraints on egg size
related to the onset of incubation.

H-Span affected clutch proﬁle in the way predicted by Parsons’s hormonal hypo-
thesis: between samples of long and short H-Span (early and late incubation onset), the
proﬁle’s linear trend changed from negative to positive, and the relative size of ﬁnal eggs
increased. The same trend was observed between S-egg and 4-egg clutches (which have
median H-Span values of 4.5 and 3.0 (1, respectively), but only among high-AvVol nests
(where resources for egg formation were presumably abundant). AvVol had a positive
effect on the linear trend of clutch proﬁle and on relative last-egg size (as the nutritional
hypothesis predicted), but only in 4-egg clutches, where H-Span was relatively short (and
hormonal constraints were presumably weak). Therefore, it seems that both hormonal and
nutritional factors can inﬂuence the relative size of last eggs in this heron, but each one’s
effects become apparent only in the absence of interference from the other. The hormonal
and nutritional hypotheses also predicted, respectively, that H-Span should decline, while
AvVol should increase along a gradient of clutch proﬁle types where relative last-egg size
progressively increases. The former prediction was conﬁrmed, but the trend in AvVol
was quadratic, and clutches with irregular proﬁles also had a low mean AvVol. These
results further support the hypothesis of hormonal constraints on ﬁnal-egg size that

interact with nutritional limitations.

To my Parents

iv

ACKNOWLEDGMENTS

I would like to thank several people whose help has been invaluable to me in
completing this dissertation.

Don Beaver, my academic adviser, has not only been a great mentor, but also
gave me his unstinting support throughout this long endeavor, far beyond the call of duty.

My advisory committee members, Fred Dyer, Don Straney, and Scott Winterstein,
have shown great patience, and Scott Winterstein also advised me on data analyses and
the interpretation of their results.

Vassilis Goutner has been a great mentor and friend over the years, and he and Iris
Charalambidou assisted me in ﬁeldwork.

Finally, I want to thank my parents for their love and unconditional support

during this long period of separation.

TABLE OF CONTENTS

LIST OF TABLES .................................................. viii
LIST OF FIGURES .................................................. xi
LIST OF TERMS AND ABBREVIATIONS .............................. xii
I. INTRODUCTION ................................................... 1
II. GENERAL METHODS .............................................. 9
l. ARDEOLA RALLOIDES ................................................ 9
2. STUDY AREA AND POPULATION .................................... l l
3. DATA COLLECTION ................................................ l3
4. STATISTICAL ANALYSES ........................................ 16
III. PRELIMINARY ANALYSES ........................................ 17
l. EGG-LAYING INTERVALS ........................................... 18
2. LAYING DATE ..................................................... 23
3. CLUTCH SIZE ...................................................... 3l
4. ESTIMATION OF EGG VOLUME ..................... ' ................. 3 8
5. AVERAGE EGG VOLUME OF CLUTCHES ........................... 39
Female resource availability and egg size ............................ 40
Results and Discussion .......................................... 43
6. HATCHING INTERVALS ......................................... 48
Estimation of sibling ages and hatching intervals ...................... 48
Suitability of Hatch Span as an index of female incubation .............. 52
Statistical Methods .............................................. 54
Results and Discussion .......................................... 54
7. RELATIONSHIP BETWEEN AV.VOL AND HATCH SPAN ............... 60
IV. CLUTCH PROFILE ANALYSES ..................................... 63
1. STATISTICAL METHODS ...................................... 64
2. PREDICTIONS TESTED ........................................ 69
3. RESULTS .................................................... 72
A. AVERAGE CLUTCH PROFILE ..................................... 72
B. CLUTCH PROFILE ANALYSES .................................. 75
Analysis by Hatch Span and Clutch Size ........................... 75
Analysis by Year and Season .................... ‘ ................ 78
Analysis by Egg Size (Egsz) and Clutch Size (C lsz) .................. 81
C. CLUTCH PROFILE TYPES ...................................... 87
4. DISCUSSION ................................................. 96

vi

‘w.———-—-—-—--—u—

V. SUMMARY AND CONCLUSIONS .................................. l 1 1
APPENDIX ....................................................... 1 19

REFERENCES .................................................... 127

vii

__

"-5 C"

 

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

Table 6.

Table 7.

Table 8.

Table 9.

Table 10.

Table 1 1.

Table 12.

Table 13.

Table 14.

Table 15.

LIST OF TABLES

Frequencies of cases where different numbers of eggs
were laid during 3-d and 4-d nest-check intervals ..................... 20

Frequencies of cases where 2 or 3 eggs were laid in 4 d for
early and late clutches, and result of statistical comparison .............. 21

Frequencies of cases where 2 or 3 eggs were laid in 4 d
by year, and result of statistical comparison .......................... 22

Comparison between clutch sizes of frequencies of cases where
different numbers of eggs were laid during 3-d or 4-d intervals ........... 22

Comparison between early and late eggs of frequencies of cases where

different numbers of eggs were laid during 3-d or 4-d intervals ........... 23
Statistics on incubation intervals by clutch size and egg-laying order ...... 26
Overall and pairwise statistical comparisons of laying date among years . . . 29

Daily mean air temperature (°C) during the ﬁrst 2 weeks of May by year . . . 29
Clutch size (Clsz) frequencies during the main laying season by year ...... 33

Percentage of small and large clutches, sample size, and mean and median
clutch size in the early and late parts of each year’s main laying season . . . . 35

Frequencies of small and large clutches, and mean and median
clutch size of main and delayed nests from 1992 and 1993 .............. 35

Frequencies, mean and median of total and ﬁnal clutch size (Clsz) in cases
where initial eggs were lost, and in regular clutches from 1992 and 1994 . . . 37

Statistics on AvVol (cm3) by year and season ........................ 44
ANOVA on AvVol by year (random factor) and season (ﬁxed) .......... 44

Statistics on AvVol (cm3) by clutch size (Clsz) and in clutches
laid under certain resource limitations (explanations in text) ............. 46

viii

Table 16.

Table 17.

Table 18.

Table 19.

Table 20.

Table 21.

Table 22.

Table 23.

Table 24.

Table 25.

Table 26.

Table 27.

Table 28.

Table 29.

Table 30.

Statistics on body measurements from nestlings at ages of 0.5 and 1.5 d
in 1992, and values (Criteria) of equal probability (P-value)
of belonging to either age’s distribution ............................. 51

Successive and cumulative hatching intervals (in d) by clutch size (Clsz).
( Explanations in text.) .......................................... 55

Median hatch span (H-Span) and standardized laying date (Std L D)
by season in various samples (explanations in text).
Sample sizes are given in H-Span row .............................. 58

Sizes of samples used in clutch proﬁle analyses (RMAs).
Factor levels and name abbreviations are explained in the text ........... 68

Summary of nutritional and hormonal hypotheses’ predictions
(descriptive and referring to RMA results) ........................... 71

RMA on egg volume by egg Rank in C4 (A) and C5 (B).

( Original 5 ranks used in C5.) .................................... 73
RMA on egg volume by Rank (within-clutches factor),

and by H-Span and C152 (among-clutches factors) ..................... 76
RMAS on egg volume by Year and Season (among-nests factors),

and by Rank (within-clutches factor) in each clutch size (C4 and C5) ...... 78
Statistics on H-Span(d) by Season and by Hatch Span in 5-egg clutches . . . 81
RMA on egg volume by Rank (within-clutches factor),

and by Egsz and C132 (among-clutches factors) ....................... 82
Separate RMAs on egg volume by Egsz (among-clutches factor),

and by Rank (within-clutches factor), in each clutch size (Clsz).
Bonferroni-corrected critical P—level: 0.0125 ......................... 85
Separate RMAs on egg volume by Clsz (among-clutches factor),

and by Rank (within-clutches factor), in each level of factor Egsz.
Bonferroni-corrected critical P-level: 0.0125 ......... ' ................ 85
Numbers and percentages of clutches with different proﬁle types ......... 89
Frequencies of cases where 2 or 3 eggs were laid in 4 d for

pairs of proﬁle types, and result of statistical comparison ............... 92
ANOVA on AvVol by clutch size and proﬁle type (excluding Irregular) . . . 95

ix

Table 31. Avian species with arched clutch proﬁles (marginal eggs smaller than middle).
............................................................. 98

Table 32. Logarithmic regression of hatchling weight (g) and bill+head length (mm) on
egg volume (cm3). P-levels pertain to slope comparisons with values of O and 1.
............................................................ 100

Table A-1 RMA on egg volume by Rank (within-clutches factor), and by

Year and Season (among-clutches factors), using 4-egg nests.
Results are used in text Tables 21A and 23A ....................... 119

Table A-2 RMA on egg volume by Rank (within-clutches factor),_and by
Year and Season (among-clutches factors), using 5-egg nests.
Results are used in text Tables 21B and 23B ....................... 120

Table A-3 RMA on egg volume by Rank (within-clutches factor),
and by clutch size (Clsz) and H-Span (among-clutches factors).
Clsz included only in order to examine the RXHXC interaction.
Results are used in text Table 22 ................................ 121

Table A-4 RMA on egg volume by Rank (within-clutches factor), and by
Egsz and C152 (among-clutches factors). Contrasts were not
performed, because RXHXC interaction approached signiﬁcance.
Results are used in text Table 25 ................................ 122

Table A-5 RMA on egg volume by Rank (within-clutches factor),
and by Egsz (among-clutches factor), in 4-egg clutches.
Bonferroni-corrected critical P-level: 0.0125.
Results are used in text Table 26A ............................... 123

Table A-6 RMA on egg volume by Rank (within-clutches factor),-
and by Egsz (among-clutches factor), in 5-egg clutches.
Bonferroni-corrected critical P-level: 0.0125.
Results are used in text Table 268 ............................... 124

Table A-7 RMA on egg volume by Rank (within-clutches factor),
and by Clsz (among-clutches factor), in Low-Egsz nests.
Bonferroni-corrected critical P-level: 0.0125.
Results are used in text Table 27A ............................... 125

Table A-8 RMA on egg volume by Rank (within-clutches factor),
and by Clsz (among-clutches factor), in High-Egsz nests.

Bonferroni-corrected critical P-level: 0.0125.
Results are used in text Table 27B ............................... 126

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 7.

Figure 8.

Figure 9.

Figure 10.

Figure l 1.

Figure 12.

LIST OF FIGURES

Laying curves by year. Line M: median laying date of main clutches.

Line L (1992 and 1993): limit between main and delayed clutches.
Line E (1994): end of observations. ( See text for explanations.) . . .

Regression of egg Volume on the product of egg Length X (Width)2

Median hatch span by clutch size and year.
Sample sizes indicated above or below data points ..............

Average egg-size proﬁle by clutch size (Clsz).
Error bars represent 95% conﬁdence intervals.

Regression lines indicate each proﬁle’s linear trend .............

Average egg-shape (width/length) proﬁle by clutch size (Clsz).
Error bars represent 95% conﬁdence intervals .................

Clutch proﬁle by H-Span (pooling clutch sizes).
Error bars and regression lines as in Figure 4 ..................

Clutch proﬁle in 5-egg nests by Year.
Error bars and regression lines as in Figure 4 ..................

Clutch proﬁle by clutch size (Clsz) and mean egg size (AvVol).
Error bars and regression lines as in Figure 4 ..................

Clutch proﬁle types. Error bars represent 95% conﬁdence intervals .

Median H-Span by proﬁle type and clutch size.
Sample sizes are indicated next to data points ..................

AvVol by proﬁle type (means with 95% conﬁdence intervals).
Sample sizes are indicated next to data points ..................

Average egg-size proﬁles of 4- and 5-egg Little Egret clutches.
Error bars represent 95% conﬁdence intervals .................

xi

...... 28

...... 39

...... 57

...... 73

...... 74

...... 76

...... 79

...... 83

...... 88

...... 91

...... 94

...... 97

AvVol
Clsz
Egsz
FSH

Gn-RH

H-Span

LH

Proﬁle

of a
clutch

Rank
of an

egg

LIST OF TERMS AND ABBREVIATIONS

Average egg size of a clutch.

Clutch size ; 4- and 5-egg clutches are symbolized as C4 and C5, respectively.
Egg Size ; among-nests factor representing level of AvVol (low vs high).
Follicle-stimulating hormone.

Gonadotropin-stimulating hormone.

Hatch span of a clutch (interval between the hatching of ﬁrst and last eggs).

Luteinizing hormone.

Pattern of intraclutch egg-size variation.

Clutch proﬁle types:
a) Decline (decrease in egg size with laying order);
b) I-max (clutch proﬁle is arched and intermediate egg is largest);
c) P-max (clutch proﬁle is arched and penultimate egg is largest);
d) Rise (increase in egg size with laying order);
e) Irregular (clutch proﬁle does not ﬁt any of the above patterns).

Position in the laying sequence. Symbolized by A-D in C4 and by A-E in C5.

Modiﬁed egg ranks in C5 are ﬁrst, intermediate (average of B and C), penulti-
mate, and last. First and last can be referred to as marginal, others as middle.

xii

I. INTRODUCTION

The avian egg provides a protective environment, the genetic information and all
necessary nutrients for the development of avian embryos, which only requires the input
of thermal energy and gas exchange with the outside world (Burley and Vadehra 1989).
Given the egg’s importance for embryonic growth, biologists have been interested both in
the factors that inﬂuence its size and quality (Perrins 1996; Meijer and Drent 1999), and
in the effects of the latter on nestling ﬁtness (Williams 1994). A

Differences in egg size among females of the same species may reﬂect individual
variation in body size (Ojanen et al.1979; Nol et al.1984), age (Reid 1988; Willebrand
1992), and nutritional condition (Houston et al.1983; Leblanc 1989; Wiebe and Bortolotti
1995). Egg size also shows consistent patterns of intraclutch variability (clutch proﬁles)
in diverse avian taxa, including cormorants (Coulson et al.1969), herons (Custer and
Frederick 1990), anatids (Owen and West 1988; Williams et al.1993), coots (Arnold
1991), shorebirds (Nol et al.1984; Amat et al.2001), larids (Runde and Barrett 1981;
Meathrel and Ryder 1987; Kilpi et al.1996), and passerines (Ryden 1978; Ojanen et al.
1981; Rofstad and Sandvik 1985; Heeb 1994).

Early research on clutch proﬁles focused on the relative size of last eggs and its
potential involvement in reproductive strategies, in association with hatching asynchrony
(Howe 1976; Parsons 1976; Clark and Wilson 1981; Slagsvold et al. 1984). Lack (1947)
proposed that, in species that suffer low nest predation but depend on unpredictable food
resources for raising young, hatching asynchrony is an adaptation that creates competitive

asymmetries among broodmates, which ensure the successful growth of senior chicks at

the cost of their junior siblings’ starvation when food is scarce. Lack’s arguments have
generated a wealth of studies and alternative hypotheses, adaptive and not (Stoleson and
Beissinger 1995), and the evolutionary reasons for the occurrence of hatching asynchrony
may vary among species (Magrath 1990). Parental reproductive strategies may then be
indicated by the relative size of last eggs (Slagsvold et al.1984): a) small ﬁnal eggs may
further facilitate the demise of junior chicks in brood-reductionist species, while b) large
ﬁnal eggs may reduce the risk of maladaptive brood reduction in species where hatching
asynchrony serves functions unrelated to sibling competition (Clark and Wilson 1981).
For clutch proﬁle to play such a role, egg size must have signiﬁcant effects on nestling
performance. Egg size has been found to affect the early growth (Howe 1976) and the
survival (Grant 1991; Amat et a1. 2001) of same-age siblings, and also the early survival
of junior chicks in gulls (Parsons 1975a). Other authors, however, have concluded that
the effects of clutch proﬁle on siblings’ fates are insigniﬁcant when compared with those
of hatching asynchrony (Stokland and Amundsen 1988; Bollinger 1994; Vinuela 1996).

Since many studies of clutch proﬁle focused on its possible effects on nestling
ﬁtness, when such effects were not apparent, authors often attributed the existence of egg-
size variation within clutches to unspeciﬁed physiological constraints. All traits, however,
whether adaptive or incidental, have a physiological basis, the study of which may lead to
their better understanding, without prejudice to their possible ultimate functions.

Parsons (1976) investigated the proximate causation of clutch proﬁle in the
Herring Gull (Larus argentatus) by removing eggs as females laid them. In this species,
most females lay 3 eggs, of which the last is signiﬁcantly smaller than the other 2 (by

about 10%), mainly due to a deﬁciency in albumen. Egg removals delayed the onset of

incubation and allowed most females to lay replacement eggs, a phenomenon that has
been observed in a large variety of birds (reviews in Kennedy 1991, and in Haywood
1993a). In the enlarged clutches, only ﬁnal eggs (4th, 5th, etc.) were signiﬁcantly smaller
than early ones, and of similar size to normal last (3rd) eggs. Third or even later eggs
were not signiﬁcantly smaller, as long as they were earlier than last.

Parsons (1976) predicted that, if female resource depletion was responsible for
the smaller size of normal ﬁnal (3rd) eggs, all 3rd and later eggs would be progressively
smaller. Since, however, no eggs before the last were affected, whatever their absolute
laying order, this resource-limitations hypothesis was refuted. Instead, Parsons (1976)
suggested that the Herring Gull’s clutch proﬁle is mainly the result of physiological
constraints imposed on the female’s reproductive system by hormonal changes associated
with the onset of incubation. ( In my presentation of his proposed mechanism below,

I also use supportive evidence from more recent studies.)

The development of incubation behavior is associated with an increase in plasma
prolactin in many birds (Goldsmith 1983; Buntin 1996; Vleck 2002). High prolactin
levels inhibit the secretion of gonadotropins (Lea et a1. 1981; Dawson and Goldsmith
1982), which promote the growth of the ovary and its gametogenic and endocrine
functions (Murton and Westwood 1977; Scanes 1986). Thus, increased prolactin levels
are linked with reduced estradiol secretion (Zadworny et al.1989; Sockman and Schwabl
1999), follicular atresia (Opel and Proudman 1980), and the cessation of egg-laying
(Youngren et al.1991). Estardiol stimulates oviduct growth (Murton and Westwood 1977)
and albumen protein synthesis (Johnson 1986), whereas high levels of progesterone (as

those released by atretic follicles) elicit oviduct regression (Murton and Westwood 1977).

Thus, the suit of hormonal changes triggered by the onset of incubation eventually results
in the suppression of ovarian and oviductal function (Buntin and Tcsch 1985).

In the Herring Gull, eggs are laid every 2 d, and incubation starts >12 h after the
lst egg is laid (Parsons 1972). By that time, albumen deposition in the 2nd egg is well
advanced, and the 3rd egg’s yolk is almost complete, but its albumen will not begin to be
synthesized for another 1.5 (1 (Parsons 1976). Therefore, hormonal constraints can limit
signiﬁcantly only the last (3rd) egg’s albumen, thus affecting both this egg’s size and its
composition (Parsons 1976). In species who lay large clutches and begin low-intensity
incubation early in the laying sequence, gradual hormonal changes may reduce (without
completely suppressing) the function of both the ovary and the oviduct, thus causing a
progressive decrease in the size of eggs laid after the onset of incubation without
affecting their composition (Arnold 1991; Williams et al.1996).

Houston, Jones and Sibly (1983) proposed an alternative explanation for the
smaller size and albumen deﬁciency of last (3rd) eggs in the Lesser Black-backed Gull
(Larusfuscus). They found that female protein reserves positively affected clutch size
and the dry weight of egg yolks, but not the amount of egg albumen. Assuming that
female gulls have to acquire nutrients for egg albumen from their daily diet, like some
passerines (Jones and Ward 1979), they proposed that the apparent association between
the onset of incubation and the smaller size of ﬁnal eggs results from the constraints
imposed by the former on female foraging time and food ingestion;

More recent experiments on Lesser Black-backed Gulls (Hiom et a1. 1 991; Bolton
et al.1992) have shown that food supplements can increase both the average egg size of

clutches and the relative size of ﬁnal eggs. The latter, however, is due to an increase in

yolk, rather than albumen, protein (Bolton et al.1992). These results indicate that the yolk
of last eggs may be nutritionally constrained, but their albumen is limited by some other
factor, which could be the hormonal mechanism proposed by Parsons (1976). Studies on
geese (Leblanc 1987; Williams et al.1996), coots (Arnold 1991) and crows (Heeb 1994)
were unable to attribute small last eggs to female resource limitations.

The above hypotheses (hormonal and nutritional) concern only the proximate
limitation of last-egg size. However, both ﬁrst and last eggs are signiﬁcantly smaller than
intermediate ones in various birds, including cormorants (Stokland and Amundsen 1988),
herons (Jover et al.1993), anatids (Leblanc 1987; Kennamer et al.1997), shorebirds (Nol
et al.1984), coots (Arnold 1991), and some passerines (Ojanen et al.1981; Verbeek 1990;
Magrath 1992a), while in some other passerines egg size increases with laying order and
ﬁrst eggs are the smallest in the clutch (Howe 1976; Mead and Morton 1985; Slagsvold
and Lifjeld 1989). Three proximate hypotheses have been proposed to account for small
ﬁrst eggs in such species, but they have neither been formulated nor tested as rigorously
as the ones that concern last eggs.

The size of ﬁrst eggs seems to be nutritionally constrained in some insectivorous
songbirds, who begin to lay eggs while ambient temperature and food availability are still
low, but whose food ingestion relative to metabolic needs often improves daily during the
laying period (Slagsvold and Litjeld 1989; Nilsson and Svensson 1993a). In accordance
with this hypothesis, food supplements given to Blue Tits (Parus caeruleus) before the
onset of laying eliminated the size deﬁcit of ﬁrst eggs (N ilsson and Svensson 1993a).

The more elongate shape of some anatids’ ﬁrst eggs may result from a restriction

of their width by an initial inelasticity of the oviduct, which may later expand and thus

allow subsequent eggs to be wider (Robertson and Cooke 1993). The deposition of more
albumen along the ﬁrst egg’s longitudinal axis may only partially offset the deﬁcit along
the transverse axis, as too great an elongation may interfere with embryonic development
and lower hatching success (Robertson et al.1994). Thus, both the width and the size of
ﬁrst eggs could be anatomically constrained, but this hypothesis has never been tested.

The physiological efﬁciency of the ovary (Parsons 1976) and oviduct (Leblanc
1987; Williams et al.1993) may increase as laying progresses, thus allowing subsequent
eggs to attain greater sizes than initial ones. However, no speciﬁc mechanism has been
proposed to underlie this presumed phenomenon. Some supporting evidence for this kind
of physiological "inertia" comes from coots (Arnold 1991): the proﬁles of ﬁrst clutches
are arched, but in continuation clutches (those laid within a few days of the initial clutch’s
loss), early eggs are the largest, and egg size initially declines with laying order.

Only 2 studies have described clutch proﬁles in ardeids, one in 3 North American
species (Custer and Frederick 1990), and the other in the Purple Heron (Ardea purpurea)
in Spain (Jover et al.1993). Neither of them, however, examined in detail the proximate
causes of the observed trends. I studied the clutch proﬁle of the Squacco Heron (Ardeola
ralloides), a poorly known species, and found that, usually, both ﬁrst and ﬁnal eggs are
signiﬁcantly smaller than intermediate ones. I also collected data on the variation in egg
size and hatching intervals among clutches, which allowed me to test the predictions of
the above hypotheses concerning the proximate factors that may affect clutch proﬁle.

In the gulls studied by Parsons (1976), clutch size and the timing of the onset of
incubation varied little among females. In species with strong variation in incubation

patterns (like the Squacco Heron, see Preliminary Analyses), the length of time over

which hormonal constraints will operate on the reproductive system of females, and the
consequent deﬁcit in the size of ﬁnal eggs, should also vary accordingly. The time of the
start of incubation can be inferred from a clutch’s hatch span (the interval between the
hatching of the ﬁrst and last eggs), as parents can control their eggs’ hatching patterns
through incubation in various birds, including raptors (Wiebe et al.1998), larids (Parsons
1972; Nisbet and Cohen 1975), and passerines (Magrath 1992a; Vega and Vinuela 1993).
Therefore, Parsons’s (1976) hormonal hypothesis would be supported by a negative
relationship between hatch span and the relative size of ﬁnal eggs. '

The nutritional hypothesis, as it was formulated by Parsons (1976), predicts a
positive relationship between the relative size of ﬁnal eggs and female resource avail-
ability for egg formation, independently of the onset of incubation and hatching patterns.
In testing this hypothesis, 1 considered nutritional resources for eggs as a whole, without
distinction between those that are stored in the female’s body and those acquired through
daily diet (Meijer and Drent 1999). I also inferred the level of female resources from the
mean egg size of clutches, an assumption that I will support in the respective section of
the Preliminary Analyses. This hypothesis thus predicts a positive relationship between
the mean egg size of clutches and their last eggs’ relative size, but no association between
either of these 2 variables and hatch span.

Houston, Jones and Sibly (1983) proposed that the onset of incubation, through a
restriction of female foraging time, limits the resources that can be invested in the clutch
(which should lower mean egg size), and the size of ﬁnal eggs. Their hypothesis therefore
predicts both a) a negative association between hatch span and the relative size of last

eggs (like the hormonal hypothesis), and b) a positive relationship between mean egg size

and last—egg size (like the hypothesis of independent nutritional constraints), as well as
c) a negative relationship between hatch span and the mean egg-size of clutches. This last
prediction is unique to this hypothesis, and thus allows it to be critically tested.

I also examined the variation in mean egg-size and hatch span by year, season,
and clutch size, in order to derive secondary predictions about trends in clutch proﬁle
among the levels of the latter 3 factors. For instance, I found that hatch span was longer
in S-egg than in 4-egg clutches, while mean egg size did not vary between them. Thus the
hormonal hypothesis would predict relatively smaller ﬁnal eggs in the larger clutches,
whereas Parsons’s nutritional hypothesis would predict no difference in clutch proﬁle
between them. (I will present these secondary predictions in detail in later chapters.)

The size of ﬁrst eggs has received a lot less attention than that of last eggs, and the
existing literature did not help me examine possible factors that may limit ﬁrst-egg size
as closely as those affecting ﬁnal eggs. Nevertheless, I tested the hypothesis of nutritional
constraints on ﬁrst-egg size by comparing the proﬁles of small- and large-egg clutches,
predicting relatively bigger ﬁrst eggs in the latter (N ilsson and Svensson 1993a). I also
analyzed egg shape (width/length ratio), to test the anatomical-constraint hypothesis’
prediction that ﬁrst eggs would be more elongate than subsequent ones (Robertson and
Cooke 1993). I was unable to derive predictions from and test the "physiological inertia"
hypothesis (Parsons 1976; Leblanc 1987).

In this study, which offers valuable information about the breeding biology of a
poorly known species, I examined the Squacco Heron’s clutch proﬁle, as well as certain
factors that may affect it proximately, at a level of detail never before applied to heron

research, and rarely seen in studies of wild birds.

II . GENERAL METHODS

1. ARDE OLA RALLOIDES

The Squacco Heron (Ardeola ralloides Scopoli) is a small-sized member of the
family Ardeidae, with males weighing 285 :t 37 g (mean :l: s.d.) and females 290 :1: 38 g
(Cramp and Simmons 1977). It has relatively short and strong legs and bill, and its habit
of keeping its neck retracted gives it a compact appearance. When not in ﬂight, it displays
a cryptic brownish-buff coloration, as its elongated mantle and scapular feathers drape
over its white wings (Voisin 1991). The biology of this heron has rarely been studied and
some aspects of its natural history are poorly known. The existing information has been
summarized by Cramp and Simmons (1977) and Voisin (1991).

The pond herons of the genus Ardeola are Old World species, mostly found in
tropical areas (Hancock 1999). A. ralloides is widespread in sub-Saharan Africa
throughout the year, but also has migratory populations that breed in scattered colonies
across southern and eastern Europe and western Asia, from Portugal to the Aral Sea
(Cramp and Simmons 1977). These Palearctic populations suffered signiﬁcantly in the
late 19th and early 20th centuries from excessive hunting for feather collection, and again
later in the last century due to habitat loss (Joseﬁk 1969, 1970).

Squacco Herons arrive in Europe in the spring and start nesting in late April or
early May (Cramp and Simmons 1977). They breed colonially, together with other tree-
nesting herons, both in coastal and inland areas, but avoid sites that depend exclusively
on saline feeding grounds (V oisin 1991). In Camargue, in southern France, they arrive

later than other sympatric herons, when most nesting sites in central colony areas and at

higher tree elevations are already occupied by heterospeciﬁcs. Thus they settle mostly in
peripheral areas and at lower tree heights than other species (Hafner 1980). Egg-laying
occurs in May and June, and hatching usually begins in early June; nestling care may last
into August, but the dispersal of early breeders and their ﬂedglings starts in July (Cramp
and Simmons 1977). Fall migration lasts from mid—August into October, and though
some birds stop in north Africa, most of the European population winters south of the
Sahara (Voisin 1991). .

The Squacco Heron’s mating, nesting and parental behavior has been studied by
Voisin (1980), who found that both sexes share incubation and nestling care. Incubation
lasts 22-24 days (Sterbetz 1962). According to Cramp and Simmons (1977), incubation
starts after clutch completion and hatching is synchronous, an unusual phenomenon in
ardeids (Voisin 1991). My results indicate that, in the Axios river delta, incubation started
early in the laying sequence, and hatching was partially asynchronous (see next chapter).
Females lay 3-6 eggs in Camargue, where the average clutch size is 4.7 eggs (Hafner
1980). Egg dimensions have been measured (Cramp and Simmons 1977), but egg volume
has never been estimated, and egg-size variability has never been examined before in this
heron. In Camargue, compared with 3 sympatric ardeids, Squacco Herons suffer similar
levels of egg loss, hatching failure and early chick mortality, but lower mortality of
nestlings over the age of 15 days (Hafner 1978). Young Squaccos begin ﬂying at about
the age of 30 days and leave the colony site approximately 2 weeks later (Cramp and
Simmons 1977). The growth and development of nestlings has never been studied in this
heron in any detail, though there are some data from a few chicks raised in captivity in

Madagascar (Werding 1970).

10

During the breeding season in Europe, Squacco Herons forage exclusively in
freshwater habitats, both in natural marshes and ponds, and in man-made ricefrelds and
vegetated ditches and canals (Sterbetz 1962; Voisin 1978). They are mostly solitary and
crepuscular hunters. They often remain concealed in aquatic vegetation, waiting for prey
to come within striking distance, or they wade slowly, often at the water’s edge, which
gives them the opportunity to pursue both aquatic and terrestrial prey (Voisin 1978). In a
comparative study with the Little Egret (Egretta garzetta), Hafner et al.(1982) found that
the Squacco Heron’s less active and solitary foraging methods were better suited for the
capture of larger, more elusive prey. Stomach analyses from Italy (Moltoni 1936) and
Hungary (Sterbetz 1962), as well as nestling regurgitations from Yugoslavia (Szlivka
1986), indicate that this heron’s prey consists mainly of small (< 10 cm) ﬁsh and

amphibians, and of aquatic insects and their larvae, in varying proportions.

2. STUDY AREA AND POPULATION

The Axios river delta (40°30’ N, 22°53' E) is part of a greater wetland complex,
which extends over 70 km2 along the west coast of Thermaikos Gulf in northern Greece.
(Athanasiou 1990). This area is protected under the Ramsar Convention, having been
recognized as internationally important for waterbirds, though large parts of the original
wetland system have been reclaimed for agriculture over the last century (Athanasiou
1990). Remaining natural habitats include salt- and freshwater marshes, tidal mudﬂats,
open sea, coastal lagoons and vegetated islets, riparian forest and tamarisk (Tamarix sp.)
bushland (Kazantzidis and Goutner 1996). Man-made riceﬁelds and irrigation canals also

attract various foraging waterbirds, including herons (Kazantzidis and Goutner 1996).

ll

Human activities in the area include, besides agriculture, cattle and sheep grazing, ﬁshing
and aquaculture along the coast, housing development, sewage and garbage dumping,
and hunting in the winter. Many of these activities, and at various degrees, continue to
encroach upon and degrade the area’s wildlife habitats and diminish the waterbird
populations that use them (Athanasiou 1990).

During the years of my study, a Squacco Heron population of 150-200 pairs
nested in the Axios delta as part of a 1500-pair multispeciﬁc colony, which also included
Little Egrets (Egretta garzetta), Black-crowned Night Herons (Nycticorax nycticorax),
Spoonbills (Platalea leucoroa’ia), and Great Cormorants (Phalacro'corax carbo). The
colony site was located in riparian forest along the banks of the Axios river, dominated by
6-8 m high tamarisks, with occasional stands of alder (A lnus glutinosa) and willow trees
(Salix sp.). A thick understory of 1-2 m high blackberry bushes (Rubus sp.) was often
present within 20 m of the river banks and made parts of the colony site inaccessible.

Squacco Herons started nesting at Axios in early May, when many heterospeciﬁc
pairs were already established at the site. They built their nests in tamarisks, at a height of
2-4 m, solitarily or in small clusters (personal observations). In 1993 they nested at lower
elevations and in shorter tamarisks than Little Egrets and Night Herons, and they were
concentrated in the thicker vegetation zone close to the river bank (Charalambides 1994).
Egg-laying lasted from early May until early or mid-June (see next chapter), and hatching
occurred throughout June and in early July. Young Squaccos were rarely seen around
their nests after the age of 30 days, and they dispersed from the colony site along with

their parents soon afterwards, so that few birds remained at it after the end of July.

12

Extensive riceﬁelds in the vicinity of the colony attracted foraging adults soon
after they were ﬂooded in May, when mole crickets (Gryllotalpa gryllotalpa) became
exposed by seeking refuge on levees, and again after they were colonized by amphibians
and aquatic insects (personal observations). Adult and juvenile Squaccos also foraged in
vegetated irrigation canals and freshwater marshes, but never in saline habitats. The
relative frequencies of 295 prey items found in nestling regurgitations were 35% Rana
frogs and tadpoles, 30% adult and larval aquatic insects, 20% mole crickets, 8.5% small

freshwater ﬁsh, and 6.5% various or unidentiﬁed items (Goutner et al. 2001).

3. DATA COLLECTION

I conducted this study at the Axios river delta heronry during the breeding
seasons of 1992, 1993 and 1994. I collected my data with permission from the Ministry
of Agriculture (General Secretariat of Forests and Natural Resources) of the Hellenic
Republic, and the MSU All-University Committee on Animal Use and Care.

Throughout my study, I visited the colony site in the morning (8:00-12:00) and in
the late afternoon (16:00-20:00), in order to avoid exposure of eggs and nestlings to the
intense sunlight of midday. Within the study site, I followed standard routs, so as to visit
each group of nests at approximately the same time of the day every time. I checked nests
in the egg-laying stage every 4 days, except for the early part of the 1992 season, when
I checked them every 3 days. I did so in order to minimize disturbance during this
potentially sensitive stage of the nesting cycle (Tremblay and Ellison 1979), but at the
same time to be able to estimate egg-laying order by differences in eggshell color (see

below). I marked nests with small (10><10 cm), numbered wooden tags, which I attached

l3

beneath them. I marked and measured the eggs I found in nests that I could access with a
2-m long ladder. I checked the contents of inaccessible nests with a mirror attached to a
pole (I use only clutch-size and laying-date data from such nests).

I marked eggs according to their laying order at both ends with small letters made
with a soft lead pencil. I did not mark the middle of eggs to avoid interference with the
normal egg-tuming patterns of incubating parents (Frederick and Collopy 1989), while I
marked both ends to avoid possible obliteration of single marks by parental droppings.

Whenever I found 22 new eggs, I estimated their laying order from differences in
eggshell color: fresh eggs are darker and more greenish, whereas older ones are paler and
more bluish (Custer 1991). The validity of these estimates is also supported by hatching-
order information. Reversals of egg-laying and hatching order are extremely rare in
ardeids (Dickerman and Gavino 1969; Fujioka 1984; Inoue 1985; Custer and Frederick
1990). I observed no reversals in 107 cases where both laying and hatching order were
certain (both the eggs and the hatchlings involved were ﬁrst seen on different days).
Therefore, I consider hatching order a reliable index of egg-laying order. In most cases
where hatching information was available, it conﬁrmed my laying-order estimates based
on eggshell color: 60 out of 62 cases in 1992 (97%); 50/54 cases in 1993 (93%); 64/66
cases in 1994 (97%). Because of this, I considered my laying-order estimates reliable
enough to be used even in the absence of hatch-order information (but I corrected the few
estimates that were proven wrong by hatching data). However, there are some clutches
with uncertain laying order, because I could not estimate it at the time of egg discovery
and I also lack hatch-order information from those nests. I excluded such cases from

clutch proﬁle analyses, but I used their average egg volume data.

14

I measured egg length and width with vemier calipers and 0.05-mm accuracy.
From each egg I took duplicate measurements at right angles, which I averaged, unless
they differed by > 0.2 mm, in which case I took a 3rd measurement and I averaged the
2 closer ones (Custer and Frederick 1990). From these measurements, I calculated egg
volume using the formula

Volume = 0.509 X Length X (Width)2 ,
which I estimated from a sample of 40 eggs where I also directly measured egg volume
(details in the next chapter).

After ascertaining clutch completion (the same eggs present on 2 consecutive
visits), I only checked nest contents with the mirror every 4 or 8 days (depending on
workload). I restarted monitoring nests closely 3 weeks after the laying of their lst eggs,
when hatching was imminent. In 1992 I checked nests in the hatching stage every day,
and I considered dry and wet newfound hatchlings as 0.5 and 0.0 d old, respectively
(details in the next chapter). In the other 2 years, I made hatch checks every other day, in
order to minimize potential disturbance to the colony. In all years, I took certain body
measurements from hatchlings (of ages of 0.0 to 1.5 d), in order to estimate sibling ages
and hatching intervals with greater accuracy in 1993 and 1994 (seesection 6 in the next
chapter). I weighed hatchlings with a spring scale to the nearest 0.05 g, and I measured
twice, with vernier calipers and 0.05-mm accuracy, culmen and tarsometatarsal lengths.
In 1993 and 1994, I also measured the bill+head length (from tip of culmen to occipital
condyle) of wet hatchlings, which I correlated with egg volume, but could not use for age
estimation. Again, I averaged duplicate measurements, unless they differed by >02 mm,

in which case I took a 3rd measurement and I averaged the 2 closer ones.

15

4. STATISTICAL ANALYSES

In data analyses I generally followed the guidelines of Zar (1 996) and those of the
SYSTAT 6.0 statistical manual (SYSTAT 1996). For some procedures I also consulted
additional sources, which I will cite in later sections where I present the relevant methods.

I analyzed categorical data, like clutch size frequencies, with X" or Fisher exact
tests, depending on the number of groups compared. In continuous quantitative variables,
I tested the normality of sample distributions with Lilliefors tests, and the homogeneity of
variances among samples with Bartlett’s X" tests. When these conditions were met, I used
parametric procedures (6. g. ANOVA), and when not, their non-parametric alternatives
(e.g. Kruskal-Wallis test). In the following chapters, I will not present in detail the testing
of parametric assumptions, unless they were violated.

I analyzed clutch proﬁle, and the effects of various factors on it, with Repeated
Measures ANOVA (von Ende 1993), a procedure that requires the fulﬁllment of certain
additional conditions, and which I will discuss in detail in Chapter IV.

In the presentation of results, I will abbreviate the names of certain variables.

These can be found in the List of Terms and Abbreviations (p. xii).

16

III . PRELIMINARY ANALYSES

In this chapter I present and discuss results that provide necessary background
information and pave the way for the clutch proﬁle analyses, which are the focus of my
dissertation. The text is divided into sections according to the data analyzed. The ﬁrst 4
sections concern basic aspects of breeding biology, including egg-laying intervals, clutch
laying dates, and clutch size, as well as the estimation of egg volume from linear egg
dimensions.

In sections 5 and 6, I will present the analyses of clutches’ average egg volume
(AvVol) and hatching intervals, with emphasis on the total hatching span (H-Span). The
hypotheses I examine with regard to the proximate causation of clutch proﬁle postulate
that the latter is affected by resource limitations or hormonal constraints associated with
the onset of incubation. I have been unable to evaluate these 2 factors directly, and I infer
their levels, respectively, from measurements of AvVol and H-Span. ( I will discuss the
suitability of these 2 variables as indices of the above factors later in this chapter.)

I can directly derive from the nutritional and hormonal hypotheses the predictions
that last eggs will be relatively smaller in clutches with lower AvVol and longer H-Span,
respectively (see Introduction). Furthermore, if AvVol or H-Span vary signiﬁcantly
among the levels of other factors (year, season, clutch size), the respective hypotheses
would predict a corresponding difference in clutch proﬁle among the levels of the same
factors. For example, I will show that clutches of 4 and 5 eggs have a similar AvVol,
while H-Span is signiﬁcantly longer in the larger clutches. Therefore, the nutritional

hypothesis will predict no variability in clutch proﬁle between clutch sizes, whereas the

17

hormonal hypothesis will predict relatively smaller ﬁnal eggs in the larger clutches. The
derivation of such secondary predictions about the variation in clutch proﬁle among the
levels of factors other than AvVol and H-Span is the reason for the detailed analyses of
these 2 variables in this chapter.

In the last (7th) section, I will examine a) the association between AvVol and
H-Span, and b) the hypothesis that last-egg size is constrained by resource limitations
imposed on laying females by the restriction of their feeding time due to the onset of
incubation (Houston et al.1983). In each of the following sections, I will begin with some
introductory comments (including data characteristics and statistical analyses employed),

and then I will proceed with the presentation and discussion of results.

1. EGG-LAYING INTERVALS

Studies of such disparate avian species as the Adelie Penguin Pygoscelis adeliae
(Astheimer and Grau 1985) and the Eurasian Kestrel Falco tinnunculus (Meijer et a1.
1989), indicate that egg-laying intervals correspond to the intervals between the onset of
rapid yolk deposition in successive ovarian follicles. As the laying of an egg precedes the
ovulation of the next follicle usually by <1 h (Johnson 1986), the laying interval also
represents the period when the latter egg passes through the oviduct, where albumen, the
egg and shell membranes, and the eggshell are deposited (Burley and Vadehra 1989).

Egg-laying intervals may vary among and within clutches in various bird groups,
including anatids (Schubert and Cooke 1993), ptarmigans (Wiebe and Martin 1995),
kestrels (Aparicio 1994a), and parrots (Beissinger and Waltman 1991). Longer laying

intervals in Eurasian Kestrels (Aparicio 1994a) and gaps in the normal laying sequence

18

of Blue Tits (N ilsson and Svensson 1993 b) were associated with food shortages. In the
Common Eider (Somateria mollissima), the average egg-laying interval decreases with
increasing clutch size, and the interval between the last 2 eggs is shorter than all previous
ones (Watson et al.1993).

Subsequent analyses require some information about the time interval between
the laying of successive eggs. First, an estimate of the modal laying interval is needed for
the calculation of clutch initiation dates (see next section). Second, consistent patterns of
heterogeneity in this variable could confound my hatching interval analyses (see section 6
in this chapter). Hatching intervals depend on laying intervals and on incubation patterns
(timing of onset and intensity) during egg-laying (Wiebe et al.1998). As I mentioned
earlier, I intend to use the total hatch span (H-Span) of clutches as an index of female
incubation behavior. I can do this safely only if I ascertain that laying patterns remain on
the average constant among groups of clutches that may vary in H-Span.

I was unable to measure egg-laying intervals with accuracy, because they were
shorter than the intervals between nest visits (3 d in the early 1992 season and 4 d in all
other periods). The data I use are frequencies of cases where different numbers of eggs
were laid during 3-d and 4-d nest-check intervals. Overall results, excluding cases where
only the ﬁrst or last egg was laid between nest visits, are shown in Table 1.

Results from the larger sample of 4-d visits indicate that eggs must usually be laid
every other day (in 85% of the cases, 2 eggs were laid in 4 (1). Even the laying of 3 eggs
in 4 (I does not necessarily indicate shorter laying intervals: the lst of these eggs could
sometimes have been laid soon after the lst visit, the last egg just before the 2nd visit,

and the middle egg in the middle of the 4-d interval.

19

Table 1. Frequencies of cases where different numbers of eggs
were laid during 3-d and 4-d nest-check intervals.

 

3-d nest-check intervals 4-d nest-check intervals

 

# eggs laid # cases % cases # eggs laid # cases % cases

 

l 32 27 2 231 85
2 88 73 3 40 15
Total 120 100 Total 271 100

 

 

 

 

When I visited nests every 3 d, I never found single new eggs after 2 successive
intervals. In 80% of 63 cases where I have 2 observations from each nest, a total of 3 eggs
(1,2 or 2,1) were laid during 6 d (2 successive 3-d intervals). So, again, in the majority of
cases, eggs seem to be laid every other day. In the remaining 20% of those 63 cases, a
total of 4 eggs were laid in 6 d (2 eggs in each of 2 successive 3-d intervals). Again, there
may have been cases where the lst of those 4 eggs was laid right after the lst visit, the
4th egg just before the last visit, and all egg-laying intervals were about 2 (1 long.

I calculated a rough average for egg-laying intervals for both observation regimes
by dividing the time interval between observations by the number of eggs laid in it, and
then averaging values over all cases. The resulting averages were 1.90 d for 3-d nest-
check intervals, and 1.83 d for 4-d intervals.

In conclusion, my observations indicate that eggs must usually be laid every 2 d,
and occasionally at shorter intervals, and that results are independent of the time interval
between nest visits. These results agree with data from a variety of middle-sized ardeids
studied in temperate areas (Jenni 1969; Maxwell and Kale 1977; Tremblay and Ellison

1980; Fujioka 1984). Grey Herons Ardea cinerea (Milstein et al.1970) and Cattle Egrets

20

Bubulcus ibis (Fujioka 1984) lay their eggs mostly in the early morning, so their egg-
laying intervals often take values of integer days. Little Egrets in Japan, however, may
lay eggs at various times of the day, at an average interval of 37 h (Inoue 1985).

In calculating clutch initiation dates, I will assume a constant 2-d egg-laying
interval. As long as laying intervals do not consistently vary among groups of nests, this
practice will be safe, and my hatching interval results will also be free of bias. My data
indeed suggest that laying intervals do not vary signiﬁcantly among the levels of any
factor also examined for its effect on hatching intervals.

Since I visited nests every 3 (1 only in the early 1992 season, data from this period
cannot be used for comparisons among years or between early and late clutches (initiated
before/after annual median laying dates). Moreover, since I lack 4-d visit data from the
early part of the 1992 season, I examined the seasonal variability in egg-laying intervals
using data from 1993 and 1994 only. This variability was low and non-signiﬁcant (Table
2). Therefore, the use of only late-nest data from 1992 should not bias the results of the
comparison among years. Again, the relative frequencies of cases where 2 or 3 eggs were

laid in 4 d did not vary signiﬁcantly among samples (Table 3).

Table 2. Frequencies of cases where 2 or 3 eggs were laid in 4 d for
early and late clutches, and result of statistical comparison.

 

 

Early clutches Late clutches
# % # %
2 eggs in 4 d 87 86 86 81
3eggsin4d 14 14 20 19
Total 101 100 106 100

 

 

 

2-tailed Fisher exact P = 0.354 n.s.

 

 

 

21

Table 3. Frequencies of cases where 2 or 3 eggs were laid in 4 d

by year, and result of statistical comparison.

 

 

 

 

 

 

 

 

l9 9 2 l9 9 3 l9 9 4

# °/o # °/o # ' %

2 eggs/4 d 51 88 74 80 106 88

3eggs/4d 7 12 19 20 14 12

Total 58 100 93 100 120 100
Pearson X3 = 3.623 d.f.= 2 P = 0.163 n.s.

 

In the comparison of laying intervals between clutches of 4 and 5 eggs, I used data
from both 3-d and 4-d observations, but I analyzed them separately. Both data sets failed
to show a signiﬁcant difference in laying intervals between clutch sizes (Table 4). In a
later section, I will show that hatching intervals between successive eggs increase with
laying order. Here I examine whether a similar increase in the length of laying intervals
contributes to this phenomenon. I compared laying intervals between early (1 st-3rd) and
late (3rd-5th) eggs, again analyzing data from 3-d and 4—d nest visits separately. In both

data sets, the effect of egg-laying order was weak and non-signiﬁcant (Table 5).

Table 4. Comparison between clutch sizes of frequencies of cases where

different numbers of eggs were laid during 3-d or 4-d intervals.

 

 

 

 

 

 

 

 

 

 

3-d nest-check intervals 4-d nest-check intervals
4-egg nests S-egg nests 4-egg nests 5-egg nests
# % # % # % # °/o
1 egg / 3 d 9 35 20 31 2 eggs / 4 d 75 82 156 84
2eggs/3d 17 65 45 69 3eggs/4d ll 18 29 16
Total 26 100 65 100 Total 86 - 100 185 100
2-tailed Fisher exact P = 0.805 n.s. 2-tailed Fisher exact P = 0.586 n.s.

 

22

 

Table 5. Comparison between early and late eggs of frequencies of cases where
different numbers of eggs were laid during 3-d or 4-d intervals.

 

 

3-d nest-check intervals 4-d nest-check intervals
Early eggs Late eggs Early eggs Late eggs
# % # % # % # %

 

1 egg / 3 d 16 27 15 25 2 eggs / 4 d 93 82 137 87
2 eggs / 3 d 43 73 46 75 3 eggs / 4 d 20 18 21 13
Total 59 100 61 100 Total 113 100 158 100

 

 

 

 

 

2-tailed Fisher exact P = 0.836 n.s. 2-tailed Fisher exact P = 0.390 n.s.

 

 

 

 

In summary, none of the factors I examined (year, season, clutch size, and laying
order) seems to affect egg-laying intervals. Therefore, any variability in hatching intervals

among the levels of these factors can be attributed to differences in incubation patterns.

2. LAYING DATE

The timing of avian egg-laying is generally considered to have evolved so as
to maximize breeding success by synchronizing the energetically demanding nestling
growth stage with the period of highest food availability in the environment (Lack 1954,
1968; van Noordwijk et al.1995). A seasonal decline in reproductive success may result
merely from the fact that parents of higher intrinsic or territory quality breed both earlier
and more successfully (Newton and Marquiss 1984; De Forest and Gaston 1996). Never-
theless, experimental studies have demonstrated seasonal trends in reproductive output
that track environmental variability and are independent of parental attributes (Verhulst

and Tinbergen 1991; Brinkhof et al.1993; Barba et al.1995).

23

Population variability around a common optimal laying date may result from
nutritional constraints on egg formation (Perrins 1970). Alternatively, individual optima
may vary according to parental or territory quality and the prevailing seasonal trends in
food availability and offspring value (Daan et al.1989; Aparicio 1998). In colonially-
nesting birds, like ardeids (Krebs 1978), this variation is often reduced, as nest synchron-
ization may decrease predation (Parsons 1975b; Forbes 1989; Hatchwell 1991), facilitate
mate attraction (Draulans 1988), or enhance feeding success (Krebs 1974) through the
exchange of information about proﬁtable foraging sites (Ward and Zahavi 1973).

Proximately, the hormonally controlled egg production is triggered by changes in
photoperiod, which sets a reproductive window (F ollet and Robinson 1980; Meijer 1989).
Within this window, in a variety of avian species, egg-laying may be hastened by high
ambient temperatures (Slagsvold 1976; Perdeck and Cave 1989; Meijer et al.1999),
increased food supply (Arcese and Smith 1988; Korpimaki 1989; N01 1989; Soler and
Soler 1996), and improved female nutritional condition (Jones and Ward 1976; Dijkstra
et al.1988; MacCluskie and Sedinger 2000).

My interest in the laying dates of Squacco Herons during my study is 2-fold. First,
annual variation at the colony level may reﬂect differences in environmental conditions
among years, which could also affect other reproductive variables. Second, I need clutch
laying dates and annual population median values in order to examine seasonal trends in
other clutch characteristics (egg size, hatching intervals, etc.).

I consider as the laying date of a clutch the date on which its ﬁrst egg was laid,
and I express it as number of days after April 30 (thus May 1 and June 1, for example,

take the values of l and 32, respectively). In all years, I discovered most nests during the

24

laying stage. To estimate the laying date of such clutches, I assumed that the last of the
eggs I found on my ﬁrst visit to a nest was laid on the previous day (half of the 2-d egg-
laying interval), and I counted back 2 d for every other egg present. However, in each
year, I discovered 10-20% of observed nests after clutch completion. In these cases I used
information about incubation intervals to estimate laying date.

I considered as an egg’s incubation interval the period between its laying and
hatching dates, even though each egg may not have been effectively incubated throughout
this period. Actually, within clutches, the median incubation interval decreases with egg-
laying order (Table 6). This is because Squacco Herons do not begin full incubation with
the ﬁrst egg of their clutch, but at some later point in the laying sequence (see discussion
of hatching intervals in section 6 of this chapter). In nests under observation during both
the laying and hatching stages, I calculated incubation intervals by subtracting eggs’
laying dates from their hatching dates. Statistics on this variable, which is not normally
distributed, are shown in Table 6. In all samples, the median and mode were equal. I used
the percentage of values equal to a sample’s median (%M in Table 6) as an inverse index
of sample variance. In both clutches of 4 and 5 eggs, the incubation intervals of the last
2 eggs were less variable than those of earlier ones (they have higher %M values). For
consistency, in the estimation of laying date, I used the median incubation interval of the
4th egg (20 d) in both clutch sizes, unless that egg failed to hatch. Then I used the median
interval of the 3rd egg in clutches of 4, and that of the 5th egg in clutches of 5 (both of
which are 20 d). I estimated the focal egg’s laying date by counting back 20 d from its
hatching date, and then I counted back another 2 d for every earlier egg in order to

estimate the clutch’s laying date.

25

Table 6. Statistics on incubation intervals by clutch size and egg-laying order.

 

 

 

 

 

 

 

 

A.CLSZ=4 EGG LAYING ORDER
Statistic A B C D
n 39 43 41 40
Median 23 21 20 20
% M a 38.5 34.9 61.0 60.0
B.CLSZ=5 EGG LAYING ORDER
Statistic A B C D E
n 86 87 84 87 73
Median 23 22 21 20 20
%Ma 31.4 35.6 41.7 70.1 61.6

 

 

 

 

3 Percentage of cases with a value equal to the sample median

Figure 1 shows histograms of laying date for the 3 years of my study. In 1992 and
in 1993, I kept searching for new nests until no more were constructed. In both of these
years, there is a small group of delayed clutches, separated from those initiated during the
main laying season by a number of days (1 in 1992 and 5 in 1993) when no new clutches
were started (in histograms A and B of Figure 1, line L separates main from delayed
nests). Such observations are common in ardeids, where late clutches represent either
second attempts after the loss of the initial clutch, or ﬁrst efforts by. late-arriving pairs
(Rodgers 1980a; Pratt and Winkler 1985; Campos and F emandez—Cruz 1991).

In 1994, the workload associated with other aspects of my study prevented me
from collecting data from clutches initiated after June 11 (indicated by line E in Figure

1C). Consequently, that year’s sample does not include delayed nests, and may also lack

26

some clutches initiated at the end of the main laying period. The latter are unlikely to be
numerous, because I observed no more than 20 new nests after June 11, most of which
were constructed after June 15, in a distinct cluster at the edge of the colony site, which
was typical of delayed nests in the previous 2 years. Therefore, the median laying date of
main clutches in 1994 is not likely to be underestimated by more than 1 day at the most.
Because of these characteristics of the 1994 sample, I compared among years the median
laying date of main clutches only (indicated by line M in all histograms of Figure 1).

I made comparisons with non-parametric tests, because the distribution of this variable

is not normal.

The Squacco Heron’s main laying season at the Axios delta was not very long. It
extended over 4-5 weeks in the 3 years, with 60-70% of all nests being initiated within 2
weeks only. In southern France, the species’ main laying season seems to be 1-2 weeks
longer, and delayed clutches may be laid throughout July (Hafner 1980).

The median laying date occurred progressively later over the 3 years of my study
(Figure 1). In 1992, it was 8 d earlier than in 1993, and 13 d earlierthan in 1994. These
differences are not great in absolute terms, but they represent substantial parts of the full
range of the main laying period (26-36 d). The variability in median laying date among
all years, as well as all pairwise differences, are statistically signiﬁcant (Table 7), the
latter even with the application of the Bonferroni correction for multiple comparisons
(critical P = 0.017). Moreover, since the 1994 median laying date may be slightly
underestimated, the comparisons between 1994 and the other 2 years may be

conservative.

27

A. 1992 (N=209)

 

35 l l I

 

 

 

 
  

 

 

 

 

' I I I - 0.16
30 — I | _
MI I L 0.14 g
25 — | A 0.12 3
E 20 l 10.10 3.
3 O
I -0.06 2
1° I -0.04 g!
5 I -0.02 ‘
0 ‘ ' 0.0
-1 7 15 23 31 39 47 55
LAYING DATE
B. 1993 (N=118)
16 I I l l T T I
I I ~ 0.12
1,
12— M ' ' L -0.10 3
*- I I 0 08 E
C . ..
a 8 I - s
o 0.06 8
' 004 "
4 I ’ g!
I 0.02 _.
0 0.0
-1 7 15 23 31 39 47 55
LAYING DATE

0.1994(N=142)

 

    

 

 

25 I I I l I I I I

. M , -0.16 1,
20- I l E “0.14 a
I | 0.12 '8
‘g 15 I 0.10 (3;-
3
g 10 I 0.08 U
I 0.06 2
5 1 0.04 g

0.02

0 0.0

-1 7 15 23 31 39 47 55

LAYING DATE

Figure 1. Laying curves by year. Line M: median laying date of main clutches.
Line L (1992 and 1993): limit between main and delayed clutches.

Line E (1994): end of observations. ( See text for explanations.)

28

Table 7. Overall and pairwise statistical comparisons of laying date among years.

 

 

Years compared Statistic P—value
All 3 H21 = 168.9 < 0.001
1992 and 1993 Ub = 4256 < 0.001

1993 and 1994
1992 and 1994

U= 4966 < 0.001
U: 2785 < 0.001

 

 

 

 

 

a Kruskal-Wallis test statistic. b Mann-Whitney test statistic.

Arrival at the breeding grounds and the onset of laying seems to be associated
with mean air temperature in various ardeids (Owen 1960; Pratt 1972a, 1972b; Campos
and Femandez-Cruz 1991), including the Squacco Heron (Sterbetz 1962). Data collected
during my years of study at a nearby weather station are consistent with this, as daily
mean air T° in the ﬁrst 2 weeks of May was highest in 1992 and lowest in 1994 (Table 8).
Moreover, the differences in the 2-week average T° between successive years are 1.56 °C
for 1992-1993 and 0.94 °C for 1993-1994, and their ratio (1.56 / 0.94 = 1.67) is similar to

that of the respective differences in median laying date (8d / 5d = 1.60).

Table 8. Daily mean air temperature (°C) during the ﬁrst 2 weeks of May by year.

 

 

 

 

 

P E R I O D
Statistic May 1-7 May 8-14 May 1-14 Y E A R
n 7 7 14
Mean 16.37 18.93 17.65 1992
n 7 7 14
Mean 15.89 16.29 16.09 1993
n 7 7 14 1994
Mean 14.89 15.41 15.15

 

29

 

The timing of heron egg-laying can also be inﬂuenced by food availability (Butler
1993). Great Blue Herons (Ardea herodias) in areas where feeding grounds were covered
by the 1993 Mississippi River ﬂood laid eggs signiﬁcantly later than in unaffected areas
in the same year or the same areas in a year with normal river levels (Custer et al.1996).
Food supplements have also hastened egg-laying in a variety of birds, including kestrels
(Meijer et al.1988), owls (Korpimaki 1989), corvids (Dhinsa and Boag 1990; Soler and
Soler 1996), and other passerines (Davies and Lundberg 1985; Arcese and Smith 1988;
Clamens and Isenman 1989).

I was unable to estimate Squacco Heron food availability during egg-laying, but
it is possible that it may have varied among the years of my study and contributed to the
annual differences in median laying date. If the latter is true, mean egg size and clutch
proﬁle, as well as other aspects of breeding biology, may also differ among years.
Therefore, in the following sections I will examine the annual variability of the data
under analysis, albeit without a priori expectations.

In later sections, 1 will also examine the seasonal trends of certain variables. In
order to do so, I grouped clutches initiated during the main laying period into early and
late ones, using each year’s median laying date as the dividing point. In each year, I put
clutches started on the median date in the group with the fewer cases, in order to keep
sample sizes as similar as possible. I kept the few delayed nests from 1992 and 1993 in a
separate sample, in order to make comparisons between those and the main clutches of
the same years. In some cases, I will compare laying dates among the levels of other
factors, pooling data from all years. For this, I will use values that I‘ standardized across

years by subtracting from them their year’s median.

30

3. CLUTCH SIZE

David Lack (1947) proposed that, in altricial birds, clutch size is ultimately
determined by the number of nestlings parents can successfully raise under prevailing
environmental conditions. This hypothesis predicts that a population’s average clutch size
should be the most productive one, taking into account the common trade-off between
offspring number and quality (Smith and Fretwell 1974; Smith et al.1989). However,
larger-than-average clutches can be the most productive (e.g. Perrins and Moss 1975),
and a population’s average clutch size may be evolutionarily constrained by the cost of
reproduction, that is the parental trade-off between current breeding effort and future
reproductive prospects (Williams 1966; Nur 1988). Parents are thus expected to
maximize lifetime ﬁtness by restraining their current effort in order to increase their
future survival and fecundity (Chamov and Krebs 1974; Dijkstra et al.1990). Positive
correlations between breeding success and parental survival (6. g. Coulson and Porter
1985), have led to the development of the individual optimization hypothesis of avian
clutch size (Perrins and Moss 1975). According to this, females lay the clutches that best
suit their individual circumstances, so that both experimental enlargements and
reductions of their broods lower their ﬁtness (Pettifor 1993; Pettifor et al.1988, 2001).

At the proximate level, the end of ovulation and clutch completion seem to be
related to the abrupt onset of female incubation (Mead and Morton 1985), or the gradual
development of that behavior beyond a threshold level (Haﬁom 1981; Beukeboom et a1.
1988; Meijer 1990). This probably results from a) the association between incubation and
high plasma prolactin levels (Goldsmith 1990; Buntin 1996; Vleck 2002), and b) this
hormone’s inhibition of gonadotropin secretion (Lea et al.1981; Ramsey et al.1985),

31

which 0) terminates follicular growth, ovulation, and egg-laying (Opel and Proudman
1980; Youngren et al.1991). A seasonal increase in female prolactin levels, which may
intensify the development of incubation of late nesters (Meijer et al.1990), could be
responsible for the seasonal decline in clutch size seen in many single-brooded species
(Klomp 1970), without any direct effect of female nutrition (Meijer et al.1988, 1990).
Female nutrition and body reserves, however, directly inﬂuence clutch size in many
anatids (Ankney and Afton 1988; Ankney and Alisauskas 1991; Erikstad et al.1993).
Moreover, food supplements can increase clutch size beyond the levels expected by the
hastening of laying in some raptors (Newton and Marquis 1981; Aparicio 1994b), gulls
(Bolton et al.1992), crows (Soler and Soler 1996), and other passerines (Nilsson 1991).

The examination of clutch-size variability in my study population provides some
necessary background information for my thesis. It also helped me identify certain groups
of clutches which may have been laid under resource-limited conditions, and whose egg
size may also have been nutritionally constrained (see later).

Historically, clutch size has been treated mostly as a continuous, normally-
distributed variable: authors have been reporting sample means and variances, and
subjecting data to parametric analyses (Lederle 1995). Avian clutch size, however, takes
only a limited number of discrete values, so, strictly speaking, such a treatment of data is
inappropriate. Below I will report average values of clutch size to illustrate differences
among samples, but I will make statistical comparisons with contingency tables.

In order to examine the annual and seasonal variation in clutch size, I used data
from nests initiated during the main laying period in each year. That is, I excluded from

samples delayed clutches, and also nests where the ﬁrst 1 or 2 eggs were lost before the

32

onset of incubation (I will examine the clutch size of these 2 samples at the end of this
section). Table 9 shows the frequencies of different clutch sizes during the main laying

season in each year of my study.

Table 9. Clutch size (Clsz) frequencies during the main laying season by year.

 

 

 

 

 

 

 

 

1992 1993 1994 All Years

Clsz # % # °/o # °/o # %
3 8 4.6 1 1.0 4 3.1 13 3.3
4 56 32.4 27 27.3 50 39.4 133 33.3
5 96 55.5 68 68.7 72 56.7 236 59.1
6 13 7.5 3 3.0 1 0.8 17 4.3
All 173 100.0 99 100.0 127 100.0 399 100.0

 

The observed range of clutch size was 3-6 eggs (Table 9), but only clutches of
4 and 5 eggs were common (respectively, about 33 and 60% of all cases). Clutches of 3
and 6 eggs made up about 12% of observations in 1992, and 4% in the other 2 years. The
overall average clutch size was 4.65 eggs (n = 399), which is very similar to the species’
mean clutch size of 4.7 eggs in the Camargue region of southern France (Hafner 1980).

Median clutch size was 5 eggs in all years. The 3 successive year averages varied
little, and were 4.66, 4.74 and 4.55 eggs, respectively (sample sizes in Table 9). I tested
the signiﬁcance of the annual variability in clutch size by comparing the numbers of
small (34 eggs) and large (25 eggs) clutches with Pearson’s X2 test. I followed this
grouping scheme, because, outside the common clutches of 4 and 5 eggs, observed
frequencies were usually less than 5. The test’s results approached, but did not achieve,

statistical signiﬁcance (X’= 4.88, df = 2, P = 0.087).

33

Weak ﬂuctuations in clutch size have been observed in ardeids both over a few
years (Rodgers 1980a; Moser 1986) and over much longer periods (Pratt and Winkler
1985; Campos and Fraile 1990). Signiﬁcant variation among 16 years in Little Egret
clutch size in Camargue seems to follow trends in food availability (Bennetts et al. 2000).
Food supplements also increased Great White Heron (A rdea herodias occidentalis) clutch
size, which seems to be currently constrained in Florida by habitat deterioration (Powell
and Powell 1986). The latter 2 studies, however, did not examine the possible covariation
of laying date with clutch size. Great Blue Heron clutches were smaller in areas adversely
affected by the 1993 Mississippi River ﬂood, but they were also laid later than in control
areas (Custer et al.1996), so the effect of food availability may have been only indirect.
Whether female nutrition can directly affect the size of heron clutches or not, the annual
variation in clutch size that I observed is too low to imply any strong differences in
Squacco Heron food availability among years, or to merit any special consideration in
subsequent analyses of other clutch characteristics.

Table 10 shows the percentage of small and large clutches, together with sample
size, mean and median clutch size, in the early and late parts of each year’s main laying
season. In all years there was a small seasonal decline in the percentage of large clutches
and in average clutch size. I tested the signiﬁcance of this trend with 2X2 contingency
tables (again comparing numbers of small and large clutches). Two-tailed Fisher exact
probabilities were > 0.05 in all successive years (0.349, 0.660 and 0.593, respectively),
and in the pooled sample (0.253). Data from 1992 and 1993 indicate a much greater
difference between main and delayed clutches (Table 11). The latter group had a lower

percentage of large clutches, as well as a smaller mean and median number of eggs.

34

The difference in the relative frequencies of small and large clutches between the 2

samples is statistically signiﬁcant (2-tailed Fisher exact P = 0.004).

Table 10. Percentage of small and large clutches, sample size, and mean and median

clutch size in the early and late parts of each year’s main laying season.

 

 

 

 

 

 

 

 

Clutch 1 9 9 2 1 9 9 3 1 9 9 4 All Years
Size Early Late Early Late Early Late. Early Late
3 4 33.4 40.5 26.0 30.6 40.0 45.0 33.7 39.5
2 5 66.6 59.5 74.0 69.4 60.0 56.0 66.3 60.5

n 84 89 50 49 65 62 199 200

Mean 4.75 4.58 4.80 4.67 4.58 4.52 4.71 4.58

Mdn 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0

 

Table 11. Frequencies of small and large clutches, and mean and median
clutch size of main and delayed nests from 1992 and 1993.

 

 

 

 

Clutch Main Laying Period Delayed Clutches
Size # % # %
s 4 92 33.8 20 60.6
2 5 180 66.2 13 39.4
A 11 272 100 33 100
Mean 4.7 4.2

Median 5.0 4.0

 

 

 

 

The lack of a seasonal decline in Squacco Heron clutch size during the main egg-

laying season may be partly due to this period’s short span. Similar patterns of clutch-size

variability (near constancy during most of the laying period and a drop towards its end)

have also been observed in A.ralloides in southern France (Hafner 1980), and in other

ardeids, mostly in cases where the laying curve shows a long right tail or a distinct group

of delayed nesters (Jenni 1969; Custer et al.1983; Pratt and Winkler 1985).

35

 

Small delayed clutches may be laid by young females, who both tend to nest later,
as in the Cattle Egret (McKilligan 1985), and lay fewer eggs than older ones, as in the
Grey Heron (Femandez-Cruz and Campos 1993). This combination of traits is common
in many avian groups, including anatids (Afton 1984; Forslund and Larsson 1992), gulls
(Sydeman et al.1991), owls (Korpimaki 1988), bee-eaters (Lessells and Krebs 1989), and
passerines (Perrins and McCleery 1985; Jarvinen 1991; Desrochers and Magrath 1993).
Delayed nests may also represent second attempts after the loss of the initial clutch
(Wolford and Boag 1971; Pratt and Winkler 1985). Repeat clutches tend to be smaller
than ﬁrst attempts in night herons (Wolford and Boag 1971), as well as in oystercatchers
(Nol et al.1984) and grouse (Erikstad et a1. 1985). In section 5, I will compare the average
egg size of delayed and main-period clutches, with the expectation that it will be lower in
the former group, due to increased nutritional constraints on the laying females.

In 4 nests in the early 1992 season, and in 2 in 1994, the ﬁrst egg was lost before
any others were laid. In half of these nests, a total of 5 eggs were laid and 4 remained
after the initial loss. In the other half, 6 eggs were laid in total and 5 remained. In another
2 early 1992 nests, the ﬁrst 2 eggs were lost sequentially, while they were the only ones
present in the nest. In both of these nests the total and ﬁnal clutch sizes were 7 and 5
eggs, respectively. These observations are summarized and compared with results from
regular clutches from 1992 and 1994 in Table 12. No statistical tests can be run, because
of the small number of nests where initial eggs were lost. However, comparisons of
clutch size percentages, means and medians, all indicate that the total size of clutches
with initial egg losses is greater than that of regular clutches, whereas their ﬁnal size is

very similar to that of regular clutches. Therefore, it seems likely that females were able

36

to lay extra eggs in order to replace those lost early.

There is some evidence that other ardeids can also replace lost eggs (Jenni 1969;
Maxwell and Kale 1977; McAloney 1973). Indeterminate egg-laying has been conﬁrmed
by egg-removal experiments in a wide array of avian taxa (reviews in Kennedy 1991, and
Haywood 1993a), including raptors (Porter 1975; Beukeboom et al.1988), coots (Arnold
1992a), gulls (Weidmann 1956; Parsons 1976), and passerines (Meijer 1993; Haywood
1993 b,c). The above studies, as well as others (Kennedy 1991), have shown that the
experimental manipulations must start from the beginning of egg-laying, so as to delay
the onset of incubation and the associated increase in prolactin, which is involved in the
termination of laying (see earlier in this section).

The laying of l or 2 extra eggs in the clutches where initial eggs were lost, must
have increased the nutritional and/or energetic demands on the respective females, and
may have thus constrained egg size in these clutches. Therefore, besides the delayed nests
discussed earlier, this is another group whose average egg volume I will compare with

that of regular clutches, expecting it to be lower (see section 5 in this chapter).

Table 12.- Frequencies, mean and median of total and ﬁnal clutch size (Clsz) in cases

where initial eggs were lost, and in regular clutches from 1992 and 1994.

 

 

 

 

Clutches with early losses Regular
Clutch Total Clsz Final Clsz Clutches
Size # % # % # %
s 4 0 0.0 3 37.5 118 39.3
2 5 8 100 5 62.5 182 60.7
All 8 100 8 100 300 100
Mean 5.88 4.63 4.61
Median 6.0 5.0 5.0

 

 

 

 

 

37

4. ESTIMATION OF EGG VOLUME

As I mentioned in the General Methods, 1 measured egg length and width with
vemier calipers and 0.05-mm accuracy. From these measurements I calculated egg
volume in cm3 (ml) using a formula that I estimated from a sample of 40 eggs whose
volume I also measured directly. To do the latter, I used the method of Evans (1969) and
Hoyt (1979), which is based on the weight and volume of the water displaced by the
immersion of eggs. I weighed each egg suspended in the air and in distilled water with a
spring scale to the nearest 0.05 g. The difference equals the weight of the displaced
volume of water, and the latter is the same as the immersed egg’s volume. Within the
range of ambient temperatures during data collection (15-20 0C), the speciﬁc gravity of
water deviates from 1.0 g/ml by fractions far below my measurements’ level of accuracy
(Evans 1969). Therefore, the above difference between weight measurements in g equals
egg volume in ml.

Figure 2 illustrates the relationship between egg volume and the product of the
multiplication of egg length by the square of egg width. The linear regression of egg
volume on this product ( n = 40 , P < 0.001 , R2 = 0.958 ) yielded the equation

Volume = 0.509 X Length X (Width)2 ,
which I used for the calculation of egg volume. The estimated slope is very close to the
average (0.507) calculated for 26 avian species from various families by Hoyt (1979).
The high R2 value indicates that only about 4% of the variability in egg volume cannot be
accounted for by this relationship. Nevertheless, since the values of egg volume I use are
estimates, I examined the possible overlap of conﬁdence intervals in clutches where

differences in egg volume were less than 5% (see next chapter).

38

 

Egg Volume (ml)

 

 

 

 

25 30 35 40
Linear Dimension Product (ml)

Figure 2. Regression of egg Volume on the product of egg Length X (Width)?

5. AVERAGE EGG VOLUME OF CLUTCHES

As I mentioned earlier, I will use the average egg volume of clutches (AvVol) as
an index of the amount of resources available to females for egg formation. The latter is
postulated by the nutritional hypothesis to be the main factor affecting clutch proﬁle (see
Introduction). I will thus use AvVol as a factor in the next chapter’s analyses to test this
hypothesis’ main prediction of a positive association between AvVol and the relative size
of last (and ﬁrst) eggs. In this section I will examine the variability in AvVol by year,
season and clutch size, in order to derive secondary predictions about trends in clutch
proﬁle among the levels of these factors. 1 will also discuss whether the assumption that
AvVol reﬂects female resource availability is justiﬁed, using information from the
literature (in the following paragraphs), as well as some supporting data of my own

(in the Results and Discussion part of this section).

39

Female resource availability and egg size

Egg production is a nutritionally and energetically demanding process for a wide
variety of female birds (Meijer and Drent 1999), especially when one takes into account
the costs of acquiring and metabolizing the resources that are invested in eggs (Perrins
1996). The physiological strain it exerts on females is illustrated by the fact that increased
egg production can lower their future ﬁtness (N ager et al. 2001). Many passerines invest
in eggs about the same amount of protein they use for their own metabolism during the
egg-formation period (Meijer and Drent 1999). In large non-passerines (e.g. arctic geese,
Ankney and Maclnnes 1978) the former may be >2 times the latter (Meijer and Drent
1999). The relative lipid requirements of eggs appear lower: 10-30 % of total needs in
non-passerines (Meijer and Drent 1999), and 10% of the total in the Great Blue Heron
(Butler 1993). However, the females’ own energetic needs are augmented by egg
production, as they need to forage for lipid-poor, protein- and calcium-rich foods, and the
developing eggs increase their body weight and the cost of locomotion (Jones and Ward
1976,1979; Ojanen 1983; Alisauskas and Ankney 1985; Houston et al.1995; Perrins
1996). Given these high demands of egg formation, females may well face resource
limitations during this process, which could affect their average egg size (see below).
When this is true, the latter variable can ’be used as an index of female resource
availability for egg production.

Under conditions of food plenty relative to egg-laying and metabolic require-
ments, most females in a population may be able to lay eggs of optimal size, and inter-
clutch differences in egg size may mostly reﬂect the variation in female body size in the

population. However, relatively few studies have demonstrated statistically signiﬁcant

40

(and mostly weak) correlations between uni- or multivariate indices of female body size
and average egg size in shorebirds (Miller 1979; N01 et al.1984), passerines (Ojanen et a1.
1979), and owls under high-food conditions (Hakkarainen and Korpimaki 1993). No
signiﬁcant relationship between mean egg size and multivariate indices of female size
(which are more reliable than univariate ones, Freeman and Jackson 1990) has been
found in studies of anatids (Leblanc 1989; Swennen and van der Meer 1992), kestrels
(Wiebe and Bortolotti 1995), and Eastern Kingbirds T yrannus tyrannus (Murphy 1986a).

The examination of female carcasses has shown a signiﬁcant association between
nutritional condition and egg size in gulls (Houston et al.1983) and kingbirds (Murphy
1986a). Statistically signiﬁcant (but often weak) correlations between average egg size
and female weight relative to body size (which reﬂects nutritional condition) have been
found in anatids (Leblanc 1989; Hepp et al.1987), American Kestrels Falco sparverius
(Wiebe and Bortolotti 1995), shorebirds (Galbraith 1988), and various passerine birds
(Wiggins 1990; Smith et al.1993; Nager and Zandt 1994; Horak et al.1995).

Studies of the Pied Flycatcher (Ficedula hypoleuca) in Finland (Jarvinen and
Vaisanen 1983, 1984), have indicated that the relative effects of female nutrition and
body size on mean egg size depend on environmental (food) conditions. In years and
areas with low temperatures and food (insect) abundance, egg size correlates with female
nutritional condition, which presumably depends on individuals’ foraging skills. When
and where temperatures are high and food is plentiful (so that most females can meet
their own and their clutches’ demands), egg size merely reﬂects female body size.

Egg size correlates with food availability also in other passerines (Otto 1979;

Murphy 1986b; Perrins and McCleery 1994), European Swifts Apus apus (O’Connor

41

1979), and American Kestrels (Wiebe and Bortolotti 1995). Egg size did not signiﬁcantly
vary between Great Blue Heron colonies adversely affected by the 1993 Mississippi
River ﬂood and control areas or years (Custer et al.1996), but the authors admit that their
egg-sampling protocol may have been inadequate.

Food supplements have increased mean egg size in some passerines (Hogstedt
1981; Ramsay and Houston 1997), but not in others (Hochachka and Boag 1987; Arcese
and Smith 1988; Arnold 1992b; Nilsson and Svensson 1993c). Neither did they affect the
egg size of Tengmalm’s Owls (Aegoliusfunereus) during a peak vole year (Korpimaki
1989). Extra food increased the mean egg size of American Coots (F ulica americana) in
one study (Hill 198 8), but not in another (Arnold 1994). Positive effects of supplemental
food on egg size have also been found in kestrels (Wiebe and Bortolotti 1995) and gulls
(Hiom et al.1991; Bolton et al. 1992).

Thus feeding experiments have yielded equivocal results regarding the effect of
female nutrition on egg size, but they also have their limitations and must be carefully
designed to produce meaningful results (Bolton et al.1992; Gehlbach and Roberts 1997).
In some of the above studies, the lack of signiﬁcant effects may be due to inappropriate
feeding protocols (Arnold 1992b) or high natural food availability (Korpimaki 1989).
Experiments with Lesser Black-backed Gulls have shown that extra food increases egg
size only when females face natural food shortages (Hiom et al.1991), and that results
also depend on supplemental food quality (Bolton et al.1992), which has also been
observed in Blue Tits (Ramsay and Houston 1997).

Therefore, whether mean egg size a) can be used as an index of female resources

for egg production, or b) it merely reﬂects female body size, seems largely to depend on

42

the speciﬁc environmental circumstances of the focal population during the period it was
studied. I was unable to assess Squacco Heron food availability or female nutritional

condition during my study. However, I will present below some data and arguments that
support the use of average egg volume as an index of female resources and as a factor in

the analyses of clutch proﬁle.

Results and Discussion

The average egg volume of clutches (AvVol) is a continuous quantitative variable,
and sample distributions did not deviate signiﬁcantly from normality (Lilliefors P-values
were >0.05). Therefore, I used parametric models in data analyses. Year is a random
factor with 3 levels. Season and clutch size are both ﬁxed and have 2 levels each: early vs
late (within the main laying period), and 4-egg vs 5-egg clutches. I was unable to test the
effects of all 3 factors simultaneously with a 3-way ANOVA, because sample variances
were heterogeneous (Bartlett X" = 23.734, df = 11, P = 0.014). Consequently, I performed
2 tests, using factor combinations that did not violate the model’s assumptions.

1 tested the effects of year and season on AvVol with a mixed-model 2-way
ANOVA. Both factors’ main effects, as well as their interaction were non-signiﬁcant
(Table 14). This is not surprising, as AvVol varied little among these factors’ levels
(Table 13). The largest inter-annual difference was 2% of the grand mean (16.134 cm3),
while the one between early and late clutches was only 0.6 % of that mean.

Assuming that AvVol reﬂects female resources for egg formation, these results
indicate neither annual nor seasonal variation in the latter factor. Consequently, the
nutritional hypothesis would predict no signiﬁcant differences in clutch proﬁle either
among years or between early and late nests within the main laying season.

43

Table 13. Statistics on AvVol (cm3) by year and season.

 

 

 

 

 

 

Y E A R S E A S O N
Statistic 1992 1993 1994 Early Late
n 90 69 88 132 115
Mean 15.975 16.152 16.282 16.186 16.074
S.E. 0.108 0.116 0.127 0.098 0.096

 

 

Table 14. ANOVA on AvVol by year (random factor) and season (ﬁxed).

 

SOURCE Sum Sq. d.f. Mean Sq. F-ratio P-value

 

YEAR 4.203 2 2.101 1.802 0.167
SEAS 0.830 1 0.830 4.716 > 0.400
Y * S 0.352 2 0.176 0.151 0.860
ERROR 280.995 241 1.166

 

Average egg size did not vary signiﬁcantly with clutch size in Purple Herons
(Ardea purpurea) in Spain (Jover et al.1993), or in 3 ardeids in the US. (Custer and
Frederick 1990). Egg size has also been found to remain essentially constant across
clutches of different sizes in anatids (Rohwer 1988), grouse (Moss et al.1981), coots
(Arnold 1992a), owls (Pietiainen et al.1986), corvids (Rofstad and Sandvik 1985; Heeb
et al.1994), and other passerines (Howe 1978; Slagsvold 1982; Smith et al.1993). The
lack of a possible trade-off (Smith and F retwell 1974) between these 2 variables may be
due to the proximate adjustment of clutch size to female nutritional circumstances (see
section 3, Clutch Size). Given this information from herons and other birds, my tentative
prediction was that AvVol would not vary between 4- and 5-egg Squacco Heron clutches

(the only ones from which I had adequate data for egg-size analyses).

44

As I mentioned in section 3 (Clutch Size), there are 2 groups of clutches where 1
expected resource limitations that could constrain AvVol. The ﬁrst one involves clutches
(of a ﬁnal size of 4 or 5) where extra eggs were laid to replace the loss of initial ones (see
section 3). Protracted laying of replacement eggs negatively affects their size and quality
in the Lesser Black-backed Gull (Nager et al. 2000). In my case, only 1 or 2 extra eggs
were laid, but the additional demands on the laying females may have limited egg size.

The second group consists of delayed clutches, which are generally laid by highly
resource-limited females (see section 2, Laying Date), who also tend to lay small eggs
(see earlier in this section). Late nesters may also be young individuals, as in the Cattle
Egret (McKilligan 1985). Young females lay smaller eggs, later than older breeders, in
many avian taxa, including grouse (Willebrand 1992), shorebirds (Thompson and Hale
1991), gulls (Reid 1988), terns (Nisbet et al.1984), skuas (Ratcliffe et al.1998), and
passerines (Crawford 1977; De Steven 1978). These traits of young breeders can be
attributed to resource limitations due to inferior foraging skills (Jansen 1990; Desrochers
1992) or restrained reproductive effort (Pugesek 1983; Hamer and Fumess 1991). In
herons, delayed nests may also represent second attempts after the loss of initial clutches
(Wolford and Boag 1971; Pratt and Winkler 1985). The physiological stress of laying 2
sets of eggs may be expected to constrain egg size in heron replacement clutches as in
other birds (Mills 1979; Runde and Barrett 1981; N01 et al.1984).

In summary, my predictions were that a) AvVol would not differ between regular
clutches of 4 and 5 eggs, but b) it would be lower in both extra-egg and delayed clutches
of 4 or 5 eggs than in regular clutches. These predictions are supported by the observed

differences in mean AvVol among these samples (Table 15).

45

Table 15. Statistics on AvVol (cm3) by clutch size (Clsz) and in clutches

laid under certain resource limitations (explanations in text).

 

Statistic CLSZ = 4 CLSZ = 5 Delayed Extra Eggs D + E E

 

 

 

n 94 153 6 7 13
Mean 16.220 16.081 15.274 15.493 15.392
S.E. 0.097 0.093 0.284 0.388 0.239

 

 

I tested the statistical signiﬁcance of these differences with a l-way ANOVA,
pooling extra-egg and delayed clutches in a single group (as sample sizes were small and
expected trends in AvVol were similar). The overall variability in AvVol was signiﬁcant
(F = 3.469, df = 2, 257, P = 0.033), and this was due to the lower mean of the pooled
sample of delayed and extra-egg clutches (D+EE). Tukey’s probabilities indicate that the
difference between regular 4-egg (C4) and 5-egg (C5) clutches was not signiﬁcant (P =
0.579), while the one between C4 and D+EE was (P = 0.024), and the one between C5
and D+EE approached signiﬁcance (P = 0.066).

These results support the assumption that, in my study, AvVol can be used as an
index of female heron nutritional condition, since it was lower in the group of clutches
expected to have been laid under increased resource limitations. The small size of the
D+EE sample may weaken this conclusion, but it does not invalidate it, as results were
statistically signiﬁcant. Another possible shortcoming is the small difference in mean
AvVol between the D+EE sample and the regular 4- and 5-egg clutches (about 5% of the
grand mean). However, this difference need not be very large, as the comparison was
between a group of nests (D+EE) where certain resource limitations of unknown severity

were expected, and the population at large (regular C4 and C5 nests). Within the latter,

46

resource availability for egg formation may well have varied, thus constraining the egg
size of some females and decreasing the sample’s mean AvVol. Besides, in species where
food supplements had signiﬁcant effects on egg size, the difference between experimental
and control groups was mostly of a similar magnitude (but sometimes higher): 5% of the
overall mean egg size in coots (Hill 1988); 6% in kestrels (Wiebe and Bortolotti 1995);
7% in tits (Ramsay and Houston 1997); 6-9% in magpies in 2 years (Hogstedt 1981); and
8-10% in gulls, depending on natural food availability and extra food quality (Hiom et al.
1991; Bolton et al.1992).

Squacco Heron average egg size may be inﬂuenced by female body size, as in
some other species (see earlier), but the above results indicate that it also depends on
female resource availability. Therefore, it can be used as an index of the latter, albeit an
imperfect one. Body size may inﬂuence AvVol, but no authors have ever suggested that it
can affect the allocation of resources among successive eggs and the proﬁle of a clutch.
Thus its interference may mask to some extent nutritional effects on clutch proﬁle, but it
is not expected to create any spurious association between AvVol and proﬁle. As I will
show in the next chapter, AvVol does have a signiﬁcant inﬂuence on clutch proﬁle,
which, according to the above arguments, can only be effected through an association
with female resource availability.

In conclusion, I believe that my use of AvVol as an index of female resources for
egg formation is justiﬁed. As I showed in this section, AvVol did not vary signiﬁcantly
among years, seasonally, or between clutches of 4 and 5 eggs. Therefore, the nutritional
hypothesis would predict no signiﬁcant variability in clutch proﬁle among the levels of

any of these 3 factors.

47

6. HATCHING INTERVALS

As I mentioned earlier, I will use the hatch span (H-Span) of clutches, that is the
time interval between the hatching of their ﬁrst and last eggs, as an index of the female
parents’ incubation patterns. The early development of incubation behavior, through its
hormone-mediated association with a depression of the function of reproductive organs is
postulated by Parsons’s (1976) hormonal hypothesis to be the main factor constraining
last eggs to a relatively small size. I will thus use H-Span as a factor in the next chapter’s
analyses to test this hypothesis’ main prediction of a negative association between hatch
span and the relative size of last eggs. In this section, I will present the hatching intervals
between successive eggs, whose sum equals H-Span. Then I will examine the variability
in the latter by year, season and clutch size, in order to derive secondary predictions about
trends in clutch proﬁle among the levels of these factors. First, however, I will explain
how I estimated sibling ages and hatching intervals, and I will discuss potential

problems with the use of H-Span as an index of female incubation patterns.

Estimation of sibling ages and hatching intervals

In 1992 I checked nests in the hatching stage every day. I kept a regular time
schedule, so successive visits to a nest were made approximately 24 h apart. I considered
wet hatchlings as 0.0 (1 old, because under local climatic conditions natal down dried in
1-2 hours. I discovered wet hatchlings throughout the periods I visited nests (8:00-12:00
and 16:00-20:00), which indicated that hatching could occur at anytime of the day (and,
presumably, the night). Therefore, chicks who were found dry (the majority of cases)

could have hatched at any point in the 24-h interval since my previous visit to their nest.

48

I assumed that they had hatched in the middle of that interval, and were thus 0.5 d old at
the time of discovery. In a few cases where chicks were close to emerging from their eggs
during a visit, but I ﬁrst saw and measured them the next day, I assumed that they were
1.0 (1 old on the latter day. From the above chick age estimates I calculated sibling age
differences, which correspond to their eggs’ hatching intervals.

In the other 2 years I made hatch checks every other day, in order to minimize
potential disturbance to the colony. I used the same ageing criteria as in 1992 for wet
hatchlings and eggs in an advanced hatching stage, but I estimated the ages of dry
nestlings from certain body measurements. In all years, I weighed hatchlings and I
measured their culmen and tarsus lengths (see General Methods). 1 also calculated the
ratio of body weight to egg volume (Wt/V 0]) whenever I was certain which egg each
chick had hatched from (majority of cases).

Several authors have suggested the use of multivariate techniques for the
estimation of overall body size (Rising and Somers 1989; Freeman and Jackson 1990),
nestling age (Gilliland and Ankney 1992), and adult gender (Phillips and Fumess 1997)
from morphometric data. Nevertheless, some studies have indicated that univariate
metrics may be adequate for ageing nestlings (Mineau et al.1982; Coleman and Fraser
1989; Wiklund 1996). In many cases, regression models were employed for chick-age
estimation, as the ages in question covered most of the nestling stage (Elowe and Payne
1979; Mineau et al.1982; Bortolotti 1984; Gilliland and Ankney 1992).

I was unable to employ multivariate methods, because time limitations prevented
me from measuring some hatchlings’ linear dimensions. This is true both for the 1992-

sample, which I used for the development of ageing criteria, and especially in the other 2

49

years, when I applied these criteria. Thus I estimated and applied ageing criteria for each
of the 4 morphometric variables independently.

Given my 2-d visits in 1993 and 1994, and using the same assumptions as in
1992, dry chicks could be aged as 0.5 - 1.5 d old. Within this age span, I had adequate
1992-data for the development of ageing criteria only from 0.5-d and 1.5-d nestlings
(the sample of 1.0-d old chick measurements was very small). Since there were only 2
alternative ages (0.5 and 1.5 d), I did not employ a regression model. Instead, using 1992-
data, I estimated for each variable the value that had the same probability of belonging to
either age’s distribution, and used it as an ageing criterion. Speciﬁcally, for each variable,
using sample means (’71) and standard deviations (si), 1 estimated a value X that was
associated in both ages’ samples with t-values ( ti = [X- 2,] / si ) which, given their
respective degrees of freedom (dfi = n, -1), had the same probability (P) of belonging to
the t-distribution. Therefore, this X-value (referred to as Criterion in Table 16) had the
same probability of belonging to either age’s distribution, and chicks with measurements
<X (or >X) were more likely to be 0.5 d (or 1.5 (1) old.

Table 16 shows statistics on Wt/Vol, weight, and tarsus and culmen lengths of
0.5-d and 1.5-d old nestlings from 1992. It also shows, for each variable, the estimated
criterion for making ageing decisions in the other 2 years, and its associated P-value. The
latter is proportional to the degree of overlap between the 2 ages’ distributions, so the
lower it is the more reliably that variable can be used for ageing chicks. The best ageing
variable by far was Wt/Vol (P = 0.042). I used nestling weight (P = 0.091) only when
Wt/Vol was unavailable due to uncertain egg-chick matches. The 2 linear measurements

were less reliable (their P-values were 0.120 and 0.171).

50

Table 16. Statistics on body measurements from nestlings at ages of 0.5 and 1.5 d
in 1992, and values (Criteria) of equal probability (P-value)

of belonging to either age’s distribution.

 

 

 

 

 

 

 

 

 

 

 

WT/VOL (g/cm’) WEIGHT (g)
Statistic 0.5d 1.5d 0.5d 1.5d
n 57 32 87 42
Mean 8.156 12.297 13.350 20.045
S.D. 0.774 1.538 1.742 3.213
Lill P a 0.120 0.496 0.092 0.281
Criterion 9.540 15.725
P-value 0.042 0.091
TARSUS (mm) CULMEN (mm)
Statistic 0.5d 1.5d 0.5d 1.5d
n 39 20 87 42
Mean 12.840 14.550 9.507 10.440
S.D. 0.442 0.852 0.475 0.450
Lill P a 0.142 0.654 0.805 0.495
Criterion 13.375 9.985
P-value 0.120 0.171

 

 

 

 

 

“ Lilliefors probability of random deviation from the normal distribution

Therefore, in making ageing decisions, I employed the following rules:
1) Wt/Vol (or weight) took precedence and was corroborated by linear measurements.
2) When the latter were not taken, the decision was based entirely on Wt/Vol (or weight).
3) Agreement between Wt/Vol (or weight) and one linear measurement was decisive.

4) When there was strong disagreement between Wt/Vol (or weight) and both linear

51

measurements (well into the ranges of opposite ages), I made no ageing decision.
5) When there was weak disagreement between Wt/Vol (or weight) and the linear
measurements (on opposite sides of their criteria, but close to them), the chick was
assigned an age of 1.0 day.

Once I had estimated sibling ages on the dates on my visits to their nests, I could

calculate their hatching dates and the intervals between the latter in 1993 and 1994.

Suitability of Hatch Span as an index of female incubation

Studies of larids (Parsons 1972; Nisbet and Cohen 1975), kestrels (Wiebe et al.
1998), and passerines (Magrath 1992a; Vega and Vinuela 1993) have shown that parents
(mostly females) can control their eggs’ hatching intervals through their incubation
pattern (timing of onset and rate of development). Thus it seems reasonable to assume
that a clutch’s hatch span (H-Span) generally reﬂects parental (and in most of the above
cases, female) incubation patterns.

In the Squacco Heron, however, both sexes incubate (Voisin 1980; personal
observations), which is typical of ardeids (Blaker 1969; Pratt 1970; Rodgers 1980b;
Ashkenazi and Yom-Tov 1997). Thus H-Span can be inﬂuenced by male incubation,
whereas I would like to use it as an index of female incubation alone (because only the
latter can affect clutch proﬁle). Mates generally share incubation equally in the Little
Bittem Ixobrychus minutus (Langley 1983) and in the Grey Heron (van Vessem and
Draulans 1986a), and may begin to do so before the clutch is complete. It is therefore
more likely that, in the Squacco Heron, male and female incubation patterns during egg-
laying will resemble each other (with some random variation) rather than differ in a
consistent way that would invalidate the aforementioned use of H-Span. Nevertheless,

52

the male’s involvement is likely to increase random error in H-Span and thus decrease the
probability of detecting an existing relationship between H-Span and clutch proﬁle.

Besides parental incubation, hatching patterns also depend on egg-laying intervals
(N isbet and Cohen 1975; Wiebe et al.1998). I showed in the ﬁrst section of this chapter
that laying intervals do not vary consistently among the levels of any factor used in the
analysis of H-Span. In the next chapter I will also show that they do not vary among
clutches of different proﬁle types, either. Thus the variability in laying intervals may also
add to the random error in H-Span, but it cannot bias results.

As I explained earlier, my hatching interval estimates are not very accurate. This
measurement error in the data should occur randomly with respect to clutch proﬁle and
the other factors involved in the analysis of H-Span. Thus it is also likely to increase
random error, but is not expected to confound analyses. All sources of error in H-Span I
have so far discussed may mask its possible effects on clutch proﬁle, but are not expected
to create a spurious association between these 2 variables.

There is one factor which can affect H-Span in a consistent way with respect
to clutch proﬁle, but its bias will oppose the predictions of the hormonal-constraints
hypothesis. This factor is egg size, which is positively associated with the length of the
incubation period (Parsons 1972; Bollinger 1994). The hormonal hypothesis predicts that
when incubation develops quickly (and thus H-Span is expected to be long) last eggs will
be relatively small. Their shorter incubation period, however, will shorten H-Span and
cause the rate of incubation development to be underestimated. The opposite is expected
to happen in clutches with a delayed onset of incubation. ( I will discuss this issue in

more detail when I compare H-Span among proﬁle types in the next chapter.)

53

Statistical Methods

In analyses, I used data from clutches of 4 and 5 eggs where complete sets of
successive hatching intervals and the total H-Span were known. Hatching interval data
were not normally distributed, so I performed non-parametric analyses. 1 compared the
length of successive hatching intervals with Friedman’s test, using each clutch as a block.
I tested the effects of year, season and clutch size on H-Span with Kruskal-Wallis (for
year) and Mann-Whitney U tests (for the other 2 factors).

If 2 or more factors are found to affect H-Span, their effects will be confounded, if
they also happen to be inter-correlated. As I will show below, clutch size and season have
signiﬁcant effects on H-Span. To remove any confounding effects of clutch size, I tested
the other 2 factors’ effects in 4-egg (C4) and 5-egg (C5) clutches separately. In the 2nd
case, I compared the standardized laying date of clutches (see section 2, Laying Date)

among the levels of the other 2 factors to ascertain that it did not vary signiﬁcantly.

Results and Discussion

Table 17 shows statistics on successive and cumulative hatching intervals by
clutch size, pooling data from all years. These intervals are symbolized as H(i-j), where i
and j indicate the laying order of eggs that deﬁne each hatching interval. In 4-egg nests
there is no E-egg, so there is no D-E interval and H(A-D) equals H-Span.

Hatching intervals between successive eggs increase in length as egg-laying order
progresses. This trend is statistically signiﬁcant both in 4-egg clutches (Friedman’s sz =
29.7, df=2, P < 0.001) and in 5-egg nests (sz = 171, df=3, P < 0.001). Since egg-laying
intervals did not increase with laying order (see section 1), this increasing length of
successive H(i-j) must be due to the gradual development of parental incubation. In both

54

C4 and C5, the median A-B interval is 0.0 d, which indicates that incubation usually
begins after the 2nd egg is laid. The last interval in C5 is the only one with a median
length equal to the modal egg-laying interval (2.0 (1). Thus full incubation may not often
start before the 4th egg is laid. Parental incubation developed faster in C4, as their longer
A-D interval indicates (3.0 d vs 2.5 d in C5; U = 2019, P = 0.001). Hatching patterns in
other ardeids also indicate a gradual development of parental incubation during egg-

laying (Maxwell and Kale 1977; Werschkul 1979; Inoue 1985; Mock 1985).

Table 17. Successive and cumulative hatching intervals (in d) by clutch size (Clsz).
( Explanations in text.)

 

 

 

 

 

 

CLSZ Statistic H(A-B) H(B-C) H(C-D) H(D-E) H(A-D) H-Span
n 35 35 35 . 35 35
FOUR Min 0.00 0.50 0.00 . 2.00 2.00
Max 1.50 2.00 2.50 . 5.00 5.00
Median 0.00 1.00 1.50 . 3.00 3.00
n 73 73 73 73 73 73
FIVE Min 0.00 0.00 0.00 0.50 . 0.50 1.50
Max 1.00 2.00 3.00 3.00 4.50 6.50
Median 0.00 1.00 1.00 2.00 2.50 4.50

 

Five-egg clutches have a 1.5-day (or 50%) longer hatch span than 4-egg nests
(Table 17). This highly signiﬁcant difference (U = 465, P < 0.001), is produced by the
extra D-E interval of the larger clutches (I showed above that the A-D interval is longer in
4-egg nests). Nevertheless, the fact remains that last (5th) eggs in C5 are laid signiﬁcantly
later than last (4th) eggs in C4 relative to the onset of incubation. Thus the hormonal

hypothesis would predict the former to be relatively smaller than the latter. Because of

55

this difference in H-Span between C4 and C5, I tested the annual and seasonal variability
in H-Span in each clutch size separately. H-Span also varied a lot within C4 and C5 nests
(respective ranges were 2-3 (I and 1.5 - 6.5 d). This variation, which presumably reﬂects
differences in female incubation (with some random error, see earlier), allowed me to test
the hormonal hypothesis’ predictions concerning clutch proﬁle (see later).

I will show below that there is a signiﬁcant seasonal increase in H-Span in 5-egg
clutches. The above comparisons between clutch sizes are not confounded by seasonal
effects, as median standardized laying date did not vary signiﬁcantly between samples
(U = 1398, P = 0.426), and was actually earlier in C5 (0d vs 2d in C4). ( I will further
examine this issue at the end of this section.)

The annual variability in median H-Span was low in both clutch sizes (Figure 3).
There was no signiﬁcant year-effect either in C4 (Kruskal-Wallis H = 0.656, P = 0.720)
or in C5 (H = 0.826, P = 0.662). The annual variability in the standardized laying date of
5-egg clutches was also low, statistically non-signiﬁcant (H = 2.452, P = 0.293), and
unlikely to have confounded results (see also later). These results indicate that, according
to the hormonal hypothesis, there should be no variability in clutch proﬁle among years
in either 4- or 5-egg clutches.

In order to use H-Span as a factor in the next chapter’s clutch proﬁle analyses,

1 had to divide nests into those with short and long H-Span. I made this division in each
year and clutch size separately, using the median H-Span values shown in Figure 3 as
criteria : a) short H-Span < sample’s median < long H-Span ; b) I put nests whose value
equaled their sample median in the subset with the fewer cases, in order to keep sample

sizes as similar as possible.

56

 

 

 

 

 

5-5 I I 1
5.0 - -
24 29
’uT 4 5 — ..
5 20 /
‘c’ 4 0 — -
m
a?
L 3.5— -
.9.
E 3.0— o o o -
9 7 19 .
2.5 _ Clutch Size
0 CLSZ = 4
2.0 ' 1 ' I CLSZ = 5
1992 1993 1994
YEAR

Figure 3. Median hatch span by clutch size and year.

Sample sizes indicated above or below data points.

Statistics on H-Span by clutch size and season are shown in Table 18. Using all
available data (columns C4-All and C5-All), I found a signiﬁcant seasonal increase in
H-Span in C5 (U= 395.5, P = 0.003), but no seasonal trend in C4 (U= 109.5, P = 0.166).
This apparent interaction between season and clutch size could be a mere artefact of the
greater difference in median standardized laying date between early and late nests in C5
(12d) than in C4 (8d). In order to examine this possibility, I "trimmed" the C4-sample by
excluding cases with laying dates close to the overall median value-(results are shown in
column C4-Trim of Table 18). Thus, by excluding the latest among early C4-nests, and
the earliest among the late ones, I increased the difference in median laying date between
the 2 trimmed sub-samples to the same level as in C5 (12d). The difference in median H-
Span remained the same in magnitude (0.25 d) and statistically non-signiﬁcant (U = 64.5,

P = 0.234), despite the increased disparity in standardized laying date. I also applied the

57

opposite trimming to the C5-sample, excluding the earliest among early clutches, and the
latest among the late ones, in order to reduce the difference in median laying date to the
same level as in the complete C4-sample (8d). In these trimmed samples, the difference in
median H-Span retained its original magnitude (1d, see column CS-Triml in Table 18),
and its statistical signiﬁcance (U = 238, P = 0.041), despite the loss of power due to the

decrease in sample sizes.

Table 18. Median hatch span (H-Span) and standardized laying date (Std L D)
by season in various samples (explanations in text).

Sample sizes are given in H-Span row.

 

 

 

Variable Season C4 - All C4 -Trim C5 - All CS-Trim l CS-Trim 2
Early 15 11 35 26 17
H-Span 3.00 3.00 4.0 4.0 4.5
Late 20 16 38 27 19
3.25 3.25 5.0 5.0 4.5
Std L D Early - 3.0 - 5.0 - 7.0 -‘4.0 - 2.0
Late 5.0 7.0 5.0 4.0 3.0

 

 

 

 

 

These results suggest an actual difference between 4-egg and 5-egg nests in the
seasonal trend of H-Span. This could be related to parental reproductive strategies that
involve hatching asynchrony and brood reduction (Lack 1947; Magrath 1990). Late-
season nestlings gain less weight than earlier ones in various birds, including penguins
(van Heezik et al.1993), skuas (Furness 1983), alcids (Birkhead and Nettleship 1982), and
passerines (Slagsvold 1982; Magrath 1989). Late ﬂedglings also suffer higher mortality
in shearwaters (Perrins 1966), penguins (van Heezik et al.1993), cormorants (Harris et

al.1994), gulls (Nisbet and Drury 1972), and passerines (Perrins 1965; Krementz et a1.

58

1989; Naef-Daenzer et a1. 2001). Thus late breeders may increase hatching asynchrony in
order to improve the ﬁtness of their senior offspring by limiting the competitive ability of
their junior chicks (Mock and Forbes 1994), a strategy that would be more useful in
larger, more demanding broods.

The possible adaptive signiﬁcance of the seasonal trends in hatching asynchrony
is beyond the scope of my dissertation. These results, however, suggest the following
horrnonal-hypothesis predictions about clutch proﬁle: a) no seasonal variability in clutch
proﬁle in C4, and b) relatively smaller last eggs in late (relative to early) C5 nests. Since
predictions differ between clutch sizes, in the next chapter I will test the effect of season
on proﬁle in each clutch size separately.

Since the H-Span of 5-egg clutches increases seasonally, other factors’ effects on
H-Span would be confounded by a signiﬁcant variation in laying date among their levels.
To investigate this possibility, I further constricted the C5-sample (excluding the earliest
among early, and the latest among late nests), until the difference in median laying date
between early and late nests was only 5d (column C5-Trim2 in Table 18). The new sub-
samples did not differ in median H-Span statistically (U = 153, P = 0.784), which could
be partly due to the loss of cases and power, or numerically (both early- and late-clutch
medians converged to 4.5 d). It thus seems that this level of disparity in laying date (5d)
is not enough to produce a difference in H-Span between samples. In 5-egg nests, median
standardized laying date varied by a maximum of 4d among years and 3d among clutch
proﬁle types (see next chapter), and was 2d earlier than that of C4-clutches. Therefore, it
is unlikely that the observed trends in H-Span among the levels of these factors could be

confounded by seasonal effects.

59

7. RELATIONSHIP BETWEEN AV.VOL AND HATCH SPAN

Houston, Jones and Sibly (1983) proposed that, in gulls, incubation restricts
female foraging time, which limits resource availability for egg formation, and thus
constrains the size of eggs that are not fully-forrned by the time incubation begins. Given
the small clutch size of gulls (3 eggs) and the timing of relevant events, only the size of
last eggs is expected to be negatively affected in these species.

As I showed in the previous section, hatch span (which depends on the timing of
the onset and on the rate of development of incubation) varies a lot in the Squacco Heron,
both in 4-egg clutches (range of 2-5 d) and in 5-egg nests (1.5 to 6.5 d). If incubation
limits female heron foraging opportunities, the number of eggs that are affected, and the
level of their size-limitation, should vary strongly across these ranges of H-Span values.
Such strong variation should be reﬂected in the average egg size of clutches. Therefore,
Houston, Jones and Sibly’s (1983) hypothesis (HJS—hypothesis, for short) predicts a
signiﬁcant negative relationship between H-Span and AvVol.

I tested this predicted association with Spearman rank correlations (due to the
non-normal distribution of H-Span) in each clutch size separately, because of the great
difference in median H-Span between C4 and C5 nests. The 2 variables were independent
both in C5 (rs = -0.002, n = 73, P > 0.990) and in C4 ( rs = 0.277, n = 35, P > 0.100),
where the correlation coefﬁcient was actually positive (but not signiﬁcantly so).

These results refute the validity of the HIS-hypothesis for A. ralloides. The most

probable reason for this, is that incubation does not restrict the foraging opportunities of
female herons by increasing the amount of time they spend at their nests. Both sexes tend

to share breeding responsibilities equally in ardeids (van Vessem and Draulans 1986a ;

60

Voisin 1991), including the Squacco Heron (Voisin 1980). Competition for nesting sites
(Burger 1981; Fasola and Alieri 1992) and materials (Jenni 1969; Burger 1978) may be
intense in crowded heronries, where stealing of nest materials is common (Milstein et a1.
1970; Burger 1978; Voisin 1991). Thus, species like the Little Egret need to constantly
guard their nests, beginning well before the onset of incubation, to prevent them from
being dismantled and their sites occupied by rival pairs (Voisin 1991). Actually, the mean
duration of Grey Heron trips away from their nests is minimum before egg-laying and
maximum during incubation (van Vessem and Draulans 1986 b), but the total time of
absence may not vary in the same way, as the frequency of trips may be greater in the
former period. Territory and nest defense seems to be especially pronounced in the small-
bodied Squacco Heron (Voisin, personal communication; personal observations), and, for
the above reasons, is likely to be fully developed from the early nesting stages. Therefore,
the onset of incubation should not be expected to reduce female foraging time and
nutritional levels, and, as I showed above, it does not seem to limit mean egg size.
Houston, Jones and Sibly’s hypothesis may not fully account for the small size of
ﬁnal eggs even in larids. Nisbet’s (1973) ﬁnding of a positive association between male
feeding rates of females and last-egg size in Common Tems (Sterna hirundo) seemed to
support it, until it was noticed that early eggs beneﬁtted from extra feedings even more,
so that last eggs were relatively smaller in clutches with higher mean egg size (Reid
1987). These results, however, do not conclusively refute the HIS-hypothesis, because
male feeding rates dropped during laying, so ﬁnal-egg albumen synthesis may have been
resource-limited even in highly-provisioned females. Reid (1987) was able to increase the

relative size of last (3rd) eggs in the Glaucus-winged Gull (Larus glaucescens) by

61

providing females with additional food after they laid their 1st egg. This experiment,
however, gave early eggs little or no opportunity to beneﬁt from food supplements, so its
results do not strongly support the HJS-hypothesis. Salzer and Larkin (1990) showed that
egg-removal in the same species increased a) the period of female courtship-feeding by
the male, b) clutch size, and c) the relative size of 3rd eggs (which were no longer last).
These results are in agreement with the HJS-hypothesis, but are equally consistent with
Parsons’s (1976) hormonal hypothesis, which the authors overlooked. More recent
experiments on Lesser Black-backed Gulls (Hiom et al.1991; Bolton et al.1992) have
shown that food supplementation (from 2-3 weeks before the start, and until the end of
egg-laying) can increase both the average egg size of clutches and the relative size of
ﬁnal eggs. The latter, however, is due to an increase in yolk, rather than albumen, protein
(Bolton et al.1992). These results indicate that, at least in gulls, the yolk protein of last
eggs may be nutritionally constrained, but their albumen is limited by some other factor,
which could be the hormonal mechanism proposed by Parsons (1976).

In summary, the lack of a negative relationship between AvVol and H-Span
(the indices of female nutrition and incubation, respectively), refutes Houston, Jones and
Sibly’s (1983) hypothesis of nutritional constraints on egg size associated with the onset
of incubation in the Squacco Heron. In the next chapter, I will test the predictions of
Parsons’s (1976) hypotheses of independent nutritional and hormonal constraints on this
heron’s ﬁnal-egg size. As AvVol and H-Span varied independently, there was no risk of
their confounding each other’s possible effects on clutch proﬁle. Thus I was able to test
the 2 hypotheses, and draw at least tentative conclusions about their validity, without

experimentally controlling these variables.

62

IV. CLUTCH PROFILE ANALYSES

In this chapter I will address the main issue of my dissertation, the pattern of
intraclutch egg-size variation by laying order in the Squacco Heron. In particular, I will
a) test Parsons’s (1976) nutritional and hormonal hypotheses’ predictions about the
effects of various factors on the relative size of ﬁnal eggs, and b) examine hypotheses
concerning the relative size (and shape) of ﬁrst eggs.

For brevity, I will refer to the observed patterns of intraclutch egg-size variation
as clutch profiles, and to their analyses as clutch proﬁle analyses. The latter are not to be
confused with the speciﬁc statistical methodology known as "Proﬁle Analysis", which
includes multivariate tests (von Ende 1993). As I will explain below, I used univariate
Repeated Measures ANOVA models to test various factors’ effects on clutch proﬁle. For
simplicity, I will refer to egg-laying order as egg rank. I will also capitalize the names of
factors involved in analyses, and I will use abbreviations for some of them (the latter are
presented below, and listed on page xii).

In the ﬁrst section of this chapter, I will explain the statistical methods I used in
my analyses. In the next, I will summarize the speciﬁc predictions of the nutritional and
hormonal hypotheses I will be testing. The results of my analyses (section 3) concern the
average egg-size and -shape proﬁles of all clutches, the different proﬁle types observed,
and the effects of various factors on clutch proﬁle (tests of hypotheses’ predictions). In
the ﬁnal section of this chapter, I will discuss these results and their implications for the

proximate causation of intraclutch egg-size variation in Ardeola ralloides.

63

1. STATISTICAL METHODS

In the analysis of clutch proﬁle, one must take into account the fact that the sizes
of a clutch’s eggs are not independent, but can be thought of as deviations from the
clutch’s average egg size (AvVol), which may vary signiﬁcantly among females (Jover
et al.1993). Therefore, testing the effect of egg Rank on clutch proﬁle with a completely
randomized ANOVA would violate the model’s assumption of unit independence.
Moreover, it would be inefﬁcient, as the variability in AvVol among clutches would
inﬂate the model’s error term. An appropriate alternative, which I used in these analyses,
is the Repeated Measures ANOVA (von Ende 1993; Zar 1996). This model allows for
lack of egg independence within clutches, and removes the among-clutches variation in
AvVol from the error term that tests the signiﬁcance of clutch proﬁle.

The main effect of the within-clutches factor, egg Rank, pertains to the average
proﬁle of the whole sample of clutches used. Among-clutches factors can also be
incorporated, and their main effects represent differences in AvVol among their levels
(subsets of the total sample). The Rank X among-factor interactions concern differences
in clutch proﬁle among the latter factors’ levels. For brevity, I will refer to them as the
among-factors’ effects on clutch proﬁle. They are the most important component of these
analyses, because they are the actual tests of the different hypotheses’ predictions. For
example, the hormonal hypothesis predicts a difference in clutch proﬁle between 4-egg
and 5-egg nests (see next section), that is a signiﬁcant Rank X Clutch Size (Clsz) inter-
action, or a Clsz-effect on clutch proﬁle. Conversely, as AvVol did not vary between 4-
and 5-egg nests, the nutritional hypothesis predicts no Clsz-effect on clutch proﬁle (no
Rank X Clsz interaction).

64

The Repeated Measures ANOVA (RMA) makes the usual assumptions of normal
data distribution and homogeneity of sample variances, which I continued to test. It also
assumes a) circularity of the variance-covariance matrix for all levels of the within-factor,
and b) homogeneity of such matrices for the differences among the levels of Rank across
all levels of the among-factors (von Ende 1993). I will report test result probabilities,
generated by SYSTAT (1996), that have been corrected by the Huynh-Feldt (1976)
estimator for violations of the circularity assumption. There is no method for correcting
test results for violations of assumption (b), but RMA is robust to such violations as long
as sample sizes are similar (von Ende 1993). Moreover, I ran orthogonal contrasts for the
Rank main effect and its interactions with the among-clutches factors, which are free
from assumptions (3) and (b) above, and thus more reliable (SYSTAT 1996). Contrasts
are very useful also because they concisely describe the effect of Rank on egg size (the
shape of clutch proﬁle) and the way the latter is affected by the arnong-nests factors.
SYSTAT also runs multivariate RMAs, which some authors prefer (e.g. Jover et al.1993),
but such analyses also have their limitations (James and McCullogh 1990 ; von Ende
1993). In my case, uni- and multivariate analyses always produced similar results, and I
will present only the former.

As I mentioned above, the among-clutches factors’ main effects refer to trends in
AvVol among their levels. I have already presented the results of such analyses, run on a
sample of 247 nests, in the Preliminary Analyses chapter. I ran those tests in advance, in
order to derive predictions (from the nutritional hypothesis) about various factors’ effects
on clutch proﬁle. In this chapter’s RMAs I used only 216 (or fewer) of the above 247

nests, because I excluded 31 clutches where egg-laying order was uncertain. I checked

65

the among-factors’ main effects in the RMAs in order to ascertain that the same trends in
AvVol (and the same nutritional-hypothesis predictions) still held. As this was always the
case, I will not present these tests’ results in this chapter, but I include them in complete
RMA tables that can be found in the Appendix.

For the sake of simplicity, I will also present only pertinent orthogonal contrasts.
SYSTAT (1996) runs linear, quadratic and cubic contrasts. These refer, respectively, to
a) the clutch proﬁle’s overall linear trend, b) its curvature (relationship between middle
and marginal eggs), and c) irregularities (zig-zags) in the previous trends. I never found
signiﬁcant cubic trends, so I will not present the results of such contrasts. Moreover, I
will present results of the other 2 contrasts only when the respective overall Rank-effect
or interaction with among-clutches factors was statistically signiﬁcant.

All clutches used in the same RMA must have the same number of eggs, because
the within-factor (Rank) must always have the same number of levels. These differ
between clutches of 4 and 5 eggs. Thus, in analyses that required the simultaneous use of
clutches of both sizes (e. g. when Clsz was an among-factor), I reduced the levels of Rank
in the larger clutches, by averaging the volumes of 2nd and 3rd eggs, as suggested by
Jover et al.(1993). I will refer to the resulting 4 egg ranks as First, Intermediate, Penulti-
mate, and Last, and to all eggs except the ﬁrst and the last as middle ones.

I tested the effects of 5 among-nests factors on clutch proﬁle. Two of them are
expected by the nutritional and hormonal hypotheses to directly affect clutch proﬁle :

1) Average egg size (AvVol), with 2 levels (small 3 Year X Clsz median < large). I will
refer to this factor as Egg Size (Egsz), to avoid confusion when "AvVol" denotes data.

2) Hatch span (H-Span), with 2 levels (short and long; for nest assignment see p.56).

66

Another 3 factors may be expected by these hypotheses to indirectly affect clutch proﬁle,
depending on whether AvVol and H-Span vary signiﬁcantly among their levels :

a) Year, which in this case a is ﬁxed factor, because I have speciﬁc predictions (see next
section) about the variation in clutch proﬁle among its levels (the 3 years of my study).
b) Season, with 2 levels (early and late; for nest assignment to these categories see p.30).
c) Clutch size (Clsz), with 2 levels, 4-egg (C4) and 5-egg (C5) clutches. All predictions
about these 5 factors’ effects on clutch proﬁle are reviewed in the following section.

As I mentioned earlier, I have egg-size data from 216 nests (86 C4 and 130 C5)
where egg ranks were certain. The levels of 4 of the above 5 factors are known for all of
these nests, but I have hatching data from only 108 of them (35 C4 and 73 C5). Thus I
tested the effect of H-Span on clutch proﬁle in the latter (smaller) sample, and the effects
of the other 4 factors in the former (bigger) one. I could incorporate all of these 4 factors
in a single model, but this would result in the division of nests among 24 sub-samples
(equal to the number of all Year X Season X Egsz X Clsz level combinations), each
including 3-16 nests. Model assumptions would not be met under such conditions, so I
ran one RMA with Clsz and Egsz as among-nests factors, and a different one with Year
and Season. I performed the latter analysis in each clutch size separately, because the
seasonal effect on proﬁle was expected to differ between clutch sizes (see next section),
using the original C5-data (from 5 egg-ranks). I used the 4 modiﬁed ranks in the other
2 RMAs, because Clsz was a factor in both, and C4- and C5-data had to be compatible.
The total number of cases were thus divided into fewer sub-samples, with larger and less
disparate sizes (Table 19). Sample variances were homogeneous in all cases. and

distributions did not signiﬁcantly deviate from normality.

67

The data I used in RMAs are egg volumes. In ﬁgures depicting clutch proﬁles,
however, I will use a relative egg-size variable, the % deviation (%D) of each egg-rank’s
value from the clutch’s AvVol. This should facilitate visual comparisons of patterns, as
whole-clutch means (the equivalent of AvVol) will equal 0.0 in all samples compared.

Figures will indicate each rank’s mean %D and its 95% conﬁdence interval.

Table 19. Sizes of samples used in clutch proﬁle analyses (RMAs).

Factor levels and name abbreviations are explained in the text.

A. RMA by Hatch Span and Clucth Size

 

 

 

 

 

 

Short H-Span Long H-Span All
C 4 C 5 C 4 C 5 Nests
18 36 17 37 108

 

 

B. Separate RMAs by Year and Season in each Clutch Size

 

 

 

 

 

 

 

 

 

1992 1993 1994 All

CL SZ Early Late Early Late Early Late Nests
C4 13 17 11 7 18 20 86
C 5 26 22 21 15 25 21 130

 

C. RMA by Egg Size and Clucth Size

 

 

 

 

Low AvVol High AvVol All
C 4 C 5 C 4 C 5 Nests
43 65 43 65 216

 

 

 

 

68

 

2. PREDICTIONS TESTED

In this section I review the predictions of the nutritional and hormonal hypotheses
concerning the effects of certain factors on the relative size of ﬁnal eggs. These are based
on the results of the preliminary analyses I presented in Chapter III, and are summarized
in Table 20. At the end of this section, I will also present predictions concerning the
relative size and shape of ﬁrst eggs.

The nutritional hypothesis postulates that the relative size of last eggs depends
on the amount of resources available to females for egg formation, which I infer from
AvVol. This hypothesis predicts a positive association between AvVol and %D(L), which
should appear in the respective RMA as a signiﬁcant linear Rank X Egsz interaction.
According to the hormonal hypothesis, the size of last eggs is constrained by the early
onset and development of female incubation, which is inversely proportional to a clutch’s
hatch span. This hypothesis predicts a negative association between H-Span and %D(L),
and a signiﬁcant linear Rank X H-Span interaction in the respective RMA. Since AvVol
and H-Span are not correlated (see last section of the Preliminary Analyses), their effects
on clutch proﬁle are not expected to confound each other: each hypothesis’ main factor is
not predicted by the alternative hypothesis to affect clutch proﬁle (Table 20A).

The 2 hypotheses’ secondary predictions about the effects of Year, Season, and
Clutch Size (Clsz) on clutch proﬁle depend on the variability in AvVol and H-Span
among the levels of these factors. As I showed in the respective section of the previous
chapter, none of these factors affected AvVol signiﬁcantly. Therefore, neither are any of
them expected to affect clutch proﬁle according to the nutritional hypothesis. The results

of the analyses of H-Span (see section 6 in last chapter) suggest the following hormonal

69

hypothesis predictions: a) no inter-annual variability in clutch proﬁle in either Clsz; b) no
seasonal variability in proﬁle in C4, but relatively smaller last eggs in 5-egg clutches late
in the season (signiﬁcant linear Rank X Season interaction); 0) signiﬁcant Clsz-effect on
proﬁle (linear Rank X Clsz interaction), due to lower %D(L) in C5. These predictions are
summarized in part B of Table 20.

The 2 hypotheses’ secondary predictions are not as strong as their primary ones
for 2 reasons. First, AvVol and H-Span, even when they vary signiﬁcantly among the
levels of the secondary factors, may not do so at a level sufﬁcient to produce a detectable
effect on clutch proﬁle. Second, there may be additional factors which also vary yearly,
seasonally, or between clutch sizes, but affect clutch proﬁle in ways that oppose the
expected effects of AvVol or H-Span (though no such factors are readily suggested by the
literature). Nevertheless, it is still worthwhile to test these predictions, because they
provide additional tests of the nutritional and hormonal hypotheses.

Both hypotheses predict a lack of variability in clutch proﬁle among years, and
between early and late C4 nests. Therefore, the respective tests are not diagnostic (results
will support either both hypotheses or neither). Nevertheless, Table 20 indicates that both
hypotheses’ main predictions, and half of their secondary ones, differ and thus lead to
diagnostic tests. I should mention, however, that the 2 hypotheses are not mutually
exclusive. Resource limitations and hormonal constraints may both affect clutch proﬁle,
either independently or interactively (see next section).

The hypothesis of nutritional constraints on the size of ﬁrst eggs (Slagsvold and
Lifjeld 1989) predicts a positive association between AvVol and %D(F), which can be

evaluated by the same analysis that examines the effect of AvVol on %D(L). Since no

70

other factor had a signiﬁcant effect on AvVol, this hypothesis predicts no variation in

%D(F) by Year, Season, or Clsz. The hypothesis of anatomical constraints on ﬁrst-egg

size (Robertson and Cooke 1993) predicts more elongate ﬁrst eggs than subsequent ones,

and I tested it with RMAs on egg shape (width/length ratio) in C4 and C5 clutches.

Table 20. Summary of nutritional and hormonal hypotheses’ predictions
(descriptive and referring to RMA results).

A. Primary Factors and Predictions

 

 

 

 

 

 

 

 

 

 

 

 

 

 

F a c t o r Nutritional Hypothesis Hormonal Hypothesis
. Positive association between
Egg Srze
o a
( AvVol) AvVol and /oD(L) No effect on Proﬁle
Linear Rank X Egsz Interact.
Negative association between
H-Span No effect on Proﬁle H-Span and %D(L)
Linear Rank X H-Span Interact.
B. Secondary Factors and Predictions
Factor AvVol H-Span Nutritional Hypoth. Hormonal Hypoth.
Year Constant Constant No effect on Proﬁle No effect on Proﬁle
Greater Lower %D(L) Late
No effect on Proﬁle
Season Constant Late in Lin. Rank X Seas Int.
in either clutch size
C5 only in C5 only
Clutch Constant Greater Lower %D(L) in C5
No effect on Proﬁle
Size in C5 Lin. Rank X Clsz Int.

 

8‘ Last egg’s relative size, expressed as percent deviation from clutch’s AvVol

71

 

 

3. RESULTS

A. AVERAGE CLUTCH PROFILE

The average egg-volume proﬁles of 4- and 5-egg clutches are shown in Figures
4A and 48, respectively. I tested their signiﬁcance in each clutch size separately, using
the original 5 egg ranks (A-E) in C5, with RMA-models that also included 2 among-nests
factors (Year and Season). Here I will present the main effect of Rank (laying order) on
egg size, and I will discuss its interactions with the 2 among-factors later.

The signiﬁcance of the main Rank effect, and its linear and quadratic components,
are shown in Table 21 (A and B, for C4 and C5). For simplicity, I present only the F -
ratio, d.f., and P-value (complete ANOVA tables can be found in the Appendix). The
number of error d.f. differs between the test of the overall Rank-effect and the 2 contrasts.
This is because, in the latter, it depends on the number of clutches in the sample, whereas
in the former it depends on the number of eggs (ranks X clutches).

In both clutch sizes, the average proﬁle is "arched", that is marginal eggs (ﬁrst
and last) are smaller than middle ones (Figure 4). This trend is statistically signiﬁcant in
both samples, as quadratic contrasts indicate (Table 21). In C5 there was also an overall
decline in egg size with laying order, while there was a positive, but non-signiﬁcant linear
trend in C4 (see Figure 4, and linear contrasts in Table 21).

The clutch proﬁles shown in Figure 4 seem to indicate that, on the average, egg
size does not vary greatly within clutches. The mean difference betWeen the largest and
smallest rank is 4.3% of AvVol in C4 and 6.1% of AvVol in C5. As I will show later,
however, largest and smallest eggs vary in rank among clutches. When the maximum

difference is calculated within clutches. and then is averaged, its mean value is 8.7% of

72

AvVol in C4 (twice as big as before), and 10.8 % in C5 (almost twice as big). Such egg-

size differences within clutches may affect sibling growth and survival (see Discussion).

°/o Deviation from AV.VOL

O)

A

M

O

1
N

Is

Table 21. RMA on egg volume by egg Rank in C4 (A) and C5 (B).
( Original 5 ranks used in C5.)

 

 

 

 

 

 

 

S 0 u r c e F-ratio D.F. P-value

Rank 26.697 3, 240 <'0.001
Linear 1.228 1, 80 0.271
Quadratic 97.814 1, 80 < 0.001

Rank 69.482 4, 496 < 0.001
Linear 5.353 1, 124 0.022
Quadratic 301.526 1, 124 < 0.001

A. CLSZ = 4 (N=86) B. CLSZ = 5 (N=130)
I I I I 6 I I I I I

 

 

 

l l
% DeVIatIon from AV VOL
0 N
l I
l l

 

 

 

 

 

1 1 1 1 -6 1 1 1 1 1

A B C D A B C D E
EGG RANK EGG RANK

Figure 4. Average egg-size proﬁle by clutch size (Clsz).

Error bars represent 95% conﬁdence intervals.

Regression lines indicate each proﬁle’s linear trend.

73

The intraclutch trend in egg shape (width/length) was very weak in both C4 and
C5 nests (Figures 5A and SB), and non-signiﬁcant in the former (F = 1.897, df = 3,255,
P = 0.139). Despite the statistical signiﬁcance of the egg-shape proﬁle in C5 (F = 3.565,
df= 4,516, P = 0.008), and its quadratic term (F= 14.130, df= 1,129, P< 0.001), this
pattern of variation can hardly account for the 5 times greater differences in volume
between marginal and middle eggs (Figure 48). Moreover, variation among females in
the postulated initial oviduct inelasticity (which is expected to constrain the size of ﬁrst
eggs by limiting their width) should produce a positive association between relative ﬁrst-
egg volume and shape (% deviations from clutch means). These 2 variables, however,
were not correlated in C5 nests (Pearson r = -0.064, n = 130, P = 0.794), and were
negatively associated (though weakly so) in C4 clutches (r = -0.221, n = 86, P = 0.041).

Thus, ﬁrst-egg size does not seem to be anatomically constrained in the Squacco Heron.

A. CLSZ = 4 (N=86) B. CLSZ = 5 (N=130)

 

 

% Dev from Avg Shape
O
l
L

% Dev from Avg Shape
O
I
l

 

 

 

 

 

 

-2 - — -2 - -
-4 a - 4 - -
-6 1 1 1 1 -6 1 1 1 1 1
A B C D A B C D E
EGG RANK EGG RANK

Figure 5. Average egg-shape (width/length) proﬁle by clutch size (Clsz).

Error bars represent 95% conﬁdence intervals.

74

B. CLUTCH PROFILE ANALYSES

I described above the main effect of laying order (Rank) on egg size, which
affects the average appearance of clutch proﬁle. In this section, I will examine proﬁle
differences among the levels of certain factors, represented by RMA interactions between
Rank and these among-nests factors. These are the only terms I will present in RMA
tables here (complete ANOVAs can be found in the Appendix). I used the original 5 egg
ranks in C5 in the analysis by Year and Season (which was performed in each clutch size
separately), and the modiﬁed 4 ranks (ﬁrst, intermediate, penultimate, last) in the other
RMAs (in which I made comparisons between 4- and 5-egg clutches). The error d.f. for

orthogonal contrasts will again differ from those of the RMA error terms (see earlier).

Analysis by Hatch Span and Clutch Size

I tested the main prediction of the hormonal hypothesis, which concerns the effect
of H-Span on clutch proﬁle, on the sample of 108 nests (35 C4 and 73 C5) from which
hatching data are available. I also incorporated Clsz as an among-nests factor, in order to
examine whether the effect of H-Span on proﬁle varied between clutch sizes (by looking
at the Rank X H-Span X Clsz interaction). I will not examine the Clsz-effect on proﬁle
here, because I tested it in the larger sample of 216 nests from which egg-size data are
available (I present the results of that analysis later).

The results of the relevant RMA (Table 22) show a non-signiﬁcant 3-way inter-
action, indicating that the effect of H-Span on proﬁle is similar in both clutch sizes. This
is not surprising, because, according to the hormonal hypothesis, Clsz is expected to
affect clutch proﬁle only due to the difference in H-Span between its levels (see earlier),
that is for the same reason as the other among-nests factor in the model.

75

Table 22. RMA on egg volume by Rank (within-clutches factor),
and by H-Span and Clsz (among-clutches factors).

 

 

 

S 0 u r c e F -ratio D.F. P-value
Rank X H-Span 5.811 3, 312 0.002
Linear 9.673 1, 104 0.002
Quadratic 0.032 1, 104 0.858
R X H-Sp X Clsz 0.530 3, 312 0.620
A. Short H-Span (N=54) B. Long H-Span (N=54)
6 I I I I 6 I I I I

 

 

% Devrahon from AV VOL
0 N
j I
l l

% DeVIatlon from AV VOL
0 N
l I
l l

 

 

 

 

 

 

-2 _ _ -2 .. -
-4 - d -4 - -i
-6 1 r 1 1 -6 1 1 1 1
F l P L F l P L
EGG RANK EGG RANK

Figure 6. Clutch proﬁle by H-Span (pooling clutch sizes).

Error bars and regression lines as in Figure 4.

In nests with delayed incubation development, indicated by a short H-Span, egg
size increases from the ﬁrst until the penultimate egg, last eggs are larger than ﬁrst ones,
and the overall linear trend of clutch proﬁle is positive (Figure 6A). When incubation

develops early in the laying period (and H-Span is consequently long), the linear trend of

76

clutch proﬁle becomes negative, as penultimate eggs become smaller than intermediate,
and ﬁnal eggs smaller than ﬁrst (Figure 6B). Thus, as the hormonal hypothesis predicted,
H-Span (the index of female incubation) had a signiﬁcant (Table 22) negative effect on
the relative size of last eggs and, more generally, the linear trend of clutch proﬁle. The
change in the relative sizes of intermediate and penultimate eggs is also statistically
signiﬁcant. When only these 2 ranks are used in an RMA (ﬁrst and last excluded), the
Rank X H-Span interaction is still signiﬁcant (F = 8.544, d.f.=l,106, P = 0.004). This
shift of the proﬁle’s peak to an earlier position when H-Span increases is also consistent
with the hormonal hypothesis. Constraints on egg size, induced by hormonal changes
associated with the onset of incubation, should affect all eggs that are not fully formed at
the time of the latter (Arnold 1991). Thus, when incubation starts early in the laying
sequence (and H-Span is long), there may be time for these constraints to exert their
inﬂuence on penultimate eggs. This may not be the case when incubation starts later
(and H-Span is short), and last eggs may be the only ones affected.

The quadratic component of the Rank X H-Span interaction was non-signiﬁcant
(Table 22), indicating that H-Span did not affect that trend of clutch proﬁle. Indeed, both
proﬁles depicted in Figures 5A and SB are arched, and vary little in the overall difference
between marginal (ﬁrst and last) and middle (intermediate and penultimate) eggs, even
though there were differences in the relative sizes of eggs within these two groups (see
above). A signiﬁcant quadratic Rank X among-factor interaction was not expected in this
case, neither is it predicted by either hypothesis in any of the following analyses. Thus,

I will not discuss such terms, unless they happen to be signiﬁcant against expectations.

77

Analysis by Year and Season

I tested the effects of Year and Season on the proﬁles of 4- and 5-egg clutches
separately, because of the difference between the 2 groups in the hormonal hypothesis’
predictions concerning the seasonal variability in proﬁle (see earlier). Since the 2 samples
were not pooled or compared, I used the original 5 egg ranks (A-E) in C5 nests.

The proﬁle of 4-egg clutches did not vary signiﬁcantly either among years or
between early and late nests (Table 23 A). This is as both the nutritional and the hormonal
hypotheses predicted, so results are consistent with both. There was no seasonal variation
in the proﬁle of C5 nests (Table 238), even though the hormonal hypothesis predicted
relatively smaller ﬁnal eggs late, due to the seasonal increase in H-Span. On the contrary,
the annual variation in the proﬁle of C5 nests approached statistical signiﬁcance (Table

238), something that neither the hormonal nor the nutritional hypotheses predicted.

Table 23. RMAs on egg volume by Year and Season (among-nests factors),

and by Rank (within-clutches factor) in each clutch size (C4 and C5).

 

 

 

S o u r c e F-ratio D.F. P-value

Rank X Year 0.380 6, 240 0.867

A. C 4 Rank X Season 0.072 3, 240 0.961
RXYXS 1.528 6,240 0.181

Rank X Year 2.079 8, 496 0.060

Linear 2.069 2, 124 0.080

B. C 5 Quadratic 0.544 2, 124 0.181
Rank X Season 1.046 4, 496 0.369

R X Y X S 1.149 8, 496 0.334

 

78

 

 

 

 

 

 

 

 

 

 

 

 

 

6 I l I I I
6’
>. 4" ‘
2 2- -
5 0L _, A.1992(N=48)
.6 "2" ‘
.g -4l— J
G.)
0. -6r 1
o\
-8 n 1 L 1 1

A B C D E
_I 6 I I T I I
94L _
>
< 2_ I 1 I _ 13.1993 (N=36)
E
O
a: 0— _
8
F13
o 4r ‘
°\0 _6 g l l l L

A B C D E
_J 6 l I I I I
O
>. 44 _
>
< 2_ _ C.1994(N=46)
E
O
.. 0— 4‘ —
S
E '2' ‘
>
8 4- -
°\° '6 l l l L J

A B C D E

EGG RANK

Figure 7. Clutch proﬁle in 5-egg nests by Year.

Error bars and regression lines as in Figure 4.

79

Figure 7 shows that the annual variation in the proﬁle of 5-egg clutches was not
great (besides not being highly signiﬁcant statistically), and mainly involves the pattem’s
linear trend. The latter appeared to differ the most between 1992 (negative) and 1993
(horizontal), but not as much as between nests with short and long H-Span (positive vs
negative; Figure 6). This variation may indicate that additional, unidentiﬁed factors could
weakly affect clutch proﬁle in ways that do not involve H-Span or AvVol, but it does not
argue against the hormonal and nutritional hypotheses.

The lack of a seasonal trend in the proﬁle of 5-egg clutches is intriguing, given the
signiﬁcant seasonal increase in the H-Span of C5 nests. The latter, however, was found in
a sample of 73 nests of known hatching intervals (see section 6 in Preliminary Analyses).
For this RMA, I used a larger sample (130 clutches), where H-Span may have varied less
between early and late nests. This is unlikely, though, as the difference in median laying
date between early and late clutches is the same (12 d) in both samples.

A possible explanation for the disagreement between the hormonal hypothesis’
prediction and the observed results is that the seasonal variability in H-Span is not strong
enough to affect clutch proﬁle. In Table 24, I compare the difference in H-Span between
early and late nests with that between clutches with short and long H-Span (which, as I
showed earlier, was sufﬁcient to affect proﬁle). In both cases, the difference between
medians is the same (4.0 vs 5.0 d), but this may be misleading, as these were the modal
H-Span values in the respective samples. As minimum and maximum values show, the
ranges of short- and long-H-Span nests barely overlap @ecause this is how the 2 groups
were divided), while the H-Span range of early nests completely envelops that of late

clutches. In agreement with this, the difference in mean H-Span is much greater in the

80

former (1.6 vs 0.7 (1). Therefore, the seasonal variation in H-Span may be statistically
signiﬁcant, but inadequate to produce an observable effect on clutch proﬁle. It is also
possible that some other, unknown seasonal factor opposes the effect of H-Span on

proﬁle, but neither my data nor information from the literature suggest any such factor.

Table 24. Statistics on H-Span (d) by Season and by Hatch Span in 5-egg clutches.

 

 

 

Season Hatch Span

Statistic Early Late Short Long
N 35 38 36 37
Min 1.50 2.00 1.50 4.50
Max 6.50 6.00 4.50 . 6.50
Mean 4.0 4.7 3.6 5.2
Median 4.0 5.0 4.0 5.0

 

 

 

 

 

Analysis by Egg Size (Egsz) and Clutch Size (Clsz)

The effect of Egsz (which reﬂects AvVol, the mean egg volume of clutches)
on clutch proﬁle tests the main prediction of the nutritional hypothesis of a positive
association between AvVol and the relative size of last eggs. The effect of Clsz tests
indirectly the hormonal hypothesis, as H-Span (and the time between the onset of
incubation and the laying of the last egg) is longer in the larger clutches.

The results of an RMA by Egsz and Clsz on the total sample of 216 nests (86 C4
and 130 C5) are shown in Table 25. The among-nests part of the table shows that Clsz
had no effect on AvVol (as the preliminary analyses also showed). Thus its possible

effect on clutch proﬁle will not be confounded by any variability in AvVol between its

81

levels. The signiﬁcant among-nests effect of Egsz is not surprising, as this factor’s levels
represent low- and high-AvVol clutches. The signiﬁcant Clsz X Egsz interaction indicates
that the magnitude of the difference in AvVol between small- and big-egg nests varies
between clutch sizes (I will refer back to this later).

The Rank X Egsz X Clsz interaction in the within-clutches part of Table 25
approached statistical signiﬁcance. This indicates that the Egsz-effect on proﬁle may vary
between clutch sizes, and vice versa. Therefore, I tested each factor within each level of
the other, thus running 4 separate RMAs, each with 1 among-nests factor. These are
essentially 4 pairwise comparisons among the 4 different sub-samples (Egsz X Clsz level
combinations). Therefore, I applied the Bonferroni correction to the critical P-level,
which I reduced to 0.0125 (0.05 divided by 4). The average clutch proﬁles of the 4

samples are illustrated in Figure 8 (A-D).

Table 25. RMA on egg volume by Rank (within-clutches factor),
and by Egsz and Clsz (among-clutches factors).

 

S 0 u r c e F-ratio D.F. P-value

 

Among Clutches

 

Clsz 1.058 1, 212 0.305
Egsz 329.256 1, 212 < 0.001
Clsz X Egsz 5.024 1, 212 0.026
Within Clutches
Rank 89.695 3, 636 < 0.001
Rank X Clsz 6.011 3, 636 0.002
Rank X Egsz 3.989 3, 636 0.015
R X C X E 2.702 3, 636 0.061

 

82

% Deviation from AvVol

% Deviation from AvVol

A. CLSZ = 4, Low AvVoI (N=43) B. CLSZ = 4, High AvVol (N=43)

 

 

 

 

 

 

 

 

 

 

5 I I I l 6 1 I l I

4 - - 3 4 - -
>
z

2 - - E 2 r- -
9

O '- ‘ c 0 1- -‘
.9
32’

-2 - - g -2 i- -
O

-4 - “I °\° -4 I- -I

_5 1 1 1 1 -6 1 r 1 1

F I P L F l P L
EGG RANK EGG RANK

c. CLSZ = 5, Low AvVoI (N=65) o. CLSZ = 5, High AvVol (N=65)

6 T I I I 6 I l r I

4 - — 4 - -

O N
I I
l l
% Devnation from AvVol
O N
l I
L L

 

 

 

 

 

 

-2 b a -2 —- J
.4 I- -I -4 v- -1
’6 l l I l ‘6 l l l I
F I P L F I P L
EGG RANK EGG RANK

Figure 8. Clutch proﬁle by clutch size (Clsz) and mean egg size (AvVol).

Error bars and regression lines as in Figure 4.

83

Egsz had a signiﬁcant effect on the proﬁle of 4-egg clutches (Table 26A), mainly
due to its inﬂuence on the latter’s linear trend (whose P-value was slightly greater than
the corrected critical level). Clutch proﬁle varied between samples in the direction
predicted by the nutritional hypothesis: a) in low-AvVol nests (Figure 8A), there was a
weak overall decline in egg size with laying order, while b) high-AvVol clutches (Figure
8B) showed an overall increase in egg size with Rank, and a higher relative last-egg size.
The quadratic effect of Egsz on the proﬁle of 5-egg clutches appears to be statistically
signiﬁcant, but is not protected by the signiﬁcance of the overall Egsz-effect (Table 26B).
In agreement with this, the differences between the respective samples’ proﬁles appear to
be very small (Figures 8C and 8D).

The absence of the expected Egsz-effect on the proﬁle of C5 nests could be due to
a smaller difference in mean AvVol between the levels of this factor than in C4 clutches.
However, this difference was actually greater in S-egg nests (17.06 - 15.17 = 1.89 cm3)
than in C4 nests (16.95 -15.48 = 1.47 cm3), and the trend is statistically signiﬁcant, as the
Clsz X Egsz interaction in the among-clutches part of the RMA of Table 25 indicated.
Alternatively, the difference in the response of clutch proﬁle to Egsz between the 2 clutch
sizes (which also differ in H-Span) can be interpreted as evidence that both AvVol and
H-Span can inﬂuence clutch proﬁle in an interactive way. It seems that last eggs can
beneﬁt from extra resources (as the nutritional hypothesis predicts) only in C4, where
H-Span is short, and hormonal constraints on ﬁnal-egg size are presumably weak. In C5,
the stronger hormonal constraints (inferred from the longer H-Span) apparently prevent
additional resources (indicated by higher AvVol) from being invested in ﬁnal eggs.

These results refute the hypothesis that the size of ﬁrst eggs is nutritionally

84

constrained: a) in C5 nests, overall clutch proﬁle, and the relative size of ﬁrst eggs, were
not affected by Egsz (AvVol), while b) in C4, ﬁrst eggs were actually relatively smaller

in high-AvVol clutches (where more resources were presumably available for eggs).

Table 26. Separate RMAs on egg volume by Egsz (among-clutches factor),
and by Rank (within-clutches factor), in each clutch size (Clsz).

Bonferroni-corrected critical P-level: 0.0125.

 

S 0 u r c e F-ratio D.F. P-value

 

A. Clsz = 4

 

Rank X Egsz 4.168 3, 252 0.010
Linear 6.285 1, 84 0.014
Quadratic 2.312 1, 84 0.132

 

B. Clsz = 5

 

Rank X Egsz 2.757 3, 384 0.062
Linear 0.040 1, 128 0.842
Quadratic 10.458 1, 128 0.002

 

Table 27. Separate RMAs on egg volume by Clsz (among-clutches factor),
and by Rank (within-clutches factor), in each level of factor Egsz.

Bonferroni-corrected critical P-level: 0.0125.

 

S o u r c e F -rati0 D.F. P-value

 

A. Low Egsz (AvVol)

 

Rank X Clsz 0.501 3, 318 0.627

 

B. High Egsz(AvV01)

 

 

Rank X Clsz 8.228 3, 318 < 0.001
Linear 9.788 1, 106 0.002
Quadratic 6.826 1, 106 0.010

85

When AvVol is low, proﬁle does not vary between clutch sizes (Figures 8A and
8C; Table 27A), despite their difference in H-Span, and against the hormonal hypothesis’
prediction. In high-AvVol nests, however, Clsz had a signiﬁcant overall and linear effect
on clutch proﬁle (Table 278), in the direction predicted by the hormonal hypothesis. In
C4 nests (Figure 8B) egg size increased overall with laying order, while in C5 the linear
trend was negative, and last eggs were relatively smaller (Figure 8D). The 2 proﬁles also
differ in their quadratic components (Table 27B), but this is only due to the greater drop
from the penultimate to the last egg in C5 (which is also responsible for the linear Clsz-
effect). The relative sizes of the ﬁrst 3 eggs are similar in both clutch sizes (Figures 8B
and 8D; n.s. Rank X Clsz interaction in RMA using only these 3 ranks: F = 0.220, d.f.=
1,212, P = 0.803). Thus it seems that last eggs a) can beneﬁt in the smaller clutches from
weaker hormonal constraints on their size (inferred from their shorter H-Span), as the
hormonal hypothesis predicts, only when b) their size is not nutritionally constrained
(only in high-AvVol clutches).

These results conﬁrm the implication of the Rank X Egsz X Clsz interaction in
Table 25: proﬁle’s response to Egsz does differ between clutch sizes, and its response to
Clsz does depend on the level of AvVol. Therefore, the splitting of a single RMA with 2
among-nests factors into the 4 single-factor analyses presented above was appropriate.

The partial conﬁrmation of the nutritional and hormonal hypotheses’ predictions
indicates that both AvVol and H-Span (the proﬁle-relevant factor that varies between
clutch sizes) may affect clutch proﬁle in an interactive way. Moreover, the speciﬁc cases
where each factor’s effect on proﬁle did not meet the respective hypothesis’ prediction

can be attributed to interference from the other proﬁle-relevant factor.

86

These results also reinforce my earlier statement (in Predictions Tested) that the
nutritional and hormonal hypotheses concerning clutch proﬁle need not be mutually
exclusive. 1 have shown here that both factors postulated by these hypotheses to affect

clutch proﬁle can actually inﬂuence it, and that their effects are not independent.

C. CLUTCH PROFILE TYPES

The results I have presented so far indicate that, on the average, the Squacco
Heron’s clutch proﬁle is arched, as middle eggs differ little among themselves, but are
signiﬁcantly larger than marginal ones (ﬁrst and last). However, the proﬁles of individual
clutches vary, and most of them can be classiﬁed in the following 4 types, which are also
illustrated in Figure 9 (in which 4- and 5-egg clutches are pooled):
a) Decline (Dcln), where egg size decreases from the ﬁrst egg to the last (panel C);
b) I-max, where the proﬁle is arched and the intermediate egg is the largest (panel A);
c) P-max, where the proﬁle is arched and the penultimate egg is the largest (panel B);
(1) Rise, where egg size increases from the ﬁrst egg to the last (panel D).

Since the clutches making up each of these groups were speciﬁcally selected to ﬁt
a certain proﬁle type, there was no reason to test the statistical signiﬁcance of the pattern
of egg-size variability in each sample. The average difference between the smallest and
largest eggs varies from 8% of AvVol in I-max to 12% of AvVol in Decline.

There were also 22 other clutches (10% of total sample), whose proﬁle did not ﬁt
the above patterns, and which I designated as Irregular. Approximately half of them
showed erratic ﬂuctuations in egg size, while differences among egg ranks in the rest

were very small and statistically non-signiﬁcant (overlap of conﬁdence intervals set on

87

% Deviation from AV.VOL

% Deviation from AV.VOL

1
N

.L

1
0')

A. l-max (N=83) B. P-max (N=67)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I I I I 6 I I I I
8 4
> -
>
_ <3: 2.. _
E
e
.. ... 0.. _I
C
.9.
4 .15. ~2~ -
>
(D
C)
.. .1 32 .4._ -1
1 1 1 1 -6 1 1 1 1
F l P L F I P L
EGG RANK EGG RANK
C. Decline (N=24) D. Rise (N=20)
I F I I 6 I I I I
_ _ _, _ _
o 4
— _ >.
2 2- -
E -2— .
” ‘ E
> _ _
.3 (D
“ D
I— -4 °\° ’- _’
1 1 1 -8 1 1 1 1
F I P L F I P L
EGG RANK EGG RANK

Figure 9. Clutch proﬁle types. Error bars represent 95% conﬁdence intervals.

88

egg-size estimates based on the regression of egg volume on linear egg measurements
presented in the Preliminary Analyses). The average proﬁle of Irregular clutches shows
very low variability in egg size: last eggs are smaller than the previous ones (whose sizes
are very similar) by less than 1.5 % of AvVol (average proﬁle not shown in Figure 9).
Table 28 shows the frequencies of the different proﬁle types in each clutch size,
and in the pooled sample. In both C4 and C5, the 2 arched proﬁles (I-max and P-max)
are the only common ones, and together they make up almost 70% of the pooled sample.
Each of the other 3 proﬁle types represents approximately 10% of all clutches. The
difference in relative frequencies between clutch sizes did not seem to be great, and it
only approached statistical signiﬁcance (P = 8.167, d.f.= 4, P = 0.086). Apparently, the
most striking difference involves Rise, which is overall uncommon, but particularly so in
C5, where its percentage is only about 1/3 as among 4-egg clutches. This is consistent
with the hormonal hypothesis, as the stronger hormonal constraints on the size of last
eggs in 5-egg clutches (inferred from their longer H-Span) should make this rare proﬁle

type even less common there than among C4 nests.

Table 28. Numbers and percentages of clutches with different proﬁle types.

 

Clutch Decline I-max P-max Rise Irregular
Size N # % # % # % # % # %

 

C4 86 9 10.5 27 31.5 26 30.0 13 15.0 11 13.0
C5 130 15 11.5 56 43.0 41 31.5 7 5.5 11 8.5

Both 216 24 11.0 83 38.5 67 31.0 20 9.5 22 10.0

 

89

The 4 "regular" proﬁle types, in the order in which I presented them above
(Decline, I-max, P-max, Rise), form a gradient along which the rank of the largest egg
shifts to a progressively later position in the laying order, and the relative size of the last
egg increases (Figure 9). According to the hormonal hypothesis, these phenomena mainly
depend on the timing of the onset of female incubation, which I infer from batch span
values. Therefore, this hypothesis would predict a gradual shortening in H-Span (negative
linear trend), along this proﬁle gradient. The nutritional hypothesis postulates that the
relative size of last eggs is positively associated with female resource availability for egg
formation, which I infer from clutch AvVol. Therefore, this hypothesis would predict an
increase in AvVol (positive linear trend) along this proﬁle gradient.

As I explained above, nests with Irregular proﬁles are a heterogeneous group,
comprising clutches whose proﬁle did not ﬁt any of the 4 other types. Thus I was unable
to derive predictions from the 2 hypotheses about the values of H-Span and AvVol in this
group relative to the other proﬁle types. Nevertheless, I included Irregular clutches in a
post hoc analysis of AvVol (see below). I was unable to do so in the analysis of H-Span,
due to the lack of adequate hatching data from such clutches.

The above tests of the 2 hypotheses involve the reversal of the positions of the
dependent variable (clutch proﬁle) and certain independent factors (H-Span and AvVol)
of last section’s RMAs. The beneﬁt of this approach is that the new independent factor,
clutch proﬁle type, has more levels (4), across which I can test the predicted linear trends
in H-Span and AvVol. These corroborative analyses are not merely redundant, especially
in the case of AvVol, where they provide additional insight to its proximate effects on

clutch proﬁle (see below).

90

Figure 10 shows the trends in median H-Span among proﬁle types in 4- and 5-egg

clutches. As the hormonal hypothesis predicted, there is an overall decrease in H-Span

along the proﬁle gradient in both cases. Separate Kruskal-Wallis tests in each clutch size

indicated that the observed trend approached statistical signiﬁcance in both samples (C4:

H= 6.703, d.f.= 3, P = 0.082 ; C5: H= 7.134, d.f.= 3, P = 0.068). I used non-parametric

tests, because, as I mentioned earlier, H-Span is not normally distributed. I did not pool

4- and 5-egg clutches together, despite the fact that the same trend was predicted in both,

because differences in the relative frequencies of Decline and Rise would bias results in a

way favorable to the hormonal hypothesis’ prediction. Decline, whose median H-Span

was expected to be the longest, is more common in C5, where H-Span is generally longer

than in C4. The opposite is true for Rise, whose H-Span was expected to be the shortest.

Hatch Span (days)
I: or

0)

Figure 10. Median H-Span by proﬁle type and clutch size.
Sample sizes are indicated next to data points.

 

 

 

 

I I I I
9
3O
26 5
3
13
11
6
l l l l
Dcln l-max P-max Rise

Clutch Proﬁle Type

9]

Clutch Size
0 CLSZ = 4
I CLSZ = 5

The variation in median H-Span among proﬁle types shown in Figure 10 appears
substantial in both clutch sizes, and especially so among C4 nests, where the difference in
median H-Span between Decline and Rise is 1.5 d, or 50% of the overall median (in C5
this difference is 1.0 (1). Therefore, the apparent lack of statistical signiﬁcance in these
trends may in good part be due to the expected low power of the respective tests. The
reasons for the latter are: a) the small size of the Decline and Rise samples (see Figure 9),
and b) the high random error in H-Span due to low data accuracy, the male’s contribution
in incubation, and the possible variability in egg-laying intervals (details can be found in
the Preliminary Analyses section on hatching intervals). The latter variability is not likely
to have biased results, because laying intervals did not differ signiﬁcantly between proﬁle
types that were expected to have a long H-Span (Decline and I-max) and those predicted
to have a short H-Span (P-max and Rise), as shown in Table 29. (The pooling of these
groups was necessitated by the small samples of Decline and Rise. Laying intervals were
inversely estimated from the number of eggs laid during 4-day intervals between nest

visits, as in the respective section of the Preliminary Analyses.)

Table 29. Frequencies of cases where 2 or 3 eggs were laid in 4 d for
pairs of proﬁle types, and result of statistical comparison.

 

Dcln + I-max P-max + Rise

# °/o # °/o

 

2 eggs in 4 d 77 86.5 72 80.0
3 eggs in 4 d 12 13.5 18 20.0
T 0 t a l 89 100 90 100

 

 

 

2-tailed Fisher exact P = 0.317 n.s.

 

 

 

92

In 5-egg clutches, where H-Span increases seasonally (see earlier), results could
be confounded by a strong variation in laying date among samples. Differences in median
standardized laying date, however, were not statistically signiﬁcant (Kruskal-Wallis H =
0.085, P = 0.994), and of a magnitude (the largest such difference was 3.0 d) that is not
expected to affect H-Span (see section 6 in Preliminary Analyses).

Results, however, may be biased against the predictions of the hormonal hypo-
thesis by a positive association between egg size and the length of the incubation period
(Parsons 1972; Bollinger 1994). Such a relationship has not been demonstrated in herons,
but, if present, it will have the following consequences. In Decline, the relatively large
ﬁrst eggs will take longer than average to hatch, while the small last eggs will hatch
relatively quickly. This will shorten the H-Span of clutches expected to have the longest
H-Span. The opposite is true in Rise, the proﬁle type expected to have the shortest hatch
span, where the small size of the ﬁrst eggs and the large size of the last eggs will tend to
lengthen it. The hatch spans of I-max and P-max clutches should not be affected, as both
ﬁrst and last eggs are similarly small in these groups.

In summary, the observed trend in H-Span among proﬁle types agrees with the
hormonal hypothesis’ prediction, despite the possible bias against the expected results
described above. This trend appears to be strong in both clutch sizes, and its lack of
statistical signiﬁcance may be attributed to low power in the respective tests (see earlier).
Overall, these ﬁndings corroborate the results of last section’s analyses of clutch proﬁle
(RMAs), and further support the association between H-Span and clutch proﬁle predicted

by the hormonal hypothesis.

93

 

 

 

 

18 I I I I I
E
o _ _
g 17 .
B 83 67
>
i, 15_ 20 _
2 24 22
O
8
c3 15— -
o
>
<

14 l l l l l

Dcln I-max P-max Rise Irreg
Clutch Proﬁle Type

Figure 11. AvVol by proﬁle type (means with 95% conﬁdence intervals).

Sample sizes are indicated next to data points.

The variability in AvVol among the 4 regular proﬁle types did not follow the
pattern that was predicted by the nutritional hypothesis (Figure 11; data points for these 4
groups are connected with solid lines; I will refer to Irregular clutches later). Instead of a
linear increase in AvVol along this proﬁle gradient, the predominant (and statistically
signiﬁcant) trend was quadratic (Table 30), as mean AvVol in I-max and P-max was
greater than in clutches with linear proﬁles (Figure 11). The non-signiﬁcant interaction
between Clsz and proﬁle type (Table 30) indicates that the pattern of variation in AvVol
among proﬁle types does not differ between clutch sizes.

Figure 11 indicates that Irregular clutches also have a relatively low mean AvVol.
A l-way ANOVA on AvVol (pooling clutches of both sizes) by proﬁle type including
Irregular nests (among all 5 proﬁle groups) yielded signiﬁcant results (F = 5.943, d.f.=

4,211, P < 0.001). Tukey’s pairwise-comparison probabilities conﬁrmed that the mean

94

AvVol of Irregular clutches is similar to those of Decline (P = 0.998) and Rise (P =
0.946), but signiﬁcantly smaller than those of I-max (P = 0.032) and P-max (P = 0.012).
In summary, AvVol, which I use as an index of female resource availability for
egg production, seems to be higher among clutches with the common, arched proﬁles,
than in the 3 other groups, which are relatively rare. This suggests the possibility that
Decline, Rise, and Irregular proﬁles represent deviations from the normal, arched pattern,
produced by the combination of hormonal factors and nutritional constraints on the size
of various, but mainly middle, eggs. For instance, the delayed onset of incubation
(inferred from short H-Span) may be necessary, but not always sufﬁcient to produce an
increase in egg size with laying order until the last egg. The size of the usually large
middle eggs may also need to be constrained by resource limitations. ( I will discuss

these issues in greater detail in the following section.)

Table 30. ANOVA on AvVol by clutch size and proﬁle type (excluding Irregular).

 

 

SOURCE Sum Sq. d.f. Mean Sq. F -ratio P-value
C l s 2 1.395 1 1.395 1.220 0.271
P r 0 file 18.292 3 6.097 5.332 0.002
Linear 0.728 1 0.728 0.636 0.426
Quadratic 16.826 1 16.826 14.715 < 0.001
Cubic 0.002 1 0.002 0.002 0.968
Clsz X Prof 0.511 3 0.170 0.149 0.930
Error 212.684 186 1.143

 

95

4. DISCUSSION

In the Squacco Heron, the average egg-size proﬁle of both 4-egg (C4) and 5-egg
(C5) clutches is arched (Figure 4), as marginal eggs (ﬁrst and last) are signiﬁcantly
smaller than middle ones (Table 21). In these average proﬁles, the greatest difference
between egg ranks (laying-order positions) is 4.3% of AvVol in C4 (between A and B),
and 6% of AvVol in C5 (between C and E). Arched proﬁles (Figures 9 A-B) are also the
commonest, both in C4 (>60% of cases) and in C5 (almost 75% of cases). However, egg
size sometimes increases or decreases with laying order (Figures 9 C-D), or it may vary
irregularly. Each of the latter 3 proﬁle types represents about 10% of the total sample
(pooling C4 and C5; Table 28). Due to this variation in the ranks of smallest and largest
eggs, when the maximum egg-size difference is calculated within clutches, and then
averaged, its mean value is about 9% of AvVol in C4 and 11% of AvVol in C5.

The above patterns are not unique among ardeids. Data I collected from a few
Little Egret nests at the Axios river delta, suggest that this species’ average proﬁle is
similar to the Squacco Heron’s, both in pattern (though ﬁrst eggs appear relatively
bigger) and in the level of variation among egg ranks (Figure 12). In the Purple Heron
in Spain, the mean proﬁle of C4 and C5 clutches was arched in one year, but showed
a signiﬁcant negative linear trend in another (Jover et al.1993). The largest difference
among egg-rank means appeared greater than in A.ralloides in C5 (9-12% of AvVol by
year), but this may be partly due to a lack of signiﬁcant variation in proﬁle type among
clutches. In 3 North American species, who lay only 3 or 2 eggs, egg size decreased with
laying order, on the average, but proﬁle type also varied among clutches, especially 3-egg

ones (Custer and Frederick 1990).

96

A. Clsz = 4 (N=7) B. Clsz = 5 (N=12)

 

 

 

 

 

 

 

 

 

 

 

6 I I I I 6 I I I I I
54- i 55 ~
3 2- A 2 2i— _
E 0 - E 0- -
a i a
5'2“ 4 5'2” ‘
E :6
£41- - 634- ..
32 -6- - 32-61- -

-8 1 1 1 1 -8 1 1 r r 1

A B C D A B C D E
EGG RANK EGG RANK

Figure 12. Average egg-size proﬁles of 4- and 5-egg Little Egret clutches.
Error bars represent 95% conﬁdence intervals.

Arched proﬁles, where ﬁrst and ﬁnal eggs are signiﬁcantly smaller than middle
ones, have been observed in birds from taxonomically and ecologically diverse families,
and in clutches of widely varying sizes (Table 31). In most species, both ﬁrst and last
eggs were similarly small, but there are also cases with an overall increase or decline in
egg size with laying order (but where last and ﬁrst eggs are still smaller than middle
ones). Maximum differences among egg-rank means (Max %D in Table 31) appear to
vary moderately around my values from Squacco Heron clutches. The lowest (1.7% of
AvVol), but still statistically signiﬁcant value has been observed in the Kentish Plover
C haradrius alexandrinus (Amat et al. 2001), and the highest in the Purple Heron (12%
of AvVol in one year, 9% in another; Jover et al.1993).

On the other hand, clutch proﬁle type shows strong variations within taxonomic

groups (from Orders to species). Passerine proﬁles can range from Rise (Howe 1976) to

97

Table 31. Avian species with arched clutch proﬁles (marginal eggs smaller than middle).

 

Family and Species Proﬁle “ Clsz Max %D b Source
Phalacrocoracidae Stokland and
Shag Phalacrocorax aristotelis A 3 3.0 Amundsen 1988
Ardeidae
Purple Heron Ardea purpurea A C 4,5 5.0 - 12 Jover et al.1993
Little Egret Egretta garzetta D 4.5 3.4 - 5.9 This study
Squacco Heron Ardeola ralloides A 4,5 4.3 - 6.1 This study
Anatidae
Canada Goose Branta canadensis A 4-7 3.5 - 6.5 Leblanc 1987
Barnacle Goose Branta leucopsis D 3,4 5.5 - 8.0 ' Williams et al.1996
Wood Duck Aix sponsa A 6-1 1 5.0 - 6.0 Kennamer et al.1997
Rallidae
American Coot F ulica americana A 7-13 4.0 - 8.0 Arnold 1991
Haematopodidae
American Oystercatcher R 3 3.6 Nol et al.1984
Haematopus palliatus
Charadriidae
Kentish Plover A 3 1.7 Amat et al. 2001

C haradrius alexandrinus

Sturnidae
European Starling Sturnus vulgaris A 5,6 3.0 - 6.0 Ojanen et al.1981

Turdidae
European Blackbird Turdus merula R 3-5 2.5 - 4.5 Magrath 1992 a

Corvidae
Northwestern Crow Corvus caurinus D 4,5 6.0 - 8.0 ' Verbeek 1990

 

a A: True Arch (ﬁrst and last eggs are of similar, and small, size).
D: Arched Decline (ﬁrst eggs smaller than intermediate, but greater than last).
R: Arched Rise (increase in egg size from ﬁrst to penultimate eggs; last < penultimate).

b Difference between egg-ranks of highest and least mean size, as % of clutch average.
In all cases, intraspeciﬁc variation is among clutches of different sizes.

c In 1 year only. In another year, only the proﬁle’s negative linear trend was signiﬁcant.

98

Decline (Rofstad and Sandvik 1985), through a number of intermediate patterns (Ojanen
et al.1981), and may even vary within populations (Mead and Morton 1985). Proﬁle type
also varies among shorebird (Charadrii) species (Vaisanen et al.1972), and within the
genus F ulica: it is arched in F. americana (Arnold 1991), but egg size increases with rank
in the European Coot F. atra (Horsfall 1984). In gulls (genus Larus), egg size usually
decreases with laying order (Parsons 1976; Reid 1987; Sydeman and Emslie 1992), but
clutch proﬁles can sometimes appear arched (Meathrel et al.1987), or ﬂat (show little egg
size variation), which may be the result of abundant food (Pierotti and Bellrose 1986).

This variability in proﬁle type, especially within populations (Mead and Morton
1985; Custer and Frederick 1990; this study) suggests that clutch proﬁle is amenable to
the inﬂuence of proximate factors, which I examined in this study. Whether this also
reﬂects adaptive adjustments of reproductive strategy to environmental conditions and/or
parental attributes (Slagsvold et al.1984), is beyond the scope of my thesis. Nevertheless,
the level of egg-size variation within Squacco Heron clutches may potentially affect
sibling ﬁtness (see below), and thus has additional ecological interest, whether it may be
an adaptation for or a constraint on parental reproduction.

Data I collected from unfed hatchlings, indicate that their weight increases in
direct proportion to egg volume, but their bill+head length increases less than proportion-
ately (log-log regression slopes were, respectively, similar to and less than 1; Table 32).
Thus egg size seems to inﬂuence Squacco hatchling body condition, possibly by affecting
the amount of remaining yolk (which I could discern below young nestlings’ abdominal
skin, but was unable to quantify), as has also been found in anatids (Ankney 1980; Pelayo

and Clark 2002), quail (Ricklefs et al.1978), shorebirds (Amat et al. 2001), gulls (Parsons

99

1970), and crows (Rofstad and Sandvik 1987). In synchronous broods, egg size can affect
nestling growth (Howe 1976; Smith et al.1995; Anderson and Alisauskas 2002) and
survival up to ﬂedging (Blomqvist et al.1997; Styrsky et a1. 2000; Amat et al. 2001). In
asynchronous broods, where junior siblings are in a competitive disadvantage, egg size
variation was deemed to have no impact on sibling hierarchies in the Shag (Stokland and
Amundsen 1988) and the European Blackbird (Magrath 1992a), where it was low (Table
31), or in the Purple Heron, where it was higher (Jover et al.1993). In the Squacco Heron,
even though a chick’s competitive status may be unaffected by relative egg size, the latter
may still inﬂuence early nestling survival through its effect on yolk reserves and the
chick’s ability to withstand starvation (Parsons 1970,1975a; Howe .1978; Williams 1994).
(The consequences of clutch proﬁle on Squacco sibling ﬁtness are currently under study,

but will be published elsewhere.)

Table 32. Logarithmic regression of hatchling weight (g) and bill+head length (mm) on
egg volume (cm3). P-levels pertain to slope comparisons with values of 0 and 1.

 

 

 

Variable n Slope S.E. Pl 0.0 P/ 1.0 R 2
Weight 46 1.073 0.068 < 0.001 n.s. 0.873
Bill + Head 46 0.215 0.026 < 0.001 < 0.001 0.615

 

 

The main focus of my thesis is the examination of certain proximate factors that
may inﬂuence clutch proﬁle, and especially the relative size of ﬁnal eggs, in A.ralloides.
The causes for small ﬁnal eggs, according to the nutritional and hormonal hypotheses, are
female resource limitations and the early onset of incubation (Parsons 1976). However,

as I was unable to directly measure these variables, I used as indices of their values the

100

average egg volume (AvVol) and the hatching span (H-Span) of clutches. As I explained
in the respective sections of the Preliminary Analyses chapter, the use of these indices is
supported both by the literature and by my own data. Moreover, potential associations
between AvVol or H-Span and other variables may increase the random error in the
relationship between each index and the proﬁle-relevant factor it represents, but are not
expected to produce any spurious associations between these indices and clutch proﬁle.
Thus, since I did ﬁnd that both H-Span and AvVol can affect clutch proﬁle (mostly in
ways predicted by the respective hypotheses), I was able to reach some (at least tentative)
conclusions about the validity of these hypotheses in the Squacco Heron. The lack of a
correlation between AvVol and H-Span (see last section in Preliminary Analyses), also
allowed me to reach such conclusions without experimentally controlling these variables,
as there was no risk of each confounding the other’s effects.

Parsons’s (1976) hormonal hypothesis postulates that the development of female
incubation is associated with an increase in circulating prolactin, which depresses the
function of the oviduct, and thus limits the size of subsequently-laid eggs (only last ones
in the case of gulls, and due to a deﬁciency in albumen). In populations where the timing
of incubation onset varies among females (as in A.ralloides, see section 6 in Preliminary
Analyses), this hypothesis would predict that in clutches with a longer H-Span a) ﬁnal
eggs would be relatively smaller, and b) the decline in egg size might begin at an earlier
point in the laying order (Arnold 1991).

Both predictions were conﬁrmed by a comparison of the egg-size proﬁles of nests
with short (< median) and long (> median) H-Span (Figure 6). The proﬁle’s overall linear

trend changed from positive in the former sample to negative in the latter, where ﬁnal

101

eggs were relatively smaller and the proﬁle’s arch peaked at an earlier egg-rank (differ-
ances are statistically signiﬁcant; see Table 22 and associated text). Hatch span (and the
onset of incubation relative to the laying of the last egg) is much longer in C5 than in C4
nests (4.5 vs 3.0 d ; Table 17). Therefore, the hormonal hypothesis would predict smaller
ﬁnal eggs in the larger clutches. This was conﬁrmed when the Clsz-effect on proﬁle was
tested using high-AvVol nests only (Table 27B): the proﬁle’s linear trend changed from
positive in C4, where last eggs were relatively large (Figure 8B), to negative in C5, where
last eggs were relatively smaller (Figure 8D). However, no Clsz-effect on clutch proﬁle
was detected when only low-AvVOl nests were involved in the comparison (Table 27A;
Figures 8A and 8C). This could be the result of interference'by resource limitations
(inferred from the low AvVol-level) on last-egg size (as predicted by the nutritional
hypothesis): if last eggs are nutritionally constrained to a small size in both C4 and C5
nests, variation in the intensity of the incubation-related hormonal constraints may have
no further effect on their relative size. This is an indication that the ﬁnal eggs in Squacco
Heron clutches may be under the inﬂuence of both nutritional and hormonal factors.

Despite the statistically signiﬁcant seasonal increase in H-Span in C5 (Table 18),
there was no corresponding variation in the proﬁle of 5-egg clutches (Table 238). This
is probably due to the fact that the seasonal trend in H-Span is of insufﬁcient magnitude
to produce a measurable effect on clutch proﬁle (see page 80). It could also stem from
interference by some additional, unknown factor (not AvVol, which showed no seasonal
variation; see Table 14), though no such factor is suggested in the literature. The proﬁle
of C5 clutches showed some annual variation (Figure 7), which was not predicted by

either hypothesis tested (see Table 20). This variation, however, was weak (Figure 7),

102

and only approached signiﬁcance (Table 23B), so it does not refute either hypothesis.

The hormonal hypothesis is also supported by the overall decrease in H-Span
along the proﬁle type gradient from Decline to Rise (Figure 10). Along this gradient, the
largest egg in the clutch is of a progressively later rank, and the relative size of the ﬁnal
egg increases (Figure 9). If these phenomena mostly depend on the timing of the onset of
incubation (as the hormonal hypothesis postulates), then this should occur progressively
later, and H-Span should consequently become shorter, along this series of proﬁle types.
Observations from both C4 and C5 clutches conﬁrmed this prediction (Figure 10), but
results only approached statistical signiﬁcance. The latter, though, may merely stem from
a) low test power (due to small sample sizes and high random error; see section 6 in the
Preliminary Analyses) and b) the fact that a positive relationship between egg size and the
length of incubation period would tend to limit H-Span values in Decline and increase
them in Rise (see page 93). A ﬁnal piece of supporting evidence for the hormonal hypo-
thesis is that the uncommon proﬁle type Rise was particularly rare in C5 nests (Table 28),
where the longer H-Span should make it even less likely for ﬁnal eggs to attain maximum
size within clutches (though the overall variation in proﬁle type frequencies between C4
and C5 nests only approached statistical signiﬁcance).

In summary, the direct test of the hormonal hypothesis (examination of H-Span
inﬂuence on clutch proﬁle) showed a strong and statistically highly signiﬁcant effect on
clutch proﬁle in the predicted direction, and most secondary tests corroborated its results.
In the 2 exceptions, the lack of the expected seasonal trend in proﬁle is probably due to
insufﬁcient variation in H-Span, while the lack of the predicted Clsz-effect in low-AvVol

nests can be attributed to interfering nutritional constraints. Thus results suggest that the

103

onset of female incubation (inferred from H-Span values) can affect the point in the
laying sequence at which egg size begins to decline, and especially the relative size of
ﬁnal eggs in Squacco Heron clutches.

A direct association between the onset of female incubation. and the relative size
of last eggs was demonstrated by Parsons (1976) in Herring gulls. In the American Coot,
where females lay clutches of 7-13 eggs with arched proﬁles (Table 31), the points in the
laying sequence where egg size peaks and incubation begins are closely associated, and
occur progressively later as clutch size increases (Arnold 1991). Moreover, several lines
of evidence suggest that nutritional limitations are not responsible for the decline in egg
size after the onset of incubation in these coots (Arnold 1991), or in some geese (Leblanc
1987; Williams et al.1996), where authors have attributed this aspect of clutch proﬁle to
the hormonal constraints proposed by Parsons (1976).

Given the small clutch size of gulls (3 eggs) and the timing of relevant events,
3rd eggs are smaller than earlier ones, and this is because they lack albumen (Parsons
1976; Houston et al.1983). This deﬁciency was not alleviated by food supplements in
Lesser Black-backed Gulls (though yolk protein did increase in all eggs, and especially in
last ones; Bolton et al.1992), and thus seems to result from hormonal constraints on the
function of the oviduct. Additional support for a depression of oviduct functions comes
from the examination of eggshell-color proﬁles. Eggshell pigments are synthesized and
deposited in the oviduct (in its shell gland area; Burley and Vadehra 1989). Therefore,
factors that can constrain albumen synthesis may also limit eggshell pigmentation. Shell
color proﬁles match those of egg size in the Glaucous-winged Gull Larus glaucescens

(Verbeek 1988) and the Northwestern Crow Corvus caurinus (Verbeek 1990). Moreover,

104

they did not seem to stem from resource limitations in either species (Verbeek 1988,
1990), while initial egg removals in the former suggested that only last eggs have lighter
shells, irrespectively of absolute laying order (Verbeek 1990). In species that lay larger
clutches, though, and where females begin low-intensity incubation early in the laying
sequence, both ovarian and oviductal functions could be hormonally constrained (without
being completely suppressed), so that egg size could decline after the onset of incubation
due to deﬁcits in both yolk and albumen, and with little change in egg composition
(Arnold 1991; Williams et al.1996). Whether this is true in A.ralloides or other herons

is not known, but can be the object of future studies.

 

In my study, I did not examine the detailed physiological pathways through which
female incubation may affect the egg-size proﬁle of clutches. Nevertheless, information
from the literature suggests certain mechanisms. First, the onset of incubation seems to
be closely associated with an increase in prolactin secretion in many avian taxa (Buntin
1996; Vleck 2002), including both domesticated anatids (Hall and Goldsmith 1983), fowl
(Proudman and Opel 1981), doves (Ramsey et al.1985) and canaries (Goldsmith et al.
1984), as well as wild albatrosses (Hector and Goldsmith 1983), kestrels (Sockman et al.
2000), grouse (Etches et al.1979), shorebirds (Oring et al.1988) and passerines (Silverin
and Goldsnrith 1983). Experimental studies have conﬁrmed the association between these
traits in various birds (El Halawani et al.1986; Sharp et al.1989; Y0ungren et al.1991;
March et al.1994), but the precise causal relationships remain uncertain (Goldsmith 1991;
Vleck 2002). Nevertheless, this information supports the ﬁrst step of Parsons’s (1976)
hormonal hypothesis, that is the connection between the onset of incubation and high

levels of circulating prolactin.

105

The second step in Parsons’s (1976) hypothesis concerns the negative effects of
increased prolactin secretion on the female’s reproductive organs. There is evidence
(Buntin et al.1998) that high prolactin levels inhibit the release of the hypothalamic
neurotransmitter Gn-RH that stimulates gonadotropin (FSH and LH) secretion by the
adenohypophysis (Scanes 1986; Etches 1996), and thus reduce the plasma levels of both
FSH (Dawson and Goldsmith 1982; Silverin and Goldsmith 1983) and LH (Lea et a1.
1981; El Halawani et al.1991). Both gonadotropins are necessary for the development
of the ovary and its gametogenic and endocrine functions (Murton and Westwood 1977;
Scanes 1986), while cyclic LH surges induce ovulation (Johnson 1986; Etches 1996).
The prolactin-related drop in FSH and LH levels leads to a decrease in ovarian estradiol
secretion (Zadworny et al.1989; Sockman and Schwabl 1999), and to follicular atresia
(Opel and Proudman 1980). Estradiol stimulates oviduct growth (Murton and Westwood
1977) and albumen synthesis (Johnson 1986), whereas high progesterone levels (as those
released by atretic follicles) elicit oviduct regression (Murton and Westwood 1977).
Thus, the series of hormonal changes set in motion by the onset of incubation and its
concomitant increase in plasma prolactin, results in the suppression of ovarian and
oviductal functions (Buntin and Tesch 1985), which may initially limit egg size
(Williams et al.1996), and eventually terminate egg-laying (Youngren et al.1991).

I tested Parsons’s (1976) hypothesis that the size of last eggs is nutritionally
constrained by comparing the proﬁles of clutches with low and high AvVol (the index
of female resource availability for egg formation). I tested the effect of Egsz (the factor
whose levels represent the variation in AvVol) on proﬁle in each clutch size separately,

because an RMA with both Egsz and Clsz as among-nests factors indicated that this

106

xmﬁ-ﬁ"

1W1Eﬁﬁ iii

effect might differ between C4 and C5 clutches (the Rank X Clsz X Egsz interaction was
marginally signiﬁcant; see Table 25). In C4 nests, where H-Span is. shorter (and hormonal
constraints on last-egg size are expected to be weaker), Egsz affected clutch proﬁle in the
way predicted by the nutritional hypothesis: a) in low-AvVol clutches (Figure 8A) the
proﬁle’s linear trend was weakly negative and last eggs had the lowest mean size, while
b) in high-AvVol nests (Figure 8B) the overall linear trend was positive and last eggs
were relatively bigger, and c) the difference in clutch proﬁle between the 2 samples was
statistically signiﬁcant (Table 26A). In C5 nests, however, where H-Span is longer, and
hormonal constraints on ﬁnal-egg size are expected to be stronger, the effect of Egsz on
clutch proﬁle appeared to be very weak (Figures 8C and 8D) and was not statistically
signiﬁcant (Table 268). Therefore, female resource availability (as-inferred from AvVol)
seems to be able to affect the size of last eggs only when the latter is not strongly limited
by hormonal constraints associated with the onset of incubation. These results, together
with the fact that the Clsz-effect on proﬁle depends on the level of AvVol (see earlier),
suggest that the egg-size proﬁle of Squacco Heron clutches can be inﬂuenced both by
hormonal and nutritional factors in an interactive way: each factor’s effect becomes
apparent only when the other factor’s impact is weak.

According to the nutritional hypothesis, no annual or seasonal variation in clutch
proﬁle was expected (Table 208), as these factors did not affect AvVol (Table 14), and
none was found (Table 23). The trend in AvVol along the Decline-to-Rise proﬁle-type
gradient did not match this hypothesis’ prediction of a linear increase, as it was quadratic
in shape (Figure l 1), and signiﬁcantly so, statistically (Table 30). Clutches with Irregular

proﬁles also appeared to have a low mean AvVol, which was similar to those of Decline

107

and Rise, but signiﬁcantly smaller than those of the arched proﬁles (I-max and P-max), as
Tukey’s probabilities indicated (see page 95). Thus AvVol tends to be low in all clutches
whose proﬁle type deviates from the common, arched types. In Decline, a low AvVol
was expected by the nutritional hypothesis, but could also be imposed on clutches of
well-provisioned females who begin to incubate very early in the laying sequence, and
are thus prevented from investing their abundant resources in their eggs. The negative
effect of H—Span on relative ﬁnal-egg size, together with the low AvVol of clutches with
Rise-type proﬁles, suggest that a delayed onset of female incubation may be necessary,
but not sufﬁcient for last eggs to attain maximum size in their clutches: the size of the
(normally large) middle eggs may also need to be constrained. This joint requirement
may help explain why Rise is so rare (15% in C4, 5.5% in C5), even though last eggs are
similar to or greater than AvVol in 45% of C4 and 22% of C5 nests. The existence of
clutches where egg size ﬂuctuates strongly and irregularly (zig-zag proﬁles) indicates that
acute (possibly nutritional) constraints may occasionally affect single eggs in the middle
of the laying sequence. Such patterns are more likely to be produced by limitations of
albumen synthesis and/or deposition, as this occurs separately in each egg (as it passes
through the oviduct), whereas yolk is deposited simultaneously in multiple eggs (e.g. in
cormorants, Grau 1996; anatids, Alisauskas and Ankney 1994; kestrels, Meijer et al.1989;
coots, Alisauskas and Ankney 1985; larids, Moore et al. 2000; passerines, Krementz and
Ankney 1985). However, since I did not analyze the composition of Squacco Heron eggs,
this possibility remains unveriﬁed.

Parsons (1976) rejected the hypothesis that the size of normal last (3rd) eggs of

Herring Gulls was nutritionally constrained, because 3rd and later eggs in clutches where

108

early eggs were removed (and the onset of incubation was delayed) were not signiﬁcantly
smaller than lst and 2nd ones, as long as they were not last. Experiments on other gulls,
however, showed that all, and especially last eggs can beneﬁt from‘food supplements, but
that this is due to an increase in yolk (rather than albumen) protein (Bolton et al.1992).
The decline in egg size between intermediate and ﬁnal eggs seen in some coots (Arnold
1991) and geese (Leblanc 1987; Williams et al.1996) cannot be explained by resource
limitations, but nutrient availability can affect the relative size of last eggs in the Euro-
pean Blackbird (Magrath 1992b). Species whose ﬁnal-egg size is particularly susceptible
to nutritional limitations, may have evolved Rise-type proﬁles, because these reduce the
risk of last eggs falling below a minimum functional size (Magrath 1992b).

The size of ﬁrst eggs, and the factors that may inﬂuence it, have received a lot less
attention. One hypothesis, proposed for small, insectivorous passerines, claims that ﬁrst-
egg size may be nutritionally constrained, because females start forming eggs in a season
when ambient temperature and food availability are low, but tend to improve signiﬁcantly
from day to day (Slagsvold and Lifjeld 1989; Nilsson and Svensson 1993a). This hypo-
thesis does not seem to apply to the Squacco Heron, as relative ﬁrst-egg size a) was not
affected by AvVol (the index of female resources for eggs) in C5 clutches (Figures 8C
and 8D; Table 26B), while b) it was negatively affected by AvVol in C4 nests (Figures
8A and 88; Table 26A), which is the opposite of this hypothesis’ prediction. Another
hypothesis, proposed for anatids, postulates that the initial inelasticity of the oviduct
limits ﬁrst-egg width, but poses no such constraints on later eggs, as the passage of the
ﬁrst egg expands this organ (Robertson and Cooke 1993). This anatomical constraint on

the width of ﬁrst eggs may also limit their volume, because birds may have evolved a

109

mechanism that prevents the extreme elongation of eggs (which could be effected by the
deposition of extra albumen along the egg’s longitudinal axis), in order to avoid abnormal
embryonic development and low hatching success (Robertson et al.1994). This hypo-
thesis was also refuted, both by the very low variation in mean egg shape (width/length
ratio) among ranks (Figure 5), and by the lack of a positive correlation between ﬁrst-egg
shape and volume (see page 74). Finally, some authors have suggested that ﬁrst-egg size r
may be limited by an initially low efﬁciency of the ovary (Parsons 1976) and the oviduct
(Leblanc 1987; Williams et al.1993), which may improve as egg production progresses.

However, no speciﬁc mechanism or relevant predictions have been proposed for this

 

physiological "inertia", so I was unable to examine this hypothesis critically. Therefore,
in summary, the relative size of ﬁrst eggs in Squacco Heron clutches does not seem to be
either nutritionally or anatomically constrained, but my data could not help me identify
any of the factors that do limit it.

A summary of my results and the conclusions that can be drawn from them can be

found in the following chapter (Summary and Conclusions).

110

V. SUMMARY AND CONCLUSIONS

I studied certain aspects of the breeding biology of the Squacco Heron (Ardeola
ralloides), a poorly known species, and I focused on factors that may proximately affect
this heron’s intraclutch egg-size variation (clutch proﬁle), an issue that has rarely been
rigorously addressed in wild birds, and never before in ardeids.

I found that the mean egg-size proﬁles of both 4-egg (C4) and 5-egg (C5) clutches
were arched, that is ﬁrst and last (marginal) eggs were signiﬁcantly smaller than middle
ones. To investigate the proximate causes of these trends, I tested Parson’s (1976) hypo-
theses of independent nutritional and hormonal constraints on last-egg size. The former
attributes the lower relative size of last eggs to female nutritional resource limitations.
Being unable to assess female resource availability directly, I inferred its level from the
average egg volume of clutches (AvVol), and predicted a positive relationship between
AvVol and relative last-egg size. The latter hypothesis postulates that the high levels of
plasma prolactin that are associated with the onset of female incubation limit the size of
last (or late) eggs through their depressive effects on the function of the oviduct (and also,
possibly, the ovary; Arnold 1991). I was not able to assess female incubation directly, so
I inferred its rate of development during egg-laying from the hatch span (H—Span) of
clutches, and I predicted a negative relationship between H-Span and the relative size of
ﬁnal eggs. I also examined Houston, Jones and Sibly’s (1983) hypothesis that the onset
of incubation restricts female foraging time and food ingestion, and thus creates resource
shortages that limit egg size. This hypotheses makes both of the above predictions, and

also forecasts a negative correlation between AvVol and H-Span.

lll

With respect to ﬁrst eggs, I examined whether their size may be resource-limited
(Slagsvold and Lifjeld 1989), in which case it should be positively associated with AvVol
(N ilsson and Svensson 1993a). This hypothesis was refuted, as relative ﬁrst-egg size was
actually lower in high-AvVol 4-egg clutches, and was not affected by AvVol in C5 nests.
First eggs might also be anatomically constrained by an initial oviduct inelasticity, which
could restrict their width (Robertson and Cooke 1993) and size (Robertson et al.1994).
This hypothesis predicts that ﬁrst eggs should be thinner than later ones, but I did not ﬁnd
them signiﬁcantly so. Variability among females in the intensity of this constraint should
produce a positive correlation between ﬁrst-egg volume and shape (width/length ratio),
but neither this prediction was conﬁrmed. Thus the size of ﬁrst eggs in Squacco Heron
clutches does not seem to be either nutritionally or anatomically constrained, but my data
did not permit any further examination of its proximate control.

Before I tested the hypotheses concerning the relative size of ﬁnal eggs, I did
some preliminary data analyses that provided necessary background information. The
median laying date of clutches occurred progressively later in each of the 3 years of my
study (1992-1994), a trend that seems to be related to annual differences in mean ambient
temperature. Clutch size did not vary substantially among years or within each main
laying season, but was signiﬁcantly lower in delayed nests. Females whose initial eggs
were lost soon after they were laid, apparently replaced them by laying 1 or 2 extra ones.
These females, as well as those laying delayed clutches, could have faced increased
nutritional limitations when producing eggs, and were thus expected to lay clutches of
lower AvVol than females who laid regular 4- and S-egg clutches. This was conﬁrmed,

but the low-AvVol sample of nests was small and the difference between sample means

112

was low (about 5% of the grand mean). Nevertheless, these results support, at least
weakly, the use of AvVol as an index of female resource availability for egg formation.
Besides, extraneous inﬂuences (e. g. that of female body size) may increase the random
error in the relationship between these 2 variables, but are not expected to create any
spurious association between AvVol and clutch proﬁle. AvVol did not vary annually,
seasonally (within the main laying period) or between C4 and C5 nests. Therefore, the
nutritional hypothesis would predict no signiﬁcant variation in clutch proﬁle among the
levels of any of these 3 factors.

Parents can control the hatching patterns of their clutches through incubation in
many birds (e.g. Parsons 1972; Magrath 1992a; Wiebe et al.1998). In my study, male
involvement in this activity, measurement error in H-Span data, and possible variation in
egg-laying intervals, may have increased the random error in the relationship between
female incubation during egg-laying and H-Span. Thus they may have made the latter an
imperfect index of the former, but none of them were expected to create any spurious
association between H-Span and clutch proﬁle. However, a possible correlation between
egg size and the length of required incubation (Parsons 1972; Bollinger 1994), may have
biased results against the predictions of the hormonal hypothesis. Hatch span was
signiﬁcantly longer in C5 (median of 4.5 (1) than in C4 (3.0 d), which is not surprising,
given the extra egg and hatching interval in the larger clutches. The hormonal hypothesis
would therefore predict relatively smaller last eggs in C5 nests. The testing of annual and
seasonal effects on H-Span revealed only a signiﬁcant positive seasonal trend within C5
clutches. The hormonal hypothesis would thus predict a seasonal decline in relative ﬁnal-

egg size in C5, but no other seasonal or annual variation in clutch proﬁle.

113

 

AvVol and H-Span were not correlated in either clutch size. This result refuted
Houston, Jones and Sibly’s (1983) hypothesis of nutritional constraints on ﬁnal-egg size
due to female feeding restrictions imposed by the onset of incubation. It also allowed me
to test these 2 factors’ effects on clutch proﬁle without risk of mutual confounding, even
in the absence of experimental manipulations.

The primary prediction of the hormonal hypothesis was conﬁrmed: between nests
with short and long H-Span (late and early incubation onset), the linear trend of clutch
proﬁle changed from positive to negative, and the relative size of last eggs decreased. The
proﬁle’s arch also peaked earlier in the latter sample, which is also consistent with this
hypothesis. There was a weak annual variation in the proﬁle of 5-egg clutches, which
approached statistical signiﬁcance, but does not constitute strong evidence against either
hypothesis (neither of which predicted it). The expected (by the hormonal hypothesis)
seasonal decrease in relative last-egg size in C5 was not observed. This probably occurred
because the seasonal increase in H-Span, though statistically signiﬁcant, was not of
sufﬁcient magnitude to produce a response in clutch proﬁle. Therefore, it does not offer
strong evidence against the hormonal hypothesis. Since this prediction concerned only a
secondary effect of H-Span, it is also possible that some other, seasonally varying factor
may have opposed H-Span’s inﬂuence on clutch proﬁle, but no such factor is suggested
in the literature.

The hormonal hypothesis’ secondary prediction of relatively smaller ﬁnal eggs in
C5 (than in C4) was also conﬁrmed, but only among high-AvVol nests. When AvVol
was low (and resource limitations were presumably high), there was no difference in

proﬁle between clutch sizes. Therefore, it seems that last eggs in C4 clutches can attain a

114

greater relative size (presumably due to weaker hormonal constraints) only when female
resources are abundant (AvVol is high). When AvVol is low (and nutritional resources
are poor), the alleviation of hormonal constraints has no apparent effect on last-egg size.
Similarly, the nutritional hypothesis’ primary prediction of a positive effect of AvVol on
relative ﬁnal-egg size and the proﬁle’s linear trend, was conﬁrmed only in 4-egg clutches.
Apparently, the longer H-Span (and stronger hormonal constraints) of 5-egg nests did not
allow last eggs in high-AvVol clutches to beneﬁt from the higher resource availability of
the laying females, but kept them as small (relatively) as in low-AvVol nests. These
results suggest that both H-Span and AvVol (and the inferred hormonal and nutritional
factors) can inﬂuence the Squacco Heron’s clutch proﬁle in an interactive way: each
factor’s effects can become apparent only in the absence of strong interference from the
other. Other studies of the proximate causes of avian clutch proﬁle have so far produced
evidence supporting one or the other of the above hypotheses (e.g. Parsons 1976; Arnold
1991). My results are the ﬁrst to indicate that both hormonal and nutritional factors may
interactively affect the relative size of birds’ ﬁnal eggs.

Trends in AvVol and H-Span among the different clutch proﬁle types also offer
some insight to the phenomena under study. H-Span decreased along the Decline-to—Rise
gradient (where the relative size of last eggs progressively increases), as was predicted by
the hormonal hypothesis. The nutritional hypothesis’ prediction of a linear increase in
AvVol along this gradient was not conﬁrmed. AvVol was high only in nests with the
common, arched proﬁles, and signiﬁcantly lower in all others. The ’short H-Span and low
mean AvVol of Rise suggest that a delayed onset of incubation may be necessary, but not

sufﬁcient for last eggs to attain maximum size: the size of the (normally large) middle

115

eggs may also need to be constrained (probably nutritionally). The existence of nests with
Irregular proﬁles also indicates that the size of middle eggs may sometimes be limited
(again, probably nutritionally). Size constraints on single eggs are more likely to result
from albumen deﬁciencies, as this is deposited separately in each egg, whereas several
eggs’ yolks grow simultaneously (e.g. Meijer et al.1989; Grau 1996; Moore et al. 2000),
but this issue cannot be resolved until egg composition is examined.

In conclusion, my study makes the ﬁrst detailed examination of clutch proﬁle in
ardeids, and has indicated certain proximate factors that seem to inﬂuence it (or not), thus
suggesting further investigations of its proximate causes. The size of ﬁrst eggs in heron
clutches does not seem to be either nutritionally or anatomically constrained. Some
authors have hypothesized that their size may be limited by an initially low physiological
efﬁciency of the female reproductive organs (Parsons 1976; Leblanc 1987). No speciﬁc
mechanisms have been proposed, however, and no predictions about the variation in
relative ﬁrst-egg size have been made, so this hypothesis needs theoretical development.

The size of last eggs in Squacco Heron clutches is negatively associated with
their clutches’ hatch span, which reﬂects the early development of parental (and female)
incubation. This relationship may be mediated by a) an association between the onset of
incubation and an increase in prolactin secretion (Buntin 1996; Vleck 2002), and b) the
latter’s involvement in the suppression of the female reproductive organs’ activities
(Buntin and Tesch 1985; Youngren et al.1991). These behavioral and physiological links
can be examined with measurements, and especially manipulation, of female prolactin
levels during egg-laying, with and without removal of initial eggs, whose presence

stimulates incubation (Williams et al.1996). Such experiments, however, may not be

116

tolerated by ardeids, who are sensitive to disturbances at their nesting sites (Tremblay
and Ellison 1979; Frederick and Collopy 1989).

The size of last eggs in Squacco Heron clutches may also be subject to nutritional
constraints, as its positive association with the mean egg size of clutches indicates. The
latter, however, becomes apparent only in 4-egg clutches, where the interval between the
onset of incubation and the laying of last eggs is shorter, and the hormonal constraints on F‘
the size of the latter are presumably weaker. Females acquire most of their resources for

egg production from their daily diet in diverse avian taxa (Meijer and Drent 1999), so this

 

may also be true in ardeids. However, my data indicated that the onset of incubation does
not restrict female feeding opportunities (as proposed by Houston et al.1983), which
agrees with the existing information about heron activity patterns during nesting (see last
section in Preliminary Analyses). The effects of female nutrition on mean egg size and
clutch proﬁle can be examined a) with appropriate food supplementation at times of low
natural resource availability (Hiom et al.1991; Bolton et al.1992), or b) with repeated
early-egg removals, which can increase nutritional demands in females who respond by
laying extra eggs (Nager et al. 2000).

The examination of egg composition (which requires collection and destruction of
eggs) can indicate whether herons’ last eggs mostly lack albumen, as those of gulls
(Parsons 1976; Houston et al.1983), or whether they are deﬁcient in both albumen and
yolk, as those of coots (Arnold 1991) and geese (Williams et al.1996). This information
pertains to the question of whether hormonal changes lower (before suppressing) only the
oviduct’s function, or if they also affect the ovary. Moreover, resource limitations may

sometimes have small effects on last eggs’ size, but signiﬁcantly lower their quality,

117

both in terms of lipid and protein content, and as inferred from hatchling survival and
ﬂedging success (Nager et al. 2000).

Finally, in a separate (and still incomplete) study, I examine the ecological effects
of clutch proﬁle in A.ralloides, that is the possible inﬂuence of egg-size variation on the
growth and survival of siblings who differ in age (due to hatching asynchrony) and in

hierarchical status in their competition for ﬁnite parental care.

118

 

APPENDIX

 

 

 

 

 

Table A-1 RMA on egg volume by Rank (within-clutches factor), and by
Year and Season (among-clutches factors), using 4-egg nests.
Results are used in text Tables 21A and 23A.
Source Sum of qu D.F. Mean Sq F-ratio P-value
Among Clutches
Year 13.29 2 6.65 1.952 0.149
Season 0.05 1 0.05 0.013 0.909
Y X S 16.35 2 8.17 2.400 0.097
Error 272.44 80 3.41
Within Clutches
Rank 28.47 3 9.49 26.697 < 0.001
R X Y 0.81 6 0.14 0.380 0.867
R X S 0.08 3 0.03 0.072 0.961
R X Y X S 3.26 6 0.54 1.528 0.181
Error 85.31 240 0.36
Linear Contrast
Rank 0.75 1 0.75 1.228 0.271
R X Y 0.69 2 0.35 0.565 0.570
R X S 0.00 1 0.00 0.002 0.968
R X Y X S 1.08 2 0.54 0.886 0.416
Error 48.92 80 0.61
Quadratic Contr.
Rank 27.41 1 27.41 97.814 < 0.001
R X Y 0.11 2 0.06 0.195 0.823
R X S 0.00 1 0.00 0.001 0.979
R X Y X S 1.04 2 0.52 1.851 0.164
Error 22.42 80 0.28

 

119

 

Table A~2

RMA on egg volume by Rank (within-clutches factor), and by
Year and Season (among-clutches factors), using 5-egg nests.
Results are used in text Tables 21B and 23B,

 

 

 

 

 

Source Sum of qu D.F. Mean Sq F-ratio P-value
Among Clutches
Year 3.66 2 1.83 0.257 0.773
Season 9.57 1 9.57 1.394 0.248
Y X S 12.48 2 6.24 0.879 0.418
Error 880.00 124 7.10
Within Clutches
Rank 101.92 4 25.48 69.482 < 0.001
R X Y 6.10 8 0.76 2.079 0.060
R X S 1.53 4 0.38 1.046 0.369
R X Y X S 3.37 8 0.42 1.149 0.334
Error 181.89 496 0.37
Linear Contrast
Rank 4.31 1 4.31 5.353 0.022
R X Y 4.14 2 2.07 2.572 0.080
R X S 0.04 1 0.04 0.053 0.818
R X Y X S 1.33 2 0.66 0.825 0.441
Error 99.75 124 0.80
Quadratic Contr.
Rank 94.58 1 94.58 301.526 < 0.001
R X Y 1.09 2 0.54 1.734 0.181
R X S 0.00 1 0.00 0.000 0.993
R X Y X S 1.82 2 0.91 2.898 0.059
Error 38.89 124 0.31

 

120

 

“I

Table A-3

Results are used in text Table 22.

RMA on egg volume by Rank (within-clutches factor),
and by clutch size (Clsz) and H-Span (among-clutches factors).

Clsz included only in order to examine the RXHXC interaction.

 

 

 

 

 

Source Sum of qu D.F. Mean Sq F -ratio P—value
Among Clutches
H-Span 6.46 1 6.46 1.253 0.266
Clsz 10.39 1 10.39 2.015 0.159
H X C 7.68 1 7.68 1.491 0.225
Error 535.94 104 5.15
Within Clutches .
Rank 56.36 3 18.79 47.385 < 0.001
R X H 6.91 3 2.30 5.811 0.002
R X C 5.97 3 1.99 5.019 0.005
R X H X C 0.63 3 0.21 0.530 0.620
Error 123.70 312 0.40
Linear Contrast
Rank 0.00 1 0.00 0.000 0.990
R X H 6.73 1 6.73 9.673 0.002
R X C 3.28 1 3.28 4.713 0.032
R X H X C 0.36 1 0.36 0.511 0.476
Error 72.32 104 0.70
Quadratic C ontr.
Rank 56.36 1 56.36 173.013 < 0.001
R X H 0.01 1 0.01 0.032 0.858
R X C 2.40 1 2.40 7.369 0.008
R X H X C 0.27 1 0.27 0.835 0.363
Error 33.88 104 0.33

 

121

-- Jurﬁﬁ: iiﬁ

-.‘..M

 

Table A-4 RMA on egg volume by Rank (within-clutches factor), and by
Egsz and Clsz (among-clutches factors). Contrasts were not
performed, because RXHXC interaction approached signiﬁcance.
Results are used in text Table 25.

 

 

 

 

Source Sum of qu D.F. Mean Sq F-ratio P—value
Among Clutches
Egsz 586.78 1 586.78 329.256 < 0.001
Clsz 1.89 1 1.89 1.058 0.305
E X C 8.95 1 8.95 0.024 0.026
Error 377.82 212 1.78
Within Clutches
Rank 101.78 3 33.93 89.695 < 0.001
R X E 4.53 3 1.51 3.989 0.015
R X C 6.82 3 2.27 6.011 0.002
R X E X C 3.07 3 1.02 2.702 0.061
Error 240.55 636 0.38

 

122

Table A-5

RMA on egg volume by Rank (within-clutches factor),

and by Egsz (among-clutches factors), in 4-egg clutches.

Bonferroni-corrected critical P-level: 0.0125.
Results are used in text Table 26A.

 

 

 

 

 

Source Sum of qu D.F. Mean Sq F -ratio P-value

Among Clutches
Egsz 188.15 1 188.15 135.15 <0.001
Error 116.94 84 1.39

Within Clutches
Rank 31.66 3 10.55 31.144 <0.001
R X E 4.24 3 1.41 4.168 0.010
Error 85.39 252 0.34

Linear Contrast
Rank 0.55 1 0.55 0.972 0.327
R X E 3.54 l 3.54 6.285 0.014
Error 47.27 84 0.56

Quadratic C ontr.
Rank 30.52 1 30.52 111.163 < 0.001
R X E 0.64 1 0.64 2.312 0.132
Error 23.06 84 0.28

 

123

 

Table A-6 RMA on egg volume by Rank (within-clutches factor),
and by Egsz (among-clutches factors), in 5-egg clutches.
Bonferroni-corrected critical P-level: 0.0125.

Results are used in text Table 26B.

 

 

 

 

 

Source Sum of qu D.F. Mean Sq F-ratio P-value

Among Clutches
Egsz 465.09 1 465.09 229.00 < 0.001
Error 259.96 128 2.03 '

Within Clutches
Rank 88.91 3 29.64 73.412 < 0.001
R X E 3.34 3 1.11 2.757 0.062
Error 155.03 384 0.40

Linear Contrast
Rank 5.89 1 5.89 7.702 0.006
R X E 0.03 1 0.03 0.040 0.842
Error 97.86 128 0.76

Quadratic C ontr.
Rank 83.02 1 83.02 284.189 < 0.001
R X E 3.06 1 3.06 10.458 0.002
Error 37.40 128 0.29

 

124

 

Table A-7

RMA on egg volume by Rank (within-clutches factor),
and by Clsz (among-clutches factors), in Low-Egsz nests.
Bonferroni-corrected critical P-level: 0.0125.

Results are used in text Table 27A.

 

 

 

 

 

Source Sum of qu D.F. Mean Sq F-ratio P-value ,3
Among Clutches
Clsz 9.53 1 9.53 5.732 0.018
Error 176.18 106 1.66
Within Clutches . '
Rank 37.84 12.62 33.282 < 0.001 4L"
R X C 0.57 0.19 0.501 0.627
Error 120.53 318 0.38
Linear Contrast
Rank 2.69 1 2.69 3.846 0.052
R X C 0.14 1 0.14 0.194 0.660
Error 74.22 106 0.70
Quadratic C ontr.
Rank 34.96 1 34.96 128.537 < 0.001
R X C 0.43 1 0.43 1.595 0.209
Error 28.83 106 0.27

 

125

 

Table A-8

RMA on egg volume by Rank (within-clutches factor),

and by Clsz (among-clutches factors), in High-Egsz nests.
Bonferroni-corrected critical P-level: 0.0125.
Results are used in text Table 27B.

 

 

 

 

 

Source Sum of qu D.F. Mean Sq F -ratio P—value

Among Clutches
Clsz 1.31 1 1.31 0.689 0.408
Error 201.64 106 1.90

Within Clutches
Rank 68.46 22.82 60.460 < 0.001
R X C 9.32 3.11 8.228 < 0.001
Error 120.02 318 0.38

Linear Contrast
Rank 0.06 1 0.06 0.086 0.770
R X C 6.56 1 6.56 9.788 0.002
Error 71.00 106 0.67

Quadratic Contr.
Rank 68.24 1 68.24 227.854 < 0.001
R X C 2.04 1 2.04 6.826 0.010
Error 31.75 106 0.30

 

126

REFERENCES

Afton AD. 1984. Inﬂuence of age and time on reproductive performance of female
Lesser Scaup. Auk 101:255-265.

Alisauskas RT, Ankney CD. 1985. Nutrient reserves and the energetics of reproduction
in American Coots. Auk 102:133-144.

Alisauskas RT, Ankney CD. 1994. Costs and rates of egg formation in Ruddy Ducks.
Condor 96:11-18. .

Amat JA, Fraga RM, Arroyo GM. 2001. Intraclutch egg-size variation and offspring
survival in the Kentish Plover Charadrius alexandrinus. Ibis 143:17-23.

Anderson VR, Alisauskas RT. 2002. Composition and growth of King Eider ducklings
in relation to egg size. Auk 119:62-70.

Ankney CD. 1980. Egg weight, survival, and growth of Lesser Snow Goose goslings.
J Wildl Manage 44: 1 74-1 82.

Ankney CD, Afton AD. 1988. Bioenergetics of breeding Northern Shovelers: diet,
nutrient reserves, clutch size, and incubation. Condor 90:459-472.

Ankney CD, Alisauskas RT. 1991. Nutrient-reserve dynamics and diet of breeding
female Gadwalls. Condor 93 2799-8 1 0.

Ankney CD, Maclnnes CD. 1978. Nutrient reserves and reproductive performance of
female Lesser Snow Geese. Auk 95:459-471.

Aparicio JM. 1994a. The effect of variation in the laying interval on proximate
determination of clutch size in the European Kestrel. J Avian Biol 25:275-280.

Aparicio JM. 1994b. The seasonal decline in clutch size: an experiment with
supplementary food in the Kestrel, Falco tinnunculus. Oikos 71:451-458.

Aparicio J M. 1998. Individual optimization may explain differences in breeding time in
the European Kestrel F alco tinunculus. J Avian Biol 29:121-128.

Arcese P, Smith JNM. 1988. Effects of population density and supplemental food on
reproduction in Song Sparrows. J Anim Ecol 57:119-136.

Arnold TW. 1991. Intraclutch variation in egg size of American Coots. Condor 93:19-27.

127

Arnold TW. 1992 a. Continuous laying by American Coots in response to partial clutch
removal and total clutch loss. Auk 109:407-421.

Arnold TW. 1992 b. Variation in laying date, clutch size, egg size, and egg composition
of Yellow-headed Blackbirds (Xanthocephalus xanthocephalus): a supplemental

feeding experiment. Can J Zool 70: 1904- 1911.

Arnold TW. 1994. Effects on supplemental food on egg production in American Coots.
Auk 111:337-350.

Ashkenazi S, Yom-Tov Y. 1997. The breeding biology of the Black-crowned Night
Heron (Nycticorax nycticorax) and the Little Egret (Egretta garzetta) at the ,
Huleh Nature Reserve, Israel. J Zool Lond 242:623-641. l

 

'E‘A‘

ifmﬂniuiér -

Astheimer LB, Grau CR. 1985. The timing and energetic consequences of egg formation
in the Adelie Penguin. Condor 87:256-268.

Athanasiou H. 1990. Wetland habitat loss in Thessaloniki plain, Greece [MSc thesis].
University College, London.

Barba E, Gil-Delgado JA, Monros J S. 1995. The costs of being late: consequences of
delaying Great Tit Parus major ﬁrst clutches. J Anim Ecol 64:642-651.

Beissinger SR, Waltman JR. 1991. Extraordinary clutch size and hatching asynchrony
of a neotropical parrot. Auk 108:863-871.

Bennetts RE, Fasola M, Hafner H, Kayser Y. 2000. Inﬂuence of environmental and
density-dependent factors on reproduction of Little Egrets. Auk 117:634-639.

Beukeboom L, Dijkstra C, Daan S, Meijer T. 1988. Seasonality of clutch size
determination in the Kestrel F alco tinnunculus: An experimental approach.

Omis Scand 19:41-48.

Birkhead TR, Nettleship DN. 1982. The adaptive signiﬁcance of egg size and laying date
in Thick-billed Murres Uria lomvia. Ecology 63:300-306.

Blaker D. 1969. Behaviour of the Cattle Egret Ardeola ibis. Ostrich 40:75-129.

Blomqvist D, Johansson 0C, Gotmark F. 1997. Parental quality and egg size affect chick
survival in a precocial bird, the Lapwing Vanellus vanellus. Oecologia 110: 1 8-24.

BOllinger PB. 1994. Relative effects of hatching order, egg-size variation, and parental
quality on chick survival in Common Tems. Auk 111:263-273.

128

Bolton M, Houston D, Monaghan P. 1992. Nutritional constraints on egg formation in the
Lesser Black-backed Gull: an experimental study. J Anim Ecol 61 :521-532.

Bortolotti GR. 1984. Criteria for determining the age and sex of nestling Bald Eagles.
J Field Ornithol 55:467-481.

Brinkhof MWG, Cave AJ, Hage FJ, Verhulst S. 1993. Timing of reproduction and
ﬂedging success in the Coot F ulica atra: evidence for a causal relationship.
J Anim Ecol 62:577—587.

Buntin JD. 1996. Neural and hormonal control of parental behavior in birds. p75]
In: Rosenblatt J S, Snowdon CT (eds), Parental Care: Evolution, Mechanisms,
and Adaptive Signiﬁcance. Acad Press, San Diego. p 161-213.

Buntin JD, E1 Halawani ME, Ottinger MA, Fan Y, Fivizzani AJ. 1998. An analysis of 1
sex and breeding stage differences in prolactin binding activity in brain and ..
hypothalamic GnRH concentration in Wilson’s Phalarope, a sex role-reversed 1.)
species. Gen Comp Endocr 109:119-132.

 

 

Buntin JD, Tesch D. 1985. Effects of intracranial prolactin administration on maintenance
of incubation readiness, ingestive behavior, and gonadal condition in Ring Doves.
Horrn Behav 19:188-203.

Burger J. 1978. Competition between Cattle Egrets and native North American herons,
egrets, and ibises. Condor 80:15-23.

Burger J. 1981. A model for the evolution of mixed-species colonies of Ciconiiformes.
Quart Rev Biol 56:143-167.

Burley RW, Vadehra DV. 1989. The Avian Egg: Chemistry and Biology.
Wiley & Sons, New York. 472 p.

Butler RW. 1993. Time of breeding in relation to food availability of female Great Blue
Herons (Ardea herodias). Auk 1 10:693-701.

Campos F, Femandez-Cruz M. 1991. The breeding biology of the Grey Heron (Ardea
cinerea) in the Duero river basin in Spain. Colon Waterbirds 14:57-60.

Campos F, Fraile B.1990. Les parametres reproducteurs du Heron cendre (Ardea cinerea)
dans 1e nord-ouest de l’Espagne. Oiseau 60:212-223.

Clark AB, Wilson DS. 1981. Avian breeding adaptations: hatching asynchrony, brood
reduction, and nest failure. Quart Rrev Biol 56:253-277.

129

Charalambides IC. 1994. Habitat partitioning in a mixed-species colony of herons at a
Ramsar wetland in northern Greece [MSc thesis]. University College of North
Wales, Bangor. 130 p.

Chamov EL, Krebs JR. 1974. On clutch-size and ﬁtness. Ibis 116:217-219.

Clamens A, Isenman P. 1989. Effect of supplemental food on the breeding of Blue and
Great Tits in Mediterranean habitats. Omis Scand 20:36-42.

Coleman J S, Fraser JD. 1989. Age estimation and growth of Black and Turkey Vultures.
J Field Omithol 60: 197-208. ‘ l'

Coulson JC, Porter JM. 1985. Reproductive success of the Kittiwake Rissa tridactyla: the l
roles of clutch size, chick growth rates and parental quality. Ibis 127:450-466. 1

Coulson JC, Potts GR, Horobin J. 1969. Variation in the eggs of the Shag (Phalacrocorax
aristotelis). Auk 86:232-245.

 

Cramp S, Simmons KEL. 1977. Handbook of the Birds of Europe, the Middle East and
North Africa: The Birds of the Western Palearctic. Vol I: Ostrich to Ducks.
Oxford Univ Press, Oxford. 722 p.

Crawford RD. 1977. Breeding biology of year-old and older female Red-winged and
Yellow-headed Blackbirds. Wilson Bull 89:73-80.

Custer TW. 1991. Green coloration of recently laid Black-crowned Night Heron eggs.
Colon Waterbirds 14:184-185.

Custer TW, Frederick PC. 1990. Egg size and laying order of Snowy Egrets, Great Egrets
and Black-crowned Night Herons. Condor 92:772-775.

Custer TW, Hensler GL, Kaiser TE. 1983. Clutch size, reproductive success, and
organochlorine contaminants in Atlantic Coast Black-crowned Night Herons.

Auk 100:699-710.

Custer TW, Hines RK, Custer CM. 1996. Nest initiation and clutch size of Great Blue
Herons on the Mississippi River in relation to the 1993 ﬂood. Condor 98:181-188.

Daan S, Dijkstra C, Drent R, Meijer T. 1989. Food supply and the annual timing of avian
reproduction. Proc Int Ornith Congr 19:392-407.

Davies NB, Lundberg A. 1985. The inﬂuence of food on time budgets and timing of
breeding of the Dunnock Prunella modularis. Ibis 127:100-110.

130

Dawson A, Goldsmith AR. 1982. Prolactin and gonadotrophin secretion in wild Starlings
(Sturnus vulgaris) during the annual cycle and in relation to nesting, incubation,
and rearing young. Gen Comp Endocr 48:213-221.

De Forest LN, Gaston AJ. 1996. The effect of age on timing of breeding and reproductive
success in the Thick-billed Murre. Ecology 77:1501-1511.

Delehanty DJ, Oring LW, Fivizzani AJ, E1 Halawani ME. 1997. Circulating prolactin of
incubating male Wilson’s Phalaropes corresponds to clutch size and environ-
mental stress. Condor 99:397-405.

Desrochers A. 1992. Age and foraging success in European Blackbirds: variation between
and within individuals. Anim Behav 43:885-894.

Desrochers A, Magrath RD. 1993. Age-speciﬁc fecundity in European Blackbirds
(Turdus merula): individual and population trends. Auk 110:255-263.

De Steven D. 1978. The inﬂuence of age on the breeding biology of the Tree Swallow
Iridoprocne bicolor. Ibis 120:516-523.

Dhinsa MS, Boag DA. 1990. The effect of food supplementation on the reproductive
success of Black-billed Magpies Pica pica. Ibis 132:595-602.

Dickerrnan RW, Gavino GT. 1969. Studies of a nesting colony of Green Herons at San
Blas, Nayarit, Mexico. Living Bird 8:95-111.

Dijkstra C, Bult A, Bijlsma S, Dan S, Meijer T, Zijlstra. 1990. Brood size manipulations
in the Kestrel (Falco tinnunculus): effects on offspring and parental survival.
J Anim Ecol 59:269-285.

Dijkstra C, Daan S, Meijer T, Cave AJ, Foppen RPB. 1988. Daily and seasonal variations
in body mass of the Kestrel in relation to food availability and reproduction.
Ardea 76: 127-140.

Draulans D. 1988. The importance of heronries for mate attraction. Ardea 76:187-192.

El Halawani ME, Silsby JL, Behnke EJ, F ehrer SC. 1986. Hormonal induction of
incubation behavior in ovariectomized female Turkeys (Meleagris gallopavo).
Biol Reprod 35:59-67.

El Halawani ME, Silsby JL, Youngren OM, Phillips RE. 1991. Exogenous prolactin

delays photo-induced luteinizing hormone secretion in the Turkey (Meleagris
gallopavo). Biol Reprod 44:420-424.

131

Elowe KD, Payne S. 1979. Ageing young Herring Gulls from measurements of body
parts. Bird Banding 50:49-55.

Erikstad KE, Bustnes JO, Moum T. 1993. Clutch-size determination in precocial birds:
a study of the Common Eider. Auk 110:623-628.

Erikstad KE, Pedersen HC, Steen JB. 1985. Clutch size and egg size variation in Willow
Grouse Lagopus l. lagopus. Omis Scand 16:88-94.

Etches RJ. 1996. Reproduction in Poultry. University Press, Cambridge. 318 p.

Etches RJ, Garbutt A, Middleton AL. 1979. Plasma concentrations of prolactin
during egg laying and incubation in the Ruffed Grouse (Bonasa umbellus).
Can J Zool 57:1624-27.

Evans RM. 1969. Speciﬁc gravity of White Pelican eggs. Auk 86:560-561.

F asola M, Alieri R. 1992. Nest site characteristics in relation to body size in herons in
Italy. Colon Waterbirds 15:185-191.

F emandez—Cruz M, Campos F. 1993. The breeding of Grey Herons (Ardea cinerea) in
western Spain: the inﬂuence of age. Colon Waterbirds 16:53-58.

Follet BK, Robinson JE. 1980. Photoperiod and gonadotrophin secretion in birds. In:
Reiter RJ, Follet BK (eds), Seasonal Reproduction in Higher Vertebrates. p 39-61.

Forbes LS. 1989. Coloniality in herons: Lack’s predation hypothesis reconsidered.
Colon Waterbirds 12:24-29.

F orslund P, Larsson K. 1992. Age-related reproductive success in the Barnacle Goose.
J Anim Ecol 61:195-204.

Frederick PC, Collopy MW. 1989. Researcher disturbance in colonies of wading birds:
effects of frequency of visit and egg-marking on reproductive parameters.
Colon Waterbirds 12: 1 52-1 57.

Freeman S, Jackson WM. 1990. Univariate metrics are not adequate to measure avian
body size. Auk 107:69-74.

F ujioka M. 1984. Asynchronous hatching, growth and survival of chicks of the Cattle
Egret Bubulcus ibis. Tori 33:1-12.

Fumess RW. 1983. Variations in size and growth of Great Skua Catharacta skua chicks
in relation to adult age, hatching date, egg volume, brood size and hatching
sequence. J Zool Lond 199:101-116.

132

 

Galbraith H. 1988. Effects of egg size and composition on the size, quality and survival
of Lapwing Vanellus vanellus chicks. J Zool Lond 214:383-398.

Gehlbach FR, Roberts JC. 1997. Experimental feeding of suburban Eastern Screech-Owls

Otus asio has few effects on reproduction apart from non-experimental factors.
J Avian Biol 28:38-46.

Gilliland SG, Ankney CD. 1992. Estimating age of young birds with a multivariate
measure of body size. Auk 109:444-450.

Goldsmith AR. 1983. Prolactin in avian reproductive cycles. In: Balthazart J, Prove E,

Gilles R (eds), Hormones and Behaviour in Higher Vertebrates. Springer-Verlag,
Berlin. p 375-387.

Goldsmith AR. 1991. Prolactin and avian reproductive systems. Proc Int Omith Congr
20:2063-2071.

Goldsmith AR, Burke S, Prosser JM. 1984. Inverse changes in plasma prolactin and LH
concentrations in female canaries after disruption and reinitiation of incubation.
J Endocr 103:251-256.

Goutner V, Fumess RW, Papakostas G. 2001. Mercury in feathers of Squacco Heron
(Ardeola ralloides) chicks in relation to age, hatching order, growth, and sampling
dates. Envir Pollution 111:107-115.

Grant MC. 1991. Relationships between egg size, chick size at hatching, and chick
survival in the Whimbrel Numenius phaeopus. Ibis 133:127-133.

Gratto-Trevor CL, Oring LW, F ivizzani AJ, El Halawani ME, Cooke F. 1990. The role
of prolactin in parental care in a monogamous and a polyandrous shorebird.
Auk 107:718-729.

Grau CR. 1996. Nutritional needs for egg formation in the Shag Phalacrocorax aristotelis
Ibis 138:756-764.

Hafner H. 1978. Le succes de reproduction de quatre especes d’ardeides (Egretta g.
garzetta L., Ardeola r. ralloides Scop., Ardeola i. ibis L., Nycticorax n. nycticorax
L.) en Camargue. Terre Vie 32:279-289.

Hafner H. 1980. Etude ecologique des colonies de herons arboricoles (Egretta g. garzetta
L., Ardeola r. ralloides Scop., Ardeola i. ibis L., Nycticorax n. nycticorax L.) en
Camargue. Bonn zool Beitr 31:249-287.

133

Haﬁier H, Boy V, Gory G. 1982. Feeding methods, ﬂock size and feeding success in the
Little Egret Egretta garzetta and the Squacco Heron Ardeola ralloides in
Camargue, southern France. Ardea 70:45-54.

Haftom S. 1981. Incubation during the egg-laying period in relation to clutch-size
and other aspects of reproduction in the Great Tit Parus major.
Omis Scand 12:169-185.

Hakkarainen H, Korpimaki E. 1993. The effect of female body size on clutch volume of

Tengmalm’s Owls Aegoliusﬁmereus in varying food conditions.
Omis F ennica 70:189-195. 5

Hall MR, Goldsmith AR. 1983. Factors affecting prolactin secretion during breeding
and incubation in the Domestic Duck (Anas platyrhynchos). Gen Comp Endocr
49:270-276.

Hamer KC, F urness RW. 1991. Age-speciﬁc breeding performance and reproductive é
effort in Great Skuas Catharacta skua. J Anim Ecol 60:693-704. '~

 

Hancock J. 1999. Herons and Egrets of the World. Academic Press, San Diego. 208 p.

Harris MP, Buckland ST, Russell SM, Wanless S. 1994. Post ﬂedging survival to
breeding age of Shags Phalacrocorax aristotelis in relation to year, date of
ﬂedging and brood size. J Avian Biol 25:268-274.

Hatchwell BJ. 1991. An experimental study of the effects of timing of breeding on the
reproductive success of Common Guillemots (Uria aalge).

J Anim Ecol 60:721-736.

Haywood S. 1993a. Sensory and hormonal control of clutch size in birds.
Quart Rev Biol 68:34-60.

Haywood S. 1993b. Sensory control of clutch size in the Zebra Finch (Taeniopygia
guttata). Auk 110:778-786.

Haywood S. 1993c. Role of extrinsic factors in the control of clutch—size in the Blue Tit
Parus caeruleus. Ibis 135279-84.

Hector JAL, Goldsmith AR. 1985. The role of prolactin during incubation: comparative
studies of three Diomedea albatrosses. Gen Comp Endocr 60:236-243.

Heeb P. 1994. Intraclutch egg-mass variation and hatching asynchrony in the J ackdaw
Corvus monedula. Ardea 82:287-297.

134

Hepp GR, Stangohr DJ, Baker LA, Kennamer RA. 1987. Factors affecting variation in
the egg and duckling components of Wood Ducks. Auk 104:435-443.

Hill WL. 1988. The effect of food abundance on the reproductive patterns of coots.
Condor 90:324-331.

Hiom L, Bolton M, Monaghan P, Worrall D. 1991. Experimental evidence for food
limitation of egg production in gulls. Omis Scand 22:94-97.

Hochachka WM, Boag DA. 1987. Food shortage for breeding Black-billed Magpies
(Pica pica): an experiment using supplemental food. Can J Zool 65: 1270-1274.

Hogstedt G. 1981. Effect of additional food on reproductive success in the Magpie
(Pica pica). J Anim Ecol 50:219-229.

Horak P, Mand R, Ots I, Leivits A. 1995. Egg size in the Great Tit Parus major :
individual, habitat and geographic differences. Omis Fennica 72:97-114.

Horsfall JA. 1984. Food supply and egg mass variation in the European Coot.
Ecology 65:89-95.

Houston DC, Donnan D, Jones PJ. 1995. The source of the nutrients required for egg
production in Zebra Finches Poephila guttata. J Zool Lond 235:469-483.

Houston DC, Jones PJ, Sibly RM. 1983. The effect of female body condition on egg
laying in Lesser Black-backed Gulls Larusfuscus. J Zool Lond 200:509-520.

Howe HF. 1976. Egg size, hatching asynchrony, sex, and brood reduction in the
Common Grackle. Ecology 57:1195-1207.

Howe HF. 1978. Initial investment, clutch size, and brood reduction in the Common
Grackle (Quiscalus quiscula L.). Ecology 59:1109-1122.

Hoyt DF. 1979. Practical methods for estimating volume and fresh weight of bird eggs.
Auk 96:73-77.

Huynh H, F eldt LS. 1976. Estimation of the Box correction for degrees of freedom from
sample data in randomized block and split-plot designs. J Educ Stat 1:69-82.

Inoue Y. 1985. The process of asynchronous hatching and sibling competition in the
Little Egret Egretta garzetta. Colon Waterbirds 8:1-12.

James FC, McCullogh CE. 1990. Multivariate analysis in ecology and systematics:
panacea or Pandora’s box? Ann Rev Ecol Syst 21 :129-166.

135

 

Jansen A. 1990. Acquisition of foraging skills by Heron Island Silvereyes Zosterops
lateralis chlorocephala. Ibis 132295-101 .

Jarvinen A. 1991. A meta-analytic study of the effects of female age on laying-date and
clutch-size in the Great Tit Parus major and the Pied Flycatcher F icedula
hypoleuca. Ibis 133 :62-66.

J arvinen A, Vaisanen RA. 1983. Egg size and related reproductive traits in a southern

passerine, F icedula hypoleuca, breeding in an extreme northern environment.
Omis Scand 14:253-262.

J arvinen A, Vaisanen RA. 1984. Reproduction of Pied Flycatchers (F icedula hypoleuca) E]
in good and bad breeding seasons in a northern marginal area. Auk 101:439-450. ‘

Jenni DA. 1969. A study of the ecology of four species of herons during the breeding
season at Lake Alice, Alachua County, Florida. Ecol Monogr 39:245-270.

 

Johnson AL. 1986. Reproduction in the female. In: Sturkie PD (ed), Avian Physiology. 'L

Springer-Verlag, New York. p 403-431.

Jones PJ, Ward P. 1976. The level of reserve protein as the proximate factor controlling
the timing of breeding and clutch-size in the Red-billed Quelea Quelea quelea.
Ibis 118:547-574.

Jones PJ, Ward P. 1979. A physiological basis for colony desertion by Red-billed
Queleas (Quelea quelea). J Zool Lond 189:1-19.

Joseﬁk M. 1969. Studies on the Squacco Heron, Ardeola ralloides (Scopoli), 1.
Acta Omithol 11:135-262.

Joseﬁk M. 1970. Studies on the Squacco Heron, Ardeola ralloides (Scopoli), II-IV.
Acta Omithol 12:57-102, 393-443, 445-504.

Jover L, Ruiz X, Gonzales-Martin M. 1993. Signiﬁcance of intraclutch egg size variation
in the Purple Heron. Omis Scand 24:127-134.

Kazantzidis S, Goutner V. 1996. Foraging ecology and conservation of feeding habitats
of Little Egrets (Egretta garzetta) in the Axios river delta, Macedonia, Greece.
Colon Waterbirds 19(Spec Publ 1):1 15-121.

Kennamer RA, Alsum SK, Colwell SV. 1997. Composition of Wood Duck eggs in

relation to egg size, laying sequence, and skipped days of laying.
Auk 114:479-487.

136

Kennedy ED. 1991. Deterrninate and indeterminate egg-laying patterns: A review.
Condor 93: 106-124.

Kilpi M, Hillstrom L, Lindstrom K. 1996. Egg-size variation and reproductive success
in the Herring Gull Larus argentatus: adaptive or constrained size of the last egg?
Ibis 138:212-217.

Klomp H. 1970. The determination of clutch size in birds: a review. Ardea 58:1-125.

Korpimaki E. 1988. Effects of age on breeding performance of Tengmalm’s Owl
Aegolius funereus in western Finland. Omis Scand 19:21-26.

Korpimaki E. 1989. Breeding performance of Tengmalm’s Owl Aegolius/unereus:
effects of supplementary feeding in a peak vole year. Ibis 131:51-56.

Krebs JR. 1974. Colonial nesting and social feeding as strategies exploiting food
resources in the Great Blue Heron (Ardea herodias). Behaviour 51 :99-134.

Krebs JR. 1978. Colonial nesting in birds, with special reference to the Ciconiiformes.
Wading Birds 7:299-314.

Krementz DG, Ankney CD. 1986. Bioenergetics of egg production by female House
Sparrows. Auk 103:299-305.

Krementz DG, Nichols JD, Hines JE. 1989. Postﬂedging survival of European Starlings.
Ecology 70:646-655.

Lack D. 1947. The signiﬁcance of clutch size. Ibis 89:302-352.

Lack D. 1954. The natural regulation of animal numbers. Clarendon Press, Oxford. 343 p.

Lack D. 1968. Population studies of birds. Clarendon Press, Oxford. 341 p.

Langley CH. 1983. Biology of the Little Bittem in the southwestern Cape.
Ostrich 54:83-94. '

Lea RW, Dods ASM, Sharp PJ, Chadwick A. 1981. The possible role of prolactin in
the regulation of nesting behaviour and the secretion of luteinizing hormone in

broody Bantams. J Endocr 91:89-97.

Leblanc Y. 1987. Intraclutch variation in egg size of Canada Geese. Can J 2001
65:3044-47.

Leblanc Y. 1989. Variation in size of eggs of captive and wild Canada Geese.
Omis Scand 20:93-98.

137

 

ICI‘ -v' .' .. _
_ Alt";

 

Lederle PE. 1995. Effects of female age on reproduction, parental care and growth of
young in Tree Swallows (T achycineta bicolor) [PhD dissertation]. Michigan
State University. 143 p.

Lessells CM, Krebs JR. 1989. Age and breeding performance of European Bee-eaters.
Auk 106:375-382.

MacCluskie MC, Sedinger J S. 2000. Nutrient reserves and clutch-size regulation of
Northern Shovelers in Alaska. Auk 117:971-979.

Magrath RD. 1989. Hatching asynchrony and reproductive success in the Blackbird. I?!
Nature 339:536-538. '

Magrath RD. 1990. Hatching asynchrony in altricial birds. Biol Review 65:587-622.

Magrath RD. 1992a. Roles of egg mass and incubation pattern in establishment of
hatching hierarchies in the Blackbird (T urdus merula). Auk 109:474-487.

 

\t;.._»

Magrath RD. 1992b. Seasonal changes in egg-mass within and among clutches of birds:
general explanations and a ﬁeld study of the Blackbird Turdus merula.
Ibis 134:171-179.

March JB, Sharp PJ, Wilson PW, Sang HM. 1994. Effect of active immunization against
recombinant-derived chicken prolactin fusion protein on the onset of broodiness
and photoinduced egg laying in bantam hens. J Reprod Fertility 101:227-233.

Maxwell GR, Kale HW. 1977. Breeding biology of ﬁve species of herons in coastal
Florida. Auk 94:689-700.

McAloney K. 1973. The breeding biology of the Great Blue Heron on Tobacco Island,
Nova Scotia. Can Field-Nat 87:137-140.

McKilligan NG. 1985. The breeding success of the Indian Cattle Egret Ardeola ibis in
eastern Australia. Ibis 127:530-536.

Mead PS, Morton ML. 1985. Hatching asynchrony in the Mountain White-crowned
Sparrow (Zonotrichia leucophrys oriantha): a selected or incidental trait?
Auk 102:781-792.

Meathrel CE, Ryder JP. 1987. Intraclutch variation in the size, mass and composition
of Ring-billed Gull eggs. Condor 89:364-368.

Meathrel CE, Ryder JP, Termaat BM. 1987. Size and composition of Herring Gull eggs:

relationship to position in the laying sequence and the body condition of females.
Colon Waterbirds 10:55-63.

138

Meijer T. 1989. Photoperiodic control of reproduction and molt in the Kestrel, F alco
tinnunculus. J Biol Rhythms 42351-364.

Meijer T. 1990. Incubation development and clutch size inthe Starling.
Omis Scand 21:163-168.

Meijer T. 1993. Is the Starling Sturnus vulgaris a determinate layer? Ibis 135:315-319.

Meijer T, Daan S, Dijkstra C. 1988. Female condition and reproduction: effects of food
manipulation in free-living and captive Kestrels. Ardea 76:141-154.

‘1

Meijer T, Daan S, Hall M. 1990. Family planning in the Kestrel (F alco tinnunculus): ! ,'
the proximate control of covariation of laying date and clutch size.
Behaviour 114:117-136.

Meij er T, Drent R. 1999. Re-examination of the capital and income dichotomy in
breeding birds. Ibis 141:399-414.

 

E

Meijer T, Masman D, Daan S. 1989. Energetics of reproduction in female Kestrels.
Auk 106:549-559.

Meijer T, Nienaber U, Langer U, Trillmich F. 1999. Temperature and timing of egg-
laying of European Starlings. Condor 101:124-132.

Miller EH. 1979. Egg size in the Least Sandpiper Calidris minutilla on Sable Island,
Nova Scotia, Canada. Omis Scand 10:10-16.

Mills JA. 1979. Factors affecting the egg size of Red-billed Gulls Larus novaehollandiae
scopulinus. Ibis 121:53-67.

Milstein P le S, Prestt 1, Bell AA. 1970. The breeding cycle of the Grey Heron.
Ardea 58:198-253.

Mineau P, Smith GEJ, Markel R, Lam C-S. 1982. Ageing Herring Gulls from hatching to
ﬂedging. J Field Omithol 53:394-402.

Mock DW. 1985. Siblicidal brood reduction: the prey-size hypothesis.
Am Nat 125:327-343. ‘

Mock DW, Forbes LS. 1994. Life-history consequences of avian brood reduction.
Auk 111:115-123.

Moltoni E. 1936. Le Garzaie in Italia. Riv Ital Omithol 62210-269.

139

 

Moore DJ, Williams TD, Morris RD. 2000. Mate provisioning, nutritional requirements
for egg production, and primary reproductive effort of female Common Terns
Sterna hirundo. J Avian Biol 31:183-196.

Moser ME. 1986. Breeding strategies of Purple Herons in the Camargue, France.
Ardea 74:91-100.

Moss R, Watson A, Rothery P, Glennie WW. 1981. Clutch size, egg size, hatch weight

and laying date in relation to early mortality in Red Grouse Lagopus lagopus
scoticus chicks. Ibis 123:450-462.

Murphy MT. 1986 a. Body size and condition, timing of breeding, and aspects of egg
production in Eastern Kingbirds. Auk 103:465-476.

Murphy MT. 1986 b. Temporal components of reproductive variability in Eastern
Kingbirds (Tyrannus tyrannus). Ecology 67: 1483-92.

Murton RK, Westwood NJ .1977. Avian Breeding Cycles. Oxford Univ Press,
Oxford. 594 p.

Naef-Daenzer B, Widmer F, Nuber M. 2001. Differential post-ﬂedging survival of
Great and Coal Tits in relation to their condition and ﬂedging date.
J Anim Ecol 70:730-738.

Nager RG, Monaghan P, Houston DC. 2000. Within-clutch trade-offs between the
number and quality of eggs: experimental manipulations in gulls.

Ecology 8121339-50.

Nager RG, Monaghan P, Houston DC. 2001. The cost of egg production: increased egg
production reduces future ﬁtness in gulls. J Avian Biol 32:1-59-166.

Nager RG, Zandt HS. 1994. Variation in egg size in Great Tits. Ardea 82:315-328.

Newton 1, Marquiss M. 1981. Effect of additional food on laying dates and clutch sizes
of Sparrowhawks. Omis Scand 12:224-229.

Newton 1, Marquiss M. 1984. Seasonal trend in the breeding performance of Sparrow-
hawks. J Anim Ecol 532809-829.

Nilsson J-A. 1991. Clutch size determination in the Marsh Tit (Parus palustris).
Ecology 72:1757-1762.

Nilsson J-A, Svensson E. 1993 a. Causes and consequences of egg mass variation
between and within Blue Tit clutches. J Zool Lond 230:469-481.

140

 

 

 

Nilsson J-A, Svensson E. 1993 b. The frequency and timing of laying gaps.
Omis Scand 24: 122-126.

Nilsson J-A, Svensson E. 1993 c. Energy constraints and ultimate decisions during egg-
laying in the Blue Tit. Ecology 74:244-251.

Nisbet ICT. 1973. Courtship-feeding, egg-size and breeding success in Common Terns.
Nature 2412141-142.

Nisbet ICT. 1978. Dependence of ﬂedging success on egg-size, parental performance
and egg-composition among Common and Roseate Terns, Sterna hirundo and FE]
S. dougalli. Ibis 120:207-215. I

Nisbet ICT, Cohen ME. 1975. Asynchronous hatching in Common and Roseate Terns, I'
Sterna hirundo and S. dougalli. Ibis 1172374-379. 1'

 

Nisbet ICT, Drudy WH. 1972. Post-ﬂedging survival in Herring Gulls in relation to L!
brood-size and date of hatching. Bird-Banding 43:161-172. .-

Nisbet ICT, Winchell JM, Heise AE. 1984. Inﬂuence of age on the breeding biology of
Common Terns. Colon Waterbirds 72117-126.

Nol E. 1989. Food supply and reproductive performance of the American Oystercatcher
in Virginia. Condor 91:429-435.

N01 E, Baker AJ, Cadman MD. 1984. Clutch initiation dates, clutch size, and egg size of
the American Oystercatcher in Virginia. Auk 1012855-867.

Nur N. 1988. The cost of reproduction in birds: an examination of the evidence.
Ardea 76:155-168.

O’Connor RJ. 1979. Egg weights and brood reduction in the European Swift (Apus apus).
Condor 81:133-145.

Oj anen M. 1983. Egg development and the related nutrient reserve depletion in the
Pied Flycatcher, F icedula hypoleuca. Ann Zool Fennici 20:293-300.

Ojanen M, Orell M, Vaisanen RA.1979. Role of heredity in egg size variation in the
Great Tit Parus major and the Pied Flycatcher F icedula hypoleuca.
Ornis Scand 10:22-28.

Ojanen M, Orell M, Vaisanen RA.1981. Egg size variation within passerine clutches:
effects of ambient temperature and laying sequence. Omis Fennica 58:93-108.

141

 

Opel H, Proudman JA. 1980. Failure of mammalian prolactin to induce incubation
behavior in Chickens and Turkeys. Poultry Sci 5922550-58.

Oring LW, Fivizzani AJ, Colwell MA, El Halawani ME. 1988. Hormonal changes
associated with natural and manipulated incubation in the sex-role reversed
Wilson’s Phalarope. Gen Comp Endocr 72:247-256.

Otto, C. 1979. Environmental factors affecting egg weight within and between colonies
of Fieldfare T urdus pilaris. Omis Scand 102111-116.

Owen DF. 1960. The nesting success of the Heron Ardea cinerea in relation to the !"|
availability of food. J Zool Lond 133:597-617. ‘

Owen M, West J. 1988. Variation in egg composition in semi-captive Barnacle Geese.
Ornis Scand 19:58-62.

 

ﬁrm-.w

Parsons J. 1970. Relationship between egg size and post-hatching chick mortality in the
Herring Gull (Larus argentatus). Nature 228:1221-22.

115*.

Parsons J. 1972. Egg size, laying date and incubation period in the Herring Gull.
Ibis 114:536-541.

Parsons J. 1975 a. Asynchronus hatching and chick mortality in the Herring Gull
Larus argentatus. Ibis 1 17:517-520.

Parsons J. 1975 b. Seasonal variation in the breeding success of the Herring Gull:
an experimental approach to pre-ﬂedging success. J Anim Ecol 44:553-573.

Parsons J. 1976. Factors determining the number and size of eggs laid by the Herring
Gull. Condor 78:481-492.

Pelayo JT, Clark RG. 2002. Variation in size, composition, and quality of Ruddy Duck
eggs and ducklings. Condor 104:457-462.

Perdeck AC, Cave AJ. 1989. Inﬂuence of temperature on the laying date of the Coot:
between pairs and within individuals relationships. Ardea 77199-105.

Perrins CM. 1965. Population ﬂuctuations and clutch-size in the Great Tit Parus major L.
J Anim Ecol 34:601-647.

Perrins CM. 1966. Survival of young Manx Shearwaters Puﬂinus puﬂinus in relation to
their presumed date of hatching. Ibis 108:132-135. '

Perrins CM. 1970. The timing of birds’ breeding seasons. Ibis 112:242-255.

142

Perrins CM. 1996. Eggs, egg formation and the timing of breeding. Ibis 138:2-15.

Perrins CM, McCleery RH. 1985. The effect of age and pair bond on the breeding success
of Great Tits Parus major. Ibis 127:306-315.

Perrins CM, McCleery RH. 1994. Competition and egg-weight in the Great Tit Parus
major. Ibis 136:454-456.

Perrins CM, Moss D. 1975. Reproductive rates in the Great Tit. J Anim Ecol 44:695-706.

Pettifor RA.1993. Brood-manipulation experiments. I. The number of offspring surviving
per nest in Blue Tits (Parus caeruleus). J Anim Ecol 62:131-144.

Pettifor RA, Perrins CM, McCleery RH. 1988. Individual optimization of clutch size in
Great Tits. Nature 336:160-162.

Pettifor RA, Perrins CM, McCleery RH. 2001. The individual optimization of ﬁtness:
variation in reproductive output, including clutch size, mean nestling mass and

offspring recruitment, in manipulated broods of Great Tits Parus major.
J Anim Ecol 70:62-79.

Phillips RA, Fumess RW. 1997. Predicting the sex of Parasitic Jaegers by discriminant
analysis. Colon Waterbirds 20:14-23.

Pierotti R, Bellrose CA. 1986. Proximate and ultimate causation of egg size and the
"third-chick disadvantage" in the Western Gull. Auk 1032401 -407.

Pietiainen H, Saurola P, Vaisanen RA. 1986. Parental investment in clutch size and egg
size in the Ural Owl Strix uralensis. Omis Scand 17:309-325.

Porter RD. 1975. Experimental alterations of clutch-size of captive American Kestrels
Falco sparverius. Ibis 1 17:510-515.

Powell GVN, Powell AH. 1986. Reproduction by Great White Herons Ardea herodias
in Florida Bay as an indicator of habitat quality. Biol Conservation 36:101-1 13.

Pratt HM. 1970. Breeding biology of Great Blue Herons and Common Egrets in central
California. Condor 72:407-416.

Pratt HM. 1972a. Nesting success of Common Egrets and Great Blue Herons in the
San Francisco Bay region. Condor 74:447-453.

Pratt HM. 1972b. Nesting success of Great Blue Herons and Common Egrets at
Audubon Canyon Ranch in 1971. Amer Birds 26:699-702.

143

 

Pratt HM, Winkler DW. 1985. Clutch size, timing of laying and reproductive success in a
colony of Great Blue Herons and Great Egrets. Auk 102249-63.

Proudman JA, Opel H. 1981. Turkey prolactin: validation of a radioimmunoassay and
measurement of changes associated with broodiness. Biol Reprod 25:573-580.

Pugesek BH. 1983. The relationship between parental age and reproductive effort in the
California Gull (Larus californicus). Behav Ecol Soc 13:161-171.

Ramsay SL, Houston DC. 1997. Nutritional constraints on egg production in the Blue
Tit: a supplementary feeding study. J Anim Ecol 66:649-657.

Ramsey SM, Goldsmith AR, Silver R. 1985. Stimulus requirements for prolactin and LH
secretion in incubating Ring Doves. Gen Comp Endocr 59:246-256.

Ratcliffe N, F umess RW, Hamer KC. 1998. The interactive effects of age and food
supply on the breeding ecology of Great Skuas. J Anim Ecol 67:853-862.

 

Reid WV. 1987. Constraints on clutch size in the Glaucus-winged Gull.
Stud Avian Biol 1028-25.

Reid WV. 1988. Age-speciﬁc patterns of reproduction in the Glaucus-winged Gull:
increased effort with age? Ecology 69: 1454-1465.

Ricklefs RE, Hahn DC, Montevecchi WA. 1978. The relationship between egg size and
chick size in the Laughing Gull and the Japanese Quail. Auk 95:135-144.

Rising JD, Somers KM. 1989. The measurement of overall body size in birds.
Auk 106:666-674.

Robertson GJ, Cooch EG, Lank DB, Rockwell RF, Cooke F. 1994. Female age and egg
size in the Lesser Snow Goose. J Avian Biol 25:149-155.

Robertson GJ, Cooke F. 1993. Intraclutch egg-size variation and hatching success in the
Common Eider. Can J 2001 71:544-549.

Rodgers JA. 1980a. Breeding ecology of the Little Blue Heron on the West Coast of
Florida. Condor 82: 164-169.

Rodgers JA.1980b. Little Blue Heron breeding behavior. Auk 97:371-3 84.

Rofstad G, Sandvik J. 1985. Variation in egg size of the Hooded Crow Corvus corone
cornix. Omis Scand 16:38-44.

144

 

Rofstad G, Sandvik J. 1987. Morphology of hatchling Hooded Crows and its relation
to egg volume. Condor 89:494-499.

Rohwer F C. 1988. Inter- and intraspeciﬁc relationships between egg size and clutch size
in waterfowl. Auk 105:161-176.

Runde OJ, Barrett RT. 1981. Variations in egg size and incubation period of the
Kittiwake Rissa tridactyla in Norway. Ornis Scand 12:80-86.

Ryden O. 1978. Egg weight in relation to laying sequence in a South Swedish urban
population of the Blackbird T urdus merula. Omis Scand 9:172-177.

Salzer DW, Larkin GJ. 1990. IMpact of courtship feeding on clutch and third-egg size
in Glaucus-winged Gulls. Anim Bahav 3921149-62.

Scanes CG. 1986. Pituitary gland. In: Sturkie PD (ed), Avian physiology.
Springer-Verlag, New York, USA. p 383-402.

Schubert CA, Cooke F. 1993. Egg-laying intervals in the Lesser Snow Goose.
Wilson Bull 105:414-426.

Sharp PJ, Sterling RJ, Talbot RT, Huskisson NS. 1989. The role of hypothalamic
vasoactive intestinal polypeptide in the maintenance of prolactin secretion in
incubating bantam hens: observations using passive immunization, radio-
immunoassay and immunohistochemistry. J Endocrinol 12225-13.

Silverin B, Goldsmith AR. 1983. Reproductive endocrinology of free living Pied
F lycatchers (Ficedula hypoleuca): prolactin and F SH secretion in relation to
incubation and clutch size. J Zool Lond 200:119-130.

Slagsvold T. 1976. Annual and geographical variation in the time of breeding of the
Great Tit Parus major and the Pied Flycatcher F icedula hypoleuca in relation to

environmental phenology and spring temperature. Omis Scand 7: 127-145.

Slagsvold T. 1982. Clutch size, nest size, and hatching asynchrony in birds: experiments
with the Fieldfare (T urdus pilaris). Ecology 63:1389-99.

Slagsvold T, Lifjeld JT. 1989. Constraints on hatching asynchrony and egg size in Pied
Flycatchers. J Anim Ecol 58:837-849.

Slagsvold T, Sandvik J, Rofstad G, Lorentsen O, Husby M. 1984. On the adaptive value
of egg-size variation in birds. Auk 1012685-697.

Smith CC, Fretwell SD.1974. The optimal balance between size and number of offspring.
Am Nat 108:499-505.

145

 

Smith HG, Kallander H, Nilsson J-A. 1989. The trade-off between offspring number and
quality in the Great Tit Parus major. J Anim Ecol 58:383-401.

Smith HG, Ohlsson T, Wettermark K-J. 1995. Adaptive signiﬁcance of egg size in the
European Starling: experimental tests. Ecology 7621-7.

Smith HG, Ottosson U, Ohlsson T. 1993. Interclutch variation in egg mass among
Starlings Sturnus vulgaris reﬂects female condition. Omis Scand 24:311-316.

Sockman KW, Schwabl H. 1999. Daily estradiol and progesterone levels relative to
laying and onset of incubation in Canaries. Gen Comp Endocr 114:257-268.

Sockman KW, Schwabl H, Sharp PJ. 2000. The role of prolactin in the regulation of
clutch size and onset of incubation behavior in the American Kestrel.
Hormone Behav 38:168-176.

Soler M, Soler JJ. 1996. Effects of experimental food provisioning on reproduction in the
Jackdaw Corvus monedula, a semi-colonial species. Ibis 138:377-3 83.

Sterbetz I. 1962. The Squacco Heron in the Saser bird-sanctuary. Aquila 67-68239-70.

Stokland JN, Amundsen T. 1988. Initial size hierarchy in broods of the Shag: relative
signiﬁcance of egg size and hatching asynchrony. Auk 105:308-315.

Stoleson SH, Beissinger SR. 1995. Hatching asynchrony and the onset of incubation
in birds, revisited. Current Omithol 12:191-270.

Styrsky JD, Dobbs RC, Thompson CF. 2000. Food-supplementation does not override
the effect of egg mass on ﬁtness-related traits of nestling House Wrens.

J Anim Ecol 69:690-702.

Swennen C, van der Meer J. 1992. Variation in egg size of Common Eiders.
Ardea 80:363-373.

Sydeman WJ, Emslie SD. 1992. Effects on parental age on hatching asynchrony, egg size
and third-chick disadvantage in Western Gulls. Auk 109:242-248.

Sydeman WJ, Penniman JF, Penniman TM, Pyle P, Ainley DG. 1991. Breeding perform-
ance in the Western Gull: effects of parental age, timing of breeding and year in

relation to food availability. J Anim Ecol 60:135-149.

SYSTAT 1996. SYSTAT 6.0 for Windows: Statistics. SPSS Inc, Chicago. 751 p.

146

 

Szlivka L. 1986. Data on the food of the Purple (Ardea purpurea), Night (Nycticorax
nycticorax), and Squacco (Ardeola ralloides) herons on Lake Ludas.

Larus 36-372 175-182.

Thompson PS, Hale WG. 1991. Age-related reproductive variation in the Redshank
T ringa totanus. Ornis Scand 22:353-359.

Tremblay J, Ellison LN. 1979. Effects of human disturbance on breeding of Black-
crowned Night Herons. Auk 96:364-369.

Tremblay J, Ellison LN. 1980. Breeding success of the black-crowned night heron in the
St. Lawrence Estuary. Can J Zool 5821259-63.

Vaisanen RA, Hilden O, Soikkeli M, Vuolanto S. 1972. Egg dimension variation in ﬁve
wader species: the role of heredity. Omis Fennica 49:25-44.

van Heezik YM, Seddon PJ, Du Plessis CJ, Adams NJ. 1993. Differential growth of King
Penguin chicks in relation to date of hatching. Colon Waterbirds 16:71-76.

van Noordwijk AJ, McCleery RH, Perrins CM. 1995. Selection for the timing of
Great Tit breeding in relation to catterpillar growth and temperature.
J Anim Ecol 642451-458.

van Vessem J, Draulans D. 1986 a. Nest attendance by male and female Gray Herons.
J Field Omithol 557:34-41.

van Vessem J, Draulans D. 1986 b. Factors affecting the length of the breeding cycle
and the frequency of nest attendance by Grey Herons Ardea cinerea.
Bird Study 33298-104.

Vega JP, Vinuela J. 1993. Hatching asynchrony and hatching success in the House
Sparrow: evidence for the egg viability hypothesis. Omis Scand 242237-242.

Verbeek NAM. 1988. Differential predation of eggs in clutches of Glaucous-winged
Gulls Larus glaucescens. Ibis 130:512-518.

Verbeek NAM. 1990. Differential predation on eggs in clutches of Northwestern Crows:
the importance of egg color. Condor 92:695-701.

Verhulst S, Tinbergen JM. 1991. Experimental evidence for a causal relationship between

timing and success of reproduction in the Great Tit Parus m. major.
J Anim Ecol 60:269-282.

Vinuela J. 1996. Establishment of mass hierarchies in broods of the Black Kite.
Condor 98:93-99.

147

Vleck CM. 2002. Hormonal control of incubation behaviour. In: Deeming DC (ed),

Avian Incubation: Behaviour, Environment, and Evolution. Oxford Univ Press,
New York. p 54-62.

Voisin C. 1978. Utilization des zones humides du delta rhodanien par les ardeides.
L’Oiseau 48:329-380.

Voisin C. 1980. Etude du comportement du Heron crabier (Ardeola ralloides) en periode
de reproduction. L’Oiseau 50:149-160.

Voisin C. 1991. The Herons of Europe. T&AD Poyser, London. 372 p.
von Ende CN. 1993. Repeated-measures analysis: Growth and other time-dependent
measures. In: Scheiner SM, Gurevitch J (eds), Design and Analysis of Ecological

Experiments. Chapman & Hall, New York. p 113-137.

Ward P, Zahavi A. 1973. The importance of certain assemblages of birds as "information-
centres" for food-ﬁnding. Ibis 115:517-534.

Watson MD, Robertson GJ, Cooke F. 1993. Egg-laying time and laying interval in the
Common Eider. Condor 95:869-878.

Weidmann U. 1956. Observations and experiments on egg-laying in the Black-headed
Gull (Larus ridibundus L.). Brit J Anim Behav 42150-161.

Werding G. 1970. Observations du comportement pendant la couvee des herons gardes-
boeufs et des crabiers chevelus et de la croissance des jeunes. Bull Madagascar

2922 753-768.

Werschkul DF. 1979. Nestling mortality and the adaptive signiﬁcance of early
locomotion in the Little Blue Heron. Auk 962116-130.

Wiebe KL, Bortolotti GR. 1995. Egg size and clutch size in the reproductive investment
of American Kestrels. J Zool Lond 237:285-301.

Wiebe KL, Martin K. 1995. Ecological and physiological effects on egg laying intervals
in Ptarmigan. Condor 97:708-717.

Wiebe KL, Wiehn J, Korpimaki E. 1998. The onset of incubation in birds: can females
control hatching patterns? Anim Behav 55: 1043-52.

Wiggins DA. 1990. Sources of variation in egg mass of Tree Swallows Tachycineta
bicolor. Omis Scand 21:157-160.

148

 

Wiklund CG. 1996. Body length and wing length provide univariate estimates of overall
body size in the Merlin. Condor 98:581-588.

Willebrand T. 1992. Breeding and age in female Black Grouse Tetrao tetrix.
Omis Scand 23:29-32.

Williams GC. 1966. Natural selection, the costs of reproduction, and a reﬁnement of
Lack’s principle. Am Nat 100:687-690.

Williams TD. 1994. Intraspeciﬁc variation in egg size and egg composition in birds:
effects on offspring ﬁtness. Biol Review 68:35-59.

Williams TD, Jeffs C, Murray KA, Choudhury S. 1996. Intraclutch egg-size variation
in the Barnacle Goose Branta leucopsis: an egg-removal experiment.
Ibis 1382499-505.

Williams TD, Lank DB, Cooke F. 1993. Is intraclutch egg-size variation adaptive in the
Lesser Snow Goose? Oikos 672250-256.

Wolford JW, Boag DA. 1971. Distribution and biology of Black-crowned Night Herons
in Alberta. Can F ield-Nat 85213-19.

Youngren OM, El Halawani ME, Silsby JL, Phillips RE. 1991. Intracranial prolactin
perfusion induces incubation behavior in Turkey hens. Biol Reprod 44:425-431.

Zadwomy D, Shimada K, Ishida H, Sato K. 1989. Gonadotropin-stimulated estradiol
production in small ovarian follicles of the Hen is suppressed by physiological

concentrations of prolactin in vitro. Gen Comp Endocr 742468-473.

Zar JH. 1996. Biostatistical analysis. 3rd ed. Prentice-Hall, New Jersey. 662 p.

149

 

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

llI]llljjlljljjltllljjwlll