Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Detecting the drivers of divergence : identifying and estimating natural selection in threespine stickleback Rennison, Diana Jessie 2016

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

Item Metadata

Download

Media
24-ubc_2016_september_rennison_diana.pdf [ 7.39MB ]
Metadata
JSON: 24-1.0307148.json
JSON-LD: 24-1.0307148-ld.json
RDF/XML (Pretty): 24-1.0307148-rdf.xml
RDF/JSON: 24-1.0307148-rdf.json
Turtle: 24-1.0307148-turtle.txt
N-Triples: 24-1.0307148-rdf-ntriples.txt
Original Record: 24-1.0307148-source.json
Full Text
24-1.0307148-fulltext.txt
Citation
24-1.0307148.ris

Full Text

DETECTING THE DRIVERS OF DIVERGENCE: IDENTIFYING AND ESTIMATING NATURAL SELECTION IN THREESPINE STICKLEBACK  by Diana Jessie Rennison   M.Sc., The University of Victoria, 2010 B.Sc.H., The University of Victoria, 2008   A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Zoology) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  July 2016  © Diana Jessie Rennison, 2016   ii Abstract Differences in ecological factors between habitats drive evolutionary divergence and can lead to the generation of new species. While many studies provide evidence suggesting a given trait or genetic locus is adaptive few studies are able to elucidate the direct mechanisms responsible for differences in fitness. I use experimental and observational studies in threespine stickleback (Gasterosteus aculeatus) to identify factors that have driven divergence and to disentangle the mechanisms by which differential fitness arises.  To disentangle direct selection on a phenotype from selection on correlated characters encoded by its underlying gene, I applied statistical methods for the estimation of selection on correlated characters (Chapter 2). These data provided the first evidence of direct selection on the lateral plate phenotype and suggested that pleiotropy at the Ectodysplasin locus is likely an important factor driving the rapid and repeated evolution of armour phenotypes.  Spectral sensitivity is thought to evolve to match features of the local light environment. Marine and freshwater threespine stickleback inhabit divergent light environments and therefore provide an opportunity to test the hypothesis of spectral matching. I surveyed the opsin gene expression and spectral sensitivity of multiple marine and freshwater populations to test this hypothesis (Chapter 3). While I find weak evidence for spectral matching, I do find evidence suggesting adaptive divergence of spectral sensitivity between populations inhabiting different light environments.  Competition is widely appreciated to play a direct role in driving trait divergence; however, it can also have indirect effects mediated through differential exposure to predators. To test for the contribution of species interactions to phenotypic and genetic divergence I conducted an experiment that manipulated exposure of threespine stickleback to a predator, coastal   iii cutthroat trout (Chapter 4). After one generation of differential exposure to trout there was evidence of phenotypic and genetic divergence between treatments. These results suggest that cutthroat trout are an important source of divergent selection between populations of threespine stickleback.  These studies suggest that bony armour and visual sensitivity are locally adapted and that pleiotropy and genetic architecture likely play an important role in determining evolutionary trajectory during the adaptation of threespine stickleback to freshwater habitats.       iv Preface A version of Chapter 2 has been published as “Rennison, D.J., Heilbron, K., Barrett, R.D.H., Schluter, D. 2015. Discriminating selection on lateral plate phenotype and its underlying gene, Ectodysplasin, in threespine stickleback. American Naturalist, 185:150-156”. D. Rennison, K. Heilbron, R. Barrett and D. Schluter conceived the project. D. Rennison and K. Heilbron performed data collection and data analysis. D. Rennison wrote the manuscript with input from D. Schluter and edits from K. Helibron and R. Barrett.    A version of Chapter 3 has been published as “Rennison, D.J., Owens, G.L., Heckman, N. Schluter, D., Veen, T. 2016. Rapid evolution of colour vision in the threespine stickleback adaptive radiation. Proceedings of the Royal Society B, 283(1830): 20160242”. D. Rennison, D. Schluter and T. Veen conceived the project. D. Rennison, G. Owens, and T. Veen collected the data. D. Rennison, T. Veen and N. Heckman performed the data analysis. T. Veen and N. Heckman developed the new statistical metrics presented in the chapter. D. Rennison wrote the manuscript with input from T. Veen and D. Schluter, N. Heckman and G. Owens contributed to the editing.    For Chapter 4, D. Rennison and D. Schluter conceived the project. D. Rennison bred and reared the experimental crosses. D. Rennison and S. Rudman ran the experiment and conducted the pond sampling. D. Rennison, J. Best and M. Lo did the DNA extractions. D. Rennison measured morphological phenotypes and made the GBS libraries. D. Rennison did the bioinformatics analysis to determine SNP genotypes and implemented the selection analyses. D. Schluter performed the QTL mapping. D. Rennison wrote the manuscript with input from D. Schluter.     v The following permits were obtained for the collection of wild threespine stickleback (Gasterosteus aculeatus): Species at Risk Act collection permit number 236 and British Columbia Fish Collection permit NA-SU12-76311. Permission to care for and use threespine stickleback for the studies herein was granted by the University of British Columbia Animal Care Certificate A11 - 0402. This work was supported by grants and fellowships from the National Science and Research Council of Canada, the Faculty of Graduate Studies at UBC, the Department of Zoology at UBC, and the Society for Integrative and Comparative Biology.     vi Table of Contents Abstract .......................................................................................................................................... ii	Preface ........................................................................................................................................... iv	Table of Contents ......................................................................................................................... vi	List of Tables ............................................................................................................................... xii	List of Figures ............................................................................................................................. xiii	Acknowledgements ...................................................................................................................... xv	Chapter 1: General Introduction .................................................................................................1	1.1	 Advances in the Study of Natural Selection and Local Adaptation. .................................. 1	1.2	 Study System: the Threespine Stickleback ........................................................................ 5	1.3	 Research Conducted for this Dissertation .......................................................................... 8	Chapter 2: Discriminating Selection on Lateral Plate Phenotype and its Underlying Gene, Ectodysplasin, in Threespine Stickleback. .................................................................................11	2.1	 Introduction ...................................................................................................................... 11	2.2	 Materials and Methods ..................................................................................................... 14	2.2.1	 Pond Experiment ....................................................................................................... 14	2.2.2	 Lateral Plate Phenotype ............................................................................................ 15	2.2.3	 Selection Analysis ..................................................................................................... 17	2.3	 Results .............................................................................................................................. 20	2.4	 Discussion ........................................................................................................................ 22	Chapter 3: Rapid Adaptive Evolution of Colour Vision in the Threespine Stickleback Radiation. ......................................................................................................................................26	3.1	 Introduction ...................................................................................................................... 26	  vii 3.2	 Materials and Methods ..................................................................................................... 29	3.2.1	 Sampling ................................................................................................................... 29	3.2.2	 Opsin Expression and Visual Sensitivity .................................................................. 30	3.2.3	 Association Between Visual Sensitivity and Ambient Light .................................... 32	3.3	 Results .............................................................................................................................. 34	3.3.1	 Opsin Expression and Spectral Sensitivity ............................................................... 34	3.3.2	 Laboratory Rearing ................................................................................................... 38	3.3.3	 Association Between Shifts in Visual Sensitivity and Ambient Light ..................... 39	3.3.4	 Match of Visual Sensitivity to Ambient Light .......................................................... 43	3.4	 Discussion ........................................................................................................................ 44	Chapter 4: Survival in a Cutthroat World: Estimating Natural Selection on Armour Phenotypes and Genotypes in Threespine Stickleback. ...........................................................50	4.1	 Introduction ...................................................................................................................... 50	4.2	 Materials and Methods ..................................................................................................... 53	4.2.1	 Experimental Design ................................................................................................. 53	4.2.2	 Phenotyping .............................................................................................................. 56	4.2.3	 Genotyping ................................................................................................................ 57	4.2.4	 Linkage and Quantitative Trait Locus Mapping ....................................................... 58	4.2.5	 Selection Analyses .................................................................................................... 60	4.3	 Results .............................................................................................................................. 62	4.3.1	 Phenotype .................................................................................................................. 62	4.3.2	 Genotype ................................................................................................................... 67	4.4	 Discussion ........................................................................................................................ 70	  viii 4.4.1	 Species Interactions and Divergence ........................................................................ 70	4.4.2	 Temporal Variation in Selection ............................................................................... 71	4.4.3	 Selection on Underlying Genes ................................................................................ 72	Chapter 5: General Discussion and Future Directions ............................................................75	5.1	 Local Adaptation of Threespine Stickleback ................................................................... 75	5.2	 Broader Implications ........................................................................................................ 76	5.2.1	 Adaptation to Freshwater .......................................................................................... 76	5.2.2	 Ecological Speciation ................................................................................................ 77	5.2.3	 Secondary Effects of Species Interactions ................................................................ 79	5.2.4	 Limitations of Threespine Stickleback as a Study System ....................................... 80	5.3	 Future Directions for the Study of Adaptation ................................................................ 81	5.3.1	 Strength of Natural Selection .................................................................................... 81	5.3.2	 The Genetic Basis of Adaptation .............................................................................. 83	Bibliography .................................................................................................................................86	Appendices ..................................................................................................................................101	Appendix A - Supplementary Materials for Chapter 2 ........................................................... 101	A.1	 Relationship Between Length and Number of Lateral Plates. .................................. 101	A.2	 Change in Size-Corrected Lateral Plate Number for all Genotypes Across the Sampling Period. ................................................................................................................. 102	A.3	 Standardized Partial Selection Coefficients for the Lateral Plate Phenotype, Eda Genotype, and Dominance for the September to October and October to November periods... .............................................................................................................................. 102	A.4	 Variance in Lateral Plate Number for each Eda Genotype by Month. ..................... 103	  ix A.5	 Variance – Covariance Matrix for the Standardized Traits Included in the One-Genotype Variable Lande-Arnold Analysis for the September – October Period. ............. 103	A.6	 Variance – Covariance Matrix for the Standardized Traits Included in the One-Genotype Variable Lande-Arnold Analysis for the October – November Period. ............. 103	A.7	 Variance – Covariance Matrix for the Standardized Traits Included in the Two-Genotype Variable Lande-Arnold Analysis for the September – October Period. ............. 103	A.8	 Variance – Covariance Matrix for the Standardized Traits Included in the Two-Genotype Variable Lande-Arnold Analysis for the October – November Period. ............. 104	Appendix B - Supplementary Materials and Methods for Chapter 3 ..................................... 105	B.1	 Collection and Site Information ................................................................................ 105	B.2	 RT-qPCR Protocol .................................................................................................... 106	B.3	 Deriving Spectral Sensitivity .................................................................................... 108	B.4	 Plasticity in the Laboratory Environment ................................................................. 108	B.5	 Association Between Differences in Spectral Sensitivity and Ambient Light .......... 109	B.6	 Effect of Changing Chromophore or Reference Population in the Analyses of Differences in Sensitivity and Differences in Light Environment ...................................... 112	B.7	 Quantification of Correlation Between Differences in Spectral Sensitivity and Differences in Local Light Transmission (A) and Irradiance (B) for Marine and Freshwater Populations using Little Campbell Reference Population. ................................................. 114	B.8	 Quantification of Correlation Between Differences in Spectral Sensitivity and Differences in Local Light Transmission (A) and Irradiance (B) for Benthic and Limnetic Populations using Benthic Reference. ................................................................................ 115	  x B.9	 Effect of Changing Chromophore in the Analysis of the Correlation Between Spectral Sensitivity and Ambient Light (Spectral Matching) ........................................................... 115	B.10	 Stickleback Populations used, their Locations, and Sample Sizes (number of fish) for Opsin Expression and Environmental Light Condition. ..................................................... 116	B.11	 Primer, Probe and Amplicon Sequences from RT-qPCR Assays ........................... 117	B.12	 Mean Correlation Between the Change in Spectral Sensitivity and the Shift in Ambient Light from Marine to Freshwater under Various Chromophore Scenarios. ........ 119	B.13	 Mean Correlation Between the Change in Spectral Sensitivity and Shift in Ambient Light, from Limnetic to Benthic Environment (Limnetic Reference), and from Benthic to Limnetic Environment (Benthic reference). ....................................................................... 120	B.14	 Mean Correlation Between Spectral Sensitivity and Local Light Environment (measured as transmission and irradiance) under Various Chromophore Scenarios. ......... 121	B.15	 Estimated Mean Spectral Sensitvity of Benthic and Limnetic Populations.  .......... 122	Appendix C - Supplementary Materials for Chapter 4 ........................................................... 123	C.1	 Genetic Map Estimated from Four F1 Families. ....................................................... 123	C.2	 Quantitative Trait Locus Map for First Dorsal Spine and Pelvic Spine Length across all F1 Families. .................................................................................................................... 124	C.3	 Quantitative Trait Locus Map for First Dorsal Spine Length for each F1 Family. ... 125	C.4	 Percent Variance Explained and Significance of Treatment Effect for First Dorsal Spine, Pelvic Spine and Pelvic Girdle Length in each F1 Family ....................................... 126	C.5	 Variance – Covariance Matrix for the Size Adjusted Armour Phenotypes in September 2012. ................................................................................................................. 127	  xi C.6	 Variance – Covariance Matrix for the Size Adjusted Armour Phenotypes in January 2013..................................................................................................................................... 127	C.7	 Variance – Covariance Matrix for the Size Adjusted Armour Phenotypes April 2013…................................................................................................................................. 128	C.8	 Variance – Covariance Matrix for the Size Adjusted Armour Phenotypes in September 2013. ................................................................................................................. 128	C.9	 Mean Evolutionary Response of Armour Traits for Treatment and Control Ponds. 129	C.10	 Mean Evolutionary Response of Treatment and Control Ponds in Haldanes ......... 130	C.11	 Armour Trait Trajectories through Time for Control and Treatment Ponds.  ......... 131	C.12	 Mean Treatment Effect and Standard Error for Pond Pairs in the Fall (September – January) and Winter (January – April) Seasons. ................................................................ 132	C.13	 Coefficients of Change in Trait Trajectory through Time for Control and Treatment Ponds.. ................................................................................................................................. 132	C.14	 Difference in Evolutionary Response for SNPs within the Msx2 and Pitx1Ggene Regions. .............................................................................................................................. 133	   xii List of Tables Table 2.1 Standardized partial selection coefficients for lateral plate phenotype and Eda genotype. ....................................................................................................................................... 21	Table 2.2 Standardized univariate selection intensities for lateral plate phenotype and Eda genotype. ....................................................................................................................................... 22	Table 4.1 Mean selection intensity and standard error for all ponds in the fall (September – January ) and winter (January – April) seasons. ........................................................................... 63	Table 4.2 Divergent evolutionary response, delta haldanes (Δh), of armour phenotypes to treatment after one generation of selection. .................................................................................. 67	   xiii List of Figures Figure 2.1 Eda genotype frequencies of marine threespine stickleback after introduction to freshwater ponds. .......................................................................................................................... 15	Figure 3.1 Normalized cone opsin gene expression of marine and freshwater populations. ........ 36	Figure 3.2 Estimated spectral sensitivity of marine and freshwater populations assuming both only use the A1 chromophore. ....................................................................................................... 37	Figure 3.3 Normalized cone opsin gene expression of benthic and limnetic populations. ........... 38	Figure 3.4 Opsin expression in wild and lab reared fish from a marine (Oyster Lagoon (O)) and freshwater (Priest Lake (Pr)) location. .......................................................................................... 39	Figure 3.5 Correlations between shifts in spectral sensitivity and differences in local light. ....... 41	Figure 3.6 Change in spectral sensitivity, transmission, and irradiance of freshwater populations relative to the reference marine population. ................................................................................. 42	Figure 4.1 Illustration of armour traits of interest on a stickleback specimen stained with alizarin red. ................................................................................................................................................ 53	Figure 4.2 Schematic illustrating experimental timeline and design. ........................................... 55	Figure 4.3 Selection intensity within the first pond generation for A) first dorsal spine, B) second dorsal spine, C) pelvic spine and D) lateral plates in treatment and control ponds over the fall (September – January) and winter (January – April) seasons. ...................................................... 64	Figure 4.4 Divergent evolutionary response of armour phenotypes to treatment after on generation of selection. ................................................................................................................. 66	Figure 4.5 Divergent evolutionary response of armour genotypes within the Msx2 and Pitx1 genomic regions to treatment after one generation of selection. .................................................. 68	  xiv Figure 4.6 Divergent evolutionary response to treatment after one generation of selection estimated at SNPs near first dorsal spine QTL peaks. .................................................................. 69	Figure 5.1 Comparison of absolute selection intensity (s’) on armour to the distribution compiled by Kingsolver et al., (2001). ......................................................................................................... 82	    xv Acknowledgements First and foremost I thank my supervisor, Dolph Schluter, with his guidance I have become an immeasurably better scientist and writer, I will be forever indebted to him for this. I am particularly thankful to Dolph for his unremitting push for excellence and always reminding me to focus on the big picture. I am unbelievably grateful for the opportunities I have been afforded in Dolph’s lab and for the rich intellectual environment that the Biodiversity Centre has built. I am thankful for my committee: Sally Otto, Loren Rieseberg and Eric Taylor, who generously provided insightful and constructive feedback throughout my studies. Mike Whitlock and Nancy Heckman, although not on my committee, also shared their knowledge and expertise with me and I am very grateful to them. I am also thankful for Doug Altshuler, who was my seminar buddy, as well as, a consistent source of practical wisdom and encouragement.   In my opinion great colleagues are the key to success in academia and I have had the absolute privilege to work with some of the best. I have been blessed for the last nine years to have Greg Owens as a collaborator and dear friend; I cannot imagine all these years of grad school without him by my side. Without Seth Rudman Chapter 4 of this dissertation would have been insurmountable, I am grateful for all of the intellectual and physical labor that Seth contributed to this project and to the friendship he has shown me over the years. I consider myself very fortunate to have had Thor Veen as a friend and collaborator for the last five years; Thor’s brilliance and twisted sense of humor inspire me. I am thankful for Kira Delmore, my friend and sympathetic ear, it would have been very lonely without you. My daily chats with Sara Miller kept me going, it was heartening to have someone to commiserate with throughout this journey. Kieran Samuk’s quick wit and cleverness enriched my PhD experience immeasurably. Gwylim Blackburn, Mannfred Boehm, Haley Kenyon, Jenny Munoz were the   xvi most amazing office mates, and I loved all the intense chats we had about science and all the laughs we had about bizarre non-science topics.  I honestly believe that my parents, Pat and Darcy Windsor, think that I am capable of anything, and that is one reason that I have fearlessly strived for excellence. My parents generously put me through undergrad and have encouraged me in all of my academic pursuits. There really are no words capable of describing how thankful I am for their love and support. My parents in-law, Larry and Terry Rennison have also been an unwavering source of support over the last fourteen years, and I am always amazed by their boundless love and generosity. My brothers Keith Windsor and Owen Rennison have always been there for me, and I can’t thank them enough. Lastly, I am beyond lucky to have found my husband Graeme. I cannot imagine a better partner to go through life with. He is the definition of selfless and has endured many lonely evenings and weekends while I worked. I am unspeakably thankful for his love and support, I really don’t know why he puts up with me.              xvii       Dedicated to Graeme,  the first and last love of my life, and my closest truest friend.     1 Chapter 1: General Introduction The biological world is spectacularly diverse. Darwin himself marveled at the fact that “from so simple a beginning endless forms most beautiful and most wonderful have been, and are being, evolved” (Darwin, 1859). For hundreds of years biologists and naturalists have sought to understand the processes and mechanisms responsible for generating and maintaining this biodiversity. The last hundred and fifty years of evolutionary research have seen remarkable advances in our understanding of how ecological factors drive diversification. It is now widely accepted that adaptation to different ecological environments, due to divergent natural selection, can lead to the evolution of phenotypic and genetic differences between populations and the evolution of reproductive isolation (Schluter, 2001).  My dissertation focuses on the identification of biotic and abiotic factors that are sources of divergent natural selection, the traits that are affected, and the mechanisms by which these factors influence fitness. I utilize both experimental and observational methods to achieve a greater understanding of the processes of adaptation. Below I give an overview of the study of natural selection and adaptation and highlight some gaps in our knowledge of these processes; I then summarize the utility of threespine stickleback as a model for studying adaptation and finish by briefly outlining the chapters of my dissertation.   1.1 Advances in the Study of Natural Selection and Local Adaptation. In 1859 Darwin introduced his theory of natural selection, and the study of phenotypic adaptation took its modern form. Due to a very limited understanding of genetic inheritance in the late 1800s and early 1900s, efforts were first directed towards detecting the effects of natural selection on phenotypes. Since then biologists have sought to elucidate the ecological and   2 genetic mechanisms responsible for evolutionary change in natural populations. To make progress towards this goal, the phenotypic traits and genes associated with fitness must be identified and the contributing agents of selection isolated.  Phenotypic studies of natural selection began with attempts to ‘catch’ the process of natural selection in action. Most of the work between 1850 and 1950 investigating natural selection was observational; some studies identified natural selection directly by taking advantage of environmental perturbations and recording differences in traits before and after selection (e.g. Bumpus, 1899; Weldon, 1901; Flor, 1956). For example, Bumpus (1899) identified phenotypic differences between individual house sparrows who survived or perished after a severe storm. Other studies indirectly implicated the action of natural selection using comparative methods, such as demonstrations of consistent correlations between trait variation and environmental factors (e.g. Tutt, 1890; Poulton, 1898). Transplant experiments also played an important role in shaping our understanding of natural selection. One of the first documented transplant studies was conducted in 1874 (Bonnier and Flahault, 1878), and reciprocal transplants became increasingly popular in the mid – late 1900s. These transplants often found that populations have the highest fitness at their native sites compared to foreign sites, a pattern termed local adaptation (e.g. Ehleringer and Clark, 1988; Via, 1991; Boulding and Van Alstyne, 1993; Schluter, 1995). Together these studies and transplant experiments were fundamental in confirming Darwin’s hypothesis that in natural populations individuals possessing different phenotypes can exhibit variation in fitness.  Once traits are identified as putatively adaptive from correlations, transplant experiments or direct observations of selection, the objective shifts from identifying traits towards isolating and determining the environmental factors that underlie adaptive divergence. Selection   3 experiments are unique in their ability to isolate the contribution of a particular agent of natural selection to shifts in phenotypes (and genotypes). Some of the first selection experiments were undertaken in the early 1900s, but such experiments became more widespread in the later part of the century (e.g. Lees and Creed, 1975; Gould, 1979; Popescu, 1979). Selection experiments have manipulated both biotic and abiotic factors and have shown that both can drive natural selection on phenotypes.  The unit of inheritance, genes, must also be considered for a complete understanding of adaptation. Within the last three decades there have been significant improvements in sequencing methods, and the identification of genetic variants has been greatly simplified (Metzker, 2010; Stapley et al., 2010; Davey et al., 2011). These technological advances have allowed efforts to extend towards identifying the genes that are linked to fitness and direct estimates of selection affecting these genes (Barrett and Hoekstra, 2011). The identification of the genes underlying adaptive traits is important for understanding how the genetic architecture of a trait affects the process of adaptive evolution. Additionally, when selection is estimated on both a phenotype and its corresponding genotype(s) we have the opportunity to determine how phenotypic selection translates to changes at the genetic level. By now, hundreds, if not thousands, of studies have described patterns of genomic differentiation between populations inhabiting contrasting environments, as evidence of past natural selection (e.g. Turner et al., 2005; Minder and Widmer, 2008; Jones et al., 2012; Renaut et al., 2012; Nadeau et al., 2013). There are also a handful of transplant studies that have measured the genomic response to contemporary selection (e.g. Barrett et al., 2008; Anderson et al., 2014; Gompert et al., 2014). However, the most promising approach for determining the environmental mechanisms driving changes in allele frequencies and estimating responses is   4 controlled experimental studies (Barrett and Hoekstra, 2011). So far such experimental tests demonstrating the causative agent(s) responsible for fitness effects on both genotypes and phenotypes have been very limited, and most often laboratory based (e.g. Rokyta et al., 2005; Araya et al., 2010).     Despite over 150 years of devoted study to the process of adaptation, we still have large gaps in our knowledge. In particular we lack knowledge of the environmental factors that are important sources of natural selection and how they contribute to diversification among populations. Numerous studies have shown that resource competition is an important source of selection and have identified phenotypes underlying adaptation to this agent (e.g. Schluter and McPhail, 1992). Recent work has also identified genes involved in adaptation to resource competition (Lamichhaney et al., 2016). Adaptation to some abiotic factors such as temperature, have also been thoroughly examined at both the phenotypic and genotypic levels (e.g. Hancock et al., 2011). However, we know very little about how other species interactions such as predation and parasitism drive adaptation or what traits they influence. To address this gap in our knowledge, my dissertation seeks to determine the contribution of differential predation to trait diversification.   The study of the genetic basis of adaptation has also made considerable leaps over the last 20 years, and we have quickly accrued estimates of selection on genotypes. However, we know very little about how genetic architecture shapes phenotypic responses to natural selection. Genetic factors such as pleiotropy, genetic linkage and epistasis can constrain or facilitate evolution in response to selection depending on the context (Otto, 2004; Østman et al., 2011). It remains to be determined how the strength of selection differs when estimated on genotypes rather than phenotypes and how these genetic factors influence the observed phenotypic   5 response. My dissertation seeks to address this question with the goal of improving our understanding of the genetic basis of adaptation and the mechanisms responsible for phenotypic change in natural populations.   1.2 Study System: the Threespine Stickleback  All of the work in this dissertation uses threespine stickleback as a model system. The threespine stickleback (Gasterosteus aculeatus) comprises an adaptive radiation of fish found in coastal waters throughout the Northern Hemisphere. They inhabit both marine and freshwater environments (Bell and Foster, 1994). Marine threespine stickleback evolved 11 – 16 million years ago, whereas most freshwater populations of threespine stickleback are more recently derived, having evolved within the last 10 – 12,000 years (Bell, 1994). Marine threespine stickleback colonized newly forming lakes at the end of the last ice age and the populations they founded have since diversified in their morphology, behaviour and ecology (Bell and Foster, 1994). The phenotypic diversity of threespine stickleback populations has been recognized for over one hundred years (Reagan, 1909); however, it is only within the last 50 years that threespine stickleback have become a model for evolutionary biology.   Researchers began trying to characterize and understand the factors driving phenotypic divergence between threespine stickleback populations in the 1960s (Hagen, 1967; McPhail, 1969).  Since then hundreds of studies have been published describing phenotypic traits and ecological factors that differ between populations (reviewed in Bell and Foster 1994). In the early 2000s advances in genetic techniques facilitated the identification of the genetic architecture of divergence in threespine stickleback (Peichel et al., 2001), and with the assembly of the threespine stickleback genome (Jones et al., 2012) there is now a cottage industry   6 dedicated to describing patterns of genome-wide divergence (e.g. Hohenlohe et al., 2010; Deagle et al., 2011; Kaeuffer et al., 2012; Roesti et al., 2012) and identifying the genetic basis of phenotypic differences (e.g. Peichel et al., 2001; Colosimo et al., 2004; Shapiro et al 2004; Liu et al., 2014; Miller et al., 2015). In my thesis chapters I explore two axes of phenotypic variation among threespine stickleback populations. The first is between marine and freshwater populations. Thousands of independent freshwater populations have diverged in similar ways from their marine ancestors in trophic morphology, bony armour, pigmentation, body size and shape (McKinnon and Rundle, 2002). One of the most striking phenotypic shifts has been in lateral plates; these are calcified discs extending down the flank of the fish. While marine populations generally have 32 lateral plates the majority of freshwater populations have between zero and nine plates (Bell and Foster, 1994). Some of the genes underlying some of these traits, including lateral plates, have been identified (Colosimo et al., 2004; Shapiro et al., 2004; Miller et al., 2007) The second axis is between freshwater threespine stickleback populations, where divergence is along the benthic – limnetic phenotypic continuum (Rundle et al., 2000). Some lakes harbor limnetic morphs that primarily forage in the open water, feeding preferentially upon zooplankton (McPhail, 1992; McPhail, 1993). These fish tend to be streamlined and possess significant amounts of bony armour (McPhail, 1992; McPhail, 1993). In other lakes there are benthic morphs that are deeper-bodied, larger, and often less armoured (McPhail, 1992; McPhail, 1993); these fish forage in the shallow vegetated region of the lake on benthic invertebrates and mollusks (McPhail, 1992; McPhail, 1993). Within five lakes in southwestern British Columbia these benthic and limnetic morphs are found in sympatry and exhibit significant reproductive   7 isolation from one another (McPhail, 1992; Schluter and McPhail, 1992; McPhail, 1993; Gow et al., 2008).  Much of the divergence between populations of threespine stickleback (marine – fresh or benthic – limnetic) has evolved repeatedly among independently derived populations (Rundle et al., 2000; Conte et al., 2015), which suggests that natural selection has played an important role in this process. Differences in salinity and predation are two factors thought to be sources of divergent selection between marine and freshwater populations (Bell and Foster, 1994). Resource competition and predation are thought to have played an important role in the divergence of sympatric benthic and limnetic populations (Schluter and McPhail, 1992). However, there is little direct evidence showing that these factors impose selection on the particular traits or genes thought to be of importance.  Threespine stickleback provided an ideal system to study the process of adaptation and its genetic basis. The extensive literature on threespine stickleback ecology facilitated the identification of environmental factors that were likely to be important sources of divergent natural selection in the wild and which I could manipulate experimentally. I exploited the fact that benthic and limnetic threespine stickleback do not exhibit intrinsic post-zygotic barriers, which allowed me to create artificial hybrids to use in a selection experiment. The well-annotated genome has enabled previous researchers to identify some of the genes that code for ecologically important traits. I used these genes to ask questions about the relationship between genotypes, phenotypes and fitness. Lastly, threespine stickleback can be reared in the lab, which allowed me to estimate the degree to which some traits were genetically determined.     8 1.3 Research Conducted for this Dissertation  To understand the process of adaptation we must identify the mechanisms of natural selection. I use experiments and observational field studies in threespine stickleback to elucidate these mechanisms. In my experimental work I estimate selection on genotypes and phenotypes, in an attempt to further our understanding of how natural selection on phenotypes translates to selection on genotypes. I also test specific hypotheses regarding the role of particular agents of selection in driving trait diversification.  In chapter two of this dissertation I test whether direct and/or indirect selection has shaped the evolutionary response of a focal trait of interest. The phenotype of interest was lateral plates, which differs markedly between marine and freshwater populations of threespine stickleback (Bell and Foster, 1994). The genetic basis of lateral plate number is known, with the gene Ectodysplasin (Eda) identified as the primary causative locus (Colosimo, et al., 2005). Natural selection has long been thought to directly favour the reduction of lateral plates after freshwater colonization (reviewed in Bell and Foster, 1994). However, the Eda locus is known to exhibit pleiotropy, influencing neuronal and behavioural traits (Barrett et al., 2009a; Wark et al., 2012; Mills et al., 2014; Sadier et al., 2014), and the loss of lateral plates in freshwater could be a correlated response due to selection on these other phenotypes. To determine whether direct selection on lateral plates has contributed to changes in genotype frequency at the Eda locus I applied statistical methods for estimating selection on correlated phenotypic traits (Lande and Arnold, 1983). I find evidence of direct selection on lateral plates. I also find that selection on lateral plates alone fails to explain selection seen at the Eda locus during a freshwater introduction experiment, indicating that pleiotropy has likely played a role.    9 In chapter three I use patterns of parallel phenotypic evolution in wild threespine stickleback populations to test a long-standing prediction of visual ecology: that spectral sensitivity evolves to match ambient light conditions in order to increase the ability to catch photons and detect contrast between objects and the background (Clarke, 1936; Denton and Warren, 1957; Munz, 1958). Spectral sensitivity is predicted to be under strong divergent selection between environments that differ in their spectral properties (Endler, 1992). I find that spectral sensitivity is consistently correlated with environment type (freshwater vs. marine and littoral vs. pelagic) and that these differences are largely genetically determined. This suggests the action of natural selection. To investigate the environmental factors underlying shifts in spectral sensitivity I developed a new metric to quantify the strength of the relationship between ambient light and spectral sensitivity. I find that there is a strong relationship between changes in spectral sensitivity and differences in environmental light between environments, suggesting that features of the light environment have likely driven this shift. However, I did not find a strong relationship between spectral sensitivity and background light, which suggests other aspects of colour vision, such as contrast detection, likely also play an important role in shaping spectral divergence.  Chapter 4 tests the hypothesis that differential predation drives divergent character evolution. Previous work in threespine stickleback suggests that resource competition plays an important role in sympatric diversification (Schluter and McPhail, 1992). However, the contribution of other species interactions requires further study. For example it is unclear whether differential exposure to predators drives phenotypic divergence and if so what types of traits are under selection. There is evidence to suggest that resource competition has led to habitat differentiation between benthic and limnetic threespine stickleback. Within these habitats   10 each species encounters a unique suite of predators. To test whether coastal cutthroat trout predation is a source of divergent selection, I use hybrid benthic – limnetic threespine stickleback in an experiment where I manipulated their exposure to cutthroat trout and documented the evolutionary response at the phenotypic and genotypic levels. The traits of interest are components of bony armour, long hypothesized to be under direct selection by vertebrate predators (Reimchen, 1980). I use temporal sampling to determine how relative fitness changes over ontogeny and to further understand the observed evolutionary response. I find that armour phenotypes diverged between the threespine stickleback exposed to cutthroat trout predation and those not exposed. I also document temporal variation in the direction of selection on armour, which implies that susceptibility to predation varies over ontogeny. This work suggests that differential predation is likely an important factor driving divergent evolution in the wild.  In chapter five I discuss the general conclusions that can be drawn from my dissertation work and outline some outstanding questions in the field of adaptation.      11 Chapter 2: Discriminating Selection on Lateral Plate Phenotype and its Underlying Gene, Ectodysplasin, in Threespine Stickleback.  2.1  Introduction The study of adaptation seeks to establish a link between phenotypic variants, their underlying genotypes, and their fitness in a given environment. Recent advances in sequencing and genomics have enabled researchers to identify genes and genomic regions under natural selection using genome scans (e.g. Beaumont and Balding, 2004; Linnen et al., 2009; Jones et al., 2012; Therkildsen et al., 2013) and experimental studies of changes in allele frequency over time (e.g. Korves et al., 2007; Barrett et al., 2008; Burke et al., 2010; Fournier-Level et al., 2011; Pespeni et al., 2013; Gompert et al., 2014). However, pleiotropy and genetic linkage complicate the effort to identify phenotypic targets of selection. Even if a focal phenotype has been identified, a signature of selection at a locus may result instead via other traits determined by the same gene or linked genes, dragging the focal trait along as a correlated response. The challenge is similar to that faced by researchers attempting to identify which trait or traits, among a correlated suite, are the direct targets of phenotypic natural selection (Lande and Arnold, 1983; Price and Langen, 1992). Statistical methods for estimating selection on correlated phenotypic traits (Lande and Arnold, 1983) have been useful in identifying the phenotypes that are direct targets of natural selection (e.g. Grant, 1985; Schluter and Smith, 1986; Price and Langen, 1992; Nagy, 1997; Reznick et al., 1997). Here we show that the same approach can be used to help determine whether selection on a focal phenotypic trait has contributed to changes in genotype frequency at an underlying gene.    12 We apply the approach developed by Lande and Arnold (1983) to a study of multivariate selection on armour plates and its major underlying gene, Ectodysplasin (Eda; Colosimo et al., 2005), in threespine stickleback (Gasterosteus aculeatus). Freshwater populations established following colonization by the marine threespine stickleback after the last ice age (10,000 – 12,000 years ago) have repeatedly evolved a reduction of bony lateral armour plates (Bell and Foster, 1994). Adult marine threespine stickleback generally possess 30 – 36 lateral plates on each side of the body, whereas most freshwater populations founded by marine colonizers have 0 – 9 plates (Bell and Foster, 1994). The rate of evolution of plate loss in freshwater can be rapid (Klepaker, 1993; Kristjansson et al., 2002; Bell et al., 2004; Kristjansson, 2005). In every known case, loss of plates in freshwater has taken place via replacement of the “complete” armour allele at the Eda locus (hereafter, the C allele) by a relatively ancient “low” armour allele (the L allele; Colosimo et al., 2005). The rapid and parallel substitution of one allele by another strongly suggests that natural selection is responsible (Simpson, 1953; Schluter and Nagel, 1995), but it does not identify the mechanism of selection. In particular, we remain uncertain whether armour plating is itself the target of selection, or whether its rapid evolution is a by-product of selection on other traits affected by the same underlying gene.  The number of lateral plates has long been regarded as the direct target of natural selection (Bell and Foster, 1994), and there is some evidence in support of this hypothesis (Hagen and Gilbertson, 1973; Bell et al., 2004; Raeymaekers et al., 2007; Kitano et al., 2008; Leinonen et al., 2011; DeFaveri and Merila, 2013). However, no study has distinguished selection on lateral plates from selection on other traits that may be controlled by Eda or other tightly linked genes. In a previous study, we detected strong selection at the Eda locus in marine threespine stickleback transplanted to freshwater ponds (Barrett et al., 2008). The experiment   13 was conducted to test the growth hypothesis for the adaptive value of low armour plating in freshwater, namely, that low armour is favoured because of the high cost of mineralizing bone under reduced ion availability (Giles, 1983; Bell et al., 1993; Marchinko and Schluter, 2007). The experiment tested the prediction that low-armour phenotypes would have faster growth, leading to higher survival and earlier reproduction, by measuring relative fitness of genotypes at the underlying Eda gene. All transplanted adult marine threespine stickleback were heterozygous at the Eda locus (CL genotype), and they bred in artificial ponds to produce a cohort of all three genotypes (CC, CL, and LL), whose frequencies were then tracked over the course of a year (Barrett et al., 2008). Some of the results of the study supported the growth hypothesis. In particular, mean body length was positively correlated with the number of low alleles a fish possessed. Barrett and colleagues (2008) also found strong selection at the Eda locus. No such selection on Eda was detected in similar crosses raised in the lab (Barrett and Schluter, 2010). These results are consistent with the growth hypothesis; however, the study did not directly measure selection on lateral plates. Thus, it remains to be determined whether the observed changes in Eda genotype frequencies were the result of selection on plates or selection on other (unknown) traits affected by the Eda locus.  Here we estimate the strength of selection on lateral plates while controlling for Eda genotype. This is possible because genetic and phenotypic variation in lateral plate number is present within Eda genotypes (Colosimo et al., 2005). If selection acted on lateral plates, it should be detectable even while holding Eda genotype constant. Conversely, if genotype frequency changes at Eda resulted only via its pleiotropic effects on other selected traits, then holding Eda genotype constant should eliminate apparent selection on plates. In this case,   14 holding the plate number constant should have little effect on the estimated strength of selection on the Eda genotype.  2.2 Materials and Methods 2.2.1 Pond Experiment The original experiment is fully described in Barrett et al. (2008). Briefly, 45 or 46 wild adult marine threespine stickleback heterozygous at the Eda locus were introduced into four experimental freshwater ponds on the University of British Columbia campus. The fish were allowed to reproduce, and beginning in August 2006, 50 F1 progeny per pond per month were destructively sampled, assayed for standard length, and genotyped at a diagnostic marker for the Eda locus (Colosimo et al., 2005). The allele and genotype frequencies of the four F1 populations were then compared over the course of a year.  Here we focus on the three samples of juvenile fish taken in September, October, and November of 2006, respectively, during which the strongest changes in genotype frequencies were detected at the Eda locus (Figure. 2.1). Selection increased the CC genotype frequency by 19% between September and October and lowered the CL genotype frequency an equal amount. However, C allele frequency decreased from October to November, driven by a 15% decrease in CC genotype frequency and a 12% increase in CL genotype frequency. During both periods, the LL genotype frequency remained largely unchanged.       15 Figure 2.1 Eda genotype frequencies of marine threespine stickleback after introduction to freshwater ponds.  CC – carries two copies of the “complete” armour allele; CL – carries one complete allele and one “low” allele; LL – carries two copies of the low allele. Data points represent the mean of four ponds; error bars represent the standard error of the mean. Data re-plotted with permission from Barrett et al. 2008.  2.2.2 Lateral Plate Phenotype  We measured the number of lateral plates in a random sample of individuals from each of the three genotypes, yielding a total of between 76 and 85 fish from each of the September, October, and November samples in the Barrett et al. (2008) experiment (241 fish in total). Fish specimens were fixed in 10% formalin and stained with 0.001-0.002% w/v alizarin red S powder MonthGenotype Frequency0.00.10.20.30.40.50.6September October NovemberCCCLLL  16 in a 2% w/v potassium hydroxide solution. Stained fish were then photographed under standardized conditions and the total number of lateral plates on the left side was counted.  Lateral plate number was positively correlated with standard length, a measure of fish body size, particularly in CC and CL genotypes (CC, r  = 0.83, P = 2.07x10-13; CL, r  = 0.53, P = 4.94x 10-12; LL, r  = 0.07, P = 0.45), indicating that the number of plates had not completed development in many of the juvenile individuals in the samples. We used a breakpoint regression method (Appendix A.1) to size-adjust plate number, as follows: we fit a model in which the logarithm of plate number y increased linearly with fish length x up to a threshold value, x*, beyond which no further change took place:  Y = a + bx     for x < x*  Y = a + bx*   for x > x* The constants a, b, and x* were estimated from the data using nonlinear least squares (using the “nls” function in R, version 2.15.0; R Development Core Team 2012). Log plate counts were used in this analysis to reduce the skew of the data. The intercept and slope, a and b, were free to vary between Eda genotypes. We assumed that the threshold size x* was the same for all three Eda genotypes, because more complex models that also varied the threshold x* between genotypes did not improve the fit, as judged by AIC scores. Incorporating separate coefficients for each of the four ponds also did not improve the fit, and the analyses we present here do not include pond effects. The threshold x* value was estimated to be 34.0 mm ± 1.3 SE (Appendix A.1). In our selection analyses, log plate number for each fish was size-adjusted to a body length of 34.0 mm. Adjusted log plate counts were then back-transformed to the original and more intuitive non-log scale. Adjusted plate counts exceeding the value 32 were reduced to 32 to ensure that the range did not exceed the natural maximum seen in our data set.    17  2.2.3 Selection Analysis  We estimated selection coefficients (standardized partial regression coefficients) using the method for cross-sectional data (Lande and Arnold, 1983):   𝜷= P-1[𝐱after – 𝐱before],  where  𝜷 is the vector of estimated selection coefficients, 𝐱 is the vector of means of the focal phenotype trait and genotype scores (hereafter, “traits”) before and after selection, and P is the matrix of variances and covariances of the traits before selection. Separate analyses were carried out for the two episodes of selection, one between September and October 2006, and the other between October and November 2006 (Figure 2.1).  The vectors of trait means included the size-adjusted number of plates as the focal phenotypic trait. Genotype was scored using two genotype indicator variables. The first genotype variable (“additive”) coded the LL genotype as -1, the CL genotype as 0, and the CC genotype as 1. The second genotype variable (“dominance”) coded homozygous genotypes (LL and CC) as 0 and the heterozygotes (CL) as 1. All selection coefficients were standardized by multiplying each partial regression coefficient by the standard deviation of the trait before selection, to allow comparison between trait and genotype scores measured on different scales (Lande and Arnold, 1983). Because Eda genotype is categorical and lateral plate number is numerical, the method used here is an analysis of covariance to tease apart the contributions of genotype and plates to relative survival. Note that 𝜷 is not strictly a selection gradient because the traits do not have a multivariate normal distribution. Our analysis is within a single generation, and hence requires no assumptions about the probability distribution of breeding values of the numeric trait. Lande (1983) presented the   18 theory for predicting evolutionary response to selection on a quantitative trait influenced by a major locus and having an otherwise additive genetic basis. We do not predict evolutionary response here because our focus is on selection. We also calculated standardized univariate selection intensities (s') separately for lateral plates and the genotype variables as: s' = (𝑥after - 𝑥before)/𝜎before, where 𝑥before and 𝑥after are the phenotype or genotype trait means (coded in the same way as the dominance variable) before and after selection and 𝜎before is the estimated standard deviation of the trait before selection.   For simplicity, our analysis used size-adjusted lateral plates. However, we carried out an additional analysis in which size (standard length) was included as a trait along with unadjusted lateral plate number and Eda genotype. To accommodate the nonlinear relationship between plate number and size, it was necessary to adjust the standard length of the largest individuals to a maximum of 34.0 mm, the estimated breakpoint (Appendix A.1). The results of this analysis were not quantitatively different from the simpler analysis using size-adjusted plates and we do not present it.  To simplify further, our first analysis excluded the LL genotype and retained the CL and CC genotypes (Table 2.1). In this case there is only one genotype variable in the linear model (CL genotype is scored as 1 and CC is scored as 0), which incorporates the dominance component of Eda genotype but also half the additive component. This simplification is justified because CC and CL genotypes have high variance (Appendix A.4) and overlap in plate number, which is required to disentangle the separate effects of plates and genotypes on fitness. The variance-covariance matrices for the one-genotype analysis are given in Appendices A.5 and A.6   19 (one for each pair of months). In contrast, there is little plate number variation within the LL genotype, and little overlap in plate number with the other genotypes. A second analysis that included the LL genotype and the dominance variable is reported in the appendix (Appendix A.3; see Appendices A.7 and A.8 for the variance-covariance matrices).   We generated 95% confidence intervals for partial regression coefficients for each pair of months using a bootstrap resampling procedure. Each bootstrap replicate involved resampling with replacement, for a given month, the genotypes of n individuals from the corresponding Barrett et al. (2008) data set, where n is the number of individuals measured in our sample (the number for which we have lateral plate measurements). Next, for each genotype i we resampled ni phenotypes from the distribution of lateral plates corresponding to each genotype in the data. Resampling was done 10,000 times. Standardized partial selection coefficients were then calculated on each bootstrap replicate. The 0.025 and 0.975 quantiles of the coefficients were used to calculate the 95% confidence intervals of the parameters. The 95% confidence intervals of the selection intensities were estimated in the same way.  All our analyses incorporate pond as a fixed effect, essentially treating individual fish as the unit of replication. This is in contrast to the Barrett et al. (2008) study, which treated pond as a random effect and the unit of replication. Our approach here is justified because our goal is not to test for selection generally, as in the case of Barrett et al. (2008), but rather to investigate more narrowly the targets of selection within an experiment that has already demonstrated selection. All of the ponds responded similarly, and including pond in the analysis did not affect the results. For simplicity we present only the results for the analyses without the pond variables. All statistical analyses were conducted in R (version 2.15.0, R Development Core Team 2012).   20 2.3 Results We detected strong selection both on Eda genotype and on lateral plates (Table 2.1; Appendix A.3) using multivariate methods. Between September and October, selection favoured the Eda CC genotype over the CL genotype: point estimates of selection for the genotype variable and phenotypic trait, lateral plates, were similar in magnitude (Table 2.1), although the 95% confidence interval for lateral plates narrowly spanned zero. Between October and November the direction of selection changed, now favouring the CL genotype. During this period, selection on lateral plates and genotype was slightly weakened and the confidence intervals for both variables spanned zero (Table 2.1). Thus, selection on plates and genotype could not be disentangled in this episode. These analyses include only two genotypes and hence a single genotype variable that lumped additive and dominance components. When we analyzed all three genotypes in the September-October episode by including two genotype variables, one for each of the additive and dominance components, we detected strong selection on the dominance component of genotype (Appendix A.3). The CL genotype was disfavoured, in agreement with the findings of Barrett et al., (2008), who detected heterozygote underdominance for fitness. The point estimate of selection on plates was positive, but the confidence interval (barely) spanned zero. All confidence intervals for selection between October and November spanned zero, again indicating that selection on plates and genotype could not be disentangled in this episode (Appendix A.3).      21 Table 2.1 Standardized partial selection coefficients for lateral plate phenotype and Eda genotype.  For the September to October, and October to November, episodes. The 95% confidence intervals are reported in parentheses. Data restricted to CC and CL genotypes, where positive selection implies the L allele is favoured.   September-October October-November  𝜷 𝜷 Lateral plates  0.34 (-0.04, 0.74) -0.21 (-0.66, 0.22) Eda genotype -0.42 (-0.07, -0.79) 0.26 (0.77, -0.23)  If we had limited our analyses to univariate estimates of selection (s'), which by definition do not account for selection on correlated characters, the picture would have appeared different. The mean number of adjusted plates changed only slightly between September (18.7 +/- 0.91 SE), October (19.01 +/- 1.04 SE) and November (18.65 +/- 0.89 SE) (Appendix A.2). Consequently, the estimates of selection intensity indicated much weaker selection on lateral plates relative to selection on Eda genotype, which was at least an order of magnitude greater (Table 2.2). Additionally, unlike the selection intensity estimates for Eda genotype, the selection intensity estimates for lateral plates were near zero. Correspondingly, if selection estimates on lateral plates were only univariate (Table 2.2) the selection on lateral plates that was detected in the multivariate analysis (Table 2.1) would have been missed.        22 Table 2.2 Standardized univariate selection intensities for lateral plate phenotype and Eda genotype.  The 95% confidence intervals are reported in parentheses. Positive selection implies the L allele is favoured.   September-October October-November  s' s' Lateral plates  0.04 (-0.29, 0.38) -0.04 (-0.33, 0.25) Eda genotype -0.39 (-0.21, -0.56) 0.28 (0.07, 0.52)  2.4  Discussion We compared the magnitude of selection on a trait and its major underlying gene to investigate the targets of selection during two episodes. Univariate estimates of selection indicated strong selection on Eda genotype and only weak selection on lateral plate phenotype (Table 2.2). Our aim was to determine whether changes in Eda genotype frequency within a cohort in an experiment were due to selection on lateral plates themselves, as has been repeatedly suggested (Hagen and Gilbertson, 1973; Reimchen, 1992; Bell et al., 2004; Kitano et al., 2008), or the result of selection on some other unmeasured traits affected by the same underlying gene. Using Lande-Arnold methods for correlated characters, we found that selection on lateral plates was of similar strength as selection on Eda genotype in one of two episodes (Table 2.1, Appendix A.3), although barely non-significant. In the second episode, confidence limits for both selection coefficients spanned zero, and thus selection on plates and genotype could not be disentangled (Table 2.1 and Appendix A.3). The three-genotype analysis incorporating a dominance variable (Appendix A.3) indicates that the selection on Eda genotype (Table 2.1) was largely the result of selection against heterozygotes, rather than on the additive component of   23 genotype; this heterozygote underdominance for fitness was previously suggested by Barrett et al. (2008).  These results suggest that the rapid changes in genotype frequency at the Eda locus during experimental freshwater introduction were partially due to selection on lateral plate phenotype and partially due to selection on additional, unmeasured traits controlled by Eda or a tightly linked gene. However, genotype and phenotype are strongly correlated (r = 0.698), and their separate effects on fitness could not be separated in the second episode (October – November). While the analysis of selection on correlated traits has its limitations (discussed by Mitchell-Olds and Shaw 1987), we suggest these methods may help to disentangle whether selection is directly or indirectly influencing a phenotypic trait of interest when genotype at its major underlying locus is known.  Direct selection on lateral plate phenotype has long been thought to be the main factor driving the repeated evolution of reduced lateral plate armour in freshwater populations (Hagen and Gilbertson, 1973; Reimchen, 1992; Bell et al., 2004; Kitano et al., 2008; Leinonen et al., 2011). A number of studies have suggested that plate reduction has been favoured in freshwater due to an increased cost of mineralizing bone in freshwater due to reduced ion availability relative to marine environments (Giles, 1983; Bell et al., 1993). This hypothesis has some empirical support; Marchinko and Schluter (2007) and Barrett et al. (2008) found that low-plated genotypes grew faster in freshwater. Other work suggests that reduced predation in freshwater environments relative to the marine environment is responsible for lateral plate reduction (Moodie et al., 1973; Reimchen, 1992; Reimchen, 2000). Lateral plate reduction has also been suggested to improve swimming ability by increasing burst speed and buoyancy, which may aid in predation avoidance (Bergstrom, 2002; Myhre and Klepacker, 2009). However, until now, no   24 study had disentangled the direct effect of plates on growth and survival from the effects of Eda via other unmeasured phenotypic traits.  Two previous studies of geographic variation in lateral plates and Eda genotype frequencies found that genetic variance among populations (FST) at Eda was lower than phenotypic variance in lateral plates (QST/PST; Raeymaekers et al., 2007; DeFaveri and Merila, 2013), relative to total variance. This suggested stronger divergent selection among populations on lateral plates than on Eda frequencies. These results are consistent with our findings of selection on plates in a pond experiment. However, when we estimated the direct contribution of lateral plates to changes in Eda genotype frequencies, we found that lateral plates only partially explain these changes. Statistically controlling for number of lateral plates did not eliminate the signal of strong selection on Eda. These results highlight the value of measuring selection on a trait and its major underlying gene to help identify targets of selection.  On the basis of these results, we suggest that the rapid and repeated evolution of low-plated armour in freshwater may be the result of both selection on lateral plates and a correlated response to selection on other unmeasured traits affected by Eda. The Eda locus has been suggested to have diverse pleiotropic effects (Barrett et al., 2009a; Sadier et al., 2014). For example, Eda has been shown to affect the number of neuromasts along the lateral line (Wark et al., 2012; Mills et al., 2014). It is currently unclear whether Eda’s effect on neuromast distribution is direct or mediated indirectly through lateral plate development (Mills et al., 2014). Thus, there is potential that some of the selection we detect at the Eda locus and/or lateral plates is due to selection on the lateral line sensory system. In addition to its role in lateral plate and neuromast development, variation at Eda is associated with variation in schooling behaviour (Greenwood et al., 2013) and propensity to switch between water conditions of varying salinity   25 (Barrett et al., 2009b). Alternatively, selection might be acting on traits controlled by genes in linkage disequilibrium with Eda (Colosimo et al., 2005), which lies in a region of low recombination (Hohenlohe et al., 2011).  This study serves as a reminder that although genetic and genomic studies are informative about the evolution of traits, alone they provide insufficient evidence for selection on those traits, even when the link between a particular genotype and phenotype appears clear. Correspondingly, estimates of natural selection on phenotypes will remain an important component in genomic studies of adaptation (Travisano and Shaw, 2013) and are required to indicate whether mechanisms such as correlated selection are at work.   26 Chapter 3: Rapid Adaptive Evolution of Colour Vision in the Threespine Stickleback Radiation.  3.1 Introduction Sensory systems are often thought to be under strong natural selection (Endler, 1991), and are predicted to evolve to better correspond to signals in the local environment (Endler, 1992). For example, sensitivity of the visual system to different wavelengths of light is expected to evolve to match roughly the availability of wavelengths (Munz and McFarland, 1977; Bowmaker et al., 1994), increasing ability to catch photons and detect contrast between objects and background (Clarke, 1936; Denton and Warren, 1957; Munz, 1958). However, few studies have tested the adaptive significance of spectral sensitivity across the whole visual spectrum. The degree of matching between spectral sensitivity of organisms and their light environment across the spectrum has not been quantified. Aquatic organisms provide excellent opportunities to test for local adaptation and quantify matching (Lythgoe, 1979). This is because differential attenuation of wavelengths of light with water depth and by suspended particulates results in dramatic and predictable changes in local light spectra (Kirk, 1994; Lythgoe, 1988). For example, the transition from marine to fresh waters is usually accompanied by a large reduction in the availability of ultraviolet (UV) wavelengths, largely because of an increase in the amount of dissolved organics (Kirk, 1977; Kirk, 1994).  We used threespine stickleback (Gasterosteus aculeatus), which inhabit both marine and freshwater habitats, to investigate predicted evolutionary changes in visual adaptations of populations to the different ambient light environments. Marine threespine stickleback invaded   27 and adapted to numerous lakes and streams at the end of the last ice age (~12,000 years ago) (Bell, 1994). First, we tested for parallel evolution of opsin gene expression and spectral sensitivity over the visual light spectrum among these derived freshwater populations, which would represent strong evidence of natural selection (Schluter and Nagel, 1995).  Second, utilizing the extant marine form as a proxy for the ancestral state, we evaluated the extent to which shifts in the spectral sensitivity of freshwater populations are correlated with shifts in the ambient light environment and whether the outcome improves the match to local ambient light spectra. Finally, we tested for parallel divergence of spectral sensitivity of multiple pairs of sympatric limnetic and benthic threespine stickleback ecotypes (or “species pairs”) to finer scale heterogeneity in the light environment within lakes. In each of the three species pairs analyzed here, the benthic threespine stickleback forage in the vegetated littoral regions of the lake and deeper sediments, whereas limnetics are pelagic, found in the open water and near rocky cliffs (McPhail, 1992). The benthic environment contains relatively more long wavelengths of light than the open water (Boughman, 2001; Albert et al., 2007). We focused on expression of opsin genes, which encode the light sensitive G-protein coupled receptors that are expressed in retinal rod and cone cells. Opsins conjugate to vitamin A derived chromophores and play an important role in colour vision by mediating the conversion of photons into electrochemical signals, which initiate a neuronal response that is perceived by the brain (Schichida and Imai, 1998). The clear and well-characterized link between opsin genotype (coding sequence) and spectral phenotype (wavelength of maximal absorption, 𝜆!"#) make opsins particularly useful for studying sensory adaptation (Yokoyama, 1995). Opsin-mediated shifts in spectral sensitivity can be achieved by changes in opsin protein coding sequence (e.g. Yokoyama et al., 2000) and by changes in levels of gene expression (e.g. Carleton and Kocher,   28 2001; Fuller et al., 2004). We studied gene expression because analysis of whole genomes of marine and freshwater threespine stickleback has not found consistent differences in opsin gene coding sequence between marine and freshwater populations (Jones et al., 2012). Compared to other fish, threespine stickleback have relatively few (four) opsins, with a single functional opsin gene in each of the four cone opsin subfamilies: short-wavelength sensitive 1 (SWS1), short-wavelength sensitive 2 (SWS2), middle-wavelength sensitive (RH2), and long-wavelength sensitive (LWS) (Rennison et al., 2012). We measured expression levels of each of the four unique cone opsin genes in 11 threespine stickleback populations. We also measured expression in fish from two populations raised in a common laboratory environment to test the extent to which it is genetically determined.  We used opsin gene expression levels to estimate spectrum-wide spectral sensitivity to evaluate two general expectations. First, the advantages of photon capture and contrast should result in spectral sensitivity evolving roughly to correspond with wavelength availability (Munz and McFarland, 1977; Bowmaker et al., 1994). We measure this correspondence (“matching”) with the correlation across wavelengths between spectral sensitivity and two measures of light availability: irradiance (photons of each wavelength available at a specific water depth) and transmission (indicating the absorption of specific wavelengths by water). Large discrepancies between spectral sensitivity and light availability in specific regions of the visual spectrum might suggest specialized visual functions. Second, changes in wavelength availability from marine to fresh water should lead to similar shifts in spectral sensitivity (“local adaptation”). For example, as some wavelengths become scarce in the new environment and others common, relative to the ancestral environment, we expect spectral sensitivity to shift to correspond (Endler, 1992). Throughout, we use the whole light spectrum to study association, rather than studying   29 associations between summary measures such as the median. We introduce a new metric to quantify the correlation between shift in spectral sensitivity and the transition between light environments. Shifts in spectral sensitivity can additionally be achieved by differential use of vitamin A derived chromophores (Hárosi, 1994; Toyama et al., 2008). Conjugation of an opsin to an A1 chromophore (11-cis retinal) leads to a shorter wavelength of maximal absorption (𝜆!"#) than conjugation to an A2 chromophore (3-dehydro 11-cis retinal) (Hárosi, 1994). Switches in chromophore use have been shown to occur in fishes over ontogeny (Flamarique, 2005; Isayama and Makino, 2012) and between habitats via phenotypic plasticity (Toyama et al., 2008). Fish in the ocean generally use A1 chromophores, while freshwater fish have a mixture of A1 and A2 chromophores (varying from completely A1 to completely A2) (Toyama et al., 2008). Complete use of A2 is generally found in lakes whose waters are strongly stained with tannins (e.g. Flamarique et al., 2013), however such lakes are not included in our study. To account for possible variation in chromophore use, we model the effects of changes in chromophore and describe how this would affect our measures of local adaptation and spectral matching of opsin expression in threespine stickleback.   3.2 Materials and Methods 3.2.1 Sampling Six gravid females were collected from each of 11 populations inhabiting different breeding environments in the Strait of Georgia region of British Columbia, Canada. Collections were made under the Species at Risk Act collection permit number 236 and British Columbia Fish Collection permit number NA-SU12-76311. The samples came from two marine locations,   30 three lakes containing just a single (“solitary”) species of stickleback, and three lakes containing stickleback species pairs (see Appendices B.1 and B.10 for site details). Fish were euthanized at the collection site and eye tissue was immediately preserved in RNAlater® (Qiagen, Netherlands) and then kept at –20 °C for up to a month before RNA was extracted.  3.2.2 Opsin Expression and Visual Sensitivity The expression of each of the threespine stickleback’s four unique cone opsin genes (SWS1, SWS2A, RH2-1, and LWS (Rennison et al., 2012)) was measured using a standard reverse-transcriptase quantitative polymerase chain reaction protocol on RNA extracted from eye tissue (details in the Appendix B.2). We normalized the absolute number of transcripts for each gene from each individual by dividing the expression of a given opsin by the sum of the expression of all four opsins to get relative opsin expression (which sums to 1). We also measured gene expression of a reference gene, Beta actin, and calculated the expression of each opsin gene relative to it.  All statistical analyses in the paper were conducted in R 3.0.2 (R Development Core Team, 2015). To test for differences in mean expression of each opsin gene between marine and freshwater populations, we used a linear mixed-effects model (using the nlme package, (Pinhiro et al., 2015)) with water type (marine or fresh) as a fixed effect and location as a random effect. For this comparison, individuals from the benthic and limnetic species within a given location were combined and treated as a single population. Results were the same when only the benthics, or only the limnetics, were used instead. In separate analyses we tested for differences in gene expression between the sympatric benthic and limnetic species in three lakes, with lake as a random effect and ecotype as a fixed effect in the model. We treat lake populations as   31 independent replicates that require no phylogenetic correction. This is justified by the geological origins of lakes, which are in separate drainages and were accessible via the sea for a limited period of time. Previous studies show that the phylogeny of freshwater threespine stickleback populations in British Columbia based on putatively neutral markers is well approximated by a star phylogeny (e.g. Orti et al., 1994; Taylor and McPhail, 1999).  We bred three families of one marine population (Oyster Lagoon) and three families of one benthic population (Priest Lake) by in vitro fertilization. These fish were reared under laboratory conditions (i.e. in a common garden) in stand-alone 100 L tanks, in freshwater, with fluorescent lights. The parents of these crosses were captured from the wild in April 2011. Animals were treated in accordance with University of British Columbia Animal Care protocols (Animal Care Permit # A11-0402). After the eggs were fertilized they were reared in the lab for a year, in May 2012 a gravid adult female from each family was euthanized and her opsin expression was quantified as described above. We used linear models to test differences between lab-reared marine and freshwater fish and between lab- and wild-reared fish from the same populations.  Upon finding differences in mean opsin expression between marine and freshwater threespine stickleback, and between sympatric benthic and limnetic threespine stickleback, we estimated how they translated into differences in spectral sensitivity. We calculated a spectral sensitivity curve 𝑆! (350 – 700 nm) for each individual i based on its relative expression of the four opsin genes, and using the absorbance templates from Govardovskii et al. (2000) and estimates by Flamarique et al. (2013) of the wavelength of maximum absorbance (λ!"#) of each opsin gene (details in Appendix B.3). This model assumes that opsin expression contributes additively to spectral sensitivity; at this point in time it is a necessary simplification as we still   32 lack empirically informed models that describe and generalize any potential inhibitory interactions among opsins during signal integration and interpretation.  Chromophore (A1 and A2) ratios in the surveyed freshwater populations are not known. Based on empirical observations from Flamarique et al. (2013) we assumed that marine threespine stickleback used 100% A1 in the ocean. We estimated spectral sensitivity of stickleback in fresh water using three different chromophore ratios representing the extremes: 100% A1; 50% A1 and 50% A2; and 100% A2. We assumed that benthic and limnetic threespine stickleback have the same A1 : A2 ratio.   3.2.3 Association Between Visual Sensitivity and Ambient Light  We measured the spectral conditions of each location, with the exception of Cranby Lake and Little Quarry Lake. We used two measures to quantify the ambient light environment: irradiance and transmission. Irradiance measures the abundance of photons at each wavelength in the environment at a given point in time. Irradiance measurements of side-welling light (𝐼!) were taken at 10 cm, 20 cm, 50 cm, 100 cm and 200 cm depth at 10 or more sites within each sampling location using a cosine corrector attached to a spectrophotometer (Ocean Optics, USA). In subsequent analyses we used the irradiance at 50 cm. A limitation of using irradiance to quantify available light is that it varies with depth and with the weather and the angle of the sun. Transmission is the relative rate of loss of photons of a given wavelength per unit distance traveled through water. Transmission is a property of the body of water and may be less variable than irradiance, at least on short time scales. Transmission was measured as the light extinction coefficient with depth (Ks) (method for calculation outlined in Appendix B.5).  To test for local adaptation, we developed a statistic to quantify the association between   33 the shift in spectral sensitivity and the transition in light environment, from marine to fresh water, across all wavelengths for each lake population. First, we chose a marine population (Oyster Lagoon) to represent the ancestral phenotype and breeding environment. Next, we constructed transmission (𝐾!) and irradiance (𝐼!) curves by calculating at each wavelength (𝜆) the median from all samples within a location. At each wavelength we then subtracted the median value of the reference marine location from the median value in each freshwater location. This yielded change in transmission (Δ𝐾!) and change in irradiance (Δ𝐼!) values at every wavelength (λ) at each freshwater location. A positive value of Δ𝐼! at a given wavelength indicates that there are more photons of that wavelength (λ) present at the freshwater location relative to the marine environment. A positive value of Δ𝐾! at a given wavelength 𝜆  indicates greater light transmission (fewer photons lost as light travels through water) at the freshwater location than at the reference marine location.  Change in spectral sensitivity Δ𝑆 was calculated similarly, as follows. We calculated the median sensitivity at each wavelength 𝜆  of the sample of individuals from the reference marine population. Change in sensitivity was calculated for each freshwater individual as the difference between its sensitivity curve and the median marine curve. Finally, for each freshwater individual, we calculated the correlation coefficient (r) of the change in sensitivity (Δ𝑆) against the change in light environment (Δ𝐾! or Δ𝐼!), with each wavelength yielding a data point for each freshwater individual. A positive r indicates that regions of the spectrum with increased irradiance (or transmission) are correlated with increased spectral sensitivity, and regions of the spectrum with a decrease in irradiance (or transmission) are correlated with decreased spectral sensitivity. We used a mixed-effects model (with population as a random effect) to test whether mean correlation coefficients (r) differed significantly from zero.    34 We carried out a similar analysis of local adaptation of spectral sensitivity between the sympatric species in relation to differences in their local light environments. For each lake, we used the limnetic population and the pelagic environment as the reference. Other calculations were the same as described above for the marine and freshwater comparison (see Appendices B.5, B.6, B.7, B.8, B.12, and B.13, for further details and justification of our reference populations).  To quantify the degree to which populations are matched to their native light environments we estimated the correlation, wavelength by wavelength, between each population’s mean spectral sensitivity and the transmission and irradiance measured in its local environment. The significance of the mean correlation was tested separately for marine and freshwater populations using linear models.  Because analyses of local adaptation and matching involved a suite of tests that incorporated different measures of light environment and three chromophore scenarios, we adjusted the p-values for multiple testing in each table of results using the “BH” false discovery rate method (Benjamini and Yekutieli, 2001) and the p.adjust function in R (Appendices B.12, B.13 and B.14). Raw p-values are reported in the main paper and adjusted p-values are reported in the statistics tables in the Appendix B. In all cases significant p-values remained significant after the correction for multiple testing. All p-values are two-tailed.    3.3 Results 3.3.1 Opsin Expression and Spectral Sensitivity Freshwater stickleback populations had significantly lower expression of the SWS1 (UV) opsin gene than the marine populations (difference = -0.20 ± 0.02 SE, F1,6  = 145.2, p < 0.001)   35 and higher expression of the RH2 (green) opsin gene (difference = 0.21 ± 0.06 SE, F1,6 = 18.1, p = 0.005). We did not detect a significant difference in the other two opsin genes, LWS (red) (difference = 0.02 ± 0.04 SE, F1,6 = 0.2, p = 0.68) and SWS2 (blue) (difference = -0.009 ± 0.008 SE, F1,6 = 1.2, p = 0.31) (Figure 3.1). Differences in SWS1 and RH2 remained significant if expression was calculated relative to the reference gene Beta actin (SWS1 difference = 2.1 ± 0.3 SE, F1,6 = 49.2, p < 0.001; RH2 difference = 2.97 ± 0.9 SE, F1,6 = 10.7, p = 0.017). Thus we proceeded using cone opsin proportion as our metric of gene expression when modeling spectral sensitivity, as this has been shown to be best for making inferences about overall colour vision capacities (Fuller and Claricoates, 2011).  These differences in overall expression translated into large differences in estimated spectral sensitivity in two portions of the spectrum (Figure 3.2). Freshwater fish had reduced sensitivity in the 350-375 nm (UV and violet) region of the spectrum, and had greater sensitivity in the 450-600 nm (blue and green) region relative to both marine populations.            36 Figure 3.1 Normalized cone opsin gene expression of marine and freshwater populations. Marine populations are indicated in black, freshwater populations in grey. Horizontal lines indicate the mean of all populations; circles indicate individual fish. Location abbreviations as such: Oyster Lagoon (O), Little Campbell River (LC), Priest Lake (Pr), Paxton Lake (Pa), Little Quarry Lake (LQ), Trout Lake (T), Kirk Lake (K), and Cranby Lake (C).          0.00.10.20.30.40.50.6Normalised Expression ●●●●●●O●●●●●●LC●●●●●●●Pr●●●●●●●●●Pa●●●●●●●LQ●●●●●T●●●●●K●●●●●●CSWS1●●O●●●LC●●●●●Pr●●●●Pa●●●●●●LQ●●T●●●●●K●●●●CSWS2●●●●●O●●●●●●LC●●●●●●●●●Pr●●●●●●●Pa●●●●●●●●●●LQ●●●●T●●●●●K●●●●●CRH2●●●●●●O●●●●●LC●●●●●●●●●●●●Pr●●●●●●●●●●Pa●●●●●●●●●●LQ●●●●T●●●●●●K●●●●CLWSmarine freshwater  37 Figure 3.2 Estimated spectral sensitivity of marine and freshwater populations assuming both only use the A1 chromophore.  Marine populations are indicated in green, freshwater populations in blue. The thin lines are the fitted values of spectral sensitivity from the mixed-effects model. The shaded regions are on standard error above and below the fitted values, with standard errors also derived from the mixed-effects model.     Within lakes we found that the limnetic stickleback populations had significantly greater RH2 (green) expression than the benthics (difference = 0.05 ± 0.02 SE, F1,31 = 7, p = 0.01), and benthics had greater LWS (red) expression (0.04 ± 0.02 SE, F1,31 = 4.3, p = 0.05). However, the magnitudes of the differences were small (Figure 3.3). The expression of SWS1 and SWS2 opsins did not differ significantly (p > 0.29) between the two species (Figure 3.3). The difference in RH2 expression between the species was still significant when expression was calculated relative   38 to Beta actin gene expression (difference = 1.3 ± 0.6 SE, F1,31 = 4.4, p = 0.04), but the difference in LWS was not (difference = 0.3 ± 0.84 SE, F1,31 = 0.12, p = 0.70). These differences in expression translate to reduced sensitivity in the 525-575 nm (green) region of the spectrum and increased sensitivity in the portion of the spectrum above 600 nm (red) in benthics compared to limnetics (Appendix B.15).  Figure 3.3 Normalized cone opsin gene expression of benthic and limnetic populations.  The benthic populations are in black, limnetic populations in grey. Horizontal lines indicate the mean of all populations; triangles indicate individual fish. Location names abbreviated as: Priest Lake (Pr), Paxton Lake (Pa), Little Quarry Lake (LQ).  3.3.2 Laboratory Rearing In the lab, Oyster Lagoon (marine) and Priest benthic fish (freshwater) had similar expression differences as in the wild (Figure 3.4). SWS1 gene expression remained different between the marine and freshwater populations in the lab (difference 0.11 ± 0.02 SE, df = 1,4, F = 27.1, p = 0.01) as did RH2 (difference 0.18 ± 0.03 SE, df = 1,4, F = 40.1, p = 0.003). The difference in SWS1 was, however, greater in the wild samples, as indicated by an interaction 0.00.10.20.30.40.50.6Normalised ExpressionPr Pa LQ Pr Pa LQSWS1Pr Pa LQ Pr Pa LQSWS2Pr Pa LQ Pr Pa LQRH2Pr Pa LQ Pr Pa LQLWSbenthic limnetic  39 between rearing condition (wild or lab) and population of origin (effect size = 0.096 ± 0.039 SE, t1,4 = 2.486, p = 0.03). No other interactions were significant (all p > 0.17). Finally, we also detected a small difference in LWS expression between the two populations in the lab only (Figure 4; 0.06 ± 0.02 SE, F1,4 = 11.5, p = 0.03). Additional tests examining changes in the gene expression of lab-reared fish from each population compared to their wild counterparts are outlined in the Appendix B.4.  Figure 3.4 Opsin expression in wild and lab reared fish from a marine (Oyster Lagoon (O)) and freshwater (Priest Lake (Pr)) location.  Wild fish are indicated in black, lab reared fish in grey. Horizontal lines indicate the mean of the population, and points indicate individual fish.  3.3.3 Association Between Shifts in Visual Sensitivity and Ambient Light The shift in spectral sensitivity from marine to freshwater environments was positively correlated with the change in ambient light spectrum, when sensitivity was estimated assuming that both populations used only the A1 chromophore. On average, the correlation measured using 0.00.10.20.30.40.50.6Normalised Expression ●●●●●●O Pr●●●O PrSWS1●●O Pr●●O PrSWS2●●●●●O Pr●●O PrRH2●●●●●●O Pr●●O PrLWSwild lab  40 transmission (mean r = 0.39, ± 0.12 SE, t1,31 = 3.3, p = 0.002; Figure 3.5A) was of similar magnitude when using irradiance (mean r = 0.32, ± 0.06 SE, t1,31 = 4.95, p < 0.0001; Figure 3.5B). These correlations arose primarily from shifts in the short- (UV-blue) and middle-wavelength (green) regions (Figure 3.6). Decreased transmission of UV (350-400 nm) and violet (380-450 nm) in the freshwater environment (indicated by values below the dashed line in Figure 3.6) correspond with decreased sensitivity to these wavelengths in freshwater populations. Increased transmission of blue (450-495 nm) and green (495-570 nm) wavelengths in fresh water is correlated with increased sensitivity to these wavelengths. Freshwater populations varied considerably in the strength of the correlation (Figure 3.5).                 41 Figure 3.5 Correlations between shifts in spectral sensitivity and differences in local light.  (A) Correlations between shifts in spectral sensitivity of individuals from freshwater populations and differences in local light transmission relative to the reference marine population, Oyster Bay. (B) As in (A) but using irradiance rather than transmission to measure light environment shift. (C) Correlations between divergence in spectral sensitivity between sympatric benthic and limnetic stickleback species and differences in local light transmission. (D) As in (C) but using irradiance rather than transmission to compare light environments.     A BC D0.00.10.20.30.40.50.60.7β●●●●●●●●●●●●●●●●●●●●●●Priest Paxton Trout Kirk0.00.20.40.6β●●●●●●●●●●●●●●●●●●●●●●●●●●●●●Priest Paxton Trout Kirk−0.50.00.5β●●●●●●●●●●●●Priest Paxton−0.50.00.5β●●●●●●●●●●●Priest PaxtonCorrelation Coefficientri st     t      r t       iCorrelation CoefficientPries      Paxton     Trout       KirkCorrelation Coefficientri t               ri t               Correlation Coefficient  42 Figure 3.6 Change in spectral sensitivity, transmission, and irradiance of freshwater populations relative to the reference marine population.  Each subplot depicts the change in (A) spectral sensitivity, (B) transmission, and (C) irradiance, of four freshwater populations (Kirk, Paxton, Priest and Trout) relative to the reference marine population (Oyster Lagoon). Individual populations are labeled with a unique colour that is consistent among the three panels. Positive values indicate those wavelengths for which the freshwater populations have higher sensitivity, transmission or irradiance, than the marine reference population, and negative values indicate wavelengths for which the freshwater population has lower sensitivity, transmission or irradiance.  Change in SensitvityChange in TransmissionWavelength (nm)Change in IrradianceWavelength (nm)UVABC350 400 450 500 550 600 650 700−0.2−0.10.00.10.2350 400 450 500 550 600 650 700−0.008−0.006−0.004−0.0020.0000.002350 400 450 500 550 600 650 700−6e−04−4e−04−2e−040e+002e−044e−046e−04  43  These results isolate the effects of shifts in spectral sensitivity caused by changes in opsin gene expression in freshwater, when controlling for chromophore. We also measured the effects of these expression changes if combined with a hypothetical increase in the use of the A2 chromophore in these freshwater populations. The correlation between shifts in spectral sensitivity and transmission weakens slightly when a 50 : 50 mix of A1 and A2 chromophores is projected (mean r = 0.22, ± 0.12 SE, t1,31 = 1.85, p = 0.07). When 100% A2 chromophore is used, the correlation between shifts in sensitivity and transmission weaken further (mean r = 0.14, ± 0.09 SE, t1,31 = 1.53, p = 0.14) and the correlation between shifts in sensitivity and irradiance becomes negative (mean r = -0.48, ± 0.05 SE, t1,31 = -9.3, p = <0.0001) (see Appendix B.9 and B.12 for details, including adjusted p-values).  Within species pair lakes, there was a moderate, although not quite significant, correlation between divergence in spectral sensitivity and the difference in transmission (modeled using the A1 chromophore; Figure 3.5C; mean r = 0.27 ± 0.13 SE, t1,10 = 1.97, p = 0.077). This correlation was not significant for the difference in irradiance (Figure 3.5D; mean r = 0.18 ± 0.18 SE, t1,10 = 1.00, p = 0.339). The results were similar when other chromophore ratios were used to estimate spectral sensitivity, assuming that ratios were the same in both sympatric forms (see Appendix B.13 for details, including adjusted p-values).  3.3.4 Match of Visual Sensitivity to Ambient Light Despite strong correlations between shifts in spectral sensitivity and changes in the distribution of available wavelengths, spectral sensitivity is not closely matched to wavelength availability in either marine or freshwater environments. The mean correlation between spectral   44 sensitivity of freshwater fish and ambient light in lakes, while statistically significant, was small (0.07 ± 0.03 for transmission and 0.12 ± 0.02 for irradiance). This weak matching has arisen multiple times in parallel in lake stickleback, which suggests that natural selection favours it. Substituting the chromophore did little to alter the mean correlation for transmission (although it became statistically insignificant) and slightly changed the strength for irradiance (See Appendices B.6 and B.14. for details, including adjusted p-values). In the marine environment the mean correlations between marine spectral sensitivity and transmission or irradiance are negative (r = -0.66 ± 0.16 SE and r = -0.11 ± 0.07 SE, respectively). The main cause of the strong negative correlation in marine waters is the excessive UV sensitivity compared with UV light availability. Nevertheless, UV expression declines in fresh water, where these wavelengths are even more scarce, contributing to the observed correlation between shifts in sensitivity and the change in wavelength distribution.   3.4 Discussion Our findings indicate that there has been rapid parallel evolution of opsin gene expression and spectral sensitivity across the light spectrum in freshwater stickleback populations. All surveyed freshwater populations have their spectral sensitivity shifted towards blue and green wavelengths, and away from ultraviolet and violet, relative to the marine populations. This has been accomplished entirely by shifts in opsin gene expression rather than protein sequence changes. We provide evidence that this difference has a genetic basis, as the main differences in expression were largely maintained in two lab-reared populations. Our analyses also reveal a strong association between shifts in spectral sensitivity and changes in light transmission from marine to freshwater environments, suggesting that these shifts are in an adaptive direction. On a   45 smaller scale, we also find support for parallel adaptive divergence of gene expression and spectral sensitivity within lakes, between sympatric limnetic and benthic species. The evolution of the visual system in stickleback has been rapid, as these freshwater populations have evolved within the last 12,000 after the last glacial maximum (Bell and Foster, 1994).   The degree of phenotypic parallelism in opsin expression and spectral sensitivity that we describe is unprecedented over such a short time span. Nine independently derived populations exhibit the same direction of shift in opsin expression following the colonization of fresh water. In East African cichlids, parallel evolutionary divergence of opsin expression has been detected between species within two of the three major lake cichlid radiations (O’Quin et al., 2010), but these radiations are much older than the freshwater stickleback populations studied here. Our findings are in line with previous work in threespine stickleback, which has found extensive parallel evolution of morphological traits and patterns of genomic divergence among freshwater populations (Bell and Foster, 1994; Jones et al., 2012; Colosimo et al., 2005; Hohenlohe et al., 2010). Some but not all of this morphological parallelism involves changes at the same underlying genes, which frequently represents adaptation from a common ancestral pool of standing genetic variation (Colosimo et al., 2005). Possibly, the parallelism we observe in spectral sensitivity also represents adaptation from a common pool of standing genetic variation, which would help to explain the speed of evolution in this trait in threespine stickleback. Further genetics work is required to test this idea.    The result from our lab rearing experiment suggests a substantial genetic component to the population differences in opsin expression. This contrasts with many other systems in which differential opsin gene expression and/or spectral sensitivity is largely phenotypically plastic (e.g. Fuller et al., 2005). For example, wild Bluefin Killifish (Lucania goodie) living in clear   46 springs and tannin stained waters exhibit large differences in their opsin gene expression (Fuller et al., 2005); however, light treatment and rearing experiments in the lab have shown that most of these differences are due to environmental effects (Fuller et al., 2005). Given our breeding design, using wild caught mothers, we cannot entirely rule out the contribution of maternal effects or early developmental effects in eggs while the mothers were in their native environment, future work in stickleback will be required to rule out the contribution of these factors.  Smaller but detectable differences in opsin expression and sensitivity were found between limnetic and benthic threespine stickleback inhabiting the same lake. These differences were repeated in multiple lakes, suggesting a role for natural selection in the divergence of visual systems on a small, within-lake scale. Benthics had slightly higher estimated sensitivity to red wavelengths than did limnetics, in accord with a more red-shifted local light environment. Previous work using optomotor behavioural response assays indicated that limnetic stickleback in Enos Lake have higher red wavelength sensitivity than the benthic population from the same lake (Boughman, 2001), and similar red wavelength sensitivity to the benthic in Paxton Lake (Boughman, 2001). In contrast, we found higher expression of long wavelength opsins in benthics compared to limnetics. Unfortunately, we could not include Enos Lake in the present study because the pair has collapsed due to hybridization (Taylor et al., 2006). Future work is required to determine how these differences in opsin expression affect foraging and mate choice in stickleback, as has been suggested in Lake Victoria cichlids (Seehausen et al., 2008).  Early work in the field of visual ecology focused on the hypothesis that spectral sensitivity should evolve to maximize an individual’s photon catch (Clarke, 1936; Denton and Warren, 1957; Munz, 1958). Tests of this hypothesis have examined the relationship between the   47 λ!"# of visual pigments (opsins) and the wavelengths most prevalent in ambient environment and have often found a strong relationship (e.g. McFarland and Munz, 1975; Loew and Lythgoe, 1978; Hunt et al., 2001). However, detection of contrast and colour discrimination also likely shapes the evolution of spectral sensitivity. With multiple functions, it may be difficult to predict a priori the evolved degree of spectrum-wide matching (wavelength by wavelength correlation) of spectral sensitivity to the available light spectrum. We did not find a close match in freshwater populations, and indeed, the correlation was negative in marine populations. The low match in marines is driven by their high estimated sensitivity to short wavelengths such as UV, despite the relative rarity of these light wavelengths in the marine environment compared to mid-wavelengths. The low degree of matching suggests that increasing photon capture alone is unlikely to explain the evolution of spectral sensitivity. Predicting a shift in sensitivity with change in light spectrum may be more straightforward: reduced investment in capturing specific wavelengths that are increasingly rare is expected. For example in the deep sea, long-wavelength light is rare, and some deep-sea fish have lost long-wave sensitive opsins and shifted their sensitivity towards shorter wavelengths (Hunt et al., 2001). Similarly, we found that freshwater stickleback have reduced expression of short wavelengths, which are even scarcer in freshwater than in the sea. Nevertheless, freshwater fish retain relatively high sensitivity to UV light compared to background irradiance. One possible explanation for the low match between sensitivity and ambient wavelengths is that high expression of pigments whose sensitivity is offset from the dominant wavelengths of the environment could play an important role in contrast detection under low light conditions (McFarland and Munz, 1975; Loew and Lythgoe, 1978). For example, in stickleback UV wavelengths are important for detection of zooplankton prey against the background light (Rick   48 et al., 2012). This idea is consistent with the observed trend toward reduced UV opsin expression in freshwater threespine stickleback populations, because most are less zooplanktivorous than marine stickleback (Hagen, 1967). Experimental work in other fish species has also shown that reduced UV sensitivity coincides with reduced zooplanktivory and zooplankton foraging ability (Bowmaker and Kunz, 1987; Flamarique, 2013). A second possible explanation for the low match between spectral sensitivity and ambient light is that detection of specific wavelengths might be important for mate choice and intraspecific signaling. Short (UV-blue) and long wavelengths (yellow-red) are important signals for mate choice in threespine stickleback (Rick and Bakker, 2008), as male nuptial colouration often involves blue and red pigmentation (Rowland, 1994), as well as reflection in the UV (Rick et al., 2004). Tuning of perception towards these nuptial signals and detection of contrast among them could also contribute to the mismatch of sensitivity to available light. It is also conceivable that our estimates of sensitivity, which do not account for non-additive signal integration during neuronal processing or the translation of mRNA to opsin proteins, underestimate the environmental correlation. Seasonal variation in light environment could also explain the weak relationship; perhaps our time of sampling was not representative of spectral light conditions over the course of the year.  We find that A2 opsin chromophore complexes do not necessarily act synergistically with changes in opsin expression to produce adaptive shifts in spectral sensitivity. In the populations surveyed substitution of A1 chromophores with A2 chromophores weakens the relationship between shifts in spectral sensitivity and shifts in ambient light. While we do not know the empirical ratios of A1 and A2 in the wild for these freshwater populations, our analyses suggest that A2 domination would be unlikely. A2 dominated retinas result in shifts in spectral sensitivity that do not correlate to shifts in these environments, and thus are unlikely to be in an adaptive   49 direction. This was a somewhat surprising result as A2 chromophores are commonly used by many species of fish found in freshwater lakes or streams (Toyama et al., 2008). The potentially maladaptive shifts seen when substituting to A2 are a result of overshooting long-wavelength sensitivity relative to the prevalence of these wavelengths in the surveyed freshwater lakes. This finding is consistent with work suggesting A2 dominated retinas are common for threespine stickleback from dystrophic lakes that are strongly red-shifted relative to the marine environment, as A2 use in such an environment would likely result in shifts in an adaptive direction (Flamarique et al., 2013).  In this study we provide three lines of evidence to suggest that observed shifts in spectral sensitivity are adaptive: we show that they have evolved repeatedly, are likely genetically based and that regions of the spectrum that differ between marine and freshwater locations are largely the same regions that exhibit differences in spectral  sensitivity between populations. The methods used in this study help to understand the direction of evolution of spectral sensitivity, and its relationship with ambient light. However, our approach does not allow us to disentangle the relative contribution of selection on color discrimination, contrast detection and photon capture to shifts in spectral sensitivity. Future experimental and theoretical work will be required to determine the importance of selection on each of these functions.     50 Chapter 4: Survival in a Cutthroat World: Estimating Natural Selection on Armour Phenotypes and Genotypes in Threespine Stickleback. 4.1 Introduction Resource competition is well known to play a direct role in phenotypic diversification (e.g. Schluter et al., 1985; Bernatchez and Dodson, 1990; Losos, 1990; Schluter and McPhail, 1992; Hansen et al., 2000). When divergence in response to competition leads to changes in other species interactions, such as predation and parasitism, it can have secondary effects that further drive diversification. The importance of these secondary species interactions for divergence is unknown and has not been previously tested. We take a novel approach using a manipulative experiment to estimate phenotypic selection and evolutionary response to a species interaction by measuring changes at underlying genes. Sympatric benthic and limnetic threespine stickleback provide an ideal system to investigate the effect of these species interactions on divergence. These species pairs have evolved independently multiple times within the last 12,000 years and exhibit phenotypic parallelism, suggesting that natural selection has played a role in their formation (Rundle et al., 2000). Previous work has shown that stickleback species exhibit a pattern of character displacement of trophic traits in response to resource competition, whereby sympatric species show exaggerated divergence and solitary populations are intermediate (Schluter and McPhail, 1992). Remarkably, sympatric species also show exaggerated divergence in bony defensive armour compared to solitary populations (Vamosi and Schluter, 2004). Divergence of armour traits has been suggested to be a secondary effect of divergence in habitat-specific predation regimes and possibly an interaction between benthic and limnetic stickleback mediated by their   51 predators (Vamosi and Schluter, 2004). We used a manipulative experiment to test whether differential mortality caused by one major predator results in a divergent evolutionary response. The genes or genomic regions determining some of these armour traits have been identified (Shapiro et al., 2004; Summers, 2008), which allows us to document the response on phenotypes and their underlying genotypes.  Limnetics and benthics are exposed to distinct predation regimes in their preferred habitats outside of the breeding season. Limnetic stickleback are generally found in the pelagic region of the lake, where they encounter coastal cutthroat trout (Oncorhynchus clarkii clarkii) and common loons (Gavia immer). Benthic stickleback are found in the littoral region (Schluter and McPhail, 1992), where they are preyed upon by invertebrates, such as dragonfly nymphs (Aeschna spp.), water beetles (Dytiscid spp.), backswimmers (Notonecta spp.) and giant waterbugs (Lethocerus americanus) (Foster et al., 1988; Remichen, 1994; Vamosi, 2002; Vamosi and Schluter, 2002).  We manipulated the presence/absence of trout to test the hypothesis that differential exposure of coastal cutthroat trout is partly responsible for the divergence in bony armour seen between the benthic and limnetic species. To accomplish this we contrasted the response to selection at the phenotypic and genotypic levels on armour traits across a generation between treatments. We were interested in five armour traits: lateral plates, first and second dorsal spine, pelvic girdle and pelvic spines (Illustrated in Figure 4.1). Cutthroat trout are a gape-limited predator, and longer spines are thought to increase the effective diameter of a stickleback, which increases the difficulty of swallowing and the likelihood of injury to the trout (Reimchen, 1991; Reimchen 2000). Lateral plates are thought to increase the probability of survival post-capture by protecting against lacerations inflicted by trout (Hoogland et al., 1956; Reimchen, 1992, 1994,   52 2000). In contrast armour is believed to be detrimental in the presence of invertebrate predators. Invertebrate predators have been suggested to utilize armour to capture and grasp stickleback (Reist, 1979; Reimchen, 1980; Vamosi, 2002), and armour has been suggested to be costly to grow and maintain (Giles, 1983), which may keep fish at a size where they remain vulnerable to invertebrate predation for longer.  To test for divergent selection on armour, we estimated the change in mean armour traits between control and cutthroat trout treatment ponds between two generations and at multiple points within the first generation. Armour is generally robust in limnetic stickleback, which exhibit long spines and an intact pelvic girdle, whereas in benthic stickleback the pelvic spines and girdle, along with the first dorsal spine, are often absent or heavily reduced (McPhail, 1993). Benthic stickleback also have fewer lateral plates (zero to four) compared to limnetics, which generally have between five and nine plates (McPhail, 1993). The robust armour of a limnetic stickleback correlates with an increased likelihood of encountering cutthroat trout. However, direct evidence that selection by cutthroat trout favours the evolution of longer spines and an increased number of lateral plates is lacking. We predict that in our experiment the ponds with cutthroat trout predation will favour the limnetic phenotype of long spines, a robust pelvic girdle and more lateral plates. In control ponds, we predict that armour will either stay the same or be reduced due to selective predation imposed by invertebrates. We expect to find a similar evolutionary response for the genes/genomic regions that underlie these traits.        53 Figure 4.1 Illustration of armour traits of interest on a stickleback specimen stained with alizarin red.  A) First Dorsal Spine, B) Second Dorsal Spine, C) Pelvic Girdle, D) Pelvic Spines E) Lateral Plates.   4.2 Materials and Methods 4.2.1 Experimental Design The experiment was conducted in ten semi-natural experimental ponds located on the University of British Columbia Campus in Vancouver, Canada. The ponds contain a natural assemblage of food resources and do not exclude invertebrate or avian predators. Each pond provides a shallow littoral zone with vegetation, as well as an open water habitat (25m x 15m, with a maximum depth of 6 m). For further details of the pond structure see Arnegard et al., (2014).  The experimental fish were acquired from two sources: 1) Four F1 crosses were made in the spring of 2011, between four benthic mothers and four limnetic fathers collected from Paxton Lake on Texada Island, British Columbia, Canada. 2) Fish were collected from First Lake on Texada Island British Columbia, Canada. These First Lake fish are the descendants of F1 benthic-limnetic crosses made from benthic and limnetic fish from Paxton Lake and introduced to First Lake in 1981 (McPhail, 1993). We used these fish in addition to our lab crosses since A BEDC  54 they have undergone additional recombination events, which potentially facilitate identification of the genes underlying phenotypic traits of interest. We used hybrids as the target populations in our experiment, to maximize both variation for selection to act upon and independent segregation among traits of interest. The range of phenotypes observed in our benthic-limnetic F2 crosses encompasses much of the variation found between the benthic and limnetic ecotypes (Arnegard et al., 2014).  The ten ponds were paired prior to stickleback introduction; pairing was based on count surveys of a variety of biotic conditions including: macrophyte coverage, phytoplankton, zooplankton and insect abundance. The F1 hybrids were reared in the lab in 100 L tanks for a year prior to their introduction into the experimental ponds in May 2012. Each of the four F1 families was split between a pair of ponds, with one cross per pond pair. The First Lake fish were introduced into the fifth pair of experimental ponds in June 2012. Each pond received 21-31 individuals, with paired ponds receiving the same number. Over the spring and summer of 2012, the stickleback in all ten ponds reproduced, producing the first pond generation composed of multiple F2 hybrid families (or more advanced generation hybrids in the case of First Lake).  In September 2012, a lethal sample of offspring was taken from each pond. After this initial sampling was complete two coastal cutthroat trout (10 – 12 inches in length) were introduced to one randomly chosen pond within each pond pair. Cutthroat trout were obtained by angling in Placid Lake, southwestern British Columbia. The first pond generation was again lethally sampled in January 2013 and April 2013. In the spring and summer of 2013 the first generation of pond fish bred within the ponds creating the second pond generation. This second pond generation was lethally sampled in September 2013. During all sampling periods   55 stickleback were caught using a combination of un-baited minnow traps, open water seining, and dip netting. See Figure 4.2 for a schematic of the experimental design and sampling timeline.     Figure 4.2 Schematic illustrating experimental timeline and design.  Yellow, pink, orange and green are used to indicate individual F1 families and blue indicates First Lake fish.     56 4.2.2 Phenotyping Immediately following collection, fish were euthanized in MS-222 and placed in 95% ethanol. At a later date a portion of the caudal fin was removed and set aside for DNA extraction. The fish was then rehydrated over the course of a week in a series of ethanol - water washes (70%, 50%, 20%, 0% ethanol) and then fixed in 10% formalin. Specimens were stained with 0.001-0.002% w/v alizarin red S powder in a 2% w/v potassium hydroxide solution alizarin red to highlight bony structures, following the protocol outlined by Peichel et al., (2001). Measurements were taken on the left side of each stained specimen. Fifty individuals per pond were measured in September 2012 (with the exception of Pond 13 where 35 individuals were measured) and September 2013. Fifty individuals per pond were measured in January 2013 (with the exception of ponds 9 and 13, which were not sampled due to concerns of low population size). One hundred individuals per pond were measured in April 2013. We measured the length of the first and second dorsal spines, length of the pelvic spine and girdle, and standard length. The number of lateral plates was also counted. If a given trait was absent it was given a value of zero. The spine and pelvic traits scaled with standard length, so trait values were size corrected to compare trait values among individuals of different sizes. Traits were size corrected to an average length (32 mm, which was the mean size of individuals in September 2012 F2 cohort) using the equation:  𝑌! = 𝑋! −  𝛽(𝐿! − L),  where 𝑌! is the size-adjusted armour trait, 𝑋! is the original armour trait, 𝛽 is the regression coefficient of the un-adjusted armour trait values on standard length, 𝐿! is the standard length of the individual and L is the average length of the sample (Vamosi, 2002). There was no size correction for trait values of zero (i.e. when the trait was absent). When the trait values were   57 plotted it was apparent that pelvic girdle and pelvic spine phenotypes fell into two distinct clusters, ‘high’ or ‘low’ armour. We used Gaussian Mixture Modelling for model-based clustering, using the mclust package (Fraley and Raftery, 2012), to identify these high and low pelvic phenotypes. These two pelvic armour clusters had different relationships (β) between standard length and the focal trait value; correspondingly size correction was done independently for the two clusters. For some time periods, linear models indicated that pond had a significant effect on β, in which case size correction was done independently for each pond. Lateral plates did not scale consistently with length and were not size corrected, although applying a size correction did not qualitatively change any estimates (data not shown).   4.2.3 Genotyping  Deoxyribonucleic acid (DNA) was extracted from caudal fin clips using a standard phenol-chloroform extraction protocol. Fifty individuals were sampled per pond from September 2012 and September 2013 (1000 individuals total). DNA was prepared for Illumina sequencing using the PstI enzyme following the genotyping by sequence method of Elshire et al. (2011), with the addition of a gel size selection of fragments 500 – 700 base pairs (bp) in length. One hundred and ninety two individuals were uniquely barcoded and combined into a library, for a total of seven libraries. Libraries were sequenced at the University of British Columbia’s Biodiversity Next Generation Sequencing Centre on an Illumina HiSeq 2000. Reads were 100 bp in length and sequencing was paired end.  Sequence variants were identified using a standard, reference-based bioinformatics pipeline (see archived code for full details). After demultiplexing, Trimmomatic (Bolger et al., 2014) was used to filter out low quality sequences and adapter contamination. Reads were then   58 aligned to the stickleback reference genome (Jones et al., 2012) and a bacterial artificial chromosome (BAC) sequence containing the complete pituitary homeobox 1 (Pitx1) coding sequence (Chan et al., 2010) (GeneBank Accession: GU130435.1) using BWA v0.7.9a  (Li and Durbin, 2009), subsequent realignment was done with STAMPY v1.0.23 (Lunter and Goodson, 2011). The GATK v3.3.0 (McKenna et al., 2010) best practices workflow (DePristo et al., 2011) was followed except that the MarkDuplicates step was omitted. RealignTargetCreator and IndelRealigner were used to realign reads around indels and HaplotypeCaller identified single nucleotide polymorphisms (SNPs) in individuals. Joint genotyping was done across all individuals using GenotypeGVCFs. The results were written to a single VCF file containing all variable sites. This file was filtered for a minimum quality score (20) and depth of coverage (minimum of 8 reads and maximum of 100,000) before use in any downstream analyses.  4.2.4 Linkage and Quantitative Trait Locus Mapping The pedigree of F2 individuals was determined using the MasterBayes R package (Hadfield, 2013) using 1799 SNPs, which had minimal missing data across individuals. In order to have markers that were fully informative for linkage mapping we identified the SNPs that were homozygous for alternative alleles in the benthic and limnetic grandparents of each F2 cross. We then used these SNPs to calculate pairwise recombination frequencies and create a genetic map using JoinMap version 3.0 (Ooijen and Voorrips, 2002). In total 398 F2 progeny from four F1 crosses were used for mapping, comprised of multiple F2 families. F2 genotypes were coded according to the population code for outbred crosses, allowing segregation of up to four alleles per locus (cross-pollinator). The JMGRP module of JoinMap was used with a LOD score threshold of 4.0 to assign 2243 loci to 33 linkage groups. For each linkage group, a map   59 was created with the JMMAP module (Appendix C.1). Mapping was done using the Kosambi function with a LOD threshold of 1.0, recombination threshold of 0.499, jump threshold of 5.0, and no fixed order. Two rounds of mapping were performed, with a ripple performed after each marker was added to the map.  A total of 2243 SNP markers and the aforementioned genetic map were used for the quantitative trait locus (QTL) mapping of first dorsal spine length, pelvic spine length and pelvic girdle length. QTL mapping was done using the Haley–Knott regression with F1 family as a covariate in the R/qtl package (Broman and Wu, 2013). To test whether there was variation among families in the genetic basis of each trait the QTL mapping was done independently for each family, with a sample size of 99-100 individuals per family. The phenotype analysis was conducted on all five-pond pairs, however the genotype analysis was limited to the F1 families, since we were unable to QTL map the traits for the First Lake fish. Genotypic estimates of selection were made using SNPs within the QTLs identified for the first dorsal spine and pelvic spine/girdle phenotypes. Our multiple F1 family QTL map (Appendix C.2) was consistent with previous work which indicated that Msx2 on chromosome IV influences the length of the first dorsal spine (Summers, 2008) and Pitx1 on chromosome VII influences the length of the pelvic spine and pelvic girdle (Shapiro et al., 2004). In the QTL map with F1 family as a covariate the peak near Msx2 explained 4.0 percent of the variance (PVE) in first dorsal spine length and the peak near Pitx1 explained 52.0 percent of the variance in pelvic spine length and 56.0 percent of the variance in pelvic girdle length. We estimated the evolutionary response on genotypes using SNPs within 20 cM of these LOD peaks. QTL maps for individual F1 crosses suggested that the location of peaks contributing to first dorsal spine length was variable (Appendix C.3). To account for this variation genotypic estimates of   60 selection on first dorsal spine were also made using the LOD peaks identified within each family, a minimum LOD peak of 2 was used (> 10 percent variance explained within that family). In total there were five chromosomes (IV, VII, IX, XI, and XV) that had large contributions to variation in first dorsal spine length (see Appendix C.4 for PVE values). Two F1 families shared the peak on chromosome XV, while the peaks on other chromosomes were unique to each family. There was no variation in the location of pelvic spine or pelvic girdle QTLs (individual family PVEs reported in Appendix C.4).   4.2.5 Selection Analyses We estimated standardized univariate selection differentials (intensities, s’) between sampling periods (September to January and January to April for the five armour phenotypes of interest as:  s' = (𝑥after - 𝑥before)/𝜎pooled, where 𝑥before and 𝑥after are the phenotype trait means before and after selection and 𝜎pooled is the square root of pooled sample variance of the trait before and after selection (Lande and Arnold, 1983). We estimated the treatment effect on selection within a generation (Δs') as:  Δs' = s't - s'c, where, s't is the selection intensity in the treatment pond and s'c is the selection intensity in the control pond. Between generations, we estimate the evolutionary response in Haldanes (h) as: Δ𝑍 = (𝑧after - 𝑧before)/𝜎pooled,   61 where 𝑧before and 𝑧after are the mean phenotype or allele frequency in the generation before and after selection and 𝜎pooled is the square root of pooled sample variance of the trait or allele frequency in the first and second generation, and ℎ = !!! , where the evolutionary response, Δ𝑍, is divided by g, the number of generations of selection, which in our case is 1 (Hendry and Kinnison, 1999).  We estimate the treatment effect within a pond pair between generations (Δh) as: Δh = ht - hc   where ht is the evolutionary response in the treatment pond and hc is the evolutionary response in the control pond.              We used the same formulas and methods to estimate the evolutionary response and treatment effect for allele frequencies at SNP markers near the QTLs for pelvic spine and first dorsal spine length. For both genotype and phenotype the statistical significance of the mean selection intensity, mean evolutionary response and treatment effects were determined using a one-sample t-test using pond pairs as replicates. All p-values are two-tailed. Pelvic spine and pelvic girdle showed the same effect (due to being highly correlated see Appendices C.5 - C.8 for variance-covariance matrices); so only the results for pelvic spine are reported.             We used linear models to describe the phenotypic trait trajectories through time. These models included a quadratic term which allows us to model curvature in the trajectories through time.  All statistical analyses were conducted in R (version 3.1.2, R Development Core Team 2015).    62 4.3 Results 4.3.1 Phenotype Selection within a generation  We tracked selection on the first pond generation, which was the F2 cohort in four of the five pond pairs, and the offspring of wild hybrids from First Lake in the fifth pond pair. Selection coefficients measured across this first pond generation showed that armour was strongly selected against in juvenile stickleback and temporally variable. In the fall, both dorsal spines and pelvic spines of juvenile stickleback were strongly selected against regardless of treatment (Figure 4.3, Table 4.1). There was no significant selection on lateral plates during this period (Figure 4.3, Table 4.1). The only trait that showed a significant treatment effect for selection intensity during the fall was second dorsal spine, where there was stronger selection against length for treatment ponds relative to control ponds (see Appendix C.12 for the remaining traits and statistics). In the winter selection against armour weakened, and more armour was favoured for some armour traits. More lateral plates and a longer second dorsal were significantly favoured in both treatments (Figure 4.3, Table 4.1). During the winter period there was no significant selection on first dorsal spine or pelvic spine (Figure 4.3, Table 4.1), and no traits showed a significant treatment effect (Appendix C.12).          63 Table 4.1 Mean selection intensity and standard error for all ponds in the fall (September – January ) and winter (January – April) seasons.  Estimates significantly different from zero are in bold.    s'   T7  P-value  Fall First Dorsal Spine -0.30 ± 0.07 -4.24 0.004  Second Dorsal Spine -1.02 ± 0.11 -9.28 < 0.001  Pelvic Spine -0.15 ± 0.04 -4.26 0.004  Lateral Plates 0.01 ± 0.06 0.12 0.91 Winter First Dorsal Spine 0.03 ± 0.06 0.44 0.67  Second Dorsal Spine 0.44 ± 0.13 3.32 0.01  Pelvic Spine -0.01 ± 0.04 -0.34 0.75  Lateral Plates 0.18 ± 0.08 2.34 0.05   64 Figure 4.3 Selection intensity within the first pond generation for A) first dorsal spine, B) second dorsal spine, C) pelvic spine and D) lateral plates in treatment and control ponds over the fall (September – January) and winter (January – April) seasons.  Black circles are the mean of control ponds and black triangles are the mean of treatment ponds, with one standard error above and below the mean. Treatment ponds are indicated in triangles and control ponds are circles, each unique color indicates a pond pair.   A BDCSelection IntensitySelection IntensitySelection IntensitySelection IntensityControlTreatment-0.50-0.250.000.25September - January January - April-1.5-1.0-0.50.00.51.0September - January January - April-0.3-0.2-0.10.00.1September - January January - April-0.250.000.25September - January January - April  65 Evolutionary Response Between generations  In addition to tracking selection on the first pond generation, we characterized the evolutionary response using the offspring of the first generation of pond fish. We also used the entire time course to estimate trait trajectories, which we compared between treatment and control ponds. Our within generation estimates of selection indicated that there was a reduction in armour during the fall period in both control and treatment ponds. However, over the winter and moving into the spring and summer we see the recovery of first dorsal spine, second dorsal spine and pelvic spines, as evidenced by a significant upward (positive) curvature in the trait trajectories of treatment ponds (Appendices C.11 and C.13). As a result we see the maintenance of armour after one generation of selection in treatment ponds (C.10). In contrast for control ponds we see a negative trajectory and non-significant curvature for first dorsal spine and pelvic spines, consistent with a loss of armour over time (Appendices C.11 and C.13). The trait trajectories also indicated that there was significant heterogeneity in the evolutionary response exhibited among treatment replicates. For treatment ponds the interaction between the shape of the trait trajectory and family (replicate) was significant for all traits (p < 0.006). For control ponds the evolutionary response differed significantly among replicates for second dorsal spine and lateral plates (p < 0.03), but not first dorsal spine or pelvic spine (p > 0.18).  Estimates of evolutionary response also indicated that, after one generation of selection, on average more robust armour was favoured for all armour traits in the treatment ponds relative to control ponds (Figure 4.4, Table 4.2). This was the result of evolutionary decline in the control ponds and maintenance of armour in the treatment ponds (Appendix C.10). First dorsal spines were significantly longer in treatment ponds relative to control ponds (Table 4.2). There was a non-significant trend for the other armour traits. Second dorsal spine was longer and lateral   66 plates were more numerous for four of five treatment ponds (Table 4.2). Longer pelvic spines were favoured in three treatment ponds relative to paired control ponds (Table 4.2).   Figure 4.4 Divergent evolutionary response of armour phenotypes to treatment after on generation of selection.  The response was measured from September 2012 in the first pond generation to September 2013 in the second generation. Positive values indicate higher trait values in the treatment pond relative to its paired control pond.  Black circles are the mean of pond pairs, with one standard error above and below the mean. Each pond pair is indicated by a unique color.    −1012First Dorsal     SpineSecond Dorsal        SpinePelvic Spine Lateral PlatesDifference in Evolutionary Response (Δh)  67 Table 4.2 Divergent evolutionary response, delta haldanes (Δh), of armour phenotypes to treatment after one generation of selection.  Treatment estimates significantly different from zero are in bold.   Δh T4 P value First Dorsal Spine  0. 58 ± 0.16  3.6 0.02 Second Dorsal Spine 0.72 ± 0.44 1.6 0.18 Pelvic Spine 0.18 ± 0.23 1.1 0.48 Lateral Plates 0.19 ± 0.33 0.79 0.33  4.3.2 Genotype The difference in evolutionary response estimated for phenotypes was generally matched to changes in allele frequency change at SNPs near QTL peaks for first dorsal and pelvic spine length. Mean treatment effects estimated for SNPs at the QTL corresponding to the Msx2 locus and Pitx1 locus were not significant (Msx2 estimate = 0.16 ± 0.13 SE, T3 = 1.25, p = 0.30; Pitx1 estimate = -0.15 ± 0.22 SE, T3 = -0.67, p = 0.55) (Figure 4.5). However, we did find that the direction of evolution corresponded to that seen for these phenotypes. For Pitx1 limnetic alleles were favoured in two of the four treatment ponds relative to the paired control ponds (Figure 4.5), and these were the same two ponds in which we saw longer pelvic spines favoured (Figures 4.4). For the Msx2 region we saw limnetic alleles favoured in three of the four treatment ponds relative to the paired control ponds (Figure 4.5), whereas for the phenotype we saw longer spines favoured for all four pond pairs (Figure 4.4). Our QTL analysis suggested there was variation among F1 families in the location of the genomic regions influencing first dorsal spine length and the magnitude of their contribution (Appendix C.3). When we looked at all of the chromosomal regions accounting for at least 10% of the variation in first dorsal spine length, we found that   68 there was a significant positive treatment response for three of four families (p < 0.005) (Appendices C.4 and C.14), and all four families had at least one positive mean estimate (Figure 4.6).  Figure 4.5 Divergent evolutionary response of armour genotypes within the Msx2 and Pitx1 genomic regions to treatment after one generation of selection.  Estimates were measured from September 2012 in the first generation to September 2013 in the second generation. Positive values indicate selection for limnetic alleles (associated with longer spines) in the treatment pond relative to its paired control pond. Each colored circle is the mean response across SNPS for a pair of ponds. Black circles are the mean of all pond pairs, with one standard error above and below the mean. Each treatment – control pond pair is indicated by a unique color.   -0.50-0.250.000.250.50Msx2 Pitx1Difference in Evolutionary Response Δh  69 Figure 4.6 Divergent evolutionary response to treatment after one generation of selection estimated at SNPs near first dorsal spine QTL peaks.  Estimates were measured from September 2012 in the first generation to September 2013 in the second generation. Positive values indicate selection for limnetic alleles (associated with longer spines) in the treatment pond relative to its paired control pond. The chromosome on which the LOD peak is found is indicated on the x-axis. Each colored circle represents a SNP within 2 cM of a LOD peak for first dorsal spine. Black circles are the mean effect of the SNPs within the QTL region, with one standard error above and below the mean. Each treatment – control pond pair is indicated by a unique color.  -0.50.00.51.0ChrXI ChrXVChrXV ChrIVChrVII ChrIXDifference in Evolutionary Response ΔhFamily 1 Family 2 Family 3 Family 4  70 4.4 Discussion 4.4.1 Species Interactions and Divergence This experiment was designed to test whether differential exposure to cutthroat trout could lead to divergent evolution of armour traits. Our results suggest that the presence of trout maintained armour, which was reduced in trout-absent ponds. This was most apparent for first dorsal spine where, after one generation of selection, all treatment ponds had longer spines relative to control ponds. For other traits the trend was suggestive, with more robust spines and more lateral plates, found in three or four of the treatment ponds and significant upward trajectories for all treatment ponds over the course of the experiment. The reduction of armour in the control ponds suggests that armour is costly in the absence of vertebrate predators and selected against. The reasons for the decline are not known. This may be due to predation by invertebrates, primarily in the juvenile life stage. Previous experimental work that manipulated invertebrate predation found that dorsal spine and pelvic girdle length are selected against when invertebrates are present (Marchinko, 2009), which is consistent with our observations.  We find significant variation among treatment replicates in the response to selection, both in terms of direction and strength. Because of this the treatment effect was only statistically significant for one of the four traits, first dorsal spine. This variability in response among treatment ponds could be due to differences in the target population, for example in size distributions (which could affect susceptibility to predation), the genetic variation present for selection to act upon, or the starting population density. It could also be due to differences among replicates in the risk of predation, either due to differences in the amount of vegetation within the ponds (which were used as areas of refuge), or differences in the voraciousness of predators (due   71 to variation in diet preference or otherwise). Alternatively, it could be the case that coastal cutthroat trout do not impose significant selection on all of the components of armour. Resource competition has driven character displacement of trophic morphology in the wild and led to differential habitat use (Schluter and McPhail, 1992), which has exposed benthic and limnetic stickleback to different predation regimes (Remichen, 1994). Our results suggest that this secondary effect of differential trout predation has played an important role in the divergence of benthic and limnetic stickleback. We find decreased armour in the absence of trout predation, which is consistent with the pattern of character divergence found for armour in the wild (Vamosi and Schluter, 2004). It remains to be determined whether armour reduction in the benthic species is driven primarily through a reduction in trout predation, a relative increase in invertebrate predation or both.   4.4.2 Temporal Variation in Selection Patterns of selection within the first generation predict the evolutionary response seen between generations. We find that armour is strongly selected against in juvenile stickleback even in the presence of cutthroat trout. Later in the year we see the direction of selection turns, with selection favouring more highly armoured fish in treatment ponds, presumably due to a survival advantage in the face of trout predation. In control ponds we see a lack of turn around over the spring, which results in the reduction in armour we see after one generation of selection. The strong selection we see against armour in the fall may be due to the susceptibility of juvenile stickleback to invertebrate predators, which selectively prey upon smaller stickleback (Reimchen, 1980; Foster et al., 1988). Additionally trout predation was probably not intense in the fall season, as large trout (30 – 40 cm) do not often forage on stickleback below 30 mm   72 (Reimchen, 1990). Predation on stickleback by trout is thought to be most prevalent in the warmer months, increasing in frequency from May onwards (Reimchen, 1990). In the early spring we generally saw that selection against armour weakened or was alleviated completely. This release of negative selection could be due to the fact that most stickleback were no longer of a size where they were most susceptible to invertebrate predation (in April the smallest fish in each pond was at least 20 mm and mean was at least 34 mm). It has been previously shown that dragonfly nymphs prefer to prey upon stickleback 15 – 25 mm in size and backswimmers struggle to prey on individuals larger than 15 mm (Foster et al., 1988).   4.4.3 Selection on Underlying Genes We sought to determine whether the divergent evolutionary responses characterized for armour traits were matched by divergent allele frequency changes at their underlying genes (genomic regions). Previous work has indicated that Pitx1 is a major determinant of pelvic spine and girdle length (Shapiro et al., 2004). We find that the evolutionary response estimated from SNP allele frequencies near the Pitx1 locus is similar to that seen for the pelvic phenotypes. The correspondence between estimates at the genotypic and phenotypic levels is not surprising given the generally strong correlation between marker state and phenotype, with Pitx1 accounting for 51-63% of the variance in pelvic spine length. This pattern suggests that the response we see for pelvic spines is likely genetically determined. Our results also suggest that selection for limnetic alleles at the Msx2 locus likely contributes to the phenotypic response seen in first dorsal spine, however this is not the case for all families. Variation in the evolutionary response observed at the Msx2 locus, among families, could be due to the variation we see in the genomic regions affecting first dorsal spine length. When we look within a family at the genomic region(s) that   73 explain the most variation in phenotype we find stronger evidence of allele frequency changes that match the phenotypic shifts between treatment and control ponds, but the genetic basis of adaptation is unique for each family. Linkage disequilibrium and pleiotropy likely influenced the observed evolutionary responses of armour traits. This experiment used F2 fish from F1 families, which means that large blocks within each linkage group were co-segregating. Previous genomic and QTL mapping studies in stickleback have shown that many traits that are ecologically important, including some of the armour traits surveyed in this study, map to regions of the genome with low rates of recombination (Hohenlohe et al., 2011; Conte et al., 2015; this study). Within such regions it is unlikely that any recombination occurred during this experiment. As a result some of the observed shifts in armour may have been constrained and others enhanced by selection on genes that were in linkage disequilibrium with the genes encoding armour traits. For example in one treatment pond we see selection against limnetic alleles at a locus on chromosome VII that influences first dorsal spine despite longer first dorsal spines being favoured; this may be explained by linkage disequilibrium between this gene region and the Pitx1 region also found on chromosome VII, which also showed strong selection against limnetic alleles in the same treatment pond. Future analyses will be needed to disentangle direct selection on the focal armour traits from that on other traits also determined by the same underlying genes (pleiotropy) or other genes in tight linkage.  There are a few cases in which genotypes have been explicitly linked to fitness (e.g. Barrett et al., 2008; Gompert et al., 2014; Lamichhaney et al., 2016), however for most of these cases the agent of natural selection has not been identified. Here we looked for evidence of divergent natural selection due to predation within genomic regions shown to influence armour   74 phenotypes. We documented the mechanisms and targets of selection that drove changes in allele frequency and found evidence of the direct effects of alternative alleles on ecologically relevant phenotypes. Further work will be required to disentangle the contribution of pleiotropy and linkage to our observed patterns. However, our work suggests that the evolutionary response to predation is polygenic and influenced by the genetic architecture of each trait. Our approach has allowed a more accurate characterization of the process of adaptation to species interactions and suggests that they are indeed important drivers of divergent evolution in the wild.      75 Chapter 5: General Discussion and Future Directions My dissertation seeks to further our understanding of the ecological and genetic mechanisms responsible for evolutionary change in natural populations using the threespine stickleback as a model system. This pursuit entails testing for the evidence of natural selection on spectral sensitivity (Chapter 3), estimating the strength of selection on genotypes and phenotypes to disentangle the contribution of direct and indirect selection (Chapter 2) and testing the contribution of predation to phenotypic and genotypic diversification (Chapter 4). I focus on two suites of traits: the components of bony armour and the opsin proteins that underlie spectral sensitivity. I use observational and experimental work to link these phenotypes and their underlying genes to fitness in threespine stickleback.  5.1 Local Adaptation of Threespine Stickleback  Stickleback researchers have studied shifts in bony armour for decades (Bell and Foster, 1994). Armour is repeatedly reduced in freshwater populations relative to their marine ancestors (Bell and Foster, 1994); this pattern of parallel evolution suggests local adaptation. However, evidence that armour traits are under direct selection has been relatively elusive and the mechanism of selection is generally unknown (but see Marchinko, 2008, which suggests that insect predation selects for less armour). Three environmental factors are most often suggested to drive selection on armour: minerals, invertebrate predators and vertebrate predators (Reimchen, 1980; Giles, 1983; Bell et al., 1993).  My work in Chapters 2 and 4 provides strong evidence that variation in armour traits is a product of local adaptation and gives insights into the ecological and genetic factors shaping the evolution of these traits. In Chapter 2 I show some evidence that suggests there is direct selection on lateral plates when statistical methods are used to control for the selection on other traits   76 encoded by Ectodysplasin (Eda) locus, which is the major gene influencing lateral plate number. This suggests that changes in lateral plate number during freshwater colonization may indeed be adaptive and not just a correlated response due to selection on another trait influenced by Eda. In Chapter 4 I tested whether the presence/absence of coastal cutthroat trout predation led to divergence in armour between treatments. I found divergent evolution between treatments, with components of armour consistently reduced when cutthroat trout were absent. This pattern of divergence between treatments is consistent with the patterns of trait differentiation found in the wild, suggesting local adaptation.  Decades have also been spent studying adaptation of the visual system (reviewed in Bowmaker, 2009), although the majority of this work has not been conducted in stickleback. Particular attention has been given to characterizing shifts in the spectral sensitivity of species exposed to different light environments (e.g. Bowmaker et al., 1994; Carvalho et al., 2006; Seehausen et al., 2008). The idea behind this work has been to determine what features of the visual system respond to changes in ambient light as well as the functional explanation for these shifts. In Chapter 3 I sought to test whether there was a signature of local adaptation for spectral sensitivity. I find heritable shifts in the spectral sensitivity between marine and freshwater populations that are repeatedly correlated with shifts in ambient light between these habitats. Together these findings are also consistent with local adaptation of spectral sensitivity to ambient light.  5.2 Broader Implications 5.2.1 Adaptation to Freshwater  The marine to freshwater transition is encountered by many fish species on either an ecological (i.e. in anadromous species) or evolutionary timescale (i.e. during freshwater   77 colonization) (Bell and Andrews, 1997), and the marine to freshwater boundary is a major axis of fish diversity, having initiated speciation in many taxa (Lee and Bell, 1999; Bloom et al., 2013). Salinity always differs between these environments, but there are likely additional factors that commonly differ (e.g. light profiles, predation, temperature, mineral availability). Adaptations to any of these factors may evolve repeatedly in the many fish species that transition across this habitat gradient. One factor known to differ repeatedly between marine and fresh waters is the spectral composition of light (Kirk, 1977; Kirk, 1994). In Chapter 3, I show that there have been repeated shifts in the spectral sensitivity of freshwater threespine stickleback populations relative to their marine ancestors. I find that freshwater populations have shifted their sensitivity towards longer wavelengths of light and have significantly reduced their sensitivity to ultra-violet wavelengths. While this finding is only documented in threespine stickleback, reduced sensitivity to short wavelengths could be a general phenomenon among species that inhabit marine and freshwaters. This is of additional interest because it has the potential to affect reproductive isolation between marine and freshwater populations that still come into secondary contact (discussed in Chapter 3). Future work will be required to see if this is indeed the case.  5.2.2 Ecological Speciation  In 1859 Darwin emphasized the importance of ecologically mediated selection in the origin of species. Decades later it was shown that local adaptation, driven by divergent selection, could directly lead to reproductive isolation in a process termed ecological speciation (reviewed in Schluter, 2001). As outlined in Chapters 3 and 4, I provide strong evidence of local adaptation to divergent light environments and predation regimes. Because spectral sensitivity and armour are under divergent natural selection they have the potential to contribute to the evolution of   78 reproductive isolation. To demonstrate this, however, the traits and/or genes involved need to be shown to reduce gene flow between the species.  Previous work in East African cichlid fish suggests that adaptation of the visual system to divergent light conditions has led to the divergence of mate preference and correspondingly the evolution of pre-mating isolation in a process called sensory drive (Seehausen et al., 2008). When there are differences in signaling conditions between the environments of closely related species the sensory drive hypothesis makes two predictions: 1) That the sensory system of each species should evolve to increase perception under local ecological conditions (Levine and MacNichol, 1979) and 2) That male sexual signals should evolve to match the sensitivity of the female’s sensory system (Endler, 1992; Ryan, 1990). Although sensory drive is intuitively appealing and proposed to be an important driver of speciation, little empirical evidence exists to support the sensory drive hypothesis. Currently the best-documented example of sensory drive is found in Lake Victoria cichlids (Seehausen et al., 2008). Previous work has suggested that benthic and limnetic threespine stickleback may also exemplify this phenomenon (Boughman, 2001). In Chapter 3 I provide some support for the first prediction of the sensory drive hypothesis: my results suggest that there has been divergence of the spectral sensitivity of benthic and limnetic threespine stickleback and that this divergence corresponds to differences in the local light environment. It is possible that these differences in spectral sensitivity have affected the mate choice preferences of benthic and limnetic fish and may contribute to reproductive isolation. However, future work will be required to test the second prediction of the sensory drive hypothesis, which in this example is that male colouration has evolved to match female spectral sensitivity.    79 In Chapter 4 my results suggest that armour is favoured in the presence of trout and selected against in their absence. If this is true in the wild then differential predation regimes could also directly contribute to post-zygotic extrinsic reproductive isolation between benthic and limnetic threespine stickleback. There are two mechanisms by which reductions in gene flow could occur: 1) through selection against F1 hybrids, which possess intermediate trait values and as a consequence are poorly adapted to predation in either habitat, and 2) selection against advanced generation hybrids that exhibit negative epistasis between armour traits.  5.2.3 Secondary Effects of Species Interactions  Competition has previously been shown to be an important source of divergent selection in the sympatric benthic and limnetic species pairs (Schluter and McPhail, 1992). It has been suggested that divergence in habitat use, driven by resource competition, has also exposed each species to unique ecological factors within each environment  - so called “knock-on effects” (e.g. Vamosi and Schluter, 2004). Two ecological factors that have significant potential as knock-on effects are predation and ambient light.  Armour has been found to be more divergent between the sympatric benthic and limnetic threespine stickleback than between allopatric populations (Vamosi and Schluter, 2004), a pattern similar to character displacement. A knock-on effect and or an interaction between predation and competition may be what drives this pattern. In Chapter 4 I show that different armour phenotypes are favoured in the presence and absence of cutthroat trout. It could be that competition drove benthic and limnetic threespine stickleback into their respective habitats and within these habitats there is differential selection by predators. This is in contrast to solitary (allopatric) populations of threespine stickleback, where individuals are exposed to both types of   80 predators. Alternatively, competition may act in tandem with predation to strengthen the overall amount of divergent selection, which then exaggerates divergence in sympatry.  In Chapter 3 I describe differences between the benthic and limnetic habitats in the properties of ambient light. I then go on to show that there has been some divergence of spectral sensitivity between the species and that these differences correspond to differences in the light environment. It could be the case that this is a secondary effect of competition, where divergent selection due to resource competition has driven divergence in spectral sensitivity through the exposure of individuals to unique light environments. I suggest that it may often be the case that resource competition promotes habitat differentiation that exposes prey species to new environmental factors that act synergistically to drive phenotypic divergence and potentially the evolution of reproductive barriers. These hypothesized knock-on effects will be an interesting avenue for further study in the interaction of different sources of divergent selection. 5.2.4 Limitations of Threespine Stickleback as a Study System While the evolutionary processes underlying many of our findings likely extend to other species there are a few factors about the stickleback system that may limit the universality of our findings. For example adaptation from standing genetic variation seems particularly important in threespine stickleback (Colosimo et al., 2005); there is standing variation, for relatively ancient freshwater alleles, found in contemporary marine populations and fixation of these alleles appears to have repeatedly led to phenotypic shifts in freshwater populations (Colosimo et al., 2005). However, it is currently unclear whether adaptation from standing variation is as common in other taxa or a result of the repeated colonization of freshwater by marine stickleback after multiple episodes of glaciation.   81 In threespine stickleback the evolution of reproductive isolation and phenotypic divergence between populations has occurred very rapidly (within the last 12,000 years); strong-divergent natural selection and standing genetic variation are both thought to have played a role. In other systems the timing of the evolution of reproductive isolation and strength of natural selection is generally unknown, consequently it is uncertain whether the rate of evolution in stickleback is atypical, and thus not representative of the general process.  5.3 Future Directions for the Study of Adaptation  While this thesis makes progress understanding the link between divergent selection and trait divergence, there are still a number of outstanding questions that require further attention. While not a comprehensive list, I will give an overview of the contribution of my work in each of these areas and indicate some of the next steps for future research.  5.3.1 Strength of Natural Selection  My findings in Chapters 2 and 4 suggest that natural selection on armour phenotypes can be very strong. Across many studies phenotypic selection is usually weak; a meta-analysis of phenotypic estimates of selection from a variety of organisms has shown that the distribution of selection intensities is exponential, with most estimates close to zero (Kingsolver et al., 2001). Although there are far fewer estimates of selection estimated from genotypes, a recent meta-analysis suggests that the distribution of genotypic selection coefficients is also exponential (Thurman and Barrett, 2016).  If my estimates of selection on armour phenotypes are compared to the distribution of phenotypic selection intensities (standardized selection differentials) collected by Kingsolver et al., we see that on average they also correspond to an exponential distribution (Figure 5.1). However, a couple of my estimates are somewhat exceptional in their   82 magnitude. For example, in the fall season I find that selection on second dorsal spine falls within the top 3% of all previously reported estimates.   Figure 5.1 Comparison of absolute selection intensity (s’) on armour to the distribution compiled by Kingsolver et al., (2001).  Colored lines indicate selection intensity estimates on individual armour traits during the fall and winter seasons of the predation experiment (Chapter 4). The frequency distribution of absolute selection intensity (s’) from the Kingsolver et al. database is plotted in light grey.    After 150 years of studying natural selection, we are starting to understand the distribution of the strength of selection in nature (Kingsolver et al., 2001; Thurman and Barrett, 2016). However, there are still many outstanding questions regarding the strength of selection and how it varies. For example it is currently unknown whether the strength of selection is equivalent for abiotic and biotic factors. Both are known to play a role in diversification, but are 0501001502000 1 2 3Second Dorsal SpineFirst Dorsal SpineLateral PlatesPelvic GirdlePelvic SpineWinterFallAbsolute Selection IntensityFrequency  83 the magnitude of their contributions similar? To address this question it could be useful to update the Kingsolver et al. database and partition it into biotic and abiotic sources of selection to explicitly test whether there is a difference in the average magnitude of selection that each imposes. We also do not know whether the strength of selection in manipulative experiments is similar to that occurring in the wild.  This is an important question that should be addressed in the literature, since the utility of manipulative experiments rests on the assumption that they are representative of natural processes. Again a meta-analysis could be a useful tool to tackle this question.  My findings in Chapter 4 suggest that it is likely that both predation and competition have played a role in the divergence of benthic and limnetic threespine stickleback. However, we know very little about how different sources of divergent selection interact in this system or otherwise. Do different sources of selection act synergistically, increasing the total amount of selection along a common axis or does one weaken the contribution of the other? Preliminary work has been undertaken looking at the interaction between predation and competition (Rundle et al., 2003) and their relative contributions (Meyer and Kassen, 2007), however more of this type of work is needed. It is also an outstanding question whether strong selection by one factor can result in the same amount of divergence as relatively weak selection by multiple factors. Manipulative experiments isolating and combining multiple agents of selection will be crucial in teasing apart the interplay between these processes.  5.3.2 The Genetic Basis of Adaptation My work in Chapter 2 suggests that pleiotropy and/or tight genetic linkage may have played an important role in the divergence of armour traits between marine and freshwater environments. In nature we know very little about the prevalence of pleiotropy or the role it   84 plays in adaptive evolution. Pleiotropy has the potential to constrain adaptation by limiting the variants available for selection to act upon. Pleiotropy could contribute to repeatability of adaptive evolution (parallelism), due to a decreased number of genes available for adaptive evolution. Pleiotropy also has the potential to spur diversification if a gene under natural selection exhibits pleiotropy for reproductive isolation. More work investigating the role of pleiotropy in adaptation and speciation will be required in the coming years.    A limitation of the work in Chapters 2 and 4 is that we don’t know whether the observed trait changes in the experiments arise by selection on novel mutation or the variation present at the beginning of the experiment. In threespine stickleback, adaptation from standing genetic variation seems to be important in the wild (Colosimo et al., 2005). However, more generally we do not know the relative frequency of adaptation from standing genetic variation compared to novel mutation (Barrett and Schluter, 2008). Most theory on the genetics of adaptation has focused on adaptation from novel mutation or quantitative genetics (Orr et al., 2005). As a result we do not have strong theoretical predictions for the process of adaptation from standing genetic variation (Barrett and Schluter, 2008). Future empirical and theoretical work should focus on determining the prevalence of adaptation from standing genetic variation and characterizing how it differs from novel mutation.      Lastly, the study of adaptation is moving quickly towards obtaining a genomic view of natural selection. As a result much of the current work in the field of adaptation has moved away from direct observations of contemporary natural selection. Focus is now on using statistical methods, which look for reductions in nucleotide polymorphism, to infer the action of natural selection in the past. My work in Chapters 2 and 4 demonstrate the importance of direct measurements of selection on phenotypic traits and the genes underlying those traits for   85 elucidating the mechanisms of evolution. In our case study the true pattern of selection on lateral plates was obscured until we were able to statistically disentangle selection on Eda from the direct selection on lateral plates. So I suggest that direct estimates of selection should remain in the forefront of our minds as we fully embrace the genomic era of adaptation.         86 Bibliography Albert, A. Y., N. P. Millar, and D. Schluter. 2007. Character displacement of male nuptial colour in threespine sticklebacks (Gasterosteus aculeatus). Biological Journal of the Linnean Society 91: 37–48.  Anderson, J. T., C. R. Lee, and T. Mitchell-Olds. 2014. Strong selection genome-wide enhances fitness trade-offs across environments and episodes of selection. Evolution 68: 16–31.   Araya, C. L., C. Payen, M. J. Dunham, and S. Fields. 2010. Whole-genome sequencing of a laboratory-evolved yeast strain. BMC Genomics 11: 1.  Arnegard, M. E., M. D. McGee, B. Matthews, K. B. Marchinko, G. L. Conte, S. Kabir, N. Bedford, S. Bergek, Y. F. Chan, F. C. Jones, D. M. Kingsley, C. L. Peichel, and D. Schluter. 2014. Genetics of ecological divergence during speciation. Nature 511: 307–311.  Barrett, R. D., and H. E. Hoekstra. 2011. Molecular spandrels: tests of adaptation at the genetic level. Nature Reviews Genetics 12: 767–780.  Barrett, R. D. H., and D. Schluter. 2008. Adaptation from standing genetic variation. Trends in Ecology and Evolution 23: 38–44.  Barrett, R. D. H., S. M. Rogers, and D. Schluter. 2008. Natural selection on a major armour gene in threespine stickleback. Science 322: 255–257.  Barrett, R. D. H. 2009. Environment specific pleiotropy facilitates divergence at the Ectodysplasin locus in threespine stickleback. Evolution 63: 2831–2837.  Barrett, R. D. H., and D. Schluter. 2010. Clarifying mechanisms of evolution in stickleback using field studies of natural selection on genes. In P. Grant and R. Grant, eds. In search of the causes of evolution: from field observations to mechanisms. Princeton University Press, Princeton, NJ. pp. 332–346  Barrett, R. D. H., T. H. Vines, J. S. Bystriansky, and P. M. Schulte. 2009. Should I stay or should I go? The Ectodysplasin locus is associated with behavioural differences in threespine stickleback. Biology Letters 5: 788–791.   Beaumont, M. A., and D. J. Balding. 2004. Identifying adaptive genetic divergence among populations from genome scans. Molecular Ecology 13: 969–980.   Bell, M. A., W. E. Aguirre, and N. J. Buck. 2004. Twelve years of contemporary armour evolution in a threespine stickleback population. Evolution 58: 814–824.     87 Bell, M. A., and C. A. Andrews. 1997. Evolutionary consequences of postglacial colonization of fresh water by primitively anadromous fishes. In Evolutionary ecology of freshwater animals. Birkhäuser Basel. pp. 323–363.  Bell, M. A., and S. A. Foster, eds. 1994. The evolutionary biology of the threespine stickleback. Oxford University Press, Oxford.   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.   Benjamini, Y., and D. Yekutieli. 2001. The control of the false discovery rate in multiple testing under dependency. Annuls of Statistics 29: 1165–1188.  Bergstrom, C. A. 2002. Fast-start performance and reduction in lateral plate number in threespine stickleback. Canadian Journal of Zoology 80: 207–213.   Bernatchez, L., and J. J. Dodson. 1990. Allopatric origin of sympatric populations of lake whitefish (Coregonus clupeaformis) as revealed by mitochondrial-DNA restriction analysis. Evolution 44: 1263–1271.  Bloom, D.D., Weir, J.T., Piller, K.R. and Lovejoy, N.R., 2013. Do freshwater fishes diversify faster than marine fishes? A test using state – dependent diversification analyses and molecular phylogenetics of new world silversides (Atherinopsidae). Evolution 67: 2040–2057.  Bolger, A. M., M. Lohse, and B. Usadel. 2014. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics p.btu170.  Bonnier, G. and C. Flahault. 1878. Observations sur les modifications des vegetaux suivant les conditions physiques du milieu. Annuals Science Nature 6: 93–125.  Boughman, J. W. 2001 Divergent sexual selection enhances reproductive isolation in sticklebacks. Nature 411: 944–948.  Boulding, E. G., and K. L. Van Alstyne. 1993. Mechanisms of differential survival and growth of two species of Littorina on wave-exposed and on protected shores. Journal of Experimental Marine Biology and Ecology 169: 139–166.  Bowmaker, J. K. 2008. Evolution of vertebrate visual pigments. Vision Research 48: 2022–2041.  Bowmaker, J. K., and Y. W. Kunz. 1987. Ultraviolet receptors, tetrachromatic colour vision and retinal mosaics in the brown trout Salmo trutta: Age-dependent changes. Vision Research 27: 2101 – 2108.    88 Bowmaker, J. K., V. L. Govardovskii, S. A. Shukolyukov, L. V. Zueva, D. M. Hunt, V. G. Sideleva, and O.G. Smirnova. 1994. Visual pigments and the photic environment: the cottoid fish of Lake Baikal. Vision Research 34: 591–605.  Broman, K. W., and H. Wu. 2013. QTL: Tools for Analyzing QTL Experiments. CRAN: Comprehensive R Archive Network. Available at: http://www.r-project.org.  Bumpus, H. C. 1899. The elimination of the unfit as illustrated by the introduced sparrow, Passer domesicus. In Biological Lectures Delivered at the Marine Biological Laboratory of Wood's Holl, 1896-97. Boston: Ginn & Co pp. 209 – 226  Burke, M. K., J. P. Dunham, P. Shahrestani, K. R. Thornton, M. R. Rose, and A. D. Long. 2010. Genome-wide analysis of a long-term evolution experiment with Drosophila. Nature 467: 587–590.   Carleton, K. L., and T. D. Kocher. 2001. Cone opsin genes of African cichlid fishes: tuning spectral sensitivity by differential gene expression. Molecular Biology and Evolution 18: 1540–1550.  Carvalho, L. D. S., J. A. Cowing, S. E. Wilkie, J. K. Bowmaker, and D. M. Hunt. 2006. Shortwave visual sensitivity in tree and flying squirrels reflects changes in lifestyle. Current Biology 16: R81–R83.  Chan, Y. F., M. E. Marks, F. C. Jones, G. Villarreal, M. D. Shapiro, S. D. Brady, A. M. Southwick, D. M. Absher, J. Grimwood, J. Schmutz, Myers, R.M. et al. 2010. Adaptive evolution of pelvic reduction in sticklebacks by recurrent deletion of a Pitx1 enhancer. Science 327: 302–305.  Clarke, G. L. 1936. On the depth at which fish can see. Ecology 17: 452–456.  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.   Conte, G. L., M. E. Arnegard, J. Best, Y. F. Chan, F. C. Jones, D. M. Kingsley, D. Schluter, and C. L. Peichel. 2015. Extent of QTL reuse during repeated phenotypic divergence of sympatric threespine stickleback. Genetics 201: 1189–1200.  Darwin, C. 1859. The origin of species. Penguin, London.  Davey, J. W., P. A. Hohenlohe, P. D. Etter, J. Q. Boone, J. M. Catchen, and M. L. Blaxter, 2011. Genome-wide genetic marker discovery and genotyping using next-generation sequencing. Nature Reviews Genetics 12: 499–510.    89 Deagle, B. E., F. C. Jones, Y. F. Chan, D. M. Absher, D. M. Kingsley, and T. E. Reimchen. 2011. Population genomics of parallel phenotypic evolution in stickleback across stream–lake ecological transitions. Proceedings of the Royal Society of London B: Biological Sciences 279: 1277–1286  DeFaveri, J., and J. Merila. 2013. Evidence for adaptive phenotypic differentiation in Baltic Sea sticklebacks. Journal of Evolutionary Biology 26: 1700–1715.   Denton, E. J., and F. J. Warren. 1957 The photosensitive pigments in the retinae of deep-sea fish. Journal of the Marine Biological Association 36: 651–662.  DePristo, M. A., E. Banks, R. Poplin, K. V. Garimella, J. R. Maguire, C. Hartl, A. A. Philippakis, G. Del Angel, M. A. Rivas, M. Hanna, and A. McKenna. 2011. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nature Genetics 43: 491–498.  Ehleringer, J. R., and C. Clark. 1988. Evolution and adaptation in Encelia (Asteraceae). In Plant evolutionary biology. Springer, Netherlands. pp. 221–248  Endler, J. A. 1991. Variation in the appearance of guppy colour patterns to guppies and their predators under different visual conditions. Vision Research 31: 587–608.  Endler, J. A. 1992. Signals, signal conditions, and the direction of evolution. American Naturalist 139: S125–S153.  Flamarique, I. N. 2005. Temporal shifts in visual pigment absorbance in the retina of Pacific salmon. Journal of Comparative Physiology A 191: 37 – 49.  Flamarique, I. N. 2013. Opsin switch reveals function of the ultraviolet cone in fish foraging. Proceedings of the Royal Society B: Biological Sciences 280: 20122490.  Flamarique, I. N., C. L. Cheng, C. Bergstrom, and T. E. Reimchen. 2013. Pronounced heritable variation and limited phenotypic plasticity in visual pigments and opsin expression of threespine stickleback photoreceptors. Journal of Experimental Biology 216: 656–667.  Foster, S. A., V. B. Garcia, and M. Y. Town. 1988. Cannibalism as the cause of an ontogenetic shift in habitat use by fry of the threespine stickleback. Oecologia 74: .577–585.  Fournier-Level, A., A. Korte, M. D. Cooper, M. Nordborg, J. Schmitt, and A. M. Wilczek. 2011. A map of local adaptation in Arabidopsis thaliana. Science 334: 86–89.   Fraley, C., A. Raftery, and L. Scrucca. 2012. Package ‘mclust’. Normal mixture modeling for model-based clustering, classification, and density estimation.    90 Fuller, R. C., K. L. Carleton, J. M. Fadool, T. C. Spady, and J. Travis. 2004. Population variation in opsin expression in the bluefin killifish, Lucania goodei: a real-time PCR study. Journal of Comparative Physiology 190: 147–154.  Fuller, R. C., K. L. Carleton, J. M. Fadool, T. C. Spady, and J. Travis. 2005. Genetic and environmental variation in the visual properties of bluefin killifish, Lucania goodei. Journal of Evolutionary Biology18: 516–523.  Fuller, R. C., and K. M. Claricoates. 2011. Rapid light‐induced shifts in opsin expression: finding new opsins, discerning mechanisms of change, and implications for visual sensitivity. Molecular Ecology 20: 3321–3335.  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 199: 535–544.   Gompert, Z., A. A. Comeault, T. E. Farkas, T. L. Parchman, C. A. Buerkle, and P. Nosil. 2014. Experimental evidence for ecological selection on genome variation in the wild. Ecology Letters 17: 369–379.   Govardovskii, V. I., N. Fyhrquist, T. O. M. Reuter, D.G. Kuzmin, and K. Donner. 2000. In search of the visual pigment template. Visual Neuroscience 17: 509–528.  Gow, J. L., S. M. Rogers, M. Jackson, and D. Schluter. 2008. Ecological predictions lead to the discovery of a benthic-limnetic sympatric species pair of threespine stickleback in Little Quarry Lake, British Columbia. Canadian Journal of Zoology 86: 564–571.  Grant, B. R. 1985. Selection on bill characters in a population of Darwin’s finches: Geospiza conirostris on Isla Genovesa, Galapagos. Evolution 39:523–532.   Greenwood, A. K., A. R. Wark, K. Yoshida, and C. L. Peichel. 2013. Genetic and neural modularity underlie the evolution of schooling behavior in threespine stickleback. Current Biology 23: 1884–1888.   Gould, F., 1979. Rapid host range evolution in a population of the phytophagous mite Tetranychus urticae Koch. Evolution 33: 791–802.  Hadfield, J. 2013. Masterbayes: ML and MCMC methods for pedigree reconstruction and analysis. CRAN: Comprehensive R Archive Network. Available at: http://www.r-project.org.  Hagen D. W. 1967. Isolating mechanisms in threespine sticklebacks (Gasterosteus). Journal of the Fisheries Board of Canada 24: 1637–1692.  Hagen, D. W., and L. G. Gilbertson. 1973. Selective predation and the intensity of selection acting upon the lateral plates of threespine sticklebacks. Heredity 30: 273–287.    91  Hancock, A.M., B. Brachi, N. Faure, M. W. Horton, L. B. Jarymowycz, F. G. Sperone, C. Toomajian, F. Roux, J. Bergelson. 2011. Adaptation to climate across the Arabidopsis thaliana genome. Science. 334: 83–86.   Hansen, T. F., W. S. Armbruster, and l. Antonsen. 2000. Comparative analysis of character displacement and spatial adaptations as illustrated by the evolution of Dalechampia blossoms. American Naturalist 156: S17–S34.  Hárosi F. I. 1994. An analysis of two spectral properties of vertebrate visual pigments. Vision Research 34: 1359–1367.  Hendry, A. P., and M. T. Kinnison. 1999. Perspective: the pace of modern life: measuring rates of contemporary microevolution. Evolution 53: 1637–1653.  Hofmann, C. M., K. E. O’Quin, A. R. Smith, and K. L. Carleton. 2010. Plasticity of opsin gene expression in cichlids from Lake Malawi. Molecular Ecology 19: 2064–2074.  Hohenlohe, P. A., S. Bassham, M. C. Currey, and W. A. Cresko. 2011. Extensive linkage disequilibrium and parallel adaptive divergence across threespine stickleback genome. Philosophical Transactions of the Royal Society B: Biological Sciences 5: 395–408.   Hoogland, R., D. Morri, and N. Tinbergen. 1956. The spines of sticklebacks (Gasterosteus and Pygosteus) as means of defence against predators (Perca and Esox). Behaviour 10: 205–236.  Hunt, D. M., K. S. Dulai, J. C. Partridge, P. Cottrill, and J. K. Bowmaker. 2001. The molecular basis for spectral tuning of rod visual pigments in deep-sea fish. Journal of Experimental Biology 204: 3333-3344.  Isayama, T., and C.L. Makino. 2012. Pigment mixtures and other determinants of spectral sensitivity of vertebrate retinal photoreceptors. In Photoreceptors: physiology, types and abnormalities (eds Akutagawa E, and Ozaki K). Nova Science Publishers, New York. pp 1 – 31.  Jones, F. C., Y. F. Chan, J. Schmutz, J., Grimwood, S. D. Brady, A. M. Southwick, D. M. Absher, R. M. Myers, T. E. Reimchen, B. E. Deagle, and D. Schluter. 2012. A genome-wide SNP genotyping array reveals patterns of global and repeated species-pair divergence in sticklebacks. Current Biology 22: 83–90.  Jones, F. C., M. G. Garbherr, Y. F. Chan, P. Russel, E. Mauceli, J. Johnson, R. Swofford, et al. 2012. The genomic basis of adaptive evolution in threespine sticklebacks. Nature 484: 55–61.   Kaeuffer, R., C. L. Peichel, D. I. Bolnick, and A. P. Hendry. 2012. Parallel and nonparallel aspects of ecological, phenotypic, and genetic divergence across replicate population pairs of lake and stream stickleback. Evolution 66: 402–418.    92 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.  Kirk, J. T. O. 1977. Use of a quanta meter to measure attenuation and underwater reflectance of photosynthetically active radiation in some inland and coastal south-eastern Australian waters. Marine Freshwater Research 28: 9–21.  Kirk, J. T. O. 1994. Light and Photosynthesis in Aquatic Ecosystems. Cambridge University Press, Cambridge.  Kitano, J., D. I. Bolnick, D. A. Beauchamp, M. M. Mazur, S. Mori, T. Nakano, and C. L. Peichel. 2008. Reverse evolution of armour plates in threespine stickleback. Current Biology 18: 769–774.  Klepaker, T. 1993. Morphological changes in a marine population of threespined stickleback, Gasterosteus aculeatus, recently isolated in fresh water. Canadian Journal of Zoology 71: 1251–1258.  Korves, T., K. J. Schmid, A. L. Caicedo, C. Mays, J. R. Stinchcombe, M. D. Purugganan, and J. Schmitt. 2007. Fitness effects associated with the major flowering time gene FRIGIDA in Arabidopsis thaliana in the field. American Naturalist 169: 141–157.  Kristjansson, B. K. 2005. Rapid morphological changes in threespine stickleback, Gasterosteus aculeatus, in freshwater. Environmental Biology of Fishes 74: 357–363.  Kristjansson, B. K., S. Skulason, and D. L., Noakes. 2002. Rapid divergence in a recently isolated population of threespine stickleback (Gasterosteus aculeatus L.). Evolutionary Ecology Research 4: 659–672.  Lamichhaney, S., F. Han, J. Berglund, C. Wang, M. S. Almen, M. T. Webster, B. R. Grant, P. R. Grant and L. Anderson. 2016. A beak size locus in Darwin’s finches facilitated character displacement during a drought. Science 352: 470–474.  Lande, R. 1983. The response to selection on major and minor mutations affecting a metrical trait. Heredity 50: 47–65.  Lande, R., and S. J. Arnold. 1983. The measurement of selection on correlated characters. Evolution 37: 1210–1226.  Lee, C. E., and M. A. Bell. 1999. Causes and consequences of recent freshwater invasions by saltwater animals. Trends in Ecology and Evolution 14: 284–288.  Lees, D. R., and E. R. Creed. 1975. Industrial melanism in Biston betularia: the role of selective predation. The Journal of Animal Ecology 44: 67-83.   93  Leinonen, T., G. Herczeg, J. M. Cano, and J. Merila. 2011. Predation imposed selection on threespine stickleback (Gasterosteus aculeatus) morphology: a test of the refuge use hypothesis. Evolution 65: 2916–2926.  Li, H., and R. Durbin. 2009. Fast and accurate short read alignment with Burrows–Wheeler transform. Bioinformatics 25: 1754–1760.  Linnen, C. R., E. P. Kingsley, J. D. Jensen, and H. E. Hoekstra. 2009. On the origin and spread of an adaptive allele in deer mice. Science 235: 1095–1098.  Liu, J., T. Shikano, T. Leinonen, J. M. Cano, M. H. Li, and J. Merilä. 2014. Identification of major and minor QTL for ecologically important morphological traits in three-spined sticklebacks (Gasterosteus aculeatus). G3: Genes Genomes Genetics 4: 595–604.  Loew, E. R., and J. N. Lythgoe. 1978. The ecology of cone pigments in teleost fishes. Vision Research 18: 715–722.  Losos, J.B. 1990. A phylogenetic analysis of character displacement in Caribbean Anolis lizards. Evolution 44: 558–569.  Lunter, G., and M. Goodson, M. 2011. STAMPY: a statistical algorithm for sensitive and fast mapping of Illumina sequence reads. Genome research 21: 936–939.  Lythgoe J. N. 1979. Ecology of Vision. Oxford University Press, Oxford.  Lythgoe, J. N. 1988. Light and vision in the aquatic environment. In Sensory biology of aquatic animals (eds Atema J, R. R. Fay, A. N. Popper, W. N. Tavolga). Springer, New York. pp. 57–82.  Marchinko, K. B. 2009. Predation's role in repeated phenotypic and genetic divergence of armour in threespine stickleback. Evolution 63: 127–138.  Marchinko, K. B., and D. Schluter. 2007. Parallel evolution by correlated response: lateral plate reduction in threespine stickleback. Evolution 61: 1084–1090.  McFarland, W. N., and F. W. Munz. 1975. The evolution of photopic visual pigments in fishes. Vision Research 15: 1071–1080.  McKenna, A., M. Hanna, E. Banks, A. Sivachenko, K. Cibulskis, A. Kernytsky, K. Garimella, D. Altshuler, S. Gabriel, M. Daly, and M. A. DePristo. 2010. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Research 20: 1297–1303.  McKinnon, J.S., and H. D. Rundle. 2002. Speciation in nature: the threespine stickleback model systems. Trends in Ecology and Evolution 17: 480–488.   94  McPhail, J. D. 1969. Predation and the evolution of a stickleback (Gasterosteus). Journal of the Fisheries Board of Canada 26: 3183–3208.  McPhail J. D. 1992. Ecology and evolution of sympatric sticklebacks (Gasterosteus): evidence for a species-pair in Paxton Lake, Texada Island, British Columbia. Canadian Journal of Zoology 70: 361–369.  McPhail, J.D. 1993. Ecology and evolution of sympatric sticklebacks (Gasterosteus): origin of the species pairs. Canadian Journal of Zoology 71: 515–523.  Metzker, M. L. 2010. Sequencing technologies—the next generation. Nature reviews genetics, 11: 31–46.  Meyer, J. R., and R. Kassen. 2007. The effects of competition and predation on diversification in a model adaptive radiation. Nature 446: 432–435.  Miller, C. T., S. Beleza, A. A. Pollen, D. Schluter, R. A. Kittles, M. D. Shriver, and D. M. Kingsley. 2007. Cis-Regulatory changes in Kit ligand expression and parallel evolution of pigmentation in sticklebacks and humans. Cell 131: 1179–1189.  Miller, C. T., A. M. Glazer, B. R. Summers, B. K. Blackman, A. R. Norman, M. D. Shapiro, B. L. Cole, C. L. Peichel, D. Schluter, and D. M. Kingsley. 2014. Additive, anatomically regional, and clustered quantitative trait loci control skeletal evolution in sticklebacks. Genetics 197: 405–420.  Mills, M. G., A. K. Greenwood, and C. L. Peichel. 2014. Pleiotropic effects of a single gene on skeletal development and sensory system patterning in sticklebacks. EvoDevo 5: 5.  Minder, A. M., and A. Widmer. 2008. A population genomic analysis of species boundaries: neutral processes, adaptive divergence and introgression between two hybridizing plant species. Molecular Ecology 17: 1552–1563.  Mitchell-Olds, T., and R. G. Shaw. 1987. Regression analysis of natural selection: statistical inference and biological interpretation. Evolution 41: 1149–1161.  Moodie, G. E. E., J. D. McPhail, and D. W. Hagen. 1973. Experimental demonstration of selective predation on Gasterosteus aculeatus. Behaviour 47: 95–15.  Munz, F. W. 1958. The photosensitive retinal pigments of fishes from relatively turbid coastal waters. Journal of General Physiology 42: 445–459.  Munz, F. W., and W. N. McFarland. 1977. Evolutionary adaptations of fishes to the photic environment. In: The Visual System in Vertebrates. (ed Crescitelli F). Springer, New York. pp. 194 – 274.    95  Myhre, F., and T. Klepaker. 2009. Body armour and lateral-plate reduction in freshwater three-spined stickleback, Gasterosteus aculeatus: adaptations to a different buoyancy regime? Journal of Fish Biology 75: 2062–2074.  Nadeau, N. J., S. H. Martin, K. M. Kozak, C. Salazar, K. K. Dasmahapatra, J. W. Davey, S. W. Baxter, M. L. Blaxter, J. Mallet, and C. D. Jiggins. 2013. Genome‐wide patterns of divergence and gene flow across a butterfly radiation. Molecular Ecology 22: 814–826.  Nagy, E. S. 1997. Selection for native characters in hybrids between two locally adapted plant subspecies. Evolution 51: 1469–1480.  Ooijen, J. W., and R. E. Voorrips. 2002. JoinMap: Version 3.0: Software for the Calculation of Genetic Linkage Maps, Plant Research International. Wageningen, The Netherlands.  Orti, G., M. A. Bell, T. E. Reimchen, and A. Meyer. 1994. Global survey of mitochondrial DNA sequences in the threespine stickleback: evidence for recent migrations. Evolution 48: 608–622.  Orr, H. A. 2005. The genetic theory of adaptation: a brief history. Nature Reviews Genetics 6: 119–127.  Østman, B., A. Hintze, and C. Adami. 2011. Impact of epistasis and pleiotropy on evolutionary adaptation. Proceedings of the Royal Society of London B: Biological Sciences doi:10.1098/rspb.2011.0870  Otto, S.P. 2004. Two steps forward, one step back: the pleiotropic effects of favoured alleles. Proceedings of the Royal Society of London B: Biological Sciences 271:705–714.  O'Quin, K. E., C. M. Hofmann, H. A. Hofmann, and. K. L. Carleton. 2010. Parallel evolution of opsin gene expression in African cichlid fishes. Molecular Biology and Evolution 27: 2839–2854.  Peichel, C. L., K.S. Nereng, K. A. Ohgi, B. L. 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.  Pespeni, M. H., E. Sanford, B. Gaylord, T. M. Hill, J. D. Hosfelt, H. K. Jaris, M. LaVigne, et al. 2013. Evolutionary change during experimental ocean acidification. Proceedings of the National Academy of Sciences of the USA 110: 6937–6942.  Pinheiro, J., D. Bates, S. DebRoy, D. Sarkar, and R Core Team. 2015. nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.1-119, http://CRAN.R-project.org/package=nlme.    96 Price, T., and T. Langen. 1992. Evolution of correlated characters. Trends in Ecology and Evolution 7: 307–310.  Popescu, C. 1979. Natural selection in the industrial melanic psocid Mesopsocus unipunctatus (Mull) (Insecta: Psocoptera) in northern England. Heredity 42: 133–142.  Poulton, E. B. 1898. Natural Selection the Cause of Mimetic Resemblance and Common Warning Colours. Journal of the Linnean Society of London, Zoology 26: 558–612.  R Development Core Team. 2012-2015. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna Austria.  Raeymaekers, J. A., J. K. Van Houdt, M. H. Larmuseau, S. Geldof, and F. A. Volckaert. 2007. Divergent selection as revealed by PST and QTL-based FST in three-spined stickleback (Gasterosteus aculeatus) populations along a coastal-inland gradient. Molecular Ecology 16: 891–905.  Regan, C. T. 1909. The species of Three-spined sticklebacks (Gastrosteus). Journal of Natural History 4: 435–437.  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. 1991. Evolutionary attributes of headfirst prey manipulation and swallowing in piscivores. Canadian Journal of Zoology 69: 2912–2916.  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.  Remichen, T. E. 1994. Predators and morphological evolution in threespine stickleback. In The Evolutionary Biology of the Threespine Stickleback (Bell M. A., Foster S. A. eds) Oxford University Press, Oxford, pp 240–276  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.  Reist, J. D. 1980. Predation upon pelvic phenotypes of brook stickleback, Culaea inconstans, by selected invertebrates. Canadian Journal of Zoology 58: 1253–1258.  Renaut, S., N. Maillet, E. Normandeau, C. Sauvage, N. Derome, S. M. Rogers, and L. Bernatchez 2012. Genome-wide patterns of divergence during speciation: the lake whitefish case study. Philosophical Transactions of the Royal Society of London B: Biological Sciences 367: 354–363.    97 Rennison D. J., G. L. Owens, and J. S. Taylor. 2012. Opsin gene duplication and divergence in ray-finned fish. Molecular Phylogenetics and Evolution 62: 986–1008.  Rennison, D. J., K. Heilbron, R. D. H. Barrett, and D. Schluter. 2014. Data from: Discriminating selection on lateral plate phenotype and its underlying gene, Ectodysplasin, in threespine stickleback. American Naturalist. Dryad  Digital Repository, http://dx.doi.org /10.5061/dryad.dg82p.  Reznick, D. N., F. H. Shaw, H. F. Rodd, and R. G. Shaw. 1997. Evaluation of the rate of evolution in natural populations of guppies (Poecilia reticulata). Science 275: 1934–1937.  Rick, I. P., and T. C. Bakker. 2008. Colour signalling in conspicuous red sticklebacks: do ultraviolet signals surpass others? BMC Evolution Biology 8:189.  Rick, I. P., D. Bloemker, and T. Bakker. 2012 Spectral composition and visual foraging in the three‐spined stickleback (Gasterosteidae: Gasterosteus aculeatus L.): elucidating the role of ultraviolet wavelengths. Biological Journal of the Linnean Society 105: 359 – 368.  Rick, I. P., R. Modarressie, and T. C. Bakker. 2004. Male three-spined sticklebacks reflect in ultraviolet light. Behaviour 141: 1531 – 1541.  Roesti, M., A. P. Hendry, W. Salzburger, and D. Berner. 2012. Genome divergence during evolutionary diversification as revealed in replicate lake–stream stickleback population pairs. Molecular Ecology 21: 2852–2862.  Rokyta, D. R., P. Joyce, S. B. Caudle, and H. A. Wichman. 2005. An empirical test of the mutational landscape model of adaptation using a single-stranded DNA virus. Nature genetics 37: 441–444.  Rowland, W. J. 1994. Proximate determinants of stickleback behaviour: an evolutionary perspective. In The Evolutionary Biology of the Threespine Stickleback (eds Bell MA, Foster SA). Oxford University Press, Oxford. pp. 297–344.  Rundle, H.D., L. Nagel, J. W. Boughman, and D. Schluter. 2000. Natural selection and parallel speciation in sympatric sticklebacks. Science 287: 306–308.  Rundle, H. D., S. M. Vamosi., and D. Schluter. 2003. Experimental test of predation's effect on divergent selection during character displacement in sticklebacks. Proceedings of the National Academy of Sciences 100: 14943–14948.  Ryan, M. J. 1990. Sexual selection, sensory systems and sensory exploitation. Oxford Surveys in Evolutionary Biology 7: 156–195  Sadier, A., L. Viriot, S. Pantalacci, and V. Laudet. 2014. The Ectodysplasin pathway: from diseases to adaptations. Trends in Genetics 30: 24–31.   98  Schluter, D., 1995. Adaptive radiation in sticklebacks: trade-offs in feeding performance and growth. Ecology, 76: 82–90.  Schluter, D. 2001. Ecology and the origin of species. Trends in Ecology and Evolution 16: 372 – 380.  Schluter, D., and J. D. McPhail J.D. 1992. Ecological character displacement and speciation in sticklebacks. American Naturalist 140: 85–108.  Schluter, D., and L. M. Nagel. 1995. Parallel speciation by natural selection. American Naturalist 146: 292–301.  Schluter, D., T. D. Price, and P. R. Grant. 1985. Ecological character displacement in Darwin’s finches. Science 227: 1056–1059.  Schluter, D., and J. Smith. 1986. Natural selection on beak and body size in the song sparrow. Evolution 40: 221–231.  Seehausen, O., Y. Terai, I. S. Magalhaes, K. L. Carleton, H. D. Mrosso, R. Miyagi, I. van der Sluijs, M. V. Schneider, M. E. Maan, H. Tachida, H. Imai, and N. Okada. 2008. Speciation through sensory drive in cichlid fish. Nature 455: 620–626.  Shapiro, M. D. M. E. Marks. C. L. Peichel, B. K. Blackman, K. S. Nereng, B. Jónsson, D. Schluter, and D. M. Kingsley. 2004. Genetic and developmental basis of evolutionary pelvic reduction in threespine sticklebacks. Nature 428: 717–723.  Shichida, Y., and H. Imai. 1998. Visual pigment: G-protein-coupled receptor for light signals. Cellular and Molecular Life Sciences 54: 1299–1315.  Simpson, G. G. 1953. The major features of evolution. Columbia University Press, New York.  Stapley, J., J. Reger, P. G. Feulner, C. Smadja, J. Galindo, R. Ekblom, C. Bennison, A. D. Ball, A. P. Beckerman, and J. Slate, J., 2010. Adaptation genomics: the next generation. Trends in Ecology and Evolution 25: 705–712.  Summers, B. R. 2008. Molecular Genetics of Dorsal Spine Reduction in Threespine Sticklebacks (Gasterosteus Aculeatus). PhD Thesis, Stanford University.  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.    99 Taylor, E. B., J. W. Boughman, M. Groenenboom, M. Sniatynski, D. Schluter, and J. L. Gow. 2006. Speciation in reverse: morphological and genetic evidence of the collapse of a three‐spined stickleback (Gasterosteus aculeatus) species pair. Molecular Ecology 15: 343–355.  Therkildsen, N. O., J. Hemmer-Hansen, T. D. Als, D. P. Swain, M. J. Morgan, E. A. Trippel, S. R. Palumbi, D. Meldrup, and E. E. Nielsen. 2013. Microevolution in time and space: SNP analysis of historical DNA reveals dynamic signatures of selection in Atlantic cod. Molecular Ecology 22: 2424-2440.  Thurman, T. J., and R. D. H. Barrett. 2016. The genetic consequences of selection in natural populations. Molecular Ecology 25: 1429–1448.  Toyama, M., M. Hironaka, Y. Yamahama, H. Horiguchi, O. Tsukada, N. Uto, Y. Ueno, F. Tokunaga, K. Seno, and T. Hariyama. 2008. Presence of Rhodopsin and Porphyropsin in the Eyes of 164 Fishes, Representing Marine, Diadromous, Coastal and Freshwater Species—A Qualitative and Comparative Study. Photochemical and Photobiology 84: 996–1002.  Travisano, M., and R. G. Shaw. 2013. Lost in the map. Evolution 67: 305–314.  Turner, T. L., M. W. Hahn, and S. V. Nuzhdin. 2005. Genomic islands of speciation in Anopheles gambiae. PLoS Biology 3: e285.  Tutt J. W. 1890. Melanism and melanochroism in British Lepidoptera. S. Sonnenschein and Company. United Kingdom.   Vamosi S. M. 2002. Predation sharpens the adaptive peaks: survival trade-offs in sympatric sticklebacks. Annales Zoologici Fennici 39: 237–248.  Vamosi, S. M., and D. Schluter. 2004. Character shifts in the defensive armour of sympatric sticklebacks. Evolution 58: 376–385.  Via, S. 1991. The genetic structure of host plant adaptation in a spatial patchwork: demographic variability among reciprocally transplanted pea aphid clones. Evolution, 45: 827–852.  Wark, A. R., M. G. Mills, L.-H. Dang, Y. F. Chan, F. C. Jones, S. D. Brady, D. M. Absher, et al. 2012. Genetic architecture of variation in the lateral line sensory system of threespine sticklebacks. Genes, Genomes, Genetics 2: 1047–1056.  Weldon, W. F. R. 1901. A first study of natural selection in Clausilia kaminata (Montagu). Biometrika 1: 109–124  Yokoyama, S. 1995. Amino acid replacements and wavelength absorption of visual pigments in vertebrates. Molecular Biology and Evolution 12: 53–61.    100 Yokoyama, S., and R. B. Radlwimmer, N. S. Blow. 2000. Ultraviolet pigments in birds evolved from violet pigments by a single amino acid change. Proceedings of the National Academy of Science of the USA 97: 7366–7371.  Zeileeis, A., and G. Grothendieck. 2005. Zoo: S3 Infrastructure for regular and irregular time series. Journal of Statistical Software 14: 1–27.        101 Appendices  Appendix A   - Supplementary Materials for Chapter 2   A.1 Relationship Between Length and Number of Lateral Plates.   Fish reach their adult number of lateral plates at ~34 mm in length, indicated by the broken line. Solid lines indicate the fitted breakpoint regression, Black: CC genotype fish; light grey: CL genotype fish; dark grey: LL genotype fish.   15 20 25 30 35 40 4551015202530Length (mm)Number of Lateral Plates  102 A.2 Change in Size-Corrected Lateral Plate Number for all Genotypes Across the Sampling Period.   The horizontal black line indicates the sample median and the boxes denote the interquartile range.  A.3 Standardized Partial Selection Coefficients for the Lateral Plate Phenotype, Eda Genotype, and Dominance for the September to October and October to November periods.   September-October October-November  𝜷 𝜷 Lateral plates  0.35 (-0.08, 0.78) -0.16 (-0.83, 0.40) Additive genotype -0.04 (-0.43, 0.38) -0.04 (-0.65, 0.69) Dominance -0.47 (-0.83,  -0.14)  0.26 (-0.14, 0.71) 95% confidence intervals are reported in brackets. 51015202530MonthSize Adjusted Lateral PlatesSeptember October November  103 A.4 Variance in Lateral Plate Number for each Eda Genotype by Month.  September October November CC 22.1 25.8 24.3 CL 19.7 19.1 18.7 LL 9.5 8.4 8.7  A.5 Variance – Covariance Matrix for the Standardized Traits Included in the One-Genotype Variable Lande-Arnold Analysis for the September – October Period.   Lateral Plates Additive Genotype Lateral plates 0.70 0.17 Genotype 0.17 0.41  A.6 Variance – Covariance Matrix for the Standardized Traits Included in the One-Genotype Variable Lande-Arnold Analysis for the October – November Period.  Lateral Plates Additive Genotype Lateral plates 0.63 0.23 Genotype 0.23 0.40  A.7 Variance – Covariance Matrix for the Standardized Traits Included in the Two-Genotype Variable Lande-Arnold Analysis for the September – October Period.  Lateral Plates Additive Genotype Dominance Lateral plates 1.04 0.68 0.06 Additive genotype 0.68 0.10 -0.16 Dominance 0.06 -0.16 1.01    104 A.8 Variance – Covariance Matrix for the Standardized Traits Included in the Two-Genotype Variable Lande-Arnold Analysis for the October – November Period.   Lateral Plates Additive Genotype Dominance Lateral plates 1.05 0.81 0.01 Additive genotype 0.81 1.07 -0.14 Dominance 0.01 -0.14 0.94           105 Appendix B  - Supplementary Materials and Methods for Chapter 3 B.1 Collection and Site Information  Fish were collected from eleven populations from eight different lakes, rivers or lagoons (from now on referred to as location): two marine, three solitary and three species-pair locations (see Appendix B.10 for details). The following collection permits were used: Species at Risk Act permit number SARA 236 and British Columbia fish collection permit number NA-SU12-76311. In order to measure opsin gene expression, six gravid females were collected from each of the populations. All fish from a given population were taken at the same time and the collections were taken between 10 am and 12 pm during the period of May 16th to May 30th 2012. Fish were euthanized at the site using buffered Tricaine methanesulfonate (MS-222). Both eyes were removed and immediately stored in 1 ml RNAlater® (Qiagen, Netherlands) and moved to a -200 C freezer for up to a month until RNA was extracted. Irradiance was measured in July 2012.  Three families of Priest Benthic and three families of Oyster Marine fish were generated by in vitro fertilization in May and June 2012 respectively. These fish were hatched and reared in freshwater tanks under fluorescent lights on a 14 and 10 hour light-dark cycle. Animals were treated in accordance with University of British Columbia Animal Care protocols (Animal Care Permit # A11-0402). Gravid females were sacrificed using MS-222 between June 5th and 7th 2013. One fish from each family was surveyed (three fish per population). Both eyes were immediately removed and put directly into RNAlater®. Samples were stored in RNAlater® for one week at -200C until RNA was extracted.    106 B.2 RT-qPCR Protocol Left and right eyes were pooled for each individual. The pooled eyes were homogenized in a Retsch mm 400 Mixer Mill (Haan, Germany) using a carbide bead. Total RNA was extracted using the AurumTM Total RNA Fatty and Fibrous Tissue (BioRad®), which included a DNase I incubation step. The concentration and purity of the extracted RNA was assessed on a NanoDrop® Spectrophotometer (Thermo Scientific). Synthesis of cDNA was accomplished using the iScriptTM cDNA Synthesis Kit (Bio-Rad®); 1000 ng of RNA was used as the input for the cDNA synthesis of each sample. The resulting cDNA was diluted 1:100 in ultra-pure water for RT-qPCR analysis.   RT-qPCR primers and probes were designed using sequences from the stickleback genome (See Appendix B.11 for primer and probe sequences). Primer sequences were targeted to regions that were divergent between the five opsin gene subfamilies. Despite the fact the stickleback have two RH2 genes the primers were designed to pick up only one of the duplicates because there is no evidence to suggest they would have different absorption phenotypes and there is some evidence to suggest that the non-targeted duplicate may be a pseudogene (Rennison et al., 2012). For each gene one of the primers and/or the RT-qPCR probe spanned an intron, this was done to avoid amplification of genomic DNA. We used the PrimeTime® qPCR 5’ Nuclease Assays from Integrated DNA Technologies® (Iowa, USA) for each of the targeted genes. The assays used had a double-quenched probe with 5’ 6-FAMTM dye, internal ZENTM and 3’ Iowa Black® FQ Quencher.   Quantification of gene transcript copy number was done using RT-qPCR analysis on a BioRad®IQ5 machine (BioRad, California USA). The polymerase used was the SsoFast probes supermix (BioRad®) in a 25 µl reaction. Reactions were run in 96-well plates (Fisher,   107 Massachusetts USA), which were sealed using optical sealing tape (BioRad®). Well-factors were collected from each of the experimental plates. Reactions were run in duplicate or triplicate. No-reverse transcription and no template controls were included and for every run and did not amplify. RT-qPCR conditions consisted of 1 cycle at 95 °C (3 minutes); 40 cycles of 95 °C (10 seconds) followed by 60 °C (30 seconds). We used a standardized luminance threshold value of 50 to calculate Ct values. Equation 1 was used to calculate the PCR efficiencies (E) for each of the primer pairs, ,        (1) where the slope is determined from a linear least squares regression fit to critical threshold (Ct) data from a cDNA dilution series (1:10, 1:50, 1:100, 1:500, 1:1000).   We calculated opsin expression relative to the beta actin reference gene, however for the purposes of this study we were more interested in the expression of each opsin gene relative to the total opsin levels present in the retina, rather than absolute levels of expression, so we used the proportion of total opsin expression for a given gene. The estimate of the initial amount of gene transcript (Ti) was calculated for each individual (i) using equation 2, where E is the PCR efficiency for a given gene calculated from equation 1 and Ct is the critical threshold for fluorescence.  𝑇! = ! (!!!)!!           (2) Then for each individual we summed the opsin gene expression across the four opsin genes and calculated the proportion of total expression that each gene exhibited.    € E = e−slope −1  108 Amplicons from the RT-qPCR for each gene (primer pair) were sequenced from one individual and are reported in Appendix B.11. Sanger sequencing of the amplicons was done at the NAPS Sequencing Centre at the University of British Columbia.   B.3 Deriving Spectral Sensitivity  We used the absorbance templates for A1 chromophore (unless otherwise stated) for each of the four cone opsins from Govardovskii et al. (2000) and the wavelengths of maximum absorbance (𝜆!"#) for each opsin from Flamarique et al. (2013). The spectral sensitivity of an individual (𝑖) at a given wavelength (𝜆) for a particular opsin (𝑜) is the multiplication of its absorbance (𝐴!(𝜆)) and relative expression (𝐸!,!). The overall sensitivity of an individual is the sum over of all opsins and is defined across the visible wavelength range (350 to 700 nm) by 𝑆! 𝜆 = 𝐴!!!!!! 𝜆  𝐸!,!.  B.4 Plasticity in the Laboratory Environment  To assess the effect of plasticity on opsin gene expression, we looked at the difference between wild and lab-reared fish derived from one marine and one freshwater location (Figure 3.4). While much of the differentiation in gene expression between marine and freshwater individuals was maintained in the lab, some plasticity was still seen; the level of SWS1 (UV) expression was reduced in both types of lab reared fish relative to their wild counterparts (marine difference = 0.16 ± 0.04 SE, p=0.003, F1,7=19.9; freshwater difference = 0.06 ±0.02 SE, p=0.007, F1,7= 14.3) (Figure 3.4). There was also a significant increase in SWS2 (blue) expression for the lab-reared marine fish compared to wild individuals, although the effect size was small (difference =0.013 ±0.004 SE, p=0.009, F1,7=12.9) (Figure 3.4). However, this was not seen for   109 the freshwater population (p=0.2) (Figure 3.4). There was not a significant difference in the LWS or RH2 expression of wild and lab reared fish (p>0.09) (Figure 3.4). These results indicate that there is a contribution of plasticity in stickleback opsin gene expression. Raising the fish under artificial (fluorescent) lighting that lacked UV wavelengths likely contributed to the reduction in SWS1 expression that we saw in lab-reared fish. This same pattern has been previously described in cichlids, where lab-reared individuals raised under artificial lighting had reduced SWS1 expression compared to wild caught fish (Hofmann et al., 2010). It is also important to note that some of the differences we see between the two population in the lab may arise due to maternal effects or early developmental effects when the eggs were exposed to natural conditions while developing within the wild caught mothers.    B.5 Association Between Differences in Spectral Sensitivity and Ambient Light  We use two functions of wavelength 𝜆  to characterize ambient light: the irradiance 𝐼! 𝜆  and the transmission 𝐾! 𝜆 , with the values of 𝜆 between 350 and 700 nm. Recall from the main text that the irradiance is taken to be the irradiance measured at depth 50m. To construct 𝐾! 𝜆  we use transmission coefficients, as defined by the Beer-Lambert law, which gives transmission 𝑇! at depth (𝑑) and wavelength (𝜆) as  𝑇!,! 𝜆 = 𝑏(𝜆)𝑒!!!(!)! . For each site and each value of 𝜆, we estimated the unknown parameters 𝑏(𝜆) and 𝐾! 𝜆  using the nls function in R. We then smoothed the resulting 𝐾! 𝜆  values using a rolling mean approach (as implemented in the R zoo library, Zeileis and Grothendieck, 2005, with window width 10 nm). For each site, these smoothed 𝐾! 𝜆  values were then normalized to sum to 1, as we want to compare the difference in relative absorbance between different locations. Hereafter,   110 𝐾! 𝜆  refers to the smoothed and normalized values of the site-specific transmission coefficients. For each location, we then constructed the ‘representative’ transmission coefficient curve, 𝐾!(𝜆), by calculating at each value of 𝜆 the median of the 𝐾!(𝜆)’s from all sites within that location. To quantify ambient light differences between freshwater and marine locations, we chose a reference marine location (A), and refer to its curve of transmission coefficients as 𝐾!,!(𝜆). We then calculated the difference between the transmission coefficients for each freshwater location (B) that we wanted to test and the reference marine location (A) as Δ𝐾! 𝜆 = −1( 𝐾!,! 𝜆 − 𝐾!,! 𝜆 ). We multiplied the difference by -1 to facilitate the comparison between Δ𝑆!,, Δ𝐾! and ΔI! (see Figure 3.6). Note that, in our definition, Δ𝐾! is a measure of light propagation (instead of rate of absorbance). A positive value of Δ𝐾! 𝜆  indicates more transmission of light (i.e. fewer photons are lost as light travels through water) at wavelength 𝜆 at the freshwater location B than at the reference marine location. For this analysis we used Oyster Lagoon as our marine reference location (A). We repeated this procedure (without multiplying by -1) to calculate the difference between environments in irradiance (ΔI!). Again for irradiance a positive value of ΔI! 𝜆  indicates that there are more photons present at wavelength 𝜆 at the freshwater location (B) relative to the marine reference environment (A).  To quantify differences in spectral sensitivity between each location 𝐵 and the reference marine location 𝐴, at each wavelength 𝜆 we first calculated 𝑆! 𝜆  as the median of the spectral sensitivities of all measured individuals from the reference location 𝐴. For each individual i in location 𝐵, we calculated the difference between that individual’s spectral sensitivity at each wavelength 𝑆!,! 𝜆  and the marine location 𝐴 spectral sensitivity: Δ𝑆!,! 𝜆 =  𝑆!,! 𝜆 −𝑆! 𝜆 .    111 We then proceeded with studying association between differences in spectral sensitivity and ambient light. To do this we used each fish’s spectral sensitivity and its light environment (transmission and irradiance) at each wavelength, all measured relative to the reference marine location 𝐴. A scatterplot of spectral sensitivity difference against difference of light environment at all wavelengths allows us to visually assess the association between the two variables. To proceed with a statistical analysis, for each fish, we summarized and quantified this strength of association via the correlation coefficient (r). If the correlation coefficient calculated for fish i is positive, then, for that fish, wavelengths showing elevated sensitivity (positive Δ𝑆!,!  𝜆 ’s) are associated with increased light propagation (higher transmission) in our transmission calculations and, in our irradiance calculations, are associated with more photons (higher irradiance). We calculated this association summary for every fish in location B. We tested whether the mean association (that is, the mean of the correlations in the population) in location B was zero using a one-sample t-test. We did this for all locations. Our results are contained in Appendix B.12 and reported in the main text. We also tested simultaneously the equality of the means of the individual level measures of association of all freshwater locations by fitting a mixed-effects model to the measure of association, with location as a random effect.  To study differences in environment (pelagic versus littoral) we carried out separate analyses for each of the two species pair lakes with data on light environment (Paxton and Priest). For each lake, we took as reference the limnetic population, and thus the pelagic environment. Our calculations were the same as in the marine/freshwater comparison above, with the littoral environment (with benthic fish) being substituted for ‘location B’ in our calculations. Thus, in each lake, for each benthic fish, we considered the relationship between differences in its spectral sensitivity (relative to median limnetic sensitivity) and differences in   112 light environment (relative to that lake’s median – representative pelagic environment). We summarized that relationship for each benthic fish by the correlation and then tested whether these correlations were expected to equal to 0. The results did not differ greatly if the benthic population was used as the reference. See Appendices B.8 and B.13 for complete results, including those discussed in Appendix B.6.  B.6 Effect of Changing Chromophore or Reference Population in the Analyses of Differences in Sensitivity and Differences in Light Environment  Recall that, in our initial analyses (reported in the main text), spectral sensitivity was modeled using exclusively the A1 chromophore. Our reference population for the marine-freshwater comparison was Oyster Lagoon and our reference population for the species pair analysis was the limnetic population. We studied the robustness of our results to using different chromophores and different reference populations. To study the importance of the reference location in the marine/freshwater comparison, we first made a direct comparison of Oyster Lagoon and the other marine location, Little Campbell River, using Oyster Lagoon as reference. That is, Oyster Bay served as “A” and Little Campbell River served as “B” in the analysis in Appendix B.5. We found that there was no significant association between differences in sensitivity and differences in transmission (data not shown), but there was a significant association for the irradiance. In other words, changes in sensitivity in Little Campbell relative to the reference Oyster Lagoon population did not significantly covary with changes in transmission but did significantly covary with changes in irradiance. Thus, we might infer that Little Campbell River and Oyster Lagoon are equivalent reference populations for transmission analysis, but perhaps not for irradiance analysis.   113 In order to further study whether reference population affected the results we re-did the analysis, described in the main text and in Appendix B.5, using Little Campbell River instead of Oyster Lagoon as the marine reference (Appendix B.7). The results are given in Appendix B.12. Overall the results with the two base lines agree; the mean correlations and significance levels are very similar for transmission. When the A1 chromophore is used for estimation of sensitivity the correlation for irradiance is also similar, however when other chromophores are used the correlation becomes significantly negative. We believe that Little Campbell River is a much less reliable marine reference population because the measurements were taken in the tidal (marine) part of the river where turbidity increases significantly when the tide comes in. The light measurements were taken with incoming tide and hence may give a biased view of the light environment the stickleback experience most of the time, and this may explain the odd result for irradiance.  To determine the importance of the reference population in the analyses for the two species pair locations (Priest and Paxton), we repeated the analysis using the benthic ecotype as reference instead of the limnetic (Appendix B.8). For all chromophore combinations the results are very similar to those obtained using the limnetics as a reference for both transmission (Appendix B.13).    114 B.7 Quantification of Correlation Between Differences in Spectral Sensitivity and Differences in Local Light Transmission (A) and Irradiance (B) for Marine and Freshwater Populations using Little Campbell Reference Population.    Circles indicate individuals’ correlations. All populations are presented relative to the marine reference location, Little Campbell River.   ABCorrelation CoefficientPriest     Paxton     Trout       KirkCorrelation CoefficientPriest     Paxton     Trout       Kirk0.00.20.40.60.8β●●●●●●●●●●●●●●●●●●●●●●●●●●●●Priest Paxton Trout Kirk0.00.20.40.60.8β●●●●●●●●●●●●●●●●●●●●●●●●●●●Priest Paxton Trout Kirk  115 B.8 Quantification of Correlation Between Differences in Spectral Sensitivity and Differences in Local Light Transmission (A) and Irradiance (B) for Benthic and Limnetic Populations using Benthic Reference.   Circles indicate individuals’ correlation, using the benthics as the reference population.  B.9 Effect of Changing Chromophore in the Analysis of the Correlation Between Spectral Sensitivity and Ambient Light (Spectral Matching)  To study the effect of chromophore on our spectral matching results, we repeated the analysis reported in the main text with various different chromophore combinations in the freshwater population. The combinations and the results are presented in Appendix B.13. Switching the ratio of chromophore used did little to affect the magnitude of the correlation between spectral sensitivity and Transmission. However the magnitude of the correlation strengthened for irradiance when there was a 50:50 mix of A1 and A2 used.   −0.50.00.5β●●●●●●●●●●●Priest Paxton−0.4−0.20.00.20.40.60.8β●●●●●●●●●●●Priest PaxtonCorrelation CoefficientPriest          Paxton     Correlation CoefficientPriest          Paxton     A B  116 B.10 Stickleback Populations used, their Locations, and Sample Sizes (number of fish) for Opsin Expression and Environmental Light Condition.  Name Latitude Longitude Type  Sample Size Irradiance (# of sites)      ≤6m >6m Oyster Bay 49.61210 -124.03186 Marine 6 10 0 Little Campbell River 49.01543 -122.77662 Marine 6 10 0 Trout Lake 49.50820 -123.87641 Solitary 6 10 5 Cranby Lake 49.69537 -124.50812 Solitary 6 failed Kirk Lake 49.73897 -124.58680 Solitary 6 7 5 Paxton Lake 49.70789 -124.52492 Species-pair Benthic 6 10 5 Limnetic 6 Priest Lake 49.74517 -124.56519 Species-pair Benthic 6 10 5 Limnetic 5 Little Quarry Lake  49.66266 -124.10888 Species-pair Benthic 6 failed Limnetic 6           117 B.11 Primer, Probe and Amplicon Sequences from RT-qPCR Assays Gene Probe Sequence 5’-3’ Primer Sequences 5’-3’ Amplicon Size (bp) RT-qPCR Amplicon Sequence SWS1 CCGTAGCAGGACTGGTGACAGC Forward: ACATCACCTTGGCAGGATTC Reverse: GTGGGCTGGAACAACAGATT 279 GGTGTTTGTCGCATCTGCGAGGGGTTACTACTTCCTGGGTTACACCTTGTGCGCGCTGGAGGCTGCGATGGGATCCGTAGCAGGACTGGTGACAGCCTGGTCTTTGGCTGTTTTGTCTTTCGAGAGATATCTGATCATCTGTAAACCTTTTGGAGCCTTTAAGTTTACCAGTAACCACGCTCTCGGTGCTGTCGCCTTCACCTGGTTTATGGGAATCTGTTGTTCCAGCCCA SWS2A GAAAATGGCGGCAAAGGCC Forward: TCTGCACAATTTGCTTCTGC Reverse: GGTTGTAAACTGCGGAGGAC 261 GGCGGCAAGGCCCAAGCAGAATCCGCCTCGACCCAGAAGGCGGAGCGGGAGGTGACCAGGATGGTGGTTCTCATGGTGATGGGCTTCCTGGTGTGCTGGATGCCGTACGCCTCATTCGCTCTTTGGGTGGTCAACAACCGCGGGCAGACTTTTGACCTGAGGTTTGCTTCTATTCCGTCCGTCTTTTCCAAGTCCTCCGCAGTTTACAAC RH2 TTGGCTGGTCCAGGTACCTTCC Forward: GGGATTCATGGCCACATTAG Reverse: 174 CTGGATCCTTTTCCCTGGACCATGGCTATGGCATGTGCTGCTCCCCCTCTTTTTGGTGGCCAGGTACCTTCCT  118 Gene Probe Sequence 5’-3’ Primer Sequences 5’-3’ Amplicon Size (bp) RT-qPCR Amplicon Sequence TAGTCAGGTCCACACGAGCA GAGGGCATGCAGTGCTCGTGTGGACCTGA  LWS TGGATGGAGCAGGTACTGGCC Forward: GATATGGTCTGCCGTCTGGT Reverse: GCCACAATCATGACAACGAC 297 TGGAAGTGAAGACCCTGGAGTCCAGTCCTACATGATTGTTCTCATGATCACATGCTGTCTCATTCCTCTGGCCATCATCATATTGTGCTACCTTGCAGTCTGGTTGGCTATCCGTGCTGTGGCCATGCAGCAGAAGGAATCAGAGTCAACTCAAAAAGCTGAAAGAGACGTATCCAGAATGGTCGTTGTCATGATTGTGGC Beta Actin CTGTGCTACGTCGCCCTGGA Forward: GGCTACTCCTTCACCACCAC Reverse: CAGGACTCCATACCGAGGAA 329 CACAGCTGAGAGGGAAATCGTGCGTGACATCAAGGAGAAGCTGTGCTACGTCGCCCTGGACTTCGAGCAGGAGATGGGTACCGCTGCCTCCTCCTCCTCCCTGGAGAAGAGCTACGAGCTGCCCGACGGACAGGTCATCACCATCGGCAATGAGAGGTTCCGTTGCCCAGAGGCCCTCTTCCAGCCTTCCTTCCTCGGTACGTTTCCCTACTCGAGCCTAACAGTCTCATAATGTAAATATGTTGCTCCCTTGGTTACTCTGCACCGCC  119 Gene Probe Sequence 5’-3’ Primer Sequences 5’-3’ Amplicon Size (bp) RT-qPCR Amplicon Sequence ACATGCTTACAAGTGTCATCTCCCCTCAG  B.12 Mean Correlation Between the Change in Spectral Sensitivity and the Shift in Ambient Light from Marine to Freshwater under Various Chromophore Scenarios.  Reference Population Ambient Light Measure Chromophore Mean Corr. SE t Raw  p-value Adjusted p-value Marine Fresh  O.L. Transmission A1 A1 0.39 0.12 3.30 0.002 0.004 L.C.R. Transmission A1 A1 0.46 0.10 4.76 < 0.0001 < 0.0001 O.L. Transmission A1 A2 0.14 0.09 1.53 0.136 0.148 L.C.R. Transmission A1 A2 0.26 0.07 3.91 0.001 0.001 O.L. Transmission A1 50:50 A1/A2 0.22 0.12 1.85 0.074 0.088 L.C.R Transmission A1 50:50 A1/A2 0.55 0.09 6.19 < 0.0001 < 0.0001 O.L. Irradiance A1 A1 0.32 0.07 4.94 < 0.0001 < 0.0001 L.C.R. Irradiance A1 A1 0.6 0.05 13.3 < 0.0001 < 0.0001 O.L. Irradiance A1 A2 -0.32 0.12 -2.74 0.010 0.015 L.C.R. Irradiance A1 A2 -0.48 0.05 -9.29 < 0.0001 < 0.0001 O.L. Irradiance A1 50:50 A1/A2 -0.23 0.10 -2.15 0.039 0.052 L.C.R Irradiance A1 50:50 A1/A2 -0.13 0.09 -1.41 0.168 0.168  O.L. (Oyster Lagoon) and L.C.R (Little Campbell River) are alternative reference marine populations. Means that are significantly different from zero are in bold.     120 B.13 Mean Correlation Between the Change in Spectral Sensitivity and Shift in Ambient Light, from Limnetic to Benthic Environment (Limnetic Reference), and from Benthic to Limnetic Environment (Benthic reference).  Reference Group Ambient Light Measure Chromophore Mean Corr. SE t Raw p-value Adjusted p-value Limnetic Transmission A1 0.27 0.13 1.97 0.077 0.185 Benthic Transmission A1 0.31 0.16 1.91 0.088 0.185 Limnetic Transmission A2 0.30 0.15 1.99 0.075 0.185 Benthic Transmission A2 0.34 0.18 1.88 0.093 0.185 Limnetic Transmission 50:50 A1/A2 0.28 0.15 1.92 0.084 0.185 Benthic Transmission 50:50 A1/A2 0.33 0.17 1.88 0.094 0.185 Limnetic Irradiance A1 0.18 0.18 1.00 0.339 0.581 Benthic Irradiance A1 0.27 0.30 0.87 0.402 0.603 Limnetic Irradiance A2 0.04 0.14 0.29 0.775 0.775 Benthic Irradiance A2 0.08 0.24 0.35 0.735 0.775 Limnetic Irradiance 50:50 A1/A2 0.11 0.17 0.63 0.539 0.696 Benthic Irradiance 50:50 A1/A2 0.18 0.31 0.57 0.580 0.696          121 B.14 Mean Correlation Between Spectral Sensitivity and Local Light Environment (measured as transmission and irradiance) under Various Chromophore Scenarios.  Population Type Water  Type Ambient Light Measure Chromophore Mean Corr. SE p-value Adjusted p-value Fresh Fresh Transmission A1 0.07 0.03 0.0184 0.029 Fresh Fresh Transmission 50:50 A1/A2 0.04 0.04 0.2983 0.341 Fresh Fresh Transmission A2 -0.04 0.04 0.3859 0.386 Fresh Fresh Irradiance A1 0.12 0.02 < 0.0001 0.0002 Fresh Fresh Irradiance 50:50 A1/A2 0.15 0.02 < 0.0001 0.0002 Fresh Fresh Irradiance A2 0.09 0.03 < 0.0001 0.0002 Marine Marine Transmission A1 -0.66 0.11 0.0017 0.003 Marine Marine Irradiance A1 -0.16 0.05 0.0230 0.031  Values that are significantly different from zero are in bold.   122 B.15 Estimated Mean Spectral Sensitvity of Benthic and Limnetic Populations.  The center lines are the fitted values of spectral sensitivity from the mixed effects model. The shaded regions are one standard error above and below the fitted values, with standard error also derived from the mixed-effects model.          0.00.10.20.30.40.50.60.7Sensitivity350 400 450 500 550 600 650 700Wavelength (nm)limneticbenthic  123 Appendix C  - Supplementary Materials for Chapter 4   C.1 Genetic Map Estimated from Four F1 Families.           806040200Chromosome (Chr)Location (cM)1a 2 4 6a 7 8b 10 11b 13a 13c 15a 15c 17a 18 19b 20 Un1b 3 5 6b 8a 9 11a 12 13b 14 15b 16 17b 19a 19c 21  124 C.2 Quantitative Trait Locus Map for First Dorsal Spine and Pelvic Spine Length across all F1 Families.   Panel A is the QTL map for first dorsal spine length. Panel B is the QTL map for pelvic spine length.  0102030405060ChromosomeLOD Score1a 2 4 6a 7 8b 10 11b 13a13c15a 15c 17a 18 19b 20 Un1b 3 5 6b 8a 9 11a 12 13b 14 15b 16 17b 19a19c 2101234ChromosomeLOD Score1a 2 4 6a 7 8b 10 11b13a 13c 15a15c 17a 18 19b 20 Un1b 3 5 6b 8a 9 11a 12 13b 14 15b 16 17b 19a 19c 21AB  125 C.3 Quantitative Trait Locus Map for First Dorsal Spine Length for each F1 Family.  Each panel represents the QTL for one F1 family, with 99-100 fish used per family.    0.00.51.01.52.02.5ChromosomeLOD Score1a 2 4 6a 7 8b 10 11b 13a 13c 15a15c 17a 18 19b 20 Un1b 3 5 6b 8a 9 11a 12 13b 14 15b 16 17b 19a 19c 2101234ChromosomeLOD Score1a 2 4 6a 7 8b 10 11b 13a13c 15a 15c 17a 18 19b 20 Un1b 3 5 6b 8a 9 11a 1213b 14 15b 16 17b 19a19c 2101234ChromosomeLOD Score1a 2 4 6a 7 8b 10 11b 13a13c15a15c 17a 18 19b 20 Un1b 3 5 6b 8a 9 11a 12 13b 14 15b 16 17b 19a19c 210123ChromosomeLOD Score1a 2 4 6a 7 8b 10 11b 13a13c 15a15c 17a 18 19b 20 Un1b 3 5 6b 8a 9 11a 12 13b 14 15b 16 17b 19a 19c 21Family 1 Family 2Family 3 Family 4  126 C.4 Percent Variance Explained and Significance of Treatment Effect for First Dorsal Spine, Pelvic Spine and Pelvic Girdle Length in each F1 Family Family Trait Chromosome      PVE Treatment  P-value   1 First Dorsal 11b 12.0    0.248 1 First Dorsal 15c  17.0 < 0.001 2 First Dorsal 7 10.0 < 0.001 2 First Dorsal 15c 11.0     0.005 3 First Dorsal 9 16.0     0.288 4 First Dorsal 4 16.0     0.009 1 Pelvic Spine 7 63.0       -  2 Pelvic Spine 7 56.0       - 3 Pelvic Spine 7 51.0       - 4 Pelvic Spine 7 57.0       - 1 Pelvic Girdle 7 65.0       - 2 Pelvic Girdle 7 58.0       - 3 Pelvic Girdle 7 48.0       - 4 Pelvic Girdle 7 63.0       -           127 C.5 Variance – Covariance Matrix for the Size Adjusted Armour Phenotypes in September 2012.   First  Dorsal Spine Second  Dorsal  Spine Pelvic Spine Pelvic Girdle Lateral Plates First Dorsal Spine 0.78 0.05 -0.13 -0.18 0.16 Second Dorsal Spine 0.05 0.10 0.02 -0.00          0.04 Pelvic Spine -0.13 0.02 2.95 2.74 0.60 Pelvic Girdle -0.18 -0.00 2.74 4.54 0.91 Lateral Plates 0.16 0.04 0.60 0.91 1.78  C.6 Variance – Covariance Matrix for the Size Adjusted Armour Phenotypes in January 2013.   First  Dorsal Spine Second  Dorsal  Spine Pelvic Spine Pelvic Girdle Lateral Plates First Dorsal Spine 0.66 0.05 0.01 -0.08 0.06 Second Dorsal Spine 0.04 0.09 0.02 -0.01          0.03 Pelvic Spine 0.01 0.02 1.60 2.16 0.20 Pelvic Girdle -0.08 -0.01 2.16 3.83 0.22 Lateral Plates 0.06 0.03 0.20 0.22 1.10    128 C.7 Variance – Covariance Matrix for the Size Adjusted Armour Phenotypes April 2013.  First  Dorsal Spine Second  Dorsal  Spine Pelvic Spine Pelvic Girdle Lateral Plates First Dorsal Spine 0.67 0.03 0.03 0.07 0.1 Second Dorsal Spine   0.03 0.09 0.02 0.00          0.04 Pelvic Spine 0.03 0.02 1.85 2.68 0.67 Pelvic Girdle 0.08 0.00 2.68 4.90 1.10 Lateral Plates 0.09 0.04 0.67 1.10 2.02  C.8 Variance – Covariance Matrix for the Size Adjusted Armour Phenotypes in September 2013.  First  Dorsal Spine Second  Dorsal  Spine Pelvic Spine Pelvic Girdle Lateral Plates First Dorsal Spine 0.99 0.11 0.06 0.17 0.23 Second Dorsal Spine 0.11 0.14 0.05 0.04          0.11 Pelvic Spine 0.07 0.05 1.88 2.51 0.67 Pelvic Girdle 0.17 0.04 2.51 4.41 1.05 Lateral Plates 0.23 0.11 0.67 1.05 2.13    129  C.9 Mean Evolutionary Response of Armour Traits for Treatment and Control Ponds.   Mean evolutionary response (Haldanes) of armour traits for treatment and control ponds.  Treatment ponds are indicated in blue and control ponds in orange. Black circles are the mean of all control ponds and black triangles are the mean of treatment ponds, both are depicted with one standard error above and below the mean.        −101First Dorsal     SpineSecond Dorsal        SpinePelvicSpineLateral  PlatesHaldanes TreatmentControlPredation  130 C.10 Mean Evolutionary Response of Treatment and Control Ponds in Haldanes    Treatment     Control First Dorsal Spine  0.20 ± 0.22 0.42   -0.39 ± 0.11 0.03 Second Dorsal Spine 0.41 ± 0.40   0.37   -0.31 ± 0.29 0.35 Pelvic Spine 0.08 ± 0.18   0.69   -0.10 ± 0.16 0.54 Lateral Plates 0.13 ± 0.17  0.48   -0.05 ± 0.13 0.70  Haldane estimates significantly different from zero are in bold. Mean values are reported with standard error, P-values are given to the right of each estimate.    131 C.11 Armour Trait Trajectories through Time for Control and Treatment Ponds.  Color indicates founding family and is shared between paired treatment and control ponds panels.    0.01.02.03.0First Dorsal0.01.02.001234Second Dorsal1.02.03.001234Pelvic Spine01234Sept2012Jan2013April2013Sept20130246Lateral Plates0246Treatment Ponds Control PondsSecond DorsalPelvic SpineLateral PlatesFirst DorsalSept2012Jan2013April2013Sept2013Sept2012Jan2013April2013Sept2013Sept2012Jan2013April2013Sept2013Sept2012Jan2013April2013Sept2013Sept2012Jan2013April2013Sept2013Sept2012Jan2013April2013Sept2013Sept2012Jan2013April2013Sept2013  132 C.12 Mean Treatment Effect and Standard Error for Pond Pairs in the Fall (September – January) and Winter (January – April) Seasons. Season   Δs’  P-value  Fall First Dorsal Spine -0.05 ± 0.09 0.64  Second Dorsal Spine -0.48  ± 0.15 0.05  Pelvic Spine -0.04 ± 0.05 0.49  Lateral Plates -0.01  ± 0.17 0.98 Winter First Dorsal Spine 0.08 ± 0.10 0.51  Second Dorsal Spine 0.32 ± 0.24 0.27  Pelvic Spine -0.17 ± 0.25 0.55  Lateral Plates 0.01 ± 0.16 0.96  Selection estimates are reported with standard error, p-values are given to the right of estimates. Significantly different from zero are in bold.  C.13 Coefficients of Change in Trait Trajectory through Time for Control and Treatment Ponds.   Treatment    Control  First Dorsal Spine  0.14 ± 0.02   < 0.001 -0.02 ± 0.02 0.76 Second Dorsal Spine 0.14 ± 0.01   < 0.001 0.06 ± 0.01 < 0.001 Pelvic Spine 0.14 ± 0.04  0.001 -0.05 ± 0.04 0.25 Lateral Plates -0.02 ± 0.04  0.01 -0.09 ± 0.04 < 0.001  Coefficients are reported with standard error and p-values are reported to the right. Coefficients significantly different from zero are in bold.   133 C.14 Difference in Evolutionary Response for SNPs within the Msx2 and Pitx1Ggene Regions.  Divergent evolutionary response estimated from SNP allele frequencies for first dorsal spine and pelvic spine length QTL after one generation of selection. Estimates were measured from September 2012 in the first generation to September 2013 in the second generation. Positive values indicate selection for limnetic alleles (associated with longer spines) in the treatment pond relative to its paired control pond. Each colored circle represents an individual SNP within 20 cM of a LOD peaks for first dorsal spine or pelvic spine near the Msx2 and Pitx1 gene regions. Black circles are the mean effect of the SNPs within the region, with one standard error above and below the mean. Each pond pair is indicated by a unique color. -0.50.00.5Msx2 Pitx1Difference in Evolutionary Response (Δh)

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0307148/manifest

Comment

Related Items