Ibis (2006), 148, 221–230
Geographical and seasonal gradients in hatching
failure in Eastern Bluebirds Sialia sialis reinforce clutch
size trends
Blackwell Publishing Ltd
CAREN B. COOPER,* WESLEY M. HOCHACHKA, TINA B. PHILLIPS & ANDRÉ A. DHONDT
Laboratory of Ornithology, Cornell University, 159 Sapsucker Woods Road, Ithaca, NY 14850, USA
Eggs untended during the laying phase can lose viability if exposed to high temperatures,
such as those common at lower latitudes and late in the nesting season. The egg-viability
hypothesis states that constraints on viability during the laying phase could account for latitudinal and seasonal gradients in clutch size. We used 7 years’ worth of data collected by
volunteers (The Birdhouse Network, co-ordinated by the Cornell Laboratory of Ornithology)
to look simultaneously at populations across the temperate breeding range of Eastern
Bluebirds Sialia sialis in order to test three predictions of the egg-viability hypothesis: the
probability of hatching failure decreases at higher latitude and increases later in the season,
and that these trends are strongest among large clutches. The overall average number of
unhatched eggs relative to the total number of eggs laid was similar to that found by other
studies (7.8%; range 6.8–8.9% annually; n = 32 567 eggs from 7231 nests from 530 study
sites). Using generalized linear mixed models that controlled for the non-independence of
eggs within a clutch, we found that the ‘per-egg’ probability of hatching failure was highest
late in the season, highest at lower latitudes, and highest for both small (three-egg) and large
(six-egg) clutches. The seasonal and geographical gradients in egg hatching failure reinforce
documented seasonal and geographical trends in clutch size. Loss of egg viability prior to
incubation currently provides the most parsimonious and consistent explanation of the
observed patterns of hatching failure. However, alternative explanations for large-scale patterns, particularly those not consistent with the egg-viability hypothesis, warrant further
research into other causes of hatching failure, such as microbes and infertility (related to
extra-pair mating). Our results demonstrate that investigating causes of variation in demography among local populations across a geographical gradient provides a potential means
of identifying selection pressures on life-history traits.
Stoleson and Beissinger (1999) hypothesized that
the relationship between ambient temperature and
egg viability may constrain latitudinal and seasonal
trends in clutch size. As in other passerines, our study
species, the Eastern Bluebird Sialia sialis, generally
initiates incubation with the laying of the penultimate or ultimate egg (Deeming 2002). Eggs laid first
in the sequence therefore have prolonged exposure
to ambient environmental conditions prior to incubation. At temperatures below physiological zero (the
temperature that triggers some embryonic development), unattended eggs in the pre-incubation phase
can remain viable for extended periods (Dhondt
*Corresponding author.
Email: [email protected]
© 2006 The Authors
Journal compilation © 2006 British Ornithologists’ Union
1968, White & Kinney 1974, Webb 1987, Ewert
1992). Physiological zero is usually assumed to be at
24–26 °C for most bird species (Webb 1987). If eggs
are exposed to extended periods of ambient temperatures above physiological zero but below optimal
incubation temperatures (36–39 °C), the embryos
will experience unsynchronized tissue growth, abnormal development and mortality (Deeming &
Ferguson 1992, and references therein). The amount of
pre-incubation exposure to unfavourable temperatures
necessary to produce embryo mortality may vary
among species, although 3 days may be the minimum
(Stoleson & Beissinger 1999). As birds generally lay
one egg each day until clutch completion, first-laid
eggs in clutches of four or more eggs could lose viability.
We also expect that the larger the clutch, beyond this
222
C. B. Cooper et al.
minimum size, the greater the probability of an
unhatched egg, because of the greater probability
of eggs receiving prolonged exposure to ambient
temperatures in the critical range.
To prevent loss of egg viability, females can initiate
incubation early. However, early incubation or contact with eggs may disrupt ovarian follicular growth
and cause the cessation of laying (Haywood 1993a,
1993b), which would result in a smaller clutch size.
In addition, early incubation may have fitness costs
by limiting female foraging time (Slagsvold 1986,
Moreno 1989). Thus, with or without early onset,
females are unlikely to hatch large clutches successfully when temperatures are suboptimal.
In a multi-species comparison, Koenig (1982) reported
a linear decrease in hatching failure with increasing
latitude, and speculated that this pattern was related
indirectly to selection for greater fecundity at higher
latitudes. Thus, large-scale trends in hatching failure
may be relevant to our understanding of the evolution
of clutch size variation, particularly variation that
exhibits specific trends across spatial or temporal
gradients. If egg viability constrains clutch size along
latitudinal and seasonal gradients, then we expect
hatching failure patterns within a species to reinforce
these clutch size trends. In this paper we examine three
predictions of the egg-viability hypothesis: the
probability of hatching failure decreases at higher
latitude and increases later in the season, and that both
of these trends are strongest among large clutches.
METHODS
Eastern Bluebirds are abundant cavity-nesting passerines that exhibit a seasonal decline in clutch size
as well as larger clutches at more northern latitudes
(Gowaty & Plissner 2000, Dhondt et al. 2002). We
examined the occurrence of hatching failure in nests
across the temperate breeding range of the Eastern
Bluebird (Fig. 1). The Birdhouse Network (TBN),
administered by the Cornell Laboratory of Ornithology, relies on volunteers who collect information
on cavity-nesting species using nestboxes. Volunteers
report details of the nesting stages, such as when eggs
were seen, the number of eggs, when eggs hatch, the
number of unhatched eggs and when nestlings fledge
(see www.birds.cornell.edu/birdhouse for more
information). Each volunteer monitors from one to
over 100 nestboxes (Fig. 1). We used data reported
from between 1998 and 2004 inclusive.
When TBN participants submit data, they are
asked a series of questions regarding the details of each
Figure 1. Observation sites span most of the temperature breeding range of the Eastern Bluebird. Filled black circles indicate < 5
nestboxes; open circles indicate 5–50 nestboxes; open squares indicate > 50 nestboxes. Not shown is a site in southern Arizona.
© 2006 The Authors
Journal compilation © 2006 British Ornithologists’ Union
Hatching failure in Eastern Bluebirds
nest attempt, including ‘Were there any unhatched
eggs present?’ Participants can select the answer Yes,
No or Unknown, or leave the answer blank. Participants can answer the next question, ‘If yes, number
of unhatched eggs?’ as any number or unknown
(unknown is the default, i.e. no blank possible).
We established several criteria for records of nest
attempts to be included in our analyses. We included
nests for which at least one nestling was observed,
thereby excluding clutches that failed completely
to hatch. We included nests for which clutch size,
number of nestlings and clutch initiation date were
reported. After eliminating data that did not fit
these criteria, only 1% of the nest attempts contained
instances of egg loss other than that attributed to
hatching failure (i.e. egg loss determined by fewer
nestlings than eggs, but no unhatched eggs reported),
and we excluded these from analyses. We also
excluded approximately 10% of clutches which contained reporting errors indicated by a reported total
clutch size unequal to the total number unhatched
eggs plus maximum number nestlings (n = 861
clutches). We excluded clutches with fewer than
three eggs (n = 57) and with seven eggs (n = 13) or
more (n = 6) because there were few clutches of this
size and we did not want those to have disproportionate effects on the model fit and parameter estimates.
Response variables
To compare the rates of hatching failure with results
from previous studies, we computed the percentage
hatching failure as the total number of unhatched
eggs over the total number of eggs laid (×100). The
distribution of hatching failure was skewed (significantly non-normal, Kolmogorov–Smirnov goodness
223
of fit test, D = 0.451, P = 0.01). Arcsine transformation (H = arcsine(√p), where p = proportion of
unhatched eggs (Zar 1984)) was not successful in
improving the fit of the data to a normal distribution
(significantly non-normal, Kolmogorov–Smirnov test,
D = 0.467, P = 0.01), and thus parametric analyses
based on an assumption of normally distributed errors
were not used. Instead, we chose to examine patterns in per-egg probability of hatching failure, using
each egg’s failure (yes/no) as a binomial variable for
all of our formal statistical analyses. A per-egg analysis
does not require assumptions of normality and allows
us to avoid the confounding influence of clutch
size, because the probability of hatching failure in a
clutch might increase with clutch size simply due to
a higher probability of finding a single unhatched egg
when more eggs were present. To analyse the Boolean
response variable we used the GLIMMIX macro in
SAS (Littell et al. 1996) with PROC MIXED, with
a binomial error and logit link. The likelihood of
individual eggs hatching is probably not independent
for eggs within the same nest. We accounted for this
non-independence by using a unique identifier of
each nesting attempt as a random-effect variable in
our statistical analyses.
Predictive variables
The independent fixed-effect variables in our statistical models were year, day of breeding season,
latitude, longitude and clutch size, as well as various
interaction terms (Table 1). Testing our predictions
required two steps: statistical model selection
followed by biological interpretation of the best
model(s). Models A and E permitted testing only
of the broadest predictions of the egg-viability
Table 1. Series of models with varying fixed effects, listed with AICC values and support indicated based on the Akaike weight of the
model within the set (w). K is the number of parameters. ∆AICC is the difference between the AICC of a model and the best model.
A
Year
Day of season
Clutch size
Latitude
Longitude
Latitude × longitude
Clutch size × latitude
Clutch size × day of season
AICC
K
∆AICC
w
✗
✗
✗
✗
✗
B
✗
✗
✗
✗
✗
✗
182 846.3
15
0
0.96
182 852.9
18
6.6
0.04
C
✗
✗
✗
✗
✗
✗
182 859.6
18
13.3
0.00
D
✗
✗
✗
✗
✗
✗
✗
182 867.0
21
20.7
0.00
E
✗
✗
✗
✗
✗
✗
182 859.9
16
13.6
0.00
F
✗
✗
✗
✗
✗
✗
✗
182 866.9
19
20.6
0.00
G
H
✗
182 872.9
19
26.6
0.00
✗
✗
✗
✗
✗
✗
✗
✗
182 881.0
22
34.7
0.00
✗
✗
✗
✗
✗
✗
© 2006 The Authors
Journal compilation © 2006 British Ornithologists’ Union
224
C. B. Cooper et al.
hypothesis, namely that hatching failure decreases
with latitude and increases with day of season.
Models B, C, D and F, G, H, each of which contained
interactions between clutch size and latitude and/or
day of season, permitted testing of the more specific
prediction of the egg-viability hypothesis, i.e. that
large clutches at low latitudes or late in the season
will experience the highest rates of hatching failure.
We compared the relative strength of evidence of
each model (Buckland et al. 1997) using a ranking
paradigm based on AIC (Akaike Information Criteria)
values and Akaike weights (Johnson 1999, Anderson
et al. 2000). After selecting the best models from
the set as judged by ∆AIC scores within 10 of the
best models, and model-averaging the results across
these models, we examined the parameter estimates
and confidence limits to test predictions of the
egg-viability hypothesis.
Year and clutch size were treated as categorical
variables so as not to impose an a priori rank. Because
we wanted to distinguish between predictions of
monotonic latitudinal variation in hatching failure
with predictions of no spatial trends, preliminary
analyses examined the latitudinal gradient in hatching failure to determine whether latitude should be
included in the model as a categorical or continuous
variable. In these preliminary analyses, we binned
our data by latitude into 2° bands (roughly 200 km)
and compared model A (Table 1) with binned latitude as a categorical variable to model A with binned
latitude as a continuous, linear variable. Based on the
large (> 4) difference in AICC values, the data were
better described by models with binned latitude as
a continuous variable. Importantly, even when latitude was treated as a categorical variable, the results
revealed a monotonic latitudinal gradient rather
than geographical variation unrelated to latitude.
Specifically, within the band 28–34°N latitude,
there were gradual, but steadily decreasing, changes
in hatching failure among the 2° bins, and moving
north within the band 36–48°N, probabilities of
hatching failure dropped at an accelerating rate.
Given the better AICC values of the continuouslatitude model and the apparent gradient across
latitude, we used latitude as a continuous variable
(without bins) in all subsequent models.
Because the start of the breeding season varies
with latitude (Peakall 1970, Dhondt et al. 2002,
Cooper et al. 2005a, 2005b), the measurement of
clutch initiation as ‘day-of-year’ is confounded by
latitude (i.e. we do not wish to compare 30 March at
30°N latitude with 30 March at 46° latitude, but
© 2006 The Authors
Journal compilation © 2006 British Ornithologists’ Union
instead compared dates relative to the start of the
breeding season at a given latitude). We used a measure of clutch initiation that controls for latitudinal
differences in the start of the breeding season
(Cooper et al. 2005b). First, we divided the TBN
data among 2° latitudinal bands, corresponding to
even-numbered lines of latitude. Then, from all
Eastern Bluebird nest attempts (n = 7310), we used
the average date of clutch initiation for the earliest
1% of nests at each latitudinal interval to develop a
regression equation to calibrate date according to
latitude (Cooper et al. 2005b).
RESULTS
Percentage hatching failure
Of the 32 567 eggs laid in 7231 nests and reported
by 530 TBN participants, 2550 (7.83%) eggs failed
to hatch. Annual values (7 years) ranged between
6.8% and 8.9%.
Per-nest probability of hatching failure
Only 23% (1665 out of 7231) of clutches contained
one or more unhatched eggs. Nests that experienced
hatching failure contained a median of one unhatched
egg, but a mean of 1.53 ± 0.798 sd unhatched eggs
and as many as five (n = 2). Approximately one in
three nests at low latitudes (below 35°N) contained
at least one unhatched egg, compared with approximately one in five at higher latitudes. Few nests
contained unhatched eggs early in the season
(approximately one in six nests before day 25 of
the breeding season contained unhatched eggs,
compared with approximately one in four later in
the season).
Per-egg probability of hatching failure
In agreement with several predictions of the eggviability hypothesis and based on AICC values,
models A and B were best supported by the data
(Table 1). Model A received the highest support by
the data, with an Akaike weight of 96%, and, in contrast to the most specific predictions of the eggviability hypothesis, this model contained no clutch
size interactions. Yet, the best model did show variation in hatching failure with latitude, day of season
and clutch size, results consistent with the eggviability hypothesis. The next best model, model B,
did contain a latitude × clutch size interaction, but
Hatching failure in Eastern Bluebirds
225
Figure 2. (a) Predicted per-egg probability of hatching failure (blocks) for each clutch size with 95% confidence intervals (lines).
(b) Predicted per-egg probability of hatching failure (heavy line) across day of season with 95% confidence intervals (lighter lines).
(c) Predicted per-egg probability of hatching failure (heavy line) across latitude with 95% confidence intervals (lighter lines).
the confidence intervals associated with this interaction overlapped zero, which suggests that differences
in latitudinal gradients among clutch sizes were at
best weak or inconsistent. Using Akaike weights to
average the two best models, the data showed that,
as predicted, hatching failure increases later in the
season and decreases at higher latitude. The relationship between hatching failure and clutch size is
more complicated. As predicted, hatching failure
was more likely in clutches of six than in clutches of
four or five eggs. However, the hatching failure rate
in clutches of three eggs was as high as in clutches of
six (Fig. 2). The percentage per-egg probability of
hatching failure varied from 1.3% in the northern
part of the Eastern Bluebird range to 4.7% in the
southern part of its range (Fig. 2), a decrease of 2.13%
for every 10° increase in latitude between 30° and
46° (Koenig 1982 found a decrease of 1% for every
10° increase in latitude from the equator to 80°N).
Estimated averages from the best-supported model
showed failure rates of 1.9% at the start and 2.6% at
the end of a 140-day nesting season, about a 0.15%
increase per month (Fig. 2).
DISCUSSION
As hatching rate and clutch size are both reduced
later in the season and further south, large-scale
gradients in number of young produced in Eastern
Bluebird nests are generated by clutch size trends
and reinforced by parallel trends in hatching rates,
with more young produced per nest in more northern
locations and in earlier nests. There is an interesting
contrast in the magnitudes of seasonal and latitudinal variation in clutch size and in hatching rate. With
regard to clutch size, a greater proportion of variation is seasonal (> 1 egg) than is latitudinal (< 1 egg;
Dhondt et al. 2002). By contrast, latitudinal variation in hatching failure is far greater than seasonal
variation (a 3.4% change in probability of hatching
failure across the range of latitudes examined, compared with only a 0.7% change across the breeding
season). The concurrent patterns of a relatively small
latitudinal trend in clutch size and relatively large
latitudinal trend in hatching failure suggest that
birds at low latitudes lay clutches that are larger
than optimal. Conversely, the strong seasonal trend
in clutch size combined with low seasonal trend in
hatching failure rates suggests that birds laid clutches
that were closer to optimal throughout the season.
Taken together, this suggests a relatively strong past
evolutionary response to selection for seasonal declines
in clutch size, with ongoing directional selection for
latitudinal gradients in clutch size but with genetic
variation possibly being maintained by gene flow
(Dhondt et al. 1990, Rytkonen & Orell 2001, Kawecki
& Holt 2002, Postma & van Noordwijk 2005) in this
partial migrant.
Given higher ambient temperatures in summer
than spring and at southern than northern latitudes,
both seasonal and latitudinal trends in Eastern Bluebird hatching failure are consistent with the notion
that the loss of egg viability prior to the full onset of
incubation creates the observed gradients (Fig. 2 and
Table 2). By this scenario, loss of egg viability acts as
an overriding constraint on clutch size with diminishing importance at northern latitudes and increasing
importance as the season progresses.
Although our correlational evidence is consistent
with many aspects of the egg-viability hypothesis, it
does not validate all aspects of this hypothesis for
three reasons. First, for our results to be consistent
with the egg-viability hypothesis we need to assume
that females cannot prevent loss of egg viability by
the early onset of incubation. However, Cooper et al.
(2005a) reported latitudinal and seasonal gradients
in the duration of the apparent incubation period,
suggesting that female Bluebirds at southern latitudes
© 2006 The Authors
Journal compilation © 2006 British Ornithologists’ Union
226
C. B. Cooper et al.
Table 2. The hypothesized causes of hatching failure and predicted types of variation (indicated by ticks and crosses). Crosses indicate
that the predictions rely on more than one assumption and/or that the direction of the predicted effect cannot be clearly predicted.
Hypothesized
cause
Egg viability
Infertility
Female quality
Lethal heating
Genetic defects
Low calcium
Microbes
Season
Latitude
✓
✗
✓
✓
✓
✗
✗
✓
and late in the season initiate incubation before
clutch completion, perhaps to prevent loss of egg
viability. Although other explanations for the gradients in apparent incubation period are possible (see
Cooper et al. 2005a), it remains plausible that Eastern Bluebird females can, behaviourally, prevent at
least some loss of egg viability. Secondly, the best
model (Table 1) did not contain clutch size interactions with latitude and day of season and thus did
not support the most specific prediction of the eggviability hypothesis: that eggs in large clutches at
southern latitudes and late in the season should
experience the highest likelihood of hatching failure.
However, sample sizes may not have been sufficient
to support models with clutch size interactions.
Thirdly, egg viability alone cannot account for all
observed trends in hatching failure. The remaining
trend, elevated failure among three-egg clutches,
must be accounted for by other factors, such as
female condition (see below).
Given the numerous, and not mutually exclusive
causes of hatching failure (see Webb 1987 for
review; also Veiga 1992, Walsberg & Schmidt 1992,
Stoleson & Beissinger 1995, 1999, Graveland &
Drent 1997, Greene 1998, Cook et al. 2003), it is
plausible that large-scale patterns in hatching failure
were the result of multiple factors. Most studies of
hatching failure have been on a local scale and have
demonstrated how site-specific variation is related to
variables such as clutch size (e.g. Moreno et al. 1991),
age (Potti & Merino 1996) and sociability (Koenig
1982). However, even within a single site, Potti and
Merino (1996) suggested multiple causes of hatching failure in Pied Flycatchers Ficedula hypoleuca:
loss of egg viability accounted for the low hatchability of eggs laid early in the sequence, and reduced
fertility due to sperm depletion accounted for the
low hatchability of eggs laid late in the sequence. Thus,
© 2006 The Authors
Journal compilation © 2006 British Ornithologists’ Union
Variation
without time/
space trends
Clutch size
Small
Large
✓
✓
✓
✓
✓
✓
✗
neither the egg-viability hypothesis nor any other
single hypothesis should be viewed as providing the
full explanation for variation in hatching failure.
Nevertheless, although local studies have advanced
numerous explanations for hatching failure and
Koenig (1982) investigated the presence of largescale patterns across species, to date only the eggviability hypothesis makes specific a priori predictions
of large-scale patterns within a species. Although
many causes of hatching failure may vary among
populations (Table 2), below we review three causes
that are likely to vary among populations in ways
that correspond to gradients in latitude, season and
clutch size. As such, the following factors might
produce latitudinal, seasonal and clutch-size trends
in hatching failure (Table 2): (1) exposure to lethal
temperatures during embryo development, (2) effects
of female age and quality, and (3) infertility.
Lethal temperatures. After embryo growth has
begun, embryos can die from exposure to the excessive heat (over 40.5 °C) common at lower latitudes
and late in the season in temperate regions. Although
latitudinal and seasonal trends in hatching failure
seem consistent with embryo overheating, there are
several problems with this explanation. First, lethal
temperatures should affect all eggs in a clutch, irrespective of clutch size, but we observed an increase
in hatching failure among extremely small and
extremely large clutches. Secondly, females may
time incubation to occur before periods of excessive
lethal heat. As part of a different study (see Cooper
et al. 2005a), we recorded temperatures in over 300
nests of the three bluebird species across the United
States throughout the breeding seasons of 2002 and
2003. We found only one nestbox in which temperatures exceeded 40.5 °C while eggs were in the nest,
which suggests that most female Eastern Bluebirds
avoid incubating during periods of high heat. Thirdly,
Hatching failure in Eastern Bluebirds
227
Table 3. Direction of selection for each trait, with positive signs
indicating a higher value (e.g. higher survival in south) and
negative signs a lower value (e.g. smaller clutch sizes in south).
Latitude
Clutch size
Adult survival
Extra-pair copulations
Infertile eggs
Figure 3. Temperature recordings in a nestbox (black line) and
nest cup (grey line) of a Western Bluebird in California from 3
to 10 June 2003. Note that nest cup temperatures are not
necessarily equivalent to egg temperatures, but represent a
measure of the temperature near the eggs.
even if Eastern Bluebird eggs encounter periods of
lethal temperatures, females may maintain egg
temperatures below ambient, as can other species
nesting in extremely hot areas (e.g. shorebirds;
Downs & Ward 1997). Yet, like other passerines,
Eastern Bluebirds are intermittent incubators, leaving
the nest frequently to forage. Therefore, even if
Eastern Bluebirds can cool their eggs, there is still the
potential for eggs to experience temperatures over
lethal limits during the short periods when they are
unattended. The one nestbox that exceeded lethal
temperatures during incubation provided anecdotal
evidence that female Western Bluebirds Sialia mexicana can actively cool the clutch, because the nest
cup temperature remained below 40 °C each day
that the box temperature peaked (sometimes going
as high as 46 °C; Fig. 3), and the entire clutch hatched.
In the same way as loss of egg viability, lethal heating
is probably preventable and may vary with female
quality or experience. Thus, if lethal temperatures
play a role in the large-scale trends, female quality
must help shape these patterns.
Female quality/age/experience. Because low-quality
and inexperienced breeders have been found to breed
later in the season (Sæther 1990), a seasonal increase
in hatching failure, especially among small clutches,
could be explained by inexperienced or low-quality
breeders providing poor prenatal attendance and thus
failing to maintain optimal incubation temperatures
(Table 2). Furthermore, a seasonal deterioration in
food supply (Winkler & Allen 1996) is also consistent
with seasonal increases in hatching failure.
Female age (quality/experience) cannot easily
explain the strong latitudinal trends in hatching
North
South
+
–
+
–
–
+
–
+
failure. One possible, but tenuous, scenario is that
different migratory tendencies among Eastern Bluebird populations across latitude correspond to different overwinter survival, resulting in populations at
lower latitudes containing higher proportions of
younger individuals. Whether overwinter survival
of short-distance migrants differs from residents is
unknown, but should clearly be a research priority.
Counter to this explanation, the latitudinal pattern
of hatching failure appears to be a continuous linear
gradient rather than a step-function between the
migratory and non-migratory portions of the Eastern
Bluebird range.
Female age also cannot account for the elevated
hatching failure among six-egg clutches. The hypothesis
that the combination of female age and egg viability
accounts for observed patterns of hatching failure is
plausible, and perhaps the most parsimonious combination of causes.
Infertility. A seasonal increase in infertile eggs
could be attributed to a seasonal decline in sperm
production (Lombardo et al. 2002). Currently, there
are no data demonstrating a latitudinal gradient in
sperm depletion or infertility. Predicting large-scale
trends in hatching failure from infertile eggs is tenuous
and dependent on several assumptions, each the
subject of much independent investigation and
thus worth exploring together (Table 3). The first
assumption is that age-specific mortality can drive the
evolution of life-history strategies (Michod 1979,
Reznick et al. 1990) where higher adult survival is
coupled with allocation of greater investment to
fewer young (Ricklefs 1997, Ghalambor & Martin
2001). Thus, we can assume a latitudinal gradient in
adult survival inverse to clutch size trends; note the
direction of the adult survival gradient has been
debated (Karr et al. 1990, Johnston et al. 1997, Martin
et al. 2000). The second assumption is that because
extra-pair copulations (EPCs) garner costs to future
survival and reproduction (Wink & Dyrcz 1999,
© 2006 The Authors
Journal compilation © 2006 British Ornithologists’ Union
228
C. B. Cooper et al.
Westneat & Rambo 2000), then EPCs can be viewed
as a form of investment in current reproduction with
prevalence inverse to adult survival (Lloyd & Martin
2003). The third assumption is that there is a link
between the frequency of EPCs and infertile eggs
(Zeh & Zeh 1996, 1997) such that higher hatching
success is associated with more EPCs (see Table 3).
Females may benefit from engaging in EPCs in order
to avoid temporary sperm depletion of males (Sheldon
1994, Birkhead et al. 1995, Gray 1997). The direction
of the link between EPCs and infertility has been
debated and remains equivocal (Kempenaers et al.
1996, Cordero et al. 1999, Lubjuhn et al. 2000,
Whitekiller et al. 2000). Koenig (1982) found hatching failure increased as the complexity of the social
structure increased and suggested other relationships between social factors and hatchability, such as
mate competition, egg neglect and behavioural
synchronization. Rather than explain hatching failure trends, the hatching failure patterns can inform
predictions about latitudinal gradients in mating
behaviour. For example, given higher adult survival
rates at lower latitudes (in correlation with smaller
clutches) combined with elevated hatching failure at
lower latitudes, we would expect lower frequencies
of EPCs at low latitudes if infertility and EPCs are
inversely related.
Other factors. There are many factors that cause
hatching failure but for which we have even less
information about latitudinal and seasonal patterns.
Microbes can play an important role in hatching
failure and may vary with climate variables, such as
temperature and humidity, along seasonal and latitudinal gradients (Cook et al. 2003). We expect other
causes of hatching failure such as genetic defects and
insufficient calcium to vary among populations relative to such factors as previous bottlenecks (Briskie
& Mackintosh 2004), inbreeding (Kempenaers et al.
1996), acid rain (Graveland & Drent 1997, Greene
1998) and humidity (Bruce & Drysdale 1994,
Baggott & Graeme-Cook 2002). Although these
factors exhibit geographical variation, we do not expect
them to exhibit a specific latitudinal or seasonal
gradient (Table 2).
CONCLUSIONS
Our results provide support for the hypothesis that
high ambient temperatures common at lower latitudes and late in the breeding season reduce egg
viability. Because loss of egg viability can account for
all patterns except elevated hatching failure among
© 2006 The Authors
Journal compilation © 2006 British Ornithologists’ Union
three-egg clutches, at least two mechanisms are
required to explain all observed patterns. Effects
of female age/experience/quality are the most likely
factors to explain hatching failure among small
clutches. Other causes of hatching failure could contribute to observed latitudinal and seasonal gradients
in egg failure. Integrating the link between EPCs and
infertility into life-history trade-offs suggests patterns
consistent with our results and warrants more attention.
Because our results are correlational, we cannot
address the issue of clutch-size constraints directly.
Yet, as suggested by Stoleson and Beissinger (1999),
it is plausible that the loss of egg viability due to
temperature conditions that predominate in the south
and late in the season affects hatching failure and
constrains clutch size. Although research has addressed
the adaptive significance of nestling mortality on
fecundity (Slagsvold et al. 1995, Slagsvold 1986,
Moreno 1989, Vi!uela 2000), less is known about
hatching failure. Given that trends in hatching failure reinforce trends in clutch size, future large-scale
investigations into all the causes of hatching failure
are needed.
Implication for large-scale research
Researchers have traditionally focused on one primary selection pressure (e.g. predation or food
supply) as an agent in the life-history trade-offs that
produce consistent large-scale patterns in life-history
traits, such as small clutches in the tropics (Moreau
1944, Lack 1947, Skutch 1949, Ashmole 1963,
Cody 1966, Royama 1969, Ricklefs 1980). Underlying this work is the assumption that a single selection
pressure arises from a single ecological phenomenon
varying along a large-scale gradient. Alternatively,
multiple selection pressures (e.g. predation rates and
food supply) may create the same evolutionary
response because the adaptive responses are constrained by incompatible physiological states
(Ricklefs & Wikelski 2002). Another related and
relatively ignored possibility is that several selection
pressures (or constraints) could create a geographical
gradient in the life-history response if the relative
importance of each selection pressure (or constraint)
varies systematically across a species’ range (Cooper
et al. 2005a). Thus, there are several ways in which
large-scale trends in hatching failure may be relevant
to our understanding of the evolution of clutch size
variation. Here we have demonstrated how the causes
of hatching failure may combine to create trends
across latitude and season within a single species.
Hatching failure in Eastern Bluebirds
We are grateful to the volunteers who contribute data to
The Birdhouse Network; obviously this research could
not have been done without them. We thank T. Kast,
C. Delong, P. Allen and P. Senesac for co-ordinating The
Birdhouse Network. NSF ISE-96277280 as well as financial
support from project participants and members of the
Cornell Laboratory of Ornithology funded The Birdhouse
Network.
REFERENCES
Anderson, D.R., Burnham, K.P. & Thomson, W.L. 2000. Null
hypothesis testing: problems, prevalence, and an alternative.
J. Wildlife Manage. 64: 912 – 921.
Ashmole, N.P. 1963. The regulation of numbers of tropical
oceanic birds. Ibis 103: 458 – 473.
Baggott, G.K. & Graeme-Cook, K. 2002. Microbiology of
natural incubation. In Deeming, D.C. (ed.) Avian Incubation:
Behaviour, Environment, and Evolution: 179 – 191. Oxford:
Oxford University Press.
Birkhead, T.R., Veiga, J.P. & Fletcher, F. 1995. Sperm competition and unhatched eggs in the House Sparrow. J. Avian
Biol. 26: 343–345.
Briskie, J.V. & Mackintosh, M. 2004. Hatching failure increases
with severity of population bottlenecks in birds. Proc. Natl
Acad. Sci. 101: 558 – 561.
Bruce, J. & Drysdale, E.M. 1994. Trans-shell transmission. In
Board, R.G. & Fuller, R. (eds) Microbiology of the Avian Egg:
63–91. London: Chapman & Hall.
Buckland, S.T., Burnham, K.P. & Augustin, N.H. 1997. Model
selection: an integrated part of inference. Biometrics 53:
603–618.
Cody, M.L. 1966. A general theory of clutch size. Evolution 20:
174–184.
Cook, M.I., Beissinger, S.R., Toranzos, G.A., Rodriguez, R.A.
& Arendt, W.J. 2003. Trans-shell infection by pathogenic
micro-organisms reduces the shelf life of non-incubated
bird’s eggs: a constraint on the onset of incubation? Proc. R.
Soc. Lond. B 270: 2233 – 2240.
Cooper, C.B., Hochachka, W.M., Butcher, G. & Dhondt, A.A.
2005a. Seasonal and latitudinal trends in clutch size: constraints and adaptations to temperature during laying and
incubation. Ecology 86: 2018– 2031.
Cooper, C.B., Hochachka, W.M. & Dhondt, A.A. 2005b. Latitudinal trends in within-year reoccupation of nest boxes and
their implications. J. Avian Biol. 36: 31– 39.
Cordero, P.J., Wetton, J.H. & Parkin, D.T. 1999. Within-clutch
patterns of egg viability and paternity in the House Sparrow.
J. Avian Biol. 30: 103 –107.
Deeming, D.C. 2002. Behavior patterns during incubation. In
Deeming, D.C. (ed.) Avian Incubation: Behaviour, Environment, and Evolution: 63– 87. Oxford: Oxford University
Press.
Deeming, D.C. & Ferguson, M.W.J. 1992. Physiological effects
of incubation temperature on embryonic development in
reptiles and birds. In Deeming, D.C. & Ferguson, M.W.J. (eds)
Egg Incubation: its Effects on Embryonic Development in
Birds and Reptiles: 147– 171. Cambridge: Cambridge University Press.
Dhondt, A.A. 1968. Langdurige legonderbreking van de leg bij
de Pimpelmees Parus caeruleus. Giervalk 58: 396.
229
Dhondt, A.A., Adriaensen, F., Matthysen, E. & Kempenaers, B.
1990. Nonadaptive clutch sizes in tits. Nature 348: 723 –725.
Dhondt, A.A., Kast, T.L. & Allen, P.E. 2002. Geographical
differences in seasonal clutch size variation in multi-brooded
bird species. Ibis 144: 646 – 651.
Downs, C.T. & Ward, D. 1997. Does shading behavior of
incubating shorebirds in hot environments cool the eggs
or the adults? Auk 114: 717 –724.
Ewert, M.A. 1992. Cold torpor, diapause, delayed hatching
and aestivation in reptiles and birds. In Deeming, D.C. &
Ferguson, M.W.J. (eds) Egg Incubation: its Effects on
Embryonic Development in Birds and Reptiles: 173 –192.
Cambridge: Cambridge University Press.
Ghalambor, C.K. & Martin, T.E. 2001. Fecundity–survival
trade-offs and parental risk-taking in birds. Science 292:
494 – 497.
Gowaty, P.A. & Plissner, J.H. 2000. Eastern Bluebirds (Sialia
sialis). In Poole, A. & Gill, F. (eds) The Birds of North America
No. 381. Philadelphia: The Academy of Natural Sciences;
Washington, DC: The American Ornithologists’ Union;
Ithaca, NY: Cornell Laboratory of Ornithology.
Graveland, J. & Drent, R.H. 1997. Calcium availability limits
breeding success of passerines on poor soils. J. Anim. Ecol.
66: 279 – 288.
Gray, E. 1997. Do female Red-winged Blackbirds benefit genetically from seeking extra-pair copulations? Anim. Behav. 53:
605 – 623.
Greene, R.E. 1998. Long-term decline in the thickness of
eggshells of thrushes, Turdus spp. in Britain. Proc. R. Soc.
Lond. B 265: 679 – 684.
Haywood, S. 1993a. Role of extrinsic factors in the control of
clutch size in Blue Tit Parus caeruleus. Ibis 135: 79 – 84.
Haywood, S. 1993b. Sensory control of clutch size in the Zebra
Finch (Taeniopygia guttata). Auk 110: 778 –786.
Johnson, D. 1999. The insignificance of statistical significance
testing. J. Wildlife Manage. 63: 763 – 772.
Johnston, J.P., Peach, W.J., Gregory, R.D. & White, S.A.
1997. Survival rates of tropical and temperate passerines: a
Trinidadian perspective. Am. Nat. 150: 771 –789.
Karr, J.R., Nichols, J.D., Klimkiewicz, M.K. & Brawn, J.D.
1990. Survival rates of birds of tropical and temperate
forests: will the dogma survive? Am. Nat. 136: 277 –291.
Kawecki, T.J. & Holt, R.D. 2002. Evolutionary consequences of
asymmetric dispersal rates. Am. Nat. 160: 333 – 347.
Kempenaers, B., Adriaensen, F., Van Noordwijk, A.J. &
Dhondt, A.A. 1996. Genetic similarity, inbreeding and hatching failure in Blue Tits: are unhatched eggs infertile? Proc. R.
Soc. Lond. B 263: 179 –185.
Koenig, W.D. 1982. Ecological and social factors affecting
hatchability of eggs. Auk 99: 526 – 536.
Lack, D. 1947. The significance of clutch size. Ibis 89: 302 –352.
Littell, R.C., Milliken, G.A., Stroup, W.W. & Wolfinger, R.D.
1996. SAS System for Mixed Models. North Carolina: SAS
Institute Inc.
Lloyd, J.D. & Martin, T.E. 2003. Sibling competition and the
evolution of prenatal development rates. Proc. R. Soc. Lond.
B 270: 735 –740.
Lombardo, M.P., Forman, A.N., Czarnowski, M.R. & Thorpe, P.A.
2002. Individual, temporal, and seasonal variation in sperm
concentration in Tree Swallows. Condor 104: 803 – 810.
Lubjuhn, T., Winkel, W., Epplen, J.T. & Bruen, J. 2000. Reproductive success of monogamous and polygynous Pied
© 2006 The Authors
Journal compilation © 2006 British Ornithologists’ Union
230
C. B. Cooper et al.
Flycatchers (Ficedula hypoleuca). Behav. Ecol. Sociobiol.
48: 12–17.
Martin, T.E., Martin, P.R., Olson, C.R., Heidinger, B.J. &
Fontaine, J.J. 2000. Parental care and clutch size in North
and South American birds. Science 287: 1482 –1485.
Michod, R.E. 1979. Evolution of life histories in response to
age-specific mortality factors. Am. Nat. 113: 531– 550.
Moreau, R.E. 1944. Clutch size: a comparative study, with
reference to African birds. Ibis 86: 286 – 347.
Moreno, J. 1989. Energetic constraints on uniparental incubation
in the Wheatear Oenanthe oenanthe L. Ardea 77: 107 –115.
Moreno, J., Gustafsson, L., Carlson, A. & Pärt, T. 1991. The
costs of incubation in relation to clutch size in the Collared
Flycatcher Ficedula albicollis. Ibis 133: 186 –193.
Peakall, D.B. 1970. The Eastern Bluebird: its breeding season,
clutch size, and nesting success. Living Bird 9: 239 – 256.
Postma, E. & van Noordwijk, A.J. 2005. Gene flow maintains
a large genetic difference in clutch size at a small spatial
scale. Nature 433: 65– 68.
Potti, J. & Merino, S. 1996. Causes of hatching failure in the
Pied Flycatcher. Condor 98: 328 – 336.
Reznick, D.A., Bryga, H. & Endler, J.A. 1990. Experimentally
induced life-history evolution in a natural population. Nature
346: 357–359.
Ricklefs, R.E. 1980. Geographic variation in clutch size among
passerine birds: Ashmole’s hypothesis. Auk 97: 38– 49.
Ricklefs, R.E. 1997. Comparative demography of New World
populations of thrushes (Turdus spp.). Ecol. Monogr. 67: 23 – 43.
Ricklefs, R.E. & Wikelski, M. 2002. The physiology/life-history
nexus. Trends Ecol. Evol. 17: 462 – 468.
Royama, T. 1969. A model for global variation of clutch size in
birds. Oikos 20: 562 – 567.
Rytkonen, S. & Orell, M. 2001. Great Tits, Parus major, lay too
many eggs: experimental evidence in mid-boreal habitats.
Oikos 93: 439–450.
Sæther, B.-E. 1990. Age-specific variation in reproductive performance of birds. In Power, D.M. (ed.) Current Ornithology
7 : 251–283. New York: Plenum Press.
Sheldon, B.C. 1994. Male phenotype, fertility, and the pursuit of
extra-pair copulations by female birds. Proc. R. Soc. Lond. B
257: 25–30.
Skutch, A.F. 1949. Do tropical birds rear as many young as they
can nourish? Ibis 91: 430 – 455.
Slagsvold, T. 1986. Asynchronous versus synchronous hatching in birds: experiments with the pied flycatcher. J. Anim.
Ecol. 55: 1115–1134.
Slagsvold, T., Amundsen, T. & Dale, S. 1995. Costs and
© 2006 The Authors
Journal compilation © 2006 British Ornithologists’ Union
benefits of hatching asynchrony in blue tits Parus caeruleus.
J. Anim. Ecol. 64: 563 – 578.
Stoleson, S.H. & Beissinger, S.R. 1995. Hatching asynchrony
and the onset of incubation in birds, revisited. When is the
Critical Period? In Power, D.M. (ed.) Current Ornithology 12:
191–270. New York: Plenum Press.
Stoleson, S.H. & Beissinger, S.R. 1999. Egg viability as a constraint on hatching synchrony at high ambient temperatures.
J. Anim. Ecol. 68: 951– 962.
Veiga, J.P. 1992. Hatching asynchrony in the House Sparrow: a
test of the egg viability hypothesis. Am. Nat. 139: 669 –675.
ViNuela, J. 2000. Opposing selective pressures on hatching
asynchrony: egg viability, brood reduction, and nestling
growth. Behav. Ecol. Sociobiol. 48: 333 – 343.
Walsberg, G.E. & Schmidt, C.A. 1992. Effects of variable
humidity on embryonic development and hatching success
of Mourning Doves. Auk 109: 309 – 314.
Webb, D.R. 1987. Thermal tolerance of avian embryos: a
review. Condor 89: 874 – 898.
Westneat, D.F. & Rambo, T.B. 2000. Copulation exposes
female Red-winged Blackbirds to bacteria in male semen.
J. Avian Biol. 31: 1–7.
White, F.N. & Kinney, J.L. 1974. Avian incubation: interactions
among behavior, environment, nest and eggs in regulation of
egg temperature. Science 189: 107 –115.
Whitekiller, R.R., Westneat, D.F., Schwagmeyer, P.L. &
Mock, D.W. 2000. Badge size and extra-pair fertilizations
in the House Sparrow. Condor 102: 342 – 348.
Wink, M. & Dyrcz, A. 1999. Mating systems in birds: a review of
molecular studies. Acta Ornithol. 34: 91 –109.
Winkler, D.W. & Allen, P.E. 1996. The seasonal decline in avian
clutch size: strategy or physiological constraints? Ecology
77: 922 – 932.
Zar, J.H. 1984. Biostatistical Analysis, 2nd edn. New Jersey:
Prentice Hall.
Zeh, J.A. & Zeh, D.W. 1996. The evolution of polyandry I:
intragenomic conflict and genetic incompatibility. Proc. R.
Soc. Lond. B 263: 1711 –1717.
Zeh, J.A. & Zeh, D.W. 1997. The evolution of polyandry II: postcopulatory defenses against genetic incompatibility. Proc. R.
Soc. Lond. B 264: 69 –75.
Received 21 February 2005;
revision accepted 22 September 2005;
published OnlineEarly 18 January 2006;
doi: 10.1111/j.1474-919x.2006.00500.x