Biol. Lett. (2006) 2, 39–42
Published online 6 December 2005
Effect of growth
fitness in green swordtails
Nick J. Royle*, Jan Lindstro ¨m
and Neil B. Metcalfe
Division of Environmental and Evolutionary Biology,
Institute of Biomedical and Life Sciences, University of Glasgow,
Graham Kerr Building, Glasgow G12 8QQ, UK
*Author for correspondence (firstname.lastname@example.org).
Early environmental conditions have been
suggested to influence subsequent locomotor
performance in a range of species, but most
measurements have been of initial (baseline)
performance. By manipulating early growth
trajectories in green swordtail fish, we show that
males that underwent compensatory growth as
juveniles had a similar baseline swimming
endurance when mature adults to ad libitum fed
controls. However, they had a reduced capacity
to increase endurance with training, which is
more likely to relate to Darwinian fitness. Com-
pensatory growth may thus result in important
locomotor costs later in life.
Keywords: growth rate; exercise; endurance;
resource allocation; swimming performance
Changes in food availability during growth and
development are likely to be common for many
animal species. If food availability is initially low but
then increases, individuals have the chance to com-
pensate for their poor start in life, so that they catch-
up and rejoin, or even exceed, their original growth
trajectory. Such an increase in size may be beneficial
for survival and reproduction, but growth compen-
sation is also associated with numerous and varied
long-term costs (Metcalfe & Monaghan 2001). One
of the potential costs of rapid growth is a reduction in
locomotor performance (Munch & Conover 2004),
possibly mediated through altered muscle develop-
ment (Galloway et al. 1999).
Most studies of locomotor performance have
simply taken initial baseline measurements, often
using laboratory-reared animals (reviewed by Kolok
1999; Irschick & Garland 2001). However, perform-
ance changes with nutritional and energetic status
(Guderley 2004) and vertebrates usually improve
their performance with repeated exposure (‘exercise
training’, Davison 1997). Despite this, there has been
little investigation of the factors that might influence
this rate of improvement, especially as the response to
exercise is arguably more important to survival than
baseline performance, since it is a measure of how
well the individual can match performance to current
Here, we test the hypothesis that early growth
trajectory affects the subsequent ability to improve
locomotor performance, using green swordtail fish
virtually cease body growth at the onset of sexual
maturation, but their adult body size is a strong
predictor of dominance in interactions with other
males (Beaugrand & Cotnoir 1996). Consequently,
any reduction in early growth rate might be expected
to induce a compensatory response prior to sexual
maturation, which is predicted to have negative long-
term effects on the ability to increase physical fitness.
2. MATERIAL AND METHODS
(a) Rearing regimes
Fry of wild parentage were reared in the laboratory under
controlled conditions (stable temperature of 23G1 8C, 16L : 8D
light regime, ad libitum food). Individual fry were placed in one-half
of an acrylic plastic rearing tank (320!170!180 mm; 10 l), which
was divided longitudinally by a transparent Perspex partition. This
ensured that fishes were physically, but not visually isolated from
their neighbours. Each tank had a gravel substrate, a plastic plant
for cover and was aerated and filtered. At two months of age, equal
numbers of fishes from each brood were allocated to one of two
(i) Good/Good (GG)—ad libitum food daily for the duration of
the experimental period (two months of age onwards).
(ii) Poor/Good (PG)—fed three times a week from two to six
months of age, then subsequently put onto the same daily ad
libitum diet as GG fishes from six months onwards.
Details of the feeding protocol and basic husbandry are given in
the electronic supplementary material. Measurements of baseline
endurance were taken on 71 males from 19 different families at 12
months of age from both treatment groups. The response to
exercise was assessed in a subsample of 13 size-matched pairs of
sibling males to control for genetic and maternal effects
on endurance (nZ13 dams), one from each of treatment groups
1 and 2 (table 1). Thirteen size-matched pairs of non-sibling males
(from different families to each other and to the sib pairs), one
from each of treatment groups 1 and 2 (nZ13), acted as untrained
controls to assess change in endurance without exercise (see
Fishes were weighed (G0.01 g) and measured initially at two
months, and then subsequently every two weeks until 10 months of
age. On each occasion, standard length (i.e. from tip of nose to tip
of caudal peduncle), total length (standard length plus length of the
caudal fin, including the ‘sword’ extension of the fin if present) and
maximum body depth were recorded (G0.1 mm). Growth rate was
quantified as the growth increment of standard length (SL, in mm)
over successive growth periods (two to six months and six to ten
(b) Measurement of endurance
Swimming endurance was measured in a modified Bla ¨zka respirom-
eter (see electronic supplementary material), which pumped a
steady flow of water through a central chamber containing a single
fish. The fish was placed into the chamber for 5 min to settle at an
initial flow rate of 21 cm sK1, then allowed to acclimatize for 1 min
at a flow rate of 27 cm sK1, before the pump was turned up to the
experimental running speed (50 cm sK1) and maintained until
fatigue. Fishes were deemed to be exhausted when they were forced
back against the back fine mesh grid for more than 5 s (Ryan 1988)
and were no longer able to continue swimming, despite tapping of
the side of the chamber (Ojanguren & Bran ˜a 2000). Once
exhausted, the pump was turned off and the fish allowed 5 min
recuperation time before being placed back in its rearing tank.
Endurance is defined as the amount of time that a fish swam at the
highest flow rate (50 cm sK1).
(c) Exercise protocol
After initial screening of endurance, sib pairs were put through the
following training regime over a three-week period. Each week
involved a single 25 min session of exercise as follows: Week
1—15 min at a low flow rate (21 cm sK1) then 5 min at a medium
The electronic supplementary material is available at http://dx.doi.
org/10.1098/rsbl.2005.0414 or via http://www.journals.royalsoc.ac.
Received 27 September 2005
Accepted 31 October 2005
q 2005 The Royal Society
flow rate (27 cm sK1), finishing with a further 5 min at the low rate.
Week 2—10 min low/10 min medium/5 min low. Week 3—5 min
low/15 min medium/5 min low. Control pairs of fishes did not go
through the exercise regime. Both sib pairs and (untrained) controls
were again tested for endurance at the end of the three-week
(d) Statistical analysis
Endurance was log transformed to normalize the data before
analysis. We used the penalized log likelihood (Akaike Information
Criteria, AIC) in mixed-effects general linear models (GLMs) to
compare the fit of different models of endurance, following
sequential dropping of non-significant terms from a full model
(Crawley 2002). The smaller the AIC, the better the model fit. The
log-likelihood ratio test was used to compare the fit of successive
models (Crawley 2002). A two-factor repeated measures ANOVA
with standard length as the response variable and age (2, 6 and 10
months) and treatment (GG versus PG) as the factors was used to
analyse growth. All statistical tests were conducted using S-PLUS 6
for Windows or SPSS 10 for Windows.
(a) Growth rate: sib males
There was, unsurprisingly, a significant effect of
age on body size for sib pairs (F2,10Z210.17,
p!0.0005; figure 1). There was no overall effect of
treatment on growth (F1,11Z0.84, pZ0.38), but
there was a significant age!treatment interaction
(F2,10Z14.27, pZ0.001; figure 1). This was because
PG males, although slightly larger at two months,
were significantly smaller than their GG brothers at
six months of age, but then (as a result of
compensatory growth) were no different in size by
10 months of age (quadratic term: F1,11Z25.81,
p!0.0005; figure 1).
Baseline endurance levels were not related to age at
maturation (t44Z0.31, pZ0.76), treatment (t50Z1.03,
pZ0.31), standard length at two months of age (t49Z
0.53, pZ0.60) or sword length (t45Z0.48, pZ0.63).
Body size (SL) was positively, but non-significantly
related to baseline endurance (t51Z1.37, pZ0.18).
After training, GG reared males had significantly
greater endurance capacity than their PG sibs
(figure 2a). A mixed-effect GLM was fitted to these
data, with the difference between final and initial
endurance capacity (log) as the response variable,
family as a random effect and initial endurance (log),
standard length, standard length at two months of
age and treatment as fixed effects. Standard length
(t8Z0.76, pZ0.47) and standard length at two
dropped from successive models without significantly
increasing the AIC. The minimum adequate model
included initial endurance (t10Z5.79, pZ0.0002),
treatment (t10Z2.33, pZ0.042) and their interaction
(t10Z2.03, pZ0.070). The interaction was retained
in the model, as its removal resulted in a significant
increase in the AIC (likelihood ratioZ4.44, pZ
0.035). GG males with the lowest initial endurance
showed the greatest change in endurance following
training. PG males showed very little response to
training (figure 2a).
In contrast, there was no effect of treatment on the
change in endurance capacity between control pairs
of males over the same period (GLM; F1,22Z0.18,
pZ0.68). There was also no significant effect of initial
endurance on endurance capacity (F1,22Z3.84,
pZ0.063), although endurance decreased overall
(figure 2b). However, there was a significant inter-
action between initial endurance and treatment
(F1,22Z12.12, pZ0.002): GG males showing the
lowest initial endurance had the biggest decrease in
endurance over time, whereas the greatest decline in
PG males was in those with the highest initial
Male swordtails experiencing an improvement in
food availability during development (PG males)
were subsequently able to compensate fully in body
Table 1. Biometrics, when tested for swimming, of males used in the experimental response to exercise and controls.
sib pairs (nZ13)non-sib controls (nZ13)
GG PGpaired t-test GG PGpaired t-test
Figure 1. Standard length (SL; mm) in relation to age
(months) for male sib pairs (nZ13). MeanG1 s.e. Circular
symbols represent GG males and triangles PG males.
40N. J. Royle and others
Growth and physical fitness in swordtails
Biol. Lett. (2006)
and tail (sword) size compared to GG males, and
had a similar baseline level of endurance. However,
their capacity to improve their performance was
significantly reduced. Since the fishes were not
exposed to unidirectional currents in their rearing
tanks, it is perhaps not surprising that differences in
swimming capacity were only revealed once they
were repeatedly forced to swim against a strong
current. Nonetheless, baseline aerobic swimming
performance is routinely used in laboratory studies
as a measure of locomotor capacity (Kolok 1999),
despite the problems of translating laboratory per-
formance to ecological function (Irschick & Garland
2001). Exercise training more readily replicates the
conditions found in the wild, where swordtails often
court females feeding on algal-covered rocks in
strong currents (Ryan 1988).
In many cases, natural selection appears to increase
physical performance. Despite this, physical perform-
ance in humans and other animals is often highly
variable. Le Galliard et al. (2004) showed that initial
endurance (at birth) in lizards was heritable, but
selection in favour of increased endurance was weak.
This was because environmental conditions following
birth determined the expression of a genetic predis-
position for high initial endurance (Le Galliard et al.
2004). Early developmental conditions may thus
greatly influence the effect of locomotor performance
on Darwinian fitness. The importance of endurance
capacity for courtship (Ryan 1988), combined with
the greater dominance of GG males in paired
encounters with size-matched PG males (Royle et al.
2005), suggests that compensation following poor
initial growth may reduce the mating success of male
swordtails much later in life, and illustrates the
importance of incorporating behaviour into a ‘per-
formance paradigm’ of locomotor function (Irschick
& Garland 2001). The challenge now is to identify
the structural (e.g. Galloway et al. 1999) or metabolic
(Davison 1997; Guderley 2004) causes for the
growth-induced effects on muscular performance.
This work was supported by BBSRC grant 17/S15807.
Thanks to John Laurie, June Freel, Graham Adam and
Helicia Lepatik for help with fish husbandry, and two
anonymous referees for comments on an earlier version of
Beaugrand, J. P. & Cotnoir, P.-A. 1996 The role of
individual differences in the formation of triadic dom-
inance orders of male green swordtail fish (Xiphophorus
helleri). Behav. Process. 38, 287–296. (doi:10.1016/
Crawley, M. J. 2002 Statistical computing: an introduction to
data analysis using S-Plus. Chichester, UK: Wiley.
Davison, W. 1997 The effects of exercise training on teleost
fish: a review of recent literature. Comp. Biochem. Physiol.
A 117, 67–75. (doi:10.1016/S0300-9629(96)00284-8)
Galloway, T. F., Kjorsvik, E. & Kryvi, H. 1999 Muscle
growth and development in Atlantic cod (Gadus morhua
L.) related to different somatic growth rates. J. Exp. Biol.
Guderley, H. 2004 Locomotor performance and muscle
metabolic capacities: impact of temperature and ener-
getic status. Comp. Biochem. Physiol. B 139, 371–382.
Irschick, D. J. & Garland, T. 2001 Integrating function and
ecology in studies of adaptation: investigations of loco-
motor capacity as a model system. Annu. Rev. Ecol. Syst.
32, 367–396. (doi:10.1146/annurev.ecolsys.32.081501.
Kolok, A. S. 1999 Interindividual variation in the prolonged
locomotor performance of ectothermic vertebrates: a
comparison of fish and herpatofaunal methodologies and
a brief review of the fish literature. Can. J. Fish. Aquat.
Sci. 56, 700–710. (doi:10.1139/cjfas-56-4-700)
Le Galliard, J.-F., Clobert, J. & Ferrie `re, R. 2004 Physical
performance and Darwinian fitness in lizards. Nature
432, 502–505. (doi:10.1038/nature03057)
Metcalfe, N. B. & Monaghan, P. 2001 Compensation for a
bad start: grow now, pay later? Trends Ecol. Evol. 16,
final–initial endurance (s)
final–initial endurance (s)
Figure 2. (a) Change in endurance in response to exercise
for male sib pairs and (b) change in endurance for non-sib
male controls (that were not exercised) over the same time
period. MeanG1 s.e. Data presented using untransformed
values for clarity. Statistics were performed on log values for
endurance (see text).
Growth and physical fitness in swordtails
N. J. Royle and others 41
Biol. Lett. (2006)
Munch, S. B. & Conover, D. O. 2004 Nonlinear growth
cost in Menidia menidia: theory and empirical evidence.
Evolution 58, 661–664.
Ojanguren, A. F. & Bran ˜a, F. 2000 Thermal dependence of
swimming endurance in juvenile brown trout. J. Fish
Biol. 56, 1342–1347. (doi:10.1111/j.1095-8649.2000.
Royle, N. J., Lindstro ¨m, J. E. & Metcalfe, N. B. 2005 A poor
start in life affects dominance status in adulthood indepen-
dent of body size in green swordtails Xiphophorus helleri. Proc.
R. Soc. B 272, 1917–1922. (doi:10.1098/rspb.2005.3190)
Ryan, M. J. 1988 Phenotype, genotype, swimming endur-
ance and sexual selection in a swordtail (Xiphophorus
nigrensis). Copeia 2, 484–487.
42N. J. Royle and others
Growth and physical fitness in swordtails
Biol. Lett. (2006)