All together now! Hatching synchrony in freshwater turtles

Abstract

Synchronous hatching is widespread among oviparous taxa. However, the adaptive significance of this phenomenon is unclear, as are the cues mediating hatching synchrony. We took a comparative approach and experimentally tested for synchronous hatching in two freshwater turtles with different life histories (Apalone spinifera and Graptemys geographica). We also aimed to disentangle the cues and mechanisms facilitating any synchronisation and explored its potential costs. For each species, we incubated eggs of different ages and from different mothers in two conditions—in direct contact with other eggs, or physically isolated but able to receive acoustic or chemical cues. We found evidence of hatching synchrony in A. spinifera, but not in G. geographica. Apalone spinifera eggs of different ages that developed in-contact hatched together, implicating mechanical cues. Younger eggs that were incubated in isolation also synchronised with in-contact eggs, which also implicates acoustic or chemosensory cues. Hatchling yolk sac size and mass were similar among treatments. Overall, A. spinifera exhibits hatching synchrony and there was no evidence of developmental costs. The lack of hatching synchrony in G. geographica may reflect their different life-history strategy, as this species can overwinter in the nest after hatching. Clutch effects explained a large and significant proportion of variance in both hatching date and incubation duration in both turtle species. Future research on hatching synchrony should control for these potentially confounding clutch effects. Variation in hatching or emergence synchrony among freshwater turtles raises questions about the selective forces that favour evolution of this behaviour.

Significance statement
Many freshwater turtles hatch from their eggs together—but how do they coordinate? We identified that if eggs of Spiny Softshell turtles develop in contact with each other, the eggs hatch together. But, Northern Map turtles do not synchronise hatching, possibly because of an important difference in their ecology—Northern Map turtles overwinter within their nest. Variable occurrence of hatching synchrony across turtle species may reflect different egg structures and/or overwintering behaviour. Research testing for the presence of hatching synchrony typically incubates less and more advanced siblings together. But we found that clutch effects explained a large and significant proportion of variance in both hatching date and incubation duration, which highlights the need to consider and control for these effects moving forward. Our study provides a promising experimental framework to study hatching synchrony that controls for clutch effects.

Full text

ORIGINAL ARTICLE
All together now! Hatching synchrony in freshwater turtles
Julia L. Riley
1
&Sean Hudson
2,3
&Coral Frenette-Ling
2
&Christina M. Davy
3,4
Received: 9 August 2019 / Revised: 30 December 2019 / Accepted: 6 January 2020
#Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract
Synchronous hatching is widespread among oviparous taxa. However, the adaptive significance of this phenomenon is unclear,
as are the cues mediating hatching synchrony. We took a comparative approach and experimentally tested for synchronous
hatching in two freshwater turtles with different life histories (Apalone spinifera and Graptemys geographica). We also aimed to
disentangle the cues and mechanisms facilitating any synchronisation and explored its potential costs. For each species, we
incubated eggs of different ages and from different mothers in two conditions—in direct contact with other eggs, or physically
isolated but able to receive acoustic or chemical cues. We found evidence of hatching synchrony in A. spinifera, but not in
G. geographica.Apalone spinifera eggs of different ages that developed in-contact hatched together, implicating mechanical
cues. Younger eggs that were incubated in isolation also synchronised with in-contact eggs, which also implicates acoustic or
chemosensory cues. Hatchling yolk sac size and mass were similar among treatments. Overall, A. spinifera exhibits hatching
synchrony and there was no evidence of developmental costs. The lack of hatching synchrony in G. geographica may reflect their
different life-history strategy, as this species can overwinter in the nest after hatching. Clutch effects explained a large and
significant proportion of variance in both hatching date and incubation duration in both turtle species. Future research on hatching
synchrony should control for these potentially confounding clutch effects. Variation in hatching or emergence synchrony among
freshwater turtles raises questions about the selective forces that favour evolution of this behaviour.
Significance statement
Many freshwater turtles hatch from their eggs together—but how do they coordinate? We identified that if eggs of Spiny Softshell
turtles develop in contact with each other, the eggs hatch together. But, Northern Map turtles do not synchronise hatching,
possibly because of an important difference in their ecology—Northern Map turtles overwinter within their nest. Variable
occurrence of hatching synchrony across turtle species may reflect different egg structures and/or overwintering behaviour.
Research testing for the presence of hatching synchrony typically incubates less and more advanced siblings together. But we
found that clutch effects explained a large and significant proportion of variance in both hatching date and incubation duration,
which highlights the need to consider and control for these effects moving forward. Our study provides a promising experimental
framework to study hatching synchrony that controls for clutch effects.
Keywords Apalone spinifera .Clutch effects .Developmental plasticity .Embryonic communication .Graptemys geographica .
Parental effects
Communicated by T. Madsen
*Christina M. Davy
christina.davy@ontario.ca
1
Department of Botany and Zoology, Stellenbosch University,
Stellenbosch, Western Cape 7600, South Africa
2
Wildlife Preservation Canada, RR#5, 5420 Highway 6 North,
Guelph, ON N1H 6J2, Canada
3
Environmental and Life Sciences Graduate Program, Trent
University, 2140 East Bank Drive, Peterborough, ON K9J 7B8,
Canada
4
Wildlife Research and Monitoring Section, Ontario Ministry of
Natural Resources and Forestry, Peterborough, ON, Canada
Behavioral Ecology and Sociobiology (2020) 74:58
https://doi.org/10.1007/s00265-020-2800-y

Introduction
Many oviparous taxa exhibit some degree of hatching syn-
chrony among embryos within a clutch and/or brood. In spe-
cies with little or no parental care the timing and coordination
of hatching can have a substantial effect on the survival of
offspring (Pearson and Warner 2018). Thus, hatching syn-
chrony has many potential benefits. It may facilitate more
efficient emergence from the nest through communal digging
(Ims 1990;Tuckeretal.2008) and mass migration post-
emergence (Carr and Hirth 1961; Spencer et al. 2001;
Warkentin and Caldwell 2009). Additionally, it may reduce
the likelihood of encountering predators through predator
swamping, reduce the probability of detection by predators
during emergence, and dilute predation risk across individuals
(Arnold and Wassersug, 1978; Santos et al. 2016). Yet, there
are also costs associated with the mechanisms that facilitate
hatching synchrony, and the trade-offs between the benefits
and costs of hatching synchrony are not well understood.
Apparent hatching synchrony results from constrained in-
cubation durations that are determined by parental effects (Ims
1990;Aubretetal.2016). In contrast, true hatching synchrony
occurs when embryos coordinate hatching events (Ims 1990;
Aubret et al. 2016), which can be facilitated through both
environmental and physiological mechanisms. For example,
fully developed embryos may delay hatching until stimulated
by an environmental cue (Doody 2011; Warkentin 2011;
Doody et al. 2012). Pig-Nosed Turtles (Carretochelys
insculpta) hatch in response to hypoxia in the laboratory
(Doody 2011), reflecting hatching and emergence behaviour
of natural nests in response to wet season flooding events
(Webb et al. 1986). Alternatively, embryos may adjust their
rate of development. Less developed (i.e. younger) embryos
may increase their developmental rate in order to hatch at the
same time as more developed embryos (i.e. the ‘catch-up’
hypothesis; Spencer et al. 2001; McGlashan et al. 2011), or
less developed eggs may simply hatch at an earlier develop-
mental stage (Spencer and Janzen 2011; McGlashan et al.
2018). More developed (i.e. older) embryos may also delay
hatching to synchronise with younger embryos (i.e. the ‘wait’
hypothesis; Spencer et al. 2001; Colbert et al. 2010). The
physiological mechanisms facilitating hatching synchrony
could be costly, particularly if altered development rates affect
individual fitness. For example, embryos that increase their
developmental rates to ‘catch-up’with older embryos can ex-
hibit traits of incomplete development, including reduced neu-
romuscular capabilities and a greater amount of unabsorbed
yolk resources (Schwagmeyer et al. 1991; Spencer et al. 2001;
Colbert et al. 2010; McGlashan et al. 2011).
The cues that mediate hatching synchrony are also not well
known. One potential driver of hatching synchrony is
embryo–embryo communication, which can include physical,
acoustic, or chemical signalling (Vergne and Mathevon 2008;
Doody 2011; Warkentin 2011; Doody et al. 2012; Ferrara
et al. 2014a,b;McKenna2016). Eggs of the Eastern Long-
Necked Turtle (Chelodina longicollis) incubated in contact
with more advanced clutch-mates synchronised hatching by
increasing their embryonic heart rate, metabolic rates, thus
coordinating their developmental rates (McGlashan et al.
2011). Eggs of C. insculpta exposed to vibrations on an elec-
tronic shaker hatched several days earlier than undisturbed
eggs (Doody et al. 2012), and eggs of the Brown
Marmorated Stink Bug (Halyomorpha halys) hatch in re-
sponse to the vibration from siblings cracking open their egg-
shells (Endo et al. 2019;Hill2019). Embryos and hatchlings
of some crocodilians, turtles and birds exhibit intra-specific
vocalisations close to hatching, and these vocalisations could
be used to synchronise hatching (Orcutt 1974; Vergne and
Mathevon 2008; Ferrara et al. 2014a,b; McKenna 2016).
Additionally, chemosensory cues can be important in
embryo–embryo communication in many taxa (Pianka and
Vitt 2003; Warkentin 2011). It is clear that mechanical, audi-
tory, and chemosensory cues all have the potential to facilitate
hatching synchrony, but the relative roles they play in this
process are unclear.
Most experimental studies that have tested potential
mechanisms for hatching synchrony in turtles have incu-
bated early-stage eggs at different temperatures for the first
part of incubation, thus inducing developmental asynchro-
ny between individuals of the same clutch (Table 1). This
experimental design effectively tests embryos’ability to
alter developmental rates within a clutch, but does not ac-
count for potentially confounding clutch effects that can
also cause apparent hatching synchrony (Ims 1990;
Steyermark and Spotila 2001). Developmental time (incu-
bation or gestation duration) may be affected by parental
genetics and phenotype (clutch or litter effects), as are
many other traits (Steyermark and Spotila 2001;Webb
et al. 2001;Nobleetal.2014; Baxter-Gilbert et al. 2018).
Parental effects on developmental time are reported in
Couch’s Spadefoot Toads (Scaphiopus couchii;Newman
1988) and European Sand Lizards (Lacerta agilis;Olsson
et al. 1996). The role clutch effects have in hatching syn-
chrony differentiates between apparent and true hatching
synchrony (Ims 1990). Thus, clutch effects should be ex-
plicitly considered, and controlled for, in experimental de-
signs that test for true hatching synchrony, and this can be
accomplished creating developmental asymmetry by co-
incubating eggs from older and younger clutches with dif-
ferent mothers (Fig. 1). This approach was used to demon-
strate true hatching synchrony in larval rhinoceros beetles
(Trypoxylus dichotomus;Kojima2015) and embryonic wa-
ter snakes (Natrix maura; Aubret et al. 2016), where youn-
ger eggs developing in contact with older embryos from
different clutches increased their developmental rate to
achieve synchronous hatching.
58 Page 2 of 15 Behav Ecol Sociobiol (2020) 74:58

We experimentally tested for true hatching synchrony, the
potential cues and mechanisms that facilitate it, and its poten-
tial fitness costs in two freshwater turtles with differing life
histories and evolutionary trajectories: the Spiny Softshell
(Apalone spinifera) and the Northern Map Turtle
(Graptemys geographica).Our first aim was to test whether
these turtles exhibit hatching synchrony by incubating older
and younger embryos from different clutches together within
each experimental container (Fig. 1), and testing whether co-
incubating eggs hatched together within a 24-h period (i.e. this
study’s definition of hatching synchrony). If hatching syn-
chrony was not present, then we predicted that older eggs
would hatch before younger eggs (Fig. 2a). Our second aim
was to identify the relative roles of mechanical, acoustic, and/
or chemosensory cues in facilitating hatching synchrony.
Within each experimental container, some eggs were
incubated in contact with one another, and others were phys-
ically isolated in single-egg compartments where they could
still receive acoustic or chemosensory cues (Fig. 1). We
hypothesised that if mechanical cues facilitate hatching syn-
chrony, then the eggs incubated in contact with each other,
regardless of their age, would hatch at the same time, but the
younger and older isolated eggs would hatch at different times
(Fig. 2c). Our experimental design also allowed us to test the
hypothesis that non-contact cues (auditory and/or
chemosensory cues) mediate hatching synchrony. In this case,
we expected that younger and older eggs would hatch at the
same time, regardless of incubation in direct contact or isola-
tion (Fig. 2b). If both contact and non-contact cues were in-
volved, we predicted that the results would reflect a combina-
tion of the scenarios described. The third aim of our study was
to explore potential physiological mechanisms of hatching
synchrony, if present, by comparing hatching date and incu-
bation duration among treatments. The fourth study aim was
to compare hatchling mass and yolk sac size among treat-
ments to explore potential fitness costs of hatching synchrony.
Materials and methods
Study species
Our study focused on two species of freshwater turtles for
which hatching synchrony has not been examined. Apalone
spinifera and G. geographica are phylogenetically distinct
(Shaffer et al. 1997; Krenz et al. 2005). They have wide,
largely sympatric distributions throughout eastern North
America (Ernst and Lovich 2009). Both species typically in-
habit large water bodies, like lakes and rivers, and are primar-
ily carnivorous but eat a wide variety of prey species (Ernst
and Lovich 2009). Both species exhibit sexual dimorphism in
size (females are larger than males; Gibbons and Lovich 1990;
Ernst and Lovich 2009), and sexes of A. spinifera exhibit
sexually dimorphic patterns on the carapace (Greenbaum
Fig. 1 Design of the experimental tubs showing older (dark blue) and
younger (light blue) eggs incubating in contact (6 eggs touching in the
central compartment) and no-contact conditions. The semi-transparent,
plastic experimental tub (22.5 cm × 13.48 cm × 5.5 cm) was separated
into compartments using Foamular™that was 1.27 cm thick
(represented by thick, grey lines), and the eggs were placed on a
substrate of wet vermiculite
Table 1 Summary of previous studies exploring mechanisms of hatching synchrony in turtles
Species Common name Experimental manipulation Proposed mechanism
of synchrony
Source
Emydura macquarii Murray River Turtle Contact/no-contact with eggs containing more
advanced siblings
Egg–egg contact Spencer et al. 2001
Chrysemys picta Painted Turtle Contact/no-contact with eggs containing more
advanced siblings
Egg–egg contact Colbert et al. 2010
Emydura macquarii Murray River Turtle Contact/no-contact with eggs containing more
advanced siblings
Egg-egg contact McGlashan et al. 2011
Carettochelys insculpta Pig-Nosed Turtle Vibrations; contact/no-contact eggs exposed to
nitrogen gas and water immersion
Vibrations Doody et al. 2012
Dermochelys coriacea Leatherback Sea Turtle Observational study; no mechanism tested Vocalisations Ferrara et al. 2014a
Chelodina longicollis Long-necked Turtle Contact/no-contact with eggs containing more
advanced siblings
Heart rate McGlashan et al. 2015
Behav Ecol Sociobiol (2020) 74:58 Page 3 of 15 58

and Carr 2001). Mating and courtship in both turtles occur in
the autumn and spring, and individuals will nest from late
May to early July depending on their geographic location
(Ernst and Lovich 2009). Clutch sizes of G. geographica vary
from 9 to 20 eggs, and 1 to 2 clutches of soft, ovoid eggs are
deposited annually per individual (Ernst and Lovich 2009).
Clutches of A. spinifera vary between 8 and 39 circular, hard
eggs, and individuals can also lay up to 2 clutches within a
year (Ernst and Lovich 2009).
In general, for North American freshwater turtles, hatch-
ing occurs from late August to October (Ernst and Lovich
2009), but incubation duration of each nest varies depend-
ing on temperature, and thus geographic location (Noble
et al. 2018). Hatching dates can be monitored in laboratory
settings, or in wild nests by placing motion-sensitive de-
vices into nest chambers (Rollinson et al. 2019), but in
most field studies the time from nesting to hatching is not
directly observed. Instead, field studies typically quantify
the date of hatchling emergence from the nest. The time of
emergence from the nest is a reasonable within-study
proxy for incubation duration but differs from the time of
hatching, because hatchlings often remain in the nest for
hours to weeks post-hatching (Diamondback Terrapins,
Malaclemys terrapin;Burger1976,andLoggerheadSea
Turtles, Caretta caretta; Christens 1990). Some species
can even overwinter in the nest and delay emergence until
the spring (Gibbons 2013;Lovichetal.2014).
In the wild, A. spinifera hatchlings typically emerge from
their nests in the late summer and early autumn (Ernst and
Lovich 2009; Baker et al. 2013), with a range of 58–77 days
between oviposition and hatchling emergence (Ernst and
Lovich 2009; Tornabene et al. 2018). Apalone spinifera over-
winters aquatically and must move from the nest to a suitable
overwintering location before temperatures decline below
freezing (Ultsch 2006). Emergence can be synchronised (i.e.
occurs over a 24-h period) within clutches. Baker et al. (2013)
found hatchlings from two A. spinifera nests emerged within
the same day. In the closely related Smooth Softshell
(A. mutica), 92% of 26 nests emerged within 1 day with an
average within-clutch emergence period of 46 min (Plummer
2007). Although nest emergence is synchronised in this spe-
cies, it is unknown if hatching is also synchronous.
Graptemys geographica eggs incubated at a constant 29 °C
in the laboratory take an average of 55 days to hatch (Banger
et al. 2013). In the wild, G. geographica emerges mainly in
the spring after overwintering within their nest (Christiansen
and Gallaway 1984; Baker et al. 2003,2013; Nagle et al.
2004). Graptemys geographica hatchlings likely survive
freezing temperatures using the physiological mechanism of
supercooling because their tolerance to freezing is poor (Baker
(a) There is no hatching
synchrony (the null
hypothesis).
(b) Chemosensory and
auditory cues
facilitate hatching
synchrony.
(c) Mechanical cues
facilitate hatching
synchrony.
Hypothesis Time 1 Time 2 Time 3
Fig. 2 Our experimental design (see Fig. 1) allows us to test multiple
hypotheses about hatching synchrony and the cues facilitating it. If
hatching synchrony is not present (a; our null hypothesis), then older
eggs (dark blue) would hatch first, followed by the younger eggs (light
blue). If chemosensory and auditory (but not mechanical) cues facilitate
hatching synchrony (b), we predict that that all eggs will hatch at a similar
time, as the chemosensory and/or auditory cues from the first hatchlings
stimulate hatching in all others. If mechanical cues alone facilitate
hatching synchrony (c), we predict that older and younger eggs that are
in contact with each other would hatch at the same time, followed by
older no-contact eggs, and finally younger no-contact eggs
58 Page 4 of 15 Behav Ecol Sociobiol (2020) 74:58

et al. 2003;Ultsch2006). Spring emergence from the nest is
largely synchronised in G. geographica (75% of 20 nests
emerged with the same day in Baker et al. 2013). However,
the ability of G. geographica to synchronise hatching has not
been tested.
Data collection
We conducted daily surveys of a turtle nesting area on the
north shore of Lake Erie (Ontario, Canada) from June 6 to
July 21, 2016. We collected clutches of eggs laid by
A. spinifera and G. geographica, and then transferred them
into one or more semi-transparent plastic tubs (22.5 cm ×
13.48 cm × 5.5 cm) containing a substrate of damp vermicu-
lite, mixed 1:1 by weight with dechlorinated water. Eggs were
incubated ex situ at a constant temperature of 30 °C within
laboratory incubators (Snake Shack; Nature’s Spirit Reptiles,
Vicksburg, MI, USA), and hydric conditions were maintained
by weighing tubs each week and adding water as needed to
maintain the initial vermiculite/water ratio.
To test for the presence of hatching synchrony, we paired
eggs from conspecific nests that were laid approximately 1
week apart, by different mothers, and incubated six older eggs
and six younger eggs in each experimental tub (a total of 12
eggs in each; Fig. 1). The older clutches of eggs were incu-
bated together (at a constant temperature of 30 °C) in another
tub until the younger eggs they were paired with had been
collected. For A. spinifera, we co-incubated older and younger
nests that were approximately 6 days apart in lay date (mean =
6.17 days, SE = 1.96; median = 6, range = 1 to 13 days). Lay
dates of older and younger G. geographica nests were about 7
days apart (mean = 7.13 days; SE = 0.13; median = 7, range =
7 to 8 days).
To quantify potential cues facilitating hatching synchrony
in turtles, older and younger eggs were also randomly
assigned to one of two incubation conditions. Eggs were either
incubated in direct contact with one another, or in physical
isolation within compartments made by placing Foamular™
‘walls’of 1.27 cm in thickness into the tubs (Fig. 1). The
contact treatment allowed older and younger eggs to exchange
potential mechanical cues. All eggs, across both incubation
conditions, could potentially exchange chemical or acoustic
cues from developing embryos or recently hatched turtles in
their tub. Visual differences between treatments (i.e. contact or
physical isolation) were obvious to researchers and restricted
our ability to conduct this study using a fully blind approach,
but we could not tell the lay date (i.e. younger orolder eggs) or
mother of an experimental egg visually. An ideal experimental
design would also have included two control treatments; one
where eggs with the same lay dates were incubated together
within an experimental tub (i.e. mimicking the natural state),
and another where individual eggs were isolated in individual
containers, unable to receive any cues from other eggs. Yet,
we were unable to accomplish this due to logistical limitations
during fieldwork. Specifically, there were not enough turtle
nests, and eggs within each nest, to split eggs from the same
mothers across our experimental treatments and the potential
control(s), so we opted to focus on contrasting the effects of
contact-based, physical cues, and non-contact cues (i.e. acous-
tic and/or olfactory communication) in this study. Future stud-
ies should consider the use of these, or similar, controls in their
experimental design.
Some of the eggs in our experiment were unfertilised (no
development), which reduced the final sample size. Our final
data included 61 A. spinifera and 70 G. geographica eggs,
divided among the four unique age and incubation condition
combinations: Contact-Older (N
Apalone
=16; N
Graptemys
=20),
Contact-Younger (N
Apalone
=15; N
Graptemys
= 16), No-
Contact-Older (N
Apalone
=16; N
Graptemys
= 17) and No-
Contact-Younger (N
Apalone
=14; N
Graptemys
= 17). The
A. spinifera eggs belonged to 10 clutches and were incubated
in 6 experimental tubs. Graptemys geographica eggs
belonged to 16 clutches and were incubated in 8 experimental
tubs.
Once the oldest eggs had incubated for 55 days, the tubs
were checked daily for pipping (when a hatchling has first
broken the shell of its egg) or hatched turtles. Turtles were
removed from the experimental tub once they had emerged
fully from the egg. For each individual, we recorded the date
of hatching, hatchling yolk sac size, hatchling weight (to the
nearest 0.01 g), and hatchling straight carapace length (SCL;
mm). We obtained an ordinal measure of yolk sac size by
scoring yolk sacs on a scale of 0 (no yolk sac remaining) to
8 through comparison to a set of eight clay balls ranging in
diameter from 3 to 22 mm. After measurements, hatchlings
were marked with visible implant elastomer and decimal-
coded wire tags (Davy et al. 2010), and were released near
their original nest site within 3 days of hatching. For analyses,
we coded hatching date in annual numerical sequence
(Wilimovsky 1990), and calculated the incubation duration
by counting the number of days between oviposition and
hatching date.
Statistical analyses
Before analyses, we explored the data following the protocol
described in Zuur et al. (2010) to ensure there were no influ-
ential outliers or collinearity between model covariates. We
used R version 3.0.3 (R Core Team 2016) to perform linear
mixed effect models (LMMs) with the function lme in the
nlme R package (Pinheiro et al. 2014). For all models, we first
pooled data for both turtle species, considering species as a
factor in the analysis. Neither date of hatching (β=−3.43 ±
3.17, t-value
1,130
=−1.08, P= 0.29) nor hatchling mass (β=
0.36 ± 0.56, t-value
1,130
=0.63, P= 0.54) differed between
species. However, there were significant interspecific
Behav Ecol Sociobiol (2020) 74:58 Page 5 of 15 58

differences in incubation duration (β= 2.59 ± 0.95, t-
value
1,130
=2.74,P= 0.01) and size of the yolk sac at hatching
(β= 2.90 ± 0.69, t-value
1,130
= 4.17, P< 0.01). We therefore
decided to analyse data from each species separately with
identical LMMs for each response variable. We also originally
included the random effect of experimental tub, but the vari-
ance it explained was negligible, and we removed this random
effect to optimise model fit. For each model, assumptions of
normality of residuals, and homogeneity of variance were ver-
ified (Zuur et al. 2010). To calculate 95% confidence intervals
for model coefficient and variance estimates we used the func-
tion intervals in the nlme R package (Pinheiro et al. 2014), and
generated contrasts among all incubation treatments (six com-
parisons in total) using the function lsmeans in the lsmeans R
package (Lenth 2016). Pvalues generated for comparisons
among incubation treatments were corrected using Tukey’s
HSD multiplicity adjustment (Lenth 2016). Data are presented
as mean ± standard error (SE) in the text, unless otherwise
specified.
Our first LMMs examined if hatching date (in annual nu-
merical sequence; response variable) varied among treatments
(fixed predictor variable; categorical with four levels). To in-
corporate dependency among observations of eggs from the
same clutch, we included the random effect of clutch identity.
Our second and third set of LMMs had the same fixed and
random effects to examine if incubation duration (days; re-
sponse variable) and yolk sac size (mm; response variable),
respectively, differed among treatments.
Our fourth set of LMMs examined if mass (g; response
variable) of hatchlings varied among treatments. We included
straight carapace length (mm) as a continuous predictor vari-
able, so that the analysis controlled for the effect of hatchling
body size on the overall weight of each individual. This is a
preferred method to examining body condition, in contrast to a
‘residual index’, as this method upholds the underlying as-
sumptions of our statistical methods (García-Berthou 2001).
These models also included the random effect of clutch iden-
tity. We initially included an interaction between treatment
and SCL, but we removed this interaction from both models
and re-fit without interactions after finding they were not
significant.
Results
Hatching date
Spiny Softshells
Hatching date did not differ significantly between older and
younger eggs that were in contact (Table 2a), and, on average,
they hatched within the same day (β= 0.93 ± 0.68, t-value
3,46-
=1.36, P
corr
=0.53; Table 3a). This is despite an average
difference of 6 days between older and younger eggs’lay
dates. Similarly, the hatching date of no-contact younger eggs
did not significantly differ from eggs incubated in contact,
regardless of age (Table 3a).
However, no-contact older eggs hatched significantly later
than all other treatments (Fig. 3a). When eggs were incubated
in isolation, older eggs hatched about 2 days later than youn-
ger eggs (β= 2.04 ± 0.65, t-value
3,46
=3.14,P
corr
=0.01).The
older eggs that were incubated in isolation hatched, on aver-
age, 1.32 ± 0.40 days later than contact older eggs, 2.26 ±
0.66 days later than contact younger eggs, and 2.04 ±
0.65 day later than no-contact younger eggs (Table 3a).
Northern Map turtles
There was no significant difference in hatching date between
incubation treatments in G. geographica (Table 3b;Fig.3b).
Older eggs hatched, on average, 6 days earlier than younger
eggs (Table 3b; Fig. 3b). This difference was not statistically
significant, but this effect size reflects the difference in lay
date between younger and older nests (7 days), and is likely
biologically important.
Incubation duration
Spiny Softshells
Incubation duration of A. spinifera was significantly af-
fected by incubation treatment (Fig. 4a;Tables2b and
3a). Incubation duration did not differ between older
and younger eggs in direct contact (β= 0.66 ± 0.65, t-
value
3,46
=1.01, P
corr
=0.74; Table 3a), which had aver-
age incubation durations of 47.29 ± 0.85 days and 46.63
±0.86 days, respectively.
The incubation duration of younger eggs that were incu-
bated in isolation did not differ from eggs incubated in contact,
regardless of age (Table 3a). The average incubation duration
of no-contact younger eggs was 46.84 ± 0.86 days, which is
within 1 day of the average incubation durations of contact
older and younger treatments (Fig. 4a).
Older eggs that were incubated in isolation had significant-
ly longer incubation durations than all other treatments
(Table 3a). No-contact older eggs incubated for an average
of 48.63 ± 0.84 days, which was at least one full day longer
than all other treatments (Fig. 4a).
Northern Map turtles
There was no effect of incubation treatment on incuba-
tion duration in G. geographica (Fig. 4b), and incuba-
tion duration of G. geographica varied by less than
2 days across all treatments (Table 3b). The average
incubation duration for contact- and no-contact-older
58 Page 6 of 15 Behav Ecol Sociobiol (2020) 74:58

Table 2 Outcomes of linear mixed effect models for Apalone spinifera that examined the effect of incubation treatment (Contact-Older, Contact-
Younger, No-Contact-Older, No-Contact-Younger) on hatching date, incubation duration (days), yolk sac size (mm) and hatchling mass (g)
(a) Hatching date
Fixed effects βLower 95% CI Upper 95% CI t-value P
Intercept (Contact-Older) 16,230.74 16,225.23 16,236.24 5926.52 <0.01
Treatment (Contact-Younger) −0.93 −2.31 0.44 −1.36 0.18
Treatment (No-Contact-Older) 1.33 0.53 2.13 3.33 <0.01
Treatment (No-Contact-Younger) −0.71 −2.06 0.64 −1.08 0.30
SCL nanananana
SCL × treatment na na na na na
Random effects σ
2
Lower 95% CI Upper 95% CI
Clutch identity 73.51 28.93 186.77
Residual 1.24 0.82 1.88
(b) Incubation Duration (days)
Fixed effects βLower 95% CI Upper 95% CI t-value P
Intercept (Contact-Older) 47.29 45.59 49.00 55.77 <0.01
Treatment (Contact-Younger) −0.66 −1.97 0.65 −1.01 0.32
Treatment (No-Contact-Older) 1.33 0.53 2.14 3.34 <0.01
Treatment (No-Contact-Younger) −0.45 −1.74 0.84 −0.71 0.48
SCL nanananana
SCL × treatment na na na na na
Random effects σ
2
Lower 95% CI Upper 95% CI
Clutch identity 5.80 2.14 15.73
Residual 1.25 0.84 2.67
(c) Mass (g)
Fixed effects βLower 95% CI Upper 95% CI t-value P
Intercept (Contact-Older) 3.74 1.24 6.23 3.01 <0.01
Treatment (Contact-Younger) 0.13 −0.59 0.86 0.37 0.71
Treatment (No-Contact-Older) −0.13 −0.57 0.30 −0.61 0.55
Treatment (No-Contact-Younger) 0.11 −0.61 0.82 0.30 0.77
SCL 0.10 0.05 0.16 3.66 <0.01
SCL × Treatment –––––
Random effects σ
2
Lower 95% CI Upper 95% CI
Clutch identity 2.03 0.77 5.35
Residual 0.37 0.24 0.55
(d) Yolk sac size (mm)
Fixed effects βLower 95% CI Upper 95% CI t-value P
Intercept (Contact-Older) 5.63 3.87 7.38 6.45 <0.01
Treatment (Contact-Younger) −0.23 −2.74 2.29 −0.18 0.86
Treatment (No-Contact-Older) 0.84 −1.63 3.32 0.68 0.50
Treatment (No-Contact-Younger) −0.91 −3.48 1.65 −0.71 0.48
SCL nanananana
SCL × treatment na na na na na
Random effects σ
2
Lower 95% CI Upper 95% CI
Clutch identity 0.00 na na
Residual 12.16 8.42 17.55
A total of 61 A. spinifera eggs were sampled from 10 clutches and 6 experimental tubs. Models examining the response variable of hatchling mass
included straight carapace length (SCL) as a covariate. We present coefficient estimates (β)and variance (σ
2
) of random effects, as well as their
associated 95% confidence intervals. If a fixed factor was not included in a model it is indicated with ‘na’. If models initially included a fixed factor,
but it was removed from the final full model due to non-significance, it is indicated with ‘–’. Bolded results indicate significance
Behav Ecol Sociobiol (2020) 74:58 Page 7 of 15 58

eggs was 50.38 ± 0.87 and 50.98 ± 0.88 days, respective-
ly. The average incubation duration for contact- and no-
contact-younger eggs was 49.09 ± 0.88 and 49.22 ±
0.87 days, respectively.
Yolk sac size and hatchling size
Spiny Softshells
Mass of hatchling A. spinifera was significantly related to
SCL (mm; β= 0.10 ± 0.03, t-value
1,42
= 3.66, P< 0.01;
Table 2c). Hatchling mass did not differ among incubation
treatments when controlling for SCL. Hatchlings were all,
on average, within less than 1 g of each other, regardless of
incubation treatment (Table 3a). Similarly, yolk sac size of
hatchling A. spinifera did not differ among incubation treat-
ments, and only varied, on average, from 4.71 ± 0.93 to 6.46
± 0.87 mm, among treatments (Table 3a).
Northern Map turtles
Mass of hatchling G. geographica was significantly related to
their SCL (mm; β= 0.22 ± 0.06, t-value
1,50
=3.96, P<0.01;
Tab le 4c). Hatchling mass, when controlled for SCL, did not
differ among incubation treatments (Table 3b). Hatchlings
were all, on average, within less than 1 g of each other, re-
gardless of incubation treatment (Table 3b). Yolk sac size of
hatchling G. geographica was not affected by incubation
treatment, and only varied, on average, from 7.56 ± 0.93 to
8.85 ± 1.01 mm among treatments (Table 3b).
Discussion
Our results provide evidence for hatching synchrony in
A. spinifera, but not G. geographica, and reveal strong effects
of clutch (i.e. parental) effects on hatching date and incubation
duration of eggs. Hatching in A. spinifera was synchronised
by multiple environmental cues; younger and older eggs that
were in contact hatched within 24-h of one another, as did
younger eggs that were incubated in isolation. These results
suggest that mechanical cues, as well as acoustic and/or
chemosensory cues synchronise hatching in this species
(Fig. 2b, c). The precise mechanisms by which A. spinifera
synchronises hatching cannot be determined by our results,
but we hypothesise that older embryos that were in direct
contact with younger embryos may have synchronised hatch-
ing by foregoing a resting period ofdevelopment (Doody et al.
2012). This hypothesis is supported by the later hatching of
the isolated older eggs, which also had a longer incubation
duration. An alternative hypothesis is that eggs in contact with
one another may experience a higher incubation temperature
than isolated eggs, which would shorten incubation duration
(Godfrey et al. 1997; Zbinden et al. 2006). Yet, in our
Table 3 Pairwise comparisons among the effect of incubation treatments (CO: Contact-Older, CY: Contact-Younger, NCO: No-Contact-Older, NCY:
No-Contact-Younger) on hatching date, incubation duration (days), yolk sac size (mm), and hatchling mass (g)
Incubation Treatment Hatching Date Incubation Duration (days) Yolk Sac Size (mm) Mass (g)
βSE t-value P
corr
βSE t-value P
corr
βSE t-value P
corr
βSE t-value P
corr
(a) Apalone spinifera (N
obs
=61;N
clutch
=10;N
tub
=6)
CO vs. CY 0.93 0.68 1.36 0.53 0.66 0.65 1.01 0.74 −0.13 1.29 −0.10 1.00 0.22 1.25 0.18 1.00
CO vs. NCO −1.32 0.40 −3.33 <0.01 −1.33 0.40 −3.34 0.01 −0.81 1.27 −0.64 0.92 −0.84 1.23 −0.68 0.90
CO vs. NCY 0.71 0.67 1.06 0.72 0.45 0.64 0.71 0.89 0.65 1.31 0.50 0.96 0.91 1.28 0.71 0.89
CY vs. NCO −2.26 0.66 −3.42 <0.01 −2.00 0.63 −3.16 0.01 −0.68 1.29 −0.53 0.95 −1.07 1.25 −0.85 0.83
CY vs. NCY −0.22 0.42 −0.53 0.95 −0.21 0.42 −0.49 0.96 0.78 1.33 0.59 0.94 0.69 1.30 0.53 0.95
NCO vs. NCY 2.04 0.65 3.14 0.01 1.79 0.62 2.88 0.03 1.46 1.31 1.12 0.68 1.75 1.28 1.38 0.52
(b) Graptemys geographica (N
obs
=70;N
clutch
=16;N
tub
=8)
CO vs. CY −5.76 3.44 −1.68 0.35 1.29 1.24 1.04 0.73 −1.02 1.31 −0.78 0.86 −1.29 1.37 −0.94 0.78
CO vs. NCO −0.54 0.55 −0.97 0.77 −0.60 0.55 −1.10 0.69 −0.89 1.17 −0.77 0.87 −0.97 1.22 −0.80 0.86
CO vs. NCY −5.82 3.43 −1.70 0.34 1.15 1.23 0.94 0.79 −1.52 1.29 −1.18 0.64 −1.65 1.35 −1.22 0.62
CY vs. NCO 5.22 3.44 1.52 0.43 −1.89 1.25 −1.52 0.43 0.13 1.35 0.10 1.00 0.32 1.42 0.22 1.00
CY vs. NCY −0.06 0.55 −0.11 1.00 −0.14 0.55 −0.25 1.00 −0.50 1.22 −0.41 0.98 −0.37 1.27 −0.29 0.99
NCO vs. NCY −5.28 3.44 −1.54 0.42 1.76 1.24 1.42 0.50 −0.63 1.34 −0.47 0.97 −0.69 1.40 −0.49 0.96
Linear mixed effect models were performed separately for each turtle species, (a) Apalone spinifera and (b) Graptemys geographica, and the models
examiningthe response variable of hatchling mass included straight carapace lengthas a covariate (seeTables 3and 4). Significant pairwise comparisons
are bolded, and Pvalues are corrected for multiple comparisons (P
corr
)
58 Page 8 of 15 Behav Ecol Sociobiol (2020) 74:58

experiment, the similar hatching date and incubation duration
of younger eggs, regardless of incubation condition (contact
or isolation), rules out a potential confounding thermal effect
of metabolic heating (Zbinden et al. 2006)amongthegrouped
eggs.
We note that although we isolated some eggs from each
another, we did not quantify transmission of vibrations (in-
cluding audible sounds) through the experimental design.
Even long-distance vibrations can stimulate hatching in
C. insculpta and Indian Flapshell Turtles (Lissemys punctata),
which have been observed hatching during the vibrations and/
or sound caused by thunder (Vijaya 1983;Doody2011;
Doody et al. 2012). Our study suggests that mechanical cues
between eggs incubating in contact and non-contact (auditory
and/or olfactory) cues drive hatching synchrony in
A. spinifera, but further research is required to explicitly un-
tangle which non-contact cue(s) are involved.
In contrast, we did not find evidence of hatching synchrony
in G. geographica. Although hatching was not statistically
different among incubation treatments, which was the predic-
tion if hatching synchrony was occurring, there was a biolog-
ically relevant difference in hatching date between younger
and older eggs. Younger G. geographica eggs hatched 6 days
earlier than older eggs, which mirrors the difference inlay date
between younger and older nests (7 days). Incubation duration
of G. geographica was also unaffected by our experiment,
suggesting that the developmental rate of G. geographica is
more constrained than that of A. spinifera. None of the incu-
bation treatments affected yolk size or body condition of
hatchlings of either species, which may reflect strong selection
for fully developed neonates. Below, we consider two hypoth-
eses to explain the two species’hatching strategies: (1) the
potential effects of different egg shell structure in transmitting
and receiving cues, and (2) the outcome of different selective
pressures on hatching phenology in species that emerge from
the nest in late summer and early autumn (A. spinifera;Ernst
and Lovich 2009), or that delay emergence to overwinter in
the nest with emergence occurring the following spring
(G. geographica; Nagle et al. 2004).
A key physical difference in embryonic development of
A. spinifera and G. geographica is the structure of the egg
shell: A. spinifera eggs are round with hard shells, while
G. geographica eggs are ovoid with flexible shells (Ernst
and Lovich 2009). We are not aware of any studies comparing
the transmission of mechanical, acoustic, or chemosensory
cues through egg shells with different structures. However,
we hypothesise that interspecific variation in egg shape and
shell structure may favour different strategies for embryo–
(a)
Treatment
Contact
Older
Contact
Older
Contact
Younger
Contact
Younger
No-Contact
Younger
No-Contact
Younger
No-Contact
Older
No-Contact
Older
*
*
*
Hatching Date
(b)
Fig. 3 Predicted hatching date (in annual numeral sequence; Wilkinson
1990) of each incubation treatment (Contact-Older, Contact-Younger,
No-Contact-Older, No-Contact-Younger) for clutches of (a)Apalone
spinifera and (b)Graptemys geographica. Younger and older eggs are
represented using light and dark blue, respectively. The contact
incubation condition is represented using solid fill, and the non-contact
incubation condition is represented using striped fill. Significant
comparisons between pairs of points are shown using a bar connecting
them and an asterisk (*). Treatment medians are represented with a white
dot, the interquartile range is represented with a thick, black rectangle,
95% confidence intervals are represented with a thin, black line. A density
plot of each treatment’s incubation duration is perpendicular, on the left
and right, to each line
Behav Ecol Sociobiol (2020) 74:58 Page 9 of 15 58

embryo communication by affecting the transmission of vi-
brations, sounds, or chemical signals. For example, both
C. insculpta and L. punctata have brittle, hard-shelled eggs
like A. spinifera, and are known to use vibrations as hatching
cues (Vijaya 1983; Doody 2011; Doody et al. 2012). It is
possible that hard shells transmit sounds more clearly than
other eggshell morphologies. Examining the impact of egg
morphology on hatching synchrony and embryo–embryo
communication is a future research path full of promise.
Graptemys geographica often remain in the nest overwin-
ter(Nagleetal.2004), whereas A. spinifera emerge from the
nest in late summer (Ernst and Lovich 2009) and must migrate
to overwintering locations before the onset of winter (Ultsch
2006). This key difference in early life history strategy may
explain the lack of hatching synchrony in G. geographica
because the benefits of synchronous emergence from the nest
(Ims 1990; Tucker et al. 2008) are not dependent on synchro-
nous hatching when hatchlings can overwinter and emerge
from the nest together the following spring. Baker et al.
(2013) found a high incidence of synchronous emergence in
G. geographica. The lack of synchronous hatching in our
study suggests this species may synchronise emergence inde-
pendently of hatching. Selection may even favour later hatch
dates in G. geographica because slowing development to
maximise over-winter yolk, fat and water reserves may in-
crease the probability of over-winter survival in the nest
(Mitchell et al. 2013; Riley et al. 2014). Conversely,
A. spinifera face pressure to find suitable overwintering sites
and build energy reserves by foraging prior to overwintering.
Similar pressure on Juvenile Jacky Dragons (Amphibolurus
muricatus) favoured early hatch dates, which predicted survi-
vorship and growth rates in the first 3 months following emer-
gence (Warner and Shine 2007). Overall, the different hatch-
ing strategies we documented between A. spinifera and
G. geographica illustrate the importance of comparative ap-
proaches that consider interspecific variation in biology and
life history strategies.
Painted Turtles (Chrysemys picta) are sympatric with
G. geographica, have similar, flexible egg shells, and can
overwinter within their natal nest (Ultsch 2006)—but hatch
synchronously (McGlashan et al. 2018). When developmental
asynchrony was induced within a clutch by controlling the
incubation temperature of individuals eggs, less developed
C. picta hatched within 1 day of clutch-mates by hatching
early, relative to their developmental stage, with no indication
of metabolic compensation (McGlashan et al. 2018).
Contact
Older
Contact
Older
Contact
Younger
Contact
Younger
No-Contact
Younger
No-Contact
Younger
No-Contact
Older
No-Contact
Older
*
*
*
(b)(a)
Fig. 4 Predicted incubation duration (days) of each incubation treatment
(Contact-Older, Contact-Younger, No-Contact-Older, No-Contact-
Younger) for (a)Apalone spinifera and (b)Graptemys geographica
clutches. Younger and older eggs are represented using light and dark
blue, respectively. The contact incubation condition is represented using
solid fill, and the non-contact incubation condition is represented using
striped fill. Significant comparisons between pairs of points are shown
using a bar connecting them and an asterisk (*). Treatment medians are
represented with a white dot, the interquartile range is represented with a
thick, black rectangle, 95% confidence intervals are represented with a
thin, black line. A density plot of each treatment’s incubation duration is
perpendicular, on the left and right, to each line
58 Page 10 of 15 Behav Ecol Sociobiol (2020) 74:58

Table 4 Outcomes of linear mixed effect models for Graptemys geographica that examined theeffect of incubation treatment (Contact-Older, Contact-
Younger, No-Contact-Older, No-Contact-Younger) on hatching date, incubation duration (days), hatchling mass (g) and yolk sac size (mm)
(a) Hatching date
Fixed effects βLower 95% CI Upper 95% CI t-value P
Intercept (Contact-Older) 16,224.20 16,219.33 16,229.07 6684.50 <0.01
Treatment (Contact-Younger) 5.78 −1.14 12.65 1.68 0.10
Treatment (No-Contact-Older) 0.54 −0.57 1.65 0.97 0.34
Treatment (No-Contact-Younger) 5.82 −1.07 12.71 1.70 0.10
SCL nanananana
SCL × treatment na na na na na
Random effects σ
2
Lower 95% CI Upper 95% CI
Clutch identity 46.01 21.71 97.48
Residual 2.35 1.60 3.46
(b) Incubation duration (days)
Fixed effects βLower 95% CI Upper 95% CI t-value P
Intercept (Contact-Older) 50.38 48.64 52.12 58.10 <0.01
Treatment (Contact-Younger) −1.29 −3.77 1.19 −1.04 0.30
Treatment (No-Contact-Older) 0.60 −0.49 1.70 1.10 0.28
Treatment (No-Contact-Younger) −1.15 −3.62 1.32 −0.94 0.35
SCL nanananana
SCL × treatment na na na na na
Random effects σ
2
Lower 95% CI Upper 95% CI
Clutch identity 4.93 2.14 11.34
Residual 2.35 1.60 3.46
(c) Mass (g)
Fixed effects βLower 95% CI Upper 95% CI t-value P
Intercept (Contact-Older) 0.67 −2.61 3.96 0.41 0.68
Treatment (Contact-Younger) −0.35 −1.18 0.49 −0.83 0.41
Treatment (No-Contact-Older) 0.09 −0.24 0.43 0.56 0.58
Treatment (No-Contact-Younger) −0.28 −1.12 0.55 −0.68 0.50
SCL 0.22 0.11 0.33 3.96 <0.01
SCL × treatment –––––
Random effects σ
2
Lower 95% CI Upper 95% CI
Clutch identity 0.57 0.23 1.39
Residual 0.22 0.15 0.32
(d) Yolk sac size (mm)
Fixed effects βLower 95% CI Upper 95% CI t-value P
Intercept (Contact-Older) 7.56 5.70 9.42 8.16 <0.01
Treatment (Contact-Younger) 1.29 −1.47 4.04 0.94 0.35
Treatment (No-Contact-Older) 0.97 −1.47 3.41 0.80 0.43
Treatment (No-Contact-Younger) 1.65 −1.06 4.37 1.22 0.23
SCL nanananana
SCL × treatment na na na na na
Random effects σ
2
Lower 95% CI Upper 95% CI
Clutch identity 1.47 0.11 19.67
Residual 13.04 8.84 19.24
A total of 70 G. geographica eggs were sampled from 16 clutches and 6 experimental tubs. Models examining the response variable of hatchling mass
included straight carapace length (SCL) as a covariate. We present coefficient estimates (β)and variance (σ
2
) of random effects, as well as their
associated 95% confidence intervals. If a fixed factor was not included in a model, it is indicated with ‘na’. If models initially included a fixed factor,
but it was removed from the final full model due to non-significance, it is indicated with ‘–’. Bolded results indicate significance
Behav Ecol Sociobiol (2020) 74:58 Page 11 of 15 58

McGlashan et al. (2018) hypothesised that synchronising
hatching by hatching early may increase the chance for
C. picta hatchlings to obtain an optimal overwintering posi-
tion in the nest and reduce the chance of mortality (Costanzo
et al. 2008). Yet, our study found no evidence of hatching
synchrony in G. geographica a species with a similar
overwintering strategy and requirements. The inconsistency
between our results suggests interspecific variation in hatch-
ing synchrony, even between species that overwinter within
the nest.
Previously, most experimental studies that have examined
hatching synchrony in turtles have induced developmental
asynchrony between individuals of the same clutch by con-
trolling incubation temperatures (Table 1). Comparing eggs
from the same clutch could be confounded by clutch effects,
which might also produce apparent hatching synchrony
(Steyermark and Spotila 2001). Our study design controlled
for age and clutch effects by comparing eggs at different de-
velopmental stages that belonged to different clutches. This
approach allowed us to specifically quantify the influence
clutch effects have on hatching date and incubation duration
to determine the magnitude of the role clutch effects may have
on incubation duration. We were able to accomplish this be-
cause all models that included clutch identity as a random
effect estimated σ
2
clutch
, which is the variation among individ-
uals due to maternal or paternal genetic and phenotypic effects
(Noble et al. 2014) and the residual variance (σ
2
r
; Tables 3and
4). Post hoc, we were able to calculate to proportion of vari-
ance explained by clutch effects by first calculating phenotyp-
ic variance (σ
2
p
), which is the sum of all variance components
including σ
2
r
(Noble et al. 2014) and then estimating clutch
effects as σ
2
clutch
/σ
2
p
(Lynch and Walsh 1998; Noble et al.
2014). We were also able to calculate 95% confidence inter-
vals for clutch effects by bootstrapping the data 1000 times
with the boot function from the R package boot (Davison and
Hinkley 1997; Canty and Ripley 2017), and then, to be con-
servative, considered clutch effects significant when the 95%
confidence intervals did not include 0. We found that clutch
effects explained a large and significant proportionof variance
in both hatching date and incubation duration in both turtle
species (Fig. 5). This is evidence that siblings have a much
higher likelihood of expressing similar hatching dates and
incubation durations than non-siblings, and suggests there
may be heritability in these traits (Arnold and Bennett 1984;
Webb et al. 2001). Thesefindings add to the growing evidence
that parental genetic and phenotypic (clutch) has significant
effects on behaviour of offspring. A recent study on
snapping turtles (Chelydra serpentina) found that nest temper-
ature did not affect hatching time, which suggests other fac-
tors, like clutch effects, may play a larger role (Rouleau et al.
2019). Our findings also underline the importance of control-
ling for clutch effects in studies of hatching synchrony,
where experimental designs commonly incubate less and
more advanced siblings together (Table 1). Given the strong
clutch effects detected in our study, that approach should not
be used.
The current study found interspecific variation in hatch-
ing synchrony exhibited by two freshwater turtles. There
was no evidence for hatching synchrony in
G. geographica,whileA. spinifera exhibited hatching syn-
chrony that was mediated through direct contact of eggs as
well as potential acoustic and chemosensory cues. This
interspecific variation implies varying selective pressures
among species, but these are difficult to identify as the
underlying adaptive significance of synchronous hatching
and emergence remains unclear. Predator dilution is often
invoked to explain synchronous emergence, but this hy-
pothesis is yet to gain empirical support (Ims 1990;
Tucker et al. 2008). An alternate explanation is that hatch-
ing synchrony serves as an initiation point for ongoing,
potentially kin-based social interactions. For example,
hatchling green iguanas (Iguana iguana) that emerge from
the nest together also continue to associate after hatching
and emergence during larger spatial moves (Burghardt
et al. 1977). Research on complex social behaviour in tur-
tles is growing (Rife et al. 2007; Wilkinson et al. 2010;
Davis and Burghardt 2011, Radzio et al. 2016), but lags
substantially behind research in other reptiles and verte-
brates (Doody et al. 2013; Gardner et al. 2016). Particular
life history traits, like longevity and philopatry, are thought
to generally predispose animals to group and family living
(Chapple 2003;WardandWebster2016). Turtles, in gen-
eral, have life history traits (i.e. longevity, slow recruitment
of adults into populations, and fidelity to particular habitat
features) that may favour the evolution of group and/or
0.0
0.2
0.4
0.6
0.8
1.0
Proportion of Variance (σ2
clutch/σ2
p)
Hatching Date Incubation Duration
Fig. 5 Estimates of the proportion of variance in hatching date (left) and
incubation duration (right) that is explained by clutch (σ
2
clutch
/σ
2
p
)effects
of Apalone spinifera (black circles; see Table 3a, b for corresponding
models) and Graptemys geographica (white circles; see Table 4a, b).
Error bars around estimates are 95% confidence intervals (CIs) and
estimates were all significant (conservatively considered significant as
their 95% CIs not include 0)
58 Page 12 of 15 Behav Ecol Sociobiol (2020) 74:58

family living, and predispose them to social associations
and behaviour. Overall, future research into synchronous
hatching should take care to control for clutch effects in the
experimental design, and must be put in context with re-
gard to the evolutionary pressures driving interspecific var-
iation in hatching phenology and strategies.
Acknowledgements We thank Anne and Ric MacArthur and the staff of
Rondeau Provincial Park for logistical support during the project.
StephanieChan, Rebecca Novack, Yehong Shi and Juliana Skuza assisted
with care of the eggs during incubation. This study was supported in part
by the Government of Ontario and Wildlife Preservation Canada. The
initial manuscript was much improved by comments from Dr Sean
Doody and an anonymous reviewer.
Author contributions S.H., C.F.-L. and C.M.D. conceptualised the study.
S.H. and C.F.-L. collected data, and C.M.D. acquired funding and per-
mits. J.L.R. conducted statistical analyses. S.H. and C.F.-L. drafted an
earlier version of the manuscript; J.L.R. led the writing of the final ver-
sion. All authors contributed to editing and finalising the final manuscript.
Funding information Collection of eggs for ex situ incubation was con-
ducted for the recovery of the target population, and was funded by the
Government of Ontario (Species at Risk Stewardship Fund grant SAR-
00094) and by Wildlife Preservation Canada.
Compliance with ethical standards
Ethical approval All applicable international, national and/or institutional
guidelines for the use of animals were followed. This research was
authorised under ESA Registry (M-102-4775033319), a Fish and
Wildlife Scientific Collector’s Authorization, a Protected Areas Research
Authorization, and an approved Animal Care Protocol (16-291) from the
Government of Ontario.
Conflict of interest The authors declare they have no conflict of interest.
Data availability The datasets generated during and analysed in the
current study are available in from Open Source Framework (OSF) at
https://osf.io/mganf/.
References
Arnold SJ, Bennett AF (1984) Behavioural variation in natural popula-
tions. III: antipredator displays in the garter snake Thamnophis
radix. Anim Behav 32:1108–1118. https://doi.org/10.1016/S0003-
3472(84)80227-4
Arnold SJ, Wassersug RJ (1978) Differential predation on metamorphic
anurans by garter snakes (Thamnophis): social behavior as a possi-
ble defence. Ecology 59:1014–1022. https://doi.org/10.2307/
1938553
Aubret F, Blanvillain G, Bignon F, Kok PJ (2016) Heartbeat, embryo
communication and hatching synchrony in snake eggs. Sci Rep 6:
23519. https://doi.org/10.1038/srep23519
Baker PJ, Costanzo JP, Iverson JB, Lee RE Jr (2003) Adaptations to
terrestrial overwintering of hatchling northern map turtles,
Graptemys geographica. J Comp Physiol B 173:643–651. https://
doi.org/10.1007/s00360-003-0373-5
Baker PJ, Costanzo JP, Iverson JB, Lee RE Jr (2013) Seasonality and
interspecific and intraspecific asynchrony in emergence from the
nest by hatchling freshwater turtles. Can J Zool 91:415–461.
https://doi.org/10.1139/cjz-2012-0335
Banger N, Blouin-Demers G, Bulté G, Lougheed SC (2013) More sires
may enhance offspring fitness in northern map turtles (Graptemys
geographica). Can J Zool 91:581–588. https://doi.org/10.1139/cjz-
2012-0320
Baxter-Gilbert J, Riley JL, Whiting MJ (2018) Runners and fighters:
clutch effects and body size drive innate antipredator behaviour in
hatchling lizards. Behav Ecol Sociobiol 71:97
Burger J (1976) Behaviour of hatchling diamondback terrapins
(Malaclemys terrapin) in the field. Copeia 1976:742–748. https://
doi.org/10.2307/1443457
Burghardt GM, Greene HW, Rand AS (1977) Social behavior in hatch-
ling green iguanas: life at a reptile rookery. Science 195:689–691.
https://doi.org/10.1126/science.195.4279.689
Canty A, Ripley B (2017) Boot: bootstrap R (S-plus) functions. R pack-
age version 1: 3–20, https://cran.r-project.org/web/packages/boot/
boot.pdf
Carr A, Hirth H (1961) Social facilitation in green turtle siblings. Anim
Behav 9:68–70. https://doi.org/10.1016/0003-3472(61)90051-3
Chapple DG (2003) Ecology, life-history, and behavior in the Australian
scincid genus Egernia, with comments on the evolution of complex
sociality in lizards. Herpetol Monogr 17:145–180. https://doi.org/
10.1655/0733-1347(2003)017[0145:ELABIT]2.0.CO;2
Christens E (1990) Nest emergence lag in loggerhead sea turtles. J
Herpetol 24:400–402. https://doi.org/10.2307/1565057
Christiansen JL, Gallaway BJ (1984) Raccoon removal, nesting success,
and hatchling emergence in Iowa turtles with a special reference to
Kinosternon falvescens (Kinosternidae). Southwest Nat 29:343–
348. https://doi.org/10.2307/3671365
Colbert PL, Spencer RJ, Janzen FJ (2010) Mechanism and cost of syn-
chronous hatching. Funct Ecol 24:112–121. https://doi.org/10.1111/
j.1365-2435.2009.01602.x
Costanzo JP, Lee RE Jr, Ultsch GR (2008) Physiological ecology of
overwintering in hatchling turtles. J Exp Zool A 309:297–379.
https://doi.org/10.1002/jez.460
Davis KM, Burghardt GM (2011) Turtles (Pseudemys nelsoni) learn
about visual cues indicating food from experienced turtles. J
Comp Psychol 125:404–410. https://doi.org/10.1037/a0024784
Davison AC, Hinkley DV (1997) Bootstrap methods and their applica-
tions. Cambridge University Press, Cambridge
Davy CM, Coombes SM, Whitear AK, MacKenzie AS (2010) Visible
implant elastomer: a simple, non-harmful method for marking
hatchling turtles. Herpetol Rev 41:442–445
Doody JS (2011) Environmentally cued hatching in reptiles. Integr Comp
Biol 51:49–61. https://doi.org/10.1093/icb/icr043
Doody JS, Stewart B, Camacho C, Christian K (2012) Good vibrations?
Sibling embryos expedite hatching in a turtle. Anim Behav 83:645–
651. https://doi.org/10.1016/j.anbehav.2011.12.006
Doody JS, Burghardt GM, Dinets V (2013) Breaking the social–non-
social dichotomy: a role for reptiles in vertebrate social behavior
research? Ethology 119:95–103. https://doi.org/10.1111/eth.12047
Endo J, Takanashi T, Mukai H, Numata H (2019) Egg-cracking vibration
as a cue for stink bug siblings to synchronize hatching. Curr Biol 29:
143–148. https://doi.org/10.1016/j.cub.2018.11.024
Ernst CH, Lovich JE (2009) Turtles of the united stated and Canada, 2nd
edn. John Hopkins University Press, Baltimore
Ferrara CR, Vogt RC, Harfush MR, Sousa-Lima RS, Albavera E, Tavera
A (2014a) First evidence of leatherback turtle (Dermochelys
coriacea) embryos and hatchlings emitting sounds. Chelonian
Conserv Biol 13:110–11 4. https://doi.org/10.2744/CCB-1045.1
Ferrara CR, Vogt R, Sousa-Lima R, Tardio B, Bernardes V (2014b)
Sound communication and social behaviour in an Amazonian river
turtle (Podocnemis expansa). Herpetologica 70:149–156. https://
doi.org/10.1655/HERPETOLOGICA-D-13-00050R2
Behav Ecol Sociobiol (2020) 74:58 Page 13 of 15 58

García-Berthou E (2001) On the misuse of residuals in ecology: testing
regression residuals vs. the analysis of covariance. J Anim Ecol 70:
708–711. https://doi.org/10.1046/j.1365-2656.2001.00524.x
Gardner MG, Pearson SK, Johnston GR, Schwarz MP (2016) Group
living in squamate reptiles: a review of evidence for stable aggrega-
tions. Biol Rev 91:925–936. https://doi.org/10.1111/brv.12201
Gibbons JW (2013) A long-term perspective of delayed emergence (aka
overwintering) in hatchling turtles: some they do and some they
don’t, and some you just can’t tell. J Herpetol 47:203–214. https://
doi.org/10.1670/12-122
Gibbons JW, Lovich JE (1990) Sexual dimorphism in turtles with em-
phasis on the slider turtle (Trachemys scripta). Herpetol Monogr 4:
1–29. https://doi.org/10.2307/1466966
Godfrey MH, Barreto R, Mrosovsky N (1997) Metabolically-generated
heat of developing eggs and its potential effect on sex ratio of sea
turtle hatchlings. J Herpetol 31:616–619. https://doi.org/10.2307/
1565626
Greenbaum R, Carr JL (2001) Sexual differentiation in the spiny softshell
turtle (Apalone spinifera) a species with genetic sex determination. J
Exp Zool 290:190–200. https://doi.org/10.1002/jez.1049
Hill PS (2019) Biotremology: the sound of one egg cracking. Curr Biol
29:R16–R17. https://doi.org/10.1016/j.cub.2018.11.035
Ims RA (1990) On the adaptive value of reproductive synchrony as a
predator-swamping strategy. Am Nat 136:485–498. https://doi.org/
10.1086/285109
Kojima W (2015) Mechanism of synchronous metamorphosis: larvae of a
rhinoceros beetle alter the timing of pupation depending on maturity
of their neighbours. Behav Ecol Sociobiol 69:415–424. https://doi.
org/10.1007/s00265-014-1854-0
Krenz JG, Naylor GJP, Shaffer HB, Janzen FJ (2005) Molecular phylo-
genetics and evolution of turtles. Mol Phylogenet Evol 37:178–191.
https://doi.org/10.1016/j.ympev.2005.04.027
Lenth RV (2016) Least-squares means: the R package lsmeans. J Stat
Softw 69:1–33. https://doi.org/10.18637/jss.v069.i01
Lovich JE, Ernst CH, Ernst EM, Riley JL (2014) A 21-year study of
seasonal and interspecific variation of hatchling emergence in a
Nearctic freshwater turtle community: to overwinter or not to over-
winter? Herpetol Monogr 28:93–109. https://doi.org/10.1655/
HERPMONOGRAPHS-D-14-00001
Lynch M, Walsh B (1998) Genetics and analysis of quantitative traits.
Sinauer, Sunderland
McGlashan J, Spencer R-J, Old J (2011) Embryonic communication in
the nest: metabolic responses of reptilian embryos to developmental
rates of siblings. Proc R Soc Lond B 279:1709–1715. https://doi.
org/10.1098/rspb.2011.2074
McGlashan J, Loudon F, Thompson M, Spencer R (2015) Hatching be-
havior of eastern long-necked turtles (Chelodina longicollis): the
influence of asynchronous environments on embryonic heart rate
and phenotype. Comp Biochem Physiol 188:58–64. https://doi.
org/10.1016/j.cbpa.2015.06.018
McGlashan JK, Thompson MB, Janzen FJ, Spencer R-J (2018)
Environmentally induced phenotypic plasticity explains hatching
synchrony in the freshwater turtle Chrysemys picta.JExpZoolA
329:362–372. https://doi.org/10.1002/jez.2217
McKenna L (2016) Vocalizations of sea turtle hatchlings and embryos.
MSc Thesis, Purdue University
Mitchell T, Warner D, Janzen F (2013) Phenotypic and fitness conse-
quences of maternal nest-site choice across multiple older life stages.
Ecology 94:336–345. https://doi.org/10.1890/12-0343.1
Nagle R, Lutz C, Pyle A (2004) Overwintering in the nest by hatchling
map turtles (Graptemys geographica). Can J Zool 82:1211–1218.
https://doi.org/10.1139/z04-096
Newman RA (1988) Genetic variation for larval anuran (Scaphiopus
couchii) development time in an uncertain environment. Evolution
42:763–773. https://doi.org/10.2307/2408867
Noble DW, McFarlane SE, Keogh JS, Whiting MJ (2014) Maternal and
additive genetic effects contribute to variation in offspring traits in a
lizard. Behav Ecol 25:633–640. https://doi.org/10.1093/beheco/
aru032
Noble DW, Stenhouse V, Schwanz LE (2018) Developmental tempera-
tures and phenotypic plasticity in reptiles: a systematic review and
meta-analysis. Biol Rev 93:72–97. https://doi.org/10.1111/br v.
12333
Olsson M, Gullberg A, Shine R, Madson T, Tegelström H (1996) Paternal
genotype influences incubation period, offspring size, and offspring
shape in an oviparous reptile. Evolution 50:1328–1333. https://doi.
org/10.1111/j.1558-5646.1996.tb02372.x
Orcutt A (1974) Sounds produced by hatching Japanese quail (Coturnix
coturnix japonica) as potential aids to synchronous hatching.
Behaviour 50:173–184. https://doi.org/10.1163/156853974X00093
Pearson PR, Warner DA (2018) Early hatching enhances survival despite
beneficial phenotypic effects of late-season developmental environ-
ments. Proc R Soc B 285:20180256. https://doi.org/10.1098/rspb.
2018.0256
Pianka E, Vitt L (2003) Lizards: windows to the evolution of diversity.
Oxford University of California Press, Berkeley
Pinheiro J, Bates D, DebRoy S, Sarkar D (2014) Nlme: linear and non-
linear mixed effects models. R package version 3.1–117, https://
CRAN.R-project.org/package=nlme
Plummer MV (2007) Nest emergence of smooth Softshell (Apalone
mutica) hatchlings. Herpetol Conserv Biol 2:61–64. https://doi.org/
10.1007/s00442-002-1109-z
R Core Team (2016) R: a language and environment for statistical com-
puting. R Foundation for Statistical Computing, Vienna, Austria
http://www.R-project.org
Radzio T, Cox J, Spotila R, O’Connor M (2016) Aggression, combat, and
apparent burrow competition in hatchling and juvenile gopher tor-
toises (Gopherus polyphemus). Chelonian Conserv Biol 15:231–
237. https://doi.org/10.2744/CCB-1181.1
Rife A, Strauss E, Auger P (2007) Social and basking behaviors in juve-
nile, captive-raised northern diamondback terrapins (Malaclemys
terrapin terrapin). BSc Honours Thesis, Boston College
Riley JL, Tattersall G, Litzgus JD (2014) Potential sources of intra-
population variation in the overwintering strategy of painted turtle
(Chrysemys picta) hatchlings. J Exp Biol 217:4174–4183. https://
doi.org/10.1242/jeb.111120
Rollinson N, Massey MD, Meron M, Leivesley J (2019) A low cost,
efficient, and precise technique to quantify key life cycle events in
nests of oviparous reptiles. J Herpetol 53:302–309. https://doi.org/
10.1670/18-168
Rouleau CJ, Massey MD, Rollinson N (2019) Temperature does not
affect hatch timing in snapping turtles (Chelydra serpentina). J
Herpetol 53:165–169. https://doi.org/10.1670/18-048
Santos RG, Pinheiro HT, Martins AS, Riul P, Bruno SC, Janzen FJ,
Ioannou CC (2016) The anti-predator role of within-nest emergence
synchrony in sea turtle hatchlings. Proc R Soc B 283:20160697.
https://doi.org/10.1098/rspb.2016.0697
Schwagmeyer P, Mock D, Lamey T, Lamey C, Beecher M (1991) Effects
of sibling contact on hatch timing in an asynchronously hatching
bird. Anim Behav 41:887–894. https://doi.org/10.1016/S0003-
3472(05)80355-0
Shaffer HB, Meylan P, McKnight ML (1997) Tests of turtle phylogeny:
molecular, morphological, and paleontological approaches. Syst
Biol 46:235–268. https://doi.org/10.1093/sysbio/46.2.235
Spencer R, Janzen FJ (2011) Hatching behaviour in turtles. Integr Comp
Biol 51:100–110. https://doi.org/10.1093/icb/icr045
Spencer R, Thompson MB, Banks PB (2001) Hatch or wait? A dilemma
in reptilian incubation. Oikos 93:401–406. https://doi.org/10.1034/j.
1600-0706.2001.930305.x
58 Page 14 of 15 Behav Ecol Sociobiol (2020) 74:58

Steyermark AC, Spotila JR (2001) Effects of maternal identity and incu-
bation temperature on snapping turtle (Chelydra serpentina)growth.
Funct Ecol 15:624–632. https://doi.org/10.1086/316743
Tornabene BJ, Bramblett RG, Zale AV, Leathe SA (2018) Factors affect-
ing nesting ecology of Apalone spinifera in a northwestern Great
Plains river of the United States. Chelonian Conserv Biol 17:63–77.
https://doi.org/10.2744/CCB-1298.1
Tucker JK, Paukstis GL, Janzen FJ (2008) Does predator swamping pro-
mote synchronous emergence of turtle hatchlings among nests?
Behav Ecol 19:35–40. https://doi.org/10.1093/beheco/arm097
Ultsch GR (2006) The ecology of overwintering among turtles: where
turtles overwinter and its consequences. Biol Rev 81:339–367.
https://doi.org/10.1017/S1464793106007032
Vergne AL, Mathevon N (2008) Crocodile egg sounds signal hatching
time. Curr Biol 18:R513–R514. https://doi.org/10.1016/j.cub.2008.
04.011
Vijaya J (1983) Auditory cues as possible stimuli for hatching eggs of the
flap-shell turtle Lissemys punctata granosa. Hamadryad 8:23
Ward A, Webster M (2016) Sociality: the behaviour of group-living an-
imals. Springer, Berlin
Warkentin KM (2011) Environmentally cued hatching across taxa: em-
bryos respond to risk and opportunity. Integr Comp Biol 51:14–25.
https://doi.org/10.1093/icb/icr017
Warkentin KM, Caldwell MS (2009) Assessing risk: embryos, informa-
tion, and escape hatching. In: Dukas R, Ratcliffe JM (eds) Cognitive
ecology II. University of Chicago Press, Chicago, pp 177–200
Warner DA, Shine R (2007) Fitness of juvenile lizards depends on sea-
sonal timing of hatching, not offspring body size. Oecologia 154:
65–73. https://doi.org/10.1007/s00442-007-0809-9
Webb GJ, Choquenot D, Whitehead PJ (1986) Nests, eggs, and embry-
onic development of Carettochelys insculpta (Chelonia:
Carettochelidae) from northern Australia. J Zool 1:521–550.
https://doi.org/10.1111/j.1096-3642.1986.tb00646.x
Webb JK, Brown GP, Shine R (2001) Body size, locomotor speed and
antipredator behaviour in a tropical snake (Tropidonophic mairii,
Cloubridae): the influence of incubation environments and genetic
factors. Funct Ecol 15:561–568. https://doi.org/10.1046/j.0269-
8463.2001.00570.x
Wilimovsky NJ (1990) Misuses of the term ‘Julian day’. Trans Am Fish
Soc 119:162. https://doi.org/10.1577/1548-8659-119.1.162
Wilkinson A, Kuenstner K, Mueller J, Huber L (2010) Social learning in a
non-social reptile (Geochelone carbonaria). Biol Lett 6:614–616.
https://doi.org/10.1098/rsbl.2010.0092
Zbinden J, Margaritoulis D, Arlettaz R (2006) Metabolic heating in
Mediterranean loggerhead sea turtle clutches. J Exp Mar Biol Ecol
334:151–157. https://doi.org/10.1016/j.jembe.2006.01.021
Zuur AF, Ieno EN, Elphick CS (2010) A protocol for data exploration to
avoid common statistical problems. Methods Ecol Evol 1:3–14.
https://doi.org/10.1111/j.2041-210X.2009.00001.x
Publisher’snoteSpringer Nature remains neutral with regard to jurisdic-
tional claims in published maps and institutional affiliations.
Behav Ecol Sociobiol (2020) 74:58 Page 15 of 15 58