Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Test of postzygotic isolation and parallel inheritance of morphological traits in threespine stickleback… Clifford, Elisabeth Anne 2002

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2002-0365.pdf [ 3.52MB ]
Metadata
JSON: 831-1.0090387.json
JSON-LD: 831-1.0090387-ld.json
RDF/XML (Pretty): 831-1.0090387-rdf.xml
RDF/JSON: 831-1.0090387-rdf.json
Turtle: 831-1.0090387-turtle.txt
N-Triples: 831-1.0090387-rdf-ntriples.txt
Original Record: 831-1.0090387-source.json
Full Text
831-1.0090387-fulltext.txt
Citation
831-1.0090387.ris

Full Text

TESTS O F P O S T Z Y G O T I C ISOLATION AND P A R A L L E L I N H E R I T A N C E O F M O R P H O L O G I C A L TRAITS IN T H R E E S P I N E S T I C K L E B A C K (GASTEROSTEUS ACULEATUS C O M P L E X ) by ELISABETH A N N E CLIFFORD B.Sc, cum laude, Colorado State University, 1996 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T O F T H E REQUIREMENTS FOR T H E D E G R E E O F M A S T E R O F SCIENCE in T H E F A C U L T Y OF G R A D U A T E STUDIES (Department of Zoology) We accept this thesis as conforming To^he^equir^d standard UNIVERSITY OF BRITISH COLUMBIA July 2002 © Elisabeth Anne Clifford, 2002 In p resent ing th is thesis in partial fu l f i lment o f t h e requ i rements fo r an advanced degree at the Univers i ty o f Brit ish C o l u m b i a , I agree that the Library shall make it f reely available f o r re ference and study. I fu r ther agree that permiss ion fo r extens ive c o p y i n g o f this thesis f o r scholar ly pu rposes may b e g ran ted by the head of my d e p a r t m e n t o r by his o r her representat ives. It is u n d e r s t o o d that c o p y i n g or pub l i ca t ion o f this thesis fo r f inancial gain shall n o t be a l l o w e d w i t h o u t my w r i t t e n permiss ion . D e p a r t m e n t of ^ o t ? U O fir 7 The Univers i ty o f Brit ish C o l u m b i a Vancouver , Canada Date Z.2- >) u Z o o 2^ DE-6 (2/88) A B S T R A C T This thesis investigates post-zygotic reproductive isolation, parallel evolution, and parallel inheritance of morphological characters using two pairs of ecologically differentiated populations of threespine stickleback. These populations represent two evolutionarily distinct lineages, one from Japan and one from Canada, that are each represented by one marine and one stream-dwelling population. I used hybridisation experiments to examine degree of postmating isolation among the four populations. I found no significant reduction in Fi or F 2 hybrid fitness relative to pure-bred controls. For my fitness measures I used fertilisation and hatching success of Fi and F2 hybrids. The lack of postzygotic isolation between populations when genetic relationships and times of divergence are considered is consistent with other postzygotic isolation studies in frogs, salamanders, and Drosophila. I used the hybrid fish created in the first experiment to answer three questions regarding the genetic causes of parallel evolution. First, do the two ecologically similar stream populations show parallel evolution of morphology? Second, are morphological characters evolved in parallel inherited in a similar way? And third, is parallel inheritance the result of changes at exactly the same loci or merely at loci that behave similarly? I used two morphological characters for these analyses: lateral plate number and body shape as quantified by a thin-plate spline analysis. These traits represent the extremes of genetic complexity with lateral plate number being under simple genetic control and body shape representing a polygenic composite character. Stream stickleback populations in Canada and Japan exhibited parallel evolution in both plate number and body shape. A line means analysis evaluating inheritance patterns also showed parallelism between the continents: additive-plus-dominance effects for plate number and additive effects for body shape. Furthermore, the parallel inheritance patterns of plate number appear to be the product of changes at the same loci in both stream populations. Although the patterns of inheritance were similar, I was unable to obtain evidence that parallel changes in shape, a more complex genetic trait, were due to changes in the same genes. This is one of very few studies seeking to quantify the genetic changes underlying parallel evolution. iii T A B L E O F C O N T E N T S A B S T R A C T ii T A B L E O F C O N T E N T S iv L I S T O F T A B L E S vi L I S T O F F I G U R E S vi A C K N O W L E D G E M E N T S vii C H A P T E R I. L A C K O F P O S T Z Y G O T I C I S O L A T I O N 1 1.1. Introduction 1 1.2. Materials and methods 4 1.3. Results 8 1.4. Discussion 8 C H A P T E R 2. P A R A L L E L I N H E R I T A N C E O F M O R P H O L O G I C A L T R A I T S 14 2.1. Introduction 14 2.2. Materials and methods 17 2.2.1. Fish populations 17 2.2.2. Measuring phenotypes 20 2.2.2.1. Lateral plate number 20 2.2.2.2. Photograhy 21 2.2.2.3. Landmark selection 21 2.2.2.4. Thin plate spline (TPS) geomorphometric analysis 21 2.2.2.5. Accounting for age and size differences in parental, Fi and F2 generations 22 2.2.2.6. Discriminant function analysis 23 2.2.2.7. Relating landmarks to shape changes described by discriminant functions 23 2.2.3. Quantifying parallel evolution 24 2.2.4. Quantifying parallel inheritance 25 2.2.4.1. Analysis of line means 30 2.2.4.2. Analysis of line variances 30 2.2.5. Comparing weak vs. strong parallel inheritance 34 2.2.5.1. Analysis of line means 34 2.2.5.2. Analysis of line variances 35 2.3. Results ". 36 2.3.1. Tests of parallel evolution 36 2.3.1.1. Lateral plate number 36 iv 2.3.1.2. Body shape 36 2.3.2. Tests of parallel inheritance 38 2.3.2.1. Line means: lateral plate number 38 2.3.2.2. Line means: body shape 39 2.3.2.3. Line variances: lateral plate number 40 2.3.2.4. Line variances: body shape 40 2.3.3. Testing weak vs. strong parallel evolution 40 2.3.3.1. Lateral plate number 40 2.3.3.2. Body shape 41 2.4. Discussion 42 2.4.1. Parallel evolution 42 2.4.1.1. Parallel evolution of ecotypes 42 2.4.1.2. Lateral plate number 45 2.4.1.3. Body shape 46 2.4.2. Parallel inheritance 48 2.4.2.1. Lateral plate number 48 2.4.2.2. Body shape 49 2.4.3. Weak vs. strong parallel inheritance 50 2.4.3.1. Lateral plate number 51 2.4.3.2. Body shape 52 2.4.4. A perspective on parallel evolution and inheritance 53 R E F E R E N C E S 67 v L I S T O F T A B L E S Table 2.1. Averages and standard errors of traits 55 Table 2.2. Line means jr2 test statistics for stream x marine hybrids 56 Table 2.3. Line variances x 2 test statistics for stream x marine hybrids 56 Table 2.4. Line means %2 test statistics for stream x stream hybrids 57 Table 2.5. Line variances x 2 test statistics for stream x stream hybrids 57 Table 2.6. Stream x stream hybrid variances of traits 58 Table 2.7. Stream x marine hybrid variances of traits 58 L I S T O F F I G U R E S Figure 1.1. Regressions showing seasonal improvement in success rates 12 Figure 1.2. Hatching and fertilisation success in all clutches 13 Figure 2.1. Landmark positions 59 Figure 2.2. Parallel and non-parallel evolution described by y 60 Figure 2.3. Discriminant function axes 61 Figure 2.4. Transitions between body shapes 62 Figure 2.5. Changes in landmarks described by discriminant functions 63 Figure 2.6. Inheritance patterns for plate number in stream x marine hybrids 64 Figure 2.7. Inheritance patterns for body shape in stream x marine hybrids 65 Figure 2.8. Inheritance patterns for both traits in stream x stream hybrids 66 vi A C K N O W L E D G E M E N T S This work would not have been possible without the contributions of many people. Researchers before me have provided the broad foundation of knowledge upon which this thesis is based. I owe my supervisor, Dolph Schluter, and committee members, Mike Whitlock and Sally Otto, many thanks for their knowledge, support, instruction, perspectives, and insight. Labmates past and present were invaluable in providing technical assistance and intelligent debate regarding many of the topics presented here. Thanks especially to Jeff McKinnon, Beren Robinson, Durrell Kapan, John Pritchard, Howard Rundle, Steve Vamosi, Jenny Boughman, Tamara Grand, Arne Mooers, Tom Bell, Kyle Young, Troy Day, and other members of the SOWD lab group. Jeff McKinnon and Maria Nemethy were collaborators on the first chapter of this thesis. Jeff, with the help of Seichi Mori, collected the wild fish used in this study. Maria and Jeff made the Fi generation of crosses I used, and without their work I would not have been able to do this project. Volunteers, technicians, and other students provided invaluable help with the routine and not always pleasant tasks that make research possible. Nelson Li , Kirsten Hughes, Jim Herbers, and Mike Melnychuk cheerfully cleaned tanks, fed fish, counted plates, and entered data. Friends and family have encouraged and supported me, providing motivation, inspiration, and balance. Thanks especially to Mum and Papa, John, Chewie, Dawn, Dorothee, Grant, Jaclyn, Jen, Jim, Meredith, Roger, and Rob—I wouldn't have done it without you. I was lucky enough to have a few teachers whose enthusiasm for science was influential enough to change the course of my life. Carl Max and Lee Fitzgerald showed me a world in biology that I never knew existed. This research was funded by a Graduate Research Fellowship from National Science Foundation, and a grant from NSERC to Dolph Schluter. And finally, I gratefully acknowledge all the fishes that gave their lives to make this thesis possible. vii C H A P T E R O N E : L A C K O F P O S T Z Y G O T I C I S O L A T I O N B E T W E E N F O U R E C O L O G I C A L L Y D I F F E R E N T I A T E D P O P U L A T I O N S O F T H R E E S P I N E S T I C K L E B A C K , GASTEROSTEUS ACULEATUS C O M P L E X 1.1 I N T R O D U C T I O N Speciation is the evolution of a reproductive isolating mechanism that prohibits the exchange of genetic material between two populations (Dobzhansky 1940). The isolating mechanisms may be behavioural or ecological preferences that prohibit them from interacting, or they may be an actual physiological incompatibility. Generally, reproductive isolation mechanisms are classified as either premating or postmating mechanisms. Premating reproductive isolation refers to reduced interbreeding between populations or species, while postmating isolation is generally quantified as hybrid inviability or sterility (Mayr 1942, 1963). The expected amount of time for the evolution of reproductive isolation varies depending on whether isolation is pre- or postzygotic and on the populations under consideration. In a study of salamanders (Tilley et al. 1990, Tilley and Mahoney 1996), researchers were unable to predict levels of prezygotic isolation using genetic distances. Both Coyne and Orr (1989, 1997, Drosophila) and Sasa et al. (1998, frogs), however, found that the degree of postzygotic isolation increases with genetic distance, and thus, presumably, with time of divergence between taxa. Furthermore, Coyne and Orr (1989, 1997) found higher levels of prezygotic isolation than postzygotic isolation in young, sympatric taxa, but no difference between levels of premating and postmating isolation among allopatric taxa. 1 Threespine stickleback are a phenotypically and ecologically diverse, circumpolar species complex. The complex consists of hundreds of differentiated populations of freshwater (stream- and lake-dwelling), anadromous, and marine fish (Bell and Foster 1994), which may be allopatric or parapatric, but are rarely sympatric (Bell 1994). The marine forms of Gasterosteus aculeatus are considered to be the ancestral condition from which all freshwater forms have evolved (McPhail and Lindsey 1970, Bell 1976, Bell 1984, Bell 1994, Bell and Foster 1994, McPhail 1994). Mitochondrial D N A evidence from marine populations in the Pacific basin suggests they are members of two distinct clades with a split between Japan/Alaska and the rest of North America (Orti et al. 1994). While the marine populations are phenotypically similar throughout their range (Bell and Foster 1994), morphology varies extensively between independently derived freshwater populations, with a wide range of possibilities for body shape, feeding morphology, and armour characteristics. However, many freshwater populations are more morphologically similar to each other than to the marine forms (Bell and Foster 1994). Despite dramatic, genetically based differences in morphology and ecology between populations, hybrid individuals show few intrinsic genetic incompatibilities manifested as hybrid sterility or inviability (Hagen 1967, McPhail 1984, Hatfield and Schluter 1996). Instead, premating mechanisms based on mate choice (Hay and McPhail 1974) and ecology (Hagen 1967) are the primary determinants of reproductive isolation between most populations examined. Fi and F2 hybrids between limnetic and benthic sticklebacks show reduced fitness in the wild but not in the laboratory, suggesting an ecological, rather than genetic, basis to postzygotic isolation (Hatfield 1997, Rundle and Schluter 1998). The only known instance of hybrid sterility is in Fi and F2 hybrids 2 between a marine population from the Japan sea and a landlocked freshwater population of the Pacific Ocean clade from the Agano River in Japan (Honma and Tamura 1984). However, all attempts to measure postzygotic isolation between populations of G. aculeatus have involved very young populations (Hagen 1967, McPhail 1984, Hatfield 1997, Rundle and Schluter 1998) or relatively old populations (Honma and Tamura 1984). Based on biogeographic evidence (Lindberg 1972, Nishimura 1980, Tada 1998, all cited in Yamada 2001), the Japan sea and Pacific freshwater populations that showed postzygotic isolation diverged at least 2 million years ago and have a Nei's D=0.680 (isozymes, Taniguchi et al. 1985, 1990, cited in Higuchi and Goto 1996) (Honma and Tamura 1984, 1985, Higuchi and Goto 1996, Yamada et al. 2001). While these findings suggest that in sticklebacks, as in Drosophila and frogs, postzygotic isolation builds with time since divergance from a common ancestor, it does not allow us to elucidate the rate. We are left wondering just how much time must elapse between the extremes of age before postzygotic isolation is apparent between populations. This study was designed to test whether postzygotic isolation, like premating isolation (McKinnon et al, manuscript), is stronger between ecologically differentiated populations than between populations little differentiated ecologically. I was interested in determining if two populations of ecologically differentiated threespine stickleback that already show strong premating isolation have accumulated enough genetic differences to also show postmating isolation. I used four populations of threespine stickleback for this investigation: a marine and stream pair from Canada, and a marine and stream pair from Japan. Divergence times between Japan and Canada date back an 3 estimated 0.9-1.3 million years (Orti et al. 1994). Genetic divergence between stream and marine populations is typically much lower (Buth and Haglund 1994). While exact genetic distances are unknown for these particular populations, divergence times between other stream and marine populations in the Pacific Basin are around 11,000-22,000 yr. based on maximum habitat age (Bell and Foster 1994), with coefficients of genetic distance (Nei's D, allozymes) of 0-0.04 (Avise 1976, Buth 1984, Higuchi and Goto 1996). 1.2 M A T E R I A L S A N D M E T H O D S Two generations (Fi and F 2) of hybrids were made between four populations of Gasterosteus aculeatus. The populations included a Japanese stream population, a Japanese marine population, a Canadian stream population, and a Canadian marine population. The Canadian populations are sympatric through a small part of their range during spawning season; the Japanese populations are allopatric. Individuals used for the parental generations were collected at the onset of each population's breeding season in 1996. Both Canadian stream and marine threespine stickleback were collected from a tributary of the Salmon River, near its juncture with the Fraser River in southwestern British Columbia, Canada. The Japanese stream threespine sticklebacks were collected from the headwaters of Nakagawa Creek, a tributary of the Ibi river system on Honshu Island. The Japanese marine threespine sticklebacks were from the Kushiro River, Hokkaido Island. The experimental populations were chosen such that fish from the two geographical regions (Canada and Japan) represented evolutionarily distinct lineages based on mitochondrial DNA analyses by Orti et al. (1994). In addition, each 4 geographical region included a population from each of the two (stream and marine) environments. From these original populations, my colleague Maria Nemethy created four Fi hybrid crosses in 1996: Canadian stream x Japanese stream (CSJSi), Canadian stream x Canadian marine (CSCMi), Canadian marine x Japanese marine (CMJMi), and Japanese stream x Japanese marine (JSJMi). At the same time, she made crosses within each population (CSi, CM], JSi, and JMi). These represented the first generation of laboratory crosses. In 1997,1 crossed fish from the first generation of laboratory crosses in order to create F 2 hybrids (CSJS2, CSCM2, C M J M 2 , and JSJM2) and a second generation of pure-type crosses (CS2, C M 2 , JS 2, and JM 2). These fish represented the second generation of laboratory crosses. Adult fish were kept in 100-L aquaria containing a 15% sterile seawater solution. The fish were segregated by habitat type and geographical region into groups of twenty to thirty fish. The aquaria were lined with a layer of coarse limestone gravel and filtered by charcoal filters. Fish were overwintered at 10°C with a 12 light hours : 12 dark hours daily regime and brought into breeding condition by gradually increasing the temperature to 15° C and the light to a 16L : 8D regime. Adults were fed once daily on a mixture of brine shrimp {Anemia spp.) and chironomid larvae. Artificial crosses were performed according to the procedure in Kassen et al. (1995). Only unrelated fish were used in crosses to avoid the effects of inbreeding. Optimal breeding condition in stickleback is signified by high coloration in males and distended, swollen abdomens in females. However, the amount of abdominal distension in females varies with individual and population and is more pronounced in stream 5 populations than in marine populations. Using gentle abdominal pressure, eggs were removed from females and then placed in a petri dish with just enough distilled water to cover them. Male fish were euthanised using carbon dioxide or tricaine methanosulfornate (MS-222), and their testes were removed and macerated in the petri dish with the eggs. Only one male was used per clutch of eggs. After a 15-minute fertilisation period, the eggs were examined under magnification, and the number of successful fertilisations was recorded. Each clutch was then placed in a separate holding cup with a fine mesh screen bottom and suspended over an airstone in a 10-L aquarium to aerate the eggs during development. Each 10-L aquarium was lined with washed blasting sand and coarse limestone gravel and filtered by an undergravel filter. All tanks were maintained at 15°C and a 16L : 8D light regime. Clutches were observed the day after fertilisation and then every two days until hatching. The number of developing eggs was recorded (using Swarup 1958 as a guide). Upon fertilisation, eggs swell and develop a visible membrane surrounding the yolk. As early as the first day following fertilisation, embryos are visible as dark spots in the centre of the egg. Development proceeds quickly, and eggs hatch at nine to ten days post-fertilisation. Dead eggs, which are cloudy and discoloured, were removed to prevent fungal infection. Upon hatching, the number of emerging larvae was recorded. I used fertilisation and hatching success as measures of fitness in the hybrids. Fertilisation success, measured for both Fi and F2 generations, is largely a property of the previous generation (that is, Fi fertilisation success can be considered a measure of isolation between the parents, and F2 fertilisation success can be considered an Fi trait). I made Fi crosses to measure compatibility of parent populations' gametes, as measured by 6 fertilisation success of Fi's, and to determine hatching success of Fi hybrids. The F2 crosses were used to test fertility of Fj's, measured by fertilisation success of F2's, and to determine hatching success of F2 hybrids. Fertilisation success was calculated as proportion of total fertilised eggs per clutch. Hatching success was calculated as proportion of number of emerging larvae per eggs fertilised. The number of clutches in each cross type were as follows for 1996: CSi=7, C M ^ I O , JSi=10, JMi=10, CSJSi=18, CSCM,=19, CMJMi=19, and JSJMi=20, and for 1997: CS2=6, CM2=13, JS2=7, JM2=4, CSJS2=20, CSCM2=23, CMJM2=24, and JSJM2=14. Data were transformed using a logistic transformation, the transformation of choice when carrying out regression analysis of proportional data (Sokal and Rohlf 1981). Because several clutches had 100% fertilisation or hatching success, I modified the proportion by multiplying it by a factor of 0.99 in the denominator. The transformation was done using the formula: , , proportion hatched or fertilised . ,. log( — ) (1.1) 1 - (0.99)(proportion hatched or fertilised) The eight cross types were compared for differences in fertilisation and hatching success of Fi and F2 generations using an analysis of variance by ranks. All statistics were calculated using S-PLUS 2000 (MathSoft, Inc. 1988-1999). In 1997, much of the variation in hatching success among groups could be explained by date. This is most likely because it took time to learn and perfect experimental techniques, but it is also possible that fertility and viability increased over time due to natural causes or to uncontrolled environmental trends. This trend was also present in the 1996 fertilisation data (Figure 1.1). Because techniques and aptitude for successfully making hybrid crosses increased steadily throughout each experimental 7 season, fertilisation and hatching data were corrected by taking the residuals of the linear least squares regression of (logit-transformed) success on date (1996 fertilisation r2=0.22, n=113, F t j 11=30.98, p<0.05; 1996 hatching r2=0.01, n=113, F,, i n=0.57, p=0.45; 1997 fertilisation r2=0.002, n=l l l , F u 0 9=0.25, p=0.62; 1997 hatching r2=0.15, n=l l l , Fi, 109=19.64, p<0.05). Corrected data, then, are equal to the mean success plus the residual, then back-transformed to get percentage values. In addition, crosses made early in the experimental season in 1997 (prior to July 1, seven crosses) were omitted from the analysis due to their high failure rate. 1.3 R E S U L T S I found no significant differences between the eight pure-type and hybrid clutches for either 1996 or 1997 (Kruskal-Wallis rank test, parental fertilisation x 2 =6.03, df=7, p=0.54; F, fertilisation %2 =6.22, df=7, p=0.51; Fj hatching x =12.04, df=7, p=0.10; F 2 hatching % =10.13, df=7, p=0.18). Fertilisation success for hybrids was near 100% for many clutches in the parental and F) generations. Likewise, hatching success was quite high in both the Fi and F 2 generations (Figure 1.2). 1.4 D I S C U S S I O N In this particular experiment, I examined (1) whether postzygotic isolation exists between ancestral marine populations and their descendant stream populations, and (2) whether postzygotic isolation exists between two independently derived stream populations. My results indicate that there is no physiological reproductive isolation such as incompatibility between sperm and eggs, sterility in hybrids, or inviability of zygotes amongst any two populations studied. G. aculeatus inter-ecotype and inter-region 8 hybrids do not show postzygotic isolation in the form of hybrid sterility or inviability in either the Fi or F2 generations in a lab setting. However, I only tested some, not all, aspects of fitness in the Fi and F2 hybrids. I measured just two aspects of fitness in the Fi's (hatching success and fertility, as measured by fertilisation success) and one in the F2's (hatching success). Furthermore, fertilisation and hatching success are only part of a host of characteristics I could have used to measure fitness. The high levels of fertilisation and hatching success among hybrids suggest there is little or no intrinsic genetic post-mating reproductive isolation amongst the experimental populations. Much of the variation in the data from this experiment is due to experimental error. This error can be attributed to two sources: perfecting experimental techniques and assessing reproductive condition. The ability to perform and perfect laboratory techniques improved steadily throughout the experimental season and was corrected using a logit regression transformation. However, a second level of error existed in 1997: my difficulty in assessing how gravid was each of the Japanese marine females. The bodies of female stream threespine stickleback become quite swollen when carrying eggs (Tinbergen 1948), but this effect is much subtler in marine females. The optimal period to extract eggs from females without compromising fertilisation and development is brief and can be difficult to judge (McPhail, pers. comm.). The continued maintenance of separate marine and stream lineages in the sympatric Canadian populations, especially in light of little or no genetic breakdown in hybrids, suggests that the populations have effective pre-mating isolation or that there are strong ecological effects acting selectively on hybrids (McPhail 1994). Indeed, it is 9 common for threespine stickleback to exhibit strong assortative mating. Studies examining pre-mating isolation in threespine stickleback suggest that size is an important isolating factor (Borland 1986, Nagel 1994). Marine threespine sticklebacks are larger than stream populations. Further, mate choice experiments using individuals from these same populations indicate that stream individuals prefer to mate with one another, regardless of their continent of origin, rather than with marine individuals from either geographic region (McKinnon et al., in prep.). Finally, viability and survival in the lab environment does not reflect viability and survival in the natural environment. Hybrids, when put in an ecological setting, may experience reduced fitness when exposed to competition for resources or mates, predation, or any one of several other sources of natural or sexual selection (Hatfield and Schluter 1996, Hatfield 1997, Rundle and Schluter 1998). The 0.9-1.3 million years divergence time (Orti et al. 1996, mtDNA) between Canadian and Japanese populations is not enough time for genetic incompatibilities to have developed, at least to the point that they are apparent in the absence of the appropriate ecological context. Although the threespine stickleback is capable of rapid morphological evolution, evidence suggests that intrinsic differences causing reproductive isolation evolve much more slowly. In a review of reproductive isolation amongst G. aculeatus populations, postmating mechanisms were rarely found—the sole exception to date is between the Japan Sea marine and landlocked populations from the Pacific Ocean lineage (McPhail 1994). Hybrids from these populations, diverged at least 2 million years ago, have some sterility in the Fi generation, and F 2 individuals, if formed, fail to reach sexual maturity (Honma and Tamura 1984). 10 Other recent reports on the relative rates of evolution of reproductive isolation are consistent with the findings of the present study. Using Nei's D as a measurement of elapsed time (D=l corresponds to roughly 5 million years of divergence, Nei 1972), Coyne and Orr (1988, 1997) estimated that sympatric populations of Drosophila require approximately 200,000 years for total (a composite measure combining both prezygotic and postzygotic mechanisms) reproductive isolation to evolve. Allopatric populations require even more time—approximately 2 million years. Sasa et al. (1998) found that frogs evolve postzygotic isolation at rates similar to those of Drosophila. They found no evidence of postzygotic isolation at Nei's D less than 0.3, or about 1.5 million years of divergence. These time estimates, however, are very rough and may not be comparable, as genetic "clocks" may run at different rates for different taxa. In addition, allozymes (used in the Drosophila and frog studies) and mtDNA (used to calculate the ages of the stickleback populations) are less reliable genetic estimators of divergence times than neutral third-position substitutions (Fitzpatrick 2002). Neither allozymes nor isozymes are entirely selectively neutral and thus are not ideal for use as "clocks". Divergence times measured using mtDNA, too, may not accurately reflect the average divergence time of nuclear DNA. Migrants between incompletely isolated populations can reduce genetic differences without necessarily replacing the mitochondrial genome, or vice-versa. Indeed, phylogenetic relationships of stickleback populations in Japan and Canada created using mtDNA data differ from those based on nuclear D N A data (Orti et al. 1994, Higuchi and Goto 1996, Yamada et al. 2001). This difference suggests that even highly divergent populations like the Japan Sea and Pacific Ocean clades do not yet have complete reproductive isolation (Yamada et al. 2001). 11 0 0 o * Q 4 • w lb • A° >< > • « 4 • r> • • \ • • \ • 0 > > E> 4 • t> >• ^ - 0 oL 0 < o > o * *> 4 0 a ^t>4 0 V 4 •a • o # < « 0 0 < \ 4 \ ^0 rt o „\ 0 < <\ 4 a • • f e z i o v 9661 (6u!L|01BM)j!6on - i 1 r r 0 \ < co 3 CD 2 O Z66I- f6u!U,ojBL|Sn6o° 03 O c o CD +-» 03 T3 >» CD CD O c o 60 c o 03 60 O fi 03 CD o o CO c CO ,4> 00 _o O co o CD -*-» o3 T3 CO CO CD CD o CO _o cs cn -T3 OX) g o 03 CD c o u O o CD J3 3 O O CD & 5 -fi o CD fi CD s n CD & CD o S f e 3 i 9661. (uo!iBS||!iJ9j)i!6on Z66H (uortBS!i!Vi9j)»!^on T3 fi 03 > . O c CD 'o 53 CD CD > o I CD s -s CD CD CD CO T3 CD CO 03 CD t-O _g Cw o T3 C CD "c3 <D 0  C V-fi 60 CD c CD 60 fi CD !-c2 T3 >. ,£> CD +-» fi CD CO CD 1-OH CD l-i CD 03 c o CD Ok CO c o O i-i CD +-» C CD l-i c o o3 CD fi CD CD CD -fi O 03 CD o 12 u O pauojBU. S J B S A JSJU. jo uonjodojd pauojBU. SJBGA puooes p uo^odojd —} r--C D C O O ) CT) O ) C D x — w o » co" w C O o Hi o w C O o w w C O M w W o w C O o C O o C O ~3 C O o C O o CO o C O o pasnnJSj SJBGA JSJJJ jo uofljodajd pesniviaf sjea/f puooes p uoiyodbjd > 13 C H A P T E R T W O : P A R A L L E L I N H E R I T A N C E O F M O R P H O L O G I C A L C H A R A C T E R S I N T H R E E S P I N E S T I C K L E B A C K , GASTEROSTEUS ACULEATUS C O M P L E X 2.1 I N T R O D U C T I O N Parallel evolution is the independent evolution of the same trait in closely related, yet separate, lineages (Simpson 1944, Futuyma 1986). Evidence of parallel evolution is one of the most convincing arguments for the role of natural selection in directing evolutionary change (Schluter and Nagel 1995). The strongest indication of parallel evolution takes place when it occurs in environments where the selective agent(s) are similar and identifiable. When such selective pressures are distinguished, the implication is that natural selection is the cause of the parallel changes, as genetic drift is unlikely to result in the same transformations in multiple environments. Examples of parallel evolution as a result of similar selective pressures are numerous and include adaptations for living in freshwater displayed by various fish (Foote et al. 1989, Pigeon et al. 1997), limb length and body size of Anolis lizards (Losos 1992), feeding polymorphism in E. coli (Treves et al. 1998), and eye reduction and albinism amongst cave organisms (Jones et al. 1992, Culver et al. 1994). However, few studies have tried to understand the genetic basis of parallel phenotypic evolution (see Schat et al. 1996 and Wichman et al. 1999 for two exceptions). As a result, we know very little about whether the parallel phenotypic changes in independent lineages are determined by parallel changes at the genetic level. It is unknown whether the same genes are involved in parallel evolution of characters under 14 simple genetic control, and the genetic basis of parallel evolution of characteristics under more complex genetic control is virtually unexplored. One way to approach the genetic basis of parallel evolution is to test whether changes in the same traits occurring in independent lineages exhibit parallel inheritance. By parallel inheritance, I mean that the phenotypic characteristics within each lineage exhibit similar patterns of inheritance when the divergent populations are crossed: similar additive, dominance, and epistasis components in traditional analysis of line means (Lynch and Walsh 1997), and similar effective number of loci (Lande 1981, Lynch and Walsh 1997). Parallel inheritance can exist in two forms. The strong form of parallel inheritance results from allele substitutions at the same loci in different lineages, whereas the weak form of parallel inheritance results merely from changes at the same or different loci that behave similarly. A sure-fire way to distinguish these alternatives is to map the genes responsible for phenotypic changes in each lineage (see Cunningham et al. 1997, Wichman et al. 1999). Nevertheless, using an analysis of line means to examine patterns of inheritance is a useful first step. Additionally, as I show below, crosses between lineages can sometimes distinguish the alternatives. Here I present a study of parallel evolution and parallel inheritance in the threespine stickleback. The threespine stickleback (Gasterosteus aculeatus) is an outstanding system in which to investigate these questions because the species complex is both phenotypically and ecologically diverse and inhabits a diversity of environments (Bell 1984). Moreover, parallel evolution in the threespine stickleback is common. Sticklebacks are Holoarctic in distribution and occur in a variety of habitats, including marine, anadromous, and freshwater (both stream- and lake-dwelling) populations. 15 Stickleback have extensive body armouring, and attributes of both armour and shape vary widely among populations, with most of this variation resulting from ecological pressures exerted in different environments (Bell and Orti 1994, Bourgeois et al. 1994, Walker 1997). I focus on the parallel evolution of marine and stream stickleback populations. Marine threespine stickleback have three free.dorsal spines, a pelvic girdle, pelvic spines, and a full series of lateral bony plates extending from the pectoral girdle to the tail. The marine form of G. aculeatus is accepted to be the ancestral state, with multiple derived stream forms having arisen independently from it (McPhail and Lindsey 1970, Bell 1974, Bell 1976, Bell and Foster 1994, Walker and Bell 2000). Stream populations show an overwhelming tendency toward reduction of all elements of body armour. Dorsal and pelvic spines in freshwater populations are usually shorter than those of marine fish and, in some cases, are absent. Many populations of stream fish have few lateral plates, whereas nearly all marine populations have a series of lateral plates extending from the pectoral girdle to the tip of the tail. Armour reduction has happened many times following colonisation of freshwater by the marine forms (Heuts 1947, Okada 1960, Hagen and Moodie 1982, Ziagunov 1983, Ziagunov et al. 1987). Other phenotypic differences between marine and stream fish include overall body shape. Marine stickleback are more slender and fusiform than their stream-resident counterparts, a body shape associated with efficient pelagic cruising (Baumgartner et al. 1988). In contrast, stream fish are generally deep-bodied and robustly shaped (O'Reilly et al. 1993, Walker and Bell 2000). 16 Here I analyse armour plate number reduction and shape changes in two pairs of adjacent anadromous marine and stream-resident populations widely separated in space. Each stream population is recently descended from nearby marine stickleback. I address two issues relevant to the understanding of parallel evolution. First, I quantify the degree of parallel evolution between the two pairs of marine and stream-resident populations using two aspects of phenotype: plate number and body shape. Second, I test for parallel inheritance of armour plate number and body shape between the two pairs of stream-resident and adjacent marine populations using analysis of line means. The traits, lateral plate number and body shape, were chosen because they are ecologically significant, and thereby subject to natural selection, and they differ considerably between marine and stream populations. They also likely represent the extremes of complexity of underlying genetic control. Plate number is probably under simple genetic control with few genes responsible for the difference between populations in the number of plates (Hatfield 1 9 9 7 , Peichel et al. 2 0 0 1 ) . Body shape is a more complex trait involving many aspects of phenotype and is therefore probably controlled by many genes. 2 . 2 M A T E R I A L S A N D M E T H O D S 2.2.1 Fish populations I examined plate number and body shape differences amongst four threespine stickleback populations: a Canadian stream population, a Canadian marine population, a Japanese stream population, and a Japanese marine population. The stream populations reside in the streams year-round, whereas the marine populations are anadromous, living 17 in the sea but returning to freshwater to breed. Both Canadian stream and marine stickleback were collected from different areas in a tributary of the Salmon River, near its juncture with the Fraser River in British Columbia, Canada using minnow traps. The Japanese stream stickleback were collected from the headwaters of Nakagawa Creek, a tributary of the Ibi river system, Honshu Island. The Japanese marine stickleback were from the Kushiro River, Hokkaido Island. Both populations of Japanese fish are from the "Pacific Ocean" lineage of threespine stickleback in Japan, not from the more divergent "Sea of Japan" lineage (McKinnon et al., manuscript). Fish from the wild were collected by Jeffery McKinnon and Seichi Mori at the onset of each population's breeding season in 1996. All fish used in my analyses were lab-reared first and second-generation progeny of these wild-caught threespine stickleback. Multiple phytogenies of stickleback confirm that stream populations in different parts of the world are independently derived from marine ancestors (Bell 1976, Bell 1982, Gach and Reimchen 1989, O'Reilly et al. 1993, Orti et al. 1994, Thompson et al. 1997), implying that armour reduction and parallel transitions in body shape have evolved repeatedly in many different isolated populations. This is the case in lineages used here: Canadian and Japanese populations occur in separate mitochondrial DNA lineages (Orti etal. 1994). From these original samples, my colleague Maria Nemethy created three Fi hybrid crosses: Canadian stream x Japanese stream (CSJSO, Canadian stream x Canadian marine (CSCMi), and Japanese stream x Japanese marine (JSJMi). At the same time, crosses within each population were made (CSi, C M i , JS], and JMi). These represented the first generation of laboratory crosses. The next year, I crossed fish from the first 18 generation of laboratory crosses to create F2 hybrids (CSJS2, C S C M 2 , and JSJM2) and a second generation of pure-type crosses (CS2, CM2, JS2, and JS2). The numbers of replicate families made in the first generation for each cross were CSi=7, CMi=10, JSi=10, JMi=10, CSJSi=18, CSCMi=19, and JSJMi=20. Numbers of families in the second generation were CS2=7, CM2=13, JS2=7, JM2=6, CSJS2=20, CSCM2=23, and JSJM2=15. Adult pairs used for making each of the clutches were chosen as each adult reached optimal breeding condition as indicated by high coloration for males and greatly swollen abdomens for females (see Chapter 1 for more details). Each adult was used only once, and only one male was used per clutch of eggs. For this study, ten individuals from each laboratory generation and cross type (CS,, CM], JS,, JMi, C S C M , , CSJS,, JSJM,, CS 2 , C M 2 , JS 2, J M 2 , C S C M 2 , CSJS 2, and JSJM2) were chosen for morphometric analyses (140 fish total). To preserve statistical independence, an attempt was made to sample randomly only one individual from each family. In a few cases, more than one fish were taken per family to maintain sample size; in this event, the additional fish was taken at random from a family already represented in the sample. Crosses that have more than one representative per family are CS, (three families sampled twice), CS2 (three families), JS2 (three families), JM2 (four families). In no case were more than two fish sampled from any family. First and second generation fish were raised according to the procedure in Kassen et al. (1995). Using gentle abdominal pressure, eggs were removed from females and then placed in a petri dish with just enough distilled water to cover them. Male fish were euthanised using carbon dioxide or tricaine methanosulfornate (MS-222), and their testes were removed and macerated in the petri dish with the eggs. After a 15-minute 19 fertilisation period, the eggs were placed in a separate 250-mL holding cup with a fine mesh screen bottom and suspended over an airstone in a 10-L aquarium in order to aerate the eggs during development. Each aquarium was lined with washed blasting sand and coarse limestone gravel and filtered by an undergravel filter. All tanks were maintained at 15° C and a 16 light hours : 8 dark hours daily regime. Following hatching, families of fish were raised separately in 10-L laboratory aquaria at approximately equal densities. At six months, the Fi fish were transferred to 50-L aquaria and overwintered at 10° C and a 10 L : 14 D light regime. At age one year, they were slowly brought into reproductive condition by increasing temperature to 15° C and light to 16 L : 8 D over four weeks in order to create a second generation of hybrids. The fish were sacrificed after approximately 18 months for first generation fish and approximately six months for second generation fish using a lethal dose of MS-222. All fish were preserved in 95% ethanol. Specimens used for morphometric analysis were then stained in a 10% potassium hydroxide-alizarin red solution and returned to 95% ethanol. 2.2.2 Measuring phenotypes 2.2.2.1 Lateral plate number The number of armour plates on the left side of each fish was recorded. Counts included all staining plates from the pectoral girdle to the end of the tail, regardless of plate size. For analyses, I used a log transformation of the number of plates. 2.2.2.2 Photography I took a digital photograph of each fish and imported it directly into Scion Image (Scion Corporation 1998). For taking photographs, specimens were set on a paraffin wax 20 mount with a ruler placed along one edge as a reference for size and distortion. Focus, contrast, and brightness were adjusted by eye. 2.2.2.3 Landmark selection For the shape analysis, I chose twelve landmarks to describe the perimeter of each fish lying on its left side (Figure 2.1). Landmarks were: 1) anterior tip of upper lip; 2) anterior junction of the first dorsal spine with the dorsal midline (DML); 3) anterior junction of second dorsal spine with D M L ; 4) base of the first dorsal fin ray at the DML; 5) insertion of the dorsal fin membrane on the D M L ; 6) origin of caudal fin membrane on the D M L ; 7) caudal border of hypural plate at lateral midline; 8) origin of caudal fin membrane on the ventral midline (VML); 9) insertion of anal fin membrane on V M L ; 10) base of first anal fin ray on V M L ; 11) anterior border of ectocoracoid on V M L ; and 12) posterior edge of angular. Landmarks were readily and reliably identifiable on all fish. Each landmark represented the convergence of two or more tissues, the most informative type of landmark (Bookstein 1990). These include many of the landmarks used in a previous morphometric analysis of stickleback by Walker (1997). Walker's study included three additional landmarks: one indicating the supraoccipital notch, one pertaining to placement of the pelvic girdle, and one indicating the posterior edge of the ectocoracoid. The (x,y) co-ordinates of the twelve landmark positions for each fish were calculated using Scion Image (Scion Corporation 1998) and saved to a spreadsheet. 2.2.2.4 Thin-plate spline (TPS) geomorphometric analysis I used relative warps analysis (TPS Relw, Version 1.18, Rohlf 1998), a thin-plate spline geomorphometric analysis, to analyse each specimen for body shape differences while controlling for geometric body size. Relative warps analysis scales a series of 21 landmarks to a centroid size of one and computes an average shape called the consensus figure for all the samples. A centroid size is analogous to the geometric area defined by all landmarks. It is equal to the square root of the sum of squared distances from each landmark to the centroid. Imagining an individual's landmarks as dots placed on a flat plane, the relative warps analysis describes the amount of stretching, compressing, or bending the plane would have to undergo to superimpose the dots on the consensus figure. These deformations of the plane are represented as relative warps, or major axes of change. The first two relative warps are in uniform directions (stretching or compressing) and the remaining components describe non-linear changes (bending or twisting). 2.2.2.5 Accounting for age and size differences in parental, Fj, and F2 generations Specimens from each generation of fish were sacrificed at the same time; as a result, the F 7 individuals were younger by one year and smaller than the Fi individuals. Similarly, the second-generation pure-type individuals were younger and smaller than those of the first generation were. I accounted for differences in size and ontogenetic development using the two generations of pure-type crosses (CSi, C M , , JSi, JMi, CS2, C M 2 , JS 2, and JM2) as controls. Following digitisation and relative warps shape calculations, I examined the averages of the 20 principal warps of individuals of each pure-type genotype in both generations separately. I calculated the intergenerational difference in the means of each warp separately between generations. Significant differences were found between generations such that the second-generation fish were, on the whole, shifted uniformly from the first-generation fish. These differences may be the result of ontogenetic shape 22 changes accompanying increased size as fish mature or of developmental differences related to different laboratory conditions between the two years. To minimise the effect of the between-generation differences on tests of parallel evolution, I added the difference in means to the scores of all the ¥2 generation fish before further analyses. 2.2.2.6 Discriminant function a?ialysis I carried out a discriminant function analysis to identify shape differences amongst the four original populations (CS, C M , JS, and JM). This discriminant function analysis used the combined data from the relative warp scores of the first generation of fish and the adjusted relative warp scores of the second generation. The resulting discriminant function scores, calculated using the pure-type populations, were then estimated for hybrids as well. The analysis was carried out using a linear discriminant analysis in S-PLUS 2000 (MathSoft 1988-1999), which discriminates between groups by first scaling the within-group variation so that each group is equal, and then describing the between-group variation as axes of variation, or discriminant functions. Percent of variation among cross means accounted for by each axis of discrimination was calculated by computing the mean of each cross type along every discriminant axis. I then calculated the variance among group means for each axis and summed this variance across all axes to get total variance. Finally, to get the percent variance accounted for by each axis, the variance of each axis was divided by the total variance. 2.2.2.7 Relating landmarks to shape changes described by discriminant functions In order to decipher the contributions of actual physical changes to the resulting discriminant functions, I correlated discriminant functions one and two with the x,y co-ordinates of the twelve scaled landmarks. I did this by calculating the slope of the 23 univariate regression of the x-and the y-position of each landmark on discriminant function one, then repeated the analysis for discriminant function two. These slopes explain the unit of change in each x and each y position that occurs with movement in increments of standard deviations away from the average along the discriminant function. I then plotted change in x by change in y for each landmark, and the resultant vectors described the net change each landmark underwent per two standard deviations for each discriminant function. Landmarks with vectors of large magnitude were deduced to be contributing to the axis of discriminant function, while landmarks without vectors, or with only very tiny ones, did not contribute to the description of the particular discriminant function. To see the physical changes represented as vectors of change between populations, I calculated the average shape (as described by the polygon formed by the x,y co-ordinates of the twelve landmarks) of each pure-type population (CS, C M , JS, and JM). The average shape polygons for each population were plotted using an S-PLUS graphing function, and the difference between each landmark from the CS, JS, and JM shapes to the corresponding landmark on the C M shape was determined. The twelve resulting vectors for each population show the direction and magnitude of change each landmark undergoes when one shape is transformed into another. 2.2.3 Quantifying parallel evolution Tests of parallel evolution were carried out using a two-factor A N O V A and comparing the relative amounts of variance due to habitat, lineage (continent of origin), and the interaction between habitat and lineage. Populations were considered to show parallel evolution of a trait if the mean square (MS) variation due to habitat (i.e., between 24 stream-resident and marine fish) was greater than the MS variation due to an interaction between the habitat and the lineage (either Japanese or Canadian). If one stream-marine pair has a net change in a trait mean between the stream and marine habitats but the other pair has no change in trait mean, the variation due to habitat will equal the variation due to lineage. That is, even though the average trait mean between habitats appears different, examining whether variance is due to habitat or due to an interaction of habitat and lineage will determine whether the observed difference is the result of changes in both lineages, or only one of them. As shown in Figure 2.2, parallel evolution involving two lineages in two habitats only occurs when MShabitat>MSinteraction- The ratio Y=MShabitat/(MShabitat+MSinteraction) provides a measure of how parallel the trajectories are. The ratio ranges from zero to one, where a value of one represents perfectly parallel evolution (Figure 2.2A), and a value of zero represents opposing trajectories (Figure 2.2B). At least some parallel evolution has occurred when Y>0.5 (Figure 2.2C). I used S-Plus 2000 (MathSoft) for analyses of variance. 2.2.4 Quantifying parallel inheritance 2.2.4.1 Analysis of line means Using joint scaling, a regression technique developed by Cavalli (1952) and Hayman (1958, 1960a, 1960b), I compared the contributions of additive and dominant genetic effects on trait inheritance to the morphological divergence between adjacent stream and marine stickleback populations. For the analyses, I used data from Fi and F 2 hybrids CSCMi, JSJMi, C S C M 2 , and JSJM 2 along with data from the four pure-type populations. I tested inheritance patterns of both armour plate number and body shape, as represented by the first discriminant function, which controls for differences in body size. 25 With additive effects alone, the mean phenotype of the Fi and F 2 generations is expected to be the average of the means of the two parental phenotypes. Dominance effects cause both generations of hybrids to resemble one parental phenotype over the other, a tendency more pronounced in the Fi than in the F 2 generation. Both additive and dominance effects may be detected simultaneously within a single cross. When epistatic effects between two or more genes are present, hybrids' phenotypes differ unpredictably from the parental phenotypes, following expectations from neither additive nor dominance models. However, these effects may be present but too weak to detect. The joint scaling technique I used is described in Mather and Jinks (1982) and Lynch and Walsh (1997). Originally developed to study genetic divergence between inbred lines, the technique can also be used to analyse interfertile wild populations (Hard et al. 1992, 1993, Hatfield 1997). The technique fits the following multiple regression model to the observed line means of parental and hybrid crosses, zt = j U + M,. 2a + M,-35+e,. (2.1) or, in matrix form, z = Ma + e. (2.2) Here zi is the trait mean in the ith line; M is a matrix of coefficients that describes the effects of a constant, additivity, and additivity-plus-dominance; is the mean of all lines; a is the additive genetic effect; 8 is the effect of dominance; and e,is the deviation of the observed mean from the prediction of the model. Here, epistasis terms (i.e., additive x additive, dominance x dominance) are incorporated into the error term ex. In matrix form (eq. [2.2]), z is a vector of the trait means in the parental and hybrid lines, a is a vector containing the three model parameters, jo,, a, and 8, and M is a matrix with the number of 26 rows equal to the number of lines being analysed. The error term e is a vector and incorporates effects of epistasis. For example, in an analysis of line means for a stream x marine cross, M is a 4 x 3 matrix whose first column contains only l's, the expectation if the trait is the same in all parental, Fi hybrid, and F 2 hybrid populations. The second column of M (the element M,2 in eq. [2.1]) contains the coefficients of line means for the additive effects. The stream parent is assigned a coefficient of 1, the marine parent is -1, and the Fi and F 2 hybrids have a coefficient of zero (under additivity the expectation for first- and second-generation hybrid phenotypes is halfway between the parental means). The third column in M (the element M,3 in eq. [2.1]) contains coefficients of line means for dominance effects. The coefficients here are 1 for both the parental forms, -1 for the Fi hybrid, and zero for the F 2 hybrid (under dominance, the Fi's are expected to shift two units from the additive expectation and the F 2's, half as much). To test for parallel inheritance using line means analysis, I modified the above procedure to analyse both Canadian and Japanese stream x marine crosses simultaneously. I fit the modified model, zt = H + M,- 2A + Mna + MiAd + et (2.3) M is now an 8 x 4 matrix whose first column contains only l's, as before, but has a second column denoting geographic origin (-1 for Canadian crosses and 1 for Japanese crosses. The corresponding "continent" parameter is A, in eq. [2.3]. The third and fourth columns of M contain the additive and dominance coefficients as before. If the data are adequately fit by the model in eq. [2.3], without the need to incorporate higher order 27 terms accounting for interactions between continent and genetic effects, then I conclude that marine and stream forms from Canada and Japan show parallel inheritance. The regression model is fit using the standard expression for generalised least squares (Lynch and Walsh 1998), a = ( M r V - ' M ) " 1 M r V" l z (2.4) and £ = Ma (2.5) where a is a vector containing estimates of the four model parameters: fx, A, a, and 8; M is a matrix of coefficients from eq. [2.3], £ i s the vector of predicted line means; and V is a diagonal weighting matrix with V„ equal to the squared standard errors of observed line means (sampling variances). Superscripts Tand -1 indicate transpose and inverse functions, respectively. Data from the stream x marine crosses is therefore contained in z and V (Lynch and Walsh 1997). I tested the goodness of fit of the data to the model by comparing the observed and predicted line means using the following test statistic (Mather and Jinks 1982, Lynch and Walsh 1997): * 2 = E ^ S ^ r ( 2 - 6 ) £i var(z,.) The test statistic, %2, has a yr2 distribution with degrees of freedom equal to the number of lines minus the number of estimated parameters. In the test for parallel inheritance, eight lines are used in the analysis of stream x marine hybrids from Canada and Japan. The number of parameters is equal to one for the model of no difference (z, =/i + e,)> two for the continent model (zi =fx + M,-2X + ei), three for the continent-plus-additive model (zi =jJL 28 + MQX + Ma a + ei), and four for the continent-plus-additive-plus-dominance model (z, =/j. + Mi2k + M,-3oc + M/48 + ei). I tested the models using Lynch and Walsh's (1997) sequential fitting method. First, I tested a model of no difference using only the first column of M , corresponding to the constant, fi. Rejection of this model using the goodness-of-fit test indicates divergence among means for the four parental populations and their hybrids. If the "no difference" model was rejected, I tested the continent model by repeating the above regressions after adding the continent column to M (eq.[2.3]). If this model was rejected, I tested the continent-plus-additive model after adding the additive effects column to M . If this model was not rejected, I conclude that the character in question has diverged between stream and marine forms at loci whose net effects are largely additive and behave similarly in both regions, although there may differences in means between continents. If the continent-plus-additive model was rejected, I tested a continent-plus-additive-plus-dominance model by adding the dominance effects column to M (eq. [2.3]). Failure to reject this model indicates that the character under examination has diverged at loci whose net effects are only dominant or additive-plus-dominant and that behave similarly in both regions although, again, there can be mean differences between the continents. Finally, rejection of the additive-plus-dominance effects model indicates that epistasis affects inheritance of that character, or that inheritance is not parallel in the two regions. The line means analysis methods used have some limitations in determining the extent of parallel inheritance. The first concern is that without accompanying molecular data, this method can only determine the weak form of parallel inheritance. Specifically, 29 this line means analysis provides no definitive way to determine whether the parallel changes are taking place at the same loci, and it is quite possible that the parallel changes could result from diversifying changes at different loci in each population. However, as I show below, in some instances it is possible to infer the strong form of parallel inheritance by examining several different hybrid lines:-2.2.4.2 Analysis of line variances Parallel inheritance requires that two conditions be met: similar patterns of inheritance in traditional analysis of line means and a similar effective number of loci. In order to indicate whether the effective number of loci separating the forms in Canada and Japan is similar, I conducted an "analysis of line variances" to compare differences in the variance in lateral plate number and body shape for the parental, Fj hybrid, and F 2 hybrid generations. Akin to the analysis of line means, the analysis of line variances used the mean absolute deviation of observations from the median of the trait. An analysis of line variances can determine whether the observed changes in a trait are a result of different alleles present at the same locus or whether they are the result of changes occurring at different loci by looking for evidence of segregation variance in the F 2 hybrids. Evidence of segregation variance is the same criteria used in other techniques for estimating effective number of loci, such as the Castle-Wright estimator (Lynch and Walsh 1998). Using lateral plate number as an example, marine populations that are presumably homozygous for the completely plated phenotype are crossed with stream populations presumably homozygous for the low-plated phenotype. Each of these populations has some associated variance around the mean for their respective numbers of plates. If differences in plate number between stream and marine populations are the product of 30 changes at a small number of loci, segregation in the F 2 generation will produce a larger variance associated with their mean than either the parental or the Fi generations (Tamarin 1993). This expansion of variance, called segregational variance, is an indication of a small number of genes underlying population differences. In practice, Fi hybrids often show increased variance relative to the parental forms due to instability in the genome from incompatibilities between parental alleles (Graham and Felley 1985, Leary et al. 1985a, 1985b, Burkhead and Williams 1991). Therefore, I tested for F 2 segregation variance by comparing it to the variance of Fi and parental forms. Finding a similar increase in variance of F 2 hybrids between stream and marine populations within both continents is further evidence of parallel inheritance. To compare variances, I fit the following multiple regression model to the means of the absolute deviations from the median (Filliben and Heckett 2002) of parental and hybrid crosses, yt = V + N,-2A + Ni3K + Ni4(p + et (2.7) or, in matrix form, y = N v + e. (2.8) yi is the mean absolute deviation of individuals from the median in the ith line. N is a matrix of coefficients that describes the effects of different processes, including segregation, on line variance. The parameter fx is the mean absolute deviation from the median for all lines, A is the effect of continent of origin, K"is the magnitude of variance seen in hybrids, often attributed to "decanalization". The parameter cp is the effect of segregation; and e,is the difference between the observed mean absolute deviation and the prediction of the model. In matrix form (eq. [2.8]), y is a vector of the mean absolute 31 deviations in the parental and hybrid lines from both Canada and Japan and v is a vector containing the four model parameters, fx, A, K and (p. N is an 8 x 4 matrix whose first column contains only l's, the expectation if the variance is the same in all parental, Fj hybrid, and F 2 hybrid populations. The second column in N (the element N/2 in eq. [2.7]) denotes geographic origin, -1 for Canada and 1 for Japan. The third column of N (the element N/3 in eq. [2.7]) accommodates the possibility that the variance of hybrids exceeds that of the parents. Parents were assigned the value of -1, while the Fi hybrids were assigned the value of 1. In the analysis presented here, I also assigned the F2 hybrids a value of 1, which assumes that the Fi and F2 hybrids have the same variance. While this is not expected under many models of inheritance, equal variances in Fi and F 2 hybrids is a useful null model when fitting the last column in N, where the variances of Fj and F2 hybrids are free to vary. The fourth column in N (the element N/4 in eq. [2.7]) contains coefficients of line variances for F2 segregation variance. Parents and Fi hybrids are assigned a value of -1, and F 2 hybrids are 1. This fit allows the variance of the F2 to differ from that of the Fi hybrids and the parents. If the data are adequately fit by eq. [2.7], without the need for higher order terms involving interaction between continent and the genetic effects, then I conclude that the data fail to provide evidence that the divergence between marine and stream forms is not due to parallel inheritance. As in the analysis of line means, the regression model was fit using the standard expression for generalised least squares, $ = ( N r W ~ 1 N ) ~ , N r W ~ 1 y (2.9) and 32 y = Nv (2.10) where v is a vector containing estimates of the four model parameters: /a, X, K, and cp; N is the matrix of coefficients from eq. [2.6]-[2.7]; y is the vector of observed line mean absolute deviations; y is the vector of predicted line mean absolute deviations from the median; and W is a diagonal weighting matrix whose diagonal elements are the squared standard errors of observed line mean absolute deviations. I tested the goodness of fit of the data to the model by comparing the observed and predicted line mean absolute deviations using the following test statistic (Mather and Jinks 1982, Lynch and Walsh 1997): x 2 = f (yi-h)2 ( 2 1 1 ) varCj,.) The test statistic, %2, has an approximate %2 distribution with degrees of freedom equal to the number of lines (eight) minus the number of estimated parameters. The number of parameters is equal to one for the model of equal variance (y,=(j, + et), two for the model including the continent term, three for the model of increased variance of Fj's, and four for the model incorporating segregation variance of the F 2 hybrids. I tested the models using Lynch and Walsh's (1997) sequential fitting method. First, I tested a model of equal variances using only the first column of N (eq. [2.7]-[2.9]). Rejection of this model using the goodness-of-fit test suggests unequal variance in traits among parental and hybrid populations (assuming that all the parental lines have equal variances). If the equal variance model was rejected, I fit the continent parameter next. If the model with the continent term was rejected, then I tested the model in which the Fi variance is greater than that of the parents by repeating the above regressions after adding column 33 N/3 to N (eq. [2.7]-[2.9]). If this model was not rejected, I conclude that the variance among populations is due to increased variance in all hybrids, possibly from the breakdown of canalization (Graham and Felley 1985, Leary et al. 1985a, 1985b, Burkhead and Williams 1991). If the model of increased variance in the Fi's was rejected, I tested a segregation variance model by adding the segregation variance column to N (eq. [2.7]-[2.9]). Failure to reject this model indicates that the differences in variance among lines include a component causing increased variance in the F 2 generation. This component is most likely segregation variance, but it could also be due to certain types of epistasis (e.g., if decanalization is more extreme when individuals have a mix of homozygous loci from each parent). 2.2.5 Testing weak vs. strong parallel inheritance 2.2.5.1 Analysis of line means I used a line means analysis of stream x stream crosses to infer whether parallel inheritance is weak or strong. When few genes are involved in the inheritance of a trait, as with lateral plates, hybrids between the parental types can be used in a complementation-type test to determine if the low-plated morph in stream fish is the consequence of changes at the same or different loci. If the loss of plates is due to recessive mutations at different loci in the parental genotypes, then the hybrid is expected to recover the plated phenotype as a result of recombination. That is, in a CSJSi hybrid, the loss of plates at one locus in the parent from the Canadian stream population will be compensated for by the presence of another allele at that locus in the parent from the Japanese stream population, and vice-versa. If no complementation occurs and the 34 hybrids remain scantily plated, then it can be inferred that the loss of plates is due to changes at the same loci. The line means analysis used to compare Canadian stream x Japanese stream (CSJS) hybrids was similar to the stream x marine hybrid analysis explained earlier, but this analysis did not include an additional "continent" column. In this case, M is a 4 x 3 matrix whose first column contains only l's, the expectation if the trait is the same in both parental, Fi hybrid, and F 2 hybrid populations. The second column of M (the element M , 2 in eq. [2.1]) contains the coefficients of line means for the additive effects: CS=1, JS=-1, and CSJS i and CSJS 2 both equal zero. The third column in M (the element M , 3 in eq. [2.1]) contains coefficients of line means for dominance effects: CS=1, JS=1, CSJS,=-1 and CSJS2=0. 2.2.5.2 Analysis of line variances The second test I used to compare weak vs. strong parallel inheritance used line variances analysis on hybrids between the stream populations from Canada and Japan. The model I used was like the line variance model used to compare stream x marine crosses but without the second "continent" column. It fits the following regression model to the means of the absolute deviations from the median of parental and hybrid crosses: yt =n + l\i2K + Ni3(p+ei (2.12) For the stream x stream crosses N is a 4 x 3 matrix whose first column contains only l's. The second column of N (the element N, 2 in eq. [2.8]) contains the coefficients of line variances for the effects of increased variance in the hybrids: CS=-1, JS=-1, and CSJSi and CSJS 2 both equal one. The third column in N (the element N, 3 in eq. [2.8]) contains 35 coefficients of line variances for segregation variance effects: CS=-1, JS=-1, CSJS]=-1 and CSJS2=1. In the line variance analysis of stream x stream hybrids, the strong form of parallel inheritance can be rejected by finding evidence of segregational variance. That is, if the changes that resulted in the phenotypic similarities between stream environments are caused by mutations at the same loci, then the parental, Fi , and F 2 generations should all have comparable variances if they are the same at those loci. However, this prediction is only supported if the stream x marine F 2 hybrids show segregation variance, indicating few genes responsible for the trait. 2.3 R E S U L T S 2.3.1 Tests of parallel evolution 2.3.1.1 Lateral plate number Anadromous marine sticklebacks from both Canada and Japan had similar numbers of plates (Table 2.1). Likewise, stream resident populations from the two regions resembled each other closely. The analysis of variance showed the two stream populations to be evolving in parallel, y= 0.9995. Two-way analysis of variance revealed a strong habitat effect, (F=3541.9, df=l, p<0.01), but only a weak continent effect, F=2.26, df=l, p=0.14), and a negligible interaction (F=1.72, df=l, p=0.19). 2.3.1.2 Body shape Total shape differences between the stream and marine populations are described in Figure 2.3. The major difference is body depth, with stream forms being much deeper-36 bodied than marine forms. Other differences are shorter head regions, dorsal fins, and anal fins in the stream forms relative to marine forms. There was significant variation amongst the four pure-type crosses in all partial warp scores (MANOVA, Pillai-Bartlett trace=1.8348, F 6 0 , i56=4.9039, P<0.0001). The discriminant function analysis of the partial warp scores separated the four stickleback populations into distinct groups based on body shape differences. The primary axis of discrimination among the four groups separated marine populations at one extreme from stream populations at the other extreme (Figure 2.4, Table 2.1). The first discriminant function accounted for 64.6 percent of the total variation among the four means. For body shape as represented by the first discriminant function, the two stream populations are evolving in parallel, y=0.998. A two-way analysis of variance found strong habitat (F=549.76, df=l, p<0.01) and continent (F=21.69, df=l, p<0.01) effects, but only a weak interaction (F=1.14, df=l, p=0.29). Because of the non-linear and multivariate nature of the partial warps, understanding exactly which attributes of body shape contribute to each axis is difficult. However, by correlating changes in the x and y position of each individual landmark with the discriminant functions, an understanding of the overall differences in body shape is reached. Major parallel changes from a marine stickleback shape to a stream shape such as those described by the first discriminant axis are an overall increase in body depth. Other changes in common between stream forms include an anterior shift of the dorsal fin, a shorter anal fin, and a longer "jawline" relative to body size (Figure 2.5A). The second discriminant function represented additional, non-parallel changes in body shape, that account for 16.6% of the total variation among the four means. The two 37 stream shapes define the extremes of this axis, indicating that stream populations have diverged, rather than evolved in parallel, in some aspects of phenotype. Marine shapes are intermediate with each of the Canadian and Japanese populations falling most closely with the stream population from the opposite continent. The non-parallel differences described by discriminant function two are longer dorsal and anal fins in the Japanese stream stickleback relative to the Canadian stream stickleback (Figure 2.5B). The ratio y= 0.002, revealing that the interaction between continent of origin and habitat accounts for almost all of the variance along this second discriminant function. A two-way anova revealed weak habitat effects, (F=0.36, df=l, p=0.55), strong continent effects, (F=36.10, df=l p<0.01), and a very strong interaction, (F=171.56, df=l, p<0.01). 2.3.2 Tests of parallel inheritance 2.3.2.1 Line means: lateral plate number Reduction of plate number has occurred in stream populations in both Canada and Japan. If the loss of plates is due to the same genes or to genes that behave similarly in crosses, then inheritance patterns in hybrids between stream and marine populations on both continents should match. The line means analysis showed similar patterns of inheritance of plate number on the two continents. Hybrids between stream and marine environments were intermediate in plate number (Figure 2.6). On both continents, however, F] hybrids resembled the more heavily plated marine phenotype than the stream phenotype, and the F 2 hybrid exhibited a phenotypic average closer to the average of the two parental types. The goodness-of-fit ~£ test statistics from the joint-scaling tests on line means from all hybrid types are summarised in Table 2.2. Analysing both regions together, the models of no difference between means and of additivity were sequentially 38 rejected. However, the model with the dominance term added could not be rejected. Divergence in plate number, therefore, followed an additive-plus-dominance mode of inheritance in both regions (Figure 2.6, Table 2.2). Because the model with additive and dominance components provided an adequate fit to the stream x marine hybrid crosses, in the absence of any interaction term involving regions, the divergence of lateral plate number between marine and stream phenotypes fits a parallel inheritance model between the two regions. 2.3.2.2 Line means: body shape Differences in body shape, even though it is a complex measure, were similar between stream and marine populations of both regions. Patterns of inheritance, too, were similar for Canadian and Japanese populations. Stream x marine hybrids from both regions were intermediate between the pure stream and pure marine parents' body shapes. The goodness-of-fit x 2 test statistics from the joint-scaling tests on line means from all hybrid types, for body shape defined by the first discriminant function, are summarised in Table 2.2. The model of no difference was soundly rejected. However, the additive effects model could not be rejected, indicating that the differences between stream and marine populations in body shape are inherited in a net additive fashion (Figure 2.7, Table 2.2). The model was satisfactorily fit to the stream x marine hybrid lines without adding interaction terms involving continent. Therefore, divergence of the two stream populations from marine ancestors in body shape shows parallel inheritance. 39 2.3.2.3 Line variances: lateral plate number In stream x marine crosses the analysis of line variance shows segregation variance expansion in the F 2 hybrids (Table 2.3). F 2 hybrids had higher levels of variance in number of lateral plates than either the parental or Fi generations did. Segregational variance is an indication that genetic changes responsible for differences between stream and marine populations are at few loci (Tamarin 1993, Schat et al. 1996). Furthermore, the line variances are parallel between continents. The segregation variance model fit the lines without adding any interaction terms involving continent (Table 2.3). 2.3.2.4 Line variances: body shape Stream x marine Fi and F 2 hybrids show an increase in variance of body shape compared with parent populations (Table 2.3). However, there was no evidence for excessively high variance in the F 2 hybrids relative to that of the Fi's, suggesting no segregation variance. This implies that hybrids have some destabilisation in phenotypic expression, perhaps due to decanalization, but gives no evidence that few genes are involved in inheritance of body shape. Interestingly, this destabilised pattern was parallel between both the continents. The model successfully fit the lines without adding a continent interaction term (Table 2.3). 2.3.3 Testing weak vs. strong parallel inheritance 2.3.3.1 Lateral plate number If reduction in plate number is due to changes at the same loci in stream populations, then the number of plates in parents and hybrids between the stream populations should be the same. This similarity is a result of a lack of complementation-40 like interactions between the genomes of the parents. Plate number averages for F] and F 2 hybrids between the stream-resident population in Canada and the stream population in Japan are similar to those of parental Canadian and Japanese stream populations (Figure 2.8). The model of no difference between Canadian and Japanese stream fish could not be rejected (%2=6.81, df=3, p=0.09), indicating that these populations are phenotypically indistinguishable concerning plate number. Therefore, the completely plated marine phenotype was not recovered in the CSJSi cross (Figure 2.8, Table 2.4). This provides the first piece of evidence that plate reduction in the stream forms between the two regions is due to changes at few loci. The second indication of strong parallel inheritance is that F 2 hybrids do not show a higher level of variation than Fi hybrids or parents (Tables 2.5 and 2.6) in contrast to Fi hybrids between stream and marine forms (Table 2.7). No differences in the variance of the parental, Fi , or F 2 generations were detected using an analysis of line variances. Lack of segregation variance in F 2 hybrids of stream x stream crosses implies that plate number is either controlled by hundreds of loci or that both the Canadian and Japanese stream populations have diverged from fully plated marine ancestors via changes at the same loci. In this case, the latter conclusion is justified because few loci are involved in the inheritance of plate number. 2.3.3.2 Body shape Although the differences in shape among Canadian and Japanese stream populations and their Fi and F 2 hybrids are small compared to those between stream and marine populations (Figure 2.8), the line means analysis shows that these populations have significantly different body shapes as described by discriminant function 1 (Table 41 2.2). Additivity of inheritance was also rejected, but the additivity-plus-dominance effects model could not be rejected, with the Japanese shape being dominant over that from Canada (Figure 2.8, Table 2.2). However, the phenotypes expressed by the hybrids were all within the range of phenotypes expressed by the parents, and the marine phenotype was not recovered in the Fi hybrids. In fact, the Fi hybrids favoured the parental phenotype that was the most "stream-like", that is, in a direction away from the average discriminant function scores of marine phenotypes (Figure 2.8). Because body shape is not a trait controlled by few loci, the predictions for an analysis of line variances are not clear. In this study, however, Fi and F 2 hybrids are not more variable than their parents are in body shape. The analysis of line variances shows no increase in the amount of variance, whether a result of phenotypic destabilisation in hybrids or segregation, exhibited by Fi or F 2 hybrids (Tables 2.5 and 2.6). This suggests that parallel changes have occurred in the stream forms, but fails to ascertain whether these changes result from changes at the same few loci or at many loci. 2.4 D I S C U S S I O N 2.4.1 Parallel evolution 2.4.1.1 Parallel evolution ofecotypes Parallel evolution of ecotypes, or populations with similar phenotypes and ecological functions, is known for a number of taxa. Anolis lizards in the Antilles islands (Losos 1992) are differentiated into well-defined ecological niches, each with associated morphologies (e.g., hind limb length and body size in lizards). These ecotypes have evolved repeatedly on different islands. That is, different ecotypes on a single island are 42 more closely related to one another than are similar ecotypes from different islands. Like islands, freshwater systems isolate fish into populations that then have the potential for morphological, ecological, and genetic diversification (Waters and Wallis 2001). Parallel evolution of freshwater-resident populations from anadromous marine populations is a repeated theme in fishes. Repeated independent colonisation events resulting in freshwater-resident populations are known for stickleback (McPhail 1994), galaxiids (McDowall 1997, Waters and Wallis 2001), and salmonids (Taylor et al. 1996, Taylor 1999). The most common difference between freshwater and marine fish is reduced body size in freshwater populations (e.g., Ricker 1940, Verspoor and Cole 1989, Bell and Foster 1994), but they also diverge in other aspects of morphology. This study presents evidence that stream ecotypes have evolved from marine ecotypes on two separate continents. The stream fish analysed here show parallel evolution in a size-corrected multivariate measure of body shape and in reduction of body armour. Changes in body shape may be an adaptive response linked to foraging (Webb 1984, Walker 2000) and predator avoidance (Walker 2000) as a result of different selection pressures in stream and marine environments. Reduction of plate number is a trait primarily associated with low densities of avian or piscine predators and/or high densities of invertebrate predators (Nelson 1969, Hagen and Gilbertson 1972, Gross 1977, Moodie and Reimchen 1976, Reimchen 1994, Walker 2000). However, possessing a complete set of lateral plates may have hydrodynamic (Taylor and McPhail 1986) or energetic costs (Bell et al. 1985) associated with it as well. Parallel evolution in sticklebacks has previously been demonstrated in various aspects of morphology (Bell 1974, Schluter and McPhail 1993), life history (Schluter and 43 Nagel 1995, Rundle et al. 2000), and reproductive isolation (Thompson et al. 1997). For the most part, these evolutionary events can be characterised as one of three types: (1) parallel divergence between anadromous and freshwater stickleback, (2) parallel divergence between lake and stream stickleback, and (3) parallel divergence between benthic and limnetic lake stickleback (Thompson et al. 1997). For example, most of the freshwater populations throughout the stickleback range are believed to be the result of independent colonisation events by anadromous and estuarine populations (reviewed in Bell and Foster 1994). In addition, many freshwater populations have strikingly similar morphology: small body size, drab coloration, reduced number of lateral plates, and a small number of dorsal and anal fin rays (Heuts 1947, Okada 1960, Hagen and Moodie 1982, Ziagunov 1983, Ziagunov et al. 1987). Several freshwater populations have lost their pelvic girdles and spines (Bell 1974). But even within different freshwater drainages, populations of stickleback exhibit a high degree of habitat partitioning—into stream-lake pairs, or within a lake, into limnetic-benthic pairs (McPhail 1994). This habitat partitioning results in several differentiated populations, each associated with characteristic morphology, feeding ecology, and mate preferences (Schluter 1996). Perhaps the best studied of these, in terms of parallel evolution, are the limnetic-benthic species pairs in the Strait of Georgia. Limnetic-benthic species pairs are comprised of two ecologically and morphologically distinct populations inhabiting a single lake. Limnetics are slender, narrow-gaped planktivores with numerous gill rakers; benthics are large, deep-bodied benthivores with wide gapes and few gill rakers (McPhail 1984). Limnetic and benthic populations of stickleback are reproductively isolated primarily by size-selective mating preferences (Nagel 1994). Amazingly, this pattern of 44 morphological, ecological, and behavioural diversification has been repeated independently in a number of lakes in this area (Schluter and McPhail 1992, McPhail 1993, 1994) and in two outside of it (Narver 1969, Ziuganov et al. 1987). Here, I quantify parallel evolution in two geographically separated populations of stream-dwelling stickleback in terms of two morphological traits representing extremes of phenotypic and, likely, underlying genetic complexity. First, I used number of lateral plates, a standard measure in morphological analyses of stickleback that is probably determined by very few genes. Second, I used a measure of body shape (represented by a multivariate axis of differences in landmark positions), a more complex and less common measure (but see Walker 1997, Pritchard 1998, and Walker et al. 2000) that is likely under polygenic control. 2.4.1.2 Lateral plate number Nearly all the variation between stream and marine hybrids was due to the effects of habitat. The ratio y, a measure of how parallel the evolutionary trajectories of two lineages are, was nearly 1, indicating almost perfectly parallel evolution in plate reduction in the stream forms. Reduction in plate number is a common trend in freshwater sticklebacks. Number of lateral plates in stickleback is an adaptive trait associated with the presence of avian and piscine predators, and reduction is common when predators are not abundant (Hagen and Gilbertson 1972, Gross 1977, Moodie and Reimchen 1976). Hagen and Gilbertson (1973b) showed that the number of plates in freshwater stickleback populations varies geographically with the distribution of different types of predators with completely plated phenotypes being favoured when predators are gape-limited (i.e., fish and birds) and reduction common with invertebrate predators that 45 handle and manipulate prey. Parallel differences in lateral plate phenology may stem from selection by similar predators on both continents, or parallel in intensity of predation on freshwater populations in freshwater (Reimchen 1994). In addition, sticklebacks with few lateral plates have increased body flexion and therefore greater burst speed (Taylor and McPhail 1986), which may enable them to effectively escape certain predators (i.e., diving birds that submerge for only brief periods). Nelson (1969) suggested that loss of plates and shorter spines may be advantageous to stickleback seeking to escape into dense vegetation or hide in muddy substrate. Finally, in some populations, loss of body armour is associated with reduced calcium levels in the water, suggesting that heavy armouring is costly to maintain when minerals are not abundant (Bell et al. 1985). 2.4.1.3 Body shape Body shape, as a complex measure, had components of both parallel and non-parallel evolution. The Canadian and Japanese pairs had close to perfect parallel evolution (y=0.998) in the first discriminant function describing body shape. The first discriminant function was a composite measure including depth of body profile, and dorsal and anal fin placement. In terms of landmarks, the differences between stream and marine forms described by the first discriminant function were a change in body depth and shifts in the length and placement of the dorsal and anal fins. Differences in body shape and fin area and placement have been linked to foraging and mobility of fishes in different environments (Webb 1984a, 1984b). The body shape of marine populations is vertically compressed, resulting in a streamlined shape causing little drag, an adaptation for prolonged swimming, advantageous for cruising in open water in search of 46 zooplankton and for migration (Baumgartner et al. 1988, Walker 1997). Marine sticklebacks also have long dorsal and anal fins placed further back on the body than freshwater sticklebacks that provide superior acceleration.capability for foraging on zooplankton (Baumgartner et al. 1988, Moyle and Cech 1988, Walker 1997). Stream-dwelling stickleback have increased body depth, which is correlated with the ability to manoeuvre in tight quarters, dense vegetation, and highly structured environments such as those found in streams (Webb 1984a, 1984b). At the same time, the stream-dwelling stickleback in Canada and Japan have accumulated enough non-parallel differences in body shape to be distinguishable from one another using discriminant function analysis. The main difference is that Japanese stream population has longer dorsal and anal fins than the Canadian stream population. Several factors, both adaptive and incidental, could contribute to these slight differences. First, the stream environments undoubtedly have some differences in history, topography, and associated floral and faunal communities that are unique to their geographic location. Long median fins (dorsal, caudal, and/or anal) are associated with the presence of predatory fishes and can functionally increase body depth (and therefore manoeuvrability) without compromising hydrodynamic efficiency (Walker 1996). Second, although the marine populations in both regions appear quite similar, it is possible that the marine ancestors from which each stream population evolved were slightly differentiated. These slight differences in genetic background could be amplified or modified in descendant stream populations, causing divergent phenotypes in the stream populations of different regions. Drift, mutation, and founder events could also have played a role in the divergence of the stream populations. 47 2.4.2 Parallel inheritance My second major objective in this study was to examine inheritance patterns to determine if parallel changes at the phenotypic level were the result of parallel changes at the genotypic level. This parallel inheritance can take two forms. In the strong form, parallel evolution results in identical allele substitutions at identical loci in the descendant populations. In the weak form, parallel evolution results from changes in different loci that behave similarly in crosses between the divergent types. 2.4.2.1 Lateral plate number In both of the stream-marine pairs, armour plate characteristics exhibit additive-plus-dominance effects. In both cases the completely plated marine phenotype is partially dominant over the low-plated stream phenotype in both regions (Figure 2.6). These effects fit the data well without any need for an interaction term describing the differences in genetic components between continents. Therefore, the stream populations show parallel genetic changes from the marine ancestral type in plate number. The finding of dominance effects in plate number is consistent with conclusions from a previous genetic study of limnetic and benthic sticklebacks (Hatfield 1997) that also found that few genes of large effect control plate number. An examination of the variances in Tables 2.3 and 2.7 indicates that few genes are responsible for plate number in these populations as well. In hybrids between stream and marine populations, the F 2 generation is expected to have increased variance relative to the parental and Fi generations if the derived stream populations differ at one or more of the few loci responsible for plate reduction. This segregational variance is a result of the segregation of different alleles from the parental lines (Lynch and Walsh 1998). The 48 stream x marine hybrids CSCM2 and JSJM2 show increased segregational variance. Therefore, parallel inheritance of lateral plate number is largely the result of changes at the same or closely linked loci. Previous work using electrophoresis (Hagen and Gilbertson 1973a, Avise 1976), and line means analysis and estimates of effective number of loci (Hatfield 1997), imply that plate number is indeed a character that is under the control of few genes. The hypothesis of few genes with large effect is further supported by preliminary molecular work by Peichel et al. (2001) that identified two quantitative trait loci with large effects on plate number affecting the limnetic and benthic sticklebacks. 2.4.2.2 Body shape Body shape characteristics have a different pattern of inheritance than plate number. The composite characteristic of body shape exhibits an additive mode of inheritance of differences between stream and marine hybrids. A similar pattern of inheritance was seen in both regions, and the model fit both the Canadian and Japanese line means well with no evidence of strong differences between the regions. Thus body shape also exhibits parallel inheritance. Because many traits—muscular, skeletal (including armour), and developmental growth pattern (Bell 1980) differences, for instance—define body shape, the trait is undoubtedly controlled by many genes. However, a finding of predominantly additive effects does not exclude the possibility of dominant genes being present and contributing, nor does it exclude the possibility of genes interacting such that epistasis could be present. Basically, the finding of net additive inheritance suggests that the overall effects of the genes are additive, but does not preclude the possibility of dominant or epistatic interactions taking place at individual 49 genes. Finally, it is possible that dominance or epistasis could be present in these populations, but that the effects are too weak to detect given the power of this study. The inability to detect any segregation variance in body shape is a further indication that many genes are responsible for its control. Using line variances analysis to observe segregational variance in F 2 hybrids is probably most effective when the number of involved loci is low (Schat 1996). As more and more genes are involved in determining a trait, as they likely are with body shape, the phenotypes exhibited by F 2 hybrids approach a continuous distribution (this is in opposition to the three—few plates, partially plated, and completely plated—relatively discrete phenotypes of lateral plate number). In addition, the extremes of phenotype seen in the parental population are represented in the hybrids with a decreasing frequency as the number of involved genes builds (Ayala 1982). Therefore, with a small sample size, my power to detect segregation variance in the F 2 hybrids is greatly diminished. 2.4.3 Weak vs. strong parallel inheritance The third objective of the study was to provide a preliminary test of whether parallel inheritance was weak (allele substitutions evolve at different loci that behave similarly in analysis of line variance) or strong (substitutions occur at the same locus in each derived lineage). To distinguish between weak and strong parallel inheritance, I analysed line means of stream x stream hybrids to look for recombination patterns similar to those found in complementation tests. I also examined line variances for evidence of segregational variance in crosses between the derived Canadian and Japanese stream populations. 50 2.4.3.1 Lateral plate number I found no evidence of complementation in the stream x stream hybrids. Because only a few genes are responsible for plate number (Hagen and Gilbertson 1973a, Avise 1976, Hatfield 1997, Peichel et al. 2001, this study), evidence from the line means analysis of crosses between Canadian and Japanese stream populations is useful for indicating whether the parallel inheritance shown in plate number is weak or strong. Complementation tests, first applied by Benzer (1961) to study genetic fine structure in bacteria, postulate that hybridising two mutants will result in a non-mutant phenotype if the mutations in each line are recessive and on different loci. If the mutations are on the same loci, then the hybrid will maintain the mutant phenotype. Applying this logic to the stream x stream crosses, in which the hybrids remained as low-plated morphs like the parents, suggests that the genes responsible for reduction of plate number are at the same loci in the Canadian and Japanese stream populations. This provides the first piece of evidence for strong parallel inheritance in plate number. The second piece comes from the analysis of line variances and builds on previous indications that few genes control plate number. I found no differences in the variances of parental, Fi and F 2 generations of stream x stream crosses. Lack of segregation variance in hybrids suggests that either the trait is controlled by many loci, or that the loci controlling the trait behave the same in both parent populations. Since there is no evidence that plate number is controlled by more than a few genes, I conclude that the major loci affecting plate number are the same in both Canadian and Japanese stream populations. 51 2.4.3.2 Body shape Like plate number, there was no complementation in body shape of the stream x stream (CSJS) hybrids. Although the means of the parental and hybrid populations were different, the body shapes displayed in the Fj hybrids were not more marine-like than those of the parents were. If body shape were controlled by only a few genes, the lack of complementation between genotypes affecting body shape would suggest that the genes may be at the same loci in the Canadian and Japanese stream populations, evidence for strong parallel inheritance in plate number. However, unlike plate number, there is no existing evidence that only a few genes control body shape. In this case there is no expectation for complementation to occur, and lack of complementation is a poor argument for the changes all taking place at the same loci. Stream x stream hybrids show an effect of destabilisation of phenotype in the analysis of line variances. Disruption of canalization resulting in individuals with supernumerary and hyper- and hypomorphic characteristics is known for hybrids in other fish taxa (Neff and Smith 1979, Leary et al 1983, Graham and Felley 1985, Leary et al. 1985a, 1985b, Burkhead and Williams 1991). Often, increased variance and decanalized inheritance patterns result from developmental instability in hybrids (Graham and Felley 1985, Leary et al. 1985a, 1985b, Burkhead and Will iams 1991). Although both generations of hybrids had greater variance than the parents did, the F 2 generation did not have an expansion of variance greater than that of the F i ' s , and I did not detect any evidence of segregation variance. There is a second reason to expect the hybrids to show increased variance relative to the parents in this particular study. When calculating the composite measure of body 52 shape, I used a discriminant function analysis on the pure-type (parental-type) populations only, and then estimated the scores for hybrids. Because the discriminant function analysis first minimises within-group variation, the pure-type crosses are expected to have a lower variance than the hybrid crosses that did not undergo the same transformation. The discrepancies in variance amongst generation, therefore, may just be an artefact of the discriminant function analysis. Although there was no complementation, a sign of strong parallel inheritance, there was no evidence of segregational variance indicating few genes involved, either. For body shape, my data are unable to differentiate between the strong and weak form of parallel evolution, and therefore only support the weaker form. 2.4.4 A perspective on parallel evolution and parallel inheritance How parallel the evolutionary trajectories of two populations are depends on the relative roles of the selective forces acting on them and the importance of their particular histories, or design constraints (Wake 1991, Travisano et al. 1995). Parallel evolution establishes a strong argument for the role of natural selection in evolution (Endler 1986, Schluter and Nagel 1995). Nonetheless, we rarely expect to see exactly parallel evolution: selective environments vary spatially and temporally, and the direction evolution will take is at least partially dependent on the frequency and presence of particular alleles in the ancestral populations (Cohan and Hoffman 1986, Travisano et al. 1995, Schat 1996). When comparing the total phenotype between two populations derived in parallel, even when they are the result of strong selective pressures, both parallel components, e.g., reduced plate number and body depth, and non-parallel components, e.g., fin lengths, are observed. 53 Parallel evolution of phenotype does not necessarily equate to parallel evolution of genotypes. Sometimes changes that appear homologous are, upon closer examination, due to modifications in different structures. For instance, both Old and New World populations of Drosophila subobscura exhibit variation in wing length increasing with latitude. However, the convergence in wing length phenotype is due to changes at analogous, not homologous, structures in different sections of the wing (Huey et al. 2000, Gilchrist et al. 2001). While selection in separate environments is generally thought to cause genetic divergence (Cohan and Hoffman 1989), our practical understanding of the genetic basis for observed parallel patterns is weak. Analysis of line crosses is one way of beginning to understand genetic causes of parallel evolution. Empirical evidence of parallel inheritance is rare, though. Is it possible that populations responding to selection pressures in similar environments could evolve the same genetic response? The answer appears to be yes, in some situations. Using molecular techniques on experimental populations of bacteriophages, Cunningham et al. (1997) found widespread parallel evolution of nucleotide deletions. Similarly, Wichman et al. (1999) were able to map identical amino acid substitutions conferring adaptive advantages to a novel host in replicate lines of viruses. Using evidence of segregation variance in hybrid lines, Schat et al. (1996) found parallel phenotypic changes are the result of parallel genotypic changes for heavy metal tolerance in Silene vulgaris. This also appears to be the case for plate reduction in stream stickleback (this study). However, both these traits are believed to be under simple genetic control. Whether closer examination of parallel evolution for other traits, more complex traits, and other taxa will show this pattern is open for exploration. 54 Table 2.1. Averages and standard errors (in parentheses) for lateral plate number and body shape based on the first discriminant function in the pure-type crosses. Plate number Body shape Stream Marine Stream Marine Canada 5.6 (0.21) 29.2 (0.70) -2.7 (0.24) 3.2 (0.28) Japan 5.6 (0.16) 31.5 (0.66) -3.5 (0.25) 1.8 (0.17) 55 Table 2.2. Goodness-of-fit %2 test statistics from joint-scaling tests on line means of lateral plate number and body shape for stream x marine crosses. No Difference (df=7) Continent (df=6) Additive (df=5) Additive + Dominance (df=4) Plate number Body Shape 3845.043*** 578.6118*** 3837.813*** 578.569*** 125.5939*** 6.809 (NS) 1.91062(NS) ***significant at p<0.001 N S - not significant Table 2.3. Goodness-of-fit %2 test statistics from joint-scaling tests on line variances of lateral plate number and body shape for stream x marine crosses. No Difference (df=7) Continent (df=6) Destabilisation (df=5) Segregation (df=4) Plate number Body Shape 27.78941*** 18.11775* 27.70453*** 15.92748* 12.38701* 9.066137 (NS) 3.988427(NS) * * * significant at p<0.001 *significant at p<0.05 N S - not significant 56 Table 2.4. Goodness-of-fit %2 test statistics from joint-scaling tests on line means of lateral plate number and body shape for stream x stream crosses. No Difference (df=3) Additive (df=2) Additive + Dominance (df=l) Plate number Body Shape 6.807246 (NS) 12.36* 6.29* 0.81 (NS) *significant at p<0.05 NS— not significant Table 2.5. Goodness-of-fit %2 test statistics from joint-scaling tests on line variances of lateral plate number and body shape for stream x stream crosses. No Difference (df=3) Destabilisation (df=2) Segregation (df=l) Plate number Body Shape 0.442924 (NS) 0.986412 (NS) NS— not significant 57 Table 2.6. Variances for log(lateral plate number) and body shape (first discriminant function) in stream x stream hybrids. log(Plate number) Body shape Canada 0.00477 1.03877 Japan 0.00231 0.97362 Fi 0.00635 0.80198 F 2 0.00283 2.34853 Table 2.7. Variances for log (lateral plate number) and body shape (first discriminant function) in stream x marine hybrids. log(Plate number) Body shape Canada Japan Canada Japan Stream 0.00477 0.00231 1.03877 0.97362 Marine 0.00241 0.00178 1.54656 0.57217 F, 0.00860 0.01221 1.20323 2.67776 F 2 0.09926 0.04026 4.90685 1.78412 58 Figure 2.1. Landmarks used for shape analyses. See text for a full description of each 59 TRAIT A y=1 TRAIT B y=0 M A R I N E S T R E A M M A R I N E S T R E A M TRAIT c y>0.5 TRAIT D o y=0.5 o M A R I N E S T R E A M M A R I N E S T R E A M Figure 2.2. Some possible outcomes i n two lineages (indicated by open and fi l led circles) evolving independently in two environments (marine and stream). The quantity Y = M Shabita/(M Shabitat + M Si nt erac(ion) i s a n i n d e x o f parallel evolution. Scenarios A and C represent complete and partial parallel evolution, respectively. N o parallel evolution has occurred when y^O.5 (scenarios B and D ) . 60 CM z o t 1 Z Z) Li. I -z -1 H IT O CO Q -3 JS o o o o o o CS o o o o CM AA A A JM -4 -2 DISCRIMINANT FUNCTION 1 Figure 2.4. Discriminant function scatter plot. Low discriminant function 1 scores correspond to a stream-resident body shape, and high scores correspond to a marine-like body shape. Traits reflected by discriminant function 1 have evolved in parallel between Canada and Japan, whereas traits reflected by discriminant function 2 have n A. CS= Canadian stream, JS= Japanese stream, CM= Canadian marine, and JM= Japanese marine. 62 Figure 2.5. Changes in landmark position as described by (A) discriminant function one and (B) discriminant function two. Dark vectors represent net change in each landmark per two standard deviations of each discriminant function starting from a stream-like shape and moving towards a marine-like shape. Vector lengths are doubled to make them visible. Discriminant function one describes parallel changes in stream forms, and discriminant function two describes non-parallel changes between stream forms. 63 ( s a i v i d i v y a i v i d o yaaiAiriN) eoi 64 65 R E F E R E N C E S Avise, J. C. 1976. Genetics of plate morphology in an unusual population of threespine sticklebacks (Gasterosteus aculeatus). Genetical Research 27:33-46. Ayala, F. J. 1982. Pp, 154-175 in Population and evolutionary genetics. Benjamin/Curnmings publishing Company, Inc. Menlo Park, California. Baumgartner, J.V., M. A. Bell, and P. H. Weinberg. 1988. Body form differences between the Enos Lake species pair of threespine sticklebacks {Gasterosteus aculeatus complex). Canadian Journal of Zoology 66:467-474. Bell, M . A. 1974. Reduction and loss of the pelvic girdle in Gasterosteus (Pisces): a case of parallel evolution. Natural History Museum of Los Angeles County Contributions in Science 257:1-36. Bell, M . A. 1976. Evolution of phenotypic diversity in Gasterosteus aculeatus superspecies on the pacific coast of North America. Systematic Zoologica 25:211-227. Bell, M . A. 1980. Lateral plate polymorphism and ontogeny of the complete plate morph of threespine sticklebacks (Gasterosteus aculeatus). Evolution 35:67-74. Bell, M . A. 1982. Differentiation of adjacent stream populations of threespine sticklebacks. Evolution 36: 189-199. Bell, M . A. 1984. Evolutionary phenetics and genetics: the threespine stickleback, Gasterosteus aculeatus, and related species. Pp. 431-528 in B. J. Turner, ed. Evolutionary genetics of fishes. Plenum, New York. Bell, M . A. 1994. Paleobiology and evolution of threespine stickleback. Pp. 438-471 in M . A. Bell and S. A. Foster, eds. The evolutionary biology of the threespine stickleback. Oxford University Press, New York. Bell, M . A. and G. Orti. 1994. Pelvic reduction in threespine stickleback from Cook Inlet lakes: geographical distribution and intrapopulation variation. Copeia. 1994:314-325. 67 Bell, M . A. and S. A. Foster. 1994. Introduction to the biology of the threespine stickleback. Pp. 1-27, 472-486 in M. A. Bell and S. A. Foster, eds. The evolutionary biology of the threespine stickleback. Oxford University Press, New York. Benzer, S. 1961. On the topography of the genetic fine structure. Proceedings of the National Academy of Science 47:403-15. Bookstein, F. L. 1990. Part UI, Section C: introduction to methods for landmark data, pp.215-225. In F J . Rohlf and F.L. Bookstein (eds.), Proceedings of the Michigan Morphometries Workshop. Special Publication No. 2, The University of Michigan Museum of Zoology, Ann Arbor, Michigan, U.S.A. Borland, M . 1986. Size-assortative mating in threespine sticklebacks from two sites on the Salmon River, British Columbia. M.Sc. thesis. University of British Columbia, Vancouver. Bourgeois, J. F., D. M . Blouw, J. P. Koenings, and M . A. Bell. 1994. Multivariate analysis of geographic covariance between phenotypes and environments in the threespine stickleback, Gasterosteus aculeatus, from the Cook Inlet area, Alaska. Canadian Journal of Zoology 62:1497-1509. Burkehead, N. M . and J. D. Williams. 1991. An intergeneric hybrid of a native minnow, the golden shiner, and an exotic minnow, the rudd. Transactions of the American Fisheries Society 120:781-795. Buth, D. G. 1984. The application of electrophoretic data in systematic studies. Annual Review of Ecology and Systematics 15:502-522. Buth, D. G. and T. R. Haglund. 1994. Allozyme variation in the in Gasterosteus aculeatus complex. Pp. 61-84 in M . A. Bell and S. A. Foster, eds. The evolutionary biology of the threespine stickleback. Oxford University Press, New York. Cavalli, L. L . 1952. An analysis of linkage in quantitative inheritance. Pp. 135-144 in E.C.R. Reeve and C H . Waddington, eds. Quantitative inheritance. Her Majesty's Stationery Office, London. 68 Cohan, F. M . , and A. A. Hoffman. 1986. Genetic divergence under uniform selection. II. Different responses to selection for knockdown resistance to ethanol among Drosophila melanogaster populations and their replicate lines. Genetics 114:145-163. Cohan, F. M . , and A. A. Hoffman. 1989. Uniform selection as a diversifying force in evolution: evidence from Drosophila. American Naturalist 134:613-637. Coyne, J. A. and H. A. Orr. 1989. Patterns of speciation in Drosophila. Evolution 43:362-381. Coyne, J. A. and H. A. Orr. 1997. "Patterns of speciation in Drosophila" revisited. Evolution 51:295-303. Culver, D. C , T. C. Kane, and D. W. Fong. 1994. Adaptation and evolution in caves. Harvard University Press, Cambridge, Mass. Cunningham, C. W., K. Jeng, H. Hutsi, M . Badgett, I. J. Molineaux, D. M . Hillis, and J. J. Bull. 1997. Parallel molecular evolution of deletions and nonsense mutations in bacteriophage T7. Molecular Biology and Evolution 14:113-116. Dobzhansky, T. 1940. Speciation as a stage in evolutionary divergence. The American Naturalist 74:312-321. Endler, J. A. 1986. Natural selection in the wild. Monographs in population biology, no. 21. Princeton University Press, Princeton, New Jersey. Filliben, J. J. and A. Heckett. 2002. Section 1.3.5.10 in Engineering Statistics Handbook, C Croarkin and P. Tobias, eds. National Institute of Standards and Technology. www.itl.nist.gov/div898/handbook/eda/section3/eda35a.htm. Fitzpatrick, B. M . 2002. Molecular correlates of reproductive isolation. Evolution 56:191-198. Foote, C. J., C. C. Wood, and R. E.Withler. 1989. Biochemical genetic comparisons of sockeye salmon and kokanee, the anadromous and nonanadromous forms of Oncorhynchus nerka. Canadian Journal of Fisheries and Aquatic Sciences 46:149-158. 69 Futuyma, D. J. 1986. Evolutionary biology. 2na ed. Sinauer, Sunderland, Mass. Gach, M . A., and T. E. Reimchen. 1989. Mitochondrial D N A patterns among endemic stickleback from the Queen Charlotte Islands: a preliminary survey. Canadian Journal of Zoology 67:1324-1328. Graham, J. H. and J. D. Felley. 1985. Genomic coadaptation and developmental stability within introgressed populations of Enneacanthus gloriosus and E. lobesus (Pisces, Centrarchidae). Evolution 39:104-114. Gross, H. P. 1977. Adaptive trends of environmentally sensitive traits in the three-spined stickleback, Gasterosteus aculeatus L. Zeitschrift fur zoologische Systematic und Evolutionsforschung 15:252-278. Gross, H. P. 1978. Natural selection by predators on the defensive apparatus of the three-spined stickleback, Gasterosteus aculeatus L . Canadian Journal of Zoology 56:398-413. Hagen, D. W. 1967. Isolating mechanisms in threespine sticklebacks {Gasterosteus). Journal of the Fisheries Research Board of Canada 24:1637-1693. Hagen, D. W., and L. G. Gilbertson. 1972. Geographic variation and environmental selection in Gasterosteus aculeatus L. in the Pacific Northwest, America. Evolution 26:32-51. Hagen, D. W., and L. G. Gilbertson. 1973a. The genetics of plate morphs in freshwater threespine sticklebacks. Heredity 31:75-84. Hagen, D. W., and L. G. Gilbertson. 1973b. Selective predation and the intensity of selection acting upon the lateral plates of threespine sticklebacks. Heredity 30:273-287. Hagen, D.W., and J. D. McPhail. 1970. The species problem within Gasterosteus aculeatus on the pacific coast of North America. Journal of the Fisheries Research Board of Canada 27:147-155. 70 Hagen, D. W., and G. E. E . Moodie. 1982. Polymorphism for plate morphs in Gasterosteus aculeatus on the East Coast of Canada and a hypothesis for their global distribution. Canadian Journal of Zoology 60:1032-1042. Hard, J. J., W. E . Bradshaw, and C. M . Holzapfel. 1992. Epistasis and the genetics of photoperiodism between populations of the pitcher-plant mosquito, Wyeomyia smithii. Genetics 131:389-396. Hard, J. J., W. E. Bradshaw, and C. M . Holzapfel. 1993. The genetic basis of photoperiodism and its evolutionary divergence among populations of the pitcher-plant mosquito, Wyeomyia smithii. American Naturalist 142:457-473. Hatfield, T. 1997. Genetic divergence in adaptive characters between sympatric species of stickleback. American Naturalist 149:1009-1029. Hatfield, T. and D. Schluter. 1996. A test for sexual selection on hybrids of two sympatric sticklebacks. Evolution 50:2429-2434. Hay, D. E. and J. D. McPhail. 1974. Mate selection in threespine sticklebacks (Gasterosteus). Canadian Journal of Zoology 53:441-450. Hayman, B. I. 1958. The separation of epistatic from additive and dominance variation in generation means. Heredity 12:371-390. Hayman, B. I. 1960a. Maximum likelihood estimation of genetic components of variation. Biometrics 16:369-381. Hayman, B. I. 1960b. The separation of epistatic from additive and dominance variation in generation means. II. Genetica 31:133-146. Heuts, M . J. 1947. Experimental studies on adaptive evolution in Gasterosteus aculeatusL. Evolution 1:89-102. Higuchi, M . and A. Goto. 1995. Genetic evidence supporting the existence of two distinct species in the genus Gasterosteus around Japan. Environmental Biology of Fishes 47:1-16. 71 Honma, Y. and E. Tamura. 1984. Anatomical and behavioural differences among threespine sticklebacks: the marine form, the landlocked form, and their hybrids. Acta Zoologica 65:79-87. Jones, R., D. C. Culver, and T. C. Kane. 1992. Are parallel morphologies of cave organisms the result of similar selection pressures? Evolution 46:353-365. Kassen, R., D. Schluter, and J. D. McPhail. 1995. Evolutionary history of threespine sticklebacks (Gasterosteus spp.) in British Columbia: insights from a physiological clock. Canadian Journal of Zoology 73:2154-2158. Lande, R. 1981. The minimum number of genes contributing to quantitative variation between and within populations. Genetics 99:541-553. Leary R. F., F. W. Allendorf, and K. L. Knudsen. 1983. Consistently high meristic counts in natural hybrids between brook trout and bull trout. Sytemetic Zoology 32:369-376. Leary R. F., F. W. Allendorf, and K. L. Knudsen. 1985a. Inheritance of meristic variation and the evolution of developmental stability in rainbow trout. Evolution 39:308-314. Leary R. F., F. W. Allendorf, and K. L. Knudsen. 1985b. Developmental instability and high meristic counts in interspecific hybrids of salmonid fishes. Evolution 39:1318-1326. Losos, J. B. 1992. The evolution of convergent structure in Caribbean Anolis communities. Systematic Biology 41:403-420. Lynch, M . , and J. B. Walsh. 1997. Quantitative genetics: biology and estimation, pp. 205-249. Sinauer, Sunderland, Mass. Mather, K., and J. L . Jinks. 1982. Biometrical genetics, the study of continuous variation. Pp. 65-81. 3 r d ed. Chapman and Hall, New York. MathSoft, Inc. 1988-1999. S-PLUS 2000 Professional Release 1. Mayr, E. 1963. Animal species and evolution. Harvard University Press, Cambridge, Massachusetts. 72 Mayr, E. 1942. Systematics and the origin of species. Columbia University Press, New York. McDowall, R. M . 1997. The evolution of diadromy in fishes (revisited) and its place in phylogenetic analysis. Review of Fish Biology and Fisheries 7:443-462. McPhail, J. D. 1984. Ecology and evolution of sympatric sticklebacks {Gasterosteus): morphological and genetic evidence for a species pair in Enos Lake, British Columbia. Canadian Journal of Zoology 62:1402-1408. McPhail, J. D. 1994. Speciation and the evolution of reproductive isolation in the stickleback (Gasterosteus) of southwestern British Columbia. Pp.399-437 in M.A. Bell and S.A. Foster, eds. The evolutionary biology of the threespine stickleback. Oxford Univ. Press, New York. McPhail, J. D., and C. C. Lindsey. 1970. Freshwater fishes of northwestern Canada and Alaska. Bulletin of the Fisheries Research Board of Canada 173:1-381. Moodie, G. E. E . , and T. E. Reimchen. 1976. Phenetic variation and habitat differences in Gasterosteus populations of the Queen Charlotte Islands. Systematic Zoology 25:49-61. Nagel, L. M . 1994. The parallel evolution of reproductive isolation in threespine sticklebacks. M.Sc. thesis. University of British Columbia, Vancouver. Narver, D.W. 1969. Phenotypic variation in the threespine sticklebacks (Gasterosteus aculeatus) of the Chignik River System, Alaska. Journal of the Fisheries Research Board of Canada 26:405-412. Neff, R. A. and G. R. Smith. 1979. Multivariate analysis of hybrid fishes. Systematic Zoology 28:176-196. Nei, M . 1972. Genetic distances between populations. American Naturalist 106:282-292. O'Reilly, P., T. E. Reimchen, R. Beech, and C. Strobeck. 1993. Mitochondrial DNA in Gasterosteus and Pleistocene glacial refugium on the Queen Charlotte Islands, British Columbia. Evolution 47:678-684. 73 Okada, Y. 1960. Studies on the freshwater fishes of Japan. Journal of the Faculty of Fisheries, Prefectual University of Mie 4:1-860. Orti, G., M . A. Bell, T. E . Reimchen, and A. Meyer. 1994. Global survey of mitochondrial DNA sequences in the threespine stickleback: evidence for recent migrations. Evolution 49:608-622. Peichel, C. L . , K.S. Nereng, K. A. Ohgi, B. L . E. Cole, P. F. Colosimo, C. A. Buerkle, D. Schluter, and D. M . Kingsley. 2001. The genetic architecture of divergence between threespine stickleback species. Nature 414:901-905. Pigeon, D., A. Chouinard, and L. Bernatchez. 1997. Multiple modes of speciation involved in the parallel evolution of sympatric morphotypes of lake whitefish (Coregonus clupeaformis, Salmonidae). Evolution 51:196-205. Pritchard, J. R. 1998. Competition and character Displacemnt in Sticklebacks. Ph.D. thesis. University of British Columbia, Vancouver, Canada. Reimchen, T. E. 1994. Predators and morphological evolution in threespine stickleback. Pp. 240-273 in M . A. Bell and S. A. Foster, eds. The evolutionary biology of the threespine stickleback. Oxford University Press, New York. Ricker, W. E. 1940. On the origin of kokanee, a freshwater type of sockeye salmon. Transactions of the Royal Society of Canada 344:121-135. Rohlf, F. J. 1998. Relative warps, Version 1.18. Ecology and Evolution, SUNY. Stonybrook, NY. Rundle, H. D. and D. Schluter. 1998. Reinforcement of stickleback mate preferences: sympatry breeds contempt. Evolution 52:200:208. Rundle, H. D., L. Nagel, J. W. Boughman, D. Schluter. 2001. Natural selection and parallel speciation in sympatric sticklebacks. Science 287:306-308. Sasa, M . M . , P. T. Chippendale, N. A. Johnson. 1998. Patterns of postzygotic isolation in frogs. Evolution 52:1811-1820. 74 Schat, H., R. Vooijs, and E. Kuiper. 1996. Identical major gene loci for heavy metal tolerances that have independently evolved in different local populations and subspecies of Silene vulgaris. Evolution 50:1888-1895. Schluter, D. 1996. Ecological speciation in postglacial fishes. Philosophical Transactions of the Royal Society of London B. 351:807-814. Schluter, D. and J. D. McPhail. 1992. Ecological character displacement and speciation in sticklebacks. American Naturalist 140:85-108. Schluter, D. and J. D. McPhail. 1993. Character displacement and replicate adaptive radiation. Trends in Ecology & Evolution 6:197-200. Schluter, D. and L. M . Nagel. 1995. Parallel speciation by natural selection. American Naturalist 146:292-301. Scion Corporation. 1998. Scion Image. Frederick, MD. http:Wwww.scioncorp.com. Simpson, G. G. 1944. Tempo and mode in speciation. Columbia University Press, New York. Sokal, R. R. and F. J. Rohlf. 1981. Biometry. 2 n d ed. W. H. Freeman, New York. Swarup, H. 1958. Stages in the development of the stickleback Gasterosteus aculeatus (L.). Journal of Embryology and Experimental Morphology 6:373-383. Tamarin, R. H. 1993. Pp. 501-521 in Principles of Genetics, 4 t h ed. Wm. C. Brown Communications, Inc., Dubuque, IA. Taylor, E. B. 1999. Species pairs of north temperate freshwater fishes: evolution, taxonomy, and conservation. Review of Fish Biology and Fisheries 9:299-324. Taylor, E. B., C. J. Foote, and C. C. Woode. 1996. Molecular genetic evidence for parallel life history evolution within a pacific salmon (sockeye salmon and kokanee, Oncorhynchus nerka). Evolution 50:401-416. Thompson, C. E. , E. B. Taylor, and J. D. McPhail. 1997. Parallel evolution of lake-stream pairs of threespine sticklebacks (Gasterosteus) inferred from mitochondrial DNA variation. Evolution 51:1955-1965. 75 Tilley, S. G., and M . J. Mahoney. 1996. Patterns of genetic differentiation in salamanders of the Desmognathus ochrophaeus complex. Herpetological Monographs 10:1-42. Tilley, S. G., P. A. Verrel, and S. J. Arnold. 1990. Correspondence between sexual isolation and allozyme differentiation: a test in the salamander Desmognathus orchrophaeus. Proceedings of the National Academy of Sciences 87:2715-2719. Tinbergen, N. 1948. Social releasers and the experimental method required for then-study. Wilson Bulletin 60:6-51. Treves, D. S., S. Manning, and J. Adams. 1998. Repeated evolution of an acetate-crossfeeding polymorphism in long-term populations of Escherichia coli. Molecular Biology and Evolutionl5:789-797. Travisano, M . , F. Vasi, and R. E. Lenski. 1995. Long term experimental evolution in Escherichia coli. IE. Variation among replicate populations in correlated responses to novel environments. Evolution 49:189-200. Verspoor, E. and L. J. Cole. 1989. Genetically distinct sympatric populations of resident and anadromous Atlantic salmon, Salmo salar. Canadian Journal of Zoology 67:1453-1461. Wake, D. B. 1991. Homoplasy: the result of natural selection or the result of design limitations? American Naturalist 138:543-567. Walker, J. A. 1997. Ecological morphology of lacustrine threespine stickleback Gasterosteus aculeatus L. (Gasterosteidae) body shape. Biological Journal of the Linnean Society 61:3-50. Walker, J. A., and M . A. Bell. 2000. Net evolutionary trajectories of body shape evolution within a microgeographic radiation of threespine sticklebacks (Gasterosteus aculeatus). Journal of Zoology, London 252:293-302. Waters, J. M . and G. P. Wallis. 2001. Cladogenesis and loss of the marine life-history phase in freshwater galaxiid fishes (Osmeriformes: Galaxiidae). Evolution 55:587-597. 76 Webb, P. W. 1984a. Body form, locomotion, and foraging in aquatic vertebrates. American Zoologist 24:107-120. Webb, P. W. 1984b. Form and function in fish swimming. Scientific American 251:58-68. Wichman, H. A., M . R. Badgett, L . A. Scott, C. M . Boulianne, and J. J. Bull. 1999. Different trajectories of parallel evolution during viral adaptation. Science 285:422-424. Wooton, R. J. 1984. A functional biology of sticklebacks. Croom Helm, London, and University of California Press, Berkeley. Yamada, M . M . Higuchi, and A. Goto. 2001. Extensive introgression of mitochondrial DNA found between two genetically divergent form of threespine stickleback, Gasterosteus aculeatup, around Japan. Environmental Biology of Fishes 61:269-284. Ziuganov, V. V. 1983. Genetics of osteal plate polymorphism and microevolution of threespine stickleback (Gasterosteus aculeatus L.). Theoretical and Applied Genetics 65:239-246. Ziuganov, V. V., G. J. Golovatjuk, K. A. Savvaitova, and V. F. Bugayev. 1987. Genetically isolated sympatric forms of threespine stickleback, Gasterosteus aculeatus, in Lake Azibachije (Kamchatka Peninsula, U.S.S.R.). Environmental Biology of Fishes 18:241-247. 77 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0090387/manifest

Comment

Related Items