Evolution of growth by genetic accommodation in Icelandic freshwater stickleback

Beren W. Robinson


Classical Darwinian adaptation to a change in environment can ensue when selection favours beneficial genetic variation. How plastic trait responses to new conditions affect this process depends on how plasticity reveals to selection the influence of genotype on phenotype. Genetic accommodation theory predicts that evolutionary rate may sharply increase when a new environment induces plastic responses and selects on sufficient genetic variation in those responses to produce an immediate evolutionary response, but natural examples are rare. In Iceland, marine threespine stickleback that have colonized freshwater habitats have evolved more rapid individual growth. Heritable variation in growth is greater for marine full-siblings reared at low versus high salinity, and genetic variation exists in plastic growth responses to low salinity. In fish from recently founded freshwater populations reared at low salinity, the plastic response was strongly correlated with growth. Plasticity and growth were not correlated in full-siblings reared at high salinity nor in marine fish at either salinity. In well-adapted lake populations, rapid growth evolved jointly with stronger plastic responses to low salinity and the persistence of strong plastic responses indicates that growth is not genetically assimilated. Thus, beneficial plastic growth responses to low salinity have both guided and evolved along with rapid growth as stickleback adapted to freshwater.

1. Introduction

Changes in environment that influence natural selection can drive the Darwinian evolution of traits in populations [1,2] and can also influence phenotype when plastic behavioural, physiological, life-history and morphological traits respond to environment within the lifetime of an individual or its offspring [3,4]. Such phenotypic ‘accommodations’ by individuals have long been recognized by biologists [3,512] but an important task will be to unravel how they shape the ecological [1316] and evolutionary responses [1723] of populations facing environmental change.

Phenotypic plasticity may influence trait evolution by affecting the relationship between genotypic and phenotypic variation and also by affecting fitness [10]. This may limit evolutionary responses to environmental change in three ways. A population of individuals that uniformly express detrimental plastic responses may be disfavoured [2426]. Alternatively, individuals that express an optimal phenotype may face stabilizing selection, and so not evolutionarily diverge from populations in ancestral environments [27]. Lastly, plastic responses that maintain a constant phenotype as environment changes may reduce phenotypic variation available to selection, and so not evolutionarily respond to environmental shifts [1,28].

Plastic responses may also promote evolutionary responses. Partially beneficial plastic responses may permit some genotypes faced with a change in environment to persist, and so provide opportunities for subsequent evolution [5,10,29,30]. Plasticity may also reveal genetic variation under extraordinary or novel conditions that can be acted on by selection [22,29,31,32]; variation that is otherwise hidden in the population under its standard conditions [3335]. When heritable variation in environmentally induced responses influences fitness, then any genetic variation in the form of the plastic response may provide an axis of variation on which selection can act. This can drive the evolution of beneficial plastic responses through a process called ‘genetic accommodation’ [2,3,23,36]. Plasticity may evolve to become more or less responsive to local environmental conditions. Selection may favour less responsiveness when a more homeostatic phenotype requires buffering from environmental variability [24] or when plasticity is costly [37]. The evolution of homeostatic phenotypes from more plastic ancestral phenotypes is referred to as the ‘genetic assimilation’ of a trait [24,29,38,39]. Genetic accommodation of mutation-induced novel phenotypes may also occur by the same process described here for environmentally induced traits [3]. If genetic accommodation is common, then it may be an important component of phenotypic evolution.

Formal models of genetic accommodation theory now include both of the evolutionary phases described above [2,18,27]. For example, in the first generation exposed to the new conditions, the mean phenotypes of some genotypes initially jump partially towards the new optimum phenotype owing to plastic (non-evolutionary) responses [3,5,6,9,11]. Rapid phenotypic trait evolution may ensue when the initial plastic responses are imperfect, genetic variation exists in the form of the initial plastic responses [2,27] and some plastic responses express phenotypes favoured by strong directional selection [26]. Under these conditions, selection on phenotypic variation in the new environment will favour a genetic correlation between phenotype and plastic response that can potentially guide phenotypic evolution [2]. The first phase is completed when the optimum phenotype under the new conditions is achieved primarily by the evolution of strong plastic responses. The second evolutionary phase may involve the genetic assimilation of the trait [38]. Here, less plastic and increasingly developmentally canalized genotypes that express the new phenotype may be favoured [2], particularly if the environment is relatively stable and plasticity is costly. The rate of genetic assimilation may be a function of fitness differences among more and less plastic genotypes, and should proceed slowly when plasticity is not costly [2,37].

Genetic accommodation is consistent with microevolutionary theory and is supported by artificial selection studies [40,41]. Uncertainty of its importance persists because little is known about how plastic responses to new conditions affect individual fitness and phenotypic variation [25,28,42]. Also, few natural examples of plastic trait responses guiding trait evolution are known [40,43,44]. Natural examples may be rare if genetic assimilation regularly erases evidence of intermediate stages where beneficial plastic responses evolved [2,39]. Species that undergo a major ecological transition though, may provide opportunities to test genetic accommodation theory because the large change in environment may induce plastic responses outside their normal range [2426] and generate strong directional selection for a new phenotypic optimum [31,45,46]. Recent transitions can also permit comparisons of plastic responses to ‘old’ and ‘new’ environments among ancestral and derived genotypes to evaluate how trait plasticity and mean coevolve.

I evaluate the interplay between environmental influences on phenotype and microevolution using threespine stickleback (Gasterosteus aculeatus) [47,48]. These small Holarctic fish are ideal because an evolutionarily conservative marine population frequently makes a major ecological transition by colonizing and rapidly diversifying in freshwater environments [47,49,50]. In Iceland, freshwater stickleback have diverged from extant marine forms in body size, shape and coloration [51,52] as elsewhere [47,48]. Phenotypic change in freshwater populations can be rapid [53,54] and is limited by deglaciation in Iceland to less than 10 000 years [55]. Lake stickleback in Iceland are zoobenthivores [51,55], and so have trophically diverged from their generally more zooplanktivore marine ancestor [56]. Rapid diversification is attributed to Darwinian evolution recruiting standing variation in freshwater-adapted alleles present in marine populations [57,58]. The persistence of freshwater alleles is a mystery, although functionally useful cryptic genetic variation in size and other traits is revealed in marine stickleback reared in freshwater [50]. This suggests that plastic responses to freshwater conditions may contribute to colonization success and adaptation in freshwater stickleback [49].

I focus on the evolution of intrinsic growth rate of individuals. Understanding how growth evolves in stickleback is important because size affects sexual and non-sexual fitness; genetic correlations between size and other traits may contribute to phenotypic diversification and size variation can contribute to reproductive isolation through size-assortative mating (reviewed in [50]). Individual growth rate may diverge between freshwater and marine organisms for at least two reasons [46,59,60]: strong selection favouring rapid growth in freshwater and a relaxation of energy trade-offs between growth and osmoregulation in freshwater. Stickleback rarely live beyond 1 year in freshwater at high latitudes and have indeterminate growth. Rapid growth may be strongly favoured in freshwaters because larger body size is thought to increase over-winter survival of juveniles, increase female fecundity [61,62] and permit earlier onset of spring reproduction [58]. Early reproduction provides juveniles with the warmest summer water temperatures and the greatest availability of prey to foster growth [56], particularly as the productive summer shortens with latitude [61,62]. Seasonal limitations on freshwater productivity at high latitudes have frequently driven the evolution of anadromous migration from freshwater to marine environments in temperate and polar fishes that exploit more stable and productive marine resources [63]. Thus, it is likely that the generally larger size of marine compared with freshwater stickleback particularly at high latitudes [50] partially reflects greater resource availability [62].

In marine stickleback, growth may also be strongly governed by trade-offs with other fitness-related traits [62]. Growth probably faces an important energy trade-off with osmoregulation because stickleback have to actively shed ions in water more than 14 ppt and keep ions in freshwater [64]. Osmoregulation is the costliest in seawater at 35 ppt [62], and so stickleback in freshwater face reduced osmotic stress, potentially allowing energy to be reallocated to growth and maturation. If osmotic stress limits energy available for growth or stresses development, then relaxing the costs of osmoregulation may also contribute to the divergence of freshwater from marine stickleback [65].

If individual growth rate rapidly evolves in populations of freshwater stickleback through a process of genetic accommodation, then: (i) standing genetic variation in plastic growth responses to low salinity should be present in ancestral marine populations; (ii) some plastic responses should be partially beneficial under freshwater conditions; (iii) rapid growth should become correlated with beneficial plastic growth responses to low salinity in freshwater populations; and (iv) rapid growth should jointly evolve with stronger plastic growth response to low salinity as populations adapt to freshwater. If growth is genetically assimilated in well-adapted freshwater populations, then the mean plastic response to salinity should be reduced in stickleback with the longest freshwater history. I compare growth responses to high and low salinity among stickleback from marine and freshwater populations in Iceland and find that rapid growth evolves in freshwater populations jointly with positive plastic responses to low salinity, but that these plastic responses do not yet appear to have become genetically assimilated.

2. Material and methods

Wild adult stickleback were sampled from three types of populations representing key stages in the ecological transition from marine to freshwaters: ancestral marine forms (MRN), adapted lake forms (hereafter ‘old freshwater’ or OFW) and a transitional ‘young’ freshwater form (YFW) recently founded by marine stickleback and so intermediate between the other two categories. Three geographically separate populations were sampled from each type of population. Non-estuarine coastal MRN sites were sampled along the west and south coasts of Iceland, and three old postglacial OFW lake populations were sampled from Hegranes Island in the Héru∂svötn river valley in north Iceland (see the electronic supplementary material, appendix S1). Little recent gene flow is assumed among the OFW populations and between OFW and either MRN or other freshwater populations based on the topographic location of lakes. Selected OFW populations were approximately 70 m above the surrounding valley floor with small high-velocity stream outflows that probably restrict migration from valley rivers into the lake populations. The distance to the nearest MRN population was approximately 500 km to the west. The YFW population type (see the electronic supplementary material, appendix S1) included populations from two artificially created ponds near Ne∂riás at the entrance to the Hjaltadalur valley in north Iceland, stocked in 2003 with marine forms from Hvassahraun (two generations in freshwater [52]). The third YFW population was sampled from Hraunsfjör∂ur, where a control dam separated a fjord from the sea in 1987 and so began its conversion into a coastal lake (approx. 18 generations at low salinity [53,54]). Reproductive adults were collected from the three replicate sites of each population type throughout June 2005 using unbaited standard minnow traps placed from the shore.

Each geographical population was represented by replicate full-sibling families made from randomly crossed pairs of adults from that locale (see the electronic supplementary material, appendix S1). Thirty minutes postfertilization, each family was haphazardly divided into two, with each half reared exclusively in either a high- or low-salinity treatment: either filtered and warmed seawater (high = 35 ppt) or filtered and warmed ‘fresh’ water (low = 5 ppt; to reduce fungal infection). Hatching commenced at approximately 10 days post-fertilization. Families were kept separate, first in small 0.5 l plastic cups (up to approx. 40 days post-fertilization) and then in 4 l buckets (up to approx. 150 days post-fertilization) with screen walls suspended in large flow-through tanks (three tanks per salinity treatment). Water temperature, lighting and diet were matched between salinity treatments and the spatial location of containers was regularly varied within tanks.

Rearing methods were intended to maximize growth. Water temperature during the rearing phase started at 10°C and was slowly raised to 18° (±2°) over the first 40 days. Lighting cycles started at 20 L : 4 D to replicate summer conditions and slowly shifted to a permanent 12 L : 12 D cycle by mid-October. Upon exogenous feeding, all fry were fed twice per day on live first instar brine shrimp nauplii for three weeks, and then switched to a high-protein diet of ground commercial juvenile arctic char food. The method focused on identifying the fastest growing individuals in each family by repeatedly culling the smallest individuals at approximately 10 days and again at 40 days post-fertilization to limit final density to the six largest individuals in each family–salinity treatment combination (initial family size range: 6–178 fry per family treatment combination; mean = 63.4; s.d. = 34). I assume that estimates of maximum family growth are related to mean growth. Age was kept constant by euthanizing each family at 150 days post-fertilization (±2 days). Final standard lengths of approximately 40 mm achieved here are comparable to wild adult sizes in Icelandic lakes (population means range from 43.4 to 64.4 mm; [51,52]) but somewhat less than the 56.7 mm of local marine adults [52]. Approximately 8% of females reached sexual maturity, and along with the large body sizes, suggests that growth was observed over a large portion of stickleback life history.

Growth was assessed for each individual as standard body length at 150 days post-fertilization (hereafter ‘SL150’). Plastic responses to salinity treatment were estimated for each family as the difference in mean SL150 values for full-siblings reared under low and under high salinity. Thus, responses to salinity represent a mean full-sibling ‘family’ rather than ‘genotypic’ reaction norm. Mixed statistical models were used to analyse variation in SL150-treating population type (MRN, YFW or OFW) and salinity treatments (high versus low) as fixed effects and family and population as random effects hierarchically nested in population type. The data on individual and mean SL150 are available from the Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.616n0).

3. Results

Rapid growth repeatedly evolved in freshwater compared with marine stickleback populations in Iceland (figure 1). A mixed-model fit to the SL150 data of 1114 fish from 106 families split with full-siblings reared at low and high salinity revealed that population type (MRN, YFW or OFW) accounted for 36% of variation in mean SL150 (F2,6 = 35.5, p = 0.001), with progeny from old freshwater lakes (mean SL150 = 42.9 mm ± 0.66 s.e.m.) significantly larger than those from young freshwater lakes (38.9 mm ± 0.54) and marine sites (37.0 mm ± 0.48).

Figure 1.

Mean standard body length at 150 days of age (SL150) of Icelandic threespine stickleback from nine populations reared in a common environment at low 5 ppt (black bars) and high 35 ppt salinity (white bars). Three types of population are compared, each represented by three replicate populations (see the electronic supplementary material, appendix S1): ancestral marine stickleback (MRN: populations C, D, E), ‘young’ freshwater lakes populations recently founded by marine stickleback (YFW: A, B, H) and old lake populations (OFW: I, J, G). Mean SL150 over all fish was 39.4 mm ± 0.33 s.d.

The effects of rearing salinity on SL150 were complex. Overall, stickleback were slightly larger when reared at low compared with high salinity (figure 1; 2% of variation in SL150; mean 5 ppt = 39.3 mm ± 0.12 s.e.m., and 35 ppt = 38.2 mm ± 0.12; F1,902 = 41.67, p = 0.00002), and so the evolution of more rapid growth in freshwater stickleback paralleled a weak positive size response to low salinity. However, greater variation (8.2%) in SL150 was accounted for by an interaction between salinity and population type (F2,6 = 39.78, p = 0.0003) and an interaction between salinity and family (F97,902 = 1.30, p = 0.03). OFW stickleback responded most strongly to rearing salinity, confirming that plastic responses to salinity also evolve during freshwater adaptation (figure 2). Replicate populations accounted for only 3% of variation in SL150 (F6,97 = 3.46, p = 0.004; replicate population–salinity interaction: p = 0.77) compared to much stronger effects of family within populations (14%; F97,902 = 3.76, p = 0.00001), and so replicate populations were combined in subsequent analyses.

Figure 2.

Family-level plastic responses (reaction norms) to salinity in mean standard length at 150 days of age (mean SL150) among full-sibling threespine stickleback reared at high (35 ppt) versus low (5 ppt) salinity. Families were created by crossing adults within each replicate population of a type of population (ancestral marine, young freshwater founded by marine stickleback and old freshwater lakes).

Genetic variation in SL150 was present in marine stickleback and was consistently enhanced in full-siblings reared at low compared to high salinity (variation among MRN families in SL150 increased from 31% at 35 ppt to 42% for full-siblings reared at 5 ppt; family[poptype] × salinity F28,247 = 1.58, p = 0.04). Broad-sense heritability of SL150 calculated from intraclass correlations for MRN stickleback reared at 5 ppt was 0.82 (s.e. = 0.089) compared with 0.68 (s.e. = 0.132) for full-siblings reared at 35 ppt. A similar increase was observed in YFW stickleback reared at 5 ppt (h2 = 0.94, s.e. = 0.034) compared with 35 ppt (h2 = 0.75, s.e. = 0.111). Heritability of growth in the OFW stickleback was considerably reduced, consistent with strong directional selection for rapid growth in lake environments, and again lower at high salinity (5 ppt: h2 = 0.36, s.e. = 0.138; 35 ppt: h2 = 0.25, s.e. = 0.131). These heritability values probably represent upper bounds because they are estimated on full-siblings, and also combine families from replicate populations of the same type. Relatively few families contributed to the increased variation observed at low salinity (figure 2).

Growth responses by marine stickleback to low salinity were variable and generally lower than OFW forms consistent with less than maximal plastic growth responses [27]. Families from YFW populations also had inconsistent growth responses to low salinity (figure 2; one-sample test of mean plasticity assessed as the difference in SL150 5–35 ppt, t31 = −0.98, p = 0.33; signed-rank p = 0.73), while there was some evidence of consistently positive growth responses to low salinity in MRN stickleback (one-sample t32 = 1.65, p = 0.11; signed-rank p = 0.002). By contrast, OFW stickleback showed uniformly strong positive growth responses to low salinity (figure 2).

A correlation between growth at low salinity and plastic growth responses to salinity was revealed in marine stickleback that had recently colonized freshwaters as expected [2]. In MRN stickleback, SL150 at low salinity was uncorrelated with plasticity (estimated as SL150 5–35 ppt), in contrast to a strong correlation in YFW populations of stickleback reared at low salinity (table 1; increase in correlation: z = 1.78, p1−sided = 0.038). Growth at high salinity was not correlated with plasticity in either MRN or YFW stickleback, suggesting that growth may have been under strong stabilizing selection in the ancestral marine environment [66].

View this table:
Table 1.

Genetic correlations (among families) for mean family growth (SL150) of full-siblings reared at low (5 ppt) and high (35 ppt) salinity, and between SL150 at 5 ppt and growth plasticity (mean SL150 of full-siblings at low minus mean SL150 at high salinity). (Families (n) were combined over replicate populations in each population type. Sequential Bonferroni-corrected significance levels for three simultaneous tests in each population type. *p < 0.001; **p < 0.01.)

Rapid growth and positive growth responses to low salinity jointly evolved in freshwater populations. The positive relationship between SL150 at low salinity and plasticity did not differ between OFW and YFW stickleback (figure 3; ANCOVA; growth plasticity–population type interaction, p = 0.09). But, the mean slope of SL150 responses to salinity was steeper in OFW compared with YFW populations (figure 2; F1,4 = 53.1, p = 0.002). SL150 at high and low salinity were positively correlated among families, particularly in ancestral MRN stickleback (table 1), but this did not limit the evolution of plastic responses in freshwater populations because the correlation declined in stickleback adapting to freshwater (z = 1.83, p1−sided = 0.034).

Figure 3.

Relationship between family-level mean SL150 at low salinity (5 ppt) and plastic responses to rearing salinity (difference in mean SL150 of full-siblings in cm: 5–35 ppt). The vertical dashed line reflects no change (plasticity) in mean SL150 between full-siblings divided between high- and low-salinity treatments. Squares represent families from young freshwater populations (recently founded by marine forms) and circles represent families from old freshwater lakes. Replicate populations within each type of stickleback population (MRN, YFW and OFW) did not differ in the relationship between plasticity and low-salinity growth (ANCOVA; population × growth interaction, all p > 0.35; population, all p > 0.08).

The evolution of stronger beneficial plastic growth responses to low salinity in freshwater stickleback was not reversed through the genetic assimilation of growth in the oldest freshwater populations. OFW stickleback expressed the greatest negative growth response to high salinity of all population types (figure 2). Families varied little in their reduced SL150 at high salinity in OFW stickleback (family–salinity interaction, p = 0.36).

4. Discussion

Studies of natural populations evolutionarily diverging across a major ecological transition provide excellent opportunities to test how environmentally sensitive phenotypes evolve because extraordinary changes in conditions probably stress development [2426] and generate strong selection [31,45,46]. Genetic accommodation theory focuses on how the plastic responses of genotypes to external ecological or internal genetic changes may influence phenotypic evolution [2,3,23,36]. Phenotypic novelty arising in many individuals responding to such a change potentially overcomes the major demographic hurdle faced by rare beneficial mutations, and so may accelerate evolutionary responses [2,3,17]. Genetic accommodation is supported by comparative studies showing parallelism between environmentally induced responses and adaptive radiation [3,67,68], the adaptive divergence of plasticity during ecotype formation [43,45,49,69] and artificial selection studies [40,41]. Here, I provide a rare natural example of how plastic responses can guide phenotypic evolution in a new environment. Assuming that maximum growth is related to mean growth (as opposed to reflecting variance in growth), threespine stickleback in Icelandic lakes evolve faster growth through the evolution of stronger plastic responses to low salinity.

The replicated evolution of rapid growth in freshwater stickleback may have arisen because of strong directional selection for larger size in freshwater environments (reviewed in [50]) or because freshwater relaxes an energy constraint related to osmoregulation at higher marine salinity [62,65]. Relaxed osmoregulatory costs are supported by an average increase in standard length of 2.5% in marine stickleback here reared at low (5 ppt) compared to high (35 ppt) salinity, similar to a 2% increase in size reported in marine stickleback in Alaska reared at 1 ppt [50]. However, selection in lakes may also favour rapid growth because body size is probably related to survival, female fecundity [61,62] and earlier onset of reproduction [58], especially in high-latitude freshwater populations faced with seasonally limited resources [63]. Comparative studies among freshwater stickleback populations over a latitudinal gradient could further test if rapid growth is favoured with latitude. Furthermore, it is too early to conclude that selection on growth is diversifying between marine and freshwaters until we evaluate how selection acts on growth in marine environments.

Standing genetic variation in environmentally induced responses exists in ancestral marine stickleback here as in other studies [50]. Salinity-specific allelic effects may be an important factor in the diversification of threespine stickleback [50,65], but their frequency in marine populations is uncertain [57]. A few families accounted for most of the variation in growth at low salinity in marine stickleback here. However, variation among families rather than individuals here may have underestimated the allelic frequency which are also higher in marine stickleback elsewhere [50]. Thus, genetic variation in salinity-specific effects appears to be available to be recruited by selection on colonizing freshwater environments [70].

Rapid phenotypic evolution in a new environment should be enhanced by a high proportion of expressed additive genetic variation that is owing to genetic variance in partially beneficial plastic responses [2]. Genetic variation in plasticity is a prerequisite for evolutionary response, but near perfect plastic responses to a new environment that could come under stabilizing selection could slow evolution [27]. In marine and young freshwater stickleback, growth was often more rapid than that of full-siblings reared at high salinity, indicating that plastic growth responses could be beneficial in freshwater environments. However, the low-salinity growth of marine and young freshwater stickleback were almost always less than that of stickleback from old freshwater populations which suggests that phenotypes induced by low salinity would be under directional rather than stabilizing selection in freshwaters.

When phenotypic variation expressed in the new environment is primarily owing to genetic variance in plastic responses, then a genetic correlation between plasticity and trait mean can be revealed that shapes evolutionary responses of expressed phenotype and plasticity [2]. Directional selection on phenotypic variation in the new environment will favour genotypes with beneficial plastic responses, and the evolution of the trait mean can be accomplished by the evolution of stronger plastic responses. In stickleback from young freshwater populations, plastic growth responses to low salinity were correlated with higher mean growth. The same correlation was present in the better-adapted old freshwater populations where mean growth was also higher. This strongly suggests that rapid growth evolved along with responses to low salinity in lake populations. This correlation was environmentally sensitive because it was not present in YFW full-siblings reared at high salinity, and can arise very quickly in freshwater populations because growth and plasticity were not correlated in marine stickleback reared at either salinity.

Genetic assimilation occurs when less plastic genotypes replace more costly plastic genotypes usually in a stable environment [38,39,44], although modelling suggests that this process is probably strongly influenced by the relative fitnesses of more and less plastic genotypes [2]. Growth at low salinity did not evolve to become increasingly developmentally unresponsive to salinity even after many thousands of generations in the constant low-salinity environment of Icelandic lakes [71]. The absence of any genetic assimilation of growth in lake stickleback in Iceland contrasts with that of freshwater demes from the St Laurence estuary which have reduced plasticity compared with local saltwater demes [65]. It is possible that in lake stickleback, plastic growth responses to salinity are not relatively costly because they are never expressed, in which case alleles reducing growth at high salinity may persist for long periods in well-adapted freshwater populations and genetic assimilation never occurs. Alternatively, the physiological effects of high marine salinity may be too costly to ever completely buffer [64], and so limit the genetic assimilation of growth. Regardless, the persistence of plastic responses to ancestral marine conditions in derived lake populations [33,50] may facilitate gene flow back into extant ancestral populations, and so help to explain the persistence of freshwater-adapted alleles in marine stickleback [50,57].

The notably repeated evolution of small body size in freshwater stickleback [50] does not appear to be owing to the evolution of slower individual growth rate, at least for Icelandic stickleback. Wild adults from lakes are 9–23% smaller than local marine forms in Iceland [52], with an exception in Lake Myvatn which hosts very large populations of aquatic chironomid insect prey [51]. However, stickleback from old freshwater lakes grew on average 16% larger than marine forms in common rearing environments here, indicating that genetic factors have evolved that enhance growth in these lake stickleback. Body size can vary among populations depending on the relative effects of genetic and environmental influences on growth. For example, when selection favours rapid growth that counteracts detrimental environmental effects on growth, then countergradient growth variation can evolve [72]. Alternatively, when selection favours genetic influences on growth that are in the same direction as environmental influences on growth, then a pattern of co-gradient growth variation is possible [26]. The net environmental effect on size in Iceland lake stickleback depends on the outcome of two opposing effects: a salinity effect where reduced osmotic stress enhances growth at low salinity, and a resource limitation effect that reduces growth in freshwaters. Countergradient variation seems common in lake stickleback because the evolved genetic factors contributing to more rapid growth appear to be weaker than the net negative effect of limited freshwater resources in most Icelandic lakes, reducing mean body size as noted above. Thus, focusing on variation in individual growth rates can provide new insights into the origins of body size variation in stickleback.

The freshwater environment may play two roles in the evolution of growth in threespine stickleback adapting to freshwaters, by revealing environmentally induced growth variants, and favouring those with strong positive responses to low salinity. It will be useful to evaluate how other traits that phenotypically respond to salinity evolve in freshwater stickleback and to compare these to the evolution of traits with little or no responses to salinity. The genetic accommodation of functional traits may be a common evolutionary feature in lineages of stickleback [49], other fishes [60,73] and other aquatic taxa [46,59,74] that have made major ecological transitions. Half of the standing diversity in fishes evolved after colonization of freshwater, and so if diversification across this or other similar ecological transitions is regularly facilitated by the evolution of plastic responses [22], then genetic accommodation may be a common feature of phenotypic evolution.

Animal use was approved by the animal care and use committees of the University of Guelph and of Iceland.

Funding statement

This research was supported by a Discovery grant from the Natural Sciences and Engineering Research Council of Canada.


Numerous friendly Icelanders allowed access onto their lands and waters to collect fish. B. Kristjánsson, S. Skúlason and H. Thorarensen of the Department of Fish Biology and Aquaculture, Hólar University College, Iceland, provided generous logistic support for sampling and rearing of Icelandic stickleback. K. Peiman and R. Sturlaugsdóttur helped in the crossing and rearing of stickleback. W. Cresko of CEEB at the University of Oregon provided analytical space, and along with P. Phillips, L. Chapman, J. Kingsolver and D. Pfennig engaged in valuable conversations about genetic accommodation theory. Three anonymous referees and the associate editor suggested many constructive comments on the manuscript.

  • Received August 22, 2013.
  • Accepted September 24, 2013.


View Abstract