DEVELOPMENTAL DYNAMICS 234:815–823, 2005

REVIEWS—A PEER REVIEWED FORUM

Fishing for the Secrets of Vertebrate Evolution in Threespine

Catherine L. Peichel*

The threespine (Gasterosteus aculeatus) is rapidly emerging as a new model genetic system to study questions at the interface of evolution and development. The relatively rapid and recent diversification of this small teleost fish, combined with the development of genetic and genomic tools for this fish, provides an unprecedented opportunity to identify the genetic and molecular basis of morphological variation in natural populations of vertebrates. Recently, the genes underlying two different adaptive morphological traits in stickleback have been identified. This work has provided answers to four longstanding questions in the field of evolution and development: (1) How many genes underlie morphological variation in natural populations? (2) What are the genes that underlie morphological variation in natural populations? (3) Do coding or regulatory mutations underlie morphological evolution? (4) What is the molecular and genetic basis of parallel morphological evolution? Because stickleback populations also display natural variation in morphology, life history, physiology, and behavior, extending the approaches used to identify the genetic basis of morphological variation in sticklebacks to other phenotypes is sure to yield further important insights into the genetic and developmental basis of diversity in natural populations. Developmental Dynamics 234:815–823, 2005. © 2005 Wiley-Liss, Inc.

Key words: threespine stickleback; genetics; evolution; pelvic reduction; plate morph

Received 3 July 2005; Revised 22 July 2005; Accepted 26 July 2005

INTRODUCTION molecular pathways used to achieve THE THREESPINE the same phenotype? What are the mechanisms that under- STICKLEBACK AS A Many of these questions have chal- lie the variation of forms found in na- GENETIC MODEL SYSTEM lenged evolutionary biologists for over ture? Are the differences between spe- IN EVOLUTION AND a century. More recently, a number of cies due to the effects of many genes, DEVELOPMENT each with a small phenotypic effect, or researchers working at the interface can differences between occur of evolution and development have Threespine sticklebacks (Gasterosteus as a result of mutations in genes with also begun to study these questions. aculeatus) are small teleost fish that large phenotypic effects? Are there In order to take a genetic approach to have been widely and intensively particular classes of genes that are answering these questions, it is neces- studied by ethologists, ecologists, and more likely to be modified to produce sary to identify natural populations evolutionary biologists, resulting in novel phenotypes? Do mutations in that have significant phenotypic vari- thousands of research papers and sev- these genes occur in protein coding ation, but can still be intercrossed in eral books on their morphology, phys- regions or regulatory regions? When the laboratory in order to identify the iology, life history, ecology, evolution, similar traits evolve in independent genetic and molecular mechanisms and behavior. Sticklebacks are widely populations, are the same genetic and that underlie morphological variation. distributed in both marine and fresh-

Division of Human Biology, Fred Hutchinson Cancer Research Center, Seattle, Washington Grant sponsor: National Institute of General Medical Sciences; Grant number: GM71854; Grant sponsor: Burroughs Wellcome Fund; Grant number: 1003370. *Correspondence to: Catherine L. Peichel, Division of Human Biology, Fred Hutchinson Cancer Research Center, 1100 Fairview Ave North, Seattle, WA 98109. E-mail: [email protected] DOI 10.1002/dvdy.20564 Published online 26 October 2005 in Wiley InterScience (www.interscience.wiley.com).

© 2005 Wiley-Liss, Inc. 816 PEICHEL water populations throughout the cis-regulatory elements that drive ex- (Baumgartner and Bell, 1984), para- northern hemisphere. The ancestral pression of genes in particular ana- site susceptibility (MacLean, 1980), sticklebacks are found in marine hab- tomical locations (DiLeone et al., increased body flexibility and changes itats and are relatively uniform in life 2000; Mortlock et al., 2003). With the in swimming performance (Taylor and history, morphology, and behavior. In development of many of these tools McPhail, 1986; Bergstrom, 2002), as contrast, the sticklebacks found in that were previously only available in well as changes in predation regime post-glacial freshwater lakes and more traditional model organisms, (Hagen and Gilbertson, 1973; Moodie streams have evolved a diversity of threespine sticklebacks are rapidly et al., 1973; Reimchen, 1992, 1995, morphologies, life histories, and be- emerging as a “supermodel” (Gibson, 2000) and climate (Hagen and Moodie, haviors from the ancestral marine 2005). 1982) have all been suggested as fac- form in response to local ecological Variation in skeletal armor, high- tors contributing to the evolution of conditions in a relatively short time lighted in Figure 2, is the most strik- lateral plate reduction. However, none scale of 12,000 years (Bell and Foster, ing phenotype found in natural popu- of these factors have been directly 1994). Despite this diversity, stickle- lations of sticklebacks. Marine proven to account for the repeated backs from virtually any two popula- sticklebacks are covered in bony ar- evolution of lateral plate morphs in tions from around the world can be mor, with three dorsal spines, two pel- thousands of freshwater populations crossed in the laboratory to generate a vic spines, and a continuous row of across the Northern hemisphere (Bell large number of viable and fertile bony lateral plates (Fig. 2A). The dor- and Foster, 1994). progeny, thereby facilitating the for- sal and pelvic spines (Fig. 2A) are Using a forward genetics approach ward genetic approach outlined in thought to be protective against gape- (Fig. 1), the genetic and molecular ba- Figure 1 to identify the molecular ba- limited predators, such as birds and sis of pelvic reduction and plate morph sis of natural phenotypic variation. piscivorous fish (Hoogland et al., 1957; variation has now been identified (Co- In order to identify the genes and Reimchen, 1983). However, several losimo et al., 2004; 2005; Cresko et al., molecular changes that underlie nat- freshwater populations from around 2004; Shapiro et al., 2004). This work ural variation, a variety of genetic and the world have evolved a complete or in sticklebacks has provided answers genomic tools are required. Despite partial loss of the pelvic skeleton (Fig. to some long-standing and fundamen- the long history of study in stickle- 2B), including sticklebacks from Pax- tal questions in evolutionary biology. backs, there were virtually no molec- ton Lake, British Columbia (Bell, In this review, I will frame each ques- ular or genetic tools available for this 1974), the Queen Charlotte Islands, tion in a historical context, briefly fish until a few years ago. There are British Columbia (Moodie and Reim- summarize some relevant work in now hundreds of microsatellite mark- chen, 1976; Reimchen, 1984), Quebec other taxa, and then describe the re- ers available, which are distributed (Edge and Coad, 1983), the Cook Inlet cent work in sticklebacks and its rel- across the genome and have been or- region of Alaska (Bell et al., 1993), evance to these questions. I will end dered into a genetic linkage map for southern California (Bell, 1987), the by discussing some of the future re- use in genetic linkage analysis in vir- Outer Hebrides in Scotland (Camp- search directions that can be taken in tually any stickleback population bell, 1979, 1985), and Iceland (Shapiro this newly emerging model for inte- (Peichel et al., 2001). Several bacterial et al., 2004). It has been hypothesized grative research in evolution and de- artificial chromosome (BAC) libraries that loss of pelvic structures in stick- velopment. have been constructed to facilitate po- lebacks could be the result of the ab- sitional cloning of genetic loci identi- sence of predatory fish, reduced levels fied by genetic linkage analysis of calcium availability, or predation by HOW MANY GENES (Kingsley et al., 2004). Nearly 100,000 macroinvertebrates (Reimchen, 1980, UNDERLIE ESTs from a variety of tissues and 1983; Reist, 1980a,b; Giles, 1983; Bell MORPHOLOGICAL developmental stages have now been et al., 1993; Ziuganov and Zotin, VARIATION IN NATURAL deposited in GenBank, and provide a 1995), although none of these hypoth- POPULATIONS? resource for the cloning of interesting eses can completely account for all candidate genes (Kingsley et al., cases of pelvic reduction. Does evolution proceed via the accu- 2004). Expression patterns of these The most commonly observed mor- mulation of small changes with minor candidate genes can be assayed using phological variation in freshwater phenotypic effects or can mutations of in situ hybridization (Ahn and Gibson, sticklebacks is a reduction in the num- large phenotypic effect underlie evolu- 1999a, 1999b; Cresko et al., 2003; ber of lateral plates (Fig. 2B). Most tionary change? This question has Shapiro et al., 2004; Tanaka et al., marine sticklebacks, although not all been debated since the time of Dar- 2005). Transgenic technologies have (Klepaker, 1996), have a complete set win, and Darwin himself believed that also been developed for sticklebacks of lateral plates along their side (Fig. natural selection could only occur via (Hosemann et al., 2004), which make 2A, complete morph), and most fresh- small steps with minor effects on fit- it possible to perform genetic rescue water populations have either re- ness (Darwin, 1859). The micromuta- experiments to demonstrate that a tained only the anterior plates (low tionist view of evolution was champi- candidate gene is actually responsible morph) or anterior plates and a poste- oned by R.A. Fisher, who created a for a phenotype (Colosimo et al., rior keel (partial morph). Low calcium theoretical framework that reconciled 2005). Additionally, transgenic ap- levels (Giles, 1983), salinity tolerance gradualism and Mendelian genetics proaches can be used to identify the (Heuts, 1947), stream gradients (Fisher, 1930). Although challenged GENETICS OF EVOLUTION IN STICKLEBACKS 817

Fig. 1. Forward genetic approach to identify the genes and molecular changes that underlie natural variation in sticklebacks. with divergent phenotypes (morphologies, physiologies, behaviors) are crossed in the laboratory to generate an F1 generation, which are then intercrossed to generate hundreds of F2 progeny. The progeny are then phenotyped for traits of interest and genotyped with a panel of polymorphic molecular markers. Region(s) of the genome in which genotypes are correlated with phenotypes are defined. Candidate genes in these regions are identified, either through positional cloning using BACs and sequencing of the genetic interval, or by genetic mapping of interesting candidate genes to these regions. These candidate genes must then be analyzed for the presence of sequence or expression differences between the starting populations, which correlate with the phenotypic differences between the starting populations. Ultimately, rescue and complementation experiments are required to prove that a sequence variant is responsible for the phenotype observed.

by scientists such as Bateson and Goldschmidt (Bateson, 1894; Gold- schmidt 1940), micromutationism re- mained the prevailing theory until nearly the end of the 20th century (re- viewed in Orr, 2005). However, the advent of quantitative trait locus (QTL) analysis in the 1980s provided an experimental approach to deter- mine both the number of loci and their relative contributions to phenotypic differences between natural popula-

Fig. 2. Threespine stickleback populations have striking differences in overall morphology and skeletal structures. A: Alizarin red–stained stickleback from a Japanese Pacific marine population. The fish has robust dorsal and pel- vic spines (arrow), and a set of fully developed lateral plates along its side. The morphology of this fish is typical for a marine population. B: Alizarin red–stained stickleback from the Pax- ton Benthic freshwater population. The fish has smaller dorsal spines, completely absent pelvic structures, and only a single lateral plate. In addition, fish from this population are smaller than the marine fish and have differences in Fig. 2 body and head shape. 818 PEICHEL tions. A growing body of experimental many more loci of smaller effect (Orr, genes can underlie morphological dif- evidence, mostly from work in insects 1998). Genetic mapping of other skel- ferences between species (Stern, 1998; and plants, suggests that relatively etal traits in sticklebacks, such as fa- Sucena and Stern, 2000; Wittkopp et few genetic loci with major effects on cial bone morphology (Kimmel et al., al., 2003); however, very few studies phenotype can contribute to morpho- 2005), gill raker number, and dorsal have been conducted. logical differences between natural spine length (Peichel et al., 2001) also The genes that underlie pelvic and populations (reviewed in Orr, 2001, support the hypothesis that relatively plate morph variation in sticklebacks 2005). few genes of major effect underlie have also been identified, providing To determine the genetic architec- morphological adaptations in stickle- the first insights into the types of ture that underlies skeletal changes backs. However, these crosses were genes that control morphological vari- in natural populations of sticklebacks, too small to detect loci of smaller ef- ation in vertebrates. The major pelvic a Japanese marine fish with a com- fect. Future studies in both stickle- locus on LG7 does not recombine with plete set of lateral plate and pelvic backs and other taxa are required to the Pitx1 gene, a homeodomain con- structures (Fig. 2A) was crossed with assess the generality of the theoretical taining protein originally identified in a Paxton benthic fish with only a few prediction about the distribution of a screen for hindlimb-specific genes in plates and a missing pelvis (Fig. 2B). loci fixed during adaptation. mouse (Shang et al., 1997). The phe- Intercrossing F1 progeny resulted in a notype of mice with a targeted muta- large number of F2 progeny with seg- tion in the Pitx1 gene is remarkably regating variation in plate number WHAT ARE THE GENES similar to the phenotype of stickle- and pelvic size, as well as in a number THAT UNDERLIE backs with pelvic reduction in two as- of other traits that differed between MORPHOLOGICAL pects. First, Pitx1 null mice have the starting populations. When the VARIATION IN NATURAL reduced hindlimbs and normal fore- progeny were genotyped with a large limbs (Lanctoˆt et al., 1999; Szeto et number of microsatellite markers POPULATIONS? al., 1999), and Paxton Lake stickle- spanning the threespine stickleback Although we still know very little backs have reduced pelvic spines and genome (Peichel et al., 2001), one ma- about the genes that underlie morpho- girdles (fish hindlimbs) and normal jor locus and four minor loci were logical variation in natural popula- pectoral fins (fish forelimbs). Second, identified that together explained a tions, work during the past 20 years in hindlimb reduction is more severe on large proportion of the variance in pel- developmental genetics has yielded the right side of the body in the Pitx1 vic phenotypes (Shapiro et al., 2004). insights into the genes that underlie mutant, due to compensation by the The major locus accounts for close to the development of many different or- closely related Pitx2 gene, which is ex- 70% of the variance in pelvic reduction gans and structures within model or- pressed at higher levels on the left in sticklebacks from Paxton Lake, ganisms such as fruit flies (Drosophila side of the body in the mouse (Marcil British Columbia, and maps to the melanogaster), nematodes (Caeno- et al., 2003). The reduction of pelvic end of linkage group (LG) 7. A similar rhabditis elegans), and mice (Mus structures in the Paxton benthic pop- analysis for the plate morph pheno- musculus). Surprisingly, most devel- ulation, as well as several other pelvic type identified a near Mendelian locus opmental regulatory genes are con- reduced stickleback populations (Bell on LG 4 that controlled nearly 80% of served across metazoans (Carroll, et al., 1985; Cole et al., 2003), is also the phenotypic variation in plate 2000). Early work in evolutionary de- more severe on the right side than the counts. Four additional loci were iden- velopmental biology correlated differ- left. This directional asymmetry also tified that contributed to significant ences in the expression patterns of maps to the Pitx1 locus in the Paxton variation in plate number and plate these conserved genes, such as Hox benthic cross (Shapiro et al., 2004). size. These loci interact in a semi-ad- genes, with morphological differences To identify the gene that underlies ditive manner to determine overall across widely divergent taxa (Averof the major plate morph locus, a com- plate counts (Colosimo et al., 2004). and Akam, 1995; Burke et al., 1995; pletely unbiased positional cloning ap- These data suggest that evolution of Carroll, 1995; Sordino et al., 1995; proach was taken, which makes no as- significant morphological differences Holland and Garcia-Fernandez, 1996; sumption about the identity of the between populations can indeed occur Averof and Patel, 1997; Cohn and genes involved (Colosimo et al., 2005). by changes in a relatively small num- Tickle, 1999; Hughes and Kaufman, The plate morph locus was genetically ber of loci with major phenotypic ef- 2002). However, this approach cannot mapped to a small (0.68 cM) genetic fects. identify the actual genetic changes interval on LG4. Bacterial artificial For both pelvic girdle and plate that contribute to the evolution of chromosome (BAC) clones covering morph variation, it was interesting morphologies. Furthermore, it is lim- the genetic interval were then identi- that a single major locus with a large ited to only looking for variation in fied and two BACs covering most of phenotypic effect and multiple loci known genes. A more direct forward the interval were sequenced. A num- with smaller phenotypic effects were genetic approach provides an unbi- ber of genes were present in this found. This experimental evidence ased search for the genes that under- 400-kb interval, but linkage disequi- supports the theoretical prediction lie evolutionary changes in natural librium mapping narrowed the region that the distribution of loci fixed dur- populations. Genetic approaches in to a small 16-kb interval that con- ing adaptation will be exponential, Drosophila have provided some evi- tained only a handful of genes (Colo- with a single locus of large effect and dence that known developmental simo et al., 2005). One of these genes, GENETICS OF EVOLUTION IN STICKLEBACKS 819

Ectodysplasin (Eda), a member of the 1999). However, with few exceptions overall viability of the organism in the tumor necrosis family, was particu- (Belting et al., 1998; Wang and Cham- wild (Stern, 2000; Davidson, 2001; Car- larly interesting. Mutations in the berlin, 2002; Gompel et al., 2005), the roll et al., 2004). Eda gene, as well as in the Ectodys- actual molecular changes that underlie Support for the cis-regulatory hy- plasin receptor (Edar) and down- evolution at cis-regulatory elements pothesis is less clear in the case of the stream signaling molecules in the have yet to be identified. Eda gene and lateral plate morph evo- pathway, result in alterations in skin, Although many cases of morpholog- lution. Although there are four pre- hair, and tooth formation in humans ical evolution are the result of regula- dicted amino acid differences between and mice (Mikkola and Thesleff, tory evolution, there are examples of the completely plated marine and low 2003). In addition, a mutation in Edar dramatic morphological changes in plated Paxton benthic in the Eda pro- causes the loss of scales in medaka natural populations that result from tein sequence, none of the changes are fish (Kondo et al., 2001). Sticklebacks amino acid changes within the pro- in particularly well-conserved resi- do not have scales, but the lateral tein. One of the most striking is the dues (Colosimo et al., 2005). In addi- plates are dermal bone, which shares occurrence of activating mutations in tion, there are no coding differences in a common developmental origin with the melanocortin1 receptor (Mc1r)in Eda between the marine population scales (Sire and Huysseune, 2003). Al- melanistic populations of birds and a second low-plated population, in though Eda proved to be a compelling (Theron et al., 2001; Mundy et al., which the low phenotype fails to com- candidate once it was discovered to lie 2004), jaguars (Eizirik et al., 2003), plement the low phenotype of a popula- within the plate morph interval, the and pocket mice (Nachman et al., tion known to carry the low-plated Eda Eda gene would probably not have 2003). The occurrence of natural vari- allele (Schluter et al., 2004; Colosimo et topped the candidate gene list for the ation in the coding region of this gene al., 2005). Taken together, these data plate morph phenotype. Therefore, may result from the fact that the role suggest that the difference in the plate this story highlights the power of an of Mc1r is largely confined to pigment phenotypes is due to a regulatory rather unbiased forward genetics approach, cells (Robbins et al., 1993); therefore, than a protein coding change in Eda. rather than relying on a candidate there may be no negative pleiotropic However, attempts to compare the ex- gene approach, to identify genes that effects on fitness when the coding re- pression of the Eda gene in complete underlie morphological evolution. Re- gion of this gene is mutated in the and low morph fish have not yet been gardless of the fact that Eda may not wild. Given the limited nature of these successful. This future work will con- have been the gene expected to under- studies, further comparative analysis tribute to a greater understanding of lie plate morph evolution, transgenic is required to determine what types of both the development of the lateral rescue experiments prove that the genes might tolerate coding or regula- plates and the mechanisms by which Eda gene is in fact responsible for the tory variation in natural populations. they are lost in evolution. loss of lateral plates in freshwater fish In the case of pelvic reduction in In the future, it will be particularly (Colosimo et al., 2005). sticklebacks, no changes in the amino exciting to identify the cis-acting ele- acid sequence of Pitx1 were found be- ments that drive expression of both tween the Japanese marine (complete Pitx1 and Eda in particular anatomi- DO CODING OR pelvis) and Paxton benthic (reduced pel- cal areas and to then compare the se- REGULATORY MUTATIONS vis) populations (Fig. 2). Therefore, the quence of these elements in different UNDERLIE expression pattern of Pitx1 was exam- stickleback populations. The ongoing MORPHOLOGICAL ined by in situ hybridization at time sequencing of the stickleback genome, points prior to the appearance of pelvic as well as the ability to make trans- EVOLUTION? structures. Pitx1 is expressed in a vari- genic sticklebacks (Hosemann et al., Mutations of cis-regulatory elements in ety of larval structures, including the 2004) makes this a worthwhile and genes have been proposed to provide the prospective pelvic region in marine lar- feasible, though challenging, goal. fodder for the evolution of morphologi- vae with full pelvic development. In cal diversity as it provides a mechanism Paxton benthic larvae, Pitx1 is ex- to alter expression in a specific struc- pressed normally in most structures but WHAT IS THE MOLECULAR ture while preserving expression at is not expressed in the presumptive pel- AND GENETIC BASIS OF other sites required for viability of the vic region. These data suggest that a PARALLEL organism (King and Wilson, 1975; cis-regulatory element, which specifi- MORPHOLOGICAL Stern, 2000; Davidson, 2001; Carroll et cally drives expression of Pitx1 in the EVOLUTION? al., 2004). Genetic studies have also pro- developing pelvis, is altered in the Pax- vided evidence for cis-regulatory evolu- ton benthic pelvic reduced sticklebacks. The evolution of similar phenotypes in tion underlying morphological variation Although this putative cis-regulatory independent populations in associa- in natural populations of fruit flies element has not yet been identified in tion with similar environmental fac- (Stern, 1998; Sucena and Stern, 2000; sticklebacks, these data provide strong tors implies that a trait evolved in re- Wittkopp et al., 2002; Sucena et al., support for the hypothesis that cis-reg- sponse to natural selection, rather 2003; Gompel et al., 2005), worms ulatory evolution of developmental con- than by genetic drift (Simpson, 1953; (Wang and Chamberlin, 2002), mam- trol genes can provide a simple mecha- Schluter and Nagel, 1995; Rundle et mals (Belting et al., 1998; Van Laere et nism for natural selection to tinker with al., 2001). The evolution of similar al., 2003), and plants (Wang et al., one structure without affecting the phenotypes can result from the same 820 PEICHEL underlying mechanisms, termed “par- locus for pelvic reduction at the end of ever, the low-plated populations do allel evolution,” or as a result of differ- LG7 (Cresko et al., 2004). Crosses be- not share a common origin when ana- ent mechanisms, termed “convergent tween pelvic reduced sticklebacks lyzed with 25 random nuclear mark- evolution” (Hodin, 2000). However, from Paxton Lake and pelvic reduced ers. The most parsimonious explana- the mechanisms underlying the evolu- sticklebacks from Iceland resulted in tion for this data is that there is an tion of similar phenotypes in indepen- progeny lacking pelvic structures ancient low-plated allele present at a dent populations have not been well (Shapiro et al., 2004). These data low frequency in the ancestral marine understood. It has been suggested suggest that pelvic reduction in population that is selected for when that genetic and developmental con- several independent populations of the fish move into fresh water. In fact, straints (Haldane, 1932; Maynard threespine sticklebacks is controlled this low-plated allele can be found at Smith et al., 1985; Wake, 1991; Shu- by the same major genetic locus. How- low frequencies in anadromous popu- bin et al., 1995; West-Eberhard, 2003) ever, the expression of the Pitx1 gene lations from both Canada and Califor- shared between closely related species has not yet been analyzed in Alaskan nia (Colosimo et al., 2005). However, may bias the direction of phenotypic or Icelandic pelvic reduced popula- there is evidence that at least one low- evolution and lead to parallel evolu- tions. The expression of Pitx1 has plated population in Japan does not tion. These hypotheses predict that been examined and was found to be share the low-plated Eda haplotype; the same genes will underlie parallel lost in a Scottish pelvic reduced popu- despite the fact that complementation evolution because there are only a lim- lation, as was seen in the Paxton crosses indicate the low-plated pheno- ited number of loci that will not have benthic population (Cole et al., 2003; type is allelic to Eda (Schluter et al., negative pleiotropic effects on fitness Shapiro et al., 2004). However, nei- 2004; Colosimo et al., 2005). There- when mutated. An alternative expla- ther genetic mapping studies nor fore, this may represent a situation in nation for the existence of parallel complementation analysis have been which new mutation at the same lo- evolution in closely related lineages is performed with this population, so it cus, rather than standing genetic vari- that cryptic genetic variation that is is currently unknown whether pelvic ation, contributes to the evolution of a present in the ancestral population is reduction in this Scottish population trait. uncovered when the organism is ex- is due to a mutation at the Pitx1 gene There are also cases of convergent posed to a novel environment (Bell, or in another gene upstream of Pitx1. evolution in which the same genes do 1974, 1988; Gibson and Dworkin, Ultimately, it will be interesting to not underlie traits in independent 2004). determine whether the same or differ- populations. Consistent with this, Parallel phenotypic evolution has ent molecular alterations at the Pitx1 there are stickleback populations in been recently demonstrated in dros- locus underlie pelvic reduction in in- which the developmental mode of pel- ophilid flies, where variation in pig- dependent stickleback populations vic reduction does not look similar to mentation and hair patterns is wide- from across the world to determine the populations that have been stud- spread, and can result from genetic whether genetic and developmental ied genetically (Bell, 1987). These may changes at the same locus (Gompel constraints or pre-existing variation represent populations in which other and Carroll, 2003; Sucena et al., plays a role in the parallel evolution of genes are responsible for pelvic reduc- 2003). However, the molecular nature pelvic reduction. tion. In addition, the minor loci that of these changes is unknown; there- The positional cloning of the plate contribute to pelvic spine and plate fore, it is unclear from these studies morph locus has revealed that pre-ex- variation do differ between stickle- whether parallel evolution is the re- isting variation in the ancestral ma- back populations (Peichel et al., 2001; sult of new mutations or pre-existing rine population can explain the rapid Colosimo et al., 2004; Cresko et al., genetic variation in an ancestral pop- (Klepaker, 1993; Bell, 2001; Kristja´ns- 2004). Convergence of pigment pat- ulation. Melanism in multiple verte- son et al., 2002; Bell et al., 2004) and terning in Drosophila is not always brate species results from mutations repeated evolution of lateral plate due to the same genetic loci (Wittkopp in the Mc1r gene (Theron et al., 2001; morphs in freshwater populations et al., 2003) and Mc1r does not always Eizirik et al., 2003; Nachman et al., from around the world (Colosimo et underlie melanism in vertebrates 2003; Mundy et al., 2004). In these al., 2005). Genetic mapping studies as (Hoekstra and Nachman, 2003). Fu- cases, the changes in Mc1r are clearly well as complementation crosses sug- ture comparative work in multiple the result of independent mutation at gested that the repeated evolution of taxa will be required to assess the rel- the same locus, providing evidence to the low morph phenotype is the result ative contributions of new mutations, support the idea that negative pleoi- of the same major genetic locus (Avise, pre-existing genetic variation, and de- tropic effects of some loci may limit 1976; Colosimo et al., 2004; Cresko et velopmental and genetic constraints the number of targets of selection. al., 2004; Schluter et al., 2004). Se- to the evolution of similar phenotypes In sticklebacks, the genetic basis of quencing around the plate morph lo- in independent populations. pelvic reduction has been investigated cus revealed a common DNA haplo- in multiple populations with pelvic re- type that is shared between 15 low- FUTURE DIRECTIONS duction. Genetic mapping and comple- plated populations from around the mentation studies were performed us- world, while a different DNA haplo- Perhaps the most exciting aspect of ing three different pelvic-reduced type is shared between 10 completely research in sticklebacks is not what populations from Alaska, and these plated populations from around the we have learned so far, but the things studies all revealed a near-Mendelian world (Colosimo et al., 2005). How- we have yet to learn using this sys- GENETICS OF EVOLUTION IN STICKLEBACKS 821 tem! This review has focused on the used different model systems to iden- case of parallel evolution. Nat Hist Mus genetic basis of morphological varia- tify different levels of causation. For LA Contrib Sci 257:1–36. tion in skeletal traits, but stickleback example, the genetic basis of adapta- Bell MA. 1987. Interacting evolutionary constraints in pelvic reduction of populations have extensive variation tion has been studied in laboratory threespine sticklebacks, Gasterosteus ac- in other morphological traits, such as organisms such as Drosophila, while uleatus (Pisces, Gasterosteidae). Biol J feeding morphologies, body shape and population level processes, such as the Linn Soc 31:347–382. size, and breeding coloration; behav- selective forces that lead to adaptation Bell MA. 1988. Stickleback fishes: bridging the gap between population biology and ioral traits, such as courtship behav- have been studied in a different set of paleobiology. Trends Ecol Evol 3:320– iors, social aggregation, levels of ag- organisms in the field. In sticklebacks, 325. gression, and predator avoidance; and we have the opportunity to under- Bell MA. 2001. Lateral plate evolution in physiological traits, such as lifespan, stand evolutionary causation at many the threespine stickleback: getting no- salinity tolerance, and temperature levels, from the selective forces that where fast. Genetica 112–113:445–461. Bell MA, Foster SA. 1994. The evolution- preference (Bell and Foster, 1994). lead to phenotypic evolution down to ary biology of the threespine stickleback. Studying multiple traits in stickle- the single nucleotide changes in the Oxford: Oxford University Press. 571 p. backs will allow us to discern if there genome upon which selection acts. It Bell MA, Francis RC, Havens AC. 1985. are more general rules concerning the is only through these integrated, mul- Pelvic reduction and its directional genetic basis of phenotypic variation tidisciplinary approaches that we will asymmetry in threespine sticklebacks from the Cook Inlet region, Alaska. Co- in natural populations. Are there par- achieve a greater understanding of peia 1985:437–444. ticular classes of traits that are more the basic principles that have resulted Bell MA, Ortı´ G, Walker JA, Koenings JP. likely to have a simple genetic basis? in the biological diversity we see 1993. Evolution of pelvic reduction in Are there differences in the genetic around us. threespine stickleback fish: a test of com- peting hypothesis. Evolution 47:906– architecture of traits that are lost dur- 914. ing evolution vs. traits that are gained Bell MA, Aguirre WE, Buck NJ. 2004. during evolution? Will we identify ACKNOWLEDGMENTS Twelve years of contemporary armor novel genes and gene functions I thank Jun Kitano for help with fig- evolution in a threespine stickleback through this approach? What are the ures, and Mike Bell, Brian Fritz, population. Evolution 58:814–824. Belting HG, Shashikant CS, Ruddle FH. relative contributions of cis- regula- Steve Froggett, and an anonymous re- 1998. Modification of expression and cis- tory and protein coding variation to viewer for helpful comments on the regulation of Hoxc8 in the evolution of evolution? Will parallel evolution al- manuscript. Work in my laboratory is diverged axial morphology. Proc Natl ways involve mutations at the same supported in part by a Career Award Acad Sci USA 95:2355–2360. in the Biomedical Sciences from the Bergstrom CA. 2002. Fast-start perfor- loci? mance and reduction in lateral plate In order to address these questions Burroughs Wellcome Fund. number in threespine stickleback. Can J and gain a fuller understanding of the Zool 80:207–213. processes driving evolution in natural Burke AC, Nelson CE, Morgan BA, Tabin populations, it will also be necessary REFERENCES C. 1995. Hox genes and the evolution of vertebrate axial morphology. Develop- to analyze multiple traits from taxo- Ahn DG, Gibson G. 1999a. Axial variation ment 121:333–346. nomically diverse species. In this post- in the threespine stickleback: relation- Campbell RN. 1979. Sticklebacks (Gaster- genomic age, genetic and molecular ship to Hox gene expression. Dev Genes osteus aculeatus [L.], and Pungitius pun- tools can be brought to virtually any Evol 209:473–481. gitius [L.]) in the Outer Hebrides, Scot- organism with interesting biology, Ahn DG, Gibson G. 1999b. Expression pat- land. Hebridean Nat 3:8–15. terns of threespine stickleback Hox Campbell RN. 1985. Morphological varia- and we need not limit our studies to a genes and insights into the evolution of tion in the three-spined stickleback small handful of traditional genetic the vertebrate body axis. Dev Genes Evol (Gasterosteus aculeatus) in Scotland. Be- model systems. The threespine stick- 209:482–494. haviour 93:161–168. leback system provides a compelling Averof M, Akam M. 1995. Hox genes and Carroll SB. 1995. Hox genes and the evo- the diversification of insect and crusta- lution of arthropods and . Na- example of what can be learned by cean body plans. Nature 376:420–423. ture 376:479–485. using a system with biology that is Averof M, Patel NH. 1997. Crustacean ap- Carroll SB. 2000. Endless forms: the evo- uniquely suited to the questions at pendage evolution associated with lution of gene regulation and morpholog- hand. changes in Hox gene expression. Nature ical diversity. Cell 101:577–580. One of the really exciting aspects of 388:682–686. Carroll SB, Grenier JK, Weatherbee SD. Avise JC. 1976. Genetics of plate morphol- 2004. From DNA to diversity: molecular studying the genetic basis of natural ogy in an unusual population of genetics and the evolution of de- variation in sticklebacks is the possi- threespine sticklebacks (Gasterosteus ac- sign, 2nd ed. Malden, MA: Blackwell Sci- bility of vertical integration across uleatus). Genet Res 27:33–46. ences. 258 p. traditionally disparate fields. For ex- Bateson W. 1894. Materials for the study of Cohn MJ, Tickle C. 1999. Developmental ample, genetic studies of skeletal vari- variation treated with especial regard to basis of limblessness and axial pattern- discontinuity in the origin of species. ing in snakes. Nature 399:474–479. ation in sticklebacks have brought London: Macmillan. 598 p. Cole NJ, Tanaka M, Prescott A, Tickle C. together developmental geneticists, Baumgartner JV, Bell MA. 1984. Lateral 2003. Expression of limb initiation genes evolutionary biologists, and field ecol- plate morph variation in California pop- and clues to the morphological diversifi- ogists. Although evolutionary biolo- ulations of the threespine stickleback, cation of threespine stickleback. Curr Gasterosteus aculeatus. Evolution 38: Biol 13:R951–952. gists have long been concerned with 665–674. Colosimo PF, Peichel CL, Nereng K, Black- identifying the causes of phenotypic Bell MA. 1974. Reduction and loss of the man BK, Shapiro MD, Schluter D, evolution, they have traditionally pelvic girdle in Gasterosteus (Pisces): a Kingsley DM. 2004. The genetic archi- 822 PEICHEL

tecture of parallel armor plate reduction threespine sticklebacks. Heredity 30:273– Lanctoˆt C, Moreau A, Chamberland M, in threespine sticklebacks. PloS Biol 2: 287. Tremblay ML, Drouin J. 1999. Hindlimb 635–641. Hagen DW, Moodie GEE. 1982. Polymor- patterning and mandible development Colosimo PF, Hosemann KE, Balabhadra phism for plate morphs in Gasterosteus require the Ptx1 gene. Development 126: S, Villareal G, Dickson M, Grimwood J, aculeatus on the east coast of Canada 1805–1810. Schmutz J, Myers RM, Schluter D, and an hypothesis for their global distri- MacLean J. 1980. Ecological genetics of Kingsley DM. 2005. Widespread parallel bution. Can J Zool 60:1032–1042. threespine sticklebacks in Heisholt evolution in sticklebacks by repeated fix- Haldane JBS. 1932. The causes of evolution. Lake. Can J Zool 58:2026–2039. ation of ectodysplasin alleles. Science 307: London: Harper and Brothers. 234 p. Marcil A, Dumontier E, Chamberland M, 1928–1933. Heuts MJ. 1947. Experimental studies on Camper SA, Drouin J. 2003. Pitx1 and Cresko WA, Yan YL, Baltrus DA, Amores adaptive evolution in Gasterosteus ac- Pitx2 are required for development of A, Singer A, Rodriguez-Mari A, Postleth- uleatus L. Evolution 1:89–102. hindlimb buds. Development 130:45–55. wait JH. 2003. Genome duplication, sub- Hodin J. 2000. Plasticity and constraints in Maynard Smith J, Burian R, Kauffman S., function partitioning, and lineage diver- development and evolution. Exp Zoolog Alberch P, Campbell J, Goodwin B, gence: Sox9 in stickleback and zebrafish. B Mol Dev Evol 288:1–20. Lande R, Raup D, Wolpert L. 1985. De- Dev Dyn 228:480–489. Hoekstra HE, Nachman MW. 2003. Differ- velopmental constraints and evolution. Cresko WA, Amores A, Wilson C, Murphy ent genes underlie adaptive melanism in Q Rev Biol 60:265–287. J, Currey M, Phillips P, Bell MA, Kim- different populations of rock pocket mice. Mikkola ML, Thesleff I. 2003. Ectodyspla- mel CB, Postlethwait JH. 2004. Parallel Mol Ecol 12:1185–1194. sin signaling in development. Cytokine genetic basis for repeated evolution of Holland PW, Garcia-Fernandez J. 1996. Growth Factor Rev 14:211–224. armor loss in Alaskan threespine stick- Hox genes and evolution. Dev Moodie GEE, Reimchen TE. 1976. Phenetic leback populations. Proc Natl Acad Sci Biol 173:382–395. variation and habitat differences in Gas- USA 101:6050–6055. Hoogland RD, Morris D, Tinbergen N. terosteus populations of the Queen Char- Darwin CR. 1859. The origin of species. 1957. The spines of sticklebacks (Gaster- lotte Islands. Syst Zool 25:49–61. London: J. Murray. 502 p. osteus and Pygosteus) as means of de- Moodie GEE, McPhail JD, Hagen DW. Davidson EH. 2001. Genomic regulatory fense against predators (Perca and 1973. Experimental demonstration of se- systems: development and evolution. Esox). Behavior 10:205–236. lective predation on Gasterosteus aculea- San Diego: Academic Press. 261 p. Hosemann KE, Colosimo PF, Summers tus. Behavior 47:95–105. DiLeone RJ, Marcus GA, Johnson MD, BR, Kingsley DM. 2004. A simple and Mortlock DP, Guenther C, Kingsley DM. Kingsley DM. 2000. Efficient studies of efficient microinjection protocol for mak- 2003. A general approach for identifying long-distance Bmp5 gene-regulation us- ing transgenic sticklebacks. Behaviour distant regulatory elements applied to the ing bacterial artificial chromosomes. 141:1345–1355. Gdf6 gene. Genome Res 13:2069–2081. Proc Natl Acad Sci USA 97:1612–1617. Hughes CL, Kaufman TC. 2002. Exploring Mundy NI, Badcock NS, Hart T, Scribner Edge TA, Coad BW. 1983. Reduction of the the myriapod body plan: expression pat- K, Janssen K, Nadeau N. 2004. Con- pelvic skeleton in the three-spined stick- terns of the ten Hox genes in a centipede. served genetic basis of a quantitative leback Gasterosteus aculeatus in 2 lakes Development 129:1225–1238. plumage trait involved in mate choice. ofQuebecCanada.CanField-Nat97:334– Kimmel CB, Ulmann B, Walker C, Wilson Science 303:1870–1873. 336. C, Currey M, Phillips PC, Bell MA, Nachman MW, Hoekstra HE, D’Agostino SL. 2003. The genetic basis of adaptive Eizirik E, Yuhki N, Johnson WE, Menotti- Postlethwait JH, Cresko WA. 2005. Evo- melanism in pocket mice. Proc Natl Acad Raymon M, Hannah SS, O’Brien SJ. lution and development of facial bone Sci USA 100:5268–5273. 2003. Molecular genetics and evolution morphology in threespine sticklebacks. Orr HA. 1998. The population genetics of of melanism in the cat family. Curr Biol Proc Natl Acad Sci USA 102:5791–5796. adaptation: the distribution of factors 13:448–453. King MC, Wilson AC. 1975. Evolution at two levels in humans and chimpanzees. fixed during adaptive evolution. Evolu- Fisher RA. 1930. The genetical theory of Science 188:107–116. tion 52:935–949. natural selection. Oxford: Oxford Uni- Kingsley DM, Zhu B, Osoegawa K, DeJong Orr HA. 2001. The genetics of species dif- versity Press. 272 p. PJ, Schein J, Marra M, Peichel C, ferences. Trends Ecol Evol 16:343–350. Gibson G. 2005. The synthesis and evolu- Amemiya C, Schluter D, Balabhadra S, Orr HA. 2005. The genetic theory of adap- tion of a supermodel. Science 307:1890– Friedlander B, Cha YM, Dickson M, tation: a brief history. Nat Rev Genet 1891. Grimwood J, Schmutz J, Talbot WS, My- 6:119–127. Gibson G, Dworkin I. 2004. Uncovering ers R. 2004. New genomic tools for mo- Peichel CL, Nereng KS, Ohgi KA, Cole BLE, cryptic genetic variation. Nature Rev lecular studies of evolutionary change in Colosimo PF, Buerkle CA, Schluter D, Genet 5:681–690. sticklebacks. Behaviour 141:1331–1344. Kingsley DM. 2001. The genetic architec- Giles N. 1983. The possible role of environ- Klepaker T. 1993. Morphological changes ture of divergence between threespine mental calcium levels during the evolu- in a marine population of threespined stickleback species. Nature 414:901–905. tion of phenotypic diversity in Outer He- stickleback, Gasterosteus aculeatus, re- Reimchen TE. 1980. Spine deficiency and bridean populations of three-spined cently isolated in fresh water. Can J Zool polymorphism in a population of Gaster- stickleback, Gasterosteus aculeatus.J 71:1231–1258. osteus aculeatus: an adaptation to pred- Zool 199:535–544. Klepaker T. 1996. Lateral plate polymor- ators. Can J Zool 58:1232–1244. Goldschmidt R. 1940. The material basis of phism in marine and estuarine popula- Reimchen TE. 1983. Structural relation- evolution. New Haven, CT: Yale Univer- tions of the threespine stickleback (Gas- ships between spines and lateral plates sity Press. 436 p. terosteus aculeatus) along the coast of in threespine stickleback (Gasterosteus Gompel N, Carroll SB. 2003. Genetic mech- Norway. Copeia 1996:832–838. aculeatus). Evolution 37:931–946. anisms and constraint governing the Kondo S, Kuwahara Y, Kondo N, Naruse Reimchen TE. 1984. Status of unarmoured evolution of correlated traits in drosophi- K, Mitani H, Wakamatsu Y, Ozato K, and spine-deficient populations (Char- lid flies. Nature 424:931–934. Asakawa S, Shimizu N, Shima A. 2001. lotte Unarmoured Stickleback) of Gompel N, Prud’homme B, Wittkopp PJ, The medaka rs-3 locus required for scale threespine stickleback, Gasterosteus sp., Kassner VA, Carroll SB. 2005. Chance development encodes ectodysplasin-A re- on the Queen Charlotte Islands, British caught on the wing: cis-regulatory evolu- ceptor. Curr Biol 11:1202–1206. Columbia. Can Field Nat 98:120–126. tion and the origin of pigment patterns Kristja´nsson BK, Sku´ lason S, Noakes Reimchen TE. 1992. Injuries on stickle- in Drosophila. Nature 433:481–487. DLG. 2002. Rapid divergence in a re- backs from attacks by a toothed predator Hagen DW, Gilberston LG. 1973. Selective cently isolated population of threespine (Oncorhyncus) and implications for the predation and the intensity of selection stickleback (Gasterosteus aculeatus). evolution of lateral plates. Evolution 46: acting upon the lateral plates of Evol Ecol Res 4:659–672. 1224–1230. GENETICS OF EVOLUTION IN STICKLEBACKS 823

Reimchen TE. 1995. Predator-induced Taricha granulosa (Caudata: Salaman- freshwater threespine stickleback, Gas- cyclical changes in lateral plate fre- dridae): evolutionary and phylogenetic terosteus aculeatus. Can J Zool 64:416– quencies of Gasterosteus. Behaviour 132: implications. Evolution 49:874–884. 420. 1079–1094. Simpson GG. 1953. The major features of Theron E, Hawkins K, Bermingham E, Reimchen TE. 2000. Predator handling evolution. New York: Columbia Univer- Ricklefs RE, Mundy NI. 2001. The mo- failures of lateral plate morphs in Gas- sity Press. 434 p. lecular basis of an avian plumage poly- terosteus aculeatus: implications for sta- Sire JY, Huysseune A. 2003. Formation of morphism in the wild: A melanocortin-1- sis and distribution of the ancestral plate dermal skeletal and dental tissues in receptor point mutation is perfectly condition. Behaviour 137:1081–1096. fish: a comparative and evolutionary ap- associated with the melanic plumage Reist JD. 1980a. Selective predation upon proach. Biol Rev Camb Phil Soc 78:219– morph of the bananaquit, Coereba fla- pelvic phenotypes of brook stickleback, 249. veola. Curr Biol 11:550–557. Culaea inconstans, by northern pike, Sordino P, van der Hoeven F, Duboule D. Van Laere A-S, Nguyen M, Braunschweig Esox lucius. Can J Zool 58:1245–1252. 1995. Hox gene expression in teleost fins M, Nezer C, Collette C, Moreau L, Reist JD. 1980b. Predation upon pelvic and the origin of vertebrate digits. Na- Archibald AL, Haley CS, Buys N, Tally phenotypes of brook stickleback, Culaea ture 375:678–681. M, Andersson G, Georges M, Andersson inconstans, by selected invertebrates. Stern DL. 1998. A role of Ultrabithorax in L. 2003. A regulatory mutation in IGF2 Can J Zool 58:1253–1258. morphological differences between Dro- causes a major QTL effect on muscle Robbins LS, Nadeau JH, Johnson KR, sophila species. Nature 396:463–466. growth in the pig. Nature 425:832–836. Kelly MA, Roselli-Rehfuss L, Baack E, Stern DL. 2000. Evolutionary developmen- Wake DB. 1991. Homoplasy: the result of Mountjoy KG, Cone RD. 1993. Pigmen- tal biology and the problem of variation. natural selection or evidence of design tation phenotypes of variant extension Evolution 54:1079–1091. limitations? Am Nat 138:543–567. locus alleles result from point mutations Sucena E, Stern D. 2000. Divergence of Wang RL, Stec A, Hey J, Lukens L, Doeb- that alter MSH receptor function. Cell larval morphology between Drosophila ley J. 1999. The limits of selection during 72:827–834. sechellia and its sibling species caused maize domestication. Nature 398:236– Rundle HD, Nagel L, Boughman JW, by cis-regulatory evolution of ovo/ 239. Schluter D. 2001. Natural selection and shaven-baby. Proc Natl Acad Sci USA parallel speciation in sympatric stickle- 97:4530–4534. Wang X, Chamberlin HM. 2002. Multiple backs. Science 287:306–308. Sucena E, Delon I, Jones I, Payre F, Stern regulatory changes contribute to the evo- Schluter D, Nagel LM. 1995. Parallel spe- DL. 2003. Regulatory evolution of shav- lution of the Caenorhabditis lin-48 ovo ciation by natural selection. Am Nat 146: enbaby/ovo underlies multiple cases of gene. Genes Dev 16:2345–2349. 292–301. morphological parallelism. Nature 424: West-Eberhard MJ. 2003. Developmental Schluter D, Clifford EA, Nemethy M, Mc- 935–938. plasticity and evolution. New York: Ox- Kinnon JS. 2004. Parallel evolution and Szeto DP, Rodriguez-Esteban C, Ryan AK, ford University Press. 794 p. inheritance of quantitative traits. Am O’Connell SM, Liu F, Kioussi C, Gleiber- Wittkopp PJ, Vaccaro K, Carroll SB. 2002. Nat 163:809–822. man AS, Izpisua-Belmonte JC, Rosen- Evolution of yellow gene regulation and Shang J, Luo Y, Clayton D. 1997. Backfoot feld MG. 1999. Role of the Bicoid-related pigmentation in Drosophila. Curr Biol is a novel homeobox gene expressed in homeodomain factor Pitx1 in specifying 12:1547–1556. the mesenchyme of developing hind hindlimb morphogenesis and pituitary Wittkopp PJ, Williams BL, Selegue JE, limb. Dev Dyn 209:242–253. development. Genes Dev 13:484–494. Carroll SB. 2003. Drosophila pigmenta- Shapiro MD, Marks ME, Peichel CL, Tanaka M, Hale LA, Amores A, Yan Y-L, tion evolution: Divergent genotypes un- Blackman BK, Nereng KS, Jonsson B, Cresko WA, Suzuki T, Postlethwait JH. derlying convergent phenotypes. Proc Schluter D, Kingsley DM. 2004. Genetic 2005. Developmental genetic basis for Natl Acad Sci USA 100:1808–1813. and developmental basis of evolutionary evolution of pelvic fin loss in the puffer- Ziuganov VV, Zotin AA. 1995. Pelvic girdle pelvic reduction in threespine stickle- fish Takifugu rubripes. Dev Biol 281:227– polymorphism and reproductive barriers backs. Nature 428:717–723. 239. in the Pungitius Shubin N, Wake DB, Crawford AJ. 1995. Taylor EB, McPhail JD. 1986. Prolonged pungitius (L.) from northwest Russia. Morphological variation in the limbs of and burst swimming in anadromous and Behaviour 132:1095–1105.