UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Mechanisms of divergence in threespine stickleback (Gasterosteus aculeatus) Marchinko, Kerry Brian 2009

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2009_fall_marchinko_kerry.pdf [ 11.37MB ]
Metadata
JSON: 24-1.0067513.json
JSON-LD: 24-1.0067513-ld.json
RDF/XML (Pretty): 24-1.0067513-rdf.xml
RDF/JSON: 24-1.0067513-rdf.json
Turtle: 24-1.0067513-turtle.txt
N-Triples: 24-1.0067513-rdf-ntriples.txt
Original Record: 24-1.0067513-source.json
Full Text
24-1.0067513-fulltext.txt
Citation
24-1.0067513.ris

Full Text

  MECHANISMS OF DIVERGENCE IN THREESPINE STICKLEBACK (GASTEROSTEUS ACULEATUS)   by  Kerry Brian Marchinko   B.Sc., The University of Alberta, 2000 M.Sc., The University of Alberta, 2002     A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF   DOCTOR OF PHILOSOPHY  in  The Faculty of Graduate Studies  (Zoology)     THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)    August 2009      © Kerry Brian Marchinko, 2009   ii ABSTRACT  One of the most impressive examples of parallel divergence in nature is the repeated loss of armour in freshwater threespine stickleback, yet we have little evidence for the mechanisms driving this evolutionary change.  I tested two ecological factors hypothesized to drive armour divergence between marine and freshwater populations and examined the mechanisms maintaining armour variation within a polymorphic population of stickleback.  I demonstrate how differences in salinity between marine and freshwater habitats might indirectly drive the reduction of lateral plate number in freshwater.  Offspring with a reduced number of armour plates grew much faster than offspring with many plates when raised in fresh water, but not salt water.  Larger size is positively associated with two measures of fitness, survival and reproductive potential, suggesting that the parallel loss of plates in fresh water arose through a correlated response to selection for faster growth during plate development.  I show that predation by freshwater aquatic insects favours armour reduction and larger body size.  Using an F2 intercross between divergent marine and freshwater populations, I measured selection on body size, spine traits and Ectodysplasin, a gene linked to quantitative trait loci for plate number, spine length, and body shape.  Insect predation, when compared to experimental controls, resulted in increased juvenile size, shorter dorsal spine and pelvic girdle length after accounting for size, and favoured the low armor Ectodysplasin allele.  My work on an Ectodysplasin polymorphism in a freshwater population revealed a rare example of a stable polymorphism with heterozygote disadvantage.  Genetic analysis revealed that selection rather than population subdivision or assortative mating accounts for the observed heterozygote deficiency.  Differences in carbon isotope signature hint that niche separation exists between Ectodysplasin homozygotes, providing an ecological mechanism (frequency-dependent selection) for the maintenance of polymorphism despite heterozygous disadvantage.  Overall, my thesis contributes to a growing body of empirical work revealing the powerful effects of divergent selection and highlights the benefits of integrating genetic and ecological approaches for understanding the mechanisms maintaining variation and promoting diversification in wild populations.  iii TABLE OF CONTENTS ABSTRACT...................................................................................................................ii TABLE OF CONTENTS .............................................................................................iii LIST OF TABLES .......................................................................................................vi LIST OF FIGURES ....................................................................................................vii ACKNOWLEDGEMENTS .......................................................................................viii CO-AUTHORSHIP STATEMENT............................................................................. ix CHAPTER 1. GENERAL INTRODUCTION ............................................................. 1 INTRODUCTION......................................................................................................... 1 ION HYPOTHESIS .............................................................................................................. 3 PREDATION HYPOTHESIS .................................................................................................. 6 REFERENCES............................................................................................................ 12 CHAPTER 2. PARALLEL EVOLUTION BY CORRELATED RESPONSE: LATERAL PLATE REDUCTION IN THREESPINE STICKLEBACK................. 19 INTRODUCTION....................................................................................................... 19 METHODS.................................................................................................................. 22 ANALYSES ..................................................................................................................... 26 RESULTS.................................................................................................................... 28 HATCHING SUCCESS....................................................................................................... 28 JUVENILE GROWTH RATE............................................................................................... 29 DISCUSSION.............................................................................................................. 29 CORRELATED RESPONSE TO SELECTION FOR FASTER GROWTH RATE .............................. 31 ECOLOGICAL AND EVOLUTIONARY MECHANISMS OF LATERAL PLATE DIVERGENCE........ 32 REFERENCES............................................................................................................ 37 CHAPTER 3. PREDATION’S ROLE IN REPEATED PHENOTYPIC AND GENETIC DIVERGENCE OF ARMOR IN THREESPINE STICKLEBACK ...... 41 INTRODUCTION....................................................................................................... 41 METHODS.................................................................................................................. 46 FISH POPULATIONS ........................................................................................................ 46 EXPERIMENTAL PROTOCOL ............................................................................................ 47  iv GENETIC ANALYSIS OF THE ECTODYSPLASIN GENE .......................................................... 51 STATISTICAL ANALYSES ................................................................................................ 52 RESULTS.................................................................................................................... 55 SELECTION ON ARMOR TRAITS ....................................................................................... 55 SELECTION ON THE ECTODYSPLASIN GENE....................................................................... 58 DISCUSSION.............................................................................................................. 60 DIVERGENT SELECTION BY PREDATORS: IMPLICATIONS FOR STICKLEBACK SPECIATION .. 64 REFERENCES............................................................................................................ 67 CHAPTER 4.  POLYMORPHISM MAINTAINED BY NATURAL SELECTION DESPITE HETEROZYGOTE DISADVANTAGE AT A MAJOR GENE.............. 75 MAIN BODY............................................................................................................... 75 METHODS.................................................................................................................. 84 COLLECTIONS ................................................................................................................ 84 STABLE ISOTOPE ANALYSIS............................................................................................ 85 GENOTYPING ................................................................................................................. 85 REPRODUCTIVE FITNESS MODEL .................................................................................... 86 RELATIVE FITNESS AND SELECTION COEFFICIENTS ......................................................... 88 STATISTICAL ANALYSES ................................................................................................ 88 REFERENCES............................................................................................................ 90 CHAPTER 5. GENERAL CONCLUSIONS.............................................................. 94 REFERENCES............................................................................................................ 99 APPENDICES........................................................................................................... 100 APPENDIX 1. LOCATIONS OF SAMPLING SITES ASSOCIATED WITH CHAPTER THREE, FIGURE 3.1............................................................................................................................... 100 APPENDIX 2.  SUPPLEMENTARY METHODS FROM CHAPTER THREE. ................................ 101 APPENDIX 3.  FAMILY SAMPLE SIZES FROM EACH F2 FAMILY IN THE PAXTON AND MCKAY LINES FOUND IN CHAPTER THREE. ................................................................................. 103 APPENDIX 4. MEANS, STANDARD DEVIATIONS AND SELECTION DIFFERENTIALS FOR STANDARD LENGTH AND ALL SIZE-ADJUSTED ARMOR TRAITS FROM EACH F2 FAMILY IN THE PAXTON AND MCKAY LINES......................................................................................... 104 APPENDIX 5. CHANGE IN FREQUENCY (∆ P) OF THE LOW MORPH ECTODYSPLASIN ALLELE, EDAL, BETWEEN NO-PREDATOR CONTROL AND PREDATOR TREATMENTS IN F2 FAMILIES GENERATED FROM THE PAXTON LINES. ......................................................................... 107 APPENDIX 6. THE ASSOCIATION BETWEEN EDA GENOTYPE AND LATERAL PLATE MORPH PHENOTYPE POOLED ACROSS ALL SAMPLE YEARS.  FISHERS' EXACT TEST P = 2.2*10-16. . 108 APPENDIX 7. LOCUS SPECIFIC TESTS FOR HARDY-WEINBERG EQUILIBRIUM (HWE) AND INBREEDING COEFFICIENTS (FIS) AT SIX NEUTRAL MICROSATELLITE LOCI GENOTYPED IN ADULT STICKLEBACK COLLECTED IN MAY 2006 FROM KENNEDY LAKE.  EACH MOLECULAR  v MARKERS USED IS NOT KNOWN TO BE LINKED TO QUANTITATIVE TRAIT LOCI FOR ANY MORPHOLOGICAL TRAITS.  BOLD VALUES INDICATED P-VALUE REMAINED SIGNIFICANT AFTER A SEQUENTIAL BONFERRONI ADJUSTMENT.......................................................... 109 APPENDIX 8. PAIRWISE TESTS FOR LINKAGE DISEQUILIBRIUM. ...................................... 110 APPENDIX 9.  RESULTS FROM ANOVA TESTING FOR EFFECTS OF LATERAL PLATE MORPH, SEX AND THEIR INTERACTION ON STANDARD LENGTH, A MEASURE OF OVERALL BODY SIZE, FROM RANDOMLY SAMPLED ADULTS TRAPPED IN MAY OF 2004. .................................... 111 APPENDIX 10. RESULTS FROM TWO-FACTOR ANOVA TESTING THE EFFECTS OF LATERAL PLATE MORPHOTYPE AND SEX ON THE RATIO OF STABLE ISOTOPES δ13C AND δ15N. ......... 112 APPENDIX 11. ANIMAL CARE CERTIFICATES. ................................................................ 113  vi LIST OF TABLES TABLE 3.1. Standardized selection differentials and significance tests for standard length and size adjusted spine lengths ........................................................................... 57 TABLE 3.2. Mixed-model ANCOVA results testing the effects of genotype at the Eda locus, family as a random factor, and their interaction on standard length...................... 61   vii LIST OF FIGURES FIGURE 1.1. Illustration of the major bony armour traits............................................... 3 FIGURE 1.2. Sampling populations in British Columbia.............................................. 11 FIGURE 2.1. The three lateral plate morphotypes of Gasterosteus aculeatus ............... 20 FIGURE 2.2. Expected patterns of fitness among lateral plate morphs ......................... 27 FIGURE 2.3. Hatching success and growth rate of reduced (low and partial) and completely plated families of Gasterosteus aculeatus.................................................... 30 FIGURE 3.1. The positive association between spine length and the number of predatory fish species.................................................................................................................... 42 FIGURE 3.2. Illustrations of the threespine stickleback................................................ 47 FIGURE 3.3. Relative survival function estimates........................................................ 56 FIGURE 3.4. Proportion of F2 juvenile stickleback missing the pelvic girdle ............... 58 FIGURE 3.5. Allele and genotype frequency at the Ectodysplasin locus....................... 59 FIGURE 4.1. Representatives of the low and complete lateral plate morphs................. 76 FIGURE 4.2. The frequencies of lateral plate morphs and Eda genotypes .................... 77 FIGURE 4.3. Estimated population structure of Kennedy Lake threespine stickleback. 78 FIGURE 4.4. Estimated relative fitness components of genotypes at Ectodysplasin ..... 79 FIGURE 4.5. Isotopic representation of the long-term diet of low morph and complete morph stickleback ......................................................................................................... 82  viii ACKNOWLEDGEMENTS  I would like to thank my Supervisor, Dolph Schluter, for the interest and enthusiasm he's shown for my work in the past few years.  None of this would have been possible without his help. Thanks also to my committee, Mike Whitlock, Jon Shurin, and Trish Schulte, for their advice and patience.  I'm deeply grateful to Matt Arnegard for unwavering help on long, cold days and nights sampling at the ponds, at Kennedy Lake and working in the lab.  No one could have substituted for you.  Sean Rogers was a great source of support academically and personally over the last few years.  Thanks to Pat Tamkee for countless hours of help in the genetics lab and for helping with field work at Kennedy Lake (and for letting us stay at your cabin).  Thanks to Tim Vines and Arianne Albert for all their help early when I knew nothing of sticklebacks.  Thanks to Blake Matthews for discussion and help with the stable isotopes analysis.  Thanks to all the grad students in the Schluter lab over the years, especially Rowan Barrett, Jason Weir, Anthony Waldron, Mirjam Barrueto and Jay Clarke for discussions and advice.  Much of this work could not have been done with out the help of many undergraduate volunteers helping feed and maintain my fish stocks, thank you all kindly.  Lastly thank you Debbie Bryant.  Your strength and resolve have empowered me to be the best I can.    ix CO-AUTHORSHIP STATEMENT  Chapter 2 was a collaboration between myself and my research supervisor Dolph Schluter.  Dolph aided in experimental design, data analysis and helped edit the manuscript.  Data was collected, analysed and the manuscript written by myself.  Chapter 4 was a collaboration between myself, Blake Matthews, Sean Rogers and Dolph Schluter.  Blake carried out the mass-spec analysis of the stable isotope samples and helped with field collections.  Sean helped choose appropriate microsatellite markers, aided in the population genetic analyses and commented on the manuscript.  Dolph aided with the development of the selection model, data analysis and manuscript editing.  I was responsible for the experimental design, collection of samples, analysis of all data, molecular genetics bench work, and writing the manuscript.  1 CHAPTER 1. GENERAL INTRODUCTION INTRODUCTION  The ecological theory of adaptive radiation states that divergent selection forms the basis of evolutionary diversification and ultimately speciation (Schluter 2000).  To properly identify ecology’s role in diversification, we must begin by demonstrating that selection is divergent between environments and that ecological mechanisms select against hybrids in natural environments (Schluter 2000).  Although populations often diverge in traits associated with habitat differences, the genetic and ecological mechanisms underlying phenotypic divergence are often not well understood in most cases.  In this thesis, I experimentally test two of the ecological factors hypothesized to govern armour divergence between marine and freshwater populations of threespine stickleback, Gasterosteus aculeatus L, as well as examine the factors maintaining polymorphism with heterozygote disadvantage at a major armour gene.  The following chapters are independent manuscripts containing more detailed background than will be covered in this brief introductory chapter.  The extensive data on natural history and the availability of genomic resources have made threespine stickleback a model system in which to ecology’s role in adaptive diversification (Gibson 2005).  In British Columbia, marine stickleback colonized freshwater lakes and streams recently, roughly 12,000 years ago.  Freshwater populations often exhibit parallel divergence in amour morphology from marine ancestors, largely in the reduction of bony armour plates and spines.  The multiple, independent colonization of coastal freshwater populations from marine ancestors is supported by sequence variation in mitochondrial DNA and microsatellite loci (Colosimo et al. 2005; McKinnon 2 et al. 2004; Taylor and McPhail 1999; Taylor and McPhail 2000; Thompson et al. 1997). Marine populations are often heavily armored with a full row of bony plates running down both sides of the body (complete morph) and long dorsal and pelvic spines (Fig. 1.1).  Freshwater populations generally have few lateral plates clustered around the pectoral fin (low morph) and short dorsal and pelvic spines (Fig. 1.1).  Recently, two major genes were discovered to govern most of the phenotypic variance in pelvic spine length and lateral plate number between marine and freshwater populations.  Shapiro et al. (2004) revealed that regulatory changes to the Pitx1 gene on chromosome 7 are responsible for the majority of phenotypic variance (up to 65%) in the length of the pelvic spine and girdle.  Colosimo et al. (2005) identified that the Ectodysplasin (Eda) gene is the major locus responsible for roughly 75% of the variance in lateral plate number (Colosimo et al. 2004).  Interestingly, most freshwater populations examined — 14/15 populations from the Eastern and Western Pacific and Atlantic — share many of same nucleotide changes within and near the Eda gene.  Colosimo et al. (2005) conclude that the parallel reduction of lateral plate number in freshwater most likely occurred through repeated selection of an ancient low lateral plate Eda allele most introduced to freshwater populations from the sea.  Molecular evidence suggests that armour divergence between marine and freshwater habitats is adaptive (Cano et al. 2006; Raeymaekers et al. 2007) and, although similar ecological conditions may have selected for reduction of armour in freshwater threespine stickleback, researchers have had little success explaining the mechanism(s) of armour divergence (Bell 2001; Bertin 1925).  Differences in lateral plate number and spine length between marine and freshwater populations are primarily thought to have 3 arisen from differences in ion concentration (Bertin 1925; Bourgeois et al. 1994; Francis et al. 1986; Giles 1983; Heuts 1947; Klepaker 1995; Klepaker 1996) or differences in predator community and predation rate (Bell 2001; Bergstrom 2002; Reimchen 1980; Reimchen 1994; Reimchen 1995; Reimchen 2000).  Ion Hypothesis  Differences in either ion concentration, specifically salinity and calcium, between marine and freshwater habitats have been suggested as a mechanism driving the divergence stickleback lateral plate number (Bell et al. 1993; Campbell 1985; Giles 1983; Heuts 1947).  In accord with one of the fundamental tenets of divergent natural selection (Schluter 2000), this hypothesis posits a trade-off in fitness advantages between fresh water and salt water environments:  freshwater fish with few lateral plates (low morph) will have higher fitness than fish with many lateral plates (complete morph) in fresh 4 water, whereas the complete morph will have higher fitness than the low morph in sea water.  Although both salinity and calcium variants of this hypothesis predict a trade-off in fitness advantages between environments, the mechanism driving phenotypic divergence in lateral plate number differs between them.  If the ability of stickleback to maintain ionic homeostasis at different salinities is associated with lateral plate number, then divergent selection could result in the general pattern of the complete morph inhabiting the sea and the low morph inhabiting freshwater lakes.  The role of salinity in divergent selection on lateral plate number between marine and freshwater populations was initially supported by Heuts’(1947) study on European threespine stickleback.  His work indicated that the complete morph survived better than the low morph in more saline water, whereas the low morph survived better than the complete morph in fresh water.  Schaarschmidt et al. (1999) revealed roughly 75% mortality for the complete morph compared to only 5% mortality for the low morph when exposed to fresh water at 4 °C.  Schaarschmidt et al. (1999) suggest this pattern is based upon differences in the regulation of hormones responsible for water and ion permeability.  However, the conclusions of both Heuts (1947) and Schaarschmidt et al. (1999) may be confounded by their collection methods: both studies collected low morphs from freshwater populations and complete morphs from brackish water populations.  Thus adaptation to local salinity conditions, rather than lateral plate number, may explain the differences in salinity tolerance observed between plate morphs.  Moreover, geographical trends in lateral plate phenotype do not always support a strong association between lateral plate number and water salinity (Baumgartner and Bell 1984; Hagen and Gilbertson 1972) and it remains 5 unclear whether differences in salinity between marine and fresh water habitats results in selection on lateral plate number (Bell 2001).  Lower levels of calcium in most freshwater habitats compared to marine and brackish habitats should favour the individuals with fewer lateral plates because of an increased cost of building calcium carbonate skeletal components at low calcium levels (Giles 1983).  Under this hypothesis of calcium limitation, possessing fewer lateral plates in fresh water of low calcium concentration is advantageous, but not necessarily disadvantageous at higher levels of calcium.  Support for this hypothesis is found primarily in the positive correlation between ionic concentration and skeletal armour including lateral plates, dorsal and pelvic spines, and the pelvic girdle among freshwater populations (Bell et al. 1993; Bourgeois et al. 1994; Francis et al. 1986; Heuts 1944).  At this point it appears no experimental work has directly tested calcium limitation as a mechanism for the reduction of armour in freshwater stickleback.  Recent molecular evidence suggests that a revisit to the role of salinity in lateral plate reduction is warranted.  Colosimo et al. (2005) detailed that Eda is closely linked to a gene possibly associated with salt secretion, Gjb1.  Interestingly, the overall selective advantage of low morphs in fresh water may arise, in part, via selection on differences in salinity tolerance that are correlated with lateral plate phenotypes.  In Chapter 2 I test if differences in salinity between marine to freshwater habitats account for the reduction of lateral plate armour in fresh water populations of threespine stickleback.  Chapter 2 uses a balanced experimental design to test if estimates for hatching success and juvenile growth rate among lateral plate morphs agree with the geographic patterns in nature: higher performance of individuals with fewer lateral plates than those with many lateral 6 plates when raised in fresh water, but lower performance of fish with fewer lateral plates than those with many lateral plates when raised in salt water.  To minimize the potential artifact of local adaptation salinity among morphs coming from different populations (Heuts 1947), I used morphs from two populations polymorphic for lateral plates (Fig. 1.2, sites 2 & 5). Predation Hypothesis  Divergent selection is thought to give rise to differences in stickleback armour through a trade-off in the mortality rate associated with differences in the predatory species present between marine and freshwater environments (Reimchen 1980; Reimchen 1994).  This hypothesis is rooted in the correlation between predator community composition or seasonality of predator activity and the level of armour found in natural populations.  In general, stickleback from marine and lake populations with many predaceous fish and bird species have long dorsal and pelvic spines (Reimchen 1994). Under these conditions lake populations are often completely plated, while nearly all marine populations consist of the complete plate morph only.  By contrast, in lakes with few predatory fish and bird species, and where predaceous insects are abundant, stickleback often have short spines and few lateral plates (Bell et al. 1993; Reimchen 1980).  Evidence for a trade-off in mortality between habitats differing in predator species was gained through experiments showing that individuals with longer dorsal and pelvic spines suffered lower mortality rates than fish with shorter spines when preyed upon by gape limited fish and birds (Reimchen 1988).  Additionally, lateral amour plates increase the chance of survival of stickleback after attack by toothed fish predators and, along with the pelvic girdle, act as buttress to the dorsal and pelvic spines when erect 7 (Reimchen 1983; Reimchen 1994; Reimchen 2000).  In freshwater habitats where non- gape limited aquatic insects are important sources of mortality, longer spines are hypothesized to increase mortality relative to short spines because they provide larger physical structures for insects to grasp while preying upon juvenile stickleback (Reimchen 1980).  Furthermore, Chapter 2 revealed that individuals with more lateral plates grew at a slower rate than fish with few lateral plates.  Faster growth at the juvenile stage may have severe consequences on juvenile predation rates in fresh water, as predaceous insects feed most frequently on the smallest individuals available (Foster et al. 1988).  Thus juveniles of the low morph may exhibit lower mortality than those of the complete morph when exposed to predaceous insects only found in fresh water.  Experimental evidence for differential susceptibility to predators is unequal with respect to the predator species studied.  Studies demonstrate the effectiveness of long spines and many lateral plates for reducing mortality due to fish or bird predation (Hoogland et al. 1956; Reimchen 1994; Reimchen 2000).  However demonstrations that shorter spines and fewer plates reduce mortality due to aquatic insects are often indirect. Reimchen and Nosil (Reimchen and Nosil 2002; Reimchen and Nosil 2004) revealed a decrease in mean spine number in the summer when predaceous insects are most active, but an increase in mean spine number in the winter when gape limited piscivores were most active.  Using benthic-limnetic stickleback species pairs, Vamosi (2002) showed that mortality due to insect predation was greatest in unarmored benthic compared to lightly armored limnetic sticklebacks.  Similar to Reimchen and Nosil (Reimchen and Nosil 2002; Reimchen and Nosil 2004), Vamosi’s (2002) results also imply a selective advantage to individuals with reduced armor when insect predation is high.  However, 8 Vamosi (2002) is unable to attribute differences in mortality to armor alone because additional traits and behaviors differentiate the wild caught species under investigation. Indeed, a selective disadvantage to individuals with more armour may be unnecessary to explain the divergence between marine and freshwater populations and an alternative hypothesis combines the effects of predation with aspects of the ion hypothesis.  In this scenario predators impose directional selection, only favouring more armour under greater levels of predation.  Amour is selected against in freshwater habitats with depauperate predator fauna because low calcium concentrations yield a selective disadvantage to individuals required to allocate more energy into building bony armour at a cost to reproductive or somatic tissues (Bell et al. 1993).  Chapter 3 tests the hypothesis that predation by aquatic insects has resulted in selection for reduced stickleback armour in fresh water using F2 hybrids generated from a cross between freshwater and marine populations divergent in armour (sites 1- 4 in Fig. 1.2).  In addition to differences in lateral plate number (low plated fresh water vs. completely plated marine), the parental populations chosen to generate F2 families were divergent in dorsal and pelvic spine length, and pelvic girdle length.  The two parental populations represent the extremes of the range of phenotypes observed in natural populations.  Two additional benefits are gained by measuring selection on hybrids rather than wild individuals differing in armour.  By crossing wild caught parental populations and raising them to the F2 generation, trait variance is increased beyond that available in a single wild population (Falconer and Mackay 1996), resulting in an increased ability to detect selection (Endler 1986).  Additionally, one episode of recombination in the F2 9 generation will reduce trait correlations arising from linkage disequilibrium between distant genomic locations, increasing the ability to isolate the effects of predation on armor from those of other traits differing between populations that may also influence predation rate (Bell et al. 2007; Huntingford et al. 1994).  However, previous work has shown that several armor traits, including the length of dorsal and pelvic spines, as well as a variety of shape traits map to the same genomic region near Eda, the major gene underlying variation in lateral plate number (Albert et al. 2008; Shapiro et al. 2004; S. M. Rogers, unpublished manuscript).  This has the advantage in that the effects of armor on mortality via predation will be more apparent if several traits covary together.  The disadvantage is that further work will be necessary to distinguish which of the armor traits in this region of the genome are the most important to survival.  Not all freshwater populations in British Columbia exhibit a complete reduction in armour.  A few freshwater populations of threespine stickleback remain polymorphic for lateral plate armour, presenting a rare opportunity to investigate the factors maintaining genetic variation for this ecologically important trait.  The maintenance of genetic variation within populations appears paradoxical given that directional and stabilizing selection are common and expected to reduce variation within populations (Fisher 1930; Kingsolver et al. 2001).  Indeed, a surprisingly large amount of variation is often found to exist within populations (Mitchell-Olds et al. 2007).  Understanding the mechanisms that maintain variation in ecologically important traits is central to our understanding of how populations persist, what generates biodiversity, and how populations may respond to rapid environmental change.  Such variation may persist because of balancing selection, recurrent mutation, or gene flow 10 from nearby populations (Mitchell-Olds et al. 2007).  However, few natural systems exist in which to undertake empirical tests of even the most basic question of how variation is maintained within wild populations and species.  In Chapter 4 I examine the possible factors maintaining the polymorphism at the major lateral plate gene, Eda, within a wild population of threespine stickleback (Fig. 1.2, site 5).  This polymorphism is unusual, however, in I discovered a deficiency of Eda heterozygotes in sampled adults.  The maintenance of polymorphism with heterozygote deficiency might be due to one of three processes, population subdivision, assortative mating, or frequency dependent selection.  Population subdivision could maintain the appearance of Eda polymorphisms within a single lake if two reproductively isolated groups were coming into secondary contact (Castric et al. 2002).  To test for the presence of population subdivision, the population genetics of six neutral loci were examined for clustering and genetic divergence.  Assortative mating in the absence of population structure (Barreto and McCartney 2008) may account for polymorphism maintained in the face of heterozygote deficiency if stickleback within the lake mate according to Eda genotype.  To test for assortative mating I genotyped the fertilized eggs of wild caught clutches at Eda to determine if genotype frequencies agreed with those expected from random mating.  Lastly, frequency-dependent selection may maintain a stable polymorphism with heterozygote disadvantage, if it favours each Eda homozygote when rare (Wilson and Turelli 1986).  I explored one possible ecological factor, niche differentiation (Ayala and Campbell 1974), that could provide such a stabilizing influence by comparing the ratio stable isotopes of carbon and nitrogen in wild caught 11 adults as indicators of the ultimate source of dietary carbon and trophic position, respectively (Post 2002). 12  REFERENCES Albert, A. Y. K., S. Sawaya, T. H. Vines, A. K. Knecht, C. T. Miller, B. R. Summers, S. Balabhadra, D. M. Kingsley, and D. Schluter.  2008.  The genetics of adaptive shape shift in stickleback: pleiotropy and effect size. Evolution 62:76-85. Ayala, F. J., and C. A. Campbell.  1974.  Frequency-dependent selection. Annual Review of Ecology and Systematics 5:115-138. Barreto, F. S., and M. A. McCartney.  2008.  Extraordinary AFLP fingerprint similarity despite strong assortative mating between reef fish color morphospecies. Evolution 62:226-233. Baumgartner, J. V., and M. A. Bell.  1984.  Lateral plate morph variation in California populations of the threespine stickleback, Gasterosteus aculeatus. Evolution 38:665-674. Bell, A. M., T. Backstrom, F. A. Huntingford, T. G. Pottinger, and S. Winberg.  2007. Variable neuroendocrine responses to ecologically-relevant challenges in sticklebacks. Physiology & Behavior 91:15-25. Bell, M. A.  2001.  Lateral plate evolution in the threespine stickleback:  Getting nowhere fast. Genetica 112:445-461. Bell, M. A., G. Orti, J. A. Walker, and J. P. Koenings.  1993.  Evolution of pelvic reduction in threespine stickleback fish:  A test of competing hypotheses. Evolution 47:906-914. Bergstrom, C. A.  2002.  Fast-start swimming performance and reduction in lateral plate number in threespine stickleback. Canadian Journal of Zoology 80:207-213. 13 Bertin, L.  1925.  Recherches bionomiques et systematiques sur les Epinoches (Gasterosteides). Annals of the Institute of Oceanography, Monaco 2:1-204. 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 72:1497-1509. Campbell, R. N.  1985.  Morphological variation in the three-spined stickleback (Gaseterosteus aculeatus) in Scotland. Behaviour 93:161-168. Cano, J. M., C. Matsuba, H. Makinen, and J. Merila.  2006.  The utility of QTL-linked markers to detect selective sweeps in natural populations — a case study of the EDA gene and a linked marker in threespine stickleback. Molecular Ecology 15:4613-4621. Castric, V., L. Bernatchez, K. Belkhir, and F. Bonhomme.  2002.  Heterozygote deficiencies in small lacustrine populations of brook charr Salvelinus Fontinalis Mitchill (Pisces, Salmonidae): a test of alternative hypotheses. Heredity 89:27-35. Colosimo, P. F., K. E. Hosemann, S. Balabhadra, V. J. Guadalupe, M. Dickson, J. Grimwood, J. Schmutz, R. M. Myers, D. Schluter, and D. M. Kingsley.  2005. Widespread parallel evolution in sticklebacks by repeated fixation of Ectodysplasin alleles. Science 307:1928-1933. Colosimo, P. F., C. L. Peichel, K. Nereng, B. K. Blackman, M. D. Shapiro, D. Schluter, and D. M. Kingsley.  2004.  The genetic architecture of parallel armor plate reduction in threespine sticklebacks. PLoS Biology 2:1-7. Endler, J.  1986.  Natural Selection in the Wild. Princeton University Press, Princeton. 14 Falconer, D. S., and T. F. C. Mackay.  1996.  Introduction to Quantitative Genetics. Pearson Prentice Hall, Harlow, England. Fisher, R. A.  1930.  The Genetical Theory of Natural Selection. Oxford University Press, Oxford. Foster, S. A., V. B. Garcia, and M. Y. Town.  1988.  Cannibalism as the cause of an ontogenetic niche shift in habitat use by fry of the threespine stickleback. Oecologia 74:577-585. Francis, R. C., J. V. Baumgartner, A. C. Havens, and M. A. Bell.  1986.  Historical and ecological sources of variation among lake populations of threespine sticklebacks, Gasterosteus aculeatus, near Cook Inlet, Alaska. Canadian Journal of Zoology 64:2257-2265. Gibson, G.  2005.  The Synthesis and Evolution of a Supermodel. Science 307:1890- 1891. Giles, N.  1983.  The possible role of environmental calcium levels during the evolution of phenotypic diversity in Outer Hebridean populations of the Three-spined stickleback, Gasterosteus aculeatus. Journal of Zoology, London 199:535-544. 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. Heuts, M. J.  1944.  Calcium ions and the geographical distribution of Gasterosteus aculeatus L. Natuurwetenschappelijk Tijdschrift 26:10-14. Heuts, M. J.  1947.  Experimental studies on adaptive evolution in Gasterosteus aculeatus L. Evolution 1:89-102. 15 Hoogland, R., D. Morris, and N. Tinbergen.  1956.  The spines of sticklebacks (Gasterosteus and Pygosteus) as means of defence against predators (Perca and Esox). Behaviour 10:205-236. Huntingford, F. A., P. J. Wright, and J. F. Tierney.  1994.  Adaptive variation in antipredator behaviour in threespine stickleback in M. A. Bell and S. A. Foster, eds. The Evolutionary Biology of the Threespine Stickleback. Oxford University Press, Oxford, UK. Kingsolver, J. G., H. E. Hoekstra, J. M. Hoekstra, D. Berrigan, S. N. Vignieri, C. E. Hill, A. Hoang, P. Gibert, and P. Beerli.  2001.  The strength of phenotypic selection in natural populations. American Naturalist 157:245-261. Klepaker, T.  1995.  Postglacial evolution in lateral plate morphs in Norwegian fresh- water populations of the threespine stickleback (Gasterosteus aculeatus). Canadian Journal of Zoology 73:898-906. Klepaker, T.  1996.  Lateral plate polymorphism in marine and estuarine populations of the threespine stickleback (Gasterosteus aculeatus) along the coast of Norway. Copeia:832-838. McKinnon, J. S., S. Mori, B. K. Blackman, L. David, D. M. Kingsley, L. Jamieson, J. Chou, and D. Schluter.  2004.  Evidence for ecology's role in speciation. Nature 429:294-298. Mitchell-Olds, T., J. H. Willis, and D. B. Goldstein.  2007.  Which evolutionary processes influence natural genetic variation for phenotypic traits? Nature Reviews Genetics 8:845-856. 16 Post, D. M.  2002.  Using stable isotopes to estimate trophic position:  Models, methods, and assumptions. Ecology 83:703-718. Raeymaekers, J. A. M., J. K. J. Van Houdt, M. H. D. Larmuseau, S. Geldof, and F. A. M. Volckaert.  2007.  Divergent selection as revealed by P-ST and QTL-based F-ST in three-spined stickleback (Gasterosteus aculeatus) populations along a coastal- inland gradient. Molecular Ecology 16:891-905. Reimchen, T. E.  1980.  Spine deficiency and polymorphism in a population of Gasterosteus aculeatus:  An adaptation to predators? Canadian Journal of Zoology 58:1232-1244. Reimchen, T. E.  1983.  Structural relationships between spines and lateral plates in threespine stickleback (Gasterosteus aculeatus). Evolution 37:931-946. Reimchen, T. E.  1988.  Inefficient predators and prey injuries in a population of giant stickleback. Canadian Journal Of Zoology 66:2036-2044. Reimchen, T. E.  1994.  Predators and morphological evolution in threespine stickleback in M. A. Bell and S. A. Foster, eds. The Evolutionary Biology of the Threespine Stickleback. Oxford University Press, Oxford. Reimchen, T. E.  1995.  Predator-induced cyclical changes in lateral plate frequencies of Gasterosteus. Behaviour 132:1079-1094. Reimchen, T. E.  2000.  Predator handling failures of lateral plate morphs in Gasterosteus aculeatus:  Functional implications for the ancestral plate condition. Behaviour 137:1081-1096. Reimchen, T. E., and P. Nosil.  2002.  Temporal variation in divergent selection on spine number in threespine stickleback. Evolution 56:2472-2483. 17 Reimchen, T. E., and P. Nosil.  2004.  Variable predation regimes predict the evolution of sexual dimorphism in a population of threespine stickleback. Evolution 58:1274- 1281. Schaarschmidt, T., E. Meyer, and K. Jurss.  1999.  A comparison of transport-related gill enzyme activities and tissue-specific free amino acid concentrations of Baltic Sea (brackish water) and freshwater threespine sticklebacks, Gasterosteus aculeatus, after salinity and temperature acclimation. Marine Biology 135:689-697. Schluter, D.  2000.  The Ecology of Adaptive Radiation. Oxford University Press, Oxford. Shapiro, M. D., M. E. Marks, C. L. Peichel, B. K. Blackman, K. Nereng, B. Jonsson, D. Schluter, and D. M. Kingsley.  2004.  Genetic and developmental basis of evolutionary pelvic reduction in threespine sticklebacks. Nature 428:717-723. Taylor, E. B., and J. D. McPhail.  1999.  Evolutionary history of an adaptive radiation in species pairs of threespine sticklebacks (Gasterosteus):  Insights from mitochondrial DNA. Biological Journal of the Linnean Society 66:271-291. Taylor, E. B., and J. D. McPhail.  2000.  Historical contingency and determinism interact to prime speciation in sticklebacks. Proceedings of the Royal Society of London B 271:2375-2384. 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. Vamosi, S. M.  2002.  Predation sharpens the adaptive peaks:  Survival trade-offs in sympatric sticklebacks. Annals Zoologica Fennici 39:237-248. 18 Wilson, D. S., and M. Turelli.  1986.  Stable Underdominance And The Evolutionary Invasion Of Empty Niches. American Naturalist 127:835-850.   19 CHAPTER 2. PARALLEL EVOLUTION BY CORRELATED RESPONSE: LATERAL PLATE REDUCTION IN THREESPINE STICKLEBACK* INTRODUCTION  One of the most impressive examples of parallel evolution in nature is the rapid and repeated loss of lateral plate armour in freshwater populations of threespine stickleback, Gasterosteus aculeatus (Colosimo et al. 2005).  Most freshwater populations from North America and Western Europe are made up of individuals with relatively few lateral plates clustered near the pectoral fins (< 10 per side = low morph; Fig. 2.1), whereas individuals from marine populations have many lateral plates extending from the head to the tip of the caudal peduncle (> 30 per side = complete morph; Fig. 2.1; Bell 2001; Wooton 1984).  A few populations are polymorphic for lateral plate armor, possessing low, complete and partial (11 to 29 lateral plates) plate morphs.  Recent molecular evidence (Colosimo et al. 2005; Colosimo et al. 2004) indicates that the recurrent transition from completely plated marine populations to low plated freshwater populations is largely controlled by the effects of a single major gene, Ectodysplasin (Eda).  The authors suggested that parallel reduction of lateral plate number in fresh water has occurred through repeated selection for low plate morph Eda alleles introduced by marine colonists, very few of whom were heterozygous for alleles at the Eda locus.    *A version of this chapter has been published. Marchinko, K. B., and D. Schluter. 2007. Parallel evolution by correlated response: Lateral plate reduction in threespine stickleback. Evolution 61:1084-1090.   20   Surprisingly, although the molecular mechanisms of lateral plate reduction are becoming established, the ecological mechanisms driving armour evolution remain uncertain.  Determining the mechanism responsible has been made difficult by the positive correlation of multiple ecological factors with differences in lateral plate number among populations (reviewed in Bell 2001).  One of the oldest and more interesting suggestions was that lateral plate number is correlated with salinity tolerance.  Heuts  21 (1947) found that individuals with greater lateral plate number survive longer and hatch more successfully in salt water, whereas those with reduced lateral plate number survive longer and hatch more successfully in fresh water.  Heuts (1947) suggested that differences in salinity tolerance between low and complete lateral plate morphs might explain the pattern of lateral plate divergence between marine and freshwater habitats. Unfortunately Heuts (1947) collected his low morphs from freshwater populations and his complete morphs from brackish water populations introducing a confounding factor: perhaps other adaptations to local salinity conditions, not lateral plates, explain the differences he observed between morphs.  Additional criticism for Heuts’ hypothesis arises from its failure to account for the presence of completely plated freshwater populations in Eastern Europe and Eastern North America (Banbura 1994; Bell 2001; Hagen and Moodie 1982) and the presence of complete morph populations in lower salinity habitats upstream of polymorphic populations in Northern California (Baumgartner and Bell 1984).  Recent physiological and molecular evidence warrants a revisit to the role of salinity in the evolution of lateral plate number in threespine stickleback.  Schaarschmidt et al. (1999) revealed dramatic differences in mortality between low and complete morphs when exposed to fresh water at 4 °C:  up to 75% mortality for completely plated morphs compared to only 5%  mortality for low morphs.  They suggest this pattern is based upon differences in the regulation of prolactin, a hormone responsible for water and ion permeability in freshwater fishes.  Additionally, Colosimo et al. (2005) revealed that Eda is closely linked to three other genes including one possibly associated with salt  22 secretion, Gjb1.  Thus, overall selective advantage of low morphs in fresh water may arise, in part, via selection on differences in salinity tolerance that are correlated with lateral plate phenotypes.  In this study, we examine the hypothesis that selection arising from changes in salinity led to the divergence in lateral plate number between marine and freshwater habitats.  Using a balanced experimental design, we raised reduced (low and partial) and complete morphs in the lab in either fresh or salt water to test if estimates for hatching success and juvenile growth rate among morphs agree with the geographic patterns in nature:  higher performance of reduced morphs than complete morphs in fresh water, but lower performance of reduced morphs than complete morphs in salt water.  To minimize the potential differences in adaptation to local salinity among morphs coming from different populations (Heuts 1947), we chose two populations possessing all three lateral plate morphs and only tested for differences in performance between morphs from the same population. METHODS Sample Populations, Fertilization And Experimental Rearing  Threespine stickleback were collected in southwestern British Columbia from two populations with a reliable frequency of all three lateral plate morphs: one marine population from Oyster Lagoon, Pender Harbour and one freshwater population from Kennedy Lake, Vancouver Island.  Lateral plate morphs within each population differed very little in the external traits (standard length, body depth, spine length; Saimoto 1993; Marchinko unpublished data) known to differentiate anadromous populations from  23 freshwater populations (McPhail 1994).  Thus, completely plated individuals from Kennedy Lake were not anadromous migrants and low plate morphs in Oyster lagoon were not accidental lake or stream migrants.  Using artificial fertilization, we made 12 low morph families and 14 complete morph families from the freshwater population, and 12 partial morph and 10 complete morph families from the marine population by crossing males and females of the same morph.  Because the frequency of low morphs in Oyster lagoon was approximately 0.1%, partial morphs were used in place of low morphs when making crosses from the marine population.  Partial-by-partial crosses will generate offspring with a range of lateral plate morphotypes.  The ratio of morphs should be approximately 3:1, reduced:complete morph respectively (Colosimo et al. 2004).  Our use of partial parents may obscure the comparison between morphs from the marine population, but in a conservative manner as 25% of each family will likely be completely plated as adults.  Thus, differences in performance between reduced and complete morph families may be even larger than if only low and complete marine morphs were compared.  Freshwater crosses will not be affected.  Crossing was accomplished by first stripping a female’s eggs into a single Petri dish, counting the eggs and then splitting the clutch in half.  One half of the eggs were placed in a Petri dish containing fresh water (0 ppt) and the other half were placed into artificial salt water (30 ppt; Instant Ocean synthetic seasalt, Aquarium Systems Inc., Mentor, OH, USA) both at a pH of 7.6 (+/- 0.1).  A male of the same morph and population was anaesthetized in clove oil and both testes were removed.  One testis was  24 placed into the freshwater Petri dish, the other in the saltwater Petri dish, and then both were crushed to release sperm.  The half clutch of eggs along with crushed testes were kept in Petri dishes for 20 minutes, then placed into separate plastic egg-cups (pint cup with fine fiberglass mesh lining the bottom) and each submerged in a separate egg-tank (20 L) according to salinity treatment. Methylene blue was added to egg-tanks to reduce fungal growth.  Eggs remained in aerated egg-tanks for eight days, after which they were transferred to one side of a divided 102 L tank of the same salinity and pH, respective to salinity treatment, for the remainder of the experiment.  After the eggs hatched and the larvae dropped into the tanks, egg-cups were removed and the number of unhatched eggs remaining was counted.  Hatching success was measured as the proportion of eggs hatched from each half clutch, one in fresh water and one in salt water.  The surviving larvae were fed live brine shrimp twice per day for four weeks, after which the number of individuals in each half tank was reduced to 15. Thereafter fish were fed 3.5 oz of frozen Daphnia cubes once per day.  After feeding stopped any remaining food was removed by filtration or manual siphoning. Thus each individual fed to satiation once per day.  Measurements of growth rate began five weeks after hatching, when individuals reached about 12 mm from snout to the tip of the caudal peduncle (standard length), the size at which lateral plates begin to form.  Standard length was measured on every individual once every two weeks for ten weeks during the period of lateral plate development.  After an individual reaches about 30 mm standard length the number of plates no longer changes (Bell 1981; Igarashi 1970).  The standard lengths of all  25 individuals from the same family in the same salinity treatment were then averaged, yielding a mean standard length for each family in each treatment.  After ten weeks of measurement, the growth rate of each family in each treatment was calculated as the slope of the ordinary least squares regression of standard length on the number of days since hatching.  Thus, for each family (the unit of replication) we obtained two measures of growth rate, one in fresh water and one in salt water.  The relationship between standard length and the number of days since hatching was linear and highly significant in all replicates.  Unfortunately, we were unable to test for differences in growth rate between low and complete morphs from the freshwater population in the saltwater treatment because of very low replication: only two families from each morph survived long enough to obtain growth rate measurements in salt water.  Not all families survived for the entire ten weeks of the experiment (see Fig. 2.3B for sample sizes) and there were families in which only six weeks, or eight weeks of growth rate measurements were available for a particular treatment.  However, there was no effect of experiment length (6, 8, or 10 weeks) on the linear relationship between standard length and the number of days since hatching (separate two-factor analysis of variance tests were completed on each morph from each habitat; dependent variable = growth rate, factors = experiment length and treatment; P experiment length > 0.16 in all tests). We have included all available data in the analyses reported in the results.  Additionally, not all families maintained 15 individuals in each treatment and any differences in growth rate may be confounded by differences in density and the influence of crowding or food availability.  However, the influence of density on growth rate appears minimal and all  26 data were used for two reasons.  First, the effect of food availability was minimized through feeding protocol; individuals were fed to satiation only once per day and there was no other opportunity for feeding.  Second, using ordinary least squares regression of the number of individuals per family on mean family growth rate, we found no significant relationship between density and growth rate in seven of eight tests (P > 0.14 except for freshwater low morphs raised in 0 ppt where P  = 0.03; separate regressions of mean family growth rate on mean number of individuals in family were completed for each morph from each population in each treatment using all data). Analyses  Our main objective was to determine if the performance of lateral plate morphs differed within each salinity treatment in the direction predicted by geographical patterns of plate morph distributions in nature.  We employed Student’s t-test to test for differences between plate morphs in mean hatching success and juvenile growth rate separately for each salinity treatment and each population (Fig. 2.2 comparisons i and ii). According to Heuts’ (1947) hypothesis that differences in salinity drive patterns of armor loss in fresh water and armor maintenance in the sea, we expect:  1) in fresh water, reduced (low and partial) morphs exhibit greater mean hatching success and growth rate than the complete morph and 2) in salt water, the complete morph exhibits greater mean hatching success and growth rate than reduced morphs (see Fig. 2.2).  Because of split- family design, evaluating the interaction between lateral plate morphotype and salinity was complicated by a lack of independence between treatments (a single family was raised in both treatments).  To test if differences in hatching and growth rate between  27 morphs depended on salinity treatment, we subtracted the values of hatching success and juvenile growth rate in fresh water from those measured in salt water yielding a single measure of mean hatching success and growth rate for each family in salt water relative to fresh water (Fig. 2.2 comparison iii).  We then tested for differences in mean relative hatching success and growth rate on each population separately, using Student's t-test. Arcsine transformation was performed on proportional hatching data and alpha values were adjusted to 0.0167 to correct for the three tests completed on each performance measure (one in fresh water, one in salt water, one on performance in salt water relative to fresh water) using a Bonferroni correction.   All statistics were analyzed using StatView 5.0 (1998 SAS Institute Inc., Cary, NC).   28 RESULTS Hatching Success  Contrary to expectation (Fig. 2.2), complete morph embryos from the marine population appeared slightly more tolerant of fresh water than embryos from reduced morph families.  Although the effect was small (4%), a significantly greater proportion of marine embryos from complete morph families hatched in fresh water than did embryos from reduced morph families (t[18] = 2.962; P = 0.008; Fig. 2.3A).  In salt water, embryos from marine reduced and complete morph families exhibited similar abilities to hatch (t[18] = -0.160; P = 0.874; Fig. 2.3A).  Complete and reduced morph marine families did not differ significantly in their ability to hatch in fresh water relative to salt water (t[18] = - 2.488; P = 0.023; nonsignificant after Bonferroni correction). Both morphs hatched proportionally more larvae in fresh water than in salt water (Fig. 2.3A).  In the freshwater population there were no significant differences in hatching success between reduced and complete morph families in either treatment (t[24] > -1.181; P > 0.081; Fig. 2.3A), or in the abilities of each morph to hatch in fresh water relative to salt water (t[24] = -1.132; P = 0.269). Hatching success was low in the saltwater treatment for both plate morphs from the freshwater population.  Furthermore, no embryo development was observed in the unhatched eggs from all of the freshwater reduced morph families raised in salt water and more than 85% of the eggs that died in the complete morph families showed no signs of embryo development.  Egg mortality may be due to low success of fertilization in both freshwater morphs in salt water.  29 Juvenile Growth Rate  Mean juvenile growth rate in the marine population was significantly different between reduced and complete morph families when raised in fresh water (t[12] = -3.06; P= 0.001), but not in salt water (t[12] = -0.382; P = 0.709; Fig. 2.3B).  When raised in fresh water, marine families from reduced morph parents grew on average 65% faster than those from complete morph parents.  Mean growth rate in salt water relative to fresh water also differed significantly between lateral plate morphs from the marine population (t[12] = 3.286; P= 0.0065; Fig. 2.3B).  Reduced morph families grew faster in fresh water relative to salt water, whereas complete morph families grew faster in salt water relative to fresh water (Fig. 2.3B).  Similarly, in the freshwater population low morph families grew significantly faster than complete morphs in the freshwater treatment (t[12] = -2.203; P = 0.0478, Fig. 2.3B).  Low sample sizes due to low levels of fertilization and hatching success prevented any comparison among freshwater plate morphs in the saltwater treatment. DISCUSSION  Our results on hatching success and juvenile growth rate contradict Heuts’ (1947) hypothesis that differences in salinity drive the pattern of lateral plate divergence between marine and freshwater populations.  Unlike Heuts (1947), we found that lateral plate morphs exhibited little difference in their ability to hatch in fresh water and salt water. This discrepancy likely arises because Heuts’ different plate morphs were collected from separate locations: reduced morph fish came from fresh water, whereas his complete morph fish were collected from brackish water.  We restricted our tests to comparisons  30 between morphs collected from the same population, likely minimizing any additional adaptations to local salinity that may have confounded Heuts’ (1947) results.  We also documented significant differences in juvenile growth rate between reduced and complete morph families from both freshwater and marine populations when raised in fresh water (Fig. 2.3B).  These differences in growth rate have two important implications for the evolution of lateral plate morphology in populations of threespine stickleback.  First, the advantage of the low lateral plate morph in fresh water may stem from the association between juvenile growth rate and lateral plate morphology.  Second,  31 the ecological mechanism of selection on lateral plate morphology likely differs between freshwater and marine habitats.  In freshwater littoral habitats, reduction of lateral plate armor may occur as a correlated response to selection for faster juvenile growth rate of individuals with lower lateral plate number.  In marine habitats and lakes with large pelagic zones, lateral plate number remains high due to selection by pelagic fish predation (Reimchen, 2000).  It is important to note that our conclusions are based on results from laboratory experiment using artificial salt water and a restricted diet.  Our results may not reflect the complete suite of differences in salinity between all marine and freshwater habitats nor the differences in growth rate found in the wild, where food may or may not be limited.  Additionally, we measured growth rate on juveniles only, leaving the possibility for compensatory growth to occur at later life stages.  Lastly, because the marine reduced morph families were created using a partial-by-partial cross, roughly 25% of the offspring will be completely plated as adults.  Despite the presence of plate morph variation within marine reduced morph families, the differences in growth rate between marine reduced and complete morphs were similar to the differences observed between true low and complete morphs from the freshwater population (Fig. 2.3B). Correlated Response to Selection for Faster Growth Rate  Faster growth at the juvenile stage has three, potentially large, consequences for individual fitness: higher overwinter survival, greater reproductive potential, and decreased predation (Arendt 1997).  Firstly, larger, faster growing juveniles posses greater lipid stores and lower metabolism, buffering individuals from lower food availability in winter (Ludsin and DeVries 1997).  Because mass specific metabolism and  32 lipid accumulation favour juvenile fish that attain a larger size at the onset of fall (Thompson et al. 1991), faster growing, reduced lateral plate morphs may suffer less overwinter mortality.  Secondly, body length is strongly positively correlated with reproductive output (clutch size and mass) in threespine stickleback (Schluter 1995; Wooton 1984) and earlier breeding in other fish species (Bagenal 1978; Schultz et al. 1991).  If differences in growth rate and body size observed here persist into the reproductive season the following spring, the higher growth rate of low and partially plated individuals may result in greater reproductive potential relative to slower growing, completely plated individuals.  Thirdly, cannibalistic adult stickleback and three major freshwater invertebrate predators feed mainly on juveniles below 20 mm standard length, and most frequently on the smallest individuals available (Foster et al. 1988).  The 65% increase in juvenile growth rate of reduced morphs over the complete morph in fresh water may translate into a substantial decrease in the risk of predation experienced by low and partially plated individuals.  In littoral freshwater habitats where juvenile predation is a considerable selection pressure (Foster et al. 1988), passing more quickly through the most vulnerable size range may be extremely advantageous and a substantial reduction in predation alone may contribute significantly to the predominance of the low morph in independently isolated freshwater populations. Ecological and Evolutionary Mechanisms of Lateral Plate Divergence  The absence of clear reciprocal differences in hatching and growth rate among lateral plate morphs when raised in fresh water vs. salt water (Fig. 2.3), suggests that a single ecological mechanism, like that of salinity tolerance (Heuts 1947; Fig. 2.2), cannot  33 fully explain the pattern of lateral plate divergence between freshwater and marine habitats.  The absence of differences between morphs when raised in salt water suggests that the predominance of complete morphs in marine habitats is not an outcome of differences in hatching or growth rate.  We suggest that divergent natural selection on differences in lateral plate morphology between marine and freshwater habitats likely occurs via two mechanisms.  Selection first operates on differences in juvenile growth rate between lateral plate morphs, and subsequently on differential mortality of lateral plate morphs as adults due to pelagic fish predation.  In adult threespine stickleback, greater lateral plate number is correlated with a greater chance of escape and survival from pelagic fish predation (Bergstrom 2002; Reimchen 2000).  The potential fitness advantages of higher juvenile growth rate in fresh water may be beneficial to reduced plate morphs only when levels of pelagic fish predation as adults are relatively low. When pelagic predation intensity becomes too high, completely plated individuals may be favored by selection.  Thus, the pattern of lateral plate differentiation may be generated by a correlated response to selection for higher juvenile growth rate in freshwater littoral habitats, but direct selection for reduced mortality from pelagic predation as adults in marine and freshwater habitats with large pelagic zones. Although our results suggest that a correlated response to selection on differences in growth rate may drive patterns of lateral plate evolution in fresh water, the mechanisms responsible for the association between salinity, juvenile growth rate and lateral plate morphology remain unclear.  One possibility is that calcium limitation drives differences in growth rate between lateral plate morphs in fresh water.  Calcium is often  34 limiting in fresh water, and correlational studies have found a reduction in plate and spine number (Giles 1983) and pelvic girdle dimensions (Bell et al. 1993) along a gradient of lakes from high to low calcium concentration.  This hypothesis may explain the absence of a difference in juvenile growth rate in the saltwater treatment (Fig. 2.3B) where calcium concentrations were roughly 300 times greater than in the freshwater treatment. The energetic cost associated with producing more bony tissue during plate development in calcium limited freshwater habitats may result in completely plated individuals growing more slowly than individuals with reduced plate number.  Bell et al. (1993) noted, however, that separating the effects of a reduction in calcium on bone growth in fresh water from the concomitant reduction of other ions in fresh water is difficult.  The low level of ions such as phosphate may also contribute to reduction in bony lateral plates as well as the pelvic girdle in fresh water (Bell et al. 1993).  Although calcium limitation may explain why the growth rate of reduced and complete morphs are similar in salt water, it cannot explain why marine reduced morphs grow faster in fresh water than their native salt water.  The differences in growth rate we observed between salinity treatments may arise from unknown pleiotropic effects of low Eda alleles reintroduced to salt water, or they could arise by correlated selection for unknown phenotypes produced by genes closely linked to the freshwater low Eda alleles (Colosimo et al., 2005).  Indeed, not all the functions of Eda are known in stickleback and low Eda alleles may exhibit deleterious effects on unknown phenotypic traits in salt water only.  In addition, if changes in the regulation or structure of a gene implicated in salt secretion, Gjb1, that is closely linked to the freshwater low Eda alleles have  35 deleterious effects in salt water, then linkage may also explain why marine reduced morphs grew faster in fresh rather than salt water.  In contrast to our results on growth, Colosimo et al. (2004) found no association between quantitative trait loci determining adult body length and lateral plate morph in the F2 generation between Japanese marine and Paxton lake cross raised in fresh water.  This lack of an association between body length and lateral plate morph in adult F2 may result from breakdown of linkage disequilibrium, possibly refuting the role of pleiotropy.  However, their results could also be due to compensatory growth of completely plated individuals after lateral plate development, or the presence of separate genes for juvenile growth rate and final adult size.  Disentangling the roles of pleiotropy and genetic linkage in maintaining the association between juvenile growth rate and lateral plate morphology in our populations requires further study. Lastly, the absence of the low morph from freshwater habitats in Eastern Europe and Eastern North America is puzzling in light of our results.  Exceptions to the pattern of reduced morphs occurring in fresh water and complete morphs in salt water can also be found on the West Coast of North America.  Baumgartner and Bell (1984) observed that stickleback populations in lagoons near the mouths of streams tend to contain all three plate morphs, but the frequency of complete morphs increased further inland in lower salinity, high gradient stream habitats.  Hagen and Moodie (1982) suggested that the global distribution of lateral plate morphs was determined by temperature rather than salinity:  areas with lower winter temperatures and large annual fluctuations most often support the complete morph, whereas areas with mild winter temperatures and low  36 annual fluctuations support the low morph.  Unfortunately this association also breaks down when examining morph distributions on a local scale (for example see Coad 1983). Although numerous exceptions to the large scale trends in the distribution of lateral plate morphs exist, these exceptions may indicate the repetition of a common theme:  selection for armor reduction in fresh water may not involve selection on the function of lateral plates.  Indeed, the parallel recurrence of low Eda alleles in fresh water occurs as part of a common haplotype, which includes genes implicated in salt secretion (Gjb1) and parasite load (Tnfs13b), among others with unknown functions (Colosimo et al., 2005). The likelihood for evolutionary change in lateral plate armor to occur via a correlated response to selection for other phenotypes appears quite high.  Selection for lateral plate reduction in fresh water through correlated response to changes in growth rate may be only one of many ways to reduce lateral plate armor in freshwater populations of threespine stickleback.  37 REFERENCES Arendt, J. D.  1997.  Adaptive intrinsic growth rates: An integration across taxa. Quarterly Review Of Biology 72:149-177. Bagenal, T. B.  1978.  Aspects of fish fecundity. Pp. 75-101 in S. D. Gerking, ed. Ecology of Freshwater Fish Production. Wiley, New York, New York, USA. Banbura, J.  1994.  Lateral plate morph differentiation of fresh-water and marine populations of the three-spined stickleback, Gasterosteus aculeatus, in Poland. Journal of Fish Biology 44:773-783. Baumgartner, J. V., and M. A. Bell.  1984.  Lateral plate morph variation in California populations of the threespine stickleback, Gasterosteus aculeatus. Evolution 38:665-674. Bell, M. A.  1981.  Lateral plate polymorphism and ontogeny of the complete plate morph of threespine sticklebacks (Gasterosteus aculeatus). Evolution 35:67-74. ———  2001.  Lateral plate evolution in the threespine stickleback:  Getting nowhere fast. Genetica 112:445-461. ——— G. Orti, J. A. Walker, and J. P. Koenings.  1993.  Evolution of pelvic reduction in threespine stickleback fish:  A test of competing hypotheses. Evolution 47:906- 914. Bergstrom, C. A.  2002.  Fast-start swimming performance and reduction in lateral plate number in threespine stickleback. Canadian Journal of Zoology 80:207-213.  38 Coad, B. W.  1983.  Plate morphs in fresh water samples of Gasterosteus aculeatus from Arctic and Atlantic Canada:  Complementary comments on a recent contribution. Canadian Journal of Zoology 61:1174-1177. Colosimo, P. F., K. E. Hosemann, S. Balabhadra, V. J. Guadalupe, M. Dickson, J. Grimwood, J. Schmutz, R. M. Myers, D. Schluter, and D. M. Kingsley.  2005. Widespread parallel evolution in sticklebacks by repeated fixation of Ectodysplasin alleles. Science 307:1928-1933. ——— C. L. Peichel, K. Nereng, B. K. Blackman, M. D. Shapiro, D. Schluter, and D. M. Kingsley.  2004.  The genetic architecture of parallel armor plate reduction in threespine sticklebacks. PLoS Biology 2:1-7. Foster, S. A., V. B. Garcia, and M. Y. Town.  1988.  Cannibalism as the cause of an ontogenetic niche shift in habitat use by fry of the threespine stickleback. Oecologia 74:577-585. Giles, N.  1983.  The possible role of environmental calcium levels during the evolution of phenotypic diversity in Outer Hebridean populations of the Three-spined stickleback, Gasterosteus aculeatus. Journal of Zoology, London 199:535-544. Hagen, D. W., and G. E. E. Moodie.  1982.  Polymorphism for plate morphs in Gasterosteus aculeatus on the east coast of Canada and an hypothesis for their global distribution. Canadian Journal of Zoology 60:1032-1042. Heuts, M. J.  1947.  Experimental studies on adaptive evolution in Gasterosteus aculeatus L. Evolution 1:89-102.  39 Igarashi, K.  1970.  Formation of the scutes in the marine form of the three-spined stickleback, Gasterosteus aculeatus (L.). Annotationes Zoologicae Japonenses 43:34-42. Ludsin, S. A., and D. R. DeVries.  1997.  First-year recruitment of largemouth bass:  The interdependency of early life stages. Ecological Applications 7:1024-1038. McPhail, J. D.  1994.  Speciation and the evolution of reproductive isolation in the sticklebacks (Gasterosteus) of south-western British Columbia in M. A. Bell and S. A. Foster, eds. The Evolutionary Biology of the Threespine Stickleback. Oxford University Press, Oxford. Reimchen, T. E.  2000.  Predator handling failures of lateral plate morphs in Gasterosteus aculeatus:  Functional implications for the ancestral plate condition. Behaviour 137:1081-1096. Saimoto, R. K.  1993.  Life history of marine threespine stickleback in Oyster Lagoon, British Columbia. Pp. 97. Zoology. University of British Columbia, Vancouver. Schaarschmidt, T., E. Meyer, and K. Jurss.  1999.  A comparison of transport-related gill enzyme activities and tissue-specific free amino acid concentrations of Baltic Sea (brackish water) and freshwater threespine sticklebacks, Gasterosteus aculeatus, after salinity and temperature acclimation. Marine Biology 135:689-697. Schluter, D.  1995.  Adaptive radiation in sticklebacks:  Trade-offs in feeding performance and growth. Ecology 76:82-90. Schultz, E. T., L. M. Clifton, and R. R. Warner.  1991.  Energetic constraints and size- based tactics:  The adaptive significance of breeding-schedule variation in a  40 marine fish (Embiotocidae:  Micrometrus minimus). American Naturalist 138:1408-1430. Thompson, J. M., E. P. Bergersen, C. A. Carlson, and L. R. Kaeding.  1991.  Role of size, condition, and lipid content in the overwinter survival of age-0 Colorado squawfish. Transactions of the American Fisheries Society 120:346-353. Wooton, R. J.  1984.  A Functional Biology of Sticklebacks. University of California Press, Berkeley.   41 CHAPTER 3. PREDATION’S ROLE IN REPEATED PHENOTYPIC AND GENETIC DIVERGENCE OF ARMOR IN THREESPINE STICKLEBACK*  INTRODUCTION  According to the ecological theory of adaptive radiation, the main driver of phenotypic differentiation is divergent natural selection between environments stemming from differences in resources, habitat structure, and predator or competitor composition (Schluter 2000).  Theory and research has focused mainly on the role of competition and differential resource use in phenotypic and genetic diversification in a wide range of organisms (Chase et al. 2002; Langerhans 2006; Schluter 2000; Vamosi 2005).  More recently, there has been an increase in theoretical and empirical studies aimed at determining the role of predation in generating biological diversity (Abrams 2000; Endler 1995; Nosil and Crespi 2006; Reznick and Endler 1982; Vermeij 1987).  Empirical evidence of predation’s role in diversification remains mostly indirect and limited to changes in phenotype between environments having different predators. The majority of evidence comes from studies on bacteria and invertebrates (Diabate et al. 2008; McPeek 1997; Meyer and Kassen 2007; Mikolajewski et al. 2006; Nosil and Crespi 2006), with few vertebrate examples (Langerhans 2006; Vamosi 2005).  Very recently, the evolutionary consequences of predator driven divergent selection were elegantly demonstrated in Bahamas mosquitofish, Gambusia hubbsi, where low and high predation regimes have driven repeated shifts in morphological traits that also form the basis of *A version of this chapter has been published. Marchinko, K. B. 2009. Predation’s role in repeated phenotypic and genetic divergence of armor in threespine stickleback. Evolution 63: 127–138.   42 assortative mating, and therefore reproductive isolation and speciation (Langerhans et al. 2007).   43  Threespine stickleback, Gasterosteus aculeatus, exhibit a correlation between predator community and the amount of external bony armor (Fig. 3.1).  Stickleback from marine and lake habitats with a high abundance of piscivorous fish and birds (Reimchen 1994) are often heavily armored with long spines and more lateral plates.  Fish and bird predators are often gape-limited, only able to consume prey smaller than the size of their mouth.  Longer spines increase the stickleback’s effective diameter, cause injury, and make it difficult for gape-limited predators to swallow their prey, while lateral plates increase the chance of survival following attack (Hoogland et al. 1956; Reimchen 1992; 1994; 2000).  By contrast, aquatic insects become important sources of predation on stickleback from freshwater habitats in which predatory fish and birds are uncommon. Here, stickleback often have short spines, few lateral plates, and some populations have lost the pelvic girdle and associated spines entirely (Bell et al. 1993; Reimchen 1980). Predaceous insects are not gape-limited, instead they use appendages to capture and hold onto prey.  However, the small size of many aquatic insects restricts their ability to capture and consume large adult stickleback, constraining them to prey predominantly upon smaller juveniles (Foster et al. 1988).  Reimchen (1980) hypothesized that spines provide points of leverage for aquatic insects to hold onto when capturing juvenile stickleback, and any reduction in the length and number of spines would reduce the predators capture success.  Taken all together, the widespread documentation of the relationship between predator regime and stickleback armor has led to the common assertion that differences in armor among populations result from predator driven divergent selection (Bell 2001; Colosimo et al. 2004; Ellegren and Sheldon 2008; Gross  44 1977; Hagen and Gilbertson 1972; Moodie et al. 1973, Reimchen 1980; 1994; Reimchen and Nosil 2002; 2004; Reist 1980; Vamosi 2002).  Although several experiments show that predation by gape-limited fish results in selection for longer spines and more lateral plates (Reimchen 1992; 2000; Vamosi and Schluter 2004), evidence that predatory aquatic insects select for armor reduction is scarce and indirect.  For example, Reimchen and Nosil (2002; 2004) revealed a correlation between stickleback spine number and season.  Spine number was lowest in the summer when predaceous insects are most active, suggesting that individuals missing spines are at a selective advantage when predatory insects predominate.  Vamosi (2002) showed experimentally that mortality due to insect predation was greatest in unarmored benthic sticklebacks compared to lightly armored limnetic sticklebacks.  His results also imply a selective advantage to individuals with reduced armor when insect predation is significant.  In these cases, however, it is impossible to attribute differences in mortality to armor alone when additional traits and behaviors segregate between species under investigation.  Alternatively, loss of armor in freshwater stickleback populations might also occur through an energetic trade-off resulting from changes in abiotic conditions. Differences in salinity, calcium and phosphate concentration among populations may result in faster growth rates of individuals with reduced bony armor because more energy can be allocated to somatic and reproductive tissues as opposed to armor plates and spines (Giles 1983).  Interactions between ion concentration and predation may also occur (Bell et al. 1993).  45  The recurrent evolution of low morph freshwater populations from completely plated marine ancestors has occurred through the fixation of a low morph allele at the Ectodysplasin, Eda, gene (Colosimo et al. 2004; 2005).  Individuals with two low EdaL alleles will typically be of the low morph, while those with two complete EdaC alleles will most often be classified as the complete morph (Fig. 3.2; Colosimo et al. 2004). Depending on the alleles present at modifier loci, heterozygous individuals with one low EdaL and one complete EdaC allele may end up as a complete, a partial, or very rarely as a low morph (Colosimo et al. 2004; Creskso et al. 2004).  In addition to the strong association of Eda with lateral plate phenotype, additional quantitative trait loci (QTL) are now known to be located near the Eda gene.  QTL associated with Eda include body shape (Albert et al. 2008), the length of the anterior dorsal spine (S. M. Rogers unpublished manuscript), pelvic spine length (Shapiro et al. 2004), and possibly growth rate (Marchinko and Schluter 2007).  Each of these traits show divergence between derived fresh water and ancestral marine populations.  I experimentally tested if predation by aquatic insects results in selection for reduced stickleback armor and changes in Eda allele frequency using the F2 generation from a cross between divergent freshwater (reduced armor) and marine (robust armor) populations.  The wealth of natural history and experimental data provide distinct predicted outcomes.  Specifically, surviving juveniles from F2 families exposed to insect predation should have larger mean body size (Foster et al. 1988), shorter mean dorsal spine, pelvic spine and pelvic girdle length (Reimchen 1980; Vamosi 2002), and a higher  46 proportion of individuals missing the pelvic girdle (Reimchen and Nosil 2002), than in families not exposed to predation.  In addition, the association of Eda with several armor traits (Colosimo et al. 2004; Shapiro et al. 2004), suggests that exposure to insect predation should result in F2 families with a higher proportion of the freshwater, low morph EdaL allele. METHODS Fish Populations  I collected threespine stickleback from four populations in southwestern British Columbia.  The two marine populations, Oyster Lagoon (49°36´43˝ N, 124°01´57˝ W) on the Sechelt peninsula and Little Campbell River (49°00´52˝ N, 122°45´33˝ W) 45 km south of Vancouver, were characterized by long spines (Fig. 3.1), and a large, robust pelvic girdle.  All individuals sampled for crossing were of the complete lateral plate morph (Fig. 3.2).  The two freshwater populations, Paxton Lake (49°42´30˝ N, 124°31´28˝ W) on Texada Island and McKay Lake (49°3´23˝ N, 123°58´3˝ W) on Vancouver Island, were characterize by short dorsal spines (Fig. 3.1), and were of the low lateral plate morph (Fig. 3.2).  In Paxton Lake, I only collected individuals of the benthic species (Schluter and McPhail 1992), of which the vast majority of individuals are missing the pelvic girdle and pelvic spines (McPhail 1994).  All individuals from McKay Lake possessed a pelvic girdle and pelvic spines, but both traits were greatly reduced compared to the marine population (Fig. 3.1).    47  Experimental Protocol  I created two sets of crosses between freshwater and marine populations. The details of crossing, fertilization, and fish husbandry can be found in appendix 2.  The first set began with Paxton benthic males and Oyster Lagoon marine females to establish six separate F1 lines (Paxton line).  A single brother-sister pair from each F1 Paxton line was  48 crossed, establishing six separate Paxton line F2 families for study.  The second set of crosses was started from a single McKay Lake male and a single Little Campbell marine female to establish one F1 line (McKay line).  Ten brother-sister pairs were then crossed to establish ten F2 families from the McKay line for study.  Sample sizes before and after predation trials for each family from each set of crosses can be found in appendix 3. Variation for dorsal and pelvic spine lengths, pelvic girdle length and the number of lateral plates was present in the F2 progeny of both sets of crosses.  Therefore, I examined the effect of predation on dorsal and pelvic spine lengths, pelvic girdle length, and Eda allele frequency in both crosses.  Although using F2 families generated from six pairs (Paxton line) or one pair (McKay lines) of wild caught parents limits the amount genetic variation available in experimental families to that present in the original parents, juvenile F2 families of both crosses exhibited similar levels of morphological variation upon which selection could act.  The coefficient of variation for body size was approximately 10% in each F2 family and the standard deviation of spine and girdle length residuals was similar in both crosses (appendix 4).  Only Paxton line F2 families exhibited segregation variance in the presence and absence of the pelvic girdle and spine.  Predation trials were conducted in 20 wooden framed enclosures built into the shallow slopes of one experimental pond (23 m by 23 m) on the University of British Columbia campus.  Five enclosures were placed on each side of the pond with the long axis perpendicular to the shoreline and sloping towards the 3 m deep center of the pond.  49 The wooden frame of each enclosure measured 1.83 m long, 0.91 m wide, 0.91 m tall on its deepest side and 0.46 m tall on its shallowest side.  The vertical sides were covered with 1 mm fine meshed door screen, sealed with silicon and buried into the sandy substrate of the pond.  The top was covered with door screen to prevent adult dragonflies from laying eggs in the enclosures.  Within each enclosure, I buried 16 floating artificial plants made from shredded green plastic bags to provide refuge for both predators and prey.  The pond was then filled so that the water level sat just above the bottom of the shallow end of the enclosure, yielding a maximum water depth of approximately 0.5 m and a final volume of approximately 380 L in each enclosure.  I seeded each enclosure with zooplankton captured by plankton tow in adjacent ponds immediately prior to introducing F2 families.  Experimental trials began by splitting each F2 family in half by randomly assigning individuals into one of two treatment enclosures: one predation treatment with two common aquatic insect predators of juvenile stickleback (Reimchen 1994), backswimmers (Notonecta sp.) and dragonfly naiads (Aeshna sp.), and one control treatment containing no predators.  Although both Notonecta and Aeshna prefer to feed on the smallest fish available (Foster et al. 1988), these predatory species differ in their ability to capture and consume stickleback prey of different sizes.  Notonecta appear constrained to feed upon juvenile stickleback less than 15 mm in standard length, whereas Aeshna have been noted to capture and consume juvenile stickleback up to 25 mm long (Foster et al. 1988).  Thus, the survival of juvenile stickleback exposed to predaceous insects in this experiment may occur through predator avoidance (avoiding  50 detection or capture), as well as escape after capture.  Juvenile F2 families were acclimated overnight before predators were introduced.  The following day Notonecta and Aeshna predators were caught in adjacent experimental ponds and added to the enclosure designated for the predator treatment.  For each predator trial, the proportion of predatory insects to stickleback prey was 0.6 to1, and the relative proportion of each predatory species was 0.62 Notonecta to 0.38 Aeshna, similar to the relative proportion of each species found in a natural stickleback lake with no natural predatory fish (Foster et al. 1988).  Initial F2 family size ranged from 62 to 98 fish in the Paxton lines and from 41 to 140 fish in the McKay lines.  The first trial began on June 1, 2006 and the last trial on September 11, 2006. Every two days, the remaining number of stickleback and predatory insects in each enclosure were counted after slightly disturbing the artificial plants and enclosure sides. Any predators found missing were replaced at each census.  Trials were stopped when roughly 50% of the stickleback introduced to the predation treatment were found missing. The mean trial length was nine days, but ranged from six to 11.  At the end of each trial, individuals from both treatments were exposed to a lethal concentration of tricaine methanesulfonate (MS-222, Syndel Laboratories), preserved in 95% ethanol, and brought back to the lab.  The caudal fin of each individual was removed, placed in a labeled 1.5 µl centrifuge tube, and preserved in 95% ethanol for genetic analysis.  The remaining body of each individual was placed in a labeled 1.5 µl centrifuge tube and then fixed in 10% formalin for two weeks.  I then stained all bony elements using Alizarin red, after which I  51 preserved each individual separately in 40% isopropyl alcohol in preparation for morphological analysis.  Each stained individual was photographed using a Nikon DH1 digital camera. Morphological measurements were made on digital JPEG files using ImageJ v1.37 (Rasband 2007).  I measured five traits (Fig. 3.2):  standard length, anterior dorsal spine length, second dorsal spine length, pelvic spine length, pelvic girdle length, and counted the number of lateral plates.  In threespine stickleback, the total number of lateral plates is not determined until fish are sub-adults, roughly 30 mm in standard length (Bell 2001; Igarashi 1970).  All of the juveniles introduced to the enclosures before trials began resembled the low plate morph and lateral plate morph could not be resolved from plate counts after the experiment ended.  Instead, I genotyped a molecular marker diagnostic for lateral plate morph (Colosimo et al. 2005) to determine the most likely lateral plate phenotype of individuals. Genetic Analysis of the Ectodysplasin Gene  I isolated total genomic DNA from caudal fin clips of all six Paxton line F2 families using standard phenol-chloroform extraction methods.  The in/del locus, Stn381, within intron six of the Ectodysplasin gene (Colosimo et al. 2005) was used to identify the genotype corresponding to lateral plate phenotype of F2 juveniles.  Ectodysplasin alleles were amplified in 10 µl PCR reactions containing 5 to 15 ng genomic DNA, 1µM of each forward and reverse primer, 1X PCR buffer, 0.25-0.125 mM of each dNTP, 1.5 mM MgCl2, and 0.25U AmpliTaq polymerase (Applied Biosystems).  Cycling conditions were as follows: 93 °C for 3 min, 95 °C 30 s, 59 °C 30s, 72 °C 30 s, 5 cycles of 94 °C 30  52 s, 59 °C 30 s, 72 °C 30 s, 35 cycles of 90 °C 30 s, 60 °C 30 s, 72 °C 30 s, followed by 72 °C 10 min, then cooled to 4 °C.  Amplified PCR products were separated by gel electrophoresis on an ABI 3733 automated sequencer using the GS500 Size Standard (Applied Biosystems) and freshwater low morph EdaL and marine complete morph EdaC alleles were scored using GENEMAPPER software (Applied Biosystems). Statistical Analyses  Standardized selection differentials (i) were calculated according to equation 6.1 in Endler (1986),  ! i = X a " X b varb  where ! X a and ! X b were the mean trait values of fish from a single family measured at the end of the predation and control treatments respectively, and varb is trait variance in the control treatment.  Selection differentials were calculated for each family separately and may be found in appendix 4.  Selection on each trait was visualized using the relative survival function, f(z) (eqn. 8 in Anderson 1995),   ! f (z) = S1 S2 h(z) 1" h(z)   53 where S1 and S2 are the total sample sizes at the end of the experiment for the control and predation treatments respectively.  h(z) is the conditional probability that a fish of phenotype z was present in the predation treatment at the end of the experiment given that it was present in the control treatment at the end of the experiment (Anderson 1995).  h(z) was estimated for each trait using non-parametric cubic spline (Schluter 1988) with family as a nominal covariate, individual phenotypes present in the control treatment set to zero (survival = 0) and phenotypes of those individuals remaining at the end of the predation treatment set to 1 (survival = 1).  The standard errors of the relative survival function f(z) were based on h(z).  To calculated standard error, S1 and S2 were fixed at observed values, and not re-sampled for each bootstrap sample.  I assume that the phenotypic distributions of two halves of an F2 family divided randomly between treatments were roughly the same prior to selection and any difference between treatments after exposure to predation results from selection.  Paired t-tests were performed separately for Paxton and McKay lines, and separately for each trait, using family as the replicate to determine the significance of selection differentials (differences in means between treatments).  Because spine and girdle lengths grow with body size, I corrected these traits for size using residuals from an ordinary least squares regressions of each trait on standard length.  Standard length, the distance from the tip of the snout to the end of the vertebral column, is highly correlated with, and serves as a common measure of body size in fish (Baumgartner et al. 1988).  Regressions were carried out separately on each family and combining fish from both treatments.  54  To analyze the Paxton line F2 cross, I removed individuals missing the pelvic girdle and spines from the data sets prior to calculating the within family regression of pelvic girdle and pelvic spine on standard length.  A separate analysis of the effect of aquatic insect predation on the survival of fish with and without the pelvic girdle was performed using paired t-tests between treatments on the difference in mean proportion of individuals without a pelvic girdle at the end of at trial.  The effect of aquatic insect predation on Eda allele frequency in the Paxton line families was analyzed using a paired t-test on the difference between treatments in the proportion of the freshwater low morph EdaL allele (∆p) present at the end of a trial. Because of linkage, any change in the frequency of Eda may be due to selection directly on Eda, or to selection on traits with genes located near Eda on the same chromosome.  I examined the association between Eda genotype and traits previously shown to be associated with the Eda region: the number of lateral plates, anterior dorsal spine length and pelvic spine length (Colosimo et al. 2004; S. M. Rogers unpublished manuscript; Shapiro et al. 2004).  The association between Eda and size-corrected lateral plate number, anterior dorsal spine and pelvic spine length was tested using mixed-model analysis of covariance (ANCOVA; fixed effect: Eda genotype, random effect: family, dependent variable: each armor trait).  In the analyses, Eda genotype was treated as numeric variable based on the number of low morph EdaL alleles present (0,1, or 2).  In addition, because low and complete lateral plate morphs differ in growth rate in fresh water (Marchinko and Schluter 2007), I also tested for and association between Eda  55 genotype and body size (standard length), using a similar mixed-model ANCOVA.  Note that because selection on Eda in the predator treatment could potentially bias the correlation of Eda genotype with morphological traits, I only used fish from the control (no predator) treatments in these tests.  Because of the specific direction of predicted outcomes, the significance level of t-tests was based on one-tailed probabilities.  However, I report all tests in which the statistical significance of one and two-tailed tests disagreed.  All proportion data were arcsine-square root transformed prior to analysis.  All analyses were carried out using R (R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org). RESULTS Selection on Armor Traits Aquatic insects preyed most heavily upon the smallest and most heavily armored individuals within F2 families.  Predation resulted in directional selection for larger body size in both crosses (Fig. 3.3): mean standard length was significantly greater in the predator treatment compared to the control (Table 3.1; difference between treatments was 0.7 & 0.4 mm in the Paxton and McKay lines respectively).  Selection resulting from insect predation produced similar patterns of shorter spine and girdle lengths in surviving juveniles from both Paxton and McKay lines (Table 3.1; Fig. 3.3).  F2 families split between treatments had shorter size-adjusted dorsal spine, and pelvic spine and girdle lengths in the predation treatment compared to the control.  However, the two sets of crosses differed with respect to which armor traits experienced significant selection (Table 3.1; Fig. 3.3), and the direction of selection varied among families in some traits  56 (appendix 4).  In McKay lines, mean size-adjusted pelvic girdle length was significantly shorter in predation treatment than in the control group (Table 3.1).  Dorsal and pelvic  57 spines were also shorter in predator compared to control treatments and differences approached significance (Table 3.1).  Similarly, in Paxton line F2 families, predation favored individuals with significantly shorter anterior dorsal spines (Table 3.1; Fig. 3.3).  Table 3.1.  Standardized selection differentials and significance tests for standard length and size adjusted (residual) spine lengths in F2 families split between two treatments: one with predatory aquatic insects, one without.  The selection differentials shown below are the median of all families within each cross type.  Mean trait values, standard deviations and selection differentials for each family can be accessed appendix 4. ✝ P < 0.1, * P < 0.05 , ** P < 0.01 for one-tailed tests.  Note that the two-tailed test of standard length in the McKay lines was not significant (P = 0.0873). median selection median selection differential (i) t0.05(1),5 differential (i) t0.05(1),9 standard length 0.403 -3.939 0.005 ** 0.352 -1.918 0.044 * anterior dorsal spine length -0.246 4.947 0.002 ** -0.103 1.078 0.154 second dorsal spine length -0.065 1.012 0.179 -0.121 1.479 0.087 pelvic spine length 0.011 0.132 0.450 -0.193 1.414 0.095 pelvic girdle length 0.011 0.246 0.408 -0.359 2.290 0.024 * Paxton lines McKay lines P P   Five of the six Paxton line F2 families exhibited segregation for pelvic girdle loss in a nearly 3:1 ratio of girdle presence to girdle absence (control treatment in Fig. 3.4).  In contrast to expectation, however, the proportion of individuals missing the pelvic girdle was lower in the predator treatment (8%) than in the control (21%; Fig. 3.4), and this difference between treatments approached significance (t0.05(2)4 = -2.308; P = 0.08). Although insect predation resulted in the higher survival of individuals with shorter  58 pelvic girdles (Fig. 3.3), the complete loss of the pelvic girdle was detrimental to survival in the presence of predatory insects (Fig. 3.4).  Selection on the Ectodysplasin Gene  Eda allele frequency differed significantly between predator and control treatments (t0.05(1)5 = 2.528; P = 0.026), and predation was greatest on individuals with complete morph EdaC alleles (Fig. 3.5).  The frequency of the low morph EdaL allele was 12% higher in predation treatments compared to that in the control (within family ∆p ranged from -0.002 to 0.25; appendix 5).  The higher frequency of the EdaL allele in predation compared to control treatments was due to an increase in the proportion of individuals with either heterozygous (EdaC/L), or homozygous low morph (EdaL/L) genotypes at the Eda locus (Fig. 3.5).  Thus, in the absence of countervailing selective  59 forces, selection on juvenile stickleback by predatory freshwater insects will lead to an increase in the frequency of the partial and low lateral plate morphs in adults.   Eda genotype was a significant covariate for size-adjusted anterior dorsal spine length (Table 3.2).  Individuals homozygous for the low morph EdaL allele had shorter size-adjusted anterior dorsal spines than individuals with two complete morph EdaC alleles.  Size-adjusted lateral plate number did not differ between predation and control treatments (t0.05(2)5 = -0.9538, P = 0.384) and Eda genotype was not a significant covariate for lateral plate number in juveniles (Table 3.2).  Eda genotype was not a significant covariate for either standard length, or size-adjusted pelvic spine length (Table 2).  60 DISCUSSION  Predation by aquatic insects resulted in higher survival of individuals with shorter spines and pelvic girdles, greater body length, and led to an increase in the frequency of the freshwater, low morph EdaL allele in F2 families.  This provides critical evidence that selection resulting from aquatic insect predation contributes to the reduction of stickleback armor during their radiation in freshwater lakes and streams.  Moreover, this study, in conjunction with ample evidence for selection for greater levels of armor when fish and bird predation are high (Reimchen 1992; 1994; 2000), strongly indicates that divergent selection, based on differences in predator regime among populations, is one mechanism contributing to the vast morphological diversification observed in threespine stickleback.  Insect predation upon juvenile stickleback clearly results in a reduction of armor traits in the direction predicted from natural populations (Fig. 3.3).  It is important to note though, that selection was demonstrated on juveniles from F2 crosses between divergent populations, rather than wild caught, or pure population lines.  As such, recombination will have reduced trait correlations arising from linkage disequilibrium between distant genomic locations.  This makes it likely that the effects of predation on armor were isolated from effects on unlinked behavioral or physiological traits relevant to predator avoidance and escape (Bell et al. 2007; Huntingford et al. 1994).  However, several armor traits map to the same genomic region near Eda, including lateral plates, dorsal and  61 Table 3.2. Mixed-model ANCOVA results testing the effects of genotype at the Eda locus, family as a random factor, and their interaction on standard length, and size-adjusted lateral plate number, anterior dorsal spine, and pelvic spine length in F2 families from the Paxton lines.  Eda genotype was treated as a numeric variable based on the number of low plate EdaL alleles present in an individual (EdaC/C = 0, EdaC/L = 1, EdaL/L = 2). source of variation df MS F P A) standard length genotype 1 0.0002 0.00006 0.994 family 5 17.165 10.165 < 0.0001 genotype * family 5 3.472 2.056 0.074 residual 150 1.689 B) number of lateral plates genotype 1 12.603 4.081 0.099 family 5 0.139 0.078 0.995 genotype * family 5 3.088 1.740 0.129 residual 150 1.775 C) anterior dorsal spine length genotype 1 0.067 11.634 0.019 family 5 0.003 0.227 0.951 genotype * family 5 0.006 0.402 0.847 residual 148 0.014 D) pelvic spine length genotype 1 0.077 1.610 0.260 family 5 0.024 0.703 0.622 genotype * family 5 0.047 1.397 0.231 residual 148 0.034   62  pelvic spines.  Covariance of several armor traits increases the ability to detect total selection on armor, but results in an inability to distinguish which trait(s) are the target of selection.  The application of multiple regression to selection analyses can often distinguish which traits are under the strongest selection, but they assume that all important characters were measured (Lande and Arnold 1983).  Because Eda genotype was correlated with dorsal spine length, and possibly with unmeasured or unknown traits also affecting fitness (Albert et al. 2008; Barrett et al. 2008), further work is necessary to distinguish which characters are the most important to surviving insect predation.  Predatory insects likely impose selection for reduced armor in freshwater populations of threespine stickleback via two mechanisms.  Firstly, insect predation may result in armor reduction through direct selection on specific traits.  Longer stickleback spines may increase the ability of predatory insects to hold onto and consume their prey (Reimchen 1980).  Whether spine length, or the absence of specific spines, influences the probability that a stickleback escapes after capture by predatory insects remains to be confirmed.  A greater number of lateral plates is associated with reduced velocity and displacement during the fast-start escape response (Bergstrom 2002), making it more likely for predaceous insects to capture stickleback with higher lateral plate counts. However, juvenile stickleback differed little in lateral plate number and the number of lateral plates did not differ between predation and control treatments in this experiment. Any direct effect of predation on lateral plate number is likely limited to adults that are largely invulnerable to predatory insects (Foster et al. 1988).  Secondly, insect predation may result in reduced armor indirectly through selection on differences in growth rate:  63 growth rate is slower in stickleback with more armor than those with reduced armor (Marchinko and Schluter 2007).  In the current study, however, selection on spine and girdle length was independent of body size, suggesting that the reduction in armor in natural populations is not entirely due to the association between growth rate and armor.  The parallel reduction of stickleback armor during the colonization of fresh water is likely due to the action of many selective agents.  Although insect predation appears important, my results do not necessarily imply that it is the sole, or even the predominant factor.  Ion concentration is often low in freshwater environments.  Reduced levels of calcium, phosphate, or other ions, may impose a trade-off in growth rate between heavily and lightly armored fish that is rooted in the development of bony armor versus soft tissues (Giles 1983).  Increased reproductive output and higher over winter survival are commonly associated with faster growing, larger fish (Ludsin and DeVries 1997; Thompson et al. 1991; Wootton 1984).  If faster growth rates of individuals with less armor yield larger adult stickleback, then body size may also feature as a target of selection during episodes of armor reduction.  As such, the relationship between armor, body size, and growth rate warrants further investigation.  Predatory insects preyed more heavily upon stickleback missing, rather than possessing, a pelvic girdle.  This result appears to contradict Reimchen’s original hypothesis that individuals lacking spine and girdle structures should experience the lowest mortality from predaceous insects (Reimchen 1980).  If the pattern I observed reflects mechanisms occurring in wild populations, then predation by insects cannot be the selective agent driving pelvic girdle loss threespine stickleback.  Alternatively,  64 negative fitness epistasis may be common in crosses between widely divergent populations with different genetic backgrounds (Leips and Mackay 2000; Rogers and Bernatchez 2007; Ungerer et al. 2003).  Negative fitness epistasis occurs when the insertion of novel alleles into a widely divergent genetic background results in decreased fitness.  For epistasis to account for the pattern observed in this study, individuals lacking a pelvic girdle, yet possessing marine alleles at a particular, but different set of loci must suffer the highest predation rates.  Only those individuals that lack a pelvic girdle, but also possess freshwater alleles at other essential loci survive.  Intriguingly, such negative epistatic interactions may account for the genetic incompatibility and decreased hybrid fitness that occurs in other wide crosses between parents from strongly divergent threespine stickleback populations (Mckinnon and Rundle 2002). Divergent Selection by Predators: Implications for Stickleback Speciation  Divergent selection on body size arising from different predator regimes may feature prominently in the evolution of reproductive isolation in threespine stickleback. In this study, insect predation had a large effect on juvenile body size.  Larger, faster growing individuals survived more often than smaller, slower growing individuals. Growth rate and body size may influence fitness in a variety of ways (Arendt 1997; Brown et al. 1993).   In addition to influencing susceptibility to predators, size is positively correlated with reproductive output (Wootton 1984) and over winter survival (Ludsin and DeVries 1997; Thompson et al. 1991).  Moreover, adult size strongly influences mate choice in reproductively isolated anadromous-stream and benthic-limnetic pairs of threespine stickleback (McKinnon et al. 2004; Nagel and  65 Schluter 1998; Vines and Schluter 2006).  Males and females most often choose to mate with individuals that most closely resemble their own size.  If differences in juvenile growth rate persist through to breeding adults, predator driven divergent selection on juvenile size may be one ecological mechanism by which size-based assortative mating is established.  Predation’s role in the evolution of reproductive isolation may also involve Eda. Mate choice in stickleback also appears to be based, at least in part, on body shape. Benthic species mate more often with benthic-shaped individuals from allopatric populations, and limnetic species with more limnetic-shaped allopatric individuals (Vines and Schluter 2006).  In an F2 cross generated from highly divergent populations similar to those used here, Albert et al. (2008) found that the genomic region in tight linkage with Eda had multiple effects on body shape in the head and pelvic regions.  It remains unknown whether changes in stickleback body shape are due to pleiotropic effects of Eda, or linkage with other genes for body shape.  However, Eda appears to have widespread pleiotropic effects on tissue and bone morphology in both mice and humans (Colosimo et al. 2005).  My work demonstrated that exposure to insect predation resulted in a higher frequency of individuals possessing the low morph EdaL allele due to decreased survival of individuals with homozygote EdaC/C genotypes.  If changes in stickleback body shape occur through the pleiotropic effects of Eda, then predation may influence assortative mating through its effects on Eda allele frequency.  This would be of particular importance when freshwater populations fixed for the low morph EdaL allele  66 still experience gene flow from, or come into secondary contact with, anadromous populations fixed for the complete morph EdaC allele.  Reproductive isolation as a by-product of adaptive differences in size and shape between geographic pairs of threespine stickleback is traditionally thought to arise via divergent (or disruptive) selection driven by resource competition (Mckinnon and Rundle 2002; McPhail 1994; Schluter and McPhail 1992; Taylor and McPhail 2000).  The effect of insect predation on juvenile size and Eda allele frequency, suggests that divergent selection arising from dissimilar predator communities may have been an important source of phenotypic differentiation upon which mate choice is based.  Along with the diversifying effects of resource competition, predation likely contributes to by-product reproductive isolation during adaptive radiation in the threespine stickleback species complex.  67 REFERENCES Abrams, P. A.  2000.  Character shifts of prey species that share predators. American Naturalist 156:S45-S61. Albert, A. Y. K., S. Sawaya, T. H. Vines, A. K. Knecht, C. T. Miller, B. R. Summers, S. Balabhadra, D. M. Kingsley, and D. Schluter.  2008.  The genetics of adaptive shape shift in stickleback: Pleiotropy and effect size. Evolution 62:76-85. Anderson, C. S.  1995.  Calculating size-dependent relative survival from samples taken before and after selection. Pp. 455–466 in D. H. Secor, J. M. Dean and S. E. Campana, eds. Recent developments in fish otolith research. University of South Carolina Press, Columbia, S.C. Arendt, J. D.  1997.  Adaptive intrinsic growth rates: An integration across taxa. Quarterly Review Of Biology 72:149-177. Barrett, R. D. H., S. M. Rogers, and D. Schluter.  2008.  Natural selection on a major armor gene in threespine stickleback. Science 322:255-257. 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).  Can. J. Zool. 66:467 474. Bell, A. M., T. Backstrom, F. A. Huntingford, T. G. Pottinger, and S. Winberg.  2007. Variable neuroendocrine responses to ecologically-relevant challenges in sticklebacks. Physiology & Behavior 91:15-25. Bell, M. A.  2001.  Lateral plate evolution in the threespine stickleback:  Getting nowhere fast. Genetica 112:445-461.  68 Bell, M. A., G. Orti, J. A. Walker, and J. P. Koenings.  1993.  Evolution of pelvic reduction in threespine stickleback fish:  a test of competing hypotheses. Evolution 47:906-914. Bergstrom, C. A.  2002.  Fast-start swimming performance and reduction in lateral plate number in threespine stickleback. Canadian Journal of Zoology 80:207-213. Brown, J. H., P. A. Marquet, and M. L. Taper.  1993.  Evolution of body-size: Consequences of an energetic definition of fitness. American Naturalist 142:573- 584. Chase, J. M., P. A. Abrams, J. P. Grover, S. Diehl, P. Chesson, R. D. Holt, S. A. Richards, R. M. Nisbet, and T. J. Case.  2002.  The interaction between predation and competition: a review and synthesis. Ecology Letters 5:302-315. Colosimo, P. F., C. L. Peichel, K. Nereng, B. K. Blackman, M. D. Shapiro, D. Schluter, and D. M. Kingsley.  2004.  The genetic architecture of parallel armor plate reduction in threespine sticklebacks. PLoS Biology 2:1-7. Colosimo, P. F., K. E. Hosemann, S. Balabhadra, V. J. Guadalupe, M. Dickson, J. Grimwood, J. Schmutz, R. M. Myers, D. Schluter, and D. M. Kingsley.  2005. Widespread parallel evolution in sticklebacks by repeated fixation of Ectodysplasin alleles. Science 307:1928-1933. Creskso, W. A., A. Amores, C. Wilson, J. Murphy, M. Currey, P. Phillips, M. A. Bell, C. B. Kimmel, and J. H. Postlethwait.  2004.  Parallel genetic basis for repeated evolution of armor loss in Alaskan threespine stickleback populations. Proceedings of the National Academy of Science USA 101:6050-6055.  69 Diabate, A., R. K. Dabire, K. Heidenberger, J. Crawford, W. O. Lamp, L. E. Culler, and T. Lehmann.  2008.  Evidence for divergent selection between the molecular forms of Anopheles gambiae: role of predation. Evolutionary Biology 8:5. Ellegren, H., and B. C. Sheldon.  2008.  Genetic basis of fitness differences in natural populations. Nature 452:169-175. Endler, J. A.  1986.  Natural Selection in the Wild. Princeton University Press, Princeton. Endler, J. A.  1995.  Multiple-trait coevolution and environmental gradients in guppies. Trends in Ecology & Evolution 10:22-29. Foster, S. A., V. B. Garcia, and M. Y. Town.  1988.  Cannibalism as the cause of an ontogenetic niche shift in habitat use by fry of the threespine stickleback. Oecologia 74:577-585. Giles, N.  1983.  The possible role of environmental calcium levels during the evolution of phenotypic diversity in Outer Hebridean populations of the Three-spined stickleback, Gasterosteus aculeatus . Journal of Zoology, London 199:535-544. Gross, H. P.  1977.  Adaptive trends of environmentally sensitive traits in the three- spined stickleback, Gasterosteus aculeatus  L. Zeitschrift für zoologische Systematik und Evolutionsforschung 15:252-278. 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.  70 Hoogland, R., D. Morris, and N. Tinbergen.  1956.  The spines of sticklebacks (Gasterosteus and Pygosteus) as means of defence against predators (Perca and Esox). Behaviour 10:205-236. Huntingford, F. A., P. J. Wright, and J. F. Tierney.  1994.  Adaptive variation in antipredator behaviour in threespine stickleback. in M. A. Bell and S. A. Foster, eds. The Evolutionary Biology of the Threespine Stickleback. Oxford University Press, Oxford, UK. Igarashi, K.  1970.  Formation of the scutes in the marine form of the three-spined stickleback, Gasterosteus aculeatus  (L.). Annotationes Zoologicae Japonenses 43:34-42. Lande, R., and S. J. Arnold.  1983.  The measurement of selection on correlated characters. Evolution 37:1210-1226. Langerhans, R. B.  2006.  Evolutionary consequences of predation:  avoidance, escape, reproduction, and diversification. Pp. 177-220 in A. M. T. Elewa, ed. Predation in organisms:  a distinct phenomenon. Springer-Verlag, Heidelberg, Germany. Langerhans, R. B., M. E. Gifford, and E. O. Joseph.  2007.  Ecological speciation in Gambusia fishes. Evolution 61:2056-2074. Leips, J., and T. F. C. Mackay.  2000.  Quantitative trait loci for life span in Drosophila melanogaster:  interactions with genetic background and larval density. Genetics 155:1773-1788. Ludsin, S. A., and D. R. DeVries.  1997.  First-year recruitment of largemouth bass:  the interdependency of early life stages. Ecological Applications 7:1024-1038.  71 Marchinko, K. B., and D. Schluter.  2007.  Parallel evolution by correlated response: lateral plate reduction in threespine stickleback. Evolution 61:1084-1090. McKinnon, J. S., S. Mori, B. K. Blackman, L. David, D. M. Kingsley, L. Jamieson, J. Chou, and D. Schluter.  2004.  Evidence for ecology's role in speciation. Nature 429:294-298. Mckinnon, J. S., and H. D. Rundle.  2002.  Speciation in nature:  The threespine stickleback model systems. Trends in Ecology and Evolution 17:480-488. McPeek, M. A.  1997.  Measuring phenotypic selection on an adaptation:  lamellae of damselflies experiencing dragonfly predation. Evolution 51:459-466. McPhail, J. D.  1994.  Speciation and the evolution of reproductive isolation in the sticklebacks (Gasterosteus) of south-western British Columbia. Pp. 399-437 in M. A. Bell and S. A. Foster, eds. The Evolutionary Biology of the Threespine Stickleback. Oxford University Press, Oxford. Meyer, J. R., and R. Kassen.  2007.  The effects of competition and predation on diversification in a model adaptive radiation.  446:432-435. Mikolajewski, D. J., F. Johansson, B. Wohlfahrt, and R. Stoks.  2006.  Invertebrate predation selects for the loss of a morphological antipredator trait. Evolution 60:1306-1310. Moodie, G. E. E., J. D. McPhail, and D. W. Hagen.  1973.  Experimental demonstration of selective predation on Gasterosteus aculeatus. Behaviour 47:95-105. Nagel, L., and D. Schluter.  1998.  Body size, natural selection, and speciation in sticklebacks. Evolution 52:209-218.  72 Nosil, P., and B. J. Crespi.  2006.  Experimental evidence that predation promotes divergence in adaptive radiation. Proceedings of the National Academy of Sciences USA 103:9090-9095. Rasband, W. S.  2007.  ImageJ. U.S. National Institutes of Health, Bethesda, Maryland, USA, http://rsb.info.nih.gov/ij/. Reimchen, T. E.  1980.  Spine deficiency and polymorphism in a population of Gasterosteus aculeatus :  an adaptation to predators? Canadian Journal of Zoology 58:1232-1244. Reimchen, T. E.  1992.  Injuries on stickleback from attacks by a toothed predator (Oncorhynchus) and implications for the evolution of lateral plates. Evolution 46:1224-1230. Reimchen, T. E.  1994.  Predators and morphological evolution in threespine stickleback in M. A. Bell and S. A. Foster, eds. The Evolutionary Biology of the Threespine Stickleback. Oxford University Press, Oxford. Reimchen, T. E.  2000.  Predator handling failures of lateral plate morphs in Gasterosteus aculeatus :  functional implications for the ancestral plate condition. Behaviour 137:1081-1096. Reimchen, T. E., and P. Nosil.  2002.  Temporal variation in divergent selection on spine number in threespine stickleback. Evolution 56:2472-2483. Reimchen, T. E., and P. Nosil.  2004.  Variable predation regimes predict the evolution of sexual dimorphism in a population of threespine stickleback. Evolution 58:1274- 1281.  73 Reist, J. D.  1980.  Predation upon pelvic phenotypes of brook stickleback, Culaea inconstans, by selected invertebrates. Canadian Journal of Zoology 58:1253-1258. Reznick, D., and J. A. Endler.  1982.  The impact of predation on life-history evolution in Trinidadian guppies (Poecilia reticulata). Evolution 36:160-177. Rogers, S. M., and L. Bernatchez.  2007.  The genetic architecture of ecological speciation and the association with signatures of selection in natural lake whitefish (Coregonas sp Salmonidae) species pairs. Molecular Biology And Evolution 24:1423-1438. Schluter, D.  1988.  Estimating the form of natural selection on a quantitative trait. Evolution 42:849-861. Schluter, D.  2000.  The Ecology of Adaptive Radiation. Oxford University Press, Oxford. Schluter, D., and J. D. McPhail.  1992.  Ecological character displacement and speciation in Sticklebacks. American Naturalist 140:85-108. Shapiro, M. D., M. E. Marks, C. L. Peichel, B. K. Blackman, K. Nereng, B. Jonsson, D. Schluter, and D. M. Kingsley.  2004.  Genetic and developmental basis of evolutionary pelvic reduction in threespine sticklebacks. Nature 428:717-723. Taylor, E. B., and J. D. McPhail.  2000.  Historical contingency and determinism interact to prime speciation in sticklebacks. Proceedings of the Royal Society of London B 271:2375-2384.  74 Thompson, J. M., E. P. Bergersen, C. A. Carlson, and L. R. Kaeding.  1991.  Role of size, condition, and lipid content in the overwinter survival of age-0 Colorado squawfish. Transactions of the American Fisheries Society 120:346-353. Ungerer, M. C., C. R. Linder, and L. H. Rieseberg.  2003.  Effects of genetic background on response to selection in experimental populations of Arabidopsis thaliana. Genetics 163:277-286. Vamosi, S. M.  2002.  Predation sharpens the adaptive peaks:  survival trade-offs in sympatric sticklebacks. Annales Zoologici Fennici 39:237-248. Vamosi, S. M.  2005.  On the role of enemies in divergence and diversification of prey: a review and synthesis. Canadian Journal of Zoology 83:894-910. Vamosi, S. M., and D. Schluter.  2004.  Character shifts in the defensive armor of sympatric sticklebacks. Evolution 58:376-385. Vermeij, G. J.  1987.  Evolution and Escalation:  An Ecological History of Life. Princeton University Press, Princeton, New Jersey. Vines, T. H., and D. Schluter.  2006.  Strong assortative mating between allopatric sticklebacks as a by-product of adaptation to different environments. Proceedings of the Royal Society of London B 273:911-916. Wootton, R. J.  1984.  A Functional Biology of Sticklebacks. University of California Press, Berkeley  75 CHAPTER 4.  POLYMORPHISM MAINTAINED BY NATURAL SELECTION DESPITE HETEROZYGOTE DISADVANTAGE AT A MAJOR GENE MAIN BODY  Single locus polymorphisms maintained in the face of heterozygote disadvantage (underdominance) are the starting point for multiple evolutionary endpoints (Wilson and Turelli 1986), including dominance and speciation (Durinx and Van Dooren 2009; Otto et al. 2008), yet examples from nature are conspicuously absent.  We report the presence of a stable polymorphism with heterozygote disadvantage at a major armour locus, Ectodysplasin (Eda), in a wild population of threespine stickleback. Armour (lateral plate) morph frequencies in adult fish have remained stable over 40 years with a deficit of intermediate phenotypes.  Genetic analysis confirms a deficit of heterozygotes at the Eda locus with an absence of population structure at neutral markers.  Random (Hardy-Weinberg) genotype frequencies are restored in wild-caught zygotes, suggesting that selection rather than assortative mating is responsible for the observed heterozygote deficiency.  Changes in genotype frequencies across a generation reveal that heterozygote disadvantage is the net result of opposing selection between the growth and reproductive phases of the life cycle.  Differences in carbon isotope signature expose niche separation between Eda homozygotes, providing an ecological mechanism for the maintenance of polymorphism despite heterozygous disadvantage.  Our study provides support for the role of disruptive selection in maintaining genetic variation for functionally important traits (Rueffler et al. 2006), and highlights the benefits of integrating genetic and *A version of this chapter will be submitted. Marchinko, K. B., B. W. Matthews, S. M. Rogers, D.Schluter. Polymorphism maintained by natural selection despite heterozygote disadvantage at a major gene.  76 ecological approaches for understanding the mechanisms maintaining variation and promoting diversification in wild populations.  We examined the genetic and ecological causes for the occurrence of a lateral plate polymorphism in a population of threespine stickleback inhabiting a large natural lake on Vancouver Island, British Columbia (Fig. 4.1).  Kennedy Lake is one of several large lakes in the region in which stickleback exhibit polymorphism in this trait.  This lateral plate polymorphism has remained stable over the last 43 years, with a lower frequency of individuals having an intermediate number of plates (partial morph) than individuals having either few (low morph) or many lateral plates (complete morph; Fig. 4.2). Genotyping of a random sample of adults from Kennedy Lake confirmed that lateral plate polymorphism is associated with variation at the Ectodysplasin locus (Cano et al. 2006; Colosimo et al. 2005), Eda (Appendix 6).  The observed scarcity of the partial  77 morph corresponded to a significant deficit of Eda heterozygotes, as indicated by a Wright’s FIS coefficient substantially greater than zero (FIS = 0.44, 95% CI [0.19, 0.67] and FIS = 0.36[0.19, 0.52] for adults in 2006 and 2008, respectively).  As a result, genotype frequencies at Eda in breeding adults showed a significant deviation from Hardy-Weinberg equilibrium, HWE, in 2006 and 2008 (Fig. 4.2; χ2 = 12.36 P = 0.002 and χ2 = 17.96 P = 0.0001, respectively).  The heterozygote deficit observed in adults is not associated with genome-wide population structure (Castric et al. 2002). To test this, we genotyped six unlinked neutral microsatellite loci located on linkage groups 2, 6, 8, 9, 18, and 20, in 63 adults collected in the May 2006 breeding season.  In sharp contrast to Eda, all six neutral loci were in HWE (Appendices 7&8).  Individual based population assignment using a Bayesian MCMC approach implemented in STRUCTURE (Pritchard et al. 2000) provided strong support for a single genetic cluster (Pr[K of 1] = 0.99; Fig. 4.3).  Moreover, the  78 differences in allele frequencies at neutral loci contrasted between the low and complete morphs explained less than 1% of the total allelic variation between low and complete morphs (FST=0.008; P= 0.112), which is similar to estimates of FST found within solitary stickleback populations (Bolnick et al. 2008).  This suggests that the stickleback within Kennedy Lake comprise a single, panmictic population.  Assortative mating among Eda genotypes also fails to account for the observed heterozygote deficiency (Barreto and McCartney 2008).  To test this we genotyped the Eda locus from six fertilized eggs (zygotes) collected from each of 71 stickleback nests during breeding season in June 2008.  If the heterozygote deficiency found in adults at Eda reflects assortative mating, then zygotes should show a similar deficiency of Eda  79 heterozygotes.  In contrast to adults, zygotes showed no deviation from HWE (χ2 = 0.433, P = 0.81), nor a significant deficiency in heterozygotes (FIS = 0.061, 95% CI [-0.071, 0.203]), implying that heterozygote deficiency is not generated by assortative mating at the Eda locus. Direct comparison shows that adults exhibit significantly more heterozygote deficiency than zygotes (95% CI [0.08 - 0.51] for difference in Fis between adults and eggs sampled in 2008).  Eda allele frequency differed somewhat between zygotes (frequency of low allele = 0.23) and adults (0.31), but the difference was not statistically significant (χ2 = 0.815; P = 0.37).   80 We estimated the selection pressures necessary to account for the observed change in genotype frequencies between life stages, from zygote to adult and from adult to zygote, across one generation (Fig. 4.4).  The frequency of EdaC/C homozygotes and heterozygotes were similar across life stages (Fig. 4.2); however, the proportion of EdaL/L homozygotes was significantly lower in zygotes when compared to adults (proportion EdaL/L in zygotes = 0.06, 95% CI [0.035, 0.098], in adults = 0.18[0.122, 0.251]).  On this basis we estimate viability selection (survival from zygote to adult) to strongly favour individuals homozygous for low alleles, EdaL/L (s = 0.67, hs = 0.7).  This is consistent with previous studies indicating a growth advantage held by individuals carrying low alleles that likely translates to a survival advantage (Barrett et al. 2008; Marchinko and Schluter 2007).  Indeed, in our Kennedy Lake samples, adult individuals of the low morph were slightly larger in standard length than individuals of the complete morph (difference of means = 1.8 mm; P = 0.048, N = 44 and 45 for the low and complete morph respectively; Appendix 9).  In contrast, selection on breeding adults during the reproductive phase favours complete EdaC/C homozygotes (s = -0.46, hs = 0.37; Fig. 4.4), returning the frequency of EdaL/L homozygotes to the lower levels observed in zygotes (see paragraph above).  Our model makes no assumption about the type of selection – sexual, fecundity, or gametic – during the reproductive phase.  The differences in relative fitness among Eda genotypes at this stage may result from mate preferences or differences in clutch size or sperm viability.  However, differential survival of zygotes prior to hatching, one form of gametic selection, is unlikely to contribute as lateral plate morphs from Kennedy Lake  81 differ little in hatching rate in freshwater (Marchinko and Schluter 2007).  Finally, the combined effect of viability and mating yielded net selection against heterozygotes at the Eda locus (hs = 0.81), favouring EdaL/L homozygotes overall (s = 0.52; Fig. 4.4). Theory suggests that frequency-dependent selection is required to maintain a stable polymorphism with heterozygote disadvantage, favouring each homozygote when rare (Wilson and Turelli 1986).  Niche differentiation among morphologically divergent individuals naturally gives rise to frequency-dependent selection, although it is not the only potential mechanism (Ayala and Campbell 1974).  We explored the possibility of niche differentiation among Eda homozygotes by comparing the ratio of stable isotopes of carbon, δ13C, and nitrogen, δ15N, between low and complete morphs.  In lakes, δ13C indicates the ultimate source of dietary carbon (littoral or pelagic), whereas δ15N estimates trophic position (Post 2002).  Our results revealed significant differences in δ13C between low and complete morphs collected in 2004 (Fig. 4.5; P < 0.0001; Appendix 10).  δ15N were similar between morphs and non-significant (P = 0.8; Appendix 10).  Based upon the δ13C signatures of sticklebacks and three additional lake species (Fig. 4.5), complete morphs feed more often upon prey with carbon derived from littoral sources than low morphs, although both appear to feed predominantly on prey from pelagic food chains.  If these niche differences provide separate density-dependent regulation, then frequency-dependent selection may act to stabilise the observed Eda polymorphism even in the presence of underdominance (Schluter 2003).    82                Polymorphisms maintained in the face of low heterozygote fitness satisfy the basic theoretical conditions required to initiate diversification and maintain genetic variation through either the modification of dominance or the evolution of assortative mating (Wilson and Turelli 1986).  Strong underdominance at a single locus yields the most likely scenario under which assortative mating should develop (Otto et al. 2008).  Subsequent reproductive isolation and speciation is a possible outcome in this population (Dobzhansky 1940; Gavrilets et al. 2007; Otto et al. 2008; Wilson and Turelli 1986).  However the strongly dissimilar allele frequencies  83 observed here may result in the extinction of the rarer allele during the establishment of assortative mating due to the disadvantage that individuals possessing the rare allele face when locating mates that are also rare (Otto et al. 2008).  Alternatively, large inequalities in allele frequencies may result in a greater chance that dominance modification will lead to a reduction of the formation maladaptive heterozygotes (Durinx and Van Dooren 2009).  At this point we are unable to predict which evolutionary trajectory is most likely proceed. Our results supply the first example of a single locus polymorphism maintained despite heterozygote disadvantage.  The maintenance of this polymorphism supports a growing body of work which suggests that selection maintains ecologically important genetic variation in the wild (Fitzpatrick et al. 2007; Mitchell-Olds et al. 2007; Subramaniam and Rausher 2000).  Our ability to detect this process at a single locus highlights the advantage of integrating genetic and ecological tools when studying the mechanisms maintaining genetic variation and promoting diversification in natural populations.  As the integration of genetics and ecology becomes more prevalent in studies of evolutionary divergence (e.g., localization of QTL for ecologically relevant traits) a better understanding of the role of population-level variation driving biological diversification may be achieved.  If the theory is robust, then patterns of underdominance leading to diversification in natural populations will be fundamental towards explaining these evolutionary patterns, particularly in cases when major genes underlie the genetics of adaptive change (Hori 1993; Smith 1993; Ueshima and Asami 2003).  84 METHODS Collections Kennedy Lake is a large (6475 ha), deep (145 m) and extremely oligotrophic lake  (Hyatt and Ringler 1989), lying six meters above sea level and approximately six kilometres from the west coast of Vancouver Island.  We collected adult threespine stickleback in minnow traps near the southern tip of the Clayqout arm of Kennedy Lake on the following dates: May 11, July 9 and Aug 17, 2004, June 11, 2006, and June 12, 2008. Specimens from the May 11, 2004 collection were frozen immediately upon capture using dry ice and transported to the University of British Columbia for storage at -80 °C prior to stable isotope analysis.  These samples were thawed briefly for morphological measurements and then stored at -80 °C until processed for stable isotope analysis. Specimens collected in July and August of 2004, and in 2006 and 2008 were given a lethal dose of tricaine methanesulfonate (MS-222, Syndel Laboratories, Qualicum Beach, BC, Canada) and then preserved in 95% ethanol.  We collected eggs by removing whole nests while snorkeling June 9-12, 2008.  Eggs were preserved in 95% ethanol.  We also collected the individual male guarding each nest (the putative fathers of clutches within a nest) using a dip net while snorkeling.  These guardian males were combined with the minnow trap samples from the same collection date to make up the June 2008 adult collection.  Preserved specimens from a sample obtained on Aug 2, 1965, were available from the University of British Columbia Fish Museum (Catalogue No. UBC 65-0506; collectors G. Haythorne and D. Hagen).  85 Stable Isotope Analysis  For the analysis of stable isotopes of Nitrogen (δ15N) and Carbon (δ13C), we removed tissue from the foot of mussels, and dorsal white muscle from threespine stickleback, peamouth chub and coho parr.  All tissue was freeze-dried and ground to a fine powder with a mortar and pestle.  Isotope samples were run on a Finnigan Delta Plus Advantage stable isotope ratio mass spectrometer at the University of Victoria.  Variation in δ13C of fish tissue can arise from individual variation in lipid content, because lipids have a lower δ13C than proteins and carbohydrates (Kiljunen et al. 2006).  However, we found no evidence for variation in lipid content among morphs (C:N ratio: p>0.05), and following chemical lipid extraction (Ingram et al. 2007), the difference in δ13C between morphs was significant (P = 0.0001), while the factor sex and the interaction between sex and morph were non-significant (P > 0.19).  This is the same pattern of isotopic differentiation between morphs that we observed with the untreated samples and presented in the text. Genotyping Total genomic DNA was isolated using standard phenol-chloroform extraction methods.  Six microsatellite loci, isolated and characterized by the Stanford Genome Research Center, were selected for genetic analyses (Genbank Accession numbers and reference in parentheses); Stn301 (BV678111, unpublished), Stn65 (G72254  (Peichel et al. 2001)), Stn216 (BV102494  (Colosimo et al. 2004)), Stn250 (BV678075, unpublished), Stn387 (BV678140, unpublished), and Stn388 (BV678141, unpublished).  86 A diagnostic in/del locus (isolated from the locus Stn381 within intron six of the Eda gene (Colosimo et al. 2005) was used to identify the genotype of lateral plate morphs. Microsatellite and Eda alleles were amplified in 10 µl PCR reactions containing 5-15 ng genomic DNA, 1 µM of each forward and reverse primer, 1X PCR buffer, 0.25-0.125 mM of each dNTP, 1.5 mM MgCl2, and 0.25U AmpliTaq polymerase (Applied Biosystems).  Cycling conditions were standardized over all loci as follows: 93 0C for 3 min, 95 0C 30 s, 59 0C 30s, 72 0C 30 s, 5 cycles of 94 0C 30 s, 59 0C 30 s, 72 0C 30 s, 35 cycles of 90 0C 30 s, 60 0C 30 s, 72 0C 30 s, followed by 72 0C 10 min, then cooled to 4 0C.  Amplified PCR products were separated by gel electrophoresis on an ABI 3733 automated sequencer using the GS500 Size Standard (Applied Biosystems) and scored with the software GENEMAPPER (Applied Biosystems).  Tests of genetic diversity and Hardy Weinberg equilibrium were estimated with GENEPOP v.4.0 (Rousset 2008). Genetic differentiation in allele frequencies between low and complete morphs was tested in ARLEQUIN (v2.001).  The most likely number of candidate source populations (from K=1 to 3) was assessed with a Bayesian individual-based assignment approach implemented in STRUCTURE  (Pritchard et al. 2000). Reproductive Fitness Model  The model of selection during the reproductive phase of the life cycle assumes that breeding is random with respect to Eda genotype, and that the genotype frequencies of females and males equals those frequencies observed at Eda in randomly sampled adults. Selection in this model results from differences in mate preferences, viability or fecundity  87 among Eda genotypes. Model calculations: p1, p2, p3 represent the observed breeding frequencies of male and female EdaL/L, EdaC/L, and EdaC/C genotypes, respectively. m, n represent the fraction of EdaC/L, and EdaC/C genotypes contributing offspring during the reproductive phase.  This is equivalent to the fitness of EdaC/L, and EdaC/C genotypes relative to EdaL/L genotypes (set to 1).  We then solve for the values of m and n that generate zygote allele frequencies (p', q') equal to those observed in zygotes sampled from wild clutches. q1, q2, q3 represent the zygote frequencies of EdaL/L, EdaC/L, and EdaC/C genotypes, respectively, calculated using the following random mating equations. Genotype frequencies: q1 = (p1• p1) + (0.5 • ((p1 • mp2) + (mp2 • p1)) + (0.25 • (mp2 • mp2))  q2 = (p1 • np3) + (np3 • p1) + (0.5 • ((p1 • mp2) + (mp2 • p1) + (mp2 • mp2) + (mp2 • np3) + (np3 • mp2))  q3 = (np3 • np3) + (0.5 • ((np3 • mp2) + (mp2 • np3)) + (0.25 • (mp2 • mp2))  Allele frequencies: p' = q1 + (0.5 • q2)   88 q' = q3 + (0.5 • q2) Relative Fitness and Selection Coefficients Relative fitness for each Eda genotype was calculated as the change in the observed frequency of a genotype as individuals aged from zygote to adult (adult frequency divided by zygote frequency), standardised to the homozygous low EdaL/L genotype.  The total relative fitness (total selection) of each genotype across one generation was calculated as the product of the relative fitness from viability and reproductive phases, standardised to the homozygous low EdaL/L genotype.  Selection coefficients were calculated as one minus the relative fitness of EdaC/C homozygote (s) and heterozygote (hs) (Hedrick 2000). Statistical Analyses All statistical analyses were carried out using R version 2.8.0 (R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org).  Confidence intervals for the Eda genotype frequencies observed in adults were calculated using the Wilson method from the Hmisc package in R.  Heterozygote deficiency was quantified using Wright’s coefficient of inbreeding (FIS), which was calculated as:    where H is the frequency of the heterozygote class, and p and q represent the allele frequencies (Hedrick 2000).  Confidence intervals of Fis in adults and zygotes were made using bootstrap re-sampling.  Confidence intervals for adult samples from 2006 and 2008 ! FIS =1" H 2pq  89 were calculated from a bootstrap distribution of Fis for each year, generated by re- sampling individuals 10000 times.  Confidence intervals for zygote genotype frequencies and Fis were estimated from a bootstrap distribution of either genotype frequency or Fis (depending on the analysis), generated from 10000 re-samples of the 71 clutches.  We also calculated the 95% confidence interval for the difference in Fis between adults and eggs sampled in 2008.  This interval did not include zero (0.08 - 0.51) confirming that the heterozgyote deficit found in adults was significantly greater than that observed in zygotes.   90 REFERENCES Ayala, F. J., and C. A. Campbell.  1974.  Frequency-dependent selection. Annual Review of Ecology and Systematics 5:115-138. Barreto, F. S., and M. A. McCartney.  2008.  Extraordinary AFLP fingerprint similarity despite strong assortative mating between reef fish color morphospecies. Evolution 62:226-233. Barrett, R. D. H., S. M. Rogers, and D. Schluter.  2008.  Natural selection on a major armor gene in threespine stickleback. Science 322:255-257. Bolnick, D. I., E. J. Caldera, and B. Matthews.  2008.  Evidence for asymmetric migration load in a pair of ecologically divergent stickleback populations. Biological Journal of the Linnean Society 94:273-287. Cano, J. M., C. Matsuba, H. Makinen, and J. Merila.  2006.  The utility of QTL-linked markers to detect selective sweeps in natural populations — a case study of the EDA gene and a linked marker in threespine stickleback. Molecular Ecology 15:4613-4621. Castric, V., L. Bernatchez, K. Belkhir, and F. Bonhomme.  2002.  Heterozygote deficiencies in small lacustrine populations of brook charr Salvelinus Fontinalis Mitchill (Pisces, Salmonidae): a test of alternative hypotheses. Heredity 89:27-35. Colosimo, P. F., K. E. Hosemann, S. Balabhadra, G. Villarreal, M. Dickson, J. Grimwood, J. Schmutz, R. M. Myers, D. Schluter, and D. M. Kingsley.  2005. Widespread parallel evolution in sticklebacks by repeated fixation of Ectodysplasin alleles. Science 307:1928-1933.  91 Colosimo, P. F., C. L. Peichel, K. Nereng, B. K. Blackman, M. D. Shapiro, D. Schluter, and D. M. Kingsley.  2004.  The genetic architecture of parallel armor plate reduction in threespine sticklebacks. PLoS Biology 2:1-7. Dobzhansky, T.  1940.  Speciation as a stage in evolutionary divergence. American Naturalist 74:312-321. Durinx, M., and T. J. M. Van Dooren.  2009.  Assortative mate choice and dominance modification:  alternative ways of removing heterozygote disadvantage. Evolution 63:306-323. Fitzpatrick, M. J., E. Feder, L. Rowe, and M. B. Sokolowski.  2007.  Maintaining a behaviour polymorphism by frequency-dependent selection on a single gene. Nature 447:210-U5. Gray, R. H. and D. D. Dauble. 2001.  Some life history characteristics of cyprinids in the Hanford Reach, mid-Columbia River. Northwest Science 75: 122-136. Gavrilets, S., A. Vose, M. Barluenga, W. Salzburger, and A. Meyer.  2007.  Case studies and mathematical models of ecological speciation. 1. Cichlids in a crater lake. Molecular Ecology 16:2893-2909. Hedrick, P. W.  2000.  Genetics of Populations. Jones and Bartlett Inc., Sudbury, MA, USA. Hori, M.  1993.  Frequency-dependent natural-selection in the handedness of scale-eating cichlid fish. Science 260:216-219. Hyatt, K. D., and N. H. Ringler.  1989.  Egg cannibalism and the reproductive srategies of threespine sicklebacks (Gasterosteus aculeatus) in a coastal British Columbia lake. Canadian journal of Zoology 67:2036-2046.  92 Ingram, T., B. Matthews, C. Harrod, T. Stephens, J. Grey, R. Markel, and A. Mazumder. 2007.  Lipid extraction has little effect on the delta N-15 of aquatic consumers. Limnology And Oceanography-Methods 5:338-343. Kiljunen, M., J. Grey, T. Sinisalo, C. Harrod, H. Immonen, and R. I. Jones.  2006.  A revised model for lipid-normalizing delta C-13 values from aquatic organisms, with implications for isotope mixing models. Journal Of Applied Ecology 43:1213-1222. Marchinko, K. B., and D. Schluter.  2007.  Parallel evolution by correlated response: Lateral plate reduction in threespine stickleback. Evolution 61:1084-1090. Mitchell-Olds, T., J. H. Willis, and D. B. Goldstein.  2007.  Which evolutionary processes influence natural genetic variation for phenotypic traits? Nature Reviews Genetics 8:845-856. Otto, S. P., M. R. Servidio, and S. L. Nuismer.  2008.  Frequency-dependent selection and the evolution of assortative mating. Genetics 179:2091-2112. 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. Post, D.  2002.  Using stable isotopes to estimate trophic position:  Models, methods, and assumptions. Ecology 83:703-718. Pritchard, J. K., M. Stephens, and P. Donnelly.  2000.  Inference of population structure using multilocus genotype data. Genetics 155:945-959. Rousset, F.  2008.  GENEPOP’007: a complete re-implementation of the GENEPOP software for Windows and Linux. Molecular Ecology Resources 8:103-106.  93 Rueffler, C., T. J. M. Van Dooren, O. Leimar, and P. A. Abrams.  2006.  Disruptive selection and then what? Trends in Ecology & Evolution 21:238-245. Schluter, D.  2003.  Frequency dependent natural selection during character displacement in sticklebacks. Evolution 57:1142-1150. Smith, T. B.  1993.  Disruptive selection and the genetic-basis sf bill size polymorphism in the african finch Pyrenestes. Nature 363:618-620. Subramaniam, B., and M. D. Rausher.  2000.  Balancing selection on a floral polymorphism. Evolution 54:691-695. Ueshima, R., and T. Asami.  2003.  Single-gene speciation by left-right reversal. Nature 425:679-679. Wilson, D. S., and M. Turelli.  1986.  Stable underdominance and the evolutionary invasion of empty niches. American Naturalist 127:835-850.    94 CHAPTER 5. GENERAL CONCLUSIONS  In chapters 2 and 3 I tested two of the ecological factors thought to drive armour divergence between marine and freshwater populations of threespine stickleback: the ion and predation hypotheses.  In chapter two I’ve shown that lateral plate morphs differ little in hatching success, but strongly in juvenile growth rate when raised at different salinities.  Offspring of parents with fewer plates grew substantially faster than offspring of parents with many lateral plates in fresh water.  Growth rates differed very little in salt water.  Results were consistent in offspring from two populations, one marine and one fresh water, strengthening the conclusion that lateral plates are disadvantageous in fresh water habitats.  These tests were restricted to comparisons between morphs collected from the same population, minimizing any additional adaptations to local salinity possibly confounding previous work that demonstrated differences in hatching and survival between plate morphs when exposed to water of different salinity (Heuts 1947; Schaarschmidt et al. 1999).  The results of Chapter 2 suggest that the repeated loss of lateral plates in fresh water may have arisen through a correlated response to selection for faster growth during lateral plate development.  Faster growth rates may translate into higher fitness through the positive effects that body size has on overwinter survival, clutch size and predator vulnerability (Arendt 1997).  In this case, the reduction of lateral plates in fresh water could be non-adaptive, with plate number decreasing because of the fitness advantages of larger size.  Chapter 2 does not, however, distinguishing whether a difference between morphs in the ability to maintain ionic homeostasis (regulating Na+, or Cl- exchange) or calcium limitation is responsible for differences in growth rate between plate morphs.  Gjb1, a  95 gene associated with salt secretion in animals lies in genomic region near Eda in the F2 progeny of a marine by fresh water cross (Colosimo et al. 2005), making it plausible that linkage between Eda and Gjb1 results in differences in ion regulatory ability between plate morphs.  However, in the polymorphic Kennedy Lake population, the recombination that has occurred over thousands of generations since being colonized by marine stickleback should have broken down any linkage between Eda and Gjb1, dissolving the correlation between lateral plates and growth rate.  The observed association between growth rate and lateral plate number in the polymorphic Kennedy Lake population might thus indicate that pleiotropic effects of Eda, not linkage between Eda and other genes, results in growth differences between plate morphs.  If pleiotropy is responsible, as suggested by Barrett et al. (Barrett et al. 2008; Cresko 2008), then calcium limitation may be a more likely mechanism by which bony lateral plates are negatively associated with growth rate in freshwater only.  Measuring the growth rates of different lateral plate morphs raised at different concentrations of calcium in fresh water will be required to test this alternative.  In Chapter 3 I tested the hypothesis that predator-driven divergent selection causes differentiation in defensive armor between marine and freshwater threespine stickleback.  Using F2 families from crosses between freshwater and marine populations, I demonstrated that predaceous insects impose selection for larger juvenile size, and shorter dorsal spine and pelvic girdle after accounting for size.  Targeted genotyping of the gene associated with building armour plates, Eda, revealed that freshwater Eda alleles are increase relative to marine alleles in the presence of insect predators.  Note however, that the experiment was conducted before the number of amour plates had reached its  96 final number during development.  Predaceous aquatic insects likely contribute to the reduction of armour in freshwater populations of threespine stickleback via direct selection on spine length.  Insects prey most heavily on the smallest individuals and we have evidence that as juveniles, the complete morph grows more slowly than the low morph.  The results from both Chapters 2 and 3 suggest that reduction of lateral plate number in fresh water may result from the indirect effect of insects preying most heavily on smaller juveniles through the association of bony armour and growth rate.  Although Chapter 3 documented selection on armour after adjusting for differences in body size, the correlation between armour traits obscures which trait or traits are the target of selection.  Indeed, the correlation of anterior dorsal spine demonstrates our inability to distinguish whether spine length, Eda, or both are under selection.  This problem also illustrates our inability to attribute whether the selection at Eda is due to the effects of Eda itself, or whether it results from linkage disequilibrium between Eda and additional armour or shape genes, or genes of unknown phenotypic or physiological effects.  Distinguishing which trait or set of traits influence predator induced mortality rates could be accomplished with careful trait manipulations, or selection experiments using advanced generation hybrids between marine and freshwater populations, instead of F2 hybrids used here, in order to breaking down linkage disequilibrium between genes affecting different traits.  One final note on Chapter 3 concerns the level of significance found for the association between pelvic girdle loss and insect predation and between Eda allele frequency and insect predation.  Tests revealed that pelvic girdle absence and Eda allele frequency were nearly significant at the 0.05 alpha level when using a two-tailed t-test (P  97 = 0.08 and 0.052 for pelvic girdle and Eda respectively).  Because of the moderate number of tests performed in this study (16) it is possible that one or both may be nearly significant by random chance.  As a result it should be acknowledged that both the association of pelvic girdle absence and insect predation rate, and Eda allele frequency and insect predation rate could be specific to this study.  Concluding that predaceous aquatic insects have played a significant role in the loss of the pelvic girdle, and the sweep of the low Eda allele in fresh water requires further examination.  In Chapter 4 I explored the ecological and genetic mechanisms maintaining phenotypic variation within a wild population.  In this chapter I demonstrated an extremely rare example of a single locus polymorphism being maintained despite heterozygote disadvantage.  Single locus underdominance is predicted to lead to the evolution of dominance or assortative mating (Durinx and Van Dooren 2009; Otto et al. 2008), yet no evidence of either pattern was found.  Monitoring long-term changes in genotype frequencies in this population may eventually capture and identify the genetic and ecological conditions resulting in dominance or possibly the early stages of reproductive isolation.  Niche separation between Eda homozygotes, observed as differentiation in carbon isotopes, suggests that frequency-dependent selection is involved in the maintenance of polymorphism despite heterozygous disadvantage.  Both resource and predator density may drive frequency-dependent selection in this system and independent confirmation that frequency-dependence maintains this polymorphism despite heterozygote disadvantage is needed.  This particular system could provide an explicit test of the theoretical prediction that frequency-dependent selection can maintain underdominant polymorphism in both lab and natural settings, through short-term  98 experimental and long-term field monitoring approaches.  For example, experimental manipulations of planktonic vs. benthic food availability and its effect on growth rate and mortality of individuals differing in Eda alleles would be useful for testing frequency-dependent selection in response to resource availability.   99 REFERENCES  Arendt, J. D.  1997.  Adaptive intrinsic growth rates: An integration across taxa. Quarterly Review Of Biology 72:149-177. Barrett, R. D. H., S. M. Rogers, and D. Schluter.  2008.  Natural selection on a major armor gene in threespine stickleback. Science 322:255-257. Colosimo, P. F., K. E. Hosemann, S. Balabhadra, V. J. Guadalupe, M. Dickson, J. Grimwood, J. Schmutz, R. M. Myers, D. Schluter, and D. M. Kingsley.  2005. Widespread parallel evolution in sticklebacks by repeated fixation of Ectodysplasin alleles. Science 307:1928-1933. Cresko, W. A.  2008.  Armor development and fitness. Science 322:204-206. Durinx, M., and T. J. M. Van Dooren.  2009.  Assortative mate choice and dominance modification:  alternative ways of removing heterozygote disadvantage. Evolution 63:306-323. Heuts, M. J.  1947.  Experimental studies on adaptive evolution in Gasterosteus aculeatus L. Evolution 1:89-102. Otto, S. P., M. R. Servidio, and S. L. Nuismer.  2008.  Frequency-dependent selection and the evolution of assortative mating. Genetics 179:2091-2112. Schaarschmidt, T., E. Meyer, and K. Jurss.  1999.  A comparison of transport-related gill enzyme activities and tissue-specific free amino acid concentrations of Baltic Sea (brackish water) and freshwater threespine sticklebacks, Gasterosteus aculeatus, after salinity and temperature acclimation. Marine Biology 135:689-697.    100 APPENDICES Appendix 1. Locations of sampling sites associated with chapter three, Figure 3.1. Numbers connected to symbols indicate the populations used in this chapter three:  Klein Lake (1), McKay Lake (2), Priest Lake benthic (3), Paxton Lake limnetic (4), Priest Lake limnetic (5),  Paxton Lake benthic (6), Paq Lake (7), North Lake (8), Cranby Lake (9), Dougan Lake (10), Beaver Lake (11), Blackjack Lake (12), Mayer Lake (13), Brannen Lake (14), Erroch Lake (15), Kennedy Lake (16),  Fairy Lake (17), Sproat Lake (18), Oyster lagoon marine (19), Little Campbell River marine (20).  All sites are located in British Columbia, Canada.    101 Appendix 2.  Supplementary methods from chapter three. Crossing, fertilization and fish husbandry  The crossing of wild caught individuals from parental populations to generate F1 hybrid lines was accomplished using standard laboratory protocol.  Fertilization was performed by first stripping a female’s eggs into a Petri dish containing a small amount of de-chlorinated tap water.  Next, a male was anaesthetized in clove oil, both testes were removed and placed into the Petri dish with the eggs, and then crushed to release sperm. The macerated testes were then removed from the Petri dish after 20 minutes, the eggs transferred to plastic egg-cups (pint cup with the bottom replaced with fine fiberglass mesh) and submerged in an egg-tank (20 L) containing methylene blue to reduce fungal growth.  Eggs remained in aerated egg-tanks for eight days, after which they were transferred to a 102 L tank. F1 hybrid lines were raised from hatchlings to adults in 102 L tanks.  Hatchlings were fed newly hatched live brine shrimp twice a day for the first six weeks, after which diets were supplemented with chopped frozen blood worms.  At about three months post-hatching, F1 lines were split into two to four 102 L tanks to maintain roughly 20 individuals per tank throughout the following winter, during which time individuals were fed frozen bloodworms once daily.  The following spring, when F1 males began to show nuptial coloration and females were gravid, brother-sister mating pairs were chosen from each F1 hybrid line. Using the protocol described above, eggs from an F1 female were stripped and fertilized with sperm from an F1 male of the same family to establish an F2 family to be used in the predation trials.  One mating pair from each of the six F1 hybrid lines was created from  102 the PBxOLM cross, resulting in six F2 families.  A total of ten brother-sister mating pairs were chosen from the single MLxLCM F1 hybrid line, resulting in ten F2 families. F2 hatchlings were fed newly hatched live brine shrimp twice a day and raised for about three weeks in 102 L tanks until they reached about 11 mm standard length.  At this time, hatchlings were transported from the laboratory to the UBC experimental ponds to begin the predation experiment.  103 Appendix 3.  Family sample sizes from each F2 family in the Paxton and McKay lines found in chapter three.                Paxton lines family 1 2 sample size 3 4 5 6 McKay lines 1 2 3 4 5 6 7 8 9 10 48(40) 48(20) before(after) before(after) 49(45) 42(42) 36(20) 31(20) 40(11) 41(31) 70(34) 21(19) 20(9) 32(28) 32(17) PredationNo-predation sample size 37(31) 37(25) 49(24) 43(21) 36(31) 36(14) 32(30) 32(21) 29(29) 28(12) 36(25) 36(15) 43(31) 44(19) 70(66) 35(7) 31(7) 39(6) 41(20)  104 Appendix 4. Means, standard deviations and selection differentials for standard length and all size-adjusted armor traits from each F2 family in the Paxton and McKay lines.                       Paxton lines selection Trait family Mean (XB) S.D. Mean (XA) S.D. differential (i) 1 14.708 1.390 15.051 1.003 0.247 2 13.909 1.192 14.217 0.895 0.258 3 13.556 1.229 14.845 1.061 1.049 4 12.451 1.130 12.847 0.995 0.351 5 15.107 0.805 16.323 1.068 1.511 6 13.967 1.616 14.702 0.944 0.455 1 0.023 0.114 -0.043 0.137 -0.577 2 0.008 0.118 -0.015 0.116 -0.194 3 0.015 0.116 -0.044 0.128 -0.511 4 0.006 0.136 -0.017 0.104 -0.164 5 0.012 0.131 -0.022 0.154 -0.256 6 0.011 0.116 -0.017 0.140 -0.236 1 0.032 0.124 -0.060 0.137 -0.741 2 0.006 0.129 -0.011 0.125 -0.132 3 0.000 0.101 0.000 0.121 0.002 4 0.007 0.095 -0.019 0.098 -0.271 5 -0.005 0.093 0.009 0.097 0.145 6 -0.008 0.157 0.012 0.120 0.128 1 0.045 0.249 -0.072 0.205 -0.470 2 0.021 0.186 -0.037 0.203 -0.313 3 0.036 0.159 -0.107 0.227 -0.896 4 -0.025 0.174 0.072 0.141 0.557 5 -0.054 0.057 0.080 0.195 2.337 6 -0.020 0.148 0.030 0.181 0.335 1 -0.018 0.776 0.033 0.480 0.065 2 0.054 0.207 -0.086 0.481 -0.676 3 0.063 0.479 -0.152 0.698 -0.450 4 -0.054 0.211 0.153 0.201 0.986 5 0.013 0.721 -0.018 0.982 -0.043 6 -0.015 0.120 0.023 0.181 0.314 PredationNo-predation standard length anterior dorsal spine length second dorsal spine length pelvic spine length pelvic girdle length  105 Appendix 4 continued.                       McKay lines selection Trait family Mean (XB) S.D. Mean (XA) S.D. differential (i) 1 15.359 1.696 15.092 1.191 -0.157 2 14.853 2.144 15.520 0.811 0.311 3 17.443 1.039 17.558 1.206 0.110 4 15.131 1.452 15.702 1.154 0.393 5 17.945 0.897 17.544 1.091 -0.447 6 16.067 1.579 15.553 1.145 -0.325 7 14.447 1.159 14.903 1.024 0.394 8 19.369 2.147 20.513 1.462 0.533 9 17.944 1.934 18.737 1.990 0.410 10 14.898 1.089 16.021 1.487 1.032 1 -0.007 0.161 0.009 0.131 0.099 2 -0.041 0.221 0.010 0.101 0.230 3 0.006 0.157 -0.009 0.127 -0.096 4 0.009 0.104 -0.021 0.087 -0.288 5 0.023 0.112 -0.038 0.094 -0.542 6 0.006 0.138 -0.009 0.111 -0.110 7 0.009 0.111 -0.018 0.137 -0.244 8 0.018 0.119 -0.037 0.096 -0.460 9 0.004 0.157 -0.007 0.192 -0.067 10 -0.009 0.110 0.018 0.119 0.244 1 -0.010 0.163 0.012 0.130 0.132 2 -0.001 0.115 0.002 0.127 0.028 3 0.023 0.144 -0.033 0.152 -0.393 4 0.004 0.106 -0.010 0.107 -0.139 5 0.034 0.107 -0.057 0.158 -0.852 6 -0.008 0.137 0.013 0.122 0.155 7 0.006 0.119 -0.012 0.127 -0.156 8 -0.006 0.164 0.013 0.140 0.116 9 0.017 0.147 -0.028 0.179 -0.307 10 0.004 0.130 -0.009 0.159 -0.103 No-predation Predation second dorsal spine length standard length anterior dorsal spine length  106 Appendix 4 continued.                       McKay lines selection Trait family Mean (XB) S.D. Mean (XA) S.D. differential (i) 1 -0.016 0.241 0.019 0.262 0.144 2 -0.010 0.238 0.022 0.215 0.132 3 -0.010 0.269 0.014 0.284 0.086 4 0.054 0.212 -0.129 0.228 -0.862 5 0.034 0.202 -0.056 0.269 -0.446 6 0.040 0.312 -0.065 0.263 -0.338 7 0.015 0.231 -0.029 0.202 -0.193 8 -0.027 0.256 0.058 0.250 0.333 9 0.053 0.314 -0.087 0.195 -0.445 10 0.001 0.264 -0.002 0.271 -0.012 1 0.001 0.166 -0.001 0.158 -0.010 2 -0.005 0.180 0.012 0.175 0.094 3 0.036 0.169 -0.052 0.119 -0.520 4 0.015 0.105 -0.036 0.148 -0.488 5 0.031 0.162 -0.051 0.184 -0.508 6 -0.024 0.190 0.040 0.188 0.338 7 0.015 0.161 -0.030 0.175 -0.278 8 0.063 0.162 -0.133 0.119 -1.211 9 0.035 0.213 -0.058 0.188 -0.439 10 0.013 0.185 -0.026 0.181 -0.214 No-predation Predation pelvic spine length pelvic girdle length  107 Appendix 5. Change in frequency (∆ p) of the low morph Ectodysplasin allele, EdaL, between no-predator control and predator treatments in F2 families generated from the Paxton lines.     family ! p t0.05(1)5 P 1 0.145 2 0.247 3 0.025 4 0.050 5 0.250 6 -0.002 -2.528 0.026  108 Appendix 6. The association between Eda genotype and lateral plate morph phenotype pooled across all sample years.  Fishers' exact test P = 2.2*10-16.       Eda  genotype Complete Partial Low CC 104 3 3 CL 42 7 3 LL 0 2 34 Plate morph  109 Appendix 7. Locus specific tests for Hardy-Weinberg equilibrium (HWE) and inbreeding coefficients (FIS) at six neutral microsatellite loci genotyped in adult stickleback collected in May 2006 from Kennedy Lake.  Each molecular marker used is not known to be linked to quantitative trait loci for any morphological traits.  Bold values indicated p-value remained significant after a sequential Bonferroni adjustment.                  Locus Number of alleles Ho He HWE (p-value) FIS Linkage group Stn381 3 0.355 0.624 < 0.0001 0.433 4 Stn301 32 0.871 0.956 0.2223 0.0874 18 Stn065 14 0.717 0.664 0.6383 -0.0807 6 Stn216 7 0.452 0.416 0.843 -0.1193 20 Stn250 37 0.919 0.946 0.9958 0.0286 8 Stn387 13 0.435 0.535 0.0423 0.1705 2 Stn388 7 0.71 0.693 0.9196 -0.0363 9  110 Appendix 8. Pairwise tests for linkage disequilibrium.          Locus#1 Locus#2 p-value SE STN381 STN301 1.0000 0.0000 STN381 STN065 0.5210 0.0330 STN301 STN065 1.0000 0.0000 STN381 STN216 0.1815 0.0200 STN301 STN216 0.2193 0.0760 STN065 STN216 0.8353 0.0432 STN381 STN250 1.0000 0.0000 STN301 STN250 1.0000 0.0000 STN065 STN250 1.0000 0.0000 STN216 STN250 0.5715 0.0733 STN381 STN387 0.2998 0.0346 STN301 STN387 0.9879 0.0121 STN065 STN387 0.9253 0.0392 STN216 STN387 0.2925 0.0490 STN250 STN387 0.8391 0.0700 STN381 STN388 0.8859 0.0104 STN301 STN388 1.0000 0.0000 STN065 STN388 0.7840 0.0590 STN216 STN388 0.6856 0.0505 STN250 STN388 1.0000 0.0000 STN387 STN388 0.0929 0.0476  111 Appendix 9.  Results from ANOVA testing for effects of lateral plate morph, sex and their interaction on standard length, a measure of overall body size, from randomly sampled adults trapped in May of 2004.                     source of variation df mean square F P lateral plate morph 1 0.02588 4.0392 0.04763 sex 1 0.00137 0.2131 0.64557 morph * sex 1 0.01473 2.2995 0.13313 residuals 85 0.00641  112 Appendix 10. Results from two-factor ANOVA testing the effects of lateral plate morphotype and sex on the ratio of stable isotopes δ13C and δ15N.             source of variation df mean square F P 13C ‰ lateral plate morph 1 30.83 17.5637 < 0.0001 sex 1 1.046 0.5957 0.4424 morph * sex 1 2.924 1.6657 0.2003 residuals 85 1.755 15N ‰ lateral plate morph 1 0.0097 0.0637 0.80128 sex 1 0.4566 3.0054 0.08661 morph * sex 1 0.0174 0.1148 0.73556 residuals 85 0.1519  113 Appendix 11. Animal care certificates.                                         114 Appendix 11. continued                                              115 Appendix 11. continued                                              116 Appendix 11. continued                                              117 Appendix 11. continued              118 Appendix 11. continued  

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}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            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:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0067513/manifest

Comment

Related Items