Author's personal copy
Ecotoxicology (2011) 20:535–542
DOI 10.1007/s10646-011-0624-2
Heritability of heat tolerance in a small livebearing fish,
Heterandria formosa
Cathleen M. Doyle • Paul L. Leberg
Paul L. Klerks
•
Accepted: 21 February 2011 / Published online: 4 March 2011
Ó Springer Science+Business Media, LLC 2011
Abstract Climate change is expected to result in an
increased occurrence of heat stress. The long-term population-level impact of this stress would be lessened in
populations able to genetically adapt to higher temperatures. Adaptation requires the presence of geneticallybased variation. At-risk populations may undergo strong
declines in population size that lower the amount of genetic
variation. The objectives of this study were to quantify the
heritability of heat tolerance in populations of the least
killifish, Heterandria formosa, and to determine if heritabilities were reduced following a population bottleneck.
Heritabilities of heat tolerance were determined for two
lines of each of two source populations; two bottlenecked
lines (established with one pair of fish) and two regular
lines. Heat tolerance was quantified as temperature-atdeath (TAD), when fish acclimated at 28°C were subjected
to an increase in water temperature of 2°C/day. Mid-parent/mean offspring regressions and full-sib analyses were
used to estimate the heritability of TAD. Heritability estimates from parent/offspring regressions ranged from 0.185
to 0.462, while those from sib analyses ranged from 0 to
0.324, with an overall estimate of 0.203 (0.230 for the
regular lines, 0.168 for bottlenecked ones). Fish from the
bottlenecked line from one source population (but not
C. M. Doyle P. L. Leberg P. L. Klerks (&)
Department of Biology, University of Louisiana at Lafayette,
Box 42451, Lafayette, LA 70504-2451, USA
e-mail: klerks@louisiana.edu
Present Address:
C. M. Doyle
Limnology Laboratory, The Ohio State University,
1522 Museum of Biological Diversity,
1315 Kinnear Rd., Columbus, OH 43212, USA
the other) had a lower heritability than did those from the
regular line. These results show that the populations tested
had some potential for adaptation to elevated water temperatures, and that this potential may be reduced following
a population bottleneck. This should not be construed as
evidence that natural populations will not suffer negative
consequences from global warming; this study only
showed that these specific populations have some potential
to adapt under a very specific set of conditions.
Keywords Heat tolerance Heritability Genetic
adaptation Population bottleneck Climate change
Evolutionary processes
Introduction
Many species find themselves present in an environment
that is changing as a consequence of human activities. One
major force behind current and especially future changes in
species’ environments is climate change. Anthropogenic
climate change is clearly occurring and global temperatures
have already increased (IPCC 2007a). There is also evidence that these climate changes are having ecological
consequences (IPCC 2007b). As specific examples, Grebmeier et al. (2006) found changes in benthic–pelagic
coupling related to the timing of seasonal ice cover in the
North Bering Sea, and Hari et al. (2006) found higher
mortality rates in brown trout due to a disease mediated by
increased water temperatures in rivers and streams in parts
of the Swiss Alps. Reviews by Walther and co-workers
(Walther 2010; Walther et al. 2002) also found ample
evidence of climate change impacts in various ecosystems
and at various levels of biological organization. Climate
change is thus likely to result in heat stress for many
123
Author's personal copy
536
populations. While these impacts may bring about local
extinctions and changes in geographic distributions, negative impacts also exert selective pressures that may ultimately result in adaptation if conditions are favorable and
if heritable variation is present for the affected trait.
This paper is focusing on fishes. Because most fishes are
ectotherms, their body temperature conforms to the surrounding water temperature (Beitinger et al. 2000; Goldspink 1995). Maintaining thermal equilibrium with the
surrounding water directly affects their internal metabolism
and physiological functions (Beitinger et al. 2000; Elliott
1981; Goldspink 1995). While many studies have addressed heat tolerance, very few have sought to determine if
heat tolerance has a genetic basis. The presence of genetic
variation in heat tolerance may allow a population to survive climate change despite the detrimental effects of high
water temperature. Population traits such as heat tolerance
are often complex traits that exhibit continuous variation
and have inheritance patterns that involve many genes (so
called ‘‘quantitative traits’’) and may also involve genotype
by environment interactions (Falconer and Mackay 1996;
Hartl 2000). A good measure of the extent to which such a
trait can respond to natural selection is its ‘‘narrow sense
heritability’’ (designated as h2 and referred to as ‘‘heritability’’ in the remainder of this paper). This heritability is
the ratio of additive genetic variance (VA) to the total
phenotypic variance (VP) in the population (Falconer and
Mackay 1996). Another heritability term in use is the
‘‘broad sense heritability’’ or ‘‘degree of genetic determination’’, designated as H2, defined as the ratio of genetic
variance to the total phenotypic variance, but not as useful
for predicting a population’s response to selection (Falconer and Mackay 1996). Heritabilities can range from 0 to
1. More in-depth introductions to quantitative genetic
variation and heritabilities are provided by Falconer and
Mackay (1996) and Roff (1997).
Few studies have estimated the heritability of thermal
tolerance in fish species. Meffe et al. (1995) studied a
population of eastern mosquitofish (Gambusia holbrooki)
and found a heritability of 0.32 for heat tolerance measured
as temperature at death. In contrast, Baer and Travis (2000)
selected the least killifish (Heterandria formosa) for an
increased resistance to heat or cold, but did not detect a
significant response to selection. They concluded that the
heritability was not likely to be more than 0.15 in their
laboratory population. Finally, Charo-Karisa et al. (2005)
estimated the heritability of cold tolerance in juvenile Nile
tilapia (Oreochromis niloticus) using a full-sib/half-sib
design, and reported a low heritability (estimated to be
0.09).
The amount of genetic variation may be reduced in a
population that has undergone a drastic reduction in population size. Such a population bottleneck may be an
123
C. M. Doyle et al.
indirect result from climate change, e.g., due to loss of
suitable habitat or due to a sudden fish kill caused by
lowered levels of dissolved oxygen. While a population
bottleneck will generally result in a loss of genetic variation (Falconer and Mackay 1996), heritabilities may actually increase. The latter was observed in populations of the
housefly (Musca domestica), where the bottleneck ‘‘converted’’ nonadditive genetic variance into additive genetic
variance (Bryant et al. 1986). Increases in additive genetic
variance following a population bottleneck appear more
likely in populations that already have a low level of
additive genetic variance (López-Fanjul et al. 1999) or a
high level of dominance variance (Saccheri et al. 2001).
Also, conversion of nonadditive into additive genetic variation is more likely for characters affected by many loci
than for those affected by few loci (Naciri-Graven and
Goudet 2003).
This study’s objectives were to quantify the heritability
of heat tolerance in populations of the least killifish (H.
formosa), and to determine whether this heritability was
affected by a population bottleneck. Heat tolerance was
determined using the chronic lethal method (CLM); a
method commonly used when determining thermal tolerance (Beitinger et al. 2000; Elliott 1981). Heritabilities of
heat tolerance were quantified on the basis of resemblance
between relatives, using the parent–offspring regressions
and full-sib analyses that are used routinely in quantitative
genetics (Falconer and Mackay 1996). The least killifish is
eurythermal, and is found in waters with temperatures
ranging from 7 to 39°C (Forster-Blouin 1989; Leips and
Travis 1999). In the context of climate change, adaptation
to elevated temperatures may be more relevant for
stenothermal species. However, species that occur over a
wide range of temperatures may be encountering stressful
temperatures at the edge of their distribution and may
have insufficient genetic variation to respond to selection
for further increases in their heat tolerance. While genetic
variation for heat tolerance had previously been studied in
the least killifish (Baer and Travis 2000), the present
study nevertheless used this same species. Heritability
estimates are population and method specific (Roff 1997),
thus the outcome of the present study would provide an
additional and partially independent estimate. Moreover,
the approach used by Baer and Travis (2000) may not
have been optimal for eliciting a response to selection.
That study used a relatively-low selection intensity (low,
at least, for observing a response in a few generations)
and the actual selection intensity was further reduced by
the inclusion of adult females that likely had mated prior
to the selection step. Also, the loss of several of that
study’s laboratory lines and the inadvertent occurrence of
population bottlenecks complicated the interpretation of
the results.
Author's personal copy
Heritability of thermal tolerance
Materials and methods
Least killifish natural history
The least killifish, H. formosa, Family Poeciliidae, is a
small livebearing fish found in the coastal southeastern
United States, ranging from the Cape Fear River drainage,
North Carolina, throughout Florida and into southern
Louisiana (Page and Burr 1991). It prefers slow-moving,
heavily vegetated, shallow, and fresh to brackish water
(Baer and Travis 2000; Page and Burr 1991). Its short
generation time (2–3 months) makes it particularly suitable
for determining heritabilities.
Experimental design and methodology
Fish were taken in 2002 from two locations situated
approximately 105 km apart; one just outside the Lacassine
Refuge (Cameron Parish, Louisiana) and the other in the
Atchafalaya Basin (St. Martin Parish, Louisiana). There
was no specific reason for collecting fish from these
localities, other than the desire to determine heritabilities
for more than a single population—as heritability estimates
may be population-specific (Falconer and Mackay 1996).
Approximately 120 fish from each location were set up in
tanks in a greenhouse at the University of Louisiana at
Lafayette’s Center for Ecology and Environmental Technology (CEET). Bottlenecked and non-bottlenecked (regular) lines were established for each source population,
approximately one generation after field-collection. Each
regular line was started with 30 males and 30 females,
while each bottlenecked line was started with one pair
consisting of a male and virgin female. The populations for
the bottlenecked lines were allowed to build up, without
further bottlenecking, to a population size of at least 120
adults. At this point, the fish populations had been at CEET
for approximately 3–4 generations since collection from
the wild. The actual breeding experiments for heritability
analyses were then started using newborn offspring for
each of the four lines.
Newborn fish (\10 mm) were set up in pairs. The fish
were sexed after a few weeks; male–female pairs were kept
as such, while fish of same-sex pairs were used to establish
additional male–female pairs (feasible since fish in samesex pairs were still virgin). These pairs formed the parental
generation. Each pair of fish was housed in a 4.73-l polyethylene bucket, and buckets were randomly placed in
water-filled tanks (213 cm L 9 56 cm W 9 30 cm D)—
approximately 40 buckets per tank. Each bucket had two or
three mesh-covered holes to allow water exchange. Each
pair of tanks had their own water system with the water
recirculating through the two tanks and a filter tank
(244 cm L 9 61 cm W 9 61 cm D). Fish were fed daily
537
with a combination of TetraminÒ flake food and newly
hatched brine shrimp. Tanks were set up in the CEET
greenhouse (Lacassine population) or in an air-conditioned
laboratory building at CEET (Atchafalaya population); this
design enhanced chances of obtaining enough offspring for
experiments from at least one of the source populations.
This meant that comparisons between the two source
populations would not be valid—but such a comparison
was not an objective of this study. Offspring were regularly
separated from the parents and raised to sexual maturity in
separate buckets (one bucket per group of offspring from a
pair of adults). This was continued till sufficient offspring
were obtained in a family. The goal was to obtain at least
ten offspring (five adult males and five adult females) from
each of 40 parental pairs for each of the four lines, but
these targets were not always reached (due to some adults
dying before sufficient offspring were produced). The following numbers of families were used for this study: 43
from the Lacassine non-bottlenecked line, 40 from the
Lacassine bottlenecked line, 19 from the Atchafalaya nonbottlenecked line, and 14 from the Atchafalaya bottlenecked line. These were the samples sizes available in the full
sib analyses. Sample sizes for the parent–offspring
regressions were somewhat lower because heat exposures
of parents were postponed till their offspring had also
reached the adult stage (so that all family-members could
be exposed at the same time). This resulted in an average of
16% of the parental fish having died before their thermal
tolerance could be quantified.
To quantify heat resistance, fish were individually
exposed to 750 ml heated water in 1 l tri-pour beakers with
screen-covered holes inside a 26.6 l container placed in a
Percival Scientific I-36LL environmental chamber (Percival Scientific, Boone IA, USA). The water in the 26.6 l
container was aerated vigorously to establish uniform
temperature conditions among the 12 beakers. A 12 h
light:12 h dark day cycle was maintained in the chamber.
Three different incubators were used, with generally three
shelves (each containing one 26.6 l container) per incubator. Fish were randomly assigned to a specific container
and incubator. Fish were fed newly hatched brine shrimp
daily throughout the exposure. The fish were acclimated at
an incubator air temperature of approximately 28°C for
7 days before the thermal gradient was started. The environmental chamber was programmed to increase the air
temperature by 2°C/day beginning at 28°C, until total
mortality was reached among the exposed fish. Criterion
for death was complete cessation of external movement
and gulping motion (cessation of opercular motion).
Because temperature-at-death (TAD) was the main
response variable, fish survival was checked every 2–3 h
once the lethal temperature was approached. More frequent
checks were not possible since heat loss from opening the
123
Author's personal copy
538
chamber door caused a temporary set-back to the desired
gradual increase in water temperature. The water temperature of each 26.6 l container was determined using a 4400
Series NIST Traceable Digital Thermometer with 0.01°C
precision and 0.02°C accuracy (Alpha Technics, Anaheim
CA, USA) each time that fish survival was checked. The
TAD of a fish was calculated as the average of the water
temperatures at the time that the fish was classified as dead
and that recorded at the previous time period. Dead fish
were removed and their standard length was determined.
To verify that the cause of mortality was temperature
and not elevated ammonia levels, ammonia concentrations
were monitored during some of the exposures. Water
samples were taken at the beginning and end of the acclimation period, at the time point when the first mortality
occurred, and at the time point where total mortality was
reached. Samples were analyzed for total ammonia using a
Lachat Instruments QuickChem FIA ? 8000 series autoanalyzer (Lachat Instruments, Milwaukee WI, USA).
Concentrations of un-ionized ammonia (NH3) were determined using the aqueous ammonia equilibrium equations
based on water pH and temperature (Emerson et al.
1975). Both total ammonia (NH4? ? NH3) and unionized
ammonia levels that the fish encountered were always well
below the U.S. Environmental Protection Agency recommended maximum levels (USEPA 1998).
Statistical analysis
Data from fish that died before the water temperature
reached 37°C (approximately 1% of all deaths) were
excluded from the analysis. It was assumed that these
deaths were not heat related.
A survival analysis was performed on the full dataset
(combining data for all four treatment types—bottlenecked
and non-bottlenecked lines for each of the two source
populations) to determine the overall effects of sex, age,
source population, and bottleneck history on TAD, using a
lognormal regression model in StatViewÒ version 4.0
(Abacus Concepts, Berkeley CA, USA).
In order to standardize results from the thermal tolerance
quantifications and to be able to get a more precise estimate
of each fish’s innate thermal tolerance, thermal tolerances
were expressed as deviations from expected temperaturesat-death. A survival analysis was done for each source
population separately (Lacassine and Atchafalaya) to
determine what factors were responsible for variation in
temperature at death. A lognormal regression model was
used in survival analysis to determine the effects of age,
sex, size, incubator, and shelf location within each incubator using StatViewÒ. All of the effects were significant
except, for the Atchafalaya population, the effects of fish
size and shelf within incubator. Therefore, a fish’s expected
123
C. M. Doyle et al.
TAD was calculated on the basis of its characteristics (age,
sex, and size) and testing conditions (shelf and incubator
used for the exposure), using the regression coefficients
from the survival analysis. The difference between the
observed TAD for a fish and the expected TAD for this fish
was used as a measure of a fish’s innate temperature tolerance. For consistency, the effects of fish size and shelf
location within the incubator were also taken into account
for the Atchafalaya line, even though these effects were not
statistically significant. Identical corrections were used for
the non-bottlenecked and bottlenecked lines of each source
population.
Heritabilities were estimated from mid-parent/mean
offspring regressions for all lines. JMP INÒ version 5.1
(SAS Institute Inc., Carey NC, USA) was used for these
regression analyses. For mid-parent/mean offspring
regressions, the regression coefficient is the heritability
estimate. These regressions are well suited for providing
estimates of the narrow sense heritability, since resemblances between parents and offspring do not contain any
non-additive genetic components (Falconer and Mackay
1996). The regressions were weighted by the number of
offspring produced by each parental pair, necessitated by
the fact that numbers of offspring per family were unequal.
The number of offspring used for a family ranged from 2 to
12, and averaged 9.9 for the Lacassine non-bottlenecked
line, 9.6 for the Lacassine bottlenecked line, 8.8 for the
Atchafalaya non-bottlenecked line, and 7.4 for the
Atchafalaya bottlenecked line. Since mid-parent/mean
offspring regressions do not provide valid heritability
estimates if variances differ between males and females
(Falconer and Mackay 1996), Fmax tests were done to
determine whether variances differed between males and
females. A Bonferroni adjustment was used since multiple
tests were conducted. The variances did not differ significantly between the sexes.
A full-sib analysis was done for each line to estimate
heritability without reference to the parents. The full-sib
analyses were based on analysis of variance (ANOVA)
using JMP INÒ. The total observed variance was partitioned into two components; between full-sib families and
within full-sib families (Falconer and Mackay 1996).
Because of unequal family sizes, a weighted estimate for
family size was used and the standard error was adjusted
for unequal family size (Roff 1997). One sample t-tests
were performed for both ANOVA and regression-based
heritability estimates to assess whether estimates were
significantly higher than zero (one-tailed test). Also, a
z-test was conducted to determine whether there were
differences between bottlenecked and non-bottlenecked
lines for each source population. Since the covariance of
full sibs is not only due to additive genetic variance but
also a portion of the dominance variance, heritability
Author's personal copy
Heritability of thermal tolerance
539
estimates based on full sibs tend to overestimate narrow
sense heritability (Falconer and Mackay 1996).
Results
Heat tolerance differed among sexes and among generations (Table 1). TAD was higher in females than it was in
males (v2 = 396.467, p \ 0.0001), higher in adult offspring than in parents (v2 = 130.474, p \ 0.0001), higher
in fish from the Atchafalaya population than in those from
the Lacassine population (v2 = 21.554, p \ 0.0001), and
higher in fish from the regular lines than in those from
the bottlenecked lines (v2 = 48.081, p \ 0.0001). Also,
TAD was negatively related to fish size (v2 = 21.946,
p \ 0.0001; data not shown).
As described earlier, an individual’s difference between
expected TAD (expected on basis of its size and other
characteristics, as well as specific testing conditions) and
observed TAD was used as the basic datum in determining
heritabilities of heat tolerance. The heritability estimates
obtained from the mid-parent/mean offspring regressions
ranged from 0.19 to 0.46 (Table 2). The estimate for the
regular Lacassine line was not significantly higher than
zero (t = 1.312, p = 0.0965), while the estimate for the
Lacassine bottlenecked line was (t = 2.007, p = 0.0279).
However, these two estimates did not differ statistically
(z = -0.5864, p = 0.5576). The heritability estimate of
the Atchafalaya non-bottlenecked line was numerically the
highest among the four lines (Table 2), though this estimate was not significantly higher than zero (t = 1.457,
p = 0.0829). In contrast, the numerically lower heritability
estimate for the Atchafalaya bottlenecked line was statistically higher than zero (t = 2.183, p = 0.0359). When
comparing the estimates between the regular and bottlenecked lines of the Atchafalaya population, the numerical
Table 1 Comparisons of average unadjusted temperature-at-death (TAD ± standard deviation, SD) in °C, for the Lacassine and Atchafalaya
H. formosa populations, non-bottlenecked (NB) and bottlenecked (B) lines
Group
Source and line
Lacassine
Atchafalaya
NB
B
NB
B
Fathers
38.27 ± 0.33 (39)
38.23 ± 0.19 (32)
38.29 ± 0.22 (19)
38.37 ± 0.17 (14)
Mothers
Sons
38.77 ± 0.21 (39)
38.71 ± 0.36 (210)
38.55 ± 0.27 (32)
38.59 ± 0.34 (190)
38.84 ± 0.21 (16)
38.88 ± 0.29 (78)
38.74 ± 0.24 (7)
38.80 ± 0.31 (59)
Daughters
39.20 ± 0.32 (215)
38.99 ± 0.45 (197)
39.32 ± 0.24 (90)
39.25 ± 0.23 (44)
All males
38.64 ± 0.41 (249)
38.54 ± 0.35 (222)
38.77 ± 0.36 (97)
38.72 ± 0.33 (73)
All females
39.13 ± 0.34 (254)
38.93 ± 0.46 (229)
39.25 ± 0.29 (106)
39.18 ± 0.29 (51)
Values are listed by combination of sex and generation and again by sex. Sample sizes are listed in parentheses. Overall, TAD was higher in
females than in males, higher in offspring than in parents, higher in Atchafalaya fish than in Lacassine fish, and higher in regular lines than in
bottlenecked lines
Table 2 Heritability estimates (with standard errors) for heat tolerance quantified as temperature-at-death (TAD), for the least killifish
H. formosa of the Lacassine and the Atchafalaya populations
Source
Non-bottlenecked line
n
Bottlenecked line
2
h ± SE
h2 ± SE
n
(A) Mid-parent/mean offspring heritability estimates
Lacassine population
35
0.185 ± 0.141 (ns)
-
26
0.307 ± 0.153 (*)
Atchafalaya population
16
0.462 ± 0.317 (ns)
-
7
0.275 ± 0.126 (*)
Lacassine population
43
0.138 ± 0.070 (*)
-
40
0.155 ± 0.077 (*)
Atchafalaya population
19
0.324 ± 0.030 (*)
?
14
Zeroa (ns)
(B) Full-sib heritability estimates
Heritabilities were obtained (A) from mid-parent/mean offspring regression analyses, and (B) from full-sib analyses. Indicated in parenthesis is
whether heritabilities were ‘‘(*)’’ or were not ‘‘(ns)’’ significantly higher than zero; ‘‘-’’ and ‘‘?’’ signs indicate whether estimates differed
significantly between non-bottlenecked and bottlenecked lines
n number of families, h2 heritability, SE standard error
a
Actual estimate was -0.107 ± 0.081
123
Author's personal copy
540
value was higher for the regular line than the bottlenecked
Atchafalaya line—though the estimates did not differ statistically (z = 0.4427, p = 0.3290).
The estimates obtained from the full-sib analyses tended
to be numerically lower than the mid-parent/mean offspring
regression estimates (Table 2). Heritability estimates ranged
from 0 to 0.324, with all three positive estimates being
significantly higher than zero in one-sample t-tests (p values
for Lacassine non-bottlenecked line, Lacassine bottlenecked
line and Atchafalaya non-bottlenecked line being respectively 0.028, 0.026 and \0.0001). The full-sib heritability
estimates did not differ statistically between the bottlenecked and non-bottlenecked Lacassine lines (z = 0.1634,
p = 0.8702). In contrast, the heritability estimates of the
non-bottlenecked and bottlenecked Atchafalaya lines differed drastically (z = 4.9898, p \ 0.0001).
Because the full-sib based estimates did not appear to
overestimate the heritabilities, combining the parent–offspring and full-sib based estimates seems reasonable and
provides a single estimate facilitating further discussion.
Combining the estimates, with averages weighted by the
sample size of each estimate, yielded heritabilities of heat
tolerance in the regular and bottlenecked lines of 0.230 and
0.168, respectively.
Discussion
This study’s encountered differences in heat tolerance
between the sexes and among least killifish populations, are
consistent with previous studies (Baer and Travis 2000;
Forster-Blouin 1989; Klerks and Blaha 2009). ForsterBlouin (1989) found heat tolerance differed between two
Florida populations of H. formosa, and that adult female
fish had a higher heat tolerance than did adult males. In the
present study, offspring also had a higher heat tolerance
than did their parents, even though all fish were exposed as
adults. However, as parents were tested after sufficient
offspring were produced, they were generally older than
their offspring at the time of heat tolerance quantification.
It was logistically impossible to test all fish at the exact
same age. Thus the difference in heat tolerance between the
two generations may have been due to age differences.
There is some evidence of age affecting thermal tolerance
in fish (Benfey et al. 1997). In the present study, differences in heat tolerance due to size, sex, testing conditions,
etc., would bias the heritability estimates (inflate the nongenetic variance). Therefore, heritabilities were determined
using the difference between each fish’s observed and
expected TAD (with the expected TAD based on the fish’s
characteristics and specific testing conditions—see
‘‘Materials and methods’’ section) rather than the raw TAD
data.
123
C. M. Doyle et al.
For all lines, the full-sib heritability estimates tended to
be numerically lower (though more often statistically
‘‘larger than zero’’ as a consequence of higher precision)
than those obtained from the mid-parent/mean offspring
regressions. This was surprising, because heritability estimates obtained from full-sib analyses tend to overestimate
the actual heritability. The overestimate is due to the
resemblance among full sibs being not only due to the
shared additive genetic variance, but also due to a shared
portion of the dominance variance and due to sibs often
having a common rearing environment (Falconer and
Mackay 1996). Dominance variance may be relatively
high, especially for traits with close ties to fitness (Crnokrak and Roff 1995). Estimates based on mid-parent/mean
offspring regressions contain only VA. This indicates that
VD and the VE from the common rearing environment were
low, thus the full-sib estimates were used as additional
estimates of heritability of thermal tolerance.
Heritability estimates for the bottlenecked and nonbottlenecked lines were not totally consistent. However,
overall averages were lower for the bottlenecked lines than
the regular lines and the one case where a statistically
significant difference was found between the bottlenecked
and regular lines involved a higher heritability in the regular line. Thus there is some evidence that the population
bottleneck had resulted in a reduced ability for natural
selection to bring about an adaptation to elevated water
temperatures. In general, one would expect a decrease in
heritable variation with loss of genetic variability from a
bottleneck or founder event (Falconer and Mackay 1996;
Wright 1931). However, this is not always the case. Bottleneck events have been found to ‘‘convert’’ non-additive
genetic variance to additive genetic variance, and this
conversion is likely in specific situations discussed earlier
in this paper (Bryant et al. 1986; López-Fanjul et al. 1999;
Naciri-Graven and Goudet 2003; Saccheri et al. 2001).
Most studies that have estimated heritability of thermal
tolerance in aquatic organisms have reported relatively low
heritabilities. For example, Elderkin et al. (2004) found an
overall heritability estimate of 0.327 in zebra mussel
veligers. However, this estimate (obtained using resemblance among sibs) appeared to be inflated by maternal
and/or environmental effects, and the more reliable heritability estimate based on the sire component was basically
zero. Also, Baer and Travis (2000) estimated heritability
for thermal tolerance in H. formosa to be B0.15, and
Charo-Karisa et al. (2005) determined that heritability for
cold tolerance in the Nile tilapia was low as well
(h2 = 0.09 ± 0.19 for TAD). Meffe et al. (1995) reported
a somewhat higher heritability; an estimated heritability of
0.32 for TAD. The overall heritability estimate of 0.20 falls
in the middle range of these various estimates, and
is consistent with other heritability studies involving
Author's personal copy
Heritability of thermal tolerance
physiological traits (Meffe et al. 1995; Mousseau and Roff
1987).
In conclusion, this study found a low-moderate heritability of heat tolerance in two laboratory populations of
H. formosa. This indicated that these fish have some
potential to adapt to global warming, since the heritabilities
indicate that these populations can show a response, albeit a
slow one, when selected for an increased thermal tolerance.
Various factors will determine whether this potential will be
realized in natural populations undergoing heat stress
resulting from climate change. It is known that heritabilities
are specific to the trait used (this study’s TAD at a temperature increase of 2°C/day is of course far removed from any
global change scenario), the specific population and species
being tested, and by the structure of the model used to
estimate heritabilities (Wilson 2008; Chown et al. 2009).
Laboratory to field extrapolation may also be an issue; while
it appears that lab and field estimates are generally similar
(Weigensberg and Roff 1996), this conclusion may not hold
for traits that are not closely related to fitness (Charmantier
and Garant 2005). Another factor that may reduce the longterm response to selection is the presence of fitness costs that
may be associated with adaptations; such costs were reported for the same least killifish species responding to selection
for resistance to cadmium (Xie and Klerks 2004). And
finally, this study’s results may not be transferable to larger
and more stenothermal fish species. Thus while the current
research indicates that certain fish populations have some
potential to genetically adapt to climate change, further
insights are needed in order to assess the long-term ecological impacts of elevated temperatures.
Acknowledgments The authors thank Susan deVries and Margo
Blaha for assistance in maintaining fish populations; the university’s
Center for Ecology and Environmental Technology for housing of the
fish; and James Albert, Johannes Rick and anonymous reviewers for
comments on earlier versions of this paper. This research was supported by a combined grant from the USEPA and the Louisiana Board
of Regents Support Fund. Although the research described in this
article has been funded, in part, by the United States Environmental
Protection Agency through grant R-82942001-0 to the Louisiana
Board of Regents, it has not been subjected to the Agency’s peer and
policy review process and therefore does not necessarily reflect the
views of the Agency and no official endorsement should be inferred.
References
Baer CF, Travis J (2000) Direct and correlated responses to artificial
selection on acute thermal stress tolerance in a livebearing fish.
Evolution 54:238–244
Beitinger TL, Bennett WA, McCauley RW (2000) Temperature
tolerances of North American freshwater fishes exposed to
dynamic changes in temperature. Environ Biol Fish 58:237–275
Benfey TJ, McCabe LE, Pepin P (1997) Critical thermal maxima of
diploid and triploid brook charr, Salvelinus fontinalis. Environ
Biol Fish 49:259–264
541
Bryant EH, McCommas SA, Combs LM (1986) The effect of an
experimental bottleneck upon quantitative genetic variation in
the housefly. Genetics 114:1191–1211
Charmantier A, Garant D (2005) Environmental quality and evolutionary potential: lessons from wild populations. Proc R Soc B
272:1415–1425
Charo-Karisa H, Rezek MA, Bovenhuis H, Komen H (2005)
Heritability of cold tolerance in Nile tilapia, Oreochromis
niloticus, juveniles. Aquaculture 249:115–123
Chown SL, Jumbam KR, Sørensen JG, Terblanche JS (2009)
Phenotypic variance, plasticity and heritability estimates of
critical thermal limits depend on methodological context. Funct
Ecol 23:133–140
Crnokrak P, Roff DA (1995) Dominance variance: associations with
selection and fitness. Heredity 75:530–540
Elderkin CL, Stoeckel JA, Klerks PL, Berg DJ (2004) Heritability of
heat tolerance in zebra mussel veligers. J Great Lakes Res
30:360–366
Elliott JM (1981) Some aspects of thermal stress on freshwater
teleosts. In: Pickering AD (ed) Stress and fish. Academic Press,
New York, pp 209–245
Emerson K, Russo RC, Lund RE, Thurston RV (1975) Aqueous
ammonia equilibrium calculations: effect of pH and temperature.
J Fish Res Board Can 32:2379–2383
Falconer DS, Mackay TFC (1996) Introduction to quantitative
genetics, 4th edn. Longman, Essex
Forster-Blouin SL (1989) Genetic and environmental components of
thermal tolerance in the least killifish, Heterandria formosa.
Dissertation, Florida State University
Goldspink G (1995) Adaptation of fish to different environmental
temperature by qualitative and quantitative changes in gene
expression. J Therm Biol 20:167–174
Grebmeier JM, Overland JE, Moore SE (2006) A major ecosystem
shift in the Northern Bering Sea. Science 311:1461–1464
Hari RE, Livingstone DM, Siber R, Burkhardt-Holm P, Güttinger H
(2006) Consequences of climatic change for water temperature
and brown trout populations in Alpine rivers and streams. Glob
Change Biol 12:10–26
Hartl DL (2000) A primer of population genetics, 3rd edn. Sinauer,
Sunderland
Intergovernmental Panel on Climate Change (IPCC) (2007a) Climate
change 2007: the physical science basis. In: Solomon S, Qin D,
Manning M et al (eds) Contribution of working group I to the
fourth assessment report of the intergovernmental panel on
climate change. Cambridge University Press, Cambridge
Intergovernmental Panel on Climate Change (IPCC) (2007b) Climate
change 2007: Impacts, adaptation and vulnerability. In: Pary
ML, Canziani OF, Palutikof JP, Van der Linden PJ, Hanson CE
(eds) Contribution of working group II to the fourth assessment
report of the intergovernmental panel on climate change.
Cambridge University Press, Cambridge
Klerks PL, Blaha MA (2009) Heat tolerance as an indicator of climate
change impacts: general considerations and a case study in
poeciliid fish. Environ Bioindic 4:46–66
Leips J, Travis J (1999) The comparative expression of life-history
traits and its relationship to the numerical dynamics of four
populations of the least killifish. J Anim Ecol 68:595–616
López-Fanjul C, Fernández A, Toro MA (1999) The role of epistasis
in the increase in additive genetic variance after population
bottlenecks. Genet Res 73:45–59
Meffe GK, Weeks SC, Mulvey M, Kandl KL (1995) Genetic
differences in thermal tolerance of eastern mosquitofish (Gambusia holbrooki; Poeciliidae) from ambient and thermal ponds.
Can J Fish Aquat Sci 52:2704–2711
Mousseau TA, Roff DA (1987) Natural selection and the heritability
of fitness components. Heredity 59:181–197
123
Author's personal copy
542
Naciri-Graven Y, Goudet J (2003) The additive genetic variance after
bottlenecks is affected by the number of loci involved in epistatic
interactions. Evolution 57:706–716
Page LM, Burr BM (1991) A field guide to freshwater fishes of North
America north of Mexico. Houghton Mifflin, New York
Roff DA (1997) Evolutionary quantitative genetics. Chapman and
Hall, New York
Saccheri IJ, Nichols RA, Brakefield PM (2001) Effects of bottlenecks
on quantitative genetic variation in the butterfly Bicyclus
anynana. Genet Res 77:167–181
United States Environmental Protection Agency (USEPA) (1998)
1998 Update of ambient water quality criteria for ammonia.
Office of Water, Washington, DC, EPA 822 R-89-008
Walther GR (2010) Community and ecosystem responses to recent
climate change. Philos Trans R Soc Lond B 365:2019–2024
123
C. M. Doyle et al.
Walther GR, Post E, Convey P, Menzel A, Parmesan C, Beebee TJC,
Fromentin JM, Hoegh-Guldberg O, Bairlein F (2002) Ecological
responses to recent climate change. Nature 416:389–395
Weigensberg I, Roff DA (1996) Natural heritabilities: can they be
reliably estimated in the laboratory? Evolution 50:2149–2157
Wilson AJ (2008) Why h2 does not always equal VA/VP. J Evol Biol
21:647–650
Wright S (1931) Evolution in Mendelian populations. Genetics
28:114–138
Xie L, Klerks PL (2004) Fitness costs of resistance to cadmium in the
least killifish (Heterandria formosa). Environ Toxicol Chem
23:1499–1503