Premature birth stunts early growth and is a possible driver of stress‐induced maternal effects in the guppy Poecilia reticulata

Abstract

We tested the effects of gestational stress, principally in the form of alarm cue extracted from the skin of conspecifics, on reproduction in female guppies (Poecilia reticulata) and the growth and behaviour of their offspring. Offspring from mothers exposed to alarm cue exhibited stunted growth in the first few days post‐partum, which appeared to be mediated by shortening of the gestation period, the length of which directly correlated with growth rate within the first 6 days post‐partum. Mature offspring did not differ in behaviour or stress responses compared with controls and so the effects of maternal predation stress did not appear to persist into adulthood. A different form of gestational stress, dietary restriction, did not significantly affect offspring growth, though brood size was reduced, suggesting that the effects of predation stress were not mediated by differences in resource demand or consumption.

Full text

1
Premature birth stunts early growth and is a possible driver
1
of stress-induced maternal effects in the guppy, Poecilia
2
reticulata
3
James Ord
1
,3*, Kelle E. Holmes1, William V. Holt2, Alireza Fazeli
2
,
3
, and Penelope J. Watt1
4
* Corresponding author: jms.ord18@gmail.com
5
6
The content herein is identical to that contained in the final (peer-reviewed) version submitted to
7
Journal of Fish Biology, with the exception that Table 1 and Figures 1-4 have been inserted among
8
the main text.
9
DOI of published version: 10.1111/jfb.14235
10
1
Department of Animal and Plant Sciences, University of Sheffield, Sheffield S10 2TN, United
Kingdom
2
Academic Unit of Reproductive and Developmental Medicine, Department of Oncology and
Metabolism, University of Sheffield, Sheffield S10 2SF, United Kingdom
3
Institute of Biomedicine and Translational Medicine, Department of Pathophysiology, University of
Tartu, Tartu, Estonia

2
Abstract
11
Exposure to elevated stress may have important consequences for the long-term fitness of animals, as
12
well as that of their offspring via maternal effects. For instance, a change in the level of stress
13
experienced by a mother, such as an increase in predation risk, may shift the levels of resources invested
14
in their offspring. Here, we tested the effects of gestational stress, principally in the form of alarm cue
15
extracted from the skin of conspecifics, on reproduction in female guppies (Poecilia reticulata) and the
16
growth and behaviour of their offspring. Offspring from mothers exposed to alarm cue exhibited stunted
17
growth in the first few days postpartum which appeared to be mediated by shortening of the gestation
18
period, the length of which directly correlated with growth rate within the first six days postpartum.
19
Mature offspring did not differ in behaviour or stress responses compared to controls and so the effects
20
of maternal predation stress did not appear to persist into adulthood. A different form of gestational
21
stress, dietary restriction, did not significantly affect offspring growth, though brood size was reduced,
22
suggesting that the effects of predation stress were not mediated by differences in resource demand or
23
consumption.
24

3
Introduction
25
As aspects of the environment such as resources, competition, and predation are in constant flux, it is
26
essential for organisms to respond plastically to environmental changes. Frequently, these changes
27
exceed the reach of homeostasis and induce stress, necessitating adaptive responses to restore balance
28
(Barton, 2006; Chrousos, 2009; Ord et al., 2017). Appropriate stress response physiology is essential
29
for survival, and plasticity in response to stressful conditions is well-documented (Schulte, 2014).
30
Moreover, stress has been identified as a driver of maternal effects, whereby environmentally-induced
31
plasticity is extended from mother to offspring (Mateo, 2014).
32
In both mammals and fish, stress responses are characterised by increases in the glucocorticoid, cortisol,
33
which interacts with receptors in various tissues to facilitate adaptive changes necessary to restore
34
homeostasis (Barton, 2006). Several studies, both in field and experimental settings and in a variety of
35
species, have shown evidence that predation, or more specifically, the perceived risk of predation, is a
36
form of stress, in that it induces physiological stress responses characterised by increases in
37
glucocorticoid hormones (Clinchy et al., 2013). Thus, predator cues, such as the scent of predators (e.g.
38
fox odour for rodents; Howerton et al. 2013), have found favour among experimental physiologists as
39
‘model’ stressors. In some species of fish, chondroitin fragments released from the skin when damaged
40
trigger an innate ‘alarm response’ (Mathuru et al., 2012) which is characterised by anti-predator
41
behaviours, such as freezing and shoaling (Speedie & Gerlai, 2008; Egan et al., 2009; Stephenson,
42
2016), and increased cortisol levels (Mathuru 2016, Eachus et al. 2017). Although providing an
43
immediate benefit to the organism in life-threatening situations, stress responses are energetically costly
44
and occur at the expense of other aspects of physiology, such as immune function (Dhabhar, 2009), and
45
thus incur a fitness cost. Therefore, sustained activation of the stress response, referred to as chronic
46
stress, may lead to a maladaptive physiological state (Prunet et al., 2008) that is further exasperated by
47
the overproduction of glucocorticoids (Lupien et al., 2009).
48
Clinchy et al. (2013) have compiled extensive evidence in support of the ‘ecology of fear’ – the notion
49
that by inducing a state of chronic stress, elevated predation risk can have important influences on
50
population demography even in the absence of direct killing. Importantly, predation stress can impact
51

4
on fitness via nutrition, firstly by suppressing foraging behaviour (an adaptive behavioural response)
52
(Torres-Dowdal et al., 2012; Elvidge et al., 2014, 2016) and thus reducing food intake, and secondly
53
because stress responses are energetically costly and thus increase the demand for food (Clinchy et al.,
54
2013). Therefore, predation stress can lead to maternal effects because it reduces the resources available
55
to offspring.
56
Several studies have shown that various forms of stress, including predation threat, social subordination,
57
and dietary restriction experienced during egg provisioning influence offspring traits in a number of
58
vertebrate genera, including birds (Coslovsky & Richner, 2011) and fish (Reznick & Yang, 1993;
59
Giesing et al., 2011; Eaton et al., 2015; Jeffrey & Gilmour, 2016). These pre-fertilisation effects could
60
be mediated by a variety of factors supplied to the egg by the mother, such as RNA (Adrian-Kalchhauser
61
et al., 2018), proteins and hormones (Blount et al., 2002; Verboven et al., 2003; Groothuis et al., 2005;
62
Niall Daisley et al., 2005), or nutrients (Blount et al., 2006). In viviparous (live-bearing) animals, the
63
gestation period is a highly sensitive phase and offspring can be susceptible to environmentally-induced
64
changes in developmental programming which are thought to have their basis in ancient adaptive
65
mechanisms (Bateson et al., 2014). Several studies on laboratory rodents have demonstrated long-term
66
influences of gestational stress on offspring behaviour and physiology (Grace et al., 2011), while
67
epidemiological studies similarly show an increased risk of psychological illness (Painter et al., 2005;
68
Roseboom et al., 2006; Khashan et al., 2008). Similar effects of stress have been observed in wild
69
mammal populations, for example, in wild chimpanzees, maternal social status influences
70
glucocorticoid levels during pregnancy and physiological traits of the offspring (Murray et al., 2018).
71
Furthermore, viviparous reptiles have been found to be susceptible to maternal effects of gestational
72
stress, for example, exposure of common lizards to snake odour during gestation results in alterations
73
to tail morphology and thermoregulatory behaviour in the offspring (Bestion et al., 2014).
74
As well as mammals and reptiles, viviparity has also evolved in teleost fish, but the effects of gestational
75
stress have been less studied in this group. The guppy (Poecilia reticulata) is a poeciliid fish with
76
lecithotrophic viviparity, whereby all necessary nutrition to support embryonic development is
77
deposited in the yolk prior to fertilisation (Reznick et al., 1996). The guppy therefore offers an
78

5
interesting model in which to study gestational influences of stress such as predation and dietary
79
restriction, which can be disentangled from direct nutritional influences on the embryo.
80
Previous work has shown that female guppies shorten their gestation time in response to elevated
81
predation stress (Evans et al., 2007), which may enhance maternal survival due to the potentially
82
detrimental effect of pregnancy on escape ability. Although this is also collectively beneficial to
83
offspring as it provides the opportunity to disperse, a reduced gestation period likely incurs a fitness
84
cost given that conditions in the maternal tract are optimised for embryonic development. Concordantly,
85
in high-predation environments, female guppies have shorter gestation lengths and produce smaller,
86
less developed offspring than their low-predation counterparts (Dial et al., 2017). Furthermore, previous
87
observations indicate that gestating guppies abort offspring in response to dietary restriction, suggesting
88
that even though embryos have already been provisioned with nutrients, they can be affected by a state
89
of physiological stress induced by resource detriment (Hester, 1964; Reznick & Yang, 1993). Although
90
Evans et al. (2007) did not find that predation stress directly affected offspring performance traits,
91
neonate swimming speed was negatively correlated with gestation length, suggesting maternal stress
92
influenced their behaviour, at least indirectly. Despite evidence that gestational stress has an effect on
93
offspring, the longer-term consequences of this in live-bearing fish have not been studied.
94
Here, we determined the effect of gestational stress, principally in the form of predation threat via
95
conspecific-derived alarm cues, on female behaviour and reproduction in the guppy, and the long-term
96
impact on the growth, behaviour and stress response of their offspring. We predicted that gestational
97
stress would have a negative impact on female reproduction, as seen previously, and that there would
98
be long-term effects on offspring phenotype. Since predation risk acts to suppress feeding activity in
99
prey species (Buskirk & Yurewicz, 1998) and increases the demand for food due to the energetic
100
expense of the stress response (Hawlena & Schmitz, 2010), maternal effects of predation stress could
101
be mediated by differences in food consumption or utilisation by the mother. We therefore also tested
102
for effects of gestational dietary restriction on female reproduction and offspring growth to assess
103
whether predation-induced effects could be mediated by nutritional influences.
104

6
Materials and methods
105
Ethics statement
106
All animal work complied with the UK Animals (Scientific Procedures) Act (ASPA) and was carried
107
out under a UK Home Office licence for the use of animals in scientific procedures (project license
108
number 40/3704). The study did not involve collection of animals from their natural habitat. All animals
109
subject to experimental procedures (predation stress and dietary restriction), and their offspring, were
110
humanely euthanised by anaesthetic overdose after the completion of measurement collection. The
111
stressors applied in the study were chosen to simulate stressful conditions that may be encountered by
112
animals in a natural setting (elevated predation risk and resource detriment) and are not known to cause
113
pain, severe distress or lasting harm to the animals. The study did not involve surgical procedures or
114
the use of neuromuscular blocking agents.
115
Animals and housing
116
Wild-type Trinidadian guppies, descended from a population obtained from The University of Leicester
117
Botanic Garden, UK, and reared in our laboratory for several generations, were housed in aquaria at
118
27˚C, maintained under a 12:12h photoperiod, and fed ad libitum twice per day as per the standard
119
aquarium feeding regime of freshly hatched brine shrimp (Artemia) nauplii (ZM Systems, Winchester,
120
UK) three days per week (Mondays, Wednesdays, and Fridays) and flake food (ZM flake; ZM Systems,
121
Winchester, UK) on every other day. All animals in the experiments were fed according to this same
122
regime with the exception of gestating females in the dietary restriction experiment (see Maternal
123
dietary restriction).
124
To obtain unmated females, fry were collected from stock tanks and reared in a separate tank, and any
125
males, identified by the presence of a gonopodium, were removed, leaving females only. At maturity
126
(> 4 months old), individuals of similar size were selected from this unmated female population for use
127
in experiments (12 and 16 for predation and diet experiments, respectively). The selected females were
128
housed individually in 1L tanks (19 x 13 x 12 cm) for one month to allow for acclimation. Each tank
129

7
was supplied with an air stone and water was changed weekly. A plastic plant was provided for
130
enrichment.
131
Each female was weighed before the start of each experimental period. For mating, each female was
132
paired with a different mature stock male on three consecutive days (males were not re-used), in order
133
to control for sperm quality and to increase the probability of mating. Paternal characteristics were not
134
evaluated but care was taken that males used for mating appeared healthy and were of similar size.
135
Alarm substance extraction and validation
136
Alarm substance was derived from mature female guppies based on established methods (Mathuru et
137
al., 2012; Schirmer et al., 2013). In brief, for 3 mL of alarm substance, five fish were euthanised and 5-
138
10 lacerations were made to the epidermis on both sides. Fish were then placed in a tube with 3 mL
139
water and gently shaken. The water containing skin extract was incubated at 95˚C for 16 hours,
140
centrifuged, and the supernatant was extracted and filtered. 200µL of prepared extract added to 600mL
141
water was sufficient to induce behavioural responses in experimentally naïve animals, in which it
142
induced significant reductions in swim distance in females (permutation test, Z = 2.09, P = 0.037; N =
143
3 alarm, 3 control) and males (Z = 2.11, P = 0.035; N = 3 alarm, 3 control) (Fig. S1b).
144
As we exposed six animals on each of 23 days (see Maternal predation stress) we prepared 30 mL of
145
extract (10 batches, total 50 fish) for exposing gestating females. An additional four batches (20 fish)
146
were prepared for exposing offspring (see Offspring behavioural stress response).
147
Maternal predation stress
148
As repeated exposure to alarm cue could be considered as a proxy for predation stress, the maternal
149
alarm cue treatment is henceforth referred to as predation stress. Twelve females were used; six were
150
assigned randomly to the predation stress treatment, and six as controls. After the three-day mating
151
period, predation females were exposed to alarm substance (200 µL in 600 mL) in separate exposure
152
tanks for 20 minutes each day for 23 days (Fig. S1, a & c), to encompass the majority of an average
153
gestation length (Reznick et al., 1996). Simultaneously to when predation females received alarm
154

8
treatment, control females were placed in separate tanks containing only aquarium water. One control
155
female died during the experiment, and another control failed to produce a brood after 60 days, so that
156
in total, six predation stress and four control broods were derived. Brood sizes ranged from 4 to 30 fry
157
(Table S1). Twelve offspring from each brood of more than 12 and all the offspring from each brood
158
of less than 12 were reared. To control for population density effects, each fry was housed in an
159
individual compartment of a breeding trap (20 x 9 x 10 cm) divided into three equal-sized compartments
160
(each compartment measuring approx. 6.6 x 9 x 10 cm). Four such divided breeding traps were kept in
161
trays (45 x 29 x 13 cm) containing 5 L water that could flow freely into each compartment. Each tray
162
was supplied with a sponge filter, and water was changed weekly. Fry were fed as per the standard
163
feeding regime (see Animals and housing). After 21 days, partitions were removed so that juveniles
164
from the same brood could interact in the breeding traps in groups of 2-3. Juveniles were checked
165
regularly for the appearance of sexual characteristics and fish were placed in separate compartments
166
with their same-sex siblings; males were housed in groups of 2-3 and females in groups of 2-4.
167
Offspring size measurements
168
Three offspring per brood were measured every three days for 15 days postpartum by placing them in
169
a petri dish over graph paper and photographing them. Photos were imported into ImageJ (Schneider et
170
al., 2012) and body area (excluding tail) was measured. Although the same three individuals were
171
measured for each brood, individual identity was not tracked, and therefore a representative mean body
172
area was taken for each brood for each measurement day (i.e. the mean body area of the three measured
173
individuals). Growth curves were derived only for the first 15 days because of the probability of body
174
size being influenced by sexual differentiation beyond this stage (Magellan & Magurran, 2009). At
175
approximately two months old, offspring were weighed individually in a plastic beaker containing 20
176
mL water.
177
Open field testing
178
Predation stress and control mothers were tested for anxiety-like behaviour seven days after the final
179
exposure to alarm substance using the open field test (Rehnberg et al., 1987; Ariyomo et al., 2013). For
180

9
each trial, an individual fish was placed into a rectangular tank (40 x 25 x 25 cm) containing 3 L of
181
water. The test tank was covered on all sides with paper to prevent external visual stimuli. Given that
182
the test is intended to assess an individual’s response to an unfamiliar environment, no acclimation time
183
was allocated. Each fish was video-recorded from above for 5 mins using a Panasonic HC-X920
184
camcorder. Water was replaced between each trial. The same testing procedure was performed on the
185
offspring of these females in a smaller tank (25 x 25 x 25 cm) at 50-70 days postpartum, and up to three
186
males and three females from each parent were used (a total of 21 control and 30 predation stress
187
offspring). Videos were scored manually for thigmotaxis, a common measure of anxiety-like behaviour,
188
which was quantified as the % time spent in the outer grid cells of an 8 x 8 grid overlaid digitally onto
189
the tank floor (3 x 5 cm rectangular cells in case of mothers and 3 x 3 cm square cells in case of offspring,
190
due to difference in shape of tank used).
191
Offspring behavioural stress response
192
We measured the effect of alarm substance exposure on swim distance in the adult offspring in order to
193
determine whether gestational stress affected the behavioural stress response. From most broods, up to
194
two male and two female offspring were exposed to either alarm substance (200 µL in 600 mL) or
195
aquarium water only (control) in tanks (19 x 13 x 12 cm) for 10 mins. See Table S2 for exact numbers
196
of offspring exposed from each brood. Due to limited numbers of offspring in some broods, animals
197
were selected randomly irrespective of whether they had been previously tested in the open field test.
198
However, for a given brood, we allowed at least three days to elapse between open field testing and
199
stress response testing. Alarm substance and control exposures were conducted simultaneously.
200
Maternal dietary restriction
201
In a separate experiment, using different females to the predation experiment, mature unmated females
202
were assigned randomly to one of two diet regimes and fed with quantified amounts of brine shrimp
203
paste (50 µL for high food diet and 10 µL for low food diet, respectively) once per day for a total of 31
204
days. Food quantities were derived from Reznick et al (1996), who reported that the high food level
205
was close to ad libitum, while the low food level was sufficient to sustain reproduction in mature female
206

10
guppies. These diets were maintained from two days before the three-day mating period (mating carried
207
out as described in Animals and housing), until 26 days after mating to cover the average gestation
208
length (Reznick et al, 1996). After this period, the diet of all females was restored to ad libitum brine
209
shrimp and flake food as per the standard feeding regime. Eight females comprised each of the high and
210
low diet groups. One high diet female and 3 low diet females failed to produce any offspring after 65
211
days, so that broods were derived from 7 high diet and 5 low diet females. Females from the high and
212
low diet groups which produced broods did not differ in wet weight at the start of the experiment
213
(permutation test, Z = 0.18, p-value = 0.855).
214
Brood sizes ranged from 2-18 fry (Table S2). Following parturition, offspring were not housed
215
individually as described for the predation experiment but were separated into groups of no more than
216
eight per 1 L tank (19 x 13 x 12 cm). Offspring were photographed on the day of birth and approximately
217
every three days for the first two weeks postpartum. All individuals in a brood were photographed
218
together. Offspring were weighed at approximately two months old, as for the predation experiment.
219
Statistical analyses
220
All statistical tests were conducted using ‘R’ version 3.5.1 (R Development Core Team, 2011). Due to
221
low sample sizes, non-parametric permutation tests (asymptotic general independence tests) from the
222
‘coin’ package (Hothorn, 2019) were used to assess statistical significance of differences in maternal
223
reproductive traits (gestation length, fecundity) for both predation and diet experiments.
224
For analyses of offspring parameters, linear mixed effects models fit by residual maximum likelihood
225
(REML) were performed, using the ‘lme4’ package (Bates et al., 2015), with nested random effect
226
terms, where applicable. Statistical significance of mixed model terms was evaluated using the
227
‘lmerTest’ (Kuznetsova et al., 2017) and ‘pbkrtest’ packages (Hakekoh & Hojsgaard, 2014) using T-
228
tests and F-tests with Kenward-Roger approximation of degrees of freedom.
229
Models with second-order polynomial regression were used to evaluate the effect of maternal treatment
230
on offspring growth, with ‘offspring age’, ‘treatment’, and ‘age x treatment’ as fixed effects terms, and
231
‘mother ID’ as a random effect term. Inclusion of the second order polynomial was found to improve
232

11
the model fit as indicated by lower AIC scores for both predation and diet models. To approximate a
233
normal distribution, offspring growth data were log-transformed prior to fitting the models.
234
For analyses of adult offspring wet weight at maturity and thigmotaxis behaviour, offspring sex was
235
included as a fixed term in the models, with mother ID included as a random effect term.
236
For behavioural stress response (total swim distance following alarm or control exposure), sex,
237
offspring treatment (alarm or control / water), maternal treatment (control or predation), and the
238
interaction between the latter two (offspring treatment x maternal treatment) were fixed effect terms,
239
while mother ID was again a random term. To approximate a normal distribution, the swim distance
240
variable was square root-transformed prior to model fitting.
241
Further details on sample sizes used in each analysis can be found in Table S3, and details of the mixed
242
effects models can be found in Table S4.
243
Results
244
Reproductive and life history parameters
245
Maternal predation stress was associated with a trend towards reduced gestation length which was on
246
average 14.2 ± 4.27 days shorter (mean ± SE) for females exposed to predation stress, although the
247
reduction was non-significant (permutation test, Z = 1.83, p = 0.066; Fig. 1a). There was no effect of
248
maternal stress on fecundity, as measured by number of offspring per unit body weight (Z = 1.36, p =
249
0.17; Fig. 1b).
250
Although there was no noticeable difference in size at birth, predation stress offspring were 0.97 ± 0.27
251
mm2 smaller (mean ± SE) than control offspring during the first two weeks postpartum. While control
252
offspring exhibited essentially linear growth, the trajectory of predation stress offspring growth was
253
noticeably more curve-shaped given that predation stress offspring exhibited very little change in size
254
for several days (Fig. 1c). A linear mixed model revealed a significant effect of treatment on offspring
255
size overall, and a significant age x treatment interaction (Table 1). The interaction appeared to be
256
driven by reduced growth in the first few days postpartum.
257

12
Due to the apparent trend towards reduced gestation length in the predation stress treatment, we also
258
visualised the change in brood mean body area with developmental age (i.e. age from conception, Fig.
259
1d). This caused the majority of brood growth trajectories to overlap and revealed that the weakest early
260
growth occurred at the youngest developmental ages, with the most premature brood even displaying a
261
reduction in mean body area during the first few days.
262
263
Figure 1. Poecilia reticulata maternal reproductive traits and offspring growth in the context of predation
264
stress encountered during gestation. (a) gestation length in days (p-value derived from a nonparametric
265
permutation test), and (b) fecundity (no. offspring per gram body weight) for control and predation stress mothers.
266
p-value derived from a nonparametric permutation test. (c) Offspring body size from days 0-15 postpartum for
267
control and predation stress broods, plotted as mean individual body area (mean of individual body areas in each
268
brood on each day) on log-scale. Curve fits with 95% CIs were derived from linear mixed effects models with
269
second-order polynomial regression. (d) Change in body size with developmental age (postpartum age plus
270
gestation length), plotted as mean individual body area for each brood on each day (log-scale).
271

13
Table 1. Linear mixed effects models describing the effect of maternal predation stress and maternal dietary
272
restriction on offspring size (mean individual body area for each brood) measured over the first two weeks post-
273
parturition. Models were fitted to log-transformed data. Both F-test and T-test results of fixed effect terms are
274
shown, with F-values derived from the effect of adding a term last to the model, while T-values derive from
275
comparing the parameter estimate to the intercept in the absence of other factors. Variance and standard deviations
276
of random effect terms are shown. Age1 and Age2 denote first and second order polynomial transformations of the
277
Age term, respectively.
278
Maternal predation stress (N = 4 control / 6 predation stress broods)
Fixed effects
ndf
ddf
F
p
Estimate
SE
T
p
Intercept
7.76
1.86
0.03
57.23
< 0.001
Treatment (Predation)
1
7.84
12.75
0.01
-0.15
0.04
3.57
0.01
Age2
2
35.83
178.49
< 0.001
Age1
35.30
1.91
0.15
13.14
< 0.001
Age2
36.24
-0.07
0.14
0.49
0.63
Treatment x Age2
2
35.83
4.27
0.02
Treatment x Age1
35.54
-0.30
0.19
1.62
0.12
Treatment x Age2
36.13
0.45
0.19
2.40
0.02
Random effects
Variance
SD
Mother
0.003
0.051
Residual
0.008
0.090
Maternal dietary restriction (N = 7 high diet / 5 low diet broods)
Fixed effects
ndf
ddf
F
p
Estimate
SE
T
p
Intercept
10.02
1.79
0.04
40.64
< 0.001
Diet (low diet)
1
9.96
1.29
0.28
0.08
0.07
1.14
0.28
Age2
2
45.18
213.53
< 0.001
Age1
45.29
1.67
0.12
14.28
< 0.001
Age2
45.38
-0.05
0.12
0.47
0.64
Diet x Age2
2
45.18
2.03
0.14
Diet x Age1
45.14
0.33
0.18
1.84
0.07
Diet x Age2
45.21
0.15
0.18
0.85
0.40
Random effects
Variance
SD
Mother
0.012
0.109
Residual
0.008
0.087
279

14
Although there remained a trend towards smaller offspring from predation stress mothers, offspring size
280
at maturity was not significantly affected by maternal predation stress (F1,8.3 = 2.05, p = 0.19; Fig. 2).
281
282
Figure. 2. Wet weight (mg) of adult female (left) and male offspring (right) from control and predation stress
283
mothers, measured at approx. two months postpartum. ‘n’ represents total number of offspring of each sex that
284
were measured from 4 control and 6 predation stress broods.
285
In the diet experiment, while there was no effect of maternal diet on gestation length (Fig. S2a), females
286
fed on the restricted diet produced significantly fewer offspring per unit body weight (permutation test,
287
Z = 2.14, p = 0.03; Fig. S2b). Offspring size throughout early development was not affected by maternal
288
diet (Fig. S2c), however, a mixed effects model (Table 1) revealed a marginally non-significant
289
interaction between maternal diet and the first order polynomial of age (T45.1 = 1.84, p = 0.07), reflecting
290
a slight increase in the linear slope of the growth line in low diet compared to high diet broods. Offspring
291
weight at maturity was unaffected by dietary restriction (F1,6.9 = 0.06, p = 0.81; Fig. S2d).
292
To further discern whether the effect of predation stress on offspring growth may have been driven by
293
reduced gestation length, we fitted separate linear regression lines (body area ~ postpartum age) for
294
each brood using data from the first six days and extracted the regression coefficient for each brood as
295
a measure of growth rate (i.e. per-day increase in the mean of individual body areas for each brood)
296

15
(Fig. S3, Table S5). When data were combined across both predation and diet experiments, there was a
297
moderate positive correlation between offspring growth rate and gestation length (Pearson’s r = 0.53,
298
N = 22, p = 0.01; Fig. 3).
299
300
Figure 3. Growth rates of Poecilia reticulata broods as a function of maternal gestation length. Correlation
301
between brood growth rates (per day increase in mean individual body area for each brood) within the first six
302
days postpartum and maternal gestation length for 22 Poecilia reticulata broods from four treatment groups across
303
two experiments. Each point represents a single brood. The experiment and treatment group from which a given
304
brood derives are indicated by the combination of shape and shade of the point: circles represent broods from the
305
diet experiment, triangles represent broods from the predation stress experiment; accordingly, black points denote
306
high diet (diet exp.) or control (pred. stress exp.) broods, while grey points denote low diet (diet exp.) or predation
307
stress (pred. stress exp.) broods depending on the experiment. Pearson’s correlation coefficient and corresponding
308
p value are shown.
309

16
Maternal and offspring behaviour (maternal predation stress)
310
One week following the final alarm exposure, thigmotaxis (% time at periphery) was significantly
311
reduced in the predation females compared to controls (permutation test, Z = 2.65, p = 0.008). However,
312
there was no effect of maternal predation on offspring thigmotaxis (T-test of mixed effects model with
313
Kenward-Roger approximation of df, T7.32 = 0.85, p = 0.421) (Fig. 4a).
314
Offspring behavioural stress response (maternal predation stress)
315
Swim distance was generally lower in offspring exposed to alarm substance (Fig. 4b). A linear mixed
316
effects model (Table S6) revealed an overall significant effect of alarm substance on offspring swim
317
distance (F1,48.3 = 7.62, p = 0.008), but no effect of maternal predation stress (F1,6.77 = 0.011, p = 0.75)
318
or interaction between maternal and offspring treatment (F1,48.3 = 0.18, p = 0.67). Sex had a significant
319
additive effect, with males travelling on average 441.8 ± 155.6 cm more than females (T49.7 = 2.31, p =
320
0.03).
321

17
322
Figure 4. Behavioural traits of adult female Poecilia reticulata and their offspring. (a) Thigmotaxis (% time
323
at periphery) in adult female (F0) Poecilia reticulata exposed to predation stress (daily exposure to conspecific-
324
derived alarm substance for 23 days during gestation) and corresponding non-exposed controls, and their mature
325
offspring (F1). For F0, ‘N’ represents the number of biological replicates (mothers), while for F1 ‘n’ represents
326
the number of pseudoreplicates (offspring) in each group, derived from 4 control and 6 predation stress mothers.
327
** p < 0.01, permutation test. (b) Behavioural stress response of Poecilia reticulata offspring derived from control
328
and predation stress mothers, measured as total distance (cm) travelled over 10 minutes of exposure to either
329
conspecific-derived alarm substance or water / control. ‘n’ represents the number of offspring in each group,
330
derived from 4 control mothers and 6 predation stress mothers. p-values derived from post-hoc T-tests of a
331
mixed effects model.
332

18
Discussion
333
By inducing sustained physiological alterations, exposure to stress in the form of predation cues may
334
have long-lasting influences on individuals and their offspring. Such changes may be driven by many
335
aspects of physiology, life history, or behaviour, or by differential utilisation of resources in the face of
336
shifting demands.
337
Using a purified form of conspecific-derived alarm substance to simulate elevated predation, we found
338
that this form of stress was associated with altered maternal behaviour and a trend towards shortened
339
gestation length (mean reduction of 14.2 days). Importantly, although non-significant, this latter
340
observation is concordant with previously reported effects of predator cues (Evans et al., 2007) and
341
high predation environments (Dial et al., 2017) on gestation length in this species. Together, these
342
observations reflect the capacity of alarm substance to induce sustained alterations to maternal
343
physiology and neurocircuitry. Offspring from mothers exposed to predator cues (henceforth predation
344
stress offspring) grew more slowly during the early period of postnatal growth, and the slowest initial
345
growth was exhibited by those broods which were born at the earliest developmental ages, suggesting
346
that premature birth contributed to reduced early growth rate. Accordingly, maternal gestation length
347
was positively correlated with early growth rate as discerned using data from both the predation and
348
diet experiments. Thus, although the effect of predation stress on gestation length was not significant,
349
it can be inferred that growth retardation was a consequence of gestational stress driven by accelerated
350
parturition. A possible explanation for this is that early release into open water, and the subsequent
351
resource expenditure on swimming, meant that predation stress offspring incurred a greater metabolic
352
deficit at an earlier developmental stage, when yolk resources would otherwise be devoted to growth
353
and other aspects of development. Females can enhance offspring swimming performance by delaying
354
parturition to allow further intrauterine development (Shine & Olsson, 2003; Evans et al., 2007). It may
355
therefore be inferred that reduced growth rate occurs because offspring with restricted intrauterine
356
development time have less well-developed sensory systems, musculature, and digestive systems at
357
birth, thus impairing their ability to exploit available food resources. Predation stress offspring growth
358

19
appeared to increase towards the end of the observation period and there was no difference in body size
359
in mature offspring, so some compensatory growth must have occurred (Auer, 2010).
360
As predation risk acts both to suppress feeding activity and increase resource demand due to the
361
energetic cost of stress responses, one hypothesis was that predation-induced effects on maternal and
362
offspring traits are mediated by resource intake or demand. However, the results of the diet experiment
363
refute this hypothesis and show that gestational dietary restriction leads to different reproductive
364
outcomes, that is, reduced brood size, but not gestation length, and no effect on offspring growth. We
365
note that the interaction between maternal diet and offspring age on offspring size during the first 15
366
days postpartum was only marginally non-significant and therefore should not be overlooked. However,
367
the trend may have been influenced by differences in population density, which was not controlled for
368
as stringently for diet experiment broods as for predation stress experiment broods. Reduction in the
369
brood size with dietary restriction is concordant with previous findings (Hester, 1964; Reznick & Yang,
370
1993), and suggests that the restricted diet was sufficient to induce a state of stress. That there was no
371
obvious effect of dietary restriction on offspring growth suggests that the effects of maternal predation
372
stress were not driven by behaviourally-mediated differences in food consumption, an increased
373
physiological demand for food (Hawlena & Schmitz, 2010; Clinchy et al., 2013), or more general stress-
374
induced physiological changes that may be induced by either predation stress or nutritional stress. Taken
375
together, the results of the two experiments illustrate that different forms of stress can influence life
376
history traits via different mechanisms.
377
As well as examining the effects of different forms of stress on reproduction and offspring life history,
378
we examined the influence of predation stress on anxiety-like behaviour (thigmotaxis) of exposed
379
females and their offspring, and the behavioural stress response of the offspring. Mothers exposed to
380
predation stress showed significantly reduced thigmotaxis, suggesting a lowered anxiety-like
381
behaviour, which was not expected. The reduction in thigmotaxis may reflect habituation following
382
repeated exposure to a stressor, as there is evidence that such habituation may arise due to plasticity of
383
the hippocampus and can lead to blunted stress responses (McEwen, 2006; Grissom & Bhatnagar,
384
2009). No changes in anxiety-like behaviour were detected in the offspring of predation stress mothers,
385

20
suggesting that maternal predation stress did not have a major impact on the long-term
386
neurodevelopment of the offspring. Similarly, marked suppression of swim distance following alarm
387
substance exposure was observed in offspring from both control and predation stress mothers to a
388
similar degree, and there was no significant interaction between maternal and offspring treatments. It
389
therefore appears that maternal predation stress did not induce appreciable changes in offspring
390
behavioural stress response.
391
We concede that because we did not track individual identity, we were unable to effectively assess
392
within-brood variability in growth. Nevertheless, the use of representative brood means was sufficient
393
to reveal a difference in growth trajectories between control and predation broods. One other possible
394
confound may have arisen because offspring were selected for the stress response test randomly
395
irrespective of their prior use in open field testing, and therefore one cannot rule out whether previous
396
exposure to behavioural testing may have impacted stress responses. However, we considered this
397
unlikely given that we allowed adequate time (no less than three days) to elapse between the two sets
398
of tests. Finally, the use of a mixed feeding regime (live food on some days and flake food on others)
399
may have contributed to differences in offspring growth due to broods being born on different days.
400
In summary, we have shown that gestational stress in the context of predation impacts of on offspring
401
early growth rates, possibly driven by a shortening of the gestation period, the length of which appears
402
to be an important determinant of early growth rate. Furthermore, the effects of predation stress did not
403
appear to be mediated by dietary influences thus impacted maternal physiology, and thus the offspring,
404
via more specific mechanisms. However, the evidence for long-term effects of gestational stress on
405
mature offspring was limited, implying that the effects were largely transient. Nevertheless, our study
406
illustrates how maternal stress at a critical reproductive stage can have an effect on offspring life history
407
and shows how predation cues can act as an ecological stressor.
408
References
409
Adrian-Kalchhauser, I., Walser, J. C., Schwaiger, M. & Burkhardt-Holm, P. (2018) RNA Sequencing
410
of Early Round Goby Embryos Reveals That Maternal Experiences Can Shape the Maternal RNA
411

21
Contribution in a Wild Vertebrate. BMC Evolutionary Biology 18, 34.
412
Ariyomo, T. O., Carter, M. & Watt, P. J. (2013) Heritability of Boldness and Aggressiveness in the
413
Zebrafish. Behavior Genetics 43, 161–167.
414
Auer, S. K. (2010) Phenotypic Plasticity in Adult Life-History Strategies Compensates for a Poor
415
Start in Life in Trinidadian Guppies (Poecilia Reticulata). The American naturalist 176, 818–829.
416
Barton, B. A. (2006) Stress in Fishes: A Diversity of Responses with Particular Reference to Changes
417
in Circulating Corticosteroids. Integrative and Comparative Biology 42, 517–525.
418
Bates, D., Maechler, M., Bolker, B. & Walker, S. (2015) Fitting Linear Mixed-Effects Models Using
419
Lme4. Journal of Statistical Software 67, 1–48.
420
Bateson, P., Gluckman, P. & Hanson, M. (2014) The Biology of Developmental Plasticity and the
421
Predictive Adaptive Response Hypothesis. The Journal of Physiology 592, 2357–2368.
422
Bestion, E., Teyssier, A., Aubret, F., Clobert, J. & Cote, J. (2014) Maternal Exposure to Predator
423
Scents: Offspring Phenotypic Adjustment and Dispersal. Proceedings of the Royal Society B:
424
Biological Sciences 281.
425
Blount, J. D., Surai, P. F., Nager, R. G., Houston, D. C., Møller, A. P., Trewby, M. L. & Kennedy, M.
426
W. (2002) Carotenoids and Egg Quality in the Lesser Blackbacked Gull Larus Fuscus: A
427
Supplemental Feeding Study of Maternal Effects. Proceedings. Biological sciences 269, 29–36.
428
Blount, J. D., Metcalfe, N. B., Arnold, K. E., Surai, P. F. & Monaghan, P. (2006) Effects of Neonatal
429
Nutrition on Adult Reproduction in a Passerine Bird. Ibis 148, 509–514.
430
Buskirk, J. Van & Yurewicz, K. L. (1998) Effects of Predators on Prey Growth Rate: Relative
431
Contributions of Thinning and Reduced Activity. Oikos 82, 20.
432
Chrousos, G. P. (2009) Stress and Disorders of the Stress System. Nat Rev Endocrinol 5, 374–381.
433
Clinchy, M., Sheriff, M. J. & Zanette, L. Y. (2013) Predator-Induced Stress and the Ecology of Fear.
434
Functional Ecology 27, 56–65.
435

22
Coslovsky, M. & Richner, H. (2011) Predation Risk Affects Offspring Growth via Maternal Effects.
436
Functional Ecology 25, 878–888.
437
Dhabhar, F. S. (2009) Enhancing versus Suppressive Effects of Stress on Immune Function:
438
Implications for Immunoprotection and Immunopathology. Neuroimmunomodulation 16, 300–317.
439
Dial, T. R., Reznick, D. N. & Brainerd, E. L. (2017) Heterochrony in the Evolution of Trinidadian
440
Guppy Offspring Size: Maturation along a Uniform Ontogenetic Trajectory. Proceedings. Biological
441
sciences 284.
442
Eachus, H., Bright, C., Cunliffe, V. T., Placzek, M., Wood, J. D. & Watt, P. J. (2017) Disrupted-in-
443
Schizophrenia-1 Is Essential for Normal Hypothalamic-Pituitary-Interrenal (HPI) Axis Function.
444
Human Molecular Genetics 26, 1992–2005.
445
Eaton, L., Edmonds, E. J., Henry, T. B., Snellgrove, D. L. & Sloman, K. A. (2015) Mild Maternal
446
Stress Disrupts Associative Learning and Increases Aggression in Offspring. Hormones and Behavior
447
71, 10–15.
448
Egan, R. J., Bergner, C. L., Hart, P. C., Cachat, J. M., Canavello, P. R., Elegante, M. F., Elkhayat, S.
449
I., Bartels, B. K., Tien, A. K., Tien, D. H., et al. (2009) Understanding Behavioral and Physiological
450
Phenotypes of Stress and Anxiety in Zebrafish. Behavioural brain research 205, 38–44.
451
Elvidge, C. K., Ramnarine, I. & Brown, G. E. (2014) Compensatory Foraging in Trinidadian Guppies:
452
Effects of Acute and Chronic Predation Threats. Current Zoology 60, 323–332.
453
Elvidge, C. K., Chuard, P. J. C. & Bbrown, G. E. (2016) Local Predation Risk Shapes Spatial and
454
Foraging Neophobia Patterns in Trinidadian Guppies. Current Zoology 62, 457–462.
455
Evans, J. P., Gasparini, C. & Pilastro, A. (2007) Female Guppies Shorten Brood Retention in
456
Response to Predator Cues. Behavioral Ecology and Sociobiology 61, 719–727.
457
Giesing, E. R., Suski, C. D., Warner, R. E. & Bell, A. M. (2011) Female Sticklebacks Transfer
458
Information via Eggs: Effects of Maternal Experience with Predators on Offspring. Proceedings of the
459
Royal Society B: Biological Sciences 278, 1753–1759.
460

23
Grace, C. E., Kim, S. J. & Rogers, J. M. (2011) Maternal Influences on Epigenetic Programming of
461
the Developing Hypothalamic-Pituitary-Adrenal Axis. Birth Defects Research Part A - Clinical and
462
Molecular Teratology 91, 797–805.
463
Grissom, N. & Bhatnagar, S. (2009) Habituation to Repeated Stress: Get Used to It. Neurobiology of
464
Learning and Memory 92, 215–224.
465
Groothuis, T. G. G., Müller, W., von Engelhardt, N., Carere, C. & Eising, C. (2005) Maternal
466
Hormones as a Tool to Adjust Offspring Phenotype in Avian Species. Neuroscience & Biobehavioral
467
Reviews 29, 329–352.
468
Hakekoh, U. & Hojsgaard, S. (2014) A Kenward-Roger Approximation and Parametric Bootstrap
469
Methods for Tests in Linear Mixed Models–the R Package Pbkrtest. Journal of Statistical Software
470
59, 1–30.
471
Hawlena, D. & Schmitz, O. J. (2010) Physiological Stress as a Fundamental Mechanism Linking
472
Predation to Ecosystem Functioning. The American naturalist 176, 537–556.
473
Hester, F. J. (1964) Effects of Food Supply on Fecundity in the Female Guppy, Lebistes Reticulatus
474
(Peters). Journal of the Fisheries Research Board of Canada 21, 757–764.
475
Hothorn, T. (2019) Package ‘Coin’.
476
Howerton, C. L., Morgan, C. P., Fischer, D. B. & Bale, T. L. (2013) O-GlcNAc Transferase (OGT) as
477
a Placental Biomarker of Maternal Stress and Reprogramming of CNS Gene Transcription in
478
Development. Proceedings of the National Academy of Sciences of the United States of America 110,
479
5169–5174.
480
Jeffrey, J. D. & Gilmour, K. (2016) Correction: Programming of the Hypothalamic-Pituitary-
481
Interrenal Axis by Maternal Social Status in Zebrafish ( Danio Rerio ). The Journal of Experimental
482
Biology 219, 2963–2963.
483
Khashan, A. S., Abel, K. M., McNamee, R., Pedersen, M. G., Webb, R. T., Baker, P. N., Kenny, L.
484
C., Mortensen, P. B., JM, L., RS, G., et al. (2008) Higher Risk of Offspring Schizophrenia Following
485

24
Antenatal Maternal Exposure to Severe Adverse Life Events. Archives of General Psychiatry 65, 146.
486
Kuznetsova, A., Brockhoff, P. B. & Christensen, R. H. B. (2017) LmerTest Package: Tests in Linear
487
Mixed Effects Models. Journal of Statistical Software 82, 1–26.
488
Lupien, S. J., McEwen, B. S., Gunnar, M. R. & Heim, C. (2009) Effects of Stress throughout the
489
Lifespan on the Brain, Behaviour and Cognition. Nat Rev Neurosci 10, 434–445.
490
Magellan, K. & Magurran, A. E. (2009) The Effect of Social Environment during Ontogeny on Life
491
History Expression in the Guppy Poecilia Reticulata. Journal of Fish Biology 74, 2329–2337.
492
Mateo, J. M. (2014) Development, Maternal Effects, and Behavioral Plasticity. Integrative and
493
Comparative Biology 54, 841–849.
494
Mathuru, A. S., Kibat, C., Cheong, W. F., Shui, G., Wenk, M. R., Friedrich, R. W. & Jesuthasan, S.
495
(2012) Chondroitin Fragments Are Odorants That Trigger Fear Behavior in Fish. Current biology :
496
CB 22, 538–544.
497
McEwen, B. S. (2006) Plasticity of the Hippocampus: Adaptation to Chronic Stress and Allostatic
498
Load. Annals of the New York Academy of Sciences 933, 265–277.
499
Murray, C. M., Stanton, M. A., Wellens, K. R., Santymire, R. M., Heintz, M. R. & Lonsdorf, E. V.
500
(2018) Maternal Effects on Offspring Stress Physiology in Wild Chimpanzees. American Journal of
501
Primatology 80, e22525.
502
Niall Daisley, J., Bromundt, V., Möstl, E. & Kotrschal, K. (2005) Enhanced Yolk Testosterone
503
Influences Behavioral Phenotype Independent of Sex in Japanese Quail Chicks Coturnix Japonica.
504
Hormones and Behavior 47, 185–194.
505
Ord, J., Fazeli, A. & Watt, P. J. (2017) Long-Term Effects of the Periconception Period on Embryo
506
Epigenetic Profile and Phenotype: The Role of Stress and How This Effect Is Mediated. In Advances
507
in Experimental Medicine and Biology pp. 117–135.
508
Painter, R. C., Roseboom, T. J. & Bleker, O. P. (2005) Prenatal Exposure to the Dutch Famine and
509

25
Disease in Later Life: An Overview. Reproductive toxicology (Elmsford, N.Y.) 20, 345–352.
510
Prunet, P., Cairns, M. T., Winberg, S. & Pottinger, T. G. (2008) Functional Genomics of Stress
511
Responses in Fish. Reviews in Fisheries Science 16, 157–166.
512
R Development Core Team. (2011) R: A Language and Environment for Statistical Computing.
513
Vienna, Austria: R Foundation for Statistical Computing.
514
Rehnberg, B. G., Smith, R. J. F. & Sloley, B. D. (1987) The Reaction of Pearl Dace (Pisces,
515
Cyprinidae) to Alarm Substance: Time-Course of Behavior, Brain Amines, and Stress Physiology.
516
Canadian Journal of Zoology 65, 2916–2921.
517
Reznick, D. & Yang, A. P. (1993) The Influence of Fluctuating Resources on Life History : Patterns
518
of Allocation and Plasticity in Female Guppies. Source: Ecology Ecology 74, 2011–2019.
519
Reznick, D., Callahan, H. & Llauredo, R. (1996) Maternal Effects on Offspring Quality in Poeciliid
520
Fishes. American Zoologist 36, 147.
521
Roseboom, T., de Rooij, S. & Painter, R. (2006) The Dutch Famine and Its Long-Term Consequences
522
for Adult Health. Early human development 82, 485–491.
523
Schirmer, A., Jesuthasan, S. & Mathuru, A. S. (2013) Tactile Stimulation Reduces Fear in Fish.
524
Frontiers in behavioral neuroscience 7, 167.
525
Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. (2012) NIH Image to ImageJ: 25 Years of Image
526
Analysis. Nature methods 9, 671–675.
527
Schulte, P. M. (2014) What Is Environmental Stress? Insights from Fish Living in a Variable
528
Environment. The Journal of experimental biology 217, 23–34.
529
Shine, R. & Olsson, M. (2003) When to Be Born? Prolonged Pregnancy or Incubation Enhances
530
Locomotor Performance in Neonatal Lizards (Scincidae). Journal of evolutionary biology 16, 823–
531
832.
532
Speedie, N. & Gerlai, R. (2008) Alarm Substance Induced Behavioral Responses in Zebrafish (Danio
533

26
Rerio). Behavioural Brain Research 188, 168–177.
534
Stephenson, J. F. (2016) Keeping Eyes Peeled: Guppies Exposed to Chemical Alarm Cue Are More
535
Responsive to Ambiguous Visual Cues. Behavioral ecology and sociobiology 70, 575–584.
536
Torres-Dowdal, J., Handelsman, C. A., Reznick, D. N. & Ghalambor, C. K. (2012) Local Adaptation
537
and the Evolution of of Phenotypic Plasticity in Trinidadian Guppies (Poecilia Reticulata). Evolution
538
66, 3432–3443.
539
Verboven, N., Monaghan, P., Evans, D. M., Schwabl, H., Evans, N., Whitelaw, C. & Nager, R. G.
540
(2003) Maternal Condition, Yolk Androgens and Offspring Performance: A Supplemental Feeding
541
Experiment in the Lesser Black-Backed Gull (Larus Fuscus). Proceedings. Biological sciences 270,
542
2223–2232.
543
Supporting information
544
Supplementary figures and tables are compiled in a separate document for online publication alongside
545
the main article.
546
Acknowledgements
547
The authors wish to acknowledge Lynsey Gregory, Allison Blake, and Phil Young for support with
548
animal husbandry. The laboratory work received funding from the University of Sheffield. J.O. was
549
supported by a University of Sheffield scholarship and by the European Union’s Horizon 2020 research
550
and innovation programme under grant agreement No. 668989.
551
Author contributions
552
P.J.W., A.F., and W.V.H. developed the initial concept; J.O. designed the experiments in consultation
553
with P.J.W., A.F., and W.V.H.; J.O. and K.E.H. performed experiments; J.O. analysed the data; J.O.
554
and P.J.W. wrote the manuscript and W.V.H. provided additional feedback on the written manuscript;
555
all authors approved the submitted manuscript.
556

27
Competing interests
557
The authors declare no competing interests
558

1
Supplementary material: Premature birth stunts early
1
growth and is a possible driver of stress-induced maternal
2
effects in the guppy, Poecilia reticulata
3
James Ord
1
,3, Kelle E. Holmes1, William V. Holt2, Alireza Fazeli
2
,
3
, and Penelope J. Watt1
4
* Corresponding author: jms.ord18@gmail.com
5
1
Department of Animal and Plant Sciences, University of Sheffield, Alfred Denny Building, Western
Bank, Sheffield S10 2TN, United Kingdom
2
Academic Unit of Reproductive and Developmental Medicine, Department of Oncology and
Metabolism, University of Sheffield, Level 4, Jessop Wing, Tree Root Walk, Sheffield S10 2SF,
United Kingdom
3
Institute of Biomedicine and Translational Medicine, Department of Pathophysiology, University of
Tartu, Tartu, Estonia

2
6
Figure S1. (a) Schematic of experimental treatment sequence. Individually-housed females were removed from
7
their home tank daily and placed in a treatment tank of 600mL water to which 200µL alarm substance was added
8
(or water as a control). Following 20 minutes of exposure, animals were transferred to a washing tank for 5 mins
9
to remove traces of alarm substance before returning to their home tank. (b) Total swim distance (cm) of
10
experimentally naïve female and male guppies during 10 mins exposure to alarm substance (p-values derived
11
from nonparametric permutation tests, N = 3 individuals per sex per group). (c) Timelines of experimental
12
manipulation in relation to mating and gestation, and phenotypic evaluation of the offspring during their early
13
growth and at maturity. After a three-day mating period, females were exposed to alarm substance (or control)
14
daily as described in (a) for the first 23 days of the gestation period.
15
16
17

3
Table S1. Weight, gestation length, and brood sizes of females which produced broods in the predation stress
18
experiment.
19
Female
Weight (g)
Treatment
Gest. length (days)
Brood size
C1
0.51
Control
56
4
C2
0.41
Control
64
22
C3
0.34
Control
51
15
C4
0.56
Control
49
20
P1
0.53
Predation
53
25
P2
0.59
Predation
49
23
P3
0.52
Predation
40
30
P4
0.46
Predation
25
18
P5
0.55
Predation
49
22
P6
0.49
Predation
29
6
20
Table S2. Weight, gestation length, and brood sizes of females which produced broods in the diet experiment.
21
Female
Weight (g)
Treatment
Gest. length (days)
Brood size
H1
0.54
High Diet
50
17
H2
0.59
High Diet
31
16
H3
0.53
High Diet
45
14
H4
0.54
High Diet
53
2
H5
0.53
High Diet
37
13
H6
0.73
High Diet
54
18
H7
0.73
High Diet
44
18
L1
0.71
Low Diet
64
3
L2
0.46
Low Diet
51
10
L3
0.58
Low Diet
44
5
L4
0.55
Low Diet
53
2
L5
0.64
Low Diet
31
8
22
23

4
Table S3. Numbers of offspring derived from each mother in the predation stress experiment, the numbers reared,
24
and the numbers of females (F) and males (M) used for each type of measurement. OF refers to the open field
25
test, while SR refers to stress response test, where the two numbers within the brackets denote the number used
26
as controls and the number exposed to alarm substance, respectively.
27
Treatment /
mother
N
offspring
N
reared
N
mortality
N weighed
N tested*
Females
Males
OF
SR-
control
SR-
alarm
Control 1
4
4
0
2
2
4(0/1/3)
2(1/1)
2(1/1)
Control 2
22
12
0
3
3
6(3/3)
4(2/2)
3(2/1)
Control 3
15
12
0
6
2
5(3/2)
3(2/1)
3(2/1)
Control 4
20
12
1
6
4
7(3/4)
4(2/2)
4(2/2)
Predation 1
25
12
0
9
2
5(3/2)
4(2/2)
4(2/2)
Predation 2
23
12
2
7
2
5(3/2)
3(2/1)
3(2/1)
Predation 3
30
12
0
8
4
6(3/3)
5(3/2)
5(3/2)
Predation 4
18
12
0
10
2
5(3/2)
3(2/1)
3(2/1)
Predation 5
22
12
4
3
2
5(3/2)
2(1/1)
2(1/1)
Predation 6
6
6
1
3
2
4(3/1)
0(0/0)
1(0/1)
* Numbers outside the brackets are the total number tested in each test category and numbers in brackets are the
28
numbers of each sex (females/males/unassigned). In one case, sex information was not used due to
29
misidentification of a male as a female at the time of testing (Control 1, OF); as we did not have a system for
30
tracking the identity of individual offspring, all three of the animals from this brood which were initially labelled
31
as females were re-classified as ‘unassigned’ for that particular test.
32

5
Table S4. Sample sizes, response variables, fixed and random effect terms for each mixed effects model.
33
Interactions are denoted by ‘x’. For analyses in which multiple individual offspring per parent were tested, the
34
number outside the brackets is the number of broods and the number inside the brackets the total number of
35
offspring in all broods.
36
Response variable
Model / fixed effect terms
Random
Group
N(n)
Mean individual
body area (log scale)
1) Maternal treatment x Age2
Mother ID
Control
4
Predation
6
Mean individual
body area (log scale)
2) Maternal diet x Age2
Mother ID
High diet
7
Low diet
5
F1 wet weight (mg)
3) Maternal treatment + Sex
Mother ID
Control
4(28)
Predation
6(54)
F1 wet weight (mg)
4) Maternal diet + Sex
Mother ID
High diet
7(30)
Low diet
5(19)
F1 thigmotaxis (%)
5) Maternal treatment + Sex
Mother ID
Control
4(21)
Predation
6(30)
F1 swim distance
(cm)
6) Maternal treatment x
Offspring treatment + Sex
Mother ID
Control-control (water)
4(13)
Control-alarm
4(12)
Predation-control
(water)
5(17)
Predation-alarm
6(18)
37
38

6
39
Figure S2. Maternal reproductive traits and offspring growth in response to gestational diet manipulation. (a) and
40
(b) gestation length in days, and fecundity (no. offspring per gram body weight), respectively for high diet and
41
low diet mothers; p-value derived from a nonparametric permutation test. (c) Body size from days 0-15 postpartum
42
for high diet and low diet broods, plotted as mean individual body area (mean of individual body areas in each
43
brood on each day) on log-scale. Curve fits with 95% CIs were derived from linear mixed effects models with
44
second-order polynomial regression. (d) Wet weight (mg) of adult female (top) and male offspring (bottom) from
45
high diet and low diet offspring, measured at approx. two months postpartum.
46
47

7
48
Fig. S3. Linear regression slopes of mean individual body area against age fitted to data from the first six days
49
postpartum for each brood in the predation stress (left) and diet (right) experiments. Different maternal treatment
50
or diet groups are discerned by point shape and line type (control or high diet = circles and solid lines, predation
51
stress or low diet = triangles and dashed lines) while different colours represent broods from different mothers
52
within those groups (thus one colour can denote up to two different mothers within each plot).
53

8
Table S5. Regression coefficients (brood mean body area ~ postpartum age) for each brood in each of the
54
predation stress and diet experiment. Gestation lengths are also shown as the regression coefficients were plotted
55
against gestation length for Fig. 3.
56
Predation stress experiment
Mother
Treatment
Regression coefficient
Gestation length
C1
Control
0.547
56
C2
Control
0.196
64
C3
Control
0.106
51
C4
Control
0.419
49
P1
Predation
0.069
53
P2
Predation
0.171
49
P3
Predation
0.121
40
P4
Predation
-0.118
25
P5
Predation
0.084
49
P6
Predation
0.016
29
Diet experiment
Mother
Treatment
Regression coefficient
Gestation length
H1
High diet
0.146
50
H2
High diet
0.262
31
H3
High diet
0.409
45
H4
High diet
0.302
53
H5
High diet
-0.032
37
H6
High diet
0.251
54
H7
High diet
0.076
44
L1
Low diet
0.386
64
L2
Low diet
0.301
51
L3
Low diet
0.147
44
L4
Low diet
0.206
53
L5
Low diet
0.189
31
57

9
Table S6. Results of a linear mixed effects model describing the effect of maternal predation stress and offspring
58
alarm treatment on swim distance in 10 mins (cm, square root-transformed). Both F-test and T-test results of fixed
59
effect terms are shown, with F-values derived from the effect of adding a term last to the model, while T-values
60
derive from comparing the parameter estimate to the intercept in the absence of other factors. Variance and
61
standard deviations of random effect terms are shown. n = 60 offspring from 10 parents (4 control and 6 predation
62
stress).
63
Fixed effects
ndf
ddf
F
p
Est
SE
T
p
Intercept
24.765
33.829
3.085
10.964
< 0.001
Offspring treatment (Alarm)
1
48.33
7.62
0.008
Offspring treatment (Alarm)
48.232
-8.049
3.862
2.084
0.04
Maternal treatment (Predation)
1
6.77
0.11
0.75
Maternal treatment (Predation)
19.141
-2.03
3.82
-0.531
0.6
Sex (Male)
1
49.701
5.32
0.03
5.832
2.528
2.306
0.03
Offspring treatment x Maternal
treatment
1
48.303
0.18
0.67
2.149
5.055
0.425
0.67
Random effects
Variance
SD
Mother
3.53
1.88
Residual
92.67
9.63
64