Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Understanding the coexistence of sperm-dependent asexual species and their sexual hosts : the role of… Mee, Jonathan Alan 2011

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

Item Metadata

Download

Media
24-ubc_2012_spring_mee_jonathan.pdf [ 7.52MB ]
Metadata
JSON: 24-1.0072467.json
JSON-LD: 24-1.0072467-ld.json
RDF/XML (Pretty): 24-1.0072467-rdf.xml
RDF/JSON: 24-1.0072467-rdf.json
Turtle: 24-1.0072467-turtle.txt
N-Triples: 24-1.0072467-rdf-ntriples.txt
Original Record: 24-1.0072467-source.json
Full Text
24-1.0072467-fulltext.txt
Citation
24-1.0072467.ris

Full Text

Understanding the coexistence of sperm-dependent asexual species and their sexual hosts: the role of biogeography, mate choice, and relative fitness in the Phoxinus eos-neogaeus (Pisces: Cyprinidae) system by Jonathan Alan Mee B.Sc.F., The University of British Columbia, 2002 M.Sc., The University of Toronto, 2005  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in The Faculty of Graduate Studies (Zoology) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) December 2011  © Jonathan Alan Mee, 2011  !"#$%&'$( In sperm-dependent asexual reproduction, sperm is not required for its genetic contribution, but it is required for stimulating zygote development. In my dissertation, I address several questions related to the coexistence of sperm-dependent asexuals and the sexually-reproducing species on which they depend. I have focused my research on a sperm-dependent asexual fish, Phoxinus eos-neogaeus, that originated via hybridization between P. eos and P. neogaeus. Using a mathematical model of mate choice among sexuals and sperm-dependent asexuals, I showed that stable coexistence can occur when there is variation among males in the strength of preference for mating with sexual females and when males with stronger preference pay a higher cost of preference. My model also predicts that coexistence is facilitated when the asexuals suffer a fitness disadvantage relative to the sexuals. Subsequent empirical work, in which I compared the repeat swimming performance, fecundity, and growth rate of asexual and sexual Phoxinus, provided results that are consistent with this prediction: the asexuals are, at best, as fit as the sexuals. I sampled Phoxinus populations from across the species’ North American distribution and the pattern of mitochondrial DNA variation across these populations suggests that all P. eos-neogaeus have originated from hybridization events that took place in a Mississippi River glacial refugium. Also, cytoplasmic hybrids (which are P. eos with P. neogaeus mitochondrial DNA) appear to have replaced ‘pure’ P. eos in all northern populations, which may reduce the disparity in fitness between P. eos-neogaeus and their sperm donors if P. neogaeus mitochondria are adapted to northern environments. The pattern of nuclear DNA variation across P. eos-neogaeus populations suggests that those using P. eos sperm are genetically distinct from those using P. neogaeus sperm. These genetically distinct populations match their host species in size, suggesting that sperm-dependent asexuals may be adapted to be proficient at soliciting sperm from particular sexual species. Persistence of sperm-dependent asexuals depends on stable coexistence with sexual species. My work highlights the importance of the relative fitness of asexual and sexual species, and of male mate choice in maintaining this stable coexistence. ii  )%*+&'*( A version of chapter 2 has been published: Mee, J.A., and Otto, S.P. (2010) Variation in the strength of male mate choice allows long-term coexistence of sperm-dependent asexuals and their sexual hosts. Evolution 64(10): 2808–2819. I did the majority of the model development and analysis, and wrote the original draft of the manuscript. Otto, S.P., contributed substantially to the model development and analysis, and she contributed revisions to the manuscript. A version of chapter 5 has been published: Mee, J.A., Brauner, C.J., and Taylor, E.B. (2011) Repeat swimming performance and its implications for inferring the relative fitness of asexual hybrid dace (Pisces: Phoxinus) and their sexually reproducing parental species. Physiological and Biochemical Zoology 84(3): 306-315. I performed all the experimental and laboratory work, I analyzed all the data, and I wrote the original draft of the manuscript. Brauner, C.J., and Taylor, E.B., contributed advice on the experimental methodology, and they contributed revisions to the manuscript. The UBC Animal Care Committee approved all research activities involving the use of animals (Animal Care Certificate Number 08-0645). The Canadian Department of Fisheries and Oceans approved the transfer of live fish to research facilities at the University of British Columbia [Introduction and Transfer Application Numbers 10838 (2006), 10889 & 11147 (2007), 11503 (2008), 12054 (2010)].  iii  ,&"-*(.+(/.0$*0$#( !"#$%&'$ 1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111122! )%*+&'* 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111222! ,&"-*(.+(/.0$*0$#1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111123! 42#$(.+(,&"-*# 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111132! 42#$(.+(5267%*#1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111322! !'80.9-*:6*;*0$#111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111112<! =*:2'&$2.0 1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 <222! >1(?0$%.:7'$2.0 1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111>! >1>1(@$7:A(@A#$*;B(!"#$%&'()*#(+&*#,-*'(. 1111111111111111111111111111111111111111111111111111111111111111111111111 C! >1D1(E*#*&%'F(G7*#$2.0#1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 H! D1(I&%2&$2.0(20($F*(#$%*06$F(.+(;&-*(;&$*('F.2'*(&--.9#(-.06J$*%;('.*<2#$*0'*(.+( #K*%;J:*K*0:*0$(&#*<7&-#(&0:($F*2%(#*<7&-(F.#$# 1111111111111111111111111111111111111111111111111111111L! D1>1(@7;;&%A 1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 L! D1D1(?0$%.:7'$2.0111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 L! D1C1(M.:*-(=*3*-.K;*0$ 111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 >N! D1O1(M.:*-(E*#7-$#(&0:(!0&-A#2#11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 >C! D1H1(=2#'7##2.0 1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 >L! C1(,F*('A"%2:(203&#2.0B(92:*#K%*&:(%*K-&'*;*0$(.+(!"#$%&'()*#((P)2#'*#B(/AK%202:&*Q( "A('A$.K-&#;2'(FA"%2:#11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111DR! C1>1(@7;;&%A 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 DR! C1D1(?0$%.:7'$2.01111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 CN! C1C1(M&$*%2&-#(&0:(M*$F.:# 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 CD! C1O1(E*#7-$# 111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 CC! C1H1(=2#'7##2.0 1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 CO! O1(S.#$J#2T*J;&$'F206(20(&(#K*%;J:*K*0:*0$(&#*<7&-(+2#F(#K*'2*# 111111111111111111111111111111CR! O1>1(@7;;&%A 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 CR! O1D1(?0$%.:7'$2.01111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 CR! O1C1(M&$*%2&-#(&0:(M*$F.:# 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 OD! O1O1(E*#7-$# 111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 OH! O1H1(=2#'7##2.0 1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 OU! H1(E*K*&$(#92;;206(K*%+.%;&0'*(&0:(2$#(2;K-2'&$2.0#(+.%(20+*%%206($F*(%*-&$23*(+2$0*##( .+(&#*<7&-(FA"%2:(:&'*(P)2#'*#B(!"#$%&'(Q(&0:($F*2%(#*<7&--A(%*K%.:7'206(K&%*0$&-( #K*'2*# 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111VH! H1>1(@7;;&%A 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 VH! H1D1(?0$%.:7'$2.01111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 VV! H1C1(M&$*%2&-#(&0:(M*$F.:# 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 VU! H1O1(E*#7-$# 111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 L>! iv  H1H1(=2#'7##2.0 1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 LC!  V1(5*'70:2$A(&0:(6%.9$F(.+(0.%$F*%0(%*:"*--A(:&'*(P!"#$%&'()*#(QW(+20*#'&-*(:&'*( P!"#$%&'()&*#,-*'(QW(&0:($F*2%(702#*<7&-(FA"%2:# 111111111111111111111111111111111111111111111111111111U>! V1>1(@7;;&%A 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 U>! V1D1(?0$%.:7'$2.01111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 U>! V1C1(M&$*%2&-#(&0:(M*$F.:# 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 UC! V1O1(E*#7-$# 111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 UV! V1H1(=2#'7##2.0 1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 UU! L1(X*0*%&-(=2#'7##2.0 1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111RR! L1>1(?;K-2'&$2.0# 1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 RR! L1D1(57$7%*(=2%*'$2.0# 111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 >NH! L1C1(520&-(,F.76F$#11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 >>N! Y2"-2.6%&KFA1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111>>>! !KK*0:2'*# 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111>DH! !KK*0:2<(!B(!0&-A#*#(@7KK-*;*0$&%A($.(/F&K$*%(D1111111111111111111111111111111111111111111111111111111111 >DH! !KK*0:2<(YB(@&;K-206(=*$&2-# 111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 >CO!  v  42#$(.+(,&"-*# Table 2.1. Frequencies of mating among sexual genotypes, E1 and E2 (with frequencies x1 and x2, respectively), and the gynogenetic genotype, G (with frequency x3) ............22 Table 2.2. Summary of simulation parameters and results ..............................................23 Table 4.1. Sampling locations and sample sizes for Phoxinus in British Columbia (BC) and Alberta (AB)............................................................................................................53 Table 4.2. Microsatellite DNA loci used to identify clonal diversity among hybrid Phoxinus sampled from British Columbia (BC) and Alberta (AB)...............................54 Table 4.3. Phoxinus body shape character loadings on all principal coordinate axes (PCO), and percent of overall shape variation explained by each axis .........................55 Table 5.1. Model summary and ANOVA (analysis of variance) results for linear mixed effects models of the effects of repeat swimming and temperature on swimming performance compared between populations of hybrid and parental Phoxinus ............77 Table 6.1. Crosses between P. eos, P. neogaeus, and hybrids from Fiesta Lake (FIE), Gunrange Lake (GUN), and Alford Lake (ALF)...........................................................91 Table 6.2. Identity of crosses combined for growth comparisons between Phoxinus species, and sample sizes for hatchlings in each combination.......................................92 Table 6.3. Proportion of viable eggs from P. eos, P. neogaeus, and hybrid females fertilized by either P. eos or P. neogaeus males ............................................................93 Table 6.4. Proportion of viable eggs from P. eos, P. neogaeus, and hybrid females fertilized by both P. eos and P. neogaeus males............................................................94 Table A1. Results of simulations showing the effects of sex ratio and variation in male mate preference on the coexistence of sperm-dependent asexuals and their hosts......131 Table A2. Frequencies of mating among sexual R. lessonae and the hybridogenic R. esculenta genotypes .....................................................................................................132 Table B1. Details of sampling locations and sample sizes for all Phoxinus samples used for this dissertation.......................................................................................................134  vi  42#$(.+(5267%*#( Figure 2.1. Summary of type and proportion of offspring produced from mating between sperm-dependent hybrids and sexually reproducing parental species ...........................24 Figure 2.2. Illustrations of the behavior in a system approaching stable equilibrium with coexistence of a gynogen, a sexual genotype with stronger male preference, and a sexual genotype with weaker male preference ..............................................................25 Figure 2.3. Parameter space allowing stable coexistence of gynogenetic and sexual individuals......................................................................................................................26 Figure 2.4. An example of the simulated evolution of mate preference..........................27 Figure 2.5. Replicate simulations of mate preference evolution with no mutation ..........28 Figure 3.1. Pure Phoxinus eos (red) have been replaced by cybrids (black) throughout much of their distribution in North America .................................................................37 Figure 3.2. Genealogy and distribution of haplotypes. ....................................................38 Figure 4.1. Sampling sites in British Columbia and Alberta ...........................................56 Figure 4.2. Plots of genetic distance (RST/(1-RST)) against geographic distance between between hybrid Phoxinus populations in Alberta and British Columbia.......................57 Figure 4.3. Plot depicting the scores for Phoxinus eos-neoageus, P. eos, and P. neogaeus on the first prinicipal component axis of shape variation in two different types of communities...................................................................................................................58 Figure 4.4. Plot depicting the standard length (mm) of Phoxinus eos-neoageus, P. eos, and P. neogaeus in two different types of communities................................................59 Figure 4.5. Hierarchical clustering of Phoxinus eos-neogaeus clones from Alberta based on average square distance (D1) ....................................................................................60 Figure 4.6. Hierarchical clustering of Phoxinus eos-neogaeus clones from BC based on average square distance (D1) .........................................................................................61 Figure 4.7. Hierarchical clustering of Phoxinus eos-neogaeus clones from BC and Alberta based on average square distance (D1).............................................................62 Figure 4.8. Plot depicting the scores for Phoxinus eos-neoageus, P. eos, and P. neogaeus on the first prinicipal coordinate axis of body shape variation in two different types of communities...................................................................................................................63  vii  Figure 4.9. Plot depicting the standard length (mm) of Phoxinus eos-neoageus, P. eos, and P. neogaeus in two different types of communities................................................64 Figure 5.1. Relationship between size (fork length, cm) and Umax (body lengths per second) for Phoxinus eos and coexisting hybrids (panel A) and for P. neogaeus and coexisting hybrids (panel B) ..........................................................................................78 Figure 5.2. Effects of repeat swimming and temperature on swimming performance (Umax) of hybrid and parental Phoxinus from four lakes ...............................................79 Figure 5.3. Swimming performance of hybrid and parental Phoxinus under the most challenging conditions (Umax2 at 25 °C) as a proportion of performance under the least challenging conditions (Umax1 at 16 °C).........................................................................80 Figure 6.1. Fecundity of P. eos, P. neogaeus, and sperm-dependent asexual hybrids in lakes where the hybrids use P. eos sperm (type 1) and in lakes where the hybrids use P. neogaeus sperm (type 2) ................................................................................................95 Figure 6.2. Weight per egg for P. eos, P. neogaeus, and sperm-dependent asexual hybrids in lakes where the hybrids use P. eos sperm (type 1) and in lakes where the hybrids use P. neogaeus sperm (type 2) ........................................................................96 Figure 6.3. Proprotion of fertilized eggs which were viable for P. eos, P. neogaeus, and sperm-dependent asexual hybrids in lakes where the hybrids use P. eos sperm (type 1 lakes) ..............................................................................................................................97 Figure 6.4. Growth of P. eos, P. neogaeus, (solid lines) and sperm-dependent asexual hybrid (dashed lines) hatchlings from parents taken from Fiesta Lake (type 1) and Alford Lake (type 2) ......................................................................................................98 Figure A1. Parameter space allowing stable coexistence of hybridogenic and sexual individuals....................................................................................................................133  viii  !'80.9-*:6*;*0$#( PhD dissertations are like children. It is often said that it takes a whole community to raise a child. Similarly, a large number of people are required for one person to complete a PhD dissertation. And while children are sometime irrational, messy, and frustrating, the process of completing a PhD dissertation is sometimes laborious, unpredictable, and humbling. Thankfully, both children and PhD dissertations are wonderfully rewarding. The most rewarding part of doing a PhD is the wealth of personal and professional connections and relationships that arise during the process. I am very grateful for all the relationships I have built during the completion of my PhD. The first person I must thank specifically is Rick Taylor. I got in touch with Rick in the fall of 2005 regarding the possibility of doing a PhD under his supervision. My idea was to continue to work on the ecology and evolution of asexual Phoxinus hybrids (I had done a project on Phoxinus for my Masters thesis), and to maybe answer a question or two about the evolution of sex. Rick took me on, and I never doubted his enthusiasm for the work I was doing. While others called my project “career suicide” (probably because I was working largely on natural history questions in a rarely-studied organism), Rick called it “fascinating”. Perhaps both descriptions are apt. Rick was also generous in suggesting that I take over the teaching of his Biogeography course (BIOL 413) – I’ve really cherished that opportunity. I’ve had a lot of fun being in Rick’s lab. I may never be able to thank my supervisory committee enough for the their help and advice. Martin Adamson was generous with his advice on working with and studying parasites. In the end, no parasitology work made it into my dissertation (probably because of my own failings), but Martin helped me figure out that hybrid and parental Phoxinus share the same species of Gyrodactylus parasite. Colin Brauner’s contribution to my dissertation was immense. All of chapter 5 was possible only because of his logistical help and academic acumen. I am also extremely grateful for Colin’s general support and guidance throughout my time as a graduate student at UBC, both in an official capacity as graduate student advisor and in an unofficial capacity as mentor and colleague. Darren Irwin was exacting in his constructive criticism of my PhD proposal. I owe Darren a lot of credit for any degree of rigour that exists in the ix  connections between my empirical work and general questions it was intended to address. Sally Otto played a huge role in the biggest intellectual leap I made during my PhD. My theoretical work in chapter 2 required the ascent of a very steep math learning curve. The ascent would not have been possible (or it would have taken much longer and been much more treacherous) if not for Sally’s guidance. Also, it would not be an exaggeration to suggest that any success I have in the academic job market in my early career will be due to the publication of chapter 2 in the journal Evolution. I am grateful to Sally for instigating this achievement by recognizing the novelty and broader implications of this work. There are plenty of people to thank for practical and logistical help. Rod Burns is the manager of the Raven Brood Trout Station near the town of Caroline, Alberta. I did most of my PhD fieldwork based out of a little house (a shack, really) on the station’s property that Rod let me work and live in for free. Rod and several other employees of Alberta Sustainable Resources Development, including Steve Herman, were very supportive of my work in the province. Justin Hanisch, a PhD student in Bill Tonn’s lab at the University of Alberta, also did his fieldwork on the lakes around Caroline. Justin and Bill helped me find and gain access to several lakes. In addition, Justin and his field assistants, especially Fred Noddin, were welcome company during the otherwise lonely weeks in the field. For my fieldwork based out of the Harkness Laboratory for Fisheries Research in Algonquin Park, Ontario, I am grateful for the help of Brad Steinberg, Glen Forward, Greg Betteridge, Gary Ridout, Mark Ridgway, Peggy Daraugh, Doug Brown, and Trevor Middel. My fieldwork in New Brunswick depended heavily on help from Rick Cunjak and his student Heather McCracken at the University of New Brunswick. Rosemary Curley and Rosanne MacFarlane at the PEI Department of Environment, Energy and Forestry provided samples from Prince Edward Island. Sandy Chalanchuk at the DFO provided samples from the Experimental Lakes Area in northern Ontario. I also benefited from the help of several field assistants including Charles Shulman, Caleb Sylvester, Les Harris, and Joey Courchesne. My molecular biology lab work would not have been possible had it not been preceded by the work of Bernard Angers and his students at the University of Montreal. Bernard and his students, especially Rachel Massicotte, have been very helpful in x  providing advice and discussion on practical issues and academic topics. Merci Bernard et Rachel! My first year or so in the lab was somewhat painful in its lack of productivity, and I credit my lab mates Jen Gow and Katriina Ilves (and of course the PCR gods) for guiding me into the land of productive and consistently successful lab work. Many of my lab mates contributed in a tangible way to the completion of my PhD dissertation by feeding my experimental fish when I was away in the field or elsewhere. For filling in for me in the thankless task of feeding my fish (they literally can’t say “thank you”), I thank Patricia Woodruff, Damon Nowosad, Sarah Northrup, Chad Ormond, and Monica Yau. I also relied heavily on Patrick Tamkee for fish husbandry and rearing advice. I was also fortunate to have the help of two fantastic undergraduate students, Carita Chan and Joanna Li Yung Lung, with both lab work and fish husbandry. There are some very clear goals in graduate school: to complete a dissertation and, if one’s ambition is to get a job in academia, to publish papers. It can seem very worklike. It’s easy to forget that the main purpose of going to graduate school and being a graduate student is to learn. I am thankful for the bounty of opportunities in the Biodiversity Research Group and the Department of Zoology at UBC to attend seminars (particularly EEB, BRS, and BLISS) and talk with visiting speakers, to participate in discussion groups (particularly EDG and DeltaTEA), and to learn widely and deeply. I am grateful to all those who participated and enriched the learning experience (particularly faculty members), but I’m especially grateful to those who helped organize all the seminars and discussion groups. I am extremely grateful for financial support from several institutions and sources. I was awarded an NSERC PGD-D scholarship for the first two years of my PhD. Between 2006 and 2010, I was awarded a series of UBC scholarships: four years of PhD Tuition Fee Awards, a Graduate Entrance Scholarship, a UBC Graduate Fellowship, a Four Year Fellowship for PhD students, the Pacific Century Graduate Scholarship, and the MacLean Fraser Memorial Scholarship. I may have been able to complete my PhD dissertation without the friendship and support of many fellow graduate students, but I doubt it. Some people experience very dark times over the course of their PhD. People often joke about the “post-comps blues”, but for many students the depth of these “blues” can be severe. At UBC, graduate xi  students in the Department of Zoology are held to very high standards of achievement. These high standards of achievement are a very good thing, but they must go hand-inhand with high standards of compassion and community support. I am extremely thankful that social traditions in the department allow for such high standards of compassion and community support. Traditions like Friday Beers and the annual Christmas skit play an enormously important role in complimenting the rigours of academic life at UBC. I am eternally grateful to all who have participated in upholding these traditions by doing duty as Beer Barons or simply acting a part in the skit or showing up to Friday Beers. Several fellow graduate students have played a particularly important role in my graduate student experience. All my office mates over the years, including Aleeza Gerstein, J.-S. Moore, Crispin Jordan, Patricia Woodruff, Gerrit Velema, Damon Nowosad, and Chad Ormond have been very supportive. I have benefited greatly from discussions with and guidance from my current office mates in the Beaty Biodiversity Centre, Aleeza Gerstein and J.-S. Moore. My discussions with Rich Fitzjohn and Leithen M’Gonigle, whose office doors down the hall were always open to me, were instrumental when I was learning biomathematical techniques. It seems odd, at the age of 32, to still be thanking one’s parents. So great, however, were my parents’ contributions to my wellbeing at every stage of my life to date, that it would be unconscionable to omit them from these acknowledgements. Mom and dad, you’re the best! Finally, I would like to thank my wife, Christie, for everything.  xii  =*:2'&$2.0( For Christie  xiii  >1(?0$%.:7'$2.0( Asexually reproducing species have a clear advantage over sexually reproducing species, all else being equal: asexuality confers the ability to reproduce without spending resources on costly male function. The cost of producing males, which is known commonly as the cost of sex, is assumed to give asexuals a two-fold advantage in population growth rate over sexuals. The cost of sex is well established theoretically (e.g., Felsenstein 1974; Barash 1976; Charlesworth 1980; Lively 1996; Hastings 1999), and has been demonstrated empirically (e.g., Jokela et al. 1997). Among evolutionary biologists (e.g., Darwin 1862; Felsenstein 1974; Williams 1975; Maynard Smith 1978; Hamilton 1980; Bell 1982; Kondrashov 1988; Otto & Barton 2001; Agrawal 2006), much effort has been directed towards reconciling the expected outcome of this clear advantage to asexual reproduction (i.e., that asexual reproduction should be abundant and common) with the empirical reality (i.e., that sexual reproduction is abundant and common). This paradoxical success of sexual reproduction is most apparent in vertebrates – there are probably fewer than 100 asexually reproducing vertebrate species (Dawley & Bogart 1989; Avise 2008). Most asexual vertebrate species are sperm-dependent, requiring heterospecific sperm to, at very least, stimulate egg development (as in, for example, gynogenesis). Sperm-dependence is common among asexual vertebrates, and it exists among asexual taxa in several phyla, including rotifers, platyhelminths, nematodes, annelids, mollusks, and arthropods (Beukeboom & Vrijenhoek 1998). Hence, spermdependent asexual reproduction is an important aspect of the diversity of asexual species, and understanding the evolution and persistence of sperm-dependent asexuals is an important aspect of understanding the evolution and maintenance of sex. Among asexual vertebrates, a variety of different reproductive modes exists, lying along a continuum from fully asexual to nearly sexual reproduction (Hellriegel & Reyer 2000). In hybridogenesis, the diploid asexual premeiotically excludes one genome, passes the other genome on to gametes without recombination, and then reconstitutes a “hemiclonal” diploid genome by mating with a sexual species (Dawley & Bogart 1989). Bisexual hybridogenic frogs (Rana esculenta), with both male and female hybridogens, fall closest to the sexual end of the continuum (Hellriegel & Reyer 2000). Certain 1  hybridogenic fish (e.g., Poeciliopsis spp.), which also produce hemiclonal offspring, fall farther from the sexual end of the continuum because they avoid the production of males (Vrijenhoek et al. 1977). Further still along the continuum are gynogenetic hybrid salamanders and fishes (e.g., Ambystoma spp., Phoxinus spp., Poecilia spp.), which, like the unisexual hybridogens, are all female and require sperm from a parental species for reproduction. Unlike the hybridogens, however, gynogens produce fully clonal diploid offspring and require sperm only to stimulate egg development (Goddard et al. 1989; Goddard et al. 1998). Certain hybrid lizards (e.g., Lacerta unisexualis) are parthenogenic and, therefore, represent the extreme asexual end of the continuum, requiring no paternal input for reproduction (Darevsky et al. 1989; Echelle et al. 1989), although reproduction in some parthenogenic lizards is facilitated by behavioral cues resembling mating (Crews et al. 1986). All known asexual vertebrates have originated from hybridization (Dawley & Bogart 1989; Avise 2008), which may have an important influence on their fitness relative to sexual vertebrates. Hybrids may suffer fitness disadvantages due to incompatibilities between the parental species’ genomes, but hybrids may also gain advantages by the creation of novel beneficial genetic variants resulting in hybrid vigour. Hybrid vigour might be expected to arise spontaneously among asexual vertebrates due to the high degree of genetic heterozygosity characteristic of interspecific hybrids (leading, perhaps, to a “general purpose genotype”) or might be expected to result from interclonal selection leading to the persistence of only the most fit hybrid genotypes (Lynch 1984). There is, in fact, evidence for hybrid vigour in some asexual vertebrate systems (Hotz et al. 1999; Kearney et al. 2005). In contrast, several theoretical models have shown that the long-term coexistence of asexual vertebrates with their sexual parental species is facilitated if asexuals do not gain a fitness benefit from hybrid vigour, and especially if the asexuals suffer a fitness loss due to hybrid disadvantage, countering the advantage accrued from avoiding the cost of sex (Moore 1976; Hellriegel & Reyer 2000). Hybrid disadvantage has been detected in several asexual vertebrate systems (Wetherington et al. 1987; Lively et al. 1990; Moritz et al. 1991; Weeks 1995; Cullum 1997; Vorburger 2001; Mee & Rowe 2006). It is important to note that these contrasting empirical results (i.e., some asexual vertebrates showing hybrid vigour, and some showing hybrid disadvantage) 2  may be due to variation in the age of these hybrid asexual lineages. Additional data on the ages of asexual vertebrate lineages and on components of fitness of asexual vertebrates and their parental species are required for a thorough analysis of the advantages and disadvantages at play in the maintenance of sex among vertebrates. All hybrid asexual vertebrates share several interesting characteristics. The hybrids are typically ecologically similar to their parental species (they generally have overlapping ecological niches and likely compete with the parental species), they must (with the exception of the parthenogens) coexist with a parental species, and in systems with all-female hybrids they gain a substantial advantage over a bisexual parental species due to the cost of sex. All else being equal, if sexual males do not limit the sperm available to asexual hybrids by discriminating between asexual and sexual females when choosing a mate, the competitive success of the asexuals will ultimately lead to the extinction of both forms. How, then, do these sperm-dependent asexuals not outcompete the parentals given the potential benefits of hybrid vigour and the demographic growth advantage from avoiding the cost of sex? Because males gain no fitness from mating with sperm-dependent asexual hybrids (Schlupp et al. 1994; Goddard et al. 1998), one should expect males to evolve a preference for conspecific sexual females. Preference for conspecific mates has been detected in a number of sperm-dependent asexual systems (Moore & McKay 1971; Abt & Reyer 1993; Schlupp et al. 1998; Engeler & Reyer 2001; Dries 2003), although it is not universal (Balsano et al. 1981; Balsano et al. 1985; Woodhead & Armstrong 1985; Schlupp et al. 1991). Of course, this preference, where it exists, is also not absolute. If absolute preferences had evolved, sperm-dependent asexuals would not persist and so would not be observed. Presumably, costs of preference and/or insufficient time have prevented the evolution of absolute preferences in those systems where asexuals and sexuals persist together. Clearly, explaining the coexistence of sperm-dependent asexuals with their sexually-reproducing hosts is a considerable challenge for evolutionary biologists.  >1>1(@$7:A(@A#$*;B(!"#$%&'()*#(+&*#,-*'(.) I have approached the challenge of explaining the coexistence of spermdependent asexuals with their sexually-reproducing hosts by focusing on a particular sperm-dependent asexual vertebrate, Phoxinus eos-neogaeus, as a model system. 3  Phoxinus eos-neogaeus is gynogenetic and originated from hybridization events between female P. neogaeus (finescale dace) and male P. eos (northern redbelly dace) that probably took place in North American glacial refugia during the last glacial maximum approximately 50,000 years before present (Angers & Schlosser 2007). Hence, all P. eos-neogaeus possess P. neogaeus mitochondrial DNA (which is inherited maternally). Hybrid and parental Phoxinus co-occur across a broad distribution in North America. Each population of P. eos-neogaeus (e.g., within one lake) consists of an assemblage of clonal lineages. Individuals within a clonal lineage (a.k.a., ‘clones’) are genetically indistinguishable at highly variable genetic markers, and often differ from individuals in other clonal lineages by very few mutational steps at these high-mutation-rate markers. There are four lines of evidence suggesting that there has been no recent or ongoing hybridization between P. neogaeus and P. eos: (i) most P. eos-neogaeus clones have widespread distributions, (ii) any given population of P. eos-neogaeus contains a low diversity of clonal lineages, (iii) P. eos-neogaeus often coexist with only one of the parental species (usually P. eos and seldom P. neogaeus), and (iv) several P. eosneogaeus mitochondrial haplotypes have never been detected in any P. neogaeus population (Schlosser et al. 1998; Doeringsfeld et al. 2004; Angers & Schlosser 2007). While P. eos-neogaeus are diploid and typically produce clonal diploid offspring via gynogenesis (Goddard et al. 1998), syngamy occasionally occurs between P. eos sperm and diploid hybrid eggs. These hybrid eggs give rise to two additional hybrid biotypes: triploid hybrids contain an additional P. eos genome, and mosaic hybrids have both diploid and triploid cells in varying proportions (Goddard et al. 1998). Triploid and mosaic hybrid biotypes, which can represent up to 50% of hybrid populations (Angers & Schlosser 2007), arise de novo from diploid hybrids in each generation and do not reproduce gynogenetically (Goddard & Schultz 1993). Instead, triploid and mosaic hybrids produce haploid P. eos eggs, which, when fertilized by P. eos, give rise to diploid individuals with fully P. eos nuclear genomes and, due to their hybrid ancestry, P. neogaeus mitochondrial genomes (Goddard & Schultz 1993). These cytoplasmic hybrids (with P. eos nuclear DNA and P. neogaeus mitochondrial DNA), known as ‘cybrids’ (Binet & Angers 2005), are morphologically indistinguishable from ‘pure’ P. eos (with P. eos mitochondrial DNA). It is not known whether a similar process leading to 4  cytoplasmic replacement in P. neogaeus can follow from syngamy between P. neogaeus sperm and diploid hybrid eggs; this would be difficult to detect because hybrids and P. neogaeus share the same or closely related mitochondrial DNA haplotypes. The Phoxinus system is a good model with which to examine the coexistence of sperm-dependent asexuals with their sexually-reproducing hosts because there is a substantial amount of ecological similarity, diet overlap, and habitat overlap between hybrid and parental Phoxinus (Cochran et al. 1988; Schlosser et al. 1998; Doeringsfeld et al. 2004; Mee & Rowe 2010), although there is some evidence that the hybrid Phoxinus are ecological generalists relative to the parentals (Schlosser et al. 1998; Doeringsfeld et al. 2004). Ecological differentiation could, on its own, promote coexistence between sexual and sperm-dependent asexual species (e.g., Vrijenhoek 1979b, 1994), but ecological similarity suggests that hybrid and parental Phoxinus must, to some degree, be in direct competition for the same resources. Consequently, the factors allowing coexistence may shed light on the processes of general importance to the coexistence of sperm-dependent asexuals and their sexual hosts in a variety of other taxa, whether or not ecological differentiation is present.  >1D1(E*#*&%'F(G7*#$2.0#( With Phoxinus eos-neogaeus as an empirical model system, I have investigated the coexistence of sperm-dependent asexuals and their sexually-reproducing hosts using three related approaches. First, I used analytical and numerical mathematical techniques to develop and refine hypotheses proposed to explain the coexistence of sperm-dependent asexuals and sexually reproducing species. Second, I used biogeographical techniques to elucidate the extent of the distribution of Phoxinus species and to explore geographic patterns of genetic and morphological variation that may provide clues about the origin, and hence the means of persistence, of P. eos-neogaeus. Third, I used experimental and observational data to compare fitness-related traits in asexual and sexual Phoxinus to test the idea that coexistence should be more difficult in systems with hybrid vigour. Using these three approaches I addressed several questions directly related to the coexistence of sperm-dependent asexuals and the sexually-reproducing species on which they depend: 1. Can mate preference maintain both asexual and sexual individuals within a community and how stable is such co-existence? (chapter 2) 5  2. How often, and in what situations, do sperm-dependent asexuals arise? (chapter 3) 3. Do the unusual processes that arise from the involvement of sperm in spermdependent asexual systems have an important role in maintaining the coexistence of sperm-dependent asexuals and their sexually-reproducing hosts? (chapter 3) 4. Does selection for proficiency at soliciting sperm lead to a limited diversity of spermdependent asexual genotypes with host-matching phenotypes? (chapter 4) 5. Do sperm-dependent asexuals show reduced performance in fitness-related traits relative to sexual hosts? (chapters 5 and 6) While the questions in all my chapters were motivated largely by a desire to understand the evolution and maintenance of sex, I am equally motivated by a desire to understand and explore the fascinating natural history of hybrid Phoxinus. Like all sperm-dependent asexual species, hybrid Phoxinus is a wonderful natural oddity. Greater knowledge of the details of the Phoxinus system will add to our appreciation of the diversity and complexity of the biological world.  6  D1(I&%2&$2.0(20($F*(#$%*06$F(.+(;&-*(;&$*('F.2'*(&--.9#(-.06J $*%;('.*<2#$*0'*(.+(#K*%;J:*K*0:*0$(&#*<7&-#(&0:($F*2%(#*<7&-( F.#$#( ( D1>1(@7;;&%A( In several asexual taxa, reproduction requires mating with related sexual species to stimulate egg development, even though genetic material is not incorporated from the sexuals (gynogenesis). In cases where gynogens do not invest in male function, they can potentially have a two-fold competitive advantage over sexuals because the asexuals avoid the cost of producing males. If unmitigated, however, the competitive success of the asexuals would ultimately lead to their own demise, following the extinction of the sexual species that stimulates egg development. I have developed a model of mate choice among sexual individuals and asexual gynogens, where males of the sexual species preferentially mate with sexual females over gynogenetic females, to determine if such mating preferences can stably maintain both gynogenetic and sexual individuals within a community. My model showed that stable coexistence of gynogens and their sexual hosts could occur when there is variation among males in the strength of preference for mating with sexual females and when males with stronger preference pay a higher cost of preference. This result helps to explain the coexistence of spermdependent asexuals with their sexual hosts.  D1D1(?0$%.:7'$2.0( Two types of hypotheses have been proposed to explain coexistence between sperm-dependent asexuals and their sexually-reproducing host species (Schlupp 2005). First, niche separation hypotheses suggest that, while sperm dependence necessitates habitat overlap between asexuals and sexual species, there may be enough niche separation to reduce competition to a degree that facilitates coexistence (Vrijenhoek 7  1979a; Schley et al. 2004). When the asexuals are of hybrid origin, their ecological niche may be intermediate between or on the margins of the parental niches. Occasionally, asexuals might develop an ecologically successful phenotype that breeds true in the absence of sex, which Vrijenhoek (1979a) called the “frozen niche variation” hypothesis for the occasional success of asexual clones. The second type of hypothesis suggests that the effect of preference by sexual males for sexual females may become more intense as asexuals become more common. The resulting frequency-dependent selection against asexuals can limit the growth of asexual populations and allow coexistence. For example, Moore (1976) showed that coexistence can be maintained if it is assumed that male preference for sexuals becomes stronger as sexuals become rarer. Moore (1976) modeled coexistence in a system with unisexual hybridogenic fish in the genus Poeciliopsis. Males in this system have been shown to prefer to mate with conspecific females, but subordinate males show reduced mate discrimination (i.e., as a plastic response to being subordinate) and will mate with hybridogenic females (Moore & McKay 1971). As a result, when the abundance of hybridogens relative to sexual fish in a given location is low, there are many subordinate males, and the hybridogen population can increase in numbers rapidly. This increase is held in check by reduced numbers of subordinate males as sexuals become rarer (males become, on average, pickier when rare). Stable coexistence in Moore’s (1976) model is dependent on a simple assumption that male preferences select more strongly against asexuals as asexuals rise in frequency, resulting in the fecundity of hybridogens exhibiting negative frequency-dependence, which stabilizes coexistence. A model by Heubel et al. (2009) of coexistence in a system with the gynogenetic fish Poecilia formosa also showed that coexistence can be maintained if the gynogens become more sperm-limited as they become more common. Their finite population model considers both the effect of the strength of male mate preference and male fertilization efficiency. When males are highly efficient, a single male can fertilize many females. The model by Heubel et al. (2009) showed that stable coexistence is possible only when sperm is limiting and when male mate preference is sufficiently strong for this limitation to affect gynogenetic females much more than sexual females. Unlike the 8  model by Moore (1976), there is no assumption in the model by Heubel et al. (2009) that male preference becomes stronger as gynogens become more common (i.e., there is no plastic behavioural response to gynogen frequency). Stable coexistence in the model by Heubel et al. (2009) depends on gynogens being more sperm-limited than sexual females when gynogens become common. Heubel et al. (2009) noted that the range of values for the preference and male efficiency parameters that leads to stable coexistence is very small and the parameter values in this range are possibly unrealistic. Also, the frequency of sexuals at the coexistence equilibrium was very low, so that coexistence is unlikely to be stable on an evolutionary time scale due to stochastic perturbations in population size. Long-term coexistence would therefore depend on metapopulation dynamics (Kokko et al. 2008); indeed, other authors have suggested that dispersal between patches can play an important role in allowing coexistence, especially if asexuals “follow” sexual species from patch to patch (Hellriegel & Reyer 2000). For this chapter, I developed a model to try and help explain the coexistence of gynogenetic hybrids and sexually reproducing parental fish in the genus Phoxinus. In my model, like the model by Heubel et al. (2009), males were not assumed to develop stronger preference as a plastic response to increased asexual frequency, and the asexuals’ ecological niche did not differ from that of the sexuals. A novel hypothesis to explain the coexistence of sperm-dependent asexuals and sexually reproducing species emerged from my model: genetic variation among males in the strength of preference for conspecific females can lead to stable coexistence as long as males with stronger preference pay a higher cost of preference. In essence, a co-evolutionary “rock-paperscissors” dynamic arises, where asexuals proliferate when males with weak preference are common, followed by the proliferation of males with stronger preference and the decline of the asexuals, followed by the decline in frequency of males bearing strong but costly preferences, leading the cycle to begin again. Moreover, in the model that I developed, this dynamic approaches a stable equilibrium with all three types present (asexuals, sexuals with weak preference, and sexuals with strong preference).  9  D1C1(M.:*-(=*3*-.K;*0$( D1C1>1(M&$*('F.2'*(20(!"#$%&'(( The P. eos-neogaeus mating system is summarized in figure 2.1 (see also chapter 1). Fertilization in Phoxinus is external, takes place among floating vegetation, and, while some degree of parental care is not impossible, there is no evidence that either parent tends the eggs after fertilization. Citing a presentation by Cooper at the first North American Wildlife Conference in 1936, Scott & Crossman (1979) reported that males, attracted to a ripe female by her darting among vegetation, will dart into a mass of vegetation to join the female and, as a pair or group, struggle against the vegetation to complete the act of spawning. There is, therefore, opportunity for both males and females to choose whether or not to mate with a particular partner. The bright ventral coloration of males during breeding season suggests the existence of female mate choice (or male-male competition), but female mate choice is not relevant to the issue of coexistence in this case, as there are no male asexuals. There are no data on the tendency of males to prefer particular females. It is possible that slight differences in the pattern of horizontal stripes and in body size between female parental and hybrid Phoxinus might serve as cues for male mate choice. D1C1D1(M.:*-(.+(6A0.6*0J#*<7&-('.*<2#$*0'*( The model borrows from the framework established by Kirkpatrick (1982) to explore the evolution of mate choice. There are three haploid genotypes: one gynogenetic (G, all female) and two sexual (E1 and E2, present in both males and females). Note that the two sexual genotypes represent a polymorphism within a single species, not two different sexual species (in my case, I arbitrarily envisioned P. eos as the polymorphic sexual species, hence the designation Ei). Males of genotype Ei prefer to mate with sexual females by a factor of ai, according to the fixed relative-preference scheme of Kirkpatrick (1982). Thus, Ei males will mate with sexual females ai times more frequently than with asexual females, if presented with equal numbers of each. It seems reasonable to assume that these preferences are manifested as a lower rate of mating with asexuals, rather than a higher rate of mating with sexuals, but the mathematical equations are the same in either case. I arbitrarily assumed that the E2 10  males have a stronger preference for sexuals than E1 males (a2 > a1). If the two types of males have equal preferences (a2 = a1), the results are equivalent to those observed with only a single sexual genotype. The effects of the E1/E2 polymorphism are assumed to be sex-limited, affecting male mating preferences but having no effect on females. Because I assumed that females exert no mating preferences, the two types of males have equal mating success (i.e., the fraction of offspring sired by E1 males is given by the frequency of E1 among males, and the same holds for E2 males, see table 2.1). Similarly, there is no difference in mating success between E1 and E2 females, which are phenotypically equivalent. By contrast, because of male mating preferences, the fraction of offspring born to sexual females is higher than the proportion of sexual females by a factor:  X = freq(E1 among males) "  a1 a + freq(E 2 among males) " 2 , a1 a2  (1)  where ai represents the average mating preference of Ei males across the current  !  ! !  population of females: ai = freq(sexual females) " ai + freq(asexual females) " 1.  The fertility advantage of sexual females, X, depends on the frequency of E1 and E2 males and the strength of male mating preferences in the current population ( ai /ai ). Importantly, an assumption of my model is that the fertility advantage of sexual females, X, does not depend on the sex ratio and is the same whether males are rare or ! plentiful (assuming a given array of females so that ai remains constant). This assumption was made in order to avoid introducing negative frequency-dependence directly into the structure of the model. That is, in my model, asexuals are not more or ! less likely to be fertilized relative to a sexual female as asexuals become common (and males become rare), either in the presence of E1 males only or E2 males only. Rather, my results are driven by the interaction among the three different genotypes (E1, E2, and G). In the next section, I discuss simulations that relax this assumption and allow the sex ratio to influence the relative fertilization success of asexual and sexual females. Within my model, a generation begins with sexual genotypes E1 and E2 and gynogenetic genotype G at frequencies x1, x2, and x3, respectively. These represent the overall offspring frequencies, but the frequencies among daughters and among sons differ 11  because of the fact that gynogens produce only daughters. Assuming an equal sex ratio among sexuals and assuming that all gynogens are female, the frequency of E1, E2, and G among female offspring at the start of each generation is x1 /F , x 2 /F , and 2x 3 /F , where  F = x1 + x 2 + 2x 3 , while the frequency of E1 and E2 among male offspring is x1 / M and x 2 / M , where M = x1 + x 2 . These assumptions give gynogens an automatic two-fold ! ! ! competitive advantage for not producing males. ! The fitnesses of gynogens and sexuals differ in two additional ways. First, due to ! viability selection, gynogens are 1 – s times as likely as sexuals to survive to the age of  ! !  reproduction (where s is the selection differential and "# < s < 1 ). Depending on the value of s, viability selection can either augment or mitigate the two-fold advantage of asexual reproduction. Second, due to the cost of preference, E2 males are 1 – c times as ! likely as E1 males to survive to reproduce (where c is the cost of preference and  0 " c < 1). After viability selection and selection against costly mate preferences, the frequencies of males, females, and all possible mating pairs are given in table 2.1. The frequencies of the sexual genotypes and the gynogen in each subsequent  !  generation were determined by iterating the following recursion equations:  x1,t +1 =  a1 x1,t " x1,t 1 x 2,t % a2 1 x 2,t (1( c) x1,t + $ '+ a1 Tm $# Tf 2 Tf '& a2 2 Tm Tf  (2a)  !  x 2,t +1 =  a1 1 x1,t x 2,t a2 x 2,t (1" c) # 1 x1,t x 2,t & + + %% ( a1 2 Tm Tf a2 Tm Tf (' $ 2 Tf  (2b)  !  x 3,t +1 =  1 x1,t 2x 3,t (1" s) 1 x 2,t (1" c ) 2x 3,t (1" s) + a1 Tm Tf a2 Tm Tf  (2c)  !  where Tm and Tf are normalizing factors for males and females, respectively, following the cost of preference, viability selection, and the two-fold cost of sex (see table 2.1). Setting xi,t+1 = xi,t and solving for xi yields the following equilibrium frequencies: {x1 = 1, x2 = 0, x3 = 0}  (3a) 12  {x1 = 0, x2 = 1, x3 = 0} # a1 ( a1 " a2 (1" c ))( a2 " 2(1" s)) % x1 = , ( a1 " a2 )(2a2 (1" c )(1" s) " a1(2 " a2c " 2s)) % % a2 ( a1 " a2 (1" c ))(2(1" s) " a1 ) % , $ x2 = ( a1 " a2 )(2a2 (1" c )(1" s) " a1(2 " a2c " 2s)) % % a1a2c % x3 = 2a2 (1" c )(1" s) " a1 (2 " a2c " 2s) %&  (3b) ' % % % % ( % % % %)  (3c)  ! D1O1(M.:*-(E*#7-$#(&0:(!0&-A#2#(  None of the above equilibria involve the coexistence of gynogens and parentals without the presence of both parental genotypes (both E1 and E2). Hence, my first result is that variation among males in the degree of preference for parental females is essential for coexistence in this model; there is no equilibrium that involves only one type of sexual (either E1 or E2) and asexuals. The stability conditions for the three equilibria were determined by setting x1 = 1 – x2 – x3 and evaluating the eigenvalues of the characteristic polynomial of the Jacobian matrix derived from equations (2b) and (2c). The stability conditions for the two equilibria lacking gynogens (3a and 3b) are straightforward and intuitive. When there are no gynogens and only E1 sexuals present (3a), the E2 genotype cannot invade because it is more costly, but the gynogens can invade as long as 2(1" s) > a1 , that is, as long as the two-fold advantage of not producing males times the fitness of the hybrids (1" s) exceeds the mating bias that E1 males have against asexuals. When no gynogens and only E2 ! sexuals are present (3b), the E1 genotype can invade as long as c > 0, and the gynogen ! can invade as long as 2(1" s) > a2 . To aid with an intuitive understanding of the above stability conditions, readers may note that the gynogen and only one of the sexual genotypes (E1 or E2) can co-exist ! neutrally if exactly. Exact equality in this equation is biologically unrealistic; both ai (preference) and s (selection differential) will most certainly vary 13  stochastically to some degree. If ai is slightly greater than 2(1 " s) , the sexual genotype will outcompete the gynogen and coexistence will be lost. If ai is slightly less than 2(1 " s) , the gynogen will outcompete the sexual genotype and both will go extinct. ! For the coexistence equilibrium (3c), the range of values for each parameter that !  yield positive frequencies for each genotype is limited to a1 < 2(1" s) < a2 and 0<c <  a 2 "a1 a2  . Under these conditions, not only is the coexistence equilibrium valid, it is  also stable. Thus, stable coexistence between gynogens and sexuals requires not only ! !  variation in preference, but pickier males must suffer a cost relative to less picky males. Although the equilibrium is stable when it exists, the eigenvalues are always complex, resulting in damped oscillations in the frequencies of sexual and gynogenetic genotypes. These dynamics can be visualized for a given set of parameter values as shown in figure 2.2; they exhibit rock-paper-scissors behavior, with the asexual genotype G spreading in populations dominated by the less picky E1 genotype, the E1 genotype spreading in populations dominated by the pickier (and more costly) E2 genotype, and the picky E2 genotype spreading in populations dominated by asexuals. Similar dynamics are observed, with a stable equilibrium reached via damped oscillations, in a diploid version of this model, regardless of whether the male preference gene is sex-linked or autosomal (simulation data not shown). A notable result of this model (figure 2.3) is that, in order to achieve and maintain coexistence, mate choice must not be too weak for a given strength of viability selection favoring gynogens (also, viability selection favoring gynogens must not be too strong for a given strength of preference among males). Note that stable coexistence is possible whether viability selection enhances (s < 0) or lessens (s > 0) the two-fold advantage gained by gynogens for not producing males. If gynogens have relatively high viability (for example, if they benefit from hybrid vigor), then E2 males must have even stronger preference for conspecific females. If gynogens have relatively low viability (for example, if they lack hybrid vigor, suffer from hybrid incompatibilities, or are disadvantaged by low population-level genetic diversity) then coexistence is facilitated, in that males need not exhibit such a high degree of preference. In addition, if the cost of preference in E2 males is high, E2 males must have stronger preference in order to support the coexistence of sexuals and asexuals. A higher cost of preference also increased the 14  equilibrium frequency of gynogens. While E1 males must be present in order to support coexistence between gynogens and parentals, having E1 males with weaker preference increased the parameter space (in terms of both s and c) where coexistence is possible. D1O1>1(!--.9206(#*<(%&$2.($.(2;K&'$(%*-&$23*(+*%$2-2T&$2.0(#7''*##(( Sex ratio is likely to be a critical determinant of fertilization rates in sexuals and asexuals. At one extreme, when males are plentiful, all females might be successfully fertilized, even though males prefer sexuals in each mating attempt. Thus, only when males are rare might an asexual female fail to be fertilized. I used individual-based simulations to explore whether my results were robust to including the potential influence of fluctuations in sex ratio, with asexuals remaining unfertilized more often when males are rare (Heubel et al. 2009). In these simulations (see appendix A), each male was allowed to have a fixed number of mating attempts. In each attempt, the male preferred sexual females over asexual females, but as long as a female had received at least one mating attempt she was considered fertilized. Consequently, when males were common, all females were likely to mate, but as males became rare, it became more likely that no male had ever attempted to mate with a given asexual female (inducing negativefrequency dependent selection on the asexuals). These simulations showed that the rockpaper-scissors dynamic is displayed over a wide variety of parameter combinations. Furthermore, stable coexistence is possible for a broader range of parameter values when mating preferences varied than when they were fixed. I thus conclude that variation in male mating preferences is likely to be an important mechanism allowing the stable coexistence between gynogens and sexuals, even when negative-frequency dependence might also be at play. (D1O1D1(@2;7-&$2.0(;.:*-(.+(6A0.6*0J#*<7&-('.*<2#$*0'*(92$F(;7$&$2.0( I designed simulations in order to investigate what would happen to the coexistence of gynogens and sexuals if male preferences were allowed to evolve via mutation. The simulations mimicked the analytical model, above, except that the preference genotype of each sexual individual (ai) was allowed to change in increments of ±0.05 due to mutations that occur at frequency ! in each generation. In addition, the cost of preference was scaled according to the strength of preference of a given male. 15  Specifically, when a male prefered to mate with conspecifics (i.e., when ai > 1), then his fitness was reduced by a factor 1" ( ai "1) /( amax "1) , so that amax was the maximum strength to which preference could evolve before fitness declined to zero. When a male prefered to mate with gynogens (i.e., when 0 < ai < 1), then his fitness was reduced by a ! factor ai. The cost of mating preferences was adjusted in my simulations by altering the value of amax (lower values of amax correspond to preferences for conspecifics being more costly and vice versa). Each simulation began with 4,000 gynogens and 6,000 sexuals, with the population size (N = 10,000) held constant over time. The sexual population began with only two preference genotypes: 2/3 of sexual individuals had a = 2, while the remainder had a = 1. As in the analytical model, the viability of gynogens was 1 – s relative to sexuals; in the simulations reported, I set s = 0.25, reducing the two-fold advantage of asexual reproduction. In my simulations, I investigated the effects of varying the mutation rate and, separately, the cost of preference. I ran 10 replicate simulations of five different parameter combinations for 20,000 generations (summarized in table 2.2). For the parameter values explored, my analytical model predicted stable coexistence of asexuals and sexuals for the initial population with two initial preference genotypes. My simulations showed that coexistence is maintained over many generations even in the face of evolving preferences. The replicates that ended in either extinction (fixation of gynogens) or loss of coexistence (fixation of sexuals) did so due to stochastic fluctuations in population numbers at the peaks or troughs of oscillations in genotype frequencies. There was no deterministic loss of coexistence with the addition of mutation. Replicate simulations with the lowest cost of preference had the lowest equilibrium frequency of gynogens, as predicted by my analytical model, and were therefore most prone to loss of asexuality (see table 2.2). In all simulations, the preference genotypes evolved towards a = 1.5, and when mutation-selection balance was reached the preferences oscillated within approximately ±0.25 of a = 1.5 (see panel E in figure 2.4). At a = 1.5, the fitnesses of sexuals and asexuals exactly balance in the analytic model, so that the sexuals and asexuals coexist neutrally. Selection drives the preference values towards a = 2(1" s) because this strength of preference is high enough to give the sexuals a sufficient fertility advantage to !  16  balance the two-fold cost of sexual reproduction (mitigated in my simulations by viability selection) and any stronger preference would be more costly. The standard deviation in preference over the last 1,000 generations was constant at about 0.06 when ! = 0.001, but increased to 0.1 when ! = 0.01. As such, the oscillations in mean preference around a = 2(1" s) (figure 2.4) were likely driven by the continual appearance of new mutations.  To elucidate the role of mutation in the above results, I ran an additional set of 10 !  replicate simulations without mutations (! = 0) but with initial variation in mating preferences (starting with equal proportions of preference genotypes a = 1.45, a = 1.5, and a = 1.55 in the sexual population and amax = 6) (figure 2.5). When either the minimum or maximum preference genotype (i.e., either a = 1.45 or a = 1.55) was lost (in 7 of 10 replicates), the other extreme genotype (i.e., either a = 1.55 or a = 1.45, respectively) was lost in short order, leaving a monomorphic population with a = 1.5. Once preference variation was lost, the frequency of gynogens drifted neutrally, leading eventually to extinction of the system or loss of the asexuals. In one replicate (indicated in bold in figure 2.5), the intermediate (“optimal”) genotype was lost first, which resulted in sustained oscillations in mean preference and gynogen frequency. These simulations thus confirm that variation in male mating preferences is key to maintaining the coexistence of sexuals and asexuals in this model and that mutation (and potentially migration) is critical to the maintenance of preference variation.  D1H1(=2#'7##2.0( The results of my model of mate choice among gynogenetic hybrids and a sexual parental species (as in the Phoxinus system) suggest that coexistence of sperm-dependent asexuals with their sexually reproducing sperm donors can be explained by the presence of heritable variation among males in their degree of preference for sexual females. Heritable variation in the degree of mate preference certainly exists in nature. Variation in the strength of male mate preference has been detected between populations of Poecilia latipinna, which act as sperm donors for the gynogenetic species Poecilia formosa (Gabor & Ryan 2001; Gabor et al. 2010). Variation in the strength of male mate preference also exists within Poecilia latipinna populations, but evidence suggests that  17  this within population variation has low heritability (Gabor & Aspbury 2008). Future studies should look explicitly for heritable variation in the strength of male mate preference within populations that act as hosts for sperm-dependent asexual species. In addition to variation in male mate preference, my model requires that there be costs associated with increased mate preference, which is well documented in other systems (Gwynne 1989; Iwasa & Pomiankowski 1991; Magnhagen & Vestergaard 1991; Pomiankowski et al. 1991; Forsgren & Magnhagen 1993; Rowe 1994; Magnhagen 1995; Godin & Briggs 1996; Mappes et al. 1996; Booksmythe et al. 2008; Sakurai & Kasuya 2008). The strength of preference required to allow coexistence depends on the relative fitness of asexuals and sexuals. For example, the presence of hybrid vigor in asexuals augments the benefits from the two-fold cost of sex and necessitates a higher degree of preference by males for conspecifics for coexistence to occur. Conversely, depressed fitness in asexuals mitigates the benefits from the two-fold cost of sex and allows a low degree of preference by males for conspecifics to maintain coexistence. There is evidence in the Phoxinus gynogenetic system that gynogens do not differ from parentals in fecundity (i.e., they lack hybrid vigor) (Das & Nelson 1990) and that the hybrids experience a higher parasite load than parentals (i.e., there may be depressed hybrid fitness) (Mee & Rowe 2006). Such depressed hybrid fitness, according to my model, facilitates coexistence. Previous authors have also modeled the coexistence of sperm-dependent asexuals and their sexual hosts (Moore 1976; Hellriegel & Reyer 2000; Heubel et al. 2009). None of these authors have included variation in male preference, but all have shown situations in which stable coexistence is possible. In the case of hybridogenic Poeciliopsis modeled by Moore (1976), the natural history of the system suggests that a heritable polymorphism in male mate preference (i.e., dominant and subordinate males) may contribute to the maintenance of coexistence. The model by Moore (1976) does not, however, explore this possibility. In a model by Hellriegel & Reyer (2000) of hybridogenic water frogs (Rana esculenta) and their sexual counterpart (Rana lessonae), coexistence is maintained by an interesting and potentially unique mechanism. Hybridogenic water frogs (R. esculenta) are not all female, and both male and female hybridogens require mating with a parental 18  species (R. lessonae) for successful reproduction (Schultz 1969). All females (both hybridogens and parentals) prefer to mate with R. lessonae males (Abt & Reyer 1993; Engeler & Reyer 2001). When R. lessonae are common, R. esculenta males mate mostly with R. lessonae females producing only female R. esculenta offspring and thereby causing the R. esculenta fraction of the population to increase. When R. esculenta are common, R. esculenta males mate mostly with R. esculenta females producing no offspring and thereby allowing the R. lessonae fraction of the population to increase. Consequently, when the preference of R. lessonae females is greater than the preference of R. esculenta females for R. lessonae males, the hybridogenic water frog mating system generates negative frequency-dependent selection against hybrids, which allows stable coexistence. A simplification of the model by Hellriegel & Reyer (2000) with nonoverlapping generations and infinite population size (see appendix A) confirms that stable coexistence of hybridogens and sexuals in this model depends on the unique mating system of hybridogenic water frogs (figure 2.1) and not on any assumption of frequency-dependent preference or variation in male (or female) preference. The model by Moore (1976) includes an explicit frequency-dependent mate preference function, where the preference by sexual males for sexual females becomes more intense as asexuals become more common. Similarly, in the model by Heubel et al. (2009) coexistence depends solely on gynogens becoming more sperm limited as they become common. In the model by Hellriegel and Reyer (2000) coexistence depends on the unusual and potentially unique mating system of hybridogenic water frogs. My model proposes an alternative mechanism for stable coexistence between sperm-dependent asexuals and sexually reproducing species. This mechanism involves non-transitive interactions arising from variation in preferences among males for sexual versus asexual mates. When gynogens are at high frequency in my model, males that strongly prefer to mate with sexuals, E2, can spread despite the cost of strong preference because of a process akin to the Fisherian runaway process. Choosy E2 males produce daughters carrying E2 who are more attractive than the asexuals to the males in the population; if the preference is strong enough, this attractiveness outweighs the advantages of asexuality and allows E2 to spread. When the frequency of the gynogens becomes low, E1 benefits from not paying a high cost of preference and increases in 19  frequency. The decrease in gynogen frequency and the replacement of E2 by E1 reduces the relative attractiveness of sexual females and allows the gynogens to increase in numbers. Thus, this mechanism involves non-transitive interactions that generate a rockpaper-scissors game, which in my model ultimately reaches a stable equilibrium (see figure 2.2). The rock-paper-scissors dynamics that I describe requires variation in preferences among males for sexual versus asexual mates. While my model applies specifically to systems with sperm-dependent asexuals, similar rock-paper-scissors dynamics have been observed in several other systems, including male mating strategies among color morphs in side-blotched lizards (Sinervo & Lively 1996), host-parasite interactions in microbial systems (Kerr et al. 2006), and toxin-resistance interactions among bacteria (Kerr et al. 2002). It is important to emphasize that my model allows high frequencies of both gynogens and sexuals at equilibrium over a wide range of parameter values. As such, the coexistence in my model is more likely to be stable over evolutionary time without the requirement for dispersal and metapopulation dynamics observed in previous models (Kokko et al. 2008; Heubel et al. 2009). Of course, the mechanisms maintaining gynogenetic-sexual systems need not be mutually exclusive; I have shown that combining variation in male mate preference with a mechanistic model of mating that causes the fertilization rate of asexual females to depend on the number of males in the population (akin to the model of Heubel et al. 2009) results in an expanded range of parameter values for which coexistence is possible (see appendix A). I also expect that including a behavioural component such that males become more likely to mate with sexuals when they are rare (as in Moore (1976)) would further promote coexistence in my model. Some additional general conclusions about coexistence in systems with spermdependent asexuals can be drawn from the models discussed above. First, in my model as well as those by Moore (1976), Hellriegel & Reyer (2000), and Heubel et al. (2009), coexistence between a sperm-dependent asexual and a sexually reproducing sperm donor species depends on the presence of mate choice. Sexual species must prefer to mate with conspecifics. Second, coexistence is facilitated by the lack of hybrid vigor. This 20  conclusion is evident in my model of the gynogenetic Phoxinus system (figure 2.3), as well as in models of a bisexual hybridogenic system (Hellriegel & Reyer 2000) and a unisexual hybridogenic system (Moore 1976). Coexistence of sexuals and spermdependent asexuals may therefore be more common in systems with hybrid asexuals where hybrid disadvantage mitigates the inherent population growth rate advantage of asexuals.  21  Table 2.1. Frequencies of mating among sexual genotypes, E1 and E2 (with frequencies x1 and x2, respectively), and the gynogenetic genotype, G (with frequency x3). The degree of preference for sexual females by males of type Ei depends on the parameter ai (a1 < a2). The frequencies of male and female genotypes prior to mating are shown as mi and fi, respectively. The Tm and Tf terms are normalizing factors for males and females, respectively, following the cost of preference (c), viability selection (s), and the two-fold cost of sex. The term ai represents the average mating preference of Ei males across the current population of females.  !  Mating pair  Mating frequency  Male  Female  E1  E1  E1  E2  E1  ! G  E2  ! E1  E2  E!2  E2  ! G  ! f1 = x1 /Tf f 2 = x 2 /Tf ! f 3 = 2x 3 (1" s) /Tf Tf = x1 + x 2 + 2x 3 (1" s)  m1 f1a1 a1 m1 f 2 a1 a1 m1 f 3 a1 m2 f1a2 a2 m2 f 2 a2 a2 m2 f 3 a2  Frequency of offspring type E1 E2 G 1  0  0  !  !  0  0  0  1  !  !  0  0  1  0  0  0  1  m1 = x1 /Tm m2 = x 2 (1" c) /Tm Tm = x1 + x 2 (1" c)  ai = ai ( f1 + f 2 ) + f 3 !  !  !  22  Table 2.2. Summary of simulation parameters and results. Each set of parameters (i.e., each column) was used in 10 replicate simulations, each of which was run for 20 000 generations, or until the coexistence was lost or the populations went extinct. Parameter values common to all simulations and replicates: s = 0.25, N = 10 000, ainit. = {1, 2} (see text for more details on parameters). Simulation parameters Mutation rate (!)  0.0001  0.001  0.001  0.001  0.01  amax  6  11  6  5  6  Cost of preference 1  moderate low  moderate high  moderate  > 20,000  12,121  > 20,000  > 20,000  > 20,000  16 773  7230  12 881  12 498  19 680  11 709  4618  7369  4603  16 795  4  10  6  4  1  2  0  3  3  0  Simulation results Max. generations to extinction or loss of coexistence Mean generations to extinction or loss of coexistence Min. generations to extinction or loss of coexistence Num. reps. with loss of coexistence Num. reps. with extinction 1  Cost of preference depends on the strength of preference of a given male, and on the maximum possible strength of preference (amax).  23  Figure 2.1. Summary of type and proportion of offspring produced from mating between sperm-dependent hybrids and sexually reproducing parental species. (A) Mating between gynogenetic P. eos-neogaeus and sexually reproducing P. eos. Note the 2:1 ratio of hybrid to parental daughters, assuming all females produce the same number of offspring. (B) Mating between hybridogenic R. esculenta and sexually-reproducing R. lessonae, adapted from Engeler & Reyer (2001). Note the 3:1 ratio of R. esculenta to R. lessonae daughters, assuming all females produce the same number of offspring.  24  Figure 2.2. Illustrations of the behavior in a system approaching stable equilibrium with coexistence of a gynogen, a sexual genotype with stronger male preference, and a sexual genotype with weaker male preference. A) A de Finetti diagram (produced with Mathematica code available at http://mathgis.blogspot.com/2008/12/how-to-make-tenary-plot.html) in which closer proximity to a point of the triangle indicates higher frequency of the genotype at that point. Cycles are illustrated in a clockwise direction, with peak genotype frequencies occurring in a particular order (E1, G, then E2). The black arrows on the sides of the triangle indicate the behavior of the system when only two genotypes are present and when the adjacent inequalities are satisfied. B) An example of damped oscillations approaching stability over 1000 generations. The curves illustrate the frequencies of the weaker preference sexual genotype E1 (blue, thick-dashed curve), the stronger preference sexual genotype E2 (green, fine-dashed curve), and the gynogen G (red, solid curve). The parameter values for the examples in both A and B are a1 = 1, a2 = 2, s = 0.25, and c = 0.2 (see text for explanation of parameters). These parameter values, as well as the equilibrium genotype frequencies shown in this figure, are equivalent to those used to parameterize and initiate the simulations (with amax = 6) presented in this chapter.  25  Figure 2.3. Parameter space allowing stable coexistence of gynogenetic and sexual individuals. Parameter combinations within the frame outside the black boundary do not allow a polymorphism to exist. In this region of the figures (pure white background), either gynogens dominate (i.e., x3 ! 1) and the community goes extinct, or sexuals dominate (with the less-costly male preference allele spreading to fixation, x1 ! 1). The darker shading represents a higher equilibrium frequency of gynogens.  26  Figure 2.4. An example of the simulated evolution of mate preference. In this example, amax = 6 and ! = 0.001. Panels A through D illustrate the course of evolution from the start of the simulation until mutation-selection balance is reached. Panels C and D represent the distribution of strengths of preference at the peak and trough, respectively, of a cycle in mean strength of preference after mutation-selection balance has been reached. Panel E shows the change in mean strength of preference over generations, with the generations represented in panels A through D indicated with arrows.  27  Figure 2.5. Replicate simulations of mate preference evolution with no mutation. Preference genotypes at the beginning of each replicate were a = 1.45, a = 1.5, and a = 1.55 in equal proportions in the sexual population. Each row shows the change in mean preference (left) and gynogen abundance (right) for a single replicate. Arrows indicate the generation in which one of the three preference genotypes was lost. The sixth replicate (shown in bold) is the only replicate where the a = 1.5 genotype was lost first.  28  C1(,F*('A"%2:(203&#2.0B(92:*#K%*&:(%*K-&'*;*0$(.+(!"#$%&'()*#(( P)2#'*#B(/AK%202:&*Q("A('A$.K-&#;2'(FA"%2:#( ( C1>1(@7;;&%A( Peculiar reproductive processes that often arise during sperm-dependent asexual reproduction may have an important, although underappreciated, role in allowing stable coexistence of sperm-dependent asexual species and their sexually-reproducing host species. Though typically clonal, non-clonal reproduction in hybrid Phoxinus has been shown to occasionally reestablish diploid P. eos nuclear genomes. These hybrid-derived P. eos have P. neogaeus mitochondria, but are otherwise indistinguishable from ‘pure’ P. eos. My aim in this chapter was to determine the abundance and extent of the distribution of these cytoplasmic hybrids (a.k.a., ‘cybrids’), and hence to determine their potential role in mechanisms of asexual-sexual coexistence. I also set out to determine the geographic origins of hybrid Phoxinus and cybrids. My sampling of Phoxinus species included locations across their range in North America, spanning from the Canadian Maritime provinces to northeastern British Columbia. This sampling revealed that cybrids have replaced ‘pure’ P. eos in the northern portion of their distribution. Also, all hybrid Phoxinus and cybrids throughout the sampled distribution had mitochondrial DNA sequences closely related to a P. neogaeus lineage originating from a Mississippi glacial refugium. A divergent P. neogaeus lineage was discovered in Atlantic Canada, but no hybrid Phoxinus or cybrids were derived from this Atlantic lineage. It is unlikely that hybrid Phoxinus, and hence cybrids, originated in any Atlantic glacial refugia. If P. neogaeus mitochondria are better-adapted to northern environments, the widespread mitochondrial DNA introgression may facilitate coexistence in northern areas by reducing any disparity in fitness between hybrid Phoxinus and P. eos.  29  C1D1(?0$%.:7'$2.0( Asexually reproducing vertebrates are uncommon, but they are a fascinating and important component of vertebrate biodiversity. All known asexually reproducing vertebrates are of hybrid origin (Dawley & Bogart 1989; Avise 2008). There are about 30 truly asexual (i.e., parthenogenic) vertebrate species, and about 50 vertebrate species that employ sperm-dependent asexual reproduction (i.e., gynogenesis, hybridogenesis, or kleptogenesis) (Dawley & Bogart 1989; Avise 2008). In sperm-dependent asexuals, sperm from a parental (sexually reproducing) species is required to stimulate egg development, but offspring are typically either fully clonal (in gynogenesis) or partially clonal (in hybridogenesis and kleptogenesis) (Dawley & Bogart 1989; Avise 2008). Due to the hypothesized costs of sexual reproduction, asexuality allows greater population growth rate than sexual reproduction, and, all else being equal, an asexual species should easily be able to outcompete a sexual species (Lively 1996). Asexual vertebrate species have commonly been studied in the context of the evolution and maintenance of sexual reproduction. For example, several studies have shown that asexuals have higher susceptibility to parasitism, which may facilitate asexual-sexual coexistence by shifting the competitive advantage in favour of sexuals (e.g., Lively et al. 1990; Moritz et al. 1991; Mee & Rowe 2006). Several theoretical models have been proposed to account for the role of spermdependence in the coexistence of closely related sexual and asexual species (Schlupp 2005). Sperm-dependence necessitates the involvement of males, and these males must prefer to mate with conspecific sexual females in order for coexistence to be stable (Moore 1976; Hellriegel & Reyer 2000; Heubel et al. 2009; Mee & Otto 2010). Some models have also shown that the stability of coexistence is influenced by the relative fitness of the sexuals and sperm-dependent asexuals (Moore 1976; Hellriegel & Reyer 2000; Mee & Otto 2010). Multiple empirical studies have evaluated mate choice and relative fitness in sperm-dependent asexual systems (Moore & McKay 1971; Balsano et al. 1981; Balsano et al. 1985; Woodhead & Armstrong 1985; Wetherington et al. 1987; Lively et al. 1990; Schlupp et al. 1991; Abt & Reyer 1993; Weeks 1995; Cullum 1997; Schlupp et al. 1998; Hotz et al. 1999; Loyning 2000; Engeler & Reyer 2001; Vorburger 2001; Dries 2003; Kearney et al. 2005; Mee & Rowe 2006; Mee et al. 2011). Relatively 30  little attention, however, has been paid to the peculiar reproductive processes that sometimes result from sperm-dependent asexual reproduction, and to how these peculiar processes may affect mate choice or relative fitness, and hence coexistence, in spermdependent asexual systems. Non-clonal reproduction in hybrid Phoxinus can occur via a peculiar process that is thought to be unique to this system. Triploid hybrids produce haploid P. eos eggs, which, when fertilized by P. eos in laboratory experiments, give rise to diploid individuals with fully P. eos nuclear genomes and, due to their hybrid ancestry, P. neogaeus mitochondrial genomes (Goddard & Schultz 1993). These cytoplasmic hybrids (with P. eos nuclear DNA and P. neogaeus mitochondrial DNA), known as ‘cybrids’ (Binet & Angers 2005), are morphologically indistinguishable from ‘pure’ P. eos (i.e., with P. eos nuclear and mitochondrial DNA). Phoxinus cybrids have been detected in natural populations (Binet & Angers 2005; Angers & Schlosser 2007), and it is possible that these natural cybrids have arisen via sexual hybridization events. This latter possibility is, however, unlikely because evidence suggests that hybrid Phoxinus arose about 50,000 years ago (Angers & Schlosser 2007), and that there has been no recent or ongoing hybridization between P. eos and P. neogaeus (Schlosser et al. 1998; Doeringsfeld et al. 2004; Angers & Schlosser 2007). Given the peculiarity of the process involved in the creation of cybrids, and the fact that cybrids were until recently known only from laboratory breeding experiments (Goddard & Schultz 1993), the reasonable expectation is that cybrids are a rare component of P. eos populations. Indeed, previous work has shown that cybrids may occur in populations with or without pure P. eos, but that most lakes contain no cybrids at all (Angers & Schlosser 2007). In the present chapter, I report a striking and unexpected result: cybrids appear to have replaced pure P. eos across an extensive area of North America. In order to understand the processes involved in driving the pattern of this replacement, I investigated the mitochondrial DNA genealogy of taxa in the P. neogaeus mitochondrial lineage (i.e., cybrids, P. eos-neogaeus, and P. neogaeus).  31  C1C1(M&$*%2&-#(&0:(M*$F.:#( C1C1>1(@&;K-*#( Fish were sampled across the range of Phoxinus in North America, spanning from the Canadian Maritime provinces, east to Montana and north to northeastern British Columbia (figures 3.1 and 3.2). A total of 1,133 fish were sampled from 37 lakes or rivers across 11 localities (see table 3.1). The Canadian Department of Fisheries and Oceans provided the samples from the Experimental Lakes Area (ELA) in northwestern Ontario, and the Prince Edward Island (PEI) Department of Environment, Energy and Forestry provided samples from PEI. In all other cases, I sampled Phoxinus using minnow traps. All fish were euthanized by immersion in an ms222 solution and preserved in 95% ethanol. I extracted DNA from each fish sampled in my study using a standard phenol-chloroform DNA isolation protocol for fish tissues. C1C1D1(Z7'-*&%(&0:(;2$.'F.0:%2&-(=Z!(2:*0$2+2'&$2.0( The methods I used to identify nuclear genotypes and mitochondrial DNA haplotypes were developed, assessed, and validated previously (Binet & Angers 2005; Angers & Schlosser 2007). In brief, I differentiated nuclear genomes based on an intron of the mesoderm specific transcript gene (Binet & Angers 2005). I differentiated mitochondrial genomes based on the cytochrome oxydase subunit I (COI) gene (Binet & Angers 2005). Sequence differences between P. eos and P. neogaeus in these genes allowed species-specific primers to be designed to amplify different fragment sizes for each species (Binet & Angers 2005). Fragments of DNA from these genes were amplified by PCR and separated using electrophoresis on a 2% agarose gel, and visualized on a UV light table after staining with SYBR® Safe DNA gel stain. C1C1C1(M2$.'F.0:%2&-(6*0*&-.6A( I used a 685 bp fragment of the COI mitochondrial gene to determine the genealogy of the P. neogaeus mitochondrial lineage. Fragments of DNA for sequencing were amplified using primers and PCR procedures developed and described previously (5’ to 3’ primer sequences: ACC AAC CAC AAA GAC ATT GG and CCT GTT AAT AGA GGG AAT CAG) (Binet & Angers 2005; Angers & Schlosser 2007). Amplified  32  fragments were purified using Qiagen PCR purification kits, and purified fragments were sent to Macrogen Korea for sequencing. Sequences were aligned manually using Se-Al v2.0a11. Aligned sequences were used to give an estimate of the genealogy of COI haplotypes based on statistical parsimony (Clement et al. 2000). To obtain as comprehensive a picture of the distribution of cybrids as possible, I included haplotype and distributional data from a previous study (Angers & Schlosser 2007).  C1O1(E*#7-$#( C1O1>1(=2#$%2"7$2.0(.+('A"%2:#( The proportions of cybrids and pure P. eos in lakes across their distribution indicate that there was very little geographic overlap between cybrids and pure P. eos (figure 3.1). Pure P. eos occur in southern Ontario, Nebraska, Minnesota, and Québec (Angers & Schlosser 2007). Mixed populations, comprised of cybrids and pure P. eos, occur only in southern Ontario and Québec. By contrast, all P. eos populations across a vast area from British Columbia, Montana, Alberta, Saskatchewan, northern Ontario, PEI, and New Brunswick were comprised entirely of cybrids (figure 3.1). C1O1D1()FA-.6*.6%&KFA(.+($F*(!.)&*#,-*'((;2$.'F.0:%2&-(-20*&6*( Within the P. neogaeus mitochondrial lineage, I found eight COI haplotypes (haplotypes A, and F through L). One of these eight haplotypes, haplotype A, had been discovered previously along with four additional haplotypes (B, C, D, and E) (Angers & Schlosser 2007). Haplotype A was the most common and widespread of all haplotypes (figure 3.2). All other haplotypes had restricted distributions, and were often confined to a single lake. A phylogenetic reconstruction of all 12 haplotypes (figure 3.2A) revealed that haplotype F, which was the only haplotype found in southern New Brunswick, was the most divergent haplotype and differed from haplotype A by 1.22% sequence divergence. The centre of the geographic range of sampled populations, including locations in Ontario and Minnesota, had the highest diversity of haplotypes (figure 3.2B3.2D).  33  C1H1(=2#'7##2.0( My data suggest that pure P. eos have been replaced by cybrids across a wide geographic area. This mitochondrial replacement in Phoxinus is an unusual case because, unlike typical cases of mitochondrial introgression, mitochondrial transfer has likely occurred without any backcrossing and subsequent introgression of nuclear DNA between P. eos and P. neogaeus (backcrossing is unlikely in the Phoxinus system because typical sexual hybrids have never been observed). There are at least two explanations for the pattern of mitochondrial replacement in Phoxinus. First, the replacement may have occurred due to a stochastic ratchet-like process: a Phoxinus community comprised of hybrids and pure P. eos can give rise to cybrids, but a Phoxinus community comprised of hybrids and cybrids cannot give rise to pure P. eos. Second, there may be a selective advantage to having a P. neogaeus mitochondrial genome in certain environments. That cybrids tend to dominate in the northern portion of the distribution suggests that P. neogaeus mitochondria might be more cold-adapted than P. eos mitochondria. Evidence for the role of mitochondrial genetic variation in cold adaptation in fishes has been documented previously (Lannig et al. 2005; Lucassen et al. 2006) and mitochondrial introgression has been implicated in the thermal adaptation of other freshwater fish species. For example, Arctic char (Salvelinus alpinus, a northern species) mitochondrial DNA has become fixed via introgression in northern populations of the typically low latitude lake trout (S. namaycush) and brook trout (S. fontinalis) (Bernatchez et al. 1995; Glemet et al. 1998; Wilson & Bernatchez 1998; Doiron et al. 2002; Cote et al. 2007), and bull trout (S. confluentus, typically a southern species) mitochondrial DNA has been found in southern populations of the typically high latitude Dolly Varden (S. malma) (Redenbach & Taylor 2002). The high diversity of mtDNA haplotypes in Minnesota and Ontario suggests that these locations are close to a putative centre of origin for the post-glacial dispersal of P. neogaeus and P. eos-neogaeus. The star-like haplotype network (figure 3.2A) is also suggestive of rapid dispersal from a single glacial refugium (Avise 2009). The clear exception is the highly divergent P. neogaeus haplotype found in southern New Brunswick. The existence of this divergent haplotype adjacent to unglaciated areas of the Atlantic coast suggests the existence of two P. neogaeus phylogroups: one associated 34  with a Mississippi glacial refugium, and one associated with an Atlantic glacial refugium. None of the mtDNA haplotypes found in P. eos-neogaeus or cybrids are, however, derived from the Atlantic P. neogaeus haplotype (figure 3.2A). Cybrids with a Mississippi mtDNA haplotype co-occur in New Brunswick with Atlantic-origin P. neogaeus. This discordance suggests that cybrids dispersed to their present distribution in the Atlantic, likely from a Mississippi glacial refugium, rather than originating from P. eos-neogaeus in situ. My data suggest, therefore, that the entire sampled distributions of hybrids and cybrids originated via rapid dispersal from a Mississippi glacial refugium, a common source of postglacial dispersal in widespread fishes (McPhail & Lindsey 1970). The abundance of cybrids in nature casts doubt on the ancestry of hybrid Phoxinus. It has been assumed that P. neogaeus is the maternal ancestor of P. eosneogaeus because all hybrids have P. neogaeus mitochondria. It is possible, however, that a female cybrid and a male P. neogaeus were the ancestors of all extant hybrids. This possibility seems remote because cybrids are rare in the unglaciated regions where hybrid Phoxinus are thought to have originated. The frequency of cybrids in refugial populations at the time of hybrid origin is, however, unknown, and the ancestry of hybrid Phoxinus remains in doubt. The existence of cybrids in the Phoxinus eos-neogaeus system may have an important role in the maintenance of asexual-sexual coexistence. Other sperm-dependent asexual systems may also have unique components or particular processes that influence coexistence. For example, in the case of the unisexual hybridogenic species Poeciliopsis monacha-lucida, Quattro et al. (1991) suggested that the hybridogenic mating system allows a feedback process of genetic exchange between the asexual and sexual species that maintains diversity in both and facilitates coexistence. In general, the particularities inherent in mechanisms of sperm-dependent asexual reproduction may have an important and underappreciated role in the coexistence of asexual and sexual species. To understand the coexistence of asexual and sexual Phoxinus, one must consider the coexistence of asexuals with both cybrids and pure P. eos. The biogeographical survey that I have completed suggests that, in essence, asexual Phoxinus have ‘changed the game’ with regards to their competition and coexistence with P. eos by causing the replacement of P. eos mitochondria with P. neogaeus mitochondria in northern 35  populations. Evaluating the physiological performance of P. eos and P. neogaeus mitochondria in different environments would certainly help to understand this fascinating example of asexual-sexual coexistence. Several studies have looked at the role of mitochondrial variation in thermal tolerance (Lannig et al. 2005; Lucassen et al. 2006; Cote et al. 2007). If P. neogaeus mitochondria are better-adapted to northern environments, this mitochondrial introgression may facilitate coexistence in northern areas by ‘leveling the playing field’ (i.e., reducing any disparity in fitness) between P. eos-neogaeus and its most common sperm donor.  36  Figure 3.1. Pure Phoxinus eos (red) have been replaced by cybrids (black) throughout much of their distribution in North America (grey area - adapted from Scott and Crossman (1979)). Boxes represent sampling locations. The circles in each box represent the lakes sampled from that location. Each circle shows the relative proportions of pure P. eos and cybrids in a single lake. An empty space in a box indicates a lake in which no P. eos or cybrids were found. Dashed lines indicate locations characterized previously (Angers & Schlosser 2007).  37  Figure 3.2. Genealogy and distribution of haplotypes. A: Genealogical relationships among mtDNA COI haplotypes that have been found in P. neogaeus (black), nuclear hybrids (grey), or cybrids (white). More common haplotypes are represented by proportionally larger symbols, but partitions within haplotypes do not represent proportions of species with that haplotype. Lines separate haplotypes that differ by at least one base pair. Each dot indicates one additional base pair difference. B-D: The geographic distribution of P. neogaeus (B), P. eos-neogaeus (C), and cybrid (D) haplotypes in North America. Each circle shows the haplotype(s) found in each species in each lake. An empty space in a box indicates a lake in which no P. neogaus, hybrids, or cybrids were found. Dashed lines indicate locations characterized previously (Angers & Schlosser 2007).  38  O1(S.#$J#2T*J;&$'F206(20(&(#K*%;J:*K*0:*0$(&#*<7&-(+2#F(#K*'2*#( ( O1>1(@7;;&%A( Sperm-dependent asexual species exist in several phyla, including chordates, rotifers, platyhelminths, nematodes, annelids, mollusks, and arthropods, and hence are an important component of biodiversity. The persistence of sperm-dependent asexuals, and coexistence with their sexually reproducing host species, is strongly influenced by the sexual species’ mate preference, and hence the propensity for the sexual species to provision the asexual species with sperm. There is likely selection for traits among asexual clones that increase the rate of sperm provisioning. This selection should favour sperm-dependent asexual clones that match their sexual host in traits that are used as cues in mate choice within the sexual species. I first set out to determine whether different clones of a sperm-dependent asexual species (Phoxinus eos-neogaeus) co-occur with different host species (P. eos and P. neogaeus). I then compared size and shape traits among populations of the sperm-dependent asexual species and their sexual hosts to test the hypothesis of host-matching. Based on genetic distances among clones, my results show that sperm-dependent asexual clones that use P. eos sperm cluster separately from those that use P. neogaeus sperm. In addition, populations of sperm-dependent asexual clones that use P. eos sperm are more genetically divergent from those that use P. neogaeus sperm than expected based on geographic distance. These genetically divergent P. eos-neogaeus clones that use different host species are also divergent in size, but not shape, and they also match the size of their host species. This study is the first to test and provide evidence for host-specialization in a sperm-dependent asexual species.  O1D1(?0$%.:7'$2.0( Asexually reproducing species have a clear advantage over sexually reproducing species, all else being equal: asexuality confers the ability to reproduce without spending resources on costly male function. Much effort has been directed by evolutionary biologists towards reconciling the expected outcome of this clear advantage to asexual 39  reproduction (i.e., that asexual reproduction should be abundant and common) with the empirical observation that sexual reproduction is abundant and common (e.g., Darwin 1862; Felsenstein 1974; Williams 1975; Maynard Smith 1978; Hamilton 1980; Bell 1982; Kondrashov 1988; Otto & Barton 2001; Agrawal 2006). This paradoxical success of sexual reproduction is most apparent in vertebrates – there are probably fewer than 100 asexually reproducing species amongst the approximately 58,000 known vertebrates (Dawley & Bogart 1989; Avise 2008). Most asexual vertebrate species are spermdependent, requiring heterospecific sperm to, at the very least, stimulate egg development (as in, for example, gynogenesis). Sperm-dependence is common among asexual vertebrates, and it also exists among asexual taxa in several phyla, including rotifers, platyhelminths, nematodes, annelids, mollusks, and arthropods (Beukeboom & Vrijenhoek 1998). Hence, sperm-dependent asexual reproduction is an important aspect of the diversity of asexual species and of biodiversity more generally. Sperm-dependence introduces new difficulties to understanding how either asexual or sexual species gain a competitive advantage. For example, if sexual males do not limit the sperm available to sperm-dependent asexuals, the asexuals should profit fully from their ability to reproduce without spending resources on male function. But if the males do limit the sperm available to the asexuals (e.g., by preferring to mate with conspecific sexual females), the asexuals may be severely limited in their ability to reproduce and may not persist. The importance of the role of mate choice in the persistence of coexisting sexual and sperm-dependent asexual species has been suggested in several theoretical studies (Moore & McKay 1971; Moore 1976; Hellriegel & Reyer 2000; Heubel et al. 2009; Mee & Otto 2010), and the existence of mate choice has been documented in several systems with sperm-dependent asexuals (Moore & McKay 1971; Abt & Reyer 1993; Schlupp et al. 1998; Engeler & Reyer 2001; Gabor & Ryan 2001; Dries 2003; Gabor & Aspbury 2008; Gabor et al. 2010), but the link between theory and mate choice in sexual-asexual systems has rarely been tested empirically. Given the apparent importance of mate choice in allowing the coexistence of sperm-dependent asexuals with their sperm-providing hosts, there are likely traits among the asexuals that are under selection to increase the rate of sperm provisioning. Variation between asexual clones in traits important for soliciting sperm may arise via mutation, 40  but it is likely that most of the genetic variation between clones would have arisen at the time the asexual lineage originated (e.g., via hybridization). Sexual selection should lead to a winnowing of this diversity of sperm-dependent asexual genotypes (i.e., clones), leaving only those that are adequately proficient at soliciting sperm from a sperm donor. A prediction of this hypothesis is that, as a result of sexual selection, sperm-dependent asexual clones should resemble the females of their sperm-donor species. This hostspecialization should be most evident in cases where the sperm-dependent asexual uses sperm from different host species in different populations. In such cases, I predicted that sexual selection should result in genetically divergent populations, each of which is adapted to use sperm from a different host species. I studied a sperm-dependent asexual freshwater fish species (Phoxinus eosneogaeus) that coexists with two sexual species (P. eos and P. neogaeus) and that can use sperm from either sexual species. Phoxinus eos-neogaeus is gynogenetic and originated from hybridization events between female P. neogaeus (finescale dace) and male P. eos (northern redbelly dace) that probably took place in North American glacial refugia during the last glacial maximum approximately 50,000 years before present (Angers & Schlosser 2007). Hybrid and parental Phoxinus co-occur across a broad distribution in North America. Each population of P. eos-neogaeus (e.g., within one lake) consists of an assemblage of clonal lineages. Individuals within a clonal lineage (a.k.a., ‘clones’) are genetically indistinguishable at highly variable genetic markers, and often differ from individuals in other clonal lineages by very few mutational steps at these high-mutationrate markers. There are four lines of evidence suggesting that there has been no recent or ongoing hybridization between P. neogaeus and P. eos: (i) most P. eos-neogaeus clones have widespread distributions, (ii) any given population of P. eos-neogaeus contains a low diversity of clonal lineages, (iii) P. eos-neogaeus often coexist with only one of the parental species (usually P. eos and seldom P. neogaeus), and (iv) several P. eosneogaeus mitochondrial haplotypes have never been detected in any P. neogaeus population (Schlosser et al. 1998; Doeringsfeld et al. 2004; Angers & Schlosser 2007). I addressed two questions related to the coexistence of asexual and sexual species in this system. First, are P. eos-neogaeus populations that use P. eos sperm genetically divergent from those that use P. neogaeus sperm? Second, do clones that use P. 41  neogaeus sperm match P. neogaeus females in size or shape while those that use P. eos sperm match P. eos females in size or shape? Answering these questions constitutes a test of the prediction that sexual selection has resulted in a limited diversity of asexual clones with specific host-matching traits (i.e., size and shape) in each asexual population.  O1C1(M&$*%2&-#(&0:(M*$F.:#( O1C1>1(@&;K-206($*'F02[7*#(&0:(#K*'2*#(2:*0$2+2'&$2.0( Fish were collected from streams and lakes across Canada (from northeastern British Columbia to New Brunswick) using seine netting, electrofishing, and minnow trapping (table 4.1; see also table B1). All fish were euthanized by immersion in an MS222 solution and preserved in 95% ethanol. I extracted DNA from each fish sampled in my study using a standard phenol-chloroform DNA isolation protocol for fish tissues. The methods I used to differentiate Phoxinus hybrids and parentals were developed, assessed, and validated previously (Binet & Angers 2005; Angers & Schlosser 2007). In brief, I differentiated species based on an intron of the mesoderm specific transcript gene (Binet & Angers 2005). Sequence differences between P. eos and P. neogaeus in this gene allowed species-specific primers to be designed to amplify different fragment sizes for each species (Binet & Angers 2005). Fragments of DNA were amplified by PCR, separated using electrophoresis on 2% agarose gels, and visualized on a UV light table after staining with SYBR® Safe DNA gel stain. O1C1D1(X*0*$2'(:23*%6*0'*(&;.06)!.)*#(+&*#,-*'((K.K7-&$2.0#( To address the question of whether P. eos-neogaeus populations that use P. eos sperm are genetically divergent from those that use P. neogaeus sperm, hybrid Phoxinus were genotyped at four high-variability microsatellite DNA loci (table 4.2). These four microsatellite loci are specific to P. eos and are therefore hemizygous in the hybrids. Diploid hybrids can only have a single allele, and triploid hybrids can have up to two alleles. Any hybrid that did not have two alleles at any locus was assumed to be a diploid hybrid because, assuming the diversity of alleles in the sexual populations (i.e., the source of the third genome in the triploids) was similar to that in the asexual populations  42  (see table 4.2), the probability of finding a triploid homozygous at all four loci was extremely low (e.g., approximately 0.00011 in Albertan populations and 0.00024 in British Columbian populations assuming all alleles are at equal frequency in the sexual population). I excluded any triploids from my analysis because it was impossible to know which of the two triploid alleles belonged to the hybrid genome and which allele came from the sperm donor. Any hybrids with the same four-locus genotype were considered to be clones. The genetic distances between clones were estimated using average square distance (D1) (Moran 1975; Goldstein et al. 1995). The genetic distances between hybrid populations were estimated using RST (Slatkin 1995). These genetic distance measures were chosen because they take account of the possibility that current microsatellite allele size retains information about the ancestral state, and, hence, that two alleles that differ by one or two repeats are likely the result of a single mutational step and, therefore, are more closely related to one another than two alleles that differ by multiple repeats (Moran 1975; Goldstein et al. 1995; Slatkin 1995). Both genetic distance measures (D1 and RST) were computed using the program Microsat (Human Population Genetics Laboratory, Stanford University). I used a partial Mantel test to estimate the correlation between genetic distance (RST(1-RST)-1) and host species distance (1 = different host, 0 = same host) among hybrid populations, and to account for the effect of spatial autocorrelation. For all populations in Alberta, no clear waterway connections between lakes existed (lakes are connected by ground water, and no inlet or outlet streams are present). I therefore used log over-land geographic distance between the Alberta populations in the partial Mantel test. For all BC populations, there were clear waterway connections and I used waterway distance (non-log-transformed) between populations in the partial Mantel test. The results of these Mantel tests did not change qualitatively (results not shown) when distance measures were computed differently (e.g., log-transformed or not; overland distance or waterway distance). O1C1C1(M.%KF.-.62'&-(:2++*%*0'*#(&;.06(!.)*#(+&*#,-*'(('-.0*#( To address the question of whether P. eos-neogaeus clones match their host Phoxinus species in size or shape, I compared shape characteristics of five “species groups” of Phoxinus: HE: hybrids that use P. eos sperm; E: P. eos that act as hosts for 43  HE; HN: hybrids that use P. neogaeus sperm; NN: P. neogaeus that act as hosts for HN; NE: P. neogaeus that coexist with E and HE, but are not hosts for HE. Eight external body shape characters, as well as standard length (SL), were measured for 172 females from the Albertan populations sampled in this study (53 E, 59 HE, 29 HN, 16 NN, and 15 NE). These individuals included those used in the genetic analyses described above, as well as 8 NN and 13 NE individuals from Albertan populations that were added to the dataset in order to ensure adequate sample sizes. Only diploid hybrids were used in these comparisons because there is some evidence that triploid hybrid Phoxinus differ from diploid hybrid Phoxinus in shape and size, and that triploids are more variable (Doeringsfeld et al. 2004). The eight shape characters measured were: mouth angle (MA – angle at the tip of the snout between the middle of the eye and the posterior of the maxillary), upper jaw length (UJL – distance from the tip of the snout to the posterior of the maxillary), pectoral fin position (PFP – distance from the tip of the snout to the anterior point of insertion of the pectoral fin), eye position (EP – distance from the tip of the snout to the anterior-most margin of the eye), eye diameter (ED – horizontal eye diameter), head length (HL – distance from the tip of the snout to the posterior margin of the operculum), caudal peduncle depth (CPD – narrowest distance between the dorsal and ventral edges of the caudal peduncle), and pre-dorsal length (PDL – distance from the tip of the snout to the anterior point of insertion of the dorsal fin). These shape characters were chosen because they had previously been used to show morphological differences between hybrid and parental Phoxinus (Schlosser et al. 1998; Doeringsfeld et al. 2004). Analysis of morphological variation followed the methods discussed by Thorpe (1976). In order to remove the allometric variation in morphology, I regressed each morphological character against standard length. Individual values for the eight morphological characters were then replaced by the residuals of these regressions against length. On top of removing allometric variation, this method also standardized the morphological characters to zero mean and unit standard deviation. Allometry-free morphological variation was then summarized using principal coordinates analysis (PCO). I obtained a score for each individual on each PCO axis. I used a linear mixed effects model (with species group as the fixed effect, and lake or stream of origin as the random effect) and analysis of variance (ANOVA) to determine the effect of Phoxinus 44  species group on standard length and on PCO axis scores, and I used Tukey's ‘honest significant difference’ method to evaluate pairwise comparisons between species groups. All statistical analyses were performed using the R programming language (R Development Core Team 2010).  O1O1(E*#7-$#( In order to compare hybrid Phoxinus that use P. eos sperm with those that use P. neogaeus sperm, it was necessary to find regions where there are hybrid populations that use P. eos sperm as well as hybrid populations that use P. neogaeus sperm. After an extensive pan-Canadian search (37 sites across six Canadian provinces), two such regions were identified: one in northeastern British Columbia (BC) near the town of Fort Nelson, and one in southwestern Alberta (AB) near the town of Caroline (figure 4.1). I determined which of the two parental species acts as the sperm donor in each hybrid population based on the relative abundance of the parental species at each sampling location, and by the presence of triploids with an extra set of P. eos chromosomes among the hybrids in each population. There were two types of Phoxinus communities in the lakes and rivers I sampled (type 1 and type 2 communities). Eight sampling locations (one in BC and seven in Alberta) contained the first type of community (type 1) comprised of P. eos and P. eos-neogaeus, with P. neogaeus occasionally present at low frequency. In all eight of these locations, P. eos outnumbered P. neogaeus. In all but one of these eight locations, even those with very low abundances of P. eos, I found triploid hybrids with an extra set of P. eos chromosomes (table 4.1), indicating that P. eos are frequently involved in hybrid reproduction. I assumed that the hybrids in type 1 lakes use P. eos sperm. Six locations (four in BC and two in Alberta) contained the second type of community (type 2) that was comprised mainly of P. neogaeus and P. eosneogaeus, with P. eos occasionally present at low frequency. In all six of these locations, P. neogaeus outnumbered P. eos. I have never found a triploid hybrid Phoxinus with an extra set of P. eos chromosomes in type 2 communities (table 4.1), indicating that any P. eos that do exist in these six locations very rarely (or never) participate in the hybrid’s reproduction. I assumed that the hybrids type 2 locations use P. neogaeus sperm (table 4.1). 45  O1O1>1(X*0*$2'(:23*%6*0'*(&;.06)!.)*#(+&*#,-*'((K.K7-&$2.0#( In general, each hybrid population was comprised of a unique set of clones. Of the 88 clones found in the 14 hybrid Phoxinus populations sampled in BC and Alberta, five clones were found in more than one population (figures 4.5 and 4.6). No clones were shared between type 1 and type 2 locations in Alberta (figures 4.5). One clone was shared between type 1 and type 2 locations in BC (figure 4.6). Many clones differed from each other by a single repeat at one or more microsatellite loci (figures 4.5 and 4.6). In Alberta, each of these closely related clusters of clones was distributed in either type 1 or type 2 locations, and never shared between location types (figure 4.5). In BC, these clusters of clones were distributed across location types (figure 4.6). Clones from type 1 lakes in BC did not cluster with clones from type 1 lakes in Alberta, and clones from type 2 lakes in BC did not cluster with clones from type 2 lakes in Alberta (figure 4.7). Clones from the same province (BC or Alberta) did not tend to cluster together (figure 4.7). Consistent with my prediction that sexual selection will result in genetically divergent asexual populations, genetic relatedness among hybrid populations in Alberta was significantly positively correlated with host species, irrespective of geographic proximity, such that populations with the same host were more closely related than populations with different hosts (n = 9, Mantel r = 0.44, two-tailed p = 0.0192; figure 4.2). By contrast, and contrary to my prediction, genetic relatedness between populations in BC was negatively correlated with host species, but not significantly so (n = 5, Mantel r = -0.52, two-tailed p = 0.152; figure 4.2).  O1O1D1(M.%KF.-.62'&-(:2++*%*0'*#(&;.06(!.)*#(+&*#,-*'(('-.0*#( My analysis of morphological variation among Phoxinus species included only Albertan samples because genetic relatedness among asexual populations was not correlated with host species in BC (see above). The first PCO axis of morphological variation (PC1) explained 45.3% of the variation in shape for Albertan Phoxinus (table 4.3). Although there were no significant differences on PC1 among Phoxinus species groups (df = 6, F = 3.22, p = 0.0977), the trend in morphological variation among species, as summarized on PC1, was consistent with previous studies of morphological 46  differences between P. eos and P. neogaeus (Schlosser et al. 1998; Doeringsfeld et al. 2004). Compared with P. eos, P. neogaeus had longer heads, larger and more posterior eyes, more posterior pectoral and dorsal fins, broader caudal peduncles, more horizontal mouths, and longer upper jaws. Hybrids were generally intermediate between P. eos and P. neogaeus in these traits. Contrary to my prediction that sexual selection will result in asexual populations that phenotypically match their host species in morphology, there was no significant difference on PC1 between hybrids in the two types of communities described above (HE and HN), and there was no clear tendency for hybrids to match the body shape of the species they use for sperm (figure 4.3). The other PCO axes (table 4.3, figure 4.3) also showed no significant differences between HE and HN hybrids. The only PCO axes that showed significant variation among Phoxinus species groups were PC3 (df = 6, F = 6.3619, p = 0.0238) and PC5 (df = 6, F = 7.782174, p = 0.0148) (figure 4.3). Principal coordinate axis 3 (PC3) showed that P. neogaeus (NN) was significantly different in shape from both groups of hybrids (NN-HE: z = 2.708, p = 0.0473; NN-HN: z = 4.465, p < 0.0001), and PC5 showed significant differences between E and HE (z = -3.739, p = 0.00140) and between NN and HN (z = 4.124, p < 0.001). Consistent with my prediction that sexual selection will result in asexual populations that phenotypically match their host species in morphology, there were clear differences in standard length between Phoxinus species groups in Alberta (df = 6, F = 6.0342, p = 0.0269; figure 4.4). Pairwise comparisons between species groups revealed that P. eos was significantly smaller than all species groups except the hybrids that use P. eos sperm (E-NN: z = 4.042, p = 0.000479; E-HN: z = 3.961, p = 0.000655; E-NE: z = 4.042, p = 0.000479; E-HE: z = 2.443, p = 0.096734). Hybrids that use P. eos sperm (HE) were closest in size to P. eos (E), and hybrids that use P. neogaeus sperm (HN) were closest in size to P. neogaeus (NN) (figure 4.4). Variation in size and shape within hybrid Phoxinus was not due to size and shape differences between clones, but rather to variation within clones (figures 4.8 and 4.9).  47  O1H1(=2#'7##2.0( For sperm-dependent asexuals that use sperm from different host species in different populations, like P. eos-neogaeus, I predicted that sexual selection should result in genetically divergent populations adapted to use the sperm of different host species. Among the populations sampled from Alberta, the P. eos-neogaeus in type 1 communities (i.e., those that use P. eos sperm) were more genetically divergent from P. eos-neogaeus in type 2 communities (i.e., those that use P. neogaeus sperm) than would be expected based on the geographic distance between lakes, and genetically-similar clones tended to be distributed in either type 1 or type 2 lakes, but not shared between lake types. These results are consistent with my prediction. This prediction follows from my hypothesis that adaptation for proficiency at soliciting sperm from sexual species is important for the persistence of sperm-dependent asexuals. The obvious alternative explanation for the genetic divergence between P. eosneogaeus in different lake types is that the divergent hybrid genotypes originated from in situ hybridization between P. eos and P. neogaeus. In addition to the pre-existing evidence suggesting that no recent or ongoing hybridization has occurred between P. eos and P. neogaeus (Schlosser et al. 1998; Doeringsfeld et al. 2004; Angers & Schlosser 2007), my results provide evidence against the possibility that in situ hybridization has produced divergent P. eos-neogaeus populations in type 1 and type 2 lakes in Alberta. First, in many of the locations sampled for this study, only one of the parental species has ever been recorded. Second, five clones were found in more than one lake, and most clusters of very closely-related clones (e.g., differing by a single mutational step at one microsatellite locus) were found in several geographically distantly separated lakes. Hence, in situ hybridization was unlikely the source of the diversity of clones found in any given lake. Populations of P. eos-neogaeus sampled from type 1 and type 2 communities in BC were not more genetically divergent than expected based on the geographic distance between populations (figure 4.2). The lack of a relationship between genetic distance and host species difference in BC P. eos-neogaeus populations might be due to a strong influence of historical biogeographic effects rather than sexual selection. The BC populations are over 900 km farther northeast from the likely source of the post-glacial 48  dispersal of all Phoxinus in North America (i.e., a Mississippi glacial refuge) than the Albertan populations. Consequently, the BC populations may have been founded more recently, and hence been more influenced by genetic drift, than the Albertan populations. The notion that the length of time that sperm-dependent asexuals have coexisted may influence mate choice, and the traits associated with mate choice, has been suggested for other systems. For instance, Gabor et al. (2010) suggested that variation in the strength of male mate choice across populations of Poecilia latipinna, which acts as the sperm donor for the gynogenetic species Poecilia formosa, may be related to the length of time of co-evolution between the host and the sperm-dependent asexual. There was a strong tendency for P. eos-neogaeus to match the body size of their host (figure 4.4). Given that size is used as a cue for assortative mating in many fish species (Foote 1988; Foote & Larkin 1988; Maekawa et al. 1994; Schluter & Nagel 1995; Hatfield & Schluter 1996; Schluter 1996; Jennions et al. 2001; McLean et al. 2005; Baube 2008; Baldauf et al. 2009; Gabor et al. 2010; Thunken et al. 2011), including species that act as hosts for sperm-dependent asexuals (Gabor et al. 2010), this result is consistent with my prediction that sperm-dependent asexual clones should resemble the females of their host species as a result of selection to be proficient at soliciting sperm from males. The pattern of size matching in Alberta lakes is, however, also consistent with the potential effects environmental adaptation. An environment that favours small fish would favour P. eos (the smaller of the two parental species in all contexts) over P. neogaeus and would select for smaller P. eos-neogaeus. Conversely, an environment that favours large fish would favour P. neogaeus over P. eos and select for larger P. eosneogaeus. Phoxinus neogaeus were smaller and less abundant in type 1 lakes than in type 2 lakes, and P. eos were abundant in type 1 lakes and absent in type 2 lakes, suggesting that small fish are favoured in type 1 lakes and large fish are favoured in type 2 lakes. Consistent with an environmental adaptation hypothesis, P. eos-neogaeus are smaller in type 1 than type 2 lakes. Hence, it is not possible with the present data to determine whether the observed host-size matching is the result of sexual selection or environmental selection on P. eos-neogaeus. Nonetheless, this uncertainty does not discount the intriguing possibility that sexual selection has influenced the evolution and persistence of this sperm-dependent asexual species. 49  There was no tendency for P. eos-neogaeus to match the shape of their host (figure 4.3). In fact, P. eos-neogaeus that use P. neogaeus sperm are more similar in shape to P. eos than to P. neogaeus (figure 4.3). This result suggests that there is no shape-based divergent selection among P. eos-neogaeus clones, which is not consistent with my prediction that sperm-dependent asexual clones should resemble the females of their host species as a result of selection to be proficient at soliciting sperm from the males. While size is a well-established cue for assortative mating in fish (see above), the role of shape in assortative mating is less well established. It is possible that shape is less important than size for mate recognition or mate preference, and that there has therefore been very little shape selection among P. eos-neogaeus clones. My results show a substantial amount of variation in body size and shape within P. eos-neogaeus. Phenotypic variation (e.g., shape, size, behaviour, etc.) within P. eosneogaeus clones can be caused by environmental or epigenetic effects (Massicotte et al. 2011). Doeringsfeld et al. (2004) previously showed that the amount of size and shape variation within clonal hybrids is on par with the amount of variation in the parental species. Similarly, in my study, there was size and shape variation within clones, and the extent of this variation was not remarkably different from that in parental species (figures 4.8 and 4.9). Also, the body size and shape variation within P. eos-neogaeus populations was not partitioned among clones, but rather each clone from the same lake type had a similar mean size and shape (figures 4.8 ad 4.9). Hence, plastic or epigenetic variation is unlikely to be the cause of the host-size matching in type 1 and type 2 lakes. My study is the first to suggest, and to provide evidence consistent with, hostspecialization in a sperm-dependent asexual species. The existence of this hostspecialization is a manifestation of the importance of mate choice in allowing the coexistence of sexual and sperm-dependent asexual species (Moore & McKay 1971; Moore 1976; Hellriegel & Reyer 2000; Heubel et al. 2009; Mee & Otto 2010). If a sexual species has very strong preference for conspecific mates (e.g., if the spermdependent asexual does not adequately match the relevant sexual mating cues), then the sperm-dependent asexual will not persist. On the other hand, if the sexual species indiscriminately provides sperm to conspecifics and to the sperm-dependent asexual (e.g., if males are not very selective about which species uses their sperm), then the sexual 50  species will be at a disadvantage because the asexual species does not spend resources producing costly males. When a sperm-dependent asexual species first arises (for example, via hybridization in asexual vertebrates) there may be antagonistic coevolution between the sexual species and the sperm-dependent asexual, with selection favouring increased sperm-provisioning via host-matching in the asexuals, while selection favours decreased sperm-provisioning via conspecific mate preference in males. The end point in this antagonistic coevolution (i.e., stable coexistence) may be reached when the diversity of asexual clones becomes sufficiently diminished, or when the cost of mate choice exceeds the benefit of conspecific mate preference. It is possible that the two types of Phoxinus communities that I described in this study (i.e., those in which the asexuals use P. eos sperm, and those in which the asexuals use P. neogaeus sperm) have reached two different end points of this antagonistic coevolution, with the asexuals in one community specializing on P. eos hosts, and the asexuals in the other community specializing on P. neogaeus hosts. A logical next step in the study of this system would be to measure the strength of preference among P. eos and P. neogaeus males for females of both sexual species, and for asexuals from both types of communities. A prediction that follows from the findings of my study is that males should prefer to mate with females of a size similar to their own species. This could lead to directional selection for smaller asexuals in populations where P. eos is the sperm donor, and directional selection for larger asexuals where P. neogaeus is the sperm donor, or stabilizing selection for a particular size in each scenario. Male Poecilia latipinna, which are hosts for the gynogenetic species Poecilia formosa, prefer to mate with conspecific females, and Poecilia latipinna females are smaller than Poecilia formosa (Gabor & Ryan 2001; Gabor et al. 2010). Gabor et al. (2010) found that mate preference in male Poecilia latipinna caused stabilizing selection for intermediate female size in one population, and directional selection for larger female size in another population. This pattern of female size preference in Poecilia latipinna, which is contrary to the prediction of selection favouring host matching, may be the result of strong selection for mate quality (of which female size is a strong predictor), which may be in conflict with selection for species recognition. Future studies, in Phoxinus or other systems, could evaluate the size preference of males of different host 51  species, as well as the nature of sexual selection on body size in sperm-dependent asexuals coexisting with different host species of different size.  52  Table 4.1. Sampling locations and sample sizes for Phoxinus in British Columbia (BC) and Alberta (AB). Bold numbers among the parental species’ sample sizes indicates which parental species is inferred to be the most likely host (i.e., sperm donor) for hybrid reproduction (see section 4.4 for how this inference was made). Locations where hybrids use P. eos are type 1 locations, and locations where hybrids use P. neogaeus are type 2 locations. Province BC BC BC BC BC AB AB AB AB AB AB AB AB AB  Sampling location (location code) Tsinhia Lake (TSL) Parker Lake (PKL) Beaver Lake (BVL) McConachie Creek (MCC) Stanolind Creek (SDC) Alford Lake (ALF) Tay Lake (TAY) Strubel Lake (STB) Ironside Lake (IRN) Mitchel Lake (MIT) Gasplant Lake (GAS) Fiesta Lake (FIE) Gunrange Lake (GUN) Tea Lake (TEA)  Number of fish sampled Hybrids P. eos P. neogaeus (% 3N) 94 14 15 (54%) 1 2 17 (0%) 0 7 4 (0%) 0 4 4 (0%) 0 0 17 (0%) 0 12 28 (0%) 0 1 31 (0%) 17 0 24 (9%) 1 0 39 (14%) 4 0 32 (27%) 29 3 18 (24%) 35 1 28 (6%) 29 1 3 (0%) 28 0 12 (43%)  Total 123 20 11 8 17 40 32 41 40 36 50 64 33 40  Lake type type 1 type 2 type 2 type 2 type 2 type 2 type 2 type 1 type 1 type 1 type 1 type 1 type 1 type 1  53  Table 4.2. Microsatellite DNA loci used to identify clonal diversity among hybrid Phoxinus sampled from British Columbia (BC) and Alberta (AB). Locus Pho1  Number of alleles AB BC (n = 182)1 (n = 56)1 17 15  Total (n = 238)1 20  Primers (5’ – 3’) Reference CTACAGTATCTATTTATCTAG Binet & Angers GCGGTTTACCGCCCAGCAC (2005) Pho2 26 22 35 CGGACAGACATACATACAG Binet & Angers GTAAATGTAGTTAATTGCCAC (2005) Pho60 13 7 15 GGAATTTATCAAACACAGACAG Angers & GGATGAGAACGGGCAGATC Schlosser (2007) Pho61 16 18 26 TATTGAATTCCAAGATAGC Angers & TGGAAAGTGTGATGAGACTG Schlosser (2007) 1 Note that these primers are specific to P. eos and are therefore hemizygous in the hybrids. Each diploid hybrid has one allele, and each triploid hybrid has up to two alleles. Hence, assuming all triploids are heterozygous, the sample size shown equals the number of chromosomes sampled (1 per diploid hybrid, 2 per triploid hybrid).  54  Table 4.3. Phoxinus body shape character loadings on all principal coordinate axes (PCO), and percent of overall shape variation explained by each axis. MA = mouth angle, UJL = upper jaw length, PFP = pectoral fin position, EP = eye position, ED = eye diameter, HL = head length, CPD = caudal peduncle depth, PDL = pre-dorsal length. Shape character MA UJL PFP EP ED HL CPD PDL  PCO axis loading PC1 PC2 0.400 -0.302 -0.374 -0.092 -0.419 -0.228 -0.440 0.037 -0.289 0.298 -0.483 -0.110 -0.087 -0.789 -0.082 0.351  PC3 0.123 -0.202 0.324 -0.150 -0.255 0.098 0.144 0.849  PC4 -0.203 0.177 0.035 0.340 -0.848 -0.073 -0.283 -0.080  PC5 0.006 -0.839 0.286 0.282 -0.005 0.221 -0.124 -0.266  PC6 -0.680 -0.267 -0.333 0.125 -0.007 -0.289 0.489 0.132  PC7 0.455 -0.006 -0.510 0.669 0.066 0.116 0.117 0.232  PC8 -0.152 -0.064 -0.462 -0.354 -0.197 0.771 -0.011 -0.025  8.90  6.86  5.32  3.58  1.26  Percent variation explained 45.34  15.66  13.09  55  Figure 4.1. Sampling sites in British Columbia and Alberta. Closed dots represent locations with type 1 Phoxinus communities comprised of P. eos and P. eos-neogaeus, with P. neogaeus occasionally present at low frequency. Open dots represent locations with type 2 Phoxinus communities, comprised mainly of P. neogaeus and P. eos-neogaeus, with P. eos occasionally present at low frequency. See table 4.1 for the key to the location names.  56  Figure 4.2. Plots of genetic distance (RST/(1-RST)) against geographic distance between hybrid Phoxinus populations in Alberta and British Columbia. Open dots represent pairs of populations that use sperm from different host species, and closed dots represent pairs of populations that both use sperm from the same host species.  57  Figure 4.3. Plot depicting the scores for Phoxinus eos-neoageus, P. eos, and P. neogaeus on the first prinicipal component axis of shape variation in two different types of communities (above and below the dotted line). In type 1 communities (below the dotted line), P. eos-neogaeus (HE) uses sperm from P. eos (E), although P. neogaeus (NE) is also present. In type 2 communities (above the dotted line), P. eosneogaeus (HN) uses sperm from P. neogaeus (NN). Dark bars represent median scores, the boxes encompass the scores between the 1st and 3rd quartile, and the whiskers extend 1.5 times the interquartile range from the box. Significant differences between groups (p < 0.05) are indicated by different lower case letters on the right of the plot.  58  Figure 4.4. Plot depicting the standard length (mm) of Phoxinus eos-neoageus, P. eos, and P. neogaeus in two different types of communities (above and below the dotted line). In type 1 communities (below the dotted line), P. eos-neogaeus (HE) uses sperm from P. eos (E), although P. neogaeus (NE) is also present. In type 2 communities (above the dotted line), P. eos-neogaeus (HN) uses sperm from P. neogaeus (NN). Dark bars represent median scores, the boxes encompass the scores between the 1st and 3rd quartile, and the whiskers extend 1.5 times the interquartile range from the box. Significant differences between groups (p < 0.001) are indicated by different lower case letters on the right of the plot.  59  Figure 4.5. Hierarchical clustering of Phoxinus eos-neogaeus clones from Alberta based on average square distance (D1) derived from variation across four microsatellite DNA loci using the ‘complete linkage’ method in the hclust function in R (R Development Core Team 2010). A filled-in box indicates that a clone (at the tip of a branch) is found in the population listed at the top of the grid. Black and red boxes indicate clones found in type 1 communities (which use P. eos sperm) and type 2 communities (which use P. neogaeus sperm), respectively. See table 4.1 for names and locations of each population.  60  Figure 4.6. Hierarchical clustering of Phoxinus eos-neogaeus clones from BC based on average square distance (D1) derived from variation across four microsatellite DNA loci using the ‘complete linkage’ method in the hclust function in R (R Development Core Team 2010). A filled-in box indicates that a clone (at the tip of a branch) is found in the population listed at the top of the grid. Black and red boxes indicate clones found in type 1 communities (which use P. eos sperm) and type 2 communities (which use P. neogaeus sperm), respectively. See table 4.1 for names and locations of each population.  61  Figure 4.7. Hierarchical clustering of Phoxinus eos-neogaeus clones from BC and Alberta based on average square distance (D1) derived from variation across four microsatellite DNA loci using the ‘complete linkage’ method in the hclust function in the R programming language (R Development Core Team 2010). A filled-in box indicates that a clone (at the tip of a branch) is found in a population that uses either P. eos (black; type 1 community) or P. neogaeus (red; type 2 community) sperm and was found in either British Columbia or Alberta, as indicated at the top of the grid.  62  Figure 4.8. Plot depicting the scores for Phoxinus eos-neoageus, P. eos, and P. neogaeus on the first prinicipal coordinate axis of body shape variation in two different types of communities (above and below the dotted line). In type 1 communities (below the dotted line), P. eos-neogaeus (HE) uses sperm from P. eos (E), although P. neogaeus (NE) is also present. In type 2 communities (above the dotted line), P. eosneogaeus (HN) uses sperm from P. neogaeus (NN). Box plots have been replaced with individual data points representing the scores for individual P. eosneogaeus. Colours differentiate clones within each community type (no clones are shared above and below the dotted line). Significant differences between groups (p < 0.001) are indicated by different lower case letters on the right of the plot.  63  Figure 4.9. Plot depicting the standard length (mm) of Phoxinus eos-neoageus, P. eos, and P. neogaeus in two different types of communities (above and below the dotted line). In type 1 communities (below the dotted line), P. eos-neogaeus (HE) uses sperm from P. eos (E), although P. neogaeus (NE) is also present. In type 2 communities (above the dotted line), P. eos-neogaeus (HN) uses sperm from P. neogaeus (NN). Box plots have been replaced with individual data points representing the length of individual P. eosneogaeus. Colours differentiate clones within each community type (no clones are shared above and below the dotted line). Significant differences between groups (p < 0.001) are indicated by different lower case letters on the right of the plot.  64  H1(E*K*&$(#92;;206(K*%+.%;&0'*(&0:(2$#(2;K-2'&$2.0#(+.%( 20+*%%206($F*(%*-&$23*(+2$0*##(.+(&#*<7&-(FA"%2:(:&'*(P)2#'*#B( !"#$%&'(Q(&0:($F*2%(#*<7&--A(%*K%.:7'206(K&%*0$&-(#K*'2*#( ( H1>1(@7;;&%A( While theories explaining the evolution and maintenance of sex are abundant, empirical data on the costs and benefits of asexual relative to sexual reproduction are less common. Asexually reproducing vertebrates, while few, provide a rare opportunity to measure differences in fitness between asexual and sexual species. All known asexually reproducing vertebrates are of hybrid origin, and hybrid disadvantage (i.e., reduced hybrid fitness) is thought to facilitate long-term coexistence between asexual and sexual species. I used repeat swimming performance as a proxy for fitness to compare the fitness of asexual hybrid dace, Phoxinus eos-neogaeus, and their sexually reproducing parental species, P. neogaeus and P. eos. I tested the prediction that, given the widespread coexistence of these hybrid and parental Phoxinus, the parental species should show at least equivalent, if not superior, repeat swimming performance relative to hybrids. A repeat constant acceleration test (Umax) was conducted at both acclimation temperature (16 °C) and at an elevated temperature (25 °C) to simulate the combined influence of a repeat swim and acute temperature change that fish might experience in the wild. The asexual hybrids performed more poorly than one of the parental species (P. neogaeus). There was a negative effect of temperature on repeat swimming performance in all fish, and the repeat performance of hybrids was more severely affected by temperature than that of P. neogaeus. No difference in the effect of temperature on repeat performance was detected between hybrids and P. eos. These results suggest that hybrids suffer physiological costs relative to P. neogaeus, and that the hybrids do not gain advantage from hybrid vigour over P. eos, which probably contributes to the coexistence of asexual and sexual species in this system.  65  H1D1(?0$%.:7'$2.0( Asexual reproduction is rare among vertebrates; depending on how one defines an asexual “species”, there are only 30 to 90 known vertebrate examples (Dawley & Bogart 1989; Avise 2008). In contrast with the low number of asexual species, the theoretical advantages of asexual reproduction are numerous and have been of interest for well over a century (Darwin 1862; Felsenstein 1974; Williams 1975; Barash 1976; Maynard Smith 1978; Charlesworth 1980; Hastings 1999). The most commonly cited, and perhaps the most intuitive and important, advantages of asexual reproduction include the fact that individual asexual parents pass all their genes to their offspring, whereas sexuals pass only half their genes on to their offspring (Charlesworth 1980), and asexual populations have double the potential growth rate of sexual populations (Lively 1996; Jokela et al. 1997). Based upon these advantages, it is perhaps surprising that asexual reproduction among vertebrates is not more common. The observation that asexuality is rare, combined with the theoretical expectation that it should be more common, has motivated the development of many hypotheses to explain how the advantages of asexual reproduction are balanced or outweighed by advantages inherent to sexual reproduction (Kondrashov 1993; Agrawal 2006). Although evolutionary biology is not lacking in theories for why sex is advantageous and common, a thorough understanding of the reasons why sexual reproduction is common, particularly among vertebrates, can be achieved through empirical study of closely related asexual and sexual species. Phoxinus eos-neogaeus and its parental species are well suited to the empirical study of asexual vertebrate species. Ecological similarity suggests that hybrid and parental Phoxinus must, to some degree, be in direct competition for the same resources (Cochran et al. 1988; Schlosser et al. 1998; Doeringsfeld et al. 2004; Mee & Rowe 2010), and coexistence must depend on the same processes that are of general importance to the coexistence of sperm-dependent asexuals and their sexual hosts in a variety of taxa. Several theoretical models have shown that the long-term coexistence of asexual vertebrates with their sexual parental species is facilitated if asexuals do not gain a fitness benefit from hybrid vigour, and especially if the asexuals suffer a fitness loss due to hybrid disadvantage (Moore 1976; Hellriegel & Reyer 2000; Mee & Otto 2010). Hence, 66  comparisons of fitness related traits in asexual vertebrates and their sexual parental species have direct relevance to understanding this coexistence. Piscivorous fishes and birds commonly prey upon Phoxinus, and the ability of these fishes to evade a predator by engaging in high intensity swimming multiple times in succession likely has important fitness consequences. An evasive response that included sustained high intensity swimming away from a predator was important for the survival of fathead minnows (Pimephales promelas) (Webb 1982), a cyprinid species that cooccurs with Phoxinus across much of its range. Acute exposure to temperatures above optimal values imposes an additional stressor that may reduce swimming performance whether the swim test is aerobic or anaerobic in nature (Beamish 1978; Randall & Brauner 1991). The ability of a fish to swim fast enough to evade a predator is likely affected by incomplete recovery from a previous evasive swim as well as by temperature, but the influence of an acute temperature change on repeat swimming performance has not previously been evaluated. The functional morphology and physiology of swimming performance is a critical aspect of the life of aquatic animals. In fact, tests of swimming ability have been used for decades to test hypotheses concerning the evolution of adaptations of migratory and stream-dwelling fishes (e.g.Brett 1964; Taylor & McPhail 1985, 1986; Taylor & Foote 1991; Brauner et al. 1994; Nelson et al. 2003). Further, tests of repeat swimming performance (i.e., when an animal is tested twice in rapid succession) have been used to study fish health and the metabolic processes involved in recovery from exhaustion (Jain et al. 1998; Jain & Farrell 2003). In this chapter, I use a test of repeat swimming performance to compare the swimming ability, as a proxy for fitness, of hybrid Phoxinus to that of its sexually reproducing parental species. I address the effects of multiple stresses, namely repeat swimming and increased temperature, on swimming ability by conducting the repeat swimming performance test at both a laboratory acclimation temperature and at an elevated temperature. I hypothesize that, given the widespread long-term coexistence of these asexual hybrid and sexually-reproducing parental Phoxinus despite the cost of sex, the asexuals should exhibit hybrid disadvantage (i.e., reduced fitness relative to the parentals) or, at least, should lack hybrid vigour. 67  H1C1(M&$*%2&-#(&0:(M*$F.:#( H1C1>1(@$7:A(K.K7-&$2.0#( Females from four lakes with populations of hybrid and parental Phoxinus were brought into the lab in the summer of 2007. These lakes included two from Alberta and two from Ontario. The Alberta populations were from Alford Lake (N 52º 03' 50'' W 115º 05' 32'') and Strubel Lake (N 52 12' 09'' W 115 0' 4'') near the town of Caroline. Alford Lake contains populations of P. neogaeus and hybrid Phoxinus (Alford Lake contains no northern redbelly dace), while Strubel Lake contains populations of P. eos and hybrid Phoxinus (Strubel lake contains no finescale dace). The Ontario populations were from Pondweed Lake (N 42º 27' 48'' W 78º 27' 30'') and the outlet creek from Sunday Lake (N 45º 35' 14'' W 78º 22' 12'') in Algonquin Provincial Park. Neither Ontario lake contained a P. neogaeus population - both contained only P. eos and hybrid Phoxinus populations. Fish from each lake were kept in 95 L tanks (all the fish from a given lake in a single tank) with no flow, and these tanks were housed in an environmental chamber with a 12hr:12hr light:dark cycle and kept at 16 °C. Fish were fed ad libitum with a mixture of frozen bloodworms and frozen daphnia once daily. One of several colours of elastomer dye was injected subcutaneously at each of four dorsal landmarks (relative to the dorsal fin: front left, front right, back left, back right) to give each fish a unique tag and track individuals throughout the protocol described below. Several fish died within the first month after being brought to the environmental chamber, but there were no mortalities in the several months leading up to the experiment, and no mortalities occurred during the experiment or as the result of tagging. H1C1D1(E*K*&$(#92;;206(K*%+.%;&0'*($*#$(K%.$.'.-( Beginning in September 2008, every fish was subjected to a repeat constant acceleration swimming performance test (Umax) at acclimation temperature (16 °C) and at an elevated temperature (25 °C). These temperatures were chosen because fish tended to remain healthy and active over the long-term at 16 °C, and 25 °C was the highest temperature at which dace acclimated to 16 °C were consistently able to perform a  68  constant acceleration test during pilot experiments. The constant acceleration test (Umax) has been shown to be equivalent to the more commonly employed critical swimming speed test (Ucrit) in juvenile pink salmon (Oncorhynchus gorbuscha) (Nendick et al. 2009), but it tends to overestimate Ucrit by up to 57% in adult rainbow trout (Farrell 2008). The Umax test is a much shorter duration test than the Ucrit test and, while neither measure can be expected to replicate predator-prey interactions in nature exactly, the former is likely a better approximation of the physiological challenge that a lake-dwelling cyprinid might face when escaping a predator. These repeat Umax tests were intended to estimate the simultaneous effects of a repeat swim and elevated temperature (i.e., the effect of multiple stressors) on the burst swimming performance, Umax, of each fish. Before being tested, fish were deprived of food for between 40 and 48 hours in order to standardize any potential influence of digestion or feeding status on swimming performance or behavioural motivation. Three submersible water heaters, immersed in the tank around the 10 L Loligo swim tunnel (Loligo Systems APS), were used to raise the temperature in the swim tunnel (over the course of approximately one hour) several hours prior to the tests at 25 °C (the heaters were left on for the duration of the 25 °C tests). For the tests at 16 °C, the heaters were turned off, but left in the tank, and either the apparatus was left for several hours to return to the ambient temperature in the environmental chamber or cold water was added to hasten the return to ambient temperature. At the beginning of each test, after the water in the swim tunnel had stabilized at the desired temperature, a single fish was transferred to the swim tunnel and left to habituate for 45 minutes at a water velocity of 7.5 cm/sec. Fish were encouraged to swim by gentle prodding with a probe as necessary during the habituation period. Following habituation, an initial measure of swimming performance, Umax1, was performed by increasing the flow rate by 2.5 cm/sec at 30 second intervals until the fish fatigued, which was defined as that point when the fish collapsed (perpendicular to the direction of flow) against the back grid of the swim tunnel for the third time. A second measure of swim performance, Umax2, was then performed immediately following Umax1 (i.e., no recovery period following Umax1) using the same procedure as Umax1 starting at 10 cm/sec. Values of Umax were calculated following Brett (1964): 69  U max = u + t 30sec " 2.5cm/sec  !  where u is the highest speed at which the fish swam for the entire 30 second interval (cm/sec) and t is the time the fish swam before becoming exhausted in the final interval (sec). The mean durations (not including habituation) of each stage of this procedure were 7 min. 22 sec. (Umax1 at 16 °C), 6 min. 2 sec. (Umax2 at 16 °C), 6 min. 43 sec. (Umax1 at 25 °C), and 5 min. 16 sec. (Umax2 at 25 °C). At least three days separated the test at the acclimation temperature and the test at the elevated temperature for each fish (giving the opportunity to recover and feed at least once). Half of the fish from each population were tested initially at the elevated temperature and half were tested initially at the acclimation temperature in order to control for any effects of learning or incomplete recovery between trials. Following the completion of all tests, a fin clip was taken from each fish to genetically identify whether it was a hybrid or one of the parental species using a DNA-based diagnostic method (see Binet & Angers 2005). H1C1C1(=&$&(&0&-A#2#( All analyses were performed using the R programming language (R Development Core Team 2010). Comparisons of swimming performance between coexisting P. eos and hybrid Phoxinus populations (three lakes) were analyzed separately from comparisons of swimming performance between coexisting P. neogaeus and hybrid Phoxinus populations (one lake). The three-lake comparison between coexisting P. eos and hybrid Phoxinus populations provided enough lake-level replication to draw general conclusions about the difference in swimming ability between these fishes. Although only one lake was available for the comparison between coexisting P. neogaeus and hybrid Phoxinus populations, the results of the analyses for this comparison are nonetheless important in identifying trends of potential general relevance. The swimming ability of all fish was analyzed in units of body lengths per second (bl / sec). In order to account for the effect of size on swimming ability, the slope of the effect of fork length on Umax was calculated from a linear mixed effects model, with temperature, repeat swim, and lake (where applicable) as random effects. The slope was  70  then used to adjust measurements of Umax to remove the effect of fish size. All subsequent analyses were performed on this size-adjusted Umax (bl / sec). In order to test the hypothesis that asexuals should display hybrid disadvantage, linear mixed effects models were used to analyze the (fixed) effects of species (hybrid versus parental dace), temperature (16 versus 25 °C), and repeat swimming (Umax1 versus Umax2) on swimming performance (Table 5.1). In these models, since Umax was measured four times for each individual (Umax1 and Umax2 at both 16 and 25 °C), individuals were treated as random effects to control for individual-level differences. In order to provide a more clear comparison of the swimming abilities of hybrid and parental dace when confronted with multiple stresses (i.e., increased temperature and exhaustion), linear models were also used to simply compare Umax under the most stressful condition (Umax2 at 25 °C) as a proportion of Umax under the least stressful condition (Umax1 at 16 °C). In these models, species was the only fixed effect, and lake was included as a random effect only for the comparison of northern redbelly and hybrid dace.  H1O1(E*#7-$#( There were differences in body size between hybrid and parental dace, and sizespecific Umax (bl / sec) declined significantly with increasing body size (figure 5.1). The values of Umax (bl / sec) were therefore size-adjusted for subsequent analyses (see below). The mean fork length (± standard error) of P. eos from populations used in these tests of repeat swimming performance was 5.50 ± 0.37 cm, and mean fork length of P. neogaeus was 6.99 ± 0.60 cm. Mean fork lengths of hybrids in populations coexisting with P. eos and P. neogaeus were 6.53 ± 0.52 cm and 7.60 ± 0.63 cm, respectively. The linear slopes of the relationship between Umax (bl / sec) and fork length were similar for the lake with populations of P. neogaeus and hybrids (slope = -1.16, df = 26, F = 39.2, p < 0.0001) and the group of lakes with populations of P. eos and hybrids (slope = -1.15, df = 63, F = 62.2, p < 0.0001). The slope of the relationship did not differ significantly between parental and hybrid populations for either lake type (P. eos lakes: df = 63, F = 1.8, p = 0.188; P. neogaeus lake: df = 26, F = 0.9, p = 0.356). I explored the possibility that a non-linear model was a better fit to the data by fitting a function of the form  71  U max =  1  ( Length ) a + b  (i.e., estimating the a and b parameters) (figure 5.1). I chose this  particular non-linear model because it has the desired non-linear attributes of the size-  !  Umax (bl / sec) relationship (i.e., an asymptotic increase in Umax to infinity as length approaches zero, and an asymptotic decline to some minimum non-negative value of Umax with increasing length). The non-linear model was not a substantially better fit to the data in the P. eos lakes ("log-likelihood = 0.67; "AIC = -1.33) and provided no improvement in the P. neogaeus lake ("log-likelihood = -0.85; "AIC = 1.71). Therefore, for simplicity, the linear slopes were used to compute the size-adjusted Umax values for all subsequent analyses. Repeat swimming and increased temperature generally reduced swimming performance. The recovery ratios (i.e., Umax2/Umax1) were calculated for repeat swimming performance tests at 16 and 25 °C. At 16 °C (i.e., the acclimation temperature), P. eos had a mean recovery ratio (± standard error) of 0.81 ± 0.04, and the hybrids coexisting with P. eos had a mean recovery ratio of 0.81 ± 0.04. Phoxinus neogaeus and their coexisting hybrids had mean recovery ratios of 0.95 ± 0.08 and 0.91 ± 0.04, respectively, at 16 °C. At 25 °C, P. eos and their coexisting hybrids had mean recovery ratios of 0.87 ± 0.03 and 0.84 ± 0.05, respectively, and mean recovery ratios for P. neogaeus and their coexisting hybrids were 0.87 ± 0.11 and 0.75 ± 0.05, respectively. To explore the potential for species bias in behavioural motivation (or willingness to perform), I analyzed contingency tables to compare the proportion of individuals of each species for which repeat swim performance was greater than initial performance (i.e., recovery ratio > 1). For the P. eos lakes, six of the 31 P. eos and six of the 39 hybrids had recovery ratios > 1 at 16 °C ("2 = 0.01, p = 0.9), and seven of the 31 P. eos and seven of the 39 hybrids had recovery ratios > 1 at 25 °C ("2 = 0.03, p = 0.9). For the P. neogaeus lake, two of the 6 P. neogaeus and five of the 24 hybrids had recovery ratios > 1 at 16 °C (Fisher’s exact test, odds ratio = 1.9, p = 0.6), and one of the 6 P. neogaeus and four of the 24 hybrids had recovery ratios > 1 at 25 °C (Fisher’s exact test, odds ratio = 1, p = 1). There was, therefore, no detectable species bias in behavioural motivation. Linear mixed effects models revealed that there was a significant negative effect of both temperature and repeat swimming on Umax in the P. eos-hybrid comparison and in the P. neogaeus-hybrid comparison (Table 5.1, figure 5.2). A significant interaction 72  between temperature and repeat swimming was detected only for the P. neogaeus-hybrid comparison such that repeat swimming had a more negative effect at the higher temperature; this was not the case for the P. eos-hybrid comparison (Table 5.1, figure 5.2). I detected no significant difference between species in the P. eos-hybrid comparison, but I found a significant interaction between species and temperature in the P. neogaeus-hybrid comparison such that temperature had a greater effect on hybrid than on P. neogaeus (Table 5.1, figure 5.2). A similar pattern was evident in the comparison of Umax between species under the most stressful conditions as a proportion of Umax under the least stressful conditions (figure 5.3). The combination of repeat swimming and increased temperature had the same effect for both species in the P. eos-hybrid comparison (df = 66, F = 0.7, p = 0.404; figure 5.3), but hybrids were more adversely affected than P. neogaeus in Alford Lake (df = 28, F = 6.6, p = 0.016; figure 5.3).  H1H1(=2#'7##2.0( A comparison of repeat swimming performance revealed very little difference between asexual hybrid dace and their sexually reproducing parental species, although there was some evidence that asexual hybrids performed worse relative to P. neogaeus. Asexual hybrid dace performed significantly worse than coexisting P. neogaeus in a repeat swim following an acute increase in temperature (figures 5.2 and 5.3). There was also a trend, although not statistically significant, of reduced repeat swimming performance of asexual hybrids relative to coexisting P. eos (figure 5.3). My study, while perhaps suggestive of lower repeat swimming performance of hybrids relative to parentals, strongly suggests that, at a minimum, hybrid dace possess no advantages over parental species in repeat swimming performance. Results from previous studies comparing swimming performance of hybrid and parental fishes are mixed with regards to evidence for hybrid vigour or hybrid disadvantage. Rouleau et al. (2010) compared Ucrit at acclimation temperature (10, 15, or 20 °C) of hybrid and pure littoral and pelagic brook trout (Salvelinus fontinalis) and found that hybrids performed worse at all temperatures. In a comparison of Ucrit at acclimation temperature (21 to 29 °C), Rosenfield et al. (2004) showed that swimming performance of hybrids between pecos pupfish (Cyprinodon pecosensis) and sheepshead 73  minnows (C. variegatus) was intermediate between the performance of the parental species. All species of Phoxinus showed, on average, reduced Umax when water temperature was increased above acclimation temperature to a value likely approaching an upper temperature limit. Although an upper lethal temperature was not determined in this study, fish could not consistently swim at temperatures higher than 25 °C. An overall reduction in swimming performance at higher temperature is expected if swimming had been fueled aerobically - an elevation in temperature causes a mismatch between water oxygen solubility and temperature-driven increase in oxygen demand (Randall & Brauner 1991). Anaerobically-fuelled swimming may also be affected when water approaches lethal temperatures, but data on the effect of temperature on the anaerobic component of swimming performance are lacking. My study provides some evidence that repeat Umax is reduced by acute changes in temperature, but the mechanism of the effect (i.e., whether it involves aerobic or anaerobic processes) is not known. Phoxinus neogaeus and the hybrid Phoxinus from Alford Lake performed worse in a repeat swim at elevated temperature than at acclimation temperature. This was not, however, the case for Phoxinus populations from the other lakes. To my knowledge, there are no other studies of the effects of acute temperature changes on repeat swimming performance in fish. Some studies have shown that recovery from exhaustion occurs faster at higher temperatures in salmonids (Kieffer et al. 1994; Wilkie et al. 1997; Kieffer 2000), and one study of repeat swimming performance in rainbow trout (Oncorhynchus mykiss) showed that recovery ratios decreased at warmer temperatures (Jain & Farrell 2003), although fish in all these studies were only tested at temperatures to which they were acclimated. Jain & Farrell (2003) suggested that the reduced recovery ratios at warmer temperatures occurred because fish swim faster and harder on their first swim and hence were unable to recover as quickly compared to repeat swimming at colder temperature where the fish were less active and did not swim as hard on their first swim. This hypothesis, however, cannot apply to the decrease in recovery ratio at higher temperature in the case of P. neogaeus and hybrid Phoxinus from Alford Lake because Umax1 was lower at the higher temperature than at the acclimation temperature and thus likely represents a direct effect of temperature on repeat 74  Umax. Given that this test likely has a large anaerobic component, a lower recovery ratio at higher temperature may indicate that a greater degree of glycogen depletion occurred following Umax1 at high temperature (Farrell 2008), but this depends on whether the fishes had recourse to anaerobic metabolic pathways during these tests. Further studies are required to address the mechanism of reduced performance at higher temperature in these tests. If hybrids have reduced overall physiological performance relative to the parental species, this might explain the greater effect of temperature on the repeat performance of hybrids relative to parentals in Alford Lake. Swimming ability is just one aspect of the ecological and physiological performance of fish that contributes to fitness. Additional measurements of other components are required to more fully characterize the relative fitness of asexual and sexual dace. One such measurement, conducted by Mee & Rowe (2006), showed that asexual dace carried heavier parasite loads than sexual dace. Further, Das & Nelson (1990) measured the fecundity of asexual and sexual dace and found no differences in this proxy measure of fitness. Taken together, these studies suggest a lack of hybrid vigour and possibly the existence of hybrid disadvantage among asexual dace. The lack of hybrid vigour, and especially the existence of hybrid disadvantage, would facilitate the coexistence of asexual and sexual dace by offsetting the advantages inherent to asexual reproduction (Moore 1976; Hellriegel & Reyer 2000; Mee & Otto 2010). It is likely that my test of a fish’s ability to repeat a high intensity swim following an acute temperature change is a good indicator of overall physiological performance (Jain et al. 1998). This could be tested directly by using methods similar to ours to compare repeat swimming performance among fish carrying various parasite loads or exposed to different levels of pollutants (McKenzie et al. 2007). Despite the lack of such confirmation, my data, which provide evidence that hybrid dace performed worse than at least one of the parental species, suggest that the hybrids do have reduced physiological performance. The use of physiological performance as a proxy for fitness might be conservative in terms of ecological relevance, especially if the previous evidence of increased parasite loads on hybrid fish (Mee & Rowe 2006) is considered. These two measures of ecological and physiological performance, both reduced in hybrid dace, suggest that a multitude of factors, perhaps acting synergistically, could result in much 75  greater performance of parental species relative to hybrid dace under natural conditions. A combination of multiple factors, and multifarious selection, has been shown to decrease hybrid fitness in other systems. For example, hybrid stickleback (Gasterosteus aculeatus) have been shown to grow more slowly (Hatfield & Schluter 1999) and have lower juvenile survival (Vamosi et al. 2000) than pure benthic or limnetic species. Similarly, hybrid whitefish (Coregonus clupeaformis) have increased embryonic mortality and more asynchronous emergence relative to pure dwarf or normal ecotypes (Rogers & Bernatchez 2006). Hybrids between Drosophila gaucha and D. pavani are not only sterile, but have reduced larval viability and inferior foraging and locomotory behaviour (Brncic & Budnik 1974; Godoy-Herrera et al. 2005). All of these hybrids differ from hybrid Phoxinus in that they were produced from contemporary or laboratory hybridization, whereas hybrid Phoxinus are descendants from historical hybridization events (Angers & Schlosser 2007), but the importance of multiple factors in reducing hybrid fitness is nonetheless a valid consideration. The fitness of asexuals and sexuals has been compared in other asexual vertebrate systems, and the evidence is mixed with regards to the existence of hybrid disadvantage or hybrid vigour. Hybrid disadvantage has been detected in a number of asexual systems including geckos (Moritz et al. 1991), Cnemidophorus lizards (Cullum 1997), European wood frogs (Vorburger 2001), and poeciliid fish (Wetherington et al. 1987; Lively et al. 1990; Weeks 1995), while hybrid vigour has been detected in asexual geckos (Kearney et al. 2005) and asexual European wood frogs (Hotz et al. 1999). Additional studies comparing fitness-related traits between asexual and sexual species in dace and other hybrid asexual systems will continue to provide information on the balance of advantages and disadvantages of sexual and asexual reproduction. Such studies are needed to further resolve the reasons for the rarity of asexual reproduction among vertebrates.  76  Table 5.1. Model summary and ANOVA (analysis of variance) results for linear mixed effects models of the effects of repeat swimming and temperature on swimming performance compared between populations of hybrid and parental Phoxinus. The organization of parameters in the models (as either fixed or random effects) is indicated in the top rows, and the effects (either alone or in an interaction) of each parameter with a fixed effect are evaluated in the bottom rows.  Model fixed effects random effects1  Comparison between Phoxinus eos and hybrid populations from Pondweed Lake, Sunday Creek, and Strubel Lake  Comparison between Phoxinus neogaeus and hybrid populations from Alford Lake  species x temperature x repeat lake/individual/temperature/repeat  species x temperature x repeat individual/temperature/repeat  Main effects species temperature repeat  df 66 68 68  F-value 0.95 9.34 101.19  p-value 0.333 0.003 < 0.001  df 28 28 28  F-value 0.88 9.88 29.70  p-value 0.357 0.004 < 0.001  Interaction effects sp. x temp. sp. x rep. temp. x rep. sp. x temp. x rep.  68 68 136 136  2.71 0.11 1.00 1.06  0.104 0.741 0.319 0.306  28 28 56 56  8.23 1.24 6.67 0.39  0.008 0.276 0.013 0.537  1  Forward slashes separate nested effects  77  Figure 5.1. Relationship between size (fork length, cm) and Umax (body lengths per second) for Phoxinus eos and coexisting hybrids (panel A) and for P. neogaeus and coexisting hybrids (panel B). In both panels, hybrids are shown in black, and parentals (P. eos or P. neogaeus) are shown in red. Number characters are used to plot the measurements, and these refer to the stages (not necessarily chronological; see main text) of the experimental procedure: 1) Umax1 at 16 °C, 2) Umax2 at 16 °C, 3) Umax1 at 25 °C, 4) Umax2 at 25 °C. Estimates of the linear and non-linear relationships (see main text) are represented by solid and dashed lines, respectively.  78  Figure 5.2. Effects of repeat swimming and temperature on swimming performance (Umax) of hybrid and parental Phoxinus from four lakes. Each line connects measurements of Umax at the four stages of the experiment for a single individual (not necessarily chronologically; see methods in section 5.3.2). Black lines represent hybrids and red lines represent parental species. Individuals from the three lakes represented separately to the left of the vertical dashed line are represented again together immediately to the right of the vertical dashed line. Bold-type labels indicate statistically significant main effects (p < 0.05). Shared superscript symbols (either * or ^) indicate a statistically significant interaction (p < 0.05).  79  Figure 5.3. Swimming performance of hybrid and parental Phoxinus under the most challenging conditions (Umax2 at 25 °C) as a proportion of performance under the least challenging conditions (Umax1 at 16 °C). Horizontal bars, boxes, and whiskers give median values, 2nd quartiles, and 3rd quartiles, respectively, of the data for hybrid and parental Phoxinus populations. The width of boxes is proportional to the samples size in each population, and the sample size is given below each box. Populations from the three lakes represented separately to the left of the vertical dashed line are pooled in the panel immediately to the right of the vertical dashed line. A significant difference (p < 0.05) between populations of hybrid and parental Phoxinus (eos = P. eos population; hyb = hybrid population; neo = P. neogaeus population) is indicated by an asterisk.  80  V1(5*'70:2$A(&0:(6%.9$F(.+(0.%$F*%0(%*:"*--A(:&'*(P!"#$%&'() *#(QW(+20*#'&-*(:&'*(P!"#$%&'()&*#,-*'(QW(&0:($F*2%(702#*<7&-( FA"%2:#( ( V1>1(@7;;&%A( Understanding the mechanism of coexistence of sperm-dependent asexuals and their sexual hosts requires an empirical evaluation of their relative fitnesses. In this chapter, I evaluated the fecundity, egg viability, and hatchling growth rate of the spermdependent asexual hybrid species Phoxinus eos-neogaeus and its sexually-reproducing parental species, P. eos and P. neogaeus. Three comparisons between asexual and sexual Phoxinus suggested a fitness advantage in the asexuals: absolute fecundity was higher in P. eos-neogaeus than P. eos, but this advantage disappeared when fecundity differences were scaled to differences in body size; a higher fraction of P. eos-neogaeus eggs than P. eos eggs were viable post-fertilization, but this advantage was slight (and not statistically significant); Phoxinus eos-neogaeus grew faster in the first 60 days post-hatching than both their P. eos and P. neogaeus hosts. All other comparisons showed no difference in fitness between P. eos-neogaeus and its parental species. Overall, my data suggest a potentially weak fitness advantage for the asexuals, which highlights the importance of male preference for sexual mates in maintaining asexual-sexual coexistence.  V1D1(?0$%.:7'$2.0( Sperm-dependent asexual reproduction occurs in several phyla, including rotifers, platyhelminths, nematodes, annelids, mollusks, arthropods, as well as in some vertebrates (Beukeboom & Vrijenhoek 1998). In vertebrates, all asexual species are of hybrid origin and the majority of asexual species are sperm-dependent. There are two general mechanisms by which the continued coexistence of sperm-dependent asexual hybrids and sexually-reproducing parental species could break down. First, the sexually-reproducing 81  parental species may have a strong preference for conspecific mating, thereby preventing the hybrid from reproducing. Second, the hybrid may gain a competitive advantage over the parental species due to hybrid vigour or because the hybrid does not produce males and hence does not pay the cost of sex, and the parental species may be driven to such low abundance that there are not enough males to provide sperm for either hybrid or parental reproduction. Several models of sperm-dependent asexual systems have shown that coexistence would break down via the first mechanism if male preference for conspecific mates is too strong and would break down via the second mechanism if the hybrids gain too much of a fitness advantage from the cost of sex or from hybrid vigour (Moore 1976; Hellriegel & Reyer 2000; Mee & Otto 2010). A general explanation for the persistence of sperm-dependent asexuals, which are an important aspect of the diversity of asexual species, depends on an empirical evaluation of mate choice and fitness in systems with sperm-depend asexual species. The processes involved in maintaining the asexual-sexual coexistence in most spermdependent asexual systems have not been empirically evaluated, but preference for conspecific females has been detected in some sperm-dependent asexual systems (Moore & McKay 1971; Abt & Reyer 1993; Schlupp et al. 1998; Engeler & Reyer 2001; Dries 2003), and not in others (Balsano et al. 1981; Balsano et al. 1985; Woodhead & Armstrong 1985; Schlupp et al. 1991). Similarly, hybrid vigour has been detected in some sperm-dependent asexual systems (Hotz et al. 1999; Kearney et al. 2005), and hybrid disadvantage has been detected in others (Wetherington et al. 1987; Lively et al. 1990; Moritz et al. 1991; Weeks 1995; Cullum 1997; Vorburger 2001; Mee & Rowe 2006; Mee et al. 2011). Northern redbelly dace (Phoxinus eos) and finescale dace (P. neogaeus) coexist with their sperm-dependent asexual hybrids across a wide distribution in North America. In the Phoxinus system, there is no data on males’ preference for conspecific females, but some data suggest that the unisexual hybrids are at a fitness disadvantage compared to the parental species in susceptibility to parasites (Mee & Rowe 2006) and in swimming performance (Mee et al. 2011). Fitness is difficult to measure, and any single fitnesscorrelated trait (e.g., either susceptibility to parasites or physiological performance) is unlikely to give an accurate estimate of actual ecological fitness (Gerstein & Otto in 82  press). In the present study, I measured additional fitness-correlated traits (fecundity, egg viability, and growth) in Phoxinus hybrids and parentals to test for hybrid vigour or hybrid disadvantage. Given the results of the previous comparisons of fitness-related traits between Phoxinus hybrids and parentals (Mee & Rowe 2006; Mee et al. 2011), as well as the predictions of the theoretical studies discussed previously (Moore 1976; Hellriegel & Reyer 2000; Mee & Otto 2010), I hypothesized that hybrids would be at a disadvantage in terms of fecundity and growth, as this would help stabilize their coexistence with sexuals.  V1C1(M&$*%2&-#(&0:(M*$F.:#( V1C1>1(5*'70:2$A( Adult P. neogaeus, P. eos, and hybrids were caught using minnow traps and dip nets from five lakes in southwestern Alberta in June of 2009 and 2010. Three of these lakes (Fiesta Lake, Gunrange Lake, and Mitchell Lake) contain P. eos and hybrids at high frequency, with P. neogaeus at low frequency and rarely used as a sperm source. The remaining two lakes (Alford Lake and Tay Lake) contain only P. neogaeus and hybrids. Henceforth, I will refer to the former three lakes as “type 1 lakes”, and the latter two lakes as “type 2 lakes”. Eggs were harvested from fully gravid females (i.e., females with distended abdomens whose eggs flow out spontaneously with gentle handling) by applying light manual pressure to the abdomen. The eggs were extruded into 20 mL glass scintillation jars and fixed in 95% ethanol. The jars were shaken vigorously immediately after the eggs were fixed in ethanol, and the jars were shaken several more times over the subsequent 48 or so hours to prevent clumping of eggs during hardening. Each adult female was euthanized by immersion in an MS222 solution, measured, and stored in 95% ethanol. The egg samples were left at room temperature for 2 to 4 weeks, after which each egg sample (including ethanol) was decanted into a Petri dish for counting using an analog click-counter and a dissecting microscope at 12x magnification. If immature eggs were present (always at very low frequency relative to the much larger mature eggs), only the mature eggs were counted. After counting, the ethanol was drained from the dish, and the eggs were quickly transferred to a jar containing a known 83  weight of ethanol. The eggs were then weighed on a precision scale (to the nearest 0.1 mg). Fecundity comparisons between Phoxinus hybrids and parental species were conducted separately for type 1 and type 2 lakes. A maximum of three triploids were found among the hybrids from any given lake (data not shown), and the effect of ploidy on fecundity could not reliably be determined. Hence, any triploid hybrids were removed from the dataset. Fecundity should vary linearly with the volume of the female abdomen. Hence, I used a linear mixed effects model of egg number on length3, and used analysis of covariance to test for a species effect (with individual females as replicates). Lake and year of sampling were included in the linear model as random effects. Pairwise differences in egg number between species were evaluated using Tukey’s test. I also analyzed fecundity differences between species using body weight as a covariate instead of length3, and the results were not qualitatively different (results not shown). In order to compare egg size between species, egg clutch weight was divided by number of eggs to give a measure of weight per egg. I used a linear mixed effects model and analysis of variance to test for an effect of species on egg size (i.e., weight per egg), including lake and year as random effects. All statistical analyses were performed using the R programming language (R Development Core Team 2010). V1C1D1(/%.##*#(&0:(I2&"2-2$A( In 2010, fully ripe females and males from Fiesta Lake (type 1), Gunrange Lake (type 1), and Alford Lake (type 2) were transported live to a nearby fish rearing facility in order to conduct crosses. From June 7th to 9th, a total of twenty females and twenty-four males were used to make forty-two crosses (table 6.1). Each cross was conducted in a single 100 mm diameter Petri dish. The eggs from a female were divided between two or three clean 100 mm diameter Petri dishes. Both testes were dissected out of a euthanized sexually mature male, and each testis from a single male was placed in a different Petri dish with a different sample of eggs. The testes were broken up using forceps, and, using the same forceps, the sperm and eggs were stirred briefly (forceps were rinsed in ethanol between crosses). After five to ten minutes, water was added to the Petri dish. Temperature at the rearing facility ranged daily between 12 °C and 20 °C during the eggrearing period. The water in the Petri dishes was changed using a plastic pipette twice 84  daily. Dead eggs were not removed from the dishes. No eggs died after the second day post-fertilization. On June 12th, a digital photo was taken of each Petri dish, and the number of dead and viable eggs in each cross was counted in these digital images (viable eggs are translucent yellow or orange, while dead eggs are opaque milky white). For the crosses among fish from the type 1 lakes, I used analysis of variance (with individual crosses as replicates) to test for differences between Phoxinus hybrids and parental species in the proportion of a mother’s eggs that were viable. No such test was possible for type 2 lakes because only one pure P. neogaeus cross from a type 2 lake was acquired (cross #22, table 6.1). Pairwise differences in viability between species in type 1 lakes were evaluated using Tukey’s test. V1C1C1(X%.9$F( On June 14th, 2010, eggs from several crosses (table 6.1) were combined to give a roughly equal proportion of P. neogaeus, P. eos, and hybrids in each of six combinations (table 6.2). Each of these combinations (each containing three crosses) was transferred into a one-gallon jar of water and transported to an environmental chamber at the University of British Columbia. Upon arrival, on June 15th, the combinations in each jar were transferred to one of six 25 gallon tanks. The water in these tanks was maintained at 16 °C, and the environmental chamber was set to a 12:12 hour light:dark cycle. Hatchlings began to emerge shortly after being released into the tanks on June 15th. The hatchlings were fed a diet of brine shrimp (Artemia sp.) and microworms (Panagrellus redivivus) twice daily (once in the morning, and once in the afternoon) ad libitum. Hatchlings were destructively sampled at 30 days and again at 60 days. At each sampling, the standard length (SL) and weight of each fish sampled was recorded in order to compare growth between hybrids and parental Phoxinus species. I used a linear model and analysis of variance to evaluate the effects of species and age (i.e., days posthatching) on hatchling length. For hatchlings from crosses between parents from Fiesta Lake (type 1), the rearing combination (i.e., tank; see table 6.2) was included in the model as a random effect, and pairwise differences in size between species were evaluated using Tukey’s test. For hatchlings from crosses between parents from Alford Lake (type 2), samples of both P. neogaeus and hybrids at both 30 and 60 days posthatching were obtained from only one tank (tank F, see table 6.2), which prevented the 85  inclusion of rearing combination (i.e., tank) in the statistical model. Hence, I included only hatchlings from tank F for the linear model for crosses from Alford lake (with individual hatchlings as replicates). V1C1O1(@K*'2*#(&0:()-.2:A(?:*0$2+2'&$2.0( I differentiated Phoxinus hybrids and parental species in my samples using DNA identification methods that were developed, assessed, and validated previously (Binet & Angers 2005; Angers & Schlosser 2007). In brief, I first extracted DNA from each fish sampled in my study (including each hatchling in the growth comparison) using a standard phenol-chloroform DNA isolation protocol for fish tissues. Species were then differentiated based on an intron of the mesoderm specific transcript gene (Binet & Angers 2005). Fragments of DNA were amplified by PCR, separated using electrophoresis on 2% agarose gels, and visualized on a UV light table after staining with SYBR® Safe DNA gel stain. For all wild-caught hybrids (i.e., those used to make crosses and those sampled for egg counting), I also assessed ploidy. I genotyped these hybrids at four highly variable microsatellite DNA loci that are specific to P. eos and are therefore hemizygous in the hybrids. Diploid hybrids can only have a single allele, and triploid hybrids can have up to two alleles. Any hybrid that did not have two alleles at any locus was assumed to be a diploid hybrid because, given the diversity of alleles at each locus (13 to 26 alleles per locus), the probability of finding a triploid homozygous at all four loci was extremely low (approximately 0.00011, assuming that the diversity of alleles in the sexual populations, which is the source of the third genome in the triploids, was similar to that in the asexual populations, and that all alleles are at equal frequency in the sexual population).  V1O1(E*#7-$#( V1O1>1(5*'70:2$A( There were substantial differences between Phoxinus species in fecundity that reflected differences between the species in body size (figure 6.1). In type 1 lakes, hybrids were intermediate in mean standard length (64.1 mm) between P. eos (52.2 mm)  86  and P. neogaeus (74.6 mm), and hybrids were intermediate in mean fecundity (367.5 eggs) between P. eos (248.3 eggs) and P. neogaeus (714.1 eggs). In type 2 lakes, hybrids were smaller (71.2 mm) than P. neogaeus (78.0 mm), and hybrids had lower mean fecundity (434.0 eggs) than P. neogaeus (530.1). In type 1 lakes, there was a significant length-independent effect of species on fecundity (df = 6, F = 8.11, p = 0.0197). After correcting for size differences, Tukey’s test of pairwise differences showed a significant fecundity difference between P. neogaeus and hybrids (z = 3.607, p < 0.001), but not between P. neogaeus and P. eos (z = 2.114, p = 0.0784) nor between P. eos and hybrids (z = -1.414, p = 0.3117) (figure 6.1). There was no length-independent fecundity difference between P. neogaeus and hybrids in type 2 lakes (df = 1, F = 2.87, p = 0.3396) (figure 6.1). There were no significant difference between species in egg size (i.e., weight per egg) in either type 1 (df= 6, F = 4.23, p = 0.0714) or type 2 lakes (df = 1, F = 0.32, p = 0.674) (figure 6.2). V1O1D1(I2&"2-2$A( There were significant differences in egg viability among species in Fiesta Lake (df = 10, F = 5.3143, p = 0.02677) (figure 6.3). Pairwise tests showed that P. neogaeus eggs were significantly more viable than P. eos eggs (t = 3.148, p = 0.251), but there was no significant viability difference between P. eos and hybrid eggs (t = 1.858, p = 0.1981) nor between P. neogaeus and hybrid eggs (t = 1.728, p = 0.2396) (figure 6.3). I also compared the viability of eggs from different mother species fertilized by the same father (table 6.3), as well as the viability of eggs from a single mother fertilized by different father species (table 6.4). This allowed species effects to be evaluated while controlling for sire or dam effects, respectively, although sample size in these comparisons (see tables 6.3 and 6.4) was too low for a meaningful statistical comparison. Hybrid females tended to have lower viability eggs than P. neogaeus females fertilized by the same father (table 6.3). Hybrid females tended to have higher viability eggs than P. eos females fertilized by the same father (table 6.3). These observations parallel the trend in the statistical analysis of egg viability in Fiesta Lake, above (see figure 6.3). Interestingly, P. eos eggs were less viable when fertilized by P. neogaeus males than P. eos males, while P. neogaeus eggs were less viable when fertilized by P. eos males than P. neogaeus males (paired t-test: t = 4.3034, df = 4, p = 0.01261; table 6.4), possibly indicating a form 87  of post-zygotic reproductive isolation. Similarly, P. neogaeus males had substantially lower success fertilizing P. eos eggs than P. neogaeus eggs, but there was no consistent difference in the success of P. eos males fertilizing P. neogaeus versus P. eos eggs (paired t-test: t = 1.2984, df = 3, p-value = 0.2849; table 6.3). V1O1C1(X%.9$F( For hatchlings from crosses between parents from Fiesta Lake (type 1), there were significant differences in growth rate among species (df = 5, F = 28.99, p = 0.0018; figure 6.4). Hybrids tended to be intermediate in growth rate between P. eos and P. neogaeus, although a significant difference was only detected between P. eos and hybrids (z = 4.568, p <0.001) and P. eos and P. neogaeus (z = 5.166, p < 0.001), and not between hybrids and P. neogaeus (z = 0.03, p = 1.000). For hatchlings from crosses between parents from Alford Lake (type 2), hybrids grew significantly faster than P. neogaeus (df = 38, F = 23.80, p < 0.001).  V1H1(=2#'7##2.0( Hybrid Phoxinus are intermediate in body size between their parental species (P. eos and P. neogaeus). Commensurate with these size differences between hybrid and parental Phoxinus species, fecundity was found to be intermediate in hybrids. When the effect of size was removed, however, the fecundity of hybrids was not intermediate between the parental species. Instead, hybrid Phoxinus were equivalent in fecundity to the species upon which they depend for sperm. For example, in lakes where P. neogaeus is rare or absent (i.e., type 1 lakes), hybrids are equivalent to P. eos females in fecundity but differ significantly from P. neogaeus females (figure 6.1). Conversely, in lakes where P. eos is absent (i.e., type 2 lakes), hybrids are equivalent to P. neogaeus females in fecundity (figure 6.1). Egg viability was significantly lower for P. eos than P. neogaeus eggs, and was intermediate for hybrid eggs (figure 6.3, tables 6.3 and 6.4). This difference in egg viability between P. eos and P. neogaeus was not predicted, but it may result from the small (but not statistically significant) difference between the median size of P. eos and P. neogaeus eggs (figure 6.2).  88  Das & Nelson (1990) also compared the fecundity of hybrid and parental Phoxinus and found no significant differences. Their study, however, focused on only a single lake in Alberta, and they measured fecundity without distinguishing between mature and immature ova (both of which were present in many specimens). My study is an improvement on that of Das & Nelson (1990) in that I collected specimens from five lakes with different assemblages of species (i.e., in type 1 and type 2 lakes), and I considered only mature ova from fully mature females in order to provide a more meaningful comparison between species. Das & Nelson (1990) found evidence for prezygotic reproductive isolation between P. eos and P. neogaeus in the form of different spawning times. My study also provides evidence for post-zygotic reproductive isolation in the form of reduced egg viability from heterospecific crosses compared to conspecific crosses (tables 6.3 and 6.4). Hybrid Phoxinus hatchlings grew faster than the hatchlings of their host species. This was true for hatchlings from both type 1 and type 2 lakes (figure 6.4), and is contrary to my prediction given that growth rate in fish, especially in early life stages, is typically assumed to be an important component of fitness (e.g., Sogard 1997; Shima & Findlay 2002; Berkeley et al. 2004; Morita & Nagasawa 2010). The faster growth of hybrids compared to P. eos in type 1 lakes may simply reflect the fact that hybrids are larger than P. eos at maturity. The faster growth of hybrids than P. neogaeus in type 2 lakes, however, does not parallel the difference in size at maturity between hybrids and P. neogaeus. Overall, the comparisons between hybrid and parental Phoxinus in fitness-related traits in this study (i.e., fecundity and growth) are variable in terms of relative hybrid fitness. If one’s null expectation is for hybrids to be intermediate between their parental species, then the slightly lower (though not significantly different) size-independent fecundity of hybrids compared to P. eos in type 1 lakes (figure 6.1) may indicate hybrid disadvantage, while the lack of a significant size-independent fecundity difference between hybrids and P. neogaeus in type 2 lakes (figure 6.1) may indicate hybrid vigour. Hybrid vigour was also evident in the growth comparison. Hybrids were, however, intermediate between the parental species in absolute fecundity (i.e., before correcting for size) and egg viability. 89  While the results of this study were variable in terms of an overall trend in the fitness comparison between hybrid and parental Phoxinus, these results allow novel predictions regarding the strength of mate preference for conspecific females among male P. eos and P. neogaeus. Mate preference for conspecific females is required for coexistence between sexuals and sperm-dependent asexuals (Moore 1976; Hellriegel & Reyer 2000; Heubel et al. 2009; Mee & Otto 2010). If mate preference is too strong (i.e., if the sperm-dependent asexuals are too sperm-limited), then the asexuals will not persist. The strength of mate preference required to maintain coexistence depends on the relative fitness of sexuals and sperm-dependent asexuals. For example, if the asexuals benefit greatly from hybrid vigour and the sexuals pay the cost of sex, then preference for conspecific mates must be very strong in order to allow coexistence. The measures of fitness-related traits in the present study could be used to set parameter values in mathematical models designed to predict the strength of mate preference required to maintain coexistence between hybrid and parental Phoxinus. This predicted strength of preference would be valuable in designing a study of mate preference in Phoxinus (e.g., for predicting effect size). Comparing the strength of preference for conspecific mates between type1 and type 2 lakes, which differ in the relative fitness of hybrid Phoxinus and their host parental species, would be a fascinating contribution to understanding the factors that promote asexual-sexual coexistence.  90  Table 6.1. Crosses between P. eos, P. neogaeus, and hybrids from Fiesta Lake (FIE), Gunrange Lake (GUN), and Alford Lake (ALF). Cross ID 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42  Male ID FIE10M01 ALF10M01 ALF10M01 ALF10M02 ALF10M02 FIE10M02 FIE10M03 ALF10M03 ALF10M03 FIE10M04 ALF10M04 ALF10M04 FIE10M05 ALF10M05 ALF10M05 FIE10M06 FIE10M06 FIE10M07 FIE10M08 FIE10M08 ALF10M06 ALF10M06 FIE10M09 FIE10M09 FIE10M10 FIE10M10 FIE10M11 FIE10M11 FIE10M12 FIE10M12 FIE10M13 FIE10M13 FIE10M14 FIE10M14 FIE10M15 FIE10M15 FIE10M16 FIE10M16 FIE10M17 FIE10M17 FIE10M18 FIE10M18  Male sp. P. eos P. neogaeus P. neogaeus P. neogaeus P. neogaeus P. eos P. eos P. neogaeus P. neogaeus P. eos P. neogaeus P. neogaeus P. eos P. neogaeus P. neogaeus P. eos P. eos P. neogaeus P. eos P. eos P. neogaeus P. neogaeus P. eos P. eos P. neogaeus P. neogaeus P. eos P. eos P. neogaeus P. neogaeus P. eos P. eos P. neogaeus P. neogaeus P. eos P. eos P. neogaeus P. neogaeus P. eos P. eos P. eos P. eos  Female ID ALF10F01 ALF10F01 ALF10F01 ALF10F02 ALF10F02 ALF10F02 ALF10F03 ALF10F03 ALF10F03 GUN10F01 GUN10F01 GUN10F01 ALF10F04 ALF10F04 ALF10F04 ALF10F05 ALF10F05 ALF10F05 ALF10F06 ALF10F07 ALF10F06 ALF10F07 FIE10F01 FIE10F02 FIE10F01 FIE10F02 FIE10F03 FIE10F04 FIE10F03 FIE10F04 FIE10F05 FIE10F06 FIE10F05 FIE10F06 FIE10F07 FIE10F08 FIE10F07 FIE10F08 FIE10F09 FIE10F10 FIE10F11 FIE10F12  Female sp. hybrid hybrid hybrid hybrid hybrid hybrid hybrid hybrid hybrid hybrid hybrid hybrid hybrid hybrid hybrid hybrid hybrid hybrid hybrid P. neogaeus hybrid P. neogaeus P. neogaeus P. eos P. neogaeus P. eos P. neogaeus P. eos P. neogaeus P. eos hybrid hybrid hybrid hybrid hybrid P. eos hybrid P. eos P. eos hybrid P. eos P. eos  Female ploidy 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n  Dead eggs 6 2 1 45 92 5 6 24 44 22 252 208 24 4 4 15 9 8 7 25 7 1 13 28 2 46 53 15 10 60 46 38 13 13 6 24 6 47 27 29 14 50  Viable eggs 115 67 72 1 0 76 52 3 1 201 4 1 102 153 100 80 76 83 77 96 74 163 109 79 264 28 237 69 185 45 149 172 257 90 89 64 88 32 80 271 187 136  91  Table 6.2. Identity of crosses combined in six tanks (F through K) for growth comparisons between Phoxinus species, and sample sizes for hatchlings in each combination. Details for each cross are listed in table 6.1. Bold text indicates crosses between individuals from Alford Lake. All other crosses included in these comparisons were between individuals from Fiesta Lake. Tank F G H I J K  P. eos (E) cross ID 24 28 36 39 41 36  P. neogaeus (N) cross ID 22 22 25 25 29 29  Hybrid (H) cross ID 21 32 14 35 15 40  Hatchlings sampled at 30 days 15 (0E, 9N, 6H) 20 (8E, 9N, 3H) 11 (3E, 5N, 3H) 10 (2E, 5N, 3H) 10 (3E, 1N, 6H) 0  Hatchlings sampled at 60 days 37 (10E, 12N, 15H) 49 (9E, 14N, 26H) 19 (5E, 7N, 7H) 16 (5E, 3N, 8H) 36 (19E, 11N, 6H) 25 (5E, 10N, 10H)  Total 52 69 30 26 46 25  92  Table 6.3. Proportion of viable eggs from P. eos, P. neogaeus, and hybrid females fertilized by either P. eos or P. neogaeus males. The values shown are for the subset of cases in which a single male fertilized the eggs of two different female species, thereby allowing a comparison between the fertilization success of the female species while controlling for sire effects. In type 1 lakes, hybrids use P. eos sperm. In type 2 lakes, hybrids use P. neogaeus sperm.  Male’s lake type 1 1 1 1 1 1 1  Male Species P. eos P. eos P. eos P. eos P. neogaeus P. neogaeus P. neogaeus  P. eos 0.738 0.821 0.727 0.748 0.378 0.429 0.405  2  P. neogaeus  -  Female Species Hybrid P. neogaeus 0.893 0.817 0.937 0.903 0.992 0.949 0.936 0.914  0.994  93  Table 6.4. Proportion of viable eggs from P. eos, P. neogaeus, and hybrid females fertilized by both P. eos and P. neogaeus males. The values shown are for the subset of crosses in which a single female’s eggs were fertilized by both male species, thereby allowing a comparison between the fertilization success of the male species while controlling for dam effects. In type 1 lakes, hybrids use P. eos sperm. In type 2 lakes, hybrids use P. neogaeus sperm. Female’s lake type 1 1 1 1 1 1 1 1  Female Species P. eos P. eos P. eos Hybrid Hybrid Hybrid P. neogaeus P. neogaeus  Male Species P. eos P. neogaeus 0.738 0.378 0.821 0.429 0.727 0.405 0.764 0.952 0.819 0.874 0.937 0.936 0.893 0.992 0.817 0.949  2 2 2 2 2  Hybrid Hybrid Hybrid Hybrid P. neogaeus  0.9501 0.8101 0.8671 0.9171 0.7931  1  0.979 0.969 0.9121 0.914 0.994  In these cases, there is a mismatch in lake type between the male's and female's lake of origin.  94  Figure 6.1. Fecundity of P. eos, P. neogaeus, and sperm-dependent asexual hybrids in lakes where the hybrids use P. eos sperm (type 1) and in lakes where the hybrids use P. neogaeus sperm (type 2). Significant differences (p < 0.05) between species in the intercepts of the linear relationships of fecundity and size are indicated by different lower case letters at the base of the lines depicting the relationships. There is no significant difference between species in the intercepts in type 2 lakes (the lines are present only for illustrative purposes).  95  Figure 6.2. Weight per egg for P. eos, P. neogaeus, and sperm-dependent asexual hybrids in lakes where the hybrids use P. eos sperm (type 1) and in lakes where the hybrids use P. neogaeus sperm (type 2). Dark bars represent median scores, the boxes encompass the scores between the 1st and 3rd quartile, and the whiskers extend 1.5 times the interquartile range from the box. Differences were not significant.  96  Figure 6.3. Proprotion of fertilized eggs which were viable for P. eos, P. neogaeus, and sperm-dependent asexual hybrids in lakes where the hybrids use P. eos sperm (type 1 lakes). Significant differences (p < 0.05) between species are indicated by different lower case letters at the bottom of the figure. Dark bars represent median scores, the boxes encompass the scores between the 1st and 3rd quartile, and the whiskers extend 1.5 times the interquartile range from the box.  97  Figure 6.4. Growth of P. eos, P. neogaeus, (solid lines) and sperm-dependent asexual hybrid (dashed lines) hatchlings from parents taken from Fiesta Lake (type 1) and Alford Lake (type 2). The point at zero days after hatching is for illustrative purposes only – the hatchlings were not measured at this early stage, but are, in general, approximately 2mm long upon hatching. Size was measured 30 and 60 days after hatching. The horizontal positions of some points at 30 days have been staggered for clearer presentation. Error bars depict ± one standard error.  98  L1(X*0*%&-(=2#'7##2.0( L1>1(?;K-2'&$2.0#( Hybridization is an important process in nature, although it is generally (or at least historically among those who study animals) considered a destructive force in terms of biodiversity (see discussion in Arnold 1997). Hybridization is also common in nature – for example, among six North American freshwater fish families, considering only sympatric taxa, hybridization occurs within 9% of genera (Hubbs 1955). A valuable approach to understanding the consequences and implications of hybridization in nature is to compare naturally occurring hybrids to their parental species. In the case of the widespread Phoxinus eos-neogaeus, instead of reducing biodiversity, hybridization has created a fascinating and rare example of an asexual vertebrate species, and the unusual reproductive processes in the Phoxinus system have further contributed to the generation of biodiversity by producing additional biotypes such as triploid hybrids, mosaic hybrids, and cybrids. Hybridization, in the case of Phoxinus eos-neogaeus, has also provided an opportunity to study the coexistence of asexual and sexual species, which has implications for understanding the evolution and maintenance of sexual reproduction. The body of theory on the evolution and maintenance of sex is very well developed, but empirical evidence lending support to any given theory is generally lacking. Asexually reproducing vertebrate species, such as hybrid Phoxinus, provide valuable opportunities to generate empirical evidence for particular theoretical benefits of sex. Although my work has focused specifically on hybrid Phoxinus and its parental species, the implications of my research apply more broadly to the origins and persistence of asexual species. My research has tested several novel and established hypotheses related to this topic and also created new hypotheses for the maintenance of sex in spermdependent asexual systems. My contributions to answering specific questions related to the maintenance of sex in sperm-dependent asexual systems are summarized below.  99  >1(/&0(;&$*(K%*+*%*0'*(#$&"-A(;&20$&20(".$F(#K*%;J:*K*0:*0$(&#*<7&-(&0:(#*<7&-( 20:232:7&-#(92$F20(&('.;;702$A\( Prior to my work on the role of mate choice in promoting stable coexistence of sexuals and sperm-dependent asexuals (see chapter 2), there were two hypotheses for how this stable coexistence is maintained (Schlupp 2005). First, niche separation hypotheses suggest that, while sperm dependence necessitates habitat overlap between asexuals and sexual species, there may be enough niche separation to reduce competition to a degree that allows coexistence (Vrijenhoek 1979a; Schley et al. 2004). Second, preference by sexual males for sexual females (or the effect of this preference) may become more intense as asexuals become more common (Moore 1976; Heubel et al. 2009). The resulting frequency-dependent selection against asexuals can limit the growth of asexual populations and allow coexistence. I have proposed a third hypothesis explaining stable coexistence of sexuals and sperm-dependent asexuals (see chapter 2). Stable coexistence can occur when there is variation among males in the degree of preference for mating with sexual females and when males with stronger preference pay a higher cost of preference. This variation in male mate preference causes a coevolutionary “rock-paper-scissors” dynamic to arise, where asexuals proliferate when males with weak preference are common, followed by the proliferation of males with stronger preference and the decline of the asexuals, followed by the decline in frequency of males bearing strong but costly preferences, leading the cycle to begin again. Variation in the strength of male mate preference has only been investigated in one sperm-dependent asexual system. Variation has been detected between populations of Poecilia latipinna, which act as sperm donors for the gynogenetic species Poecilia formosa (Gabor & Ryan 2001; Gabor et al. 2010). Variation in the strength of male mate preference also exists within Poecilia latipinna populations, but evidence suggests that this within population variation has low heritability (Gabor & Aspbury 2008). D1(S.9(.+$*0W(&0:(20(9F&$(#2$7&$2.0#W(:.(#K*%;J:*K*0:*0$(&#*<7&-#(&%2#*\(( In systems where sperm-dependent asexuals arise frequently, persistence of the asexuals will be influenced strongly by the frequent arrival of additional asexual individuals (thereby maintaining viable population sizes) and potentially also of  100  additional asexual genotypes (thereby augmenting the potential for adaptation to environmental changes). This mechanism of persistence most likely explains the coexistence of the unisexual hybridogenic species Poeciliopsis monacha-lucida with its parental species (Quattro et al. 1991). Vrijenhoek (1979a) proposed the “frozen niche variation” hypothesis to explain the success of P. monacha-lucida clones which arise frequently via hybridization, each surviving by momentarily capturing a specific subset of the parental species’ ecological niche. In contrast, in systems where sperm-dependent asexuals arise very rarely, persistence of the asexuals will depend on long-term coexistence with their sexuallyreproducing hosts. There are several examples of sperm-dependent asexual vertebrate lineages that are relatively old. A monophyletic lineage of the unisexual hybridogenic species Poeciliopsis monacha-occidentalis has been shown to be between 60,000 and 150,000 years old (Quattro et al. 1992). A monophyletic lineage of the unisexual gynogenetic species Poecilia formosa has been shown to be approximately 120,000 years old (Stock et al. 2010). There are several lines of evidence suggesting that P. eosneogaeus do not originate from current or recent hybridization (Schlosser et al. 1998; Doeringsfeld et al. 2004; Angers & Schlosser 2007). The results of my work (see chapter 3, chapter 4, figure 4.2) support and expand those of (Angers & Schlosser 2007) and suggest that the current distribution of P. eos-neogaeus originated from a small number of hybridization events in a Mississippi glacial refuge approximately 50,000 years ago. My work lends further support to the rarity of P. eos-neogaeus hybrid origins by expanding the known breadth of the distribution of P. eos-neogaeus lineages that originated in the Mississippi glacial refuge (see chapter 3). Mississippian P. eosneogaeus lineages extend to the Atlantic coast of Canada, suggesting that it is unlikely that P. eos-neogaeus originated in any Atlantic refugia. These examples of spermdependent asexual species that are relatively rare and old suggest that the persistence of sperm-dependent asexual species depends greatly on the specifics of ecological and mating interactions with sexual species. The rarity of P. eos-neogaeus origins may have to do with the complex genetic preconditions necessary to produce viable and fertile clonal genomes, as suggested by Stöck et al. (2010) in their ‘rare formation hypothesis’, which builds on earlier ideas by 101  Vrijenhoek (1989). My work (see chapter 6) suggests that the rarity of P. eos-neogaeus hybrid origins might also have to do with a general lack of hybridization success between P. eos and P. neogaeus. My results suggest that post-zygotic reproductive isolation exists between P. eos and P. neogaeus in the form of depressed viability of eggs fertilized by heterospecific males (see chapter 6). Das & Nelson (1990) have shown that P. eos and P. neogaeus spawn at different times, providing additional evidence for reproductive isolation. It is possible that pre-mating isolating barriers, particularly those affecting spawning time, were diminished due to environmental changes during the last glacial period thereby increasing the frequency of hybridization events (e.g., Taylor et al. 2006) and increasing the likelihood that the complex genetic preconditions necessary to produce viable and fertile clonal genomes would arise. It would be enlightening to know if the origins of other sperm dependent asexuals (e.g., Poeciliopsis monacha-occidentalis and Poecilia formosa) coincide with known environmental perturbations. C1(=.($F*(707#7&-(K%.'*##*#($F&$(&%2#*(+%.;($F*(203.-3*;*0$(.+(#K*%;(20(#K*%;J :*K*0:*0$(&#*<7&-(#A#$*;#(F&3*(&0(2;K.%$&0$(%.-*(20(;&20$&20206($F*('.*<2#$*0'*(.+( #K*%;J:*K*0:*0$(&#*<7&-#(&0:($F*2%(#*<7&-(F.#$#\( In the case of the unisexual hybridogenic species Poeciliopsis monacha-lucida, Quattro et al. (1991) have suggested that the hybridogenic mating system allows a feedback process of genetic exchange between the asexual and sexual species that maintains diversity in both and facilitates coexistence. In general, however, the particularities inherent in mechanisms of sperm-dependent asexual reproduction have been ignored in terms of their role in the coexistence of asexual and sexual species. Given the peculiarity of the process involved in the creation of Phoxinus cybrids (see section 1.1), and the fact that cybrids were until recently known only from laboratory breeding experiments (Goddard & Schultz 1993), the reasonable expectation is that cybrids are a rare component of P. eos populations. Indeed, previous work has shown that cybrids may occur in populations with or without pure P. eos, but that most lakes contain no cybrids at all (Angers & Schlosser 2007). My work (chapter 3) has shown that cybrids, which are essentially P. eos with P. neogaeus mitochondria, have replaced ‘pure’ P. eos in the northern portion of their distribution. Hence, the creation of cybrids  102  (likely via non-clonal reproduction in triploid hybrid Phoxinus) has potentially important implications for the coexistence of P. eos-neogaeus and its sexual hosts. If P. neogaeus mitochondria are better-adapted to northern environments, this mitochondrial introgression may facilitate coexistence in northern areas by ‘leveling the playing field’ (i.e., reducing the disparity in fitness) between P. eos-neogaeus and its most common sperm donor. O1(=.*#(#*-*'$2.0(+.%(K%.+2'2*0'A(&$(#.-2'2$206(#K*%;(-*&:($.(&(-2;2$*:(:23*%#2$A(.+( #K*%;J:*K*0:*0$(&#*<7&-(6*0.$AK*#(92$F(F.#$J;&$'F206(KF*0.$AK*#\(( Several models of coexistence between sperm-dependent asexuals and their sexual hosts, including my own, highlight the importance of the role of mate choice in maintaining this coexistence (Moore 1976; Hellriegel & Reyer 2000; Heubel et al. 2009; Mee & Otto 2010). It is, therefore, straightforward to predict that there will be selection on sperm-dependent asexuals to be proficient at soliciting sperm from their sexual hosts. Variation between asexual clones in traits important for soliciting sperm may arise via mutation, but it is likely that most of the genetic variation between clones would have arisen at the time the asexual lineages originated (e.g., via hybridization). Sexual selection should lead to a winnowing of this diversity of sperm-dependent asexual genotypes (i.e., clones), leaving those that are adequately proficient at soliciting sperm from a sperm donor. In sperm-dependent asexuals that have arisen via hybridization (i.e., all vertebrate asexuals), it is possible (if not likely) that the two parental species will differ in the phenotypes with which they prefer to mate. Hence, sperm-dependent asexual hybrids may have to specialize on being proficient at soliciting sperm from one or the other parental species. Another straightforward prediction, therefore, is that different populations of a sperm-dependent asexual hybrid that coexist with different parental host species will differ in phenotype such that they match their host species. My work (chapter 4) has shown this prediction to hold true for P. eos-neogaeus populations that use either P. eos or P. neogaeus as hosts. Populations of P. eos-neogaeus that use P. eos sperm were genetically distinct from populations that use P. neogaeus sperm (and more similar to P. eos). Also, P. neogaeus is larger than P. eos, and in samples from Alberta P. eos-neogaeus that use P. neogaeus sperm were larger than P. eos-neogaeus that use P. eos sperm. There is also evidence from my work and that of Gabor et al. 103  (2010) that increased length of time of coexistence may lead to both increased strength of preference in host species and increased host-specific specialization in sperm-dependent asexuals. H1(=.(#K*%;J:*K*0:*0$(&#*<7&-#(#F.9(%*:7'*:(K*%+.%;&0'*(20(+2$0*##J%*-&$*:($%&2$#( %*-&$23*($.(#*<7&-(F.#$#\(( There are contrasting expectations and results in the literature with regards to the fitness of sperm-dependent asexuals relative to their sexually-reproducing hosts. There is evidence for higher fitness among asexuals in some sperm-dependent asexual systems (Hotz et al. 1999; Loyning 2000; Kearney et al. 2005), and there is evidence for higher fitness among sexuals in others (Wetherington et al. 1987; Lively et al. 1990; Moritz et al. 1991; Weeks 1995; Cullum 1997; Vorburger 2001; Mee & Rowe 2006; Mee et al. 2011). Each of these studies examines only a single fitness-related trait in each spermdependent asexual species. To get a reasonably accurate estimate of fitness, however, several fitness-related traits must be considered because no single fitness-related trait is expected to correlate with overall fitness (Gerstein & Otto in press). My dissertation work (chapters 5 and 6) along with my previous work (Mee & Rowe 2006) consisted of five comparisons of relative fitness between P. eos-neogaeus and its parental sperm donors. The overall results are variable, but it seems likely that, overall, the asexuals are at least not any fitter than the sexuals. Only three comparisons show the possible existence of a fitness advantage in the asexuals: absolute fecundity is higher in P. eosneogaeus than P. eos, but this advantage disappears when fecundity differences are scaled to differences in body size (figure 6.2); a higher fraction of P. eos-neogaeus eggs than P. eos eggs are viable post-fertilization, but this advantage is slight (and nonsignificant)(figure 6.3); Phoxinus eos-neogaeus grow faster in the first 60 days posthatching than both their P. eos and P. neogaeus hosts (figure 6.4). Hybrid vigour might be expected to arise spontaneously among asexual vertebrates due to the high degree of genetic heterozygosity characteristic of interspecific hybrids (leading, perhaps, to a “general purpose genotype”) or might be expected to result from interclonal selection leading to the persistence of only the most fit hybrid genotypes (Lynch 1984). In the case of P. eos-neogaeus, it seems that any vestiges of hybrid vigour have largely eroded. Interclonal selection might have initially lead to the 104  persistence of only the most fit P. eos-neogaeus genotypes, but alternative selective forces (e.g., sexual selection – see chapter 4) or the fitness-eroding effects of asexual reproduction (e.g., Muller’s ratchet) may have negated this initial selective advantage. Consequently, my findings regarding the fitness of P. eos-neogaeus highlight the importance of considering the age of sperm-dependent asexual species when making predictions about the relative fitness of sexuals and asexuals.  L1D1(57$7%*(=2%*'$2.0#( L1D1>1(/&K$23*(%*&%206(&0:(;&$*('F.2'*(20(!"#$%&'() Given that the importance of mate choice in the coexistence of sperm-dependent asexuals and their sexual hosts is highlighted in several contexts throughout my dissertation, it is perhaps an egregious omission to have forgone any experimental investigation of mate choice among Phoxinus species. This omission is not by design. Over the course of three years at the beginning of my doctoral studies (and one year during my Masters degree), I tried unsucessfully to maintain captive breeding populations of Phoxinus species. Having individuals breed in captivity is clearly the minimum requirement for the commencement of experimental investigations into mate choice. Alas, animal husbandry is time consuming and expensive, and it was important to focus efforts on more fruitful avenues of research rather than to continue trying to establish captive breeding populations. Nonetheless, that something is difficult does not mean that it is not worthwhile. My efforts have led to the following three recommendations (based on personal observations) for future attempts at maintaining and rearing Phoxinus for experimental studies: 1. The highest standard of fish husbandry, in terms of the frequency of water changes and the quality of diet, must be maintained. Phoxinus seem particularly reluctant to enter spawning condition unless conditions are nearly ideal. 2. Future attempts should take place in reasonable proximity to wild populations of Phoxinus in order to facilitate regular restocking of  105  captive populations and occasional replacement when captive populations experience high mortality. 3. Phoxinus in captivity likely need to be stimulated in order to attain breeding condition. This should be achieved by simulating seasonal changes in light cycle (intensity and duration) and seasonal changes in water temperature. The establishment of captive breeding populations of Phoxinus would allow at least three additional avenues of investigation. First, hybrids created de novo from P. eos x P. neogaeus crosses would be valuable for inferences about the origin of P. eosneogaeus. When these de novo hybrids are raised to maturity it would become clear whether they are unisexual, and offspring from subsequent backcrosses with the parental species would indicate if the de novo hybrids were gynogenetic. If de novo hybridization rarely gives rise to gynogenetic hybrids, then it is likely that the genetic preconditions necessary to produce gynogens are complex and are rarely achieved, lending support to the rare formation hypothesis for the rarity of asexual vertebrates (Stock et al. 2010). If de novo hybridization readily gives rise to gynogenetic hybrids, then the rarity of asexual vertebrates, and the relatively old age of certain asexual lineages, likely has to do with reproductive barriers between parental species. This approach has been used in the Poecilia formosa gynogenetic system, and researchers were unable to resynthesize gynogens from crosses between the parental species (Poecilia latipinna and Poecilia mexicana) (Stock et al. 2010) thereby providing support for the rare formation hypothesis. Hybrids created de novo could also be used to investigate the origin of P. eosneogaeus in an additional sense. Hybrid vigour might be expected to arise spontaneously among asexual vertebrates due to the high degree of genetic heterozygosity characteristic of interspecific hybrids (leading, perhaps, to a “general purpose genotype”) or might be expected to result from interclonal selection leading to the persistence of only the fittest hybrid genotypes (Lynch 1984). The variation in fitness-related traits among hybrid individuals created de novo compared to naturally occurring P. eos-neogaeus might indicate whether interclonal selection has in fact led to the persistence of only the fittest hybrid genotypes. 106  A second avenue of research made available by the establishment of captive breeding populations of Phoxinus would be the evaluation of mate preferences. Male preference for conspecific mates has been detected in several systems with spermdependent asexuals. Male Poecilia latipinna have been shown to prefer to mate with conspecific females rather than gynogenetic Poecilia formosa (Gabor & Ryan 2001; Gabor et al. 2010), although earlier studies found that neither male Poecilia latipinna nor male Poecilia mexicana showed any preference for conspecific females over Poecilia formosa (Balsano et al. 1981; Balsano et al. 1985; Woodhead & Armstrong 1985; Schlupp et al. 1991). Male Poeciliopsis monacha have been shown to have a preference for mating with conspecific females rather than gynogenetic Poeciliopsis 2-monachalucida (Moore & McKay 1971), and male Poeciliopsis lucida have been shown to prefer to mate with conspecific females rather than unisexual hybridogenic Poeciliopsis monacha-lucida (Moore & McKay 1971). Females of both the bisexual hybridogenic species Rana esculenta and of its sexually-reproducing host species Rana lessonae have been shown to prefer to mate with males of the sexually-reproducing species Rana lessonae (Abt & Reyer 1993; Engeler & Reyer 2001). In all systems studied to date, the most recent work indicates that sexual species prefer to mate with conspecific mates. While the possibility of male preference for conspecific females over spermdependent asexual females is not in question, the extent and nature of the variability in this preference is still largely unknown. One hypothesis for the stable coexistence of sperm-dependent asexuals and their sexual hosts posits that stability is maintained by frequency-dependent selection against the asexuals due to increased strength of male preference (or increased effect of male preference) as the asexuals become more common (Moore 1976; Heubel et al. 2009). One testable prediction of this hypothesis is temporal or spatial variability in the strength of male preference as the frequency of asexuals varies over time or space. Another testable prediction of this hypothesis is that there should be plasticity (i.e., non-heritability) in the preference trait. An alternative hypothesis for the stable coexistence of sperm-dependent asexuals and their sexual hosts posits that stability is maintained by variation among males in the degree of preference for mating with sexual females and when males with stronger preference pay a higher cost of preference (see chapter 2). A testable prediction from this alternative hypothesis is the existence of 107  heritable variation in the strength of male preference within populations. The only available data from sperm-dependent asexual systems that are relevant to these alternative hypotheses and predictions come from work on Poecilia latipinna. There was substantial inter-population variability in the strength of preference by male Poecilia latipinna for conspecific rather than gynogenetic females (Gabor & Ryan 2001; Gabor et al. 2010), and there was low repeatability of male preference within populations, suggesting low heritability (Gabor & Aspbury 2008). The alternative hypotheses and predictions outlined above could further be evaluated using mate choice experiments with captive Phoxinus populations. A third avenue of research made available by the establishment of captive breeding populations of Phoxinus is related to the findings in chapter 4 of my dissertation. I found that P. eos-neogaeus that use different parental species as hosts are genetically distinct and match the size of their host species. A prediction that follows from this finding is that males should prefer to mate with females of a size similar to their own species. This could lead to directional selection for smaller asexuals in populations where P. eos is the sperm donor, directional selection for larger asexuals where P. neogaeus is the sperm donor, or stabilizing selection for a particular size in each scenario. Male Poecilia latipinna, which are hosts for the gynogenetic species Poecilia formosa, prefer to mate with conspecific females, though Poecilia latipinna females are smaller than Poecilia formosa (Gabor & Ryan 2001; Gabor et al. 2010). Gabor et al. (2010) found that mate preference in male Poecilia latipinna caused stabilizing selection for intermediate female size in one population, and directional selection for larger females in another population. This pattern of female size preference in Poecilia latipinna, which is contrary to the prediction of selection favouring host matching, may be the result of strong selection for mate quality (of which female size is a strong predictor), which may be in conflict with selection for species recognition. Future studies with Phoxinus could evaluate the size preference of males of different host species, as well as the nature of sexual selection on body size in sperm-dependent asexuals coexisting with different host species.  108  L1D1D1(]#206(%*K*&$(#92;;206(K*%+.%;&0'*($.(*3&-7&$*(KFA#2.-.62'&-(#$%*##*#( In chapter 5, I describe a protocol for a test of repeat swimming performance that is likely a good method to evaluate physiological performance in fishes. This test could be applied in a variety of contexts (e.g., toxicology, senescence studies, migratory performance) to evaluate the effects of stressors (e.g., environmental toxins, age, migratory status) on physiological performance. This test could also be used to follow up on a previous study showing that P. eos-neogaeus carries a higher load of Gyrodactylus parasites than its parental species (Mee & Rowe 2006). This result was interpreted as support for the Red Queen hypothesis for the maintenance of sex, but this interpretation is contingent on the parasites being pathogenic. A repeat swimming performance test, such as the one describe in chapter 5, could be used to evaluate the pathogenicity of Gyrodactylus parasites by testing the performance of fish with varying parasite loads (e.g., Nendick et al. 2011). L1D1C1(,*;K*%&$7%*J:*K*0:*0$(K*%+.%;&0'*(.+('A"%2:#(&0:(K7%*(!.)*#() In chapter 3 I have shown that cybrids, which are essentially P. eos with P. neogaeus mitochondria, have replaced ‘pure’ P. eos in the northern portion of their distribution. The fact that cybrids tend to dominate in the northern portion of the distribution suggests that P. neogaeus mitochondria might be more cold-adapted than P. eos mitochondria. Evidence for the role of mitochondrial genetic variation in cold adaptation in fishes has been proposed and documented previously (Bernatchez et al. 1995; Glemet et al. 1998; Wilson & Bernatchez 1998; Doiron et al. 2002; Redenbach & Taylor 2002; Lannig et al. 2005; Lucassen et al. 2006; Cote et al. 2007). Measures of the physiological performance of P. eos and P. neogaeus mitochondria at different temperatures would help distinguish between two alternate hypothesis for this widespread replacement of P. eos with P. neogaeus mitochondria. The replacement may have occurred due to a stochastic ratchet-like process: a Phoxinus community comprised of hybrids and pure P. eos can give rise to cybrids, but a Phoxinus community comprised of hybrids and cybrids cannot give rise to pure P. eos. Alternatively, there may be a selective advantage to having a P. neogaeus mitochondrial genome in certain environments. If P. neogaeus mitochondria are better adapted to northern environments, this mitochondrial introgression may facilitate coexistence in northern areas by ‘leveling 109  the playing field’ (i.e., reducing the disparity in fitness) between P. eos-neogaeus and its most common sperm donor.  L1C1(520&-(,F.76F$#( My work has implications on three levels. There are relatively narrow implications in the context of understanding the Phoxinus system, there are broader implications to the study of sperm-dependent asexual species, and there are general implications for understanding the evolution and maintenance of sex. The implications of my work at these three levels are hierarchical in terms of the number of species to which they are relevant, though they are not hierarchical in terms of their value. In terms of understanding the Phoxinus system, I have contributed new knowledge about the distribution of P. eos-neogaeus, about the extent of mitochondrial introgression between P. eos and P. neogaeus, about the existence of adaptive inter-clonal diversity in P. eosneogaeus, and about the importance of inter-species variation in fitness-related traits. At a more general level, I have proposed and tested new hypotheses for the coexistence of sperm-dependent asexuals and their sexual hosts. Finally, at the most general level, I have highlighted that the maintenance of sex, even in vertebrates, depends on ongoing competition with asexual forms, and that the principals and processes involved in the historical proliferation of sexual reproduction are relevant to the ongoing persistence of sexual reproduction.  110  Y2"-2.6%&KFA( Abt, G., and Reyer, H.U. 1993. Mate choice and fitness in a hybrid frog - Rana esculenta females prefer Rana lessonae males over their own. Behavioral Ecology and Sociobiology 32(4): 221-228. Agrawal, A.F. 2006. Evolution of sex: why do organisms shuffle their genotypes? Current Biology 16: R696-R704. Angers, B., and Schlosser, I.J. 2007. The origin of Phoxinus eos-neogaeus unisexual hybrids. Molecular Ecology 16(21): 4562-4571. Arnold, M.L. 1997. Natural Hybridization and Evolution. Oxford University Press, Inc., New York. Avise, J.C. 2008. Clonality: The Genetics, Ecology, and Evolution of Sexual Abstinence in Vertebrates. Oxford University Press, Oxford. Avise, J.C. 2009. Phylogeography: retrospect and prospect. Journal of Biogeography 36: 3-15. Baldauf, S.A., Kullmann, H., Schroth, S.H., Thunken, T., and Bakker, T.C.M. 2009. You can't always get what you want: size assortative mating by mutual mate choice as a resolution of sexual conflict. BMC Evolutionary Biology 9. doi:12910.1186/1471-2148-9-129. Balsano, J.S., Kucharski, K., Randle, E.J., Rasch, E.M., and Monaco, P.J. 1981. Reduction of competition between bisexual and unisexual females of Poecilia in northeastern Mexico. Environmental Biology of Fishes 6(1): 39-48. Balsano, J.S., Randle, E.J., Rasch, E.M., and Monaco, P.J. 1985. Reproductive behavior and the maintenance of all-females Poecilia. Environmental Biology of Fishes 12(4): 251-263. Barash, D.P. 1976. What does sex really cost? American Naturalist 110(975): 894-897. Baube, C.L. 2008. Body size and the maintenance of reproductive isolation in stickleback, genus Gasterosteus. Ethology 114(11): 1122-1134. doi:10.1111/j.1439-0310.2008.01564.x. Beamish, F.W.H. 1978. Swimming Capacity. In Fish Physiology. Edited by W. H. Hoar, and D. J. Randall. Academic Press, New York. pp. 101-187. 111  Bell, G. 1982. The Masterpiece of Nature: The Evolution and Genetics of Sexuality. Croom Helm Ltd., London. Berkeley, S.A., Chapman, C., and Sogard, S.M. 2004. Maternal age as a determinant of larval growth and survival in a marine fish, Sebastes melanops. Ecology 85(5): 1258-1264. doi:10.1890/03-0706. Bernatchez, L., Glemet, H., Wilson, C.C., and Danzmann, R.G. 1995. Introgression and fixation of arctic char (Salvelinus alpinus) mitochondrial genome in an allopatric population of brook trout (Salvelinus fontinalis). Canadian Journal of Fisheries and Aquatic Sciences 52(1): 179-185. doi:10.1139/f95-018. Beukeboom, L.W., and Vrijenhoek, R.C. 1998. Evolutionary genetics and ecology of sperm-dependent parthenogenesis. Journal of Evolutionary Biology 11(6): 755782. Binet, M.C., and Angers, B. 2005. Genetic identification of members of the Phoxinus eos-neogaeus hybrid complex. Journal of Fish Biology 67(4): 1169-1177. Booksmythe, I., Detto, T., and Backwell, P.R.Y. 2008. Female fiddler crabs settle for less: the travel costs of mate choice. Animal Behaviour 76: 1775-1781. doi:10.1016/j.anbehav.2008.07.022. Brauner, C.J., Iwama, G.K., and Randall, D.J. 1994. The effect of short duration seawater exposure on the swimming performance of wild and hatchery-reared juvenile coho salmon (Oncorhynchus kisutch) during smoltification. Canadian Journal of Fisheries and Aquatic Sciences 51(10): 2188-2194. Brett, J.R. 1964. The respiratory metabolism and swimming performance of young sockeye salmon. Journal of the Fisheries Research Board of Canada 21: 11831226. Brncic, D., and Budnik, M. 1974. Rate of development and viability of hybrids between Drosophila pavani and Drosophila gaucha under competitive conditions. Ecology 55(3): 662-666. Charlesworth, B. 1980. The cost of sex in relation to mating system. Journal of Theoretical Biology 84(4): 655-671. Clement, M., Posada, D., and Crandall, K.A. 2000. TCS: a computer program to estimate gene geneologies. Molecular Ecology 9(10): 1657-1660. 112  Cochran, P.A., Lodge, D.M., Hodgson, J.R., and Knapik, P.G. 1988. Diets of syntopic finescale cace, Phoxinus neogaeus, and northern redbelly dace, Phoxinus eos - a reflection of trophic morphology. Environmental Biology of Fishes 22(3): 235240. Cote, G., Perry, G., Blier, P., and Bernatchez, L. 2007. The influence of geneenvironment interactions on GHR and IGF-I expression and their association with growth in brook charr, Salvelinus fontinalis (Mitchill). BMC Genetics 8(87): 1471-2156. doi:10.1186/1471-2156-8-87. Crews, D., Grassman, M., and Lindzey, J. 1986. Behavioral facilitation of reproduction in sexual and unisexual whiptail lizards. Proceedings of the National Academy of Sciences of the United States of America 83(24): 9547-9550. Cullum, A.J. 1997. Comparisons of physiological performance in sexual and asexual whiptail lizards (genus Cnemidophorus): Implications for the role of heterozygosity. American Naturalist 150(1): 24-47. Darevsky, I.S., Danielyan, F.D., Sokolova, T.M., and Rozanov, Y.M. 1989. Intraclonal mating in the parthenogenetic lizard species Lacerta unisexualis Darevsky. In Evolution and Ecology of Unisexual Vertebrates. Edited by R. M. Dawley, and J. P. Bogart. New York State Museum, Albany, New York. pp. 228-235. Darwin, C.R. 1862. On the two forms, or dimorphic condition, in the species Primula, and on their remarkable sexual relations. Journal of the Proceedings of the Linnean Society (Botany) 6: 77-96. Das, M.K., and Nelson, J.S. 1990. Spawning time and fecundity of Northern redbelly dace, Phoxinus eos, finescale dace, Phoxinus neogaeus, and their hybrids in Upper Pierre Gray Lake, Alberta. Canadian Field-Naturalist 104(3): 409-413. Dawley, R.M., and Bogart, J.P. (eds). 1989. Evolution and Ecology of Unisexual Vertebrates. New York State Education Department, Albany, New York. Doeringsfeld, M.R., Schlosser, I.J., Elder, J.F., and Evenson, D.P. 2004. Phenotypic consequences of genetic variation in a gynogenetic complex of Phoxinus eosneogaeus clonal fish (Pisces : Cyprinidae) inhabiting a heterogeneous environment. Evolution 58(6): 1261-1273.  113  Doiron, S., Bernatchez, L., and Blier, P.U. 2002. A comparative mitogenomic analysis of the potential adaptive value of arctic charr mtDNA introgression in brook charr populations (Salvelinus fontinalis Mitchill). Molecular Biology and Evolution 19(11): 1902-1909. Dries, L.A. 2003. Peering through the looking glass at a sexual parasite: Are Amazon mollies red queens? Evolution 57(6): 1387-1396. Echelle, A.A., Dowling, T.E., Moritz, C.C., and Brown, W.M. 1989. Mitochondrial-DNA diversity and the origin of the Menidia clarkhubbsi complex of unisexual fishes (Atherinidae). Evolution 43(5): 984-993. Engeler, B., and Reyer, H.U. 2001. Choosy females and indiscriminate males: mate choice in mixed populations of sexual and hybridogenetic water frogs (Rana lessonae, Rana esculenta). Behavioral Ecology 12(5): 600-606. Farrell, A.P. 2008. Comparisons of swimming performance in rainbow trout using constant acceleration and critical swimming speed tests. Journal of Fish Biology 72(3): 693-710. Felsenstein, J. 1974. The evolutionary advantage of recombination. Genetics 78(2): 737756. Foote, C.J. 1988. Male mate choice dependent on male size in salmon. Behaviour 106: 63-80. Foote, C.J., and Larkin, P.A. 1988. The role of male choice in the assortative mating of anadromous and non-anadromous sockeye salmon (Oncorhynchus nerka). Behaviour 106: 43-62. Forsgren, E., and Magnhagen, C. 1993. Conflicting demands in sand gobies - predators influence reproductive behavior. Behaviour 126: 125-135. Gabor, C.R., and Aspbury, A.S. 2008. Non-repeatable mate choice by male sailfin mollies, Poecilia latipinna, in a unisexual-bisexual mating complex. Behavioral Ecology 19(4): 871-878. doi:10.1093/beheco/arn043. Gabor, C.R., Gonzalez, R., Parmley, M., and Aspbury, A.S. 2010. Variation in male sailfin molly, preference for female size: does sympatry with sexual parasites, drive preference for smaller conspecifics? Behavioral Ecology and Sociobiology 64(5): 783-792. doi:10.1007/s00265-009-0895-2. 114  Gabor, C.R., and Ryan, M.J. 2001. Geographical variation in reproductive character displacement in mate choice by male sailfin mollies. Proceedings of the Royal Society of London Series B-Biological Sciences 268(1471): 1063-1070. Gerstein, A.C., and Otto, S.P. in review. Diploid superiority over haploids during Saccharomcyes cerevisiae batch culture evolution is not explained by fitness assays. Plos One. Glemet, H., Blier, P., and Bernatchez, L. 1998. Geographical extent of Arctic char (Salvelinus alpinus) mtDNA introgression in brook char populations (S-fontinalis) from eastern Quebec, Canada. Molecular Ecology 7(12): 1655-1662. doi:10.1046/j.1365-294x.1998.00494.x. Goddard, K.A., Dawley, R.M., and Dowling, T.E. 1989. Origin and genetic relationships of diploid, triploid, and diploid-triploid mosaic biotypes in the Phoxinus eosneogaeus unisexual complex. In Evolution and Ecology of Unisexual Vertebrates. Edited by R. M. Dawley, and J. P. Bogart. New Yory State Education Department, Albany, New York. pp. 269-280. Goddard, K.A., Megwinoff, O., Wessner, L.L., and Giaimo, F. 1998. Confirmation of gynogenesis in Phoxinus eos-neogaeus (Pisces : Cyprinidae). Journal of Heredity 89(2): 151-157. Goddard, K.A., and Schultz, R.J. 1993. Aclonal reproduction by polyploid members of the clonal hybrid species Phoxinus eos-neogaeus (Cyprinidae). Copeia 1993(3): 650-660. Godin, J.G.J., and Briggs, S.E. 1996. Female mate choice under predation risk in the guppy. Animal Behaviour 51: 117-130. Godoy-Herrera, R., Burnet, B., and Connolly, K. 2005. Hybrid disadvantage in the larval foraging behaviour of the two neotropical species of Drosophila pavani and Drosophila gaucha. Genetica 124(1): 33-40. doi:10.1007/s10709-004-5913-8. Goldstein, D.B., Ruíz-Linares, A., Feldman, M., and Cavalli-Sforza, L.L. 1995. An evaluation of genetic distances for use with microsatellite loci. Genetics 139: 463471. Gwynne, D.T. 1989. Does copulation increase the risk of predation? Trends in Ecology & Evolution 4(2): 54-56. 115  Hamilton, W.D. 1980. Sex versus non-sex versus parasite. Oikos 35(2): 282-290. Hastings, I.M. 1999. The costs of sex due to deleterious intracellular parasites. Journal of Evolutionary Biology 12(1): 177-183. Hatfield, T., and Schluter, D. 1996. A test for sexual selection on hybrids of two sympatric sticklebacks. Evolution 50(6): 2429-2434. Hatfield, T., and Schluter, D. 1999. Ecological speciation in sticklebacks: Environmentdependent hybrid fitness. Evolution 53(3): 866-873. Hellriegel, B., and Reyer, H.U. 2000. Factors influencing the composition of mixed populations of a hemiclonal hybrid and its sexual host. Journal of Evolutionary Biology 13(6): 906-918. Heubel, K.U., Rankin, D.J., and Kokko, H. 2009. How to go extinct by mating too much: population consequences of male mate choice and efficiency in a sexual-asexual species complex. Oikos 118(4): 513-520. doi:10.1111/j.1600-0706.2009.17024.x. Hotz, H., Semlitsch, R.D., Gutmann, E., Guex, G.D., and Beerli, P. 1999. Spontaneous heterosis in larval life-history traits of hemiclonal frog hybrids. Proceedings of the National Academy of Sciences of the United States of America 96(5): 2171-2176. Hubbs, C.L. 1955. Hybridization between fish species in nature. Systematic Zoology 4: 120. Iwasa, Y., and Pomiankowski, A. 1991. The evolution of costly mate preferences 2. The handicap principle. Evolution 45(6): 1431-1442. Jain, K.E., Birtwell, I.K., and Farrell, A.P. 1998. Repeat swimming performance of mature sockeye salmon following a brief recovery period: a proposed measure of fish health and water quality. Canadian Journal of Zoology-Revue Canadienne De Zoologie 76(8): 1488-1496. Jain, K.E., and Farrell, A.P. 2003. Influence of seasonal temperature on the repeat swimming performance of rainbow trout Oncorhynchus mykiss. Journal of Experimental Biology 206(20): 3569-3579. Jennions, M.D., Moller, A.P., and Petrie, M. 2001. Sexually selected traits and adult survival: A meta-analysis. Quarterly Review of Biology 76(1): 3-36. Jokela, J., Lively, C.M., Dybdahl, M.F., and Fox, J.A. 1997. Evidence for a cost of sex in the freshwater snail Potamopyrgus antipodarum. Ecology 78(2): 452-460. 116  Kearney, M., Wahl, R., and Autumn, K. 2005. Increased capacity for sustained locomotion at low temperature in parthenogenetic geckos of hybrid origin. Physiological and Biochemical Zoology 78(3): 316-324. Kerr, B., Neuhauser, C., Bohannan, B.J.M., and Dean, A.M. 2006. Local migration promotes competitive restraint in a host-pathogen 'tragedy of the commons'. Nature 442(7098): 75-78. doi:10.1038/nature04864. Kerr, B., Riley, M.A., Feldman, M.W., and Bohannan, B.J.M. 2002. Local dispersal promotes biodiversity in a real-life game of rock-paper-scissors. Nature 418(6894): 171-174. doi:10.1038/nature00823. Kieffer, J.D. 2000. Limits to exhaustive exercise in fish. Comparative Biochemistry and Physiology a-Molecular & Integrative Physiology 126(2): 161-179. Kieffer, J.D., Currie, S., and Tufts, B.L. 1994. Effects of environmental temperature on the metabolic and acid-base responses of rainbow trout to exhaustive exercise. Journal of Experimental Biology 194: 299-317. Kirkpatrick, M. 1982. Sexual selection and the evolution of female choice. Evolution 36(1): 1-12. Kokko, H., Heubel, K.U., and Rankin, D.J. 2008. How populations persist when asexuality requires sex: the spatial dynamics of coping with sperm parasites. Proceedings of the Royal Society of London. Series B: Biological Sciences 275(1636): 817-825. doi:10.1098/rspb.2007.1199. Kondrashov, A.S. 1988. Deleterious mutations and the evolution of sexual reproduction. Nature 336: 435-440. Kondrashov, A.S. 1993. Classification of Hypotheses on the Advantage of Amphimixis. Journal of Heredity 84(5): 372-387. Lannig, G., Storch, D., and Portner, H.O. 2005. Aerobic mitochondrial capacities in Antarctic and temperate eelpout (Zoarcidae) subjected to warm versus cold acclimation. Polar Biology 28: 575-584. Lively, C.M. 1996. Host-parasite coevolution and sex. Bioscience 46(2): 107-114. Lively, C.M., Craddock, C., and Vrijenhoek, R.C. 1990. Red Queen hypothesis supported by parasitism in sexual and clonal fish. Nature 344(6269): 864-866.  117  Loyning, M.K. 2000. Reproductive performance of clonal and sexual bark beetles (Coleoptera : Scolytidae) in the field. Journal of Evolutionary Biology 13(5): 743748. Lucassen, M., Koschnick, N., Eckerle, L.G., and Portner, H.O. 2006. Mitochondrial mechanisms of cold adaptation in cod (Gadus morhua L.) populations from different climatic zones. Journal of Experimental Biology 209: 2462-2471. Lynch, M. 1984. Destabilizing hybridization, general-purpose genotypes and geographic parthenogenesis. The Quarterly Review of Biology 59: 257-290. Maekawa, K., Nakano, S., and Yamamoto, S. 1994. Spawning behavior and sizeassortative mating of japanese charr in an artificial lake-inlet stream system. Environmental Biology of Fishes 39(2): 109-117. Magnhagen, C. 1995. Sneaking behavior and nest defense are affected by predation risk in the common goby. Animal Behaviour 50: 1123-1128. Magnhagen, C., and Vestergaard, K. 1991. Risk-taking in relation to reproductive investments and future reproductive opportunities - field experiment son nestguarding common gobies, Pomatoschistus microps. Behavioral Ecology 2(4): 351-359. Mappes, J., Alatalo, R.V., Kotiaho, J., and Parri, S. 1996. Viability costs of conditiondependent sexual male display in a drumming wolf spider. Proceedings of the Royal Society of London. Series B: Biological Sciences 263(1371): 785-789. Massicotte, R., Whitelaw, E., and Angers, B. 2011. DNA methylation A source of random variation in natural populations. Epigenetics 6(4): 422-428. doi:10.4161/epi.6.4.14532. Maynard Smith, J. 1978. The Evolution of Sex. Cambridge University Press, Cambridge. McKenzie, D.J., Garofalo, E., Winter, M.J., Ceradini, S., Verweij, F., Day, N., Hayes, R., van der Oost, R., Butler, P.J., Chipman, J.K., and Taylor, E.W. 2007. Complex physiological traits as biomarkers of the sub-lethal toxicological effects of pollutant exposure in fishes. Philosophical Transactions of the Royal Society BBiological Sciences 362(1487): 2043-2059. doi:10.1098/rstb.2007.2100.  118  McLean, J.E., Bentzen, P., and Quinn, T.P. 2005. Nonrandom, size- and timing-biased breeding in a hatchery population of steelhead trout. Conservation Biology 19(2): 446-454. McPhail, J.D., and Lindsey, C.C. 1970. Freshwater fishes of northwestern Canada and Alaska. Fisheries Research Board of Canada 173: 7-26. Mee, J.A., Brauner, C.J., and Taylor, E.B. 2011. Repeat swimming performance and its implications for inferring the relative fitness of asexual hybrid dace (Pisces: Phoxinus) and their sexually reproducing parental species. Physiological and Biochemical Zoology 84(3): 306-315. doi:10.1086/659245. Mee, J.A., and Otto, S.P. 2010. Variation in the strength of male mate choice allows long-term coexistence of sperm-dependent asexuals and their sexual hosts. Evolution 64(10): 2808–2819. Mee, J.A., and Rowe, L. 2006. A comparison of parasite loads on asexual and sexual Phoxinus (Pisces : Cyprinidae). Canadian Journal of Zoology-Revue Canadienne De Zoologie 84(6): 808-816. Mee, J.A., and Rowe, L. 2010. Distribution of Phoxinus eos, Phoxinus neogaeus, and their asexually-reproducing hybrids (Pisces: Cyprinidae) in Algonquin Provincial Park, Ontario. Plos One 5(10): e13185. Moore, W.S. 1976. Components of fitness in a unisexual fish Poeciliopsis monachaoccidentalis. Evolution 30(3): 564-578. Moore, W.S., and McKay, F.E. 1971. Coexistence in unisexual species complexes of Poeciliopsis (Pisces-Poeciliidae). Ecology 52(5): 791-&. Moran, P.A.P. 1975. Wandering distributions and the electrophoretic profile. Theoretical Population Biology 8: 318-330. Morita, K., and Nagasawa, T. 2010. Latitudinal variation in the growth and maturation of masu salmon (Oncorhynchus masou) parr. Canadian Journal of Fisheries and Aquatic Sciences 67(6): 955-965. doi:10.1139/f10-028. Moritz, C., McCallum, H., Donnellan, S., and Roberts, J.D. 1991. Parasite loads in parthenogenetic and sexual lizards (Heteronotia binoei) - support for the Red Queen Hypothesis. Proceedings of the Royal Society of London. Series B: Biological Sciences 244(1310): 145-149. 119  Nelson, J.A., Gotwalt, P.S., and Snodgrass, J.W. 2003. Swimming performance of blacknose dace (Rhinichthys atratulus) mirrors home-stream current velocity. Canadian Journal of Fisheries and Aquatic Sciences 60(3): 301-308. doi:10.1139/f03-023. Nendick, L., Grant, A., Gardner, M., Sackville, M., Brauner, C.J., and Farrell, A.P. 2009. Swimming performance and associated ionic disturbance of juvenile pink salmon Oncorhynchus gorbuscha determined using different acceleration profiles. Journal of Fish Biology 75(7): 1626-1638. doi:10.1111/j.10958649.2009.02388.x. Nendick, L., Sackville, M., Tang, S., Brauner, C.J., and Farrell, A.P. 2011. Sea lice infection of juvenile pink salmon (Oncorhynchus gorbuscha): effects on swimming performance and postexercise ion balance. Canadian Journal of Fisheries and Aquatic Sciences 68(2): 241-249. doi:10.1139/f10-150. Otto, S.P., and Barton, N.H. 2001. Selection for recombination in small populations. Evolution 55(10): 1921-1931. Pomiankowski, A., Iwasa, Y., and Nee, S. 1991. The evolution of costly mate preferences 1. Fisher and biased mutation. Evolution 45(6): 1422-1430. Quattro, J.M., Avise, J.C., and Vrijenhoek, R.C. 1991. Molecular evidence for multiple origins of hybridogenetic fish clones (Poeciliidae, Poeciliopsis). Genetics 127(2): 391-398. Quattro, J.M., Avise, J.C., and Vrijenhoek, R.C. 1992. An ancient clonal lineage in the fish genus Poeciliopsis (Atheriniformes, Poeciliidae). Proceedings of the National Academy of Sciences of the United States of America 89(1): 348-352. R Development Core Team. 2010. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Randall, D., and Brauner, C.J. 1991. Effects of environmental factors on exercise in fish. Journal of Experimental Biology 160: 113-126. Redenbach, Z., and Taylor, E.B. 2002. Evidence for historical introgression along a contact zone between two species of char (Pisces : Salmonidae) in northwestern North America. Evolution 56(5): 1021-1035. doi:10.1111/j.00143820.2002.tb01413.x. 120  Rogers, S.M., and Bernatchez, L. 2006. The genetic basis of intrinsic and extrinsic postzygotic reproductive isolation jointly promoting speciation in the lake whitefish species complex (Coregonus clupeaformis). Journal of Evolutionary Biology 19(6): 1979-1994. doi:10.1111/j.1420-9101.2006.01150.x. Rosenfield, J.A., Nolasco, S., Lindauer, S., Sandoval, C., and Kodric-Brown, A. 2004. The role of hybrid vigor in the replacement of Pecos pupfish by its hybrids with sheepshead minnow. Conservation Biology 18(6): 1589-1598. Rouleau, S., Glemet, H., and Magnan, P. 2010. Effects of morphology on swimming performance in wild and laboratory crosses of brook trout ecotypes. Functional Ecology 24(2): 310-321. doi:10.1111/j.1365-2435.2009.01636.x. Rowe, L. 1994. The costs of mating and mate choice in water striders. Animal Behaviour 48(5): 1049-1056. Sakurai, G., and Kasuya, E. 2008. The costs of harassment in the adzuki bean beetle. Animal Behaviour 75: 1367-1373. doi:10.1016/j.anbehav.2007.09.010. Schley, D., Doncaster, C.P., and Sluckin, T. 2004. Population models of spermdependent parthenogenesis. Journal of Theoretical Biology 229(4): 559-572. doi:10.1016/j.jtbi.2004.04.031. Schlosser, I.J., Doeringsfeld, M.R., Elder, J.F., and Arzayus, L.F. 1998. Niche relationships of clonal and sexual fish in a heterogeneous landscape. Ecology 79(3): 953-968. Schlupp, I. 2005. The evolutionary ecology of gynogenesis. Annual Review of Ecology, Evolution and Systematics 36: 399-417. doi:10.1146/annurev.ecolsys.36.102003.152629. Schlupp, I., Marler, C., and Ryan, M.J. 1994. Benefit to male sailfin mollies of mating with heterospecific females. Science 263(5145): 373-374. Schlupp, I., Nanda, I., Dobler, M., Lamatsch, D.K., Epplen, J.T., Parzefall, J., Schmid, M., and Schartl, M. 1998. Dispensable and indispensable genes in an ameiotic fish, the Amazon molly Poecilia formosa. Cytogenetics and Cell Genetics 80(14): 193-198.  121  Schlupp, I., Parzefall, J., and Schartl, M. 1991. Male mate choice in mixed bisexual unisexual breeding complexes of Poecilia (Teleostei, Poeciliidae). Ethology 88(3): 215-222. Schluter, D. 1996. Ecological speciation in postglacial fishes. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 351(1341): 807-814. Schluter, D., and Nagel, L.M. 1995. Parallel speciation by natural selection. American Naturalist 146(2): 292-301. Schultz, R.J. 1969. Hybridization, unisexuality and polyploidy in teleost Peociliopsis (Poeciliidae) and other vertebrates. American Naturalist 103: 605-619. Scott, W.B., and Crossman, E.J. 1979. Freshwater Fishes of Canada. Minister of Supply and Services Canada, Ottawa. Shima, J.S., and Findlay, A.M. 2002. Pelagic larval growth rate impacts benthic settlement and survival of a temperate reef fish. Marine Ecology-Progress Series 235: 303-309. doi:10.3354/meps235303. Sinervo, B., and Lively, C.M. 1996. The rock-paper-scissors game and the evolution of alternative male strategies. Nature 380(6571): 240-243. Slatkin, M. 1995. A measure of population subdivision based on microsatellite allele frequencies. Genetics 139(1): 457-462. Sogard, S.M. 1997. Size-selective mortality in the juvenile stage of teleost fishes: A review. Bulletin of Marine Science 60(3): 1129-1157. Stock, M., Lampert, K.P., Moller, D., Schlupp, I., and Schartl, M. 2010. Monophyletic origin of multiple clonal lineages in an asexual fish (Poecilia formosa). Molecular Ecology 19(23): 5204-5215. doi:10.1111/j.1365-294X.2010.04869.x. Taylor, E.B., Boughman, J.W., Groenenboom, M., Sniatynski, M., Schluter, D., and Gow, J.L. 2006. Speciation in reverse: morphological and genetic evidence of the collapse of a three-spined stickleback (Gasterosteus aculeatus) species pair. Molecular Ecology 15(2): 343-355. Taylor, E.B., and Foote, C.J. 1991. Critical swimming velocities of juvenile sockeye salmon and kokanee, the anadromous and non-anadromous forms of Oncorhynchus nerka (Walbaum). Journal of Fish Biology 38(3): 407-419.  122  Taylor, E.B., and McPhail, J.D. 1985. Variation in burst and prolonger swimming performance among British Columbia populations of coho salmon, Oncorhynchus kisutch. Canadian Journal of Fisheries and Aquatic Sciences 42(12): 2029-2033. Taylor, E.B., and McPhail, J.D. 1986. Prolonged and burst swimming in anadromous and fresh water threespine stickleback, Gasterosteus aculeatus. Canadian Journal of Zoology-Revue Canadienne De Zoologie 64(2): 416-420. Thorpe, R.S. 1976. Biometric analysis of geographic variation and racial affinities. Biological Reviews of the Cambridge Philosophical Society 51(4): 407-452. Thunken, T., Baldauf, S.A., Kullmann, H., Schuld, J., Hesse, S., and Bakker, T.C.M. 2011. Size-related inbreeding preference and competitiveness in male Pelvicachromis taeniatus (Cichlidae). Behavioral Ecology 22(2): 358-362. doi:10.1093/beheco/arq217. Vamosi, S.M., Hatfield, T., and Schluter, D. 2000. A test of ecological selection against young-of-the-year hybrids of sympatric sticklebacks. Journal of Fish Biology 57(1): 109-121. Vorburger, C. 2001. Fixation of deleterious mutations in clonal lineages: Evidence from hybridogenetic frogs. Evolution 55(11): 2319-2332. Vrijenhoek, R.C. 1979a. Factors affecting clonal diversity and coexistence. American Zoologist 19(3): 787-797. Vrijenhoek, R.C. 1979b. Genetics of a sexually reproducing fish in a highly fluctuating environment. American Naturalist 113(1): 17-29. Vrijenhoek, R.C. 1989. Genetic and ecological constraints on the origins and establishment of unisexual vertebrates. In Evolution and Ecology of Unisexual Vertebrates. Edited by R. M. Dawley, and J. P. Bogart. New York State Education Department, Albany, New York. pp. 24-31. Vrijenhoek, R.C. 1994. Unisexual fish - model systems for studying ecology and evolution. Annual Review of Ecology and Systematics 25: 71-96. Vrijenhoek, R.C., Angus, R.A., and Schultz, R.J. 1977. Variation and heterozygosity in sexually vs clonally reproducing populations of Poeciliopsis. Evolution 31(4): 767-781.  123  Webb, P.W. 1982. Avoidance responses of fathead minnow to strikes by 4 teleost predators. Journal of Comparative Physiology 147(3): 371-378. Weeks, S.C. 1995. Comparisons of life-history traits between clonal and sexual fish (Poeciliopsis, Poeciliidae) raised in monoculture and mixed treatments. Evolutionary Ecology 9(3): 258-274. Wetherington, J.D., Kotora, K.E., and Vrijenhoek, R.C. 1987. A test of the spontaneous heterosis hypothesis for unisexual vertebrates. Evolution 41(4): 721-731. Wilkie, M.P., Brobbel, M.A., Davidson, K., Forsyth, L., and Tufts, B.L. 1997. Influences of temperature upon the postexercise physiology of Atlantic salmon (Salmo salar). Canadian Journal of Fisheries and Aquatic Sciences 54(3): 503-511. Williams, G.C. 1975. Sex and Evolution. Princeton University Press, Princeton, New Jersey. Wilson, C.C., and Bernatchez, L. 1998. The ghost of hybrids past: fixation of arctic charr (Salvelinus alpinus) mitochondrial DNA in an introgressed population of lake trout (S-namaycush). Molecular Ecology 7(1): 127-132. doi:10.1046/j.1365294x.1998.00302.x. Woodhead, A.D., and Armstrong, N. 1985. Aspects of the mating behavior of male mollies (Peocilia spp). Journal of Fish Biology 27(5): 593-601.  124  !KK*0:2'*#( !KK*0:2<(!B(!0&-A#*#(@7KK-*;*0$&%A($.(/F&K$*%(D( ^++*'$#(.+(#*<(%&$2.(&0:(3&%2&$2.0(20(;&-*(;&$*(K%*+*%*0'*(.0($F*('.*<2#$*0'*(.+( #K*%;J:*K*0:*0$(&#*<7&-#(&0:($F*2%(#*<7&-(F.#$#( An important assumption in my analytical model is that the fertility advantage of sexual females over gynogens does not depend on male abundance (see main text). I designed simulations to investigate the effect of relaxing this assumption. The simulations mimicked my analytical model except that each male attempted to mate with H females. While in the model described in the main text male “mate preference” was determined by the parameter a, in these simulations I replaced a with a mechanistically explicit parameter, v, that determines a male’s “visitation preference”. A given male was v times more likely to attempt to mate with a sexual female than a gynogen. Any female that was visited multiple times, either by the same male or multiple males, was fertilized with the male sire chosen randomly from among the males that had visited her. Note that when H # 1, v is roughly equivalent to a because female fertilization success is limited by male mating attempts. By contrast, when H > 1, male mate preference (measured as the relative rate of fertilization of sexual versus asexual females, i.e., a) is effectively a function of both H and v. Critically, when H > 1, the fertility advantage of sexual females over gynogens decreases with increasing male abundance. Each simulation began with 4000 gynogens and 6000 sexuals, with the population size (N = 10 000) held constant over time. The sexual population began with either one or two visitation preference genotypes. For simulations with only one preference type (i.e., no polymorphism), all sexual individuals had v = {1, 1.5, 2, 3, 4, 8, or 16}. For simulations that began with two preference types (i.e., those with polymorphism), half of the sexual individuals at the beginning had v = {1.5, 2, 3, 4, 8, or 16}, and the remainder had v = 1 (i.e., no preference). The viability of gynogens was 1 – s relative to sexuals, and the viability of any male with v > 1 was 1 – c. In the simulations explored, I set s =  125  0.25, c = 0.2, and H = 5 or 10. I ran 10 replicate simulations of 26 different parameter combinations for 1000 generations. The results of the simulations are reported in table A1, and the conclusions can be summarized as follows. Without polymorphism in male preference and with no male visitation preference (i.e. v = 1), gynogens take over and the system goes extinct. A relatively low visitation preference can prevent the gynogen from taking over, but, with low v, the sexuals remain at very low frequency. In my simulations, when v = 1.5, the sexuals were lost due to stochastic frequency fluctuations after being driven to very low frequency. The frequency of gynogens decreased with increasing v. With a high enough value of v, gynogens were lost due to stochastic frequency fluctuations. Thus, without polymorphism in male preference, there was only a limited range of values for v that allowed coexistence in the face of stochastic fluctuations (1.5 < v < 4 for H = 5 and 1.5 < v < 8 for H = 10). Even when the cost of preference was increased substantially with high v (e.g., c = 0.5), the gynogens did not gain enough of an advantage to persist (simulation data not shown). Note that the model by Heubel et al. (2009) also examines the effect of both male mating efficiency and mate preference using parameters akin to my H and v, respectively. The results of my simulations without polymorphism (described above) mirror the results of Heubel et al. (2009). By contrast, the existence of variation in male preferences increased the range of preference values allowing stable coexistence. This effect was observed when one male type was very picky and the other male type was not. In these cases, coexistence was permitted with parameter values for which coexistence was not seen in the absence of male preference variation (i.e., when 4 " v for H = 5 and when 8 " v for H = 10). On the other hand, when both male types had low preferences for sexual females, the system tended to lose one male preference type, as I would expect from my analytical model ! ! when both male preferences are weak (see figure 2.3). The mechanism allowing the maintenance of coexistence with variation in male preferences appears to be the same as that described in the main text: cycles in the frequency of preference types and gynogens appear that prevent the fixation of any one type (see main text). No such cycling is evident in simulations where coexistence is maintained without polymorphism. I thus  126  conclude that the presence of variation in preference increases the range of parameter values for which coexistence is possible. ( M*'F&02#;(.+('.*<2#$*0'*(20(FA"%2:.6*02'(9&$*%(+%.6#( Mate choice in hybridogenic water frogs: Hybridogenic water frogs (R. esculenta) require mating with a parental species (R. lessonae) for successful reproduction (Schultz 1969). This mating system is summarized in figure 2.1. All females (both hybridogens and parentals) prefer to mate with R. lessonae males (Abt & Reyer 1993; Engeler & Reyer 2001). The preference among R. lessonae females for mating with conspecific males has been postulated to be selectively motivated due to the exclusion of the R. lessonae genome when the hybridogenic frogs produce gametes (the motivation for the preference of R. esculenta females is obvious, as offspring of R. esculenta males and females do not survive beyond the larval stage). Interestingly, the males do not show a preference for mating with female R. lessonae, presumably due to the low cost of wasting sperm (relative to eggs) and to the disadvantage of mating with hybridogenic females being balanced by a general preference among male frogs for larger females (female hybridogens are larger and more fecund than sexual females) (Engeler & Reyer 2001). Model of bisexual hybridogen-sexual coexistence: In order to compare the requirements for coexistence in a gynogen-sexual system (main text) to the requirements for coexistence in the bisexual hybridogen-sexual coexistence, I analyzed a simplified version of a model by Hellriegel & Reyer (2000) that is analogous to my model. My analysis of Hellriegel & Reyer’s (2000) model has nonoverlapping generations and an infinite population size (hence, density effects on tadpole survival and differences in survival and recruitment between various life history stages are not included in my simplified version). Four different types of individuals are considered in the model: R. lessonae males and females (Lm and Lf) and R. esculenta males and females (Em and Ef). Rana lessonae males and females are assumed to occur in equal proportions so the frequency of both male and female R. lessonae is L/2. Rana 127  esculenta males and females may be found at different frequencies (Em and Ef, respectively). Female R. lessonae prefer to mate with R. lessonae males aL times more than R. esculenta males (aL $ 1). Rana esculenta females prefer to mate with R. lessonae males aE times more than R. esculenta males (aE $ 1). Because preference does not vary within either sexuals or hybridogens, costs of preference are not included in the model (such costs might contribute to the relative fitness of sexuals vs hybridogens, as measured by the parameter s). Genotype frequencies in each generation depend on the frequency of each type of mating, the proportion of each type of offspring produced from each type of mating, and the average clutch size of R. esculenta females relative to R. lessonae females, given by 1 - s (where  ) (see figure 2.1, chapter 2 text, and table A2).  Because no offspring are produced from mating between R. esculenta males and females, the following recursion equations are normalized by the sum of the offspring produced from all productive matings (Mt):  (4a)  (4b)  (4c)  Setting Lt+1 = Lt and Ei,t+1 = Ei,t and solving gives the following possible equilibria: {L = 1, Ef = 0, Em = 0}  (5a)  128  (5b)  The stability of each equilibrium was determined by setting L = 1 – (Ef + Em) and evaluating the eigenvalues of the characteristic polynomial of the Jacobian matrix given by equations (4b) and (4c). The non-coexistence equilibrium (5a) is stable when , which is satisfied for any strength of conspecific mate preference when  .  The coexistence equilibrium (5b) is valid and stable only when aL > aE and the following conditions are satisfied:  The range of parameter values that allow coexistence between bisexual hybridogens and a sexual parental species is summarized in figure A1. Persistent coexistence, in this case, requires that hybridogen fecundity (governed by the s parameter) is neither too high nor too low for a given degree of preference among R. lessonae females (the aL parameter). Unlike the gynogenetic system, stable coexistence between hybrids and parentals in the bisexual hybridogenic system does not require variation among sexual males in their degree of preference, which is not included in equations (4). Stability in the bisexual hybridogenic system arises due to the unique mating system of hybridogenic water frogs (see figure 2,1, chapter 2 text). When R. lessonae is common, R. esculenta males mate mostly with R. lessonae females, producing only female R. esculenta offspring and thereby causing the R. esculenta fraction of the population to increase. When R. esculenta is common, R. esculenta males mate mostly with R. esculenta females producing no offspring and thereby allowing the R. lessonae  129  fraction of the population to increase. Consequently, when the preference of R. lessonae females is greater than the preference of R. esculenta females for R. lessonae males (i.e., aL > aE), the hybridogenic water frog mating system generates negative frequencydependent selection against hybrids.  130  Table A1. Results of simulations showing the effects of sex ratio and variation in male mate preference on the coexistence of spermdependent asexuals and their sexual hosts. In these simulations, picky males were v times more likely to pair with (i.e., visit) a sexual rather than a gynogenetic female, and each male attempted H pairings (H = 5 in panel A, H = 10 in panel B). In the simulations with polymorphism in male mate preference, the least picky males had v = 1 (i.e., no visitation preference). All simulations ran for 1000 generations or until either the males or gynogens were lost (i.e., until extinction of loss of coexistence, respectively). The mean frequency of gynogens over the last 200 generations was averaged across replicates for each case where coexistence was maintained for > 1000 generations. Light shading indicates cases where coexistence was maintained for > 1000 generations without polymorphism. Dark shading indicates cases where coexistence was maintained for > 1000 generations with polymorphism in at least one replicate. Other parameter values were as follows: N = 10000, c = 0.2, s = 0.25 (see chapter 2 and appendix A for further explanation of parameters and simulation conditions). A (H = 5) Visitation preference of pickiest males (v): Mean generations to extinction or loss of coexistence  no poly.  Mean frequency of gynogens at equilibrium 1  no poly.  1  1.5  2  3  4  8  16  11.1  60.6  > 1000  > 1000  122.9  14.9  7.3  13.4  14.6  28.6  > 1000  > 1000  > 1000  0 (0)  0.68 (10)  0.36 (10)  0 (0)  0 (0)  0 (0)  0 (0)  0 (0)  0 (0)  0.19 (10)  0.16 (10)  0.14 (10)  1  1.5  2  3  4  8  16  12.1  55.7  > 1000  > 1000  > 1000  100.4  15.4  14.4  17.8  see 2  see 2  see 3  > 1000  0 (0)  0.84 (10)  0.73 (10)  0.66 (10)  0 (0)  0 (0)  0 (0)  0 (0)  0.73 (10) 2  0.66 (10) 2  0.11 (4) 3  0.13 (10)  poly. 0 (0)  poly.  B (H = 10) Visitation preference of pickiest males (v): Mean generations to extinction or loss of coexistence  no poly.  Mean frequency of gynogens at equilibrium 1  no poly.  poly.  poly.  0 (0)  1  Numbers in brackets indicate the number of replicates in which coexistence persisted for > 1000 generations. In all replicates the low preference allele was lost and the frequency of gynogens equilibrated to the value found without polymorphism (coexistence maintained for >1000 generations). 3 In 6 out of the 10 replicates the low preference allele was lost and the sexuals fixed, resulting in loss of coexistence (at 368 generations, on average). In 4 replicates, polymorphism was maintained and coexistence lasted for > 1000 generations. 2  131  Table A2. Frequencies of mating among sexual R. lessonae and the hybridogenic R. esculenta genotypes. The degree of preference for R. lessonae males by R. lessonae and R. esculenta females depends on the parameters aL and aE, respectively. Mating pair Female Male Lf  Lm  Lf  Em  Ef  !Lm  Ef  !Em !  Mating frequency 2 aL " L % $ ' zL # 2 & 1 L Em zL 2 aE L Ef zE 2 1 E f Em zE  Relative numbers of offspring types Lf Lm Ef Em !  !  0  0  0  0  1  0  0  0  !(1 – s)  !(1 – s)  0  0  0  0  " L %" L% zi = $ E m + ai '$ E f + ' # 2 &# 2&  !  !  132  Figure A1. Parameter space allowing stable coexistence of hybridogenic and sexual individuals. Parameter combinations within the frame outside the black boundary do not allow a polymorphism to exist. In this region of the figures (pure white background), either hybridogens dominate (i.e., E ! 1) and the community crashes, or sexuals dominate (i.e., L ! 1) and the hybridogens cannot invade. The darker shading represents a higher equilibrium frequency of hybridogens.  133  !""#$%&'()*(+,-".&$/(0#1,&.2( Table B1. Details of sampling locations and sample sizes for all Phoxinus samples used for this dissertation. Location NE BC NE BC NE BC NE BC NE BC NE BC NW Alberta NW Alberta NW Alberta SW Alberta SW Alberta SW Alberta SW Alberta SW Alberta SW Alberta SW Alberta SW Alberta SE Alberta N Saskatchewan S Saskatchewan S Saskatchewan NW Ontario NW Ontario SE Ontario SE Ontario SE Ontario SE Ontario SE Ontario SE Ontario SE Ontario SE Ontario SE Ontario  Waterbody Tsinhia Lake Parker Lake Beaver Lake Kiwigana River McConachie Creek Stanolind Creek Upper Pierre Grey's Lake Middle Pierre Grey's Lake Lower Pierre Grey's Lake Alford Lake Tay Lake Strubel Lake Ironside Lake Mitchel Lake Gasplant Lake Fiesta Lake Gunrange Lake Spruce Coulee Oscar Creek Loch Lomond Loch Leven ELA 114 ELA 375 Amikeus Lake Eos Lake Eucalia Lake Ringneck Pond Raven Lake Daisy Lake Little Eagle Lake Sunday Creek Pondweed Lake  Coordinates N 59° 33.53' W 123° 12.07' N 58° 49.59' W 122° 54.12' N 58° 58.60' W 123° 10.17' N 59° 24.41' W 123° 15.07' N 58° 53.48' W 122° 52.45' N 59° 4.32' W 123° 12.47' N 53° 54.44' W 118° 35.12' N 53° 54.17' W 118° 35.32' N 53° 54.51' W 118° 36.81' N 52° 3.91' W 115° 5.54' N 52° 3.73' W 115° 7.22' N 52° 12.16' W 114° 59.82' N 52° 14.76' W 114° 59.83' N 52° 12.94' W 115° 0.22' N 51° 59.18' W 114° 45.41' N 51° 59.69' W 114° 43.63' N 51° 59.26' W 114° 42.84' N 49° 40.71' W 110° 10.99' N 52° 46.36' W 107° 7.93' N 49° 39.35' W 109° 29.35' N 49° 39.83' W 109° 30.06' N 49° 40.44' W 93° 46.04' N 49° 44.67' W 93° 47.29' N 45° 35.29' W 78° 20.66' N 45° 35.45' W 78° 21.46' N 45° 34.65' W 78° 24.81' N 45° 35.00' W 78° 23.15' N 45° 35.28' W 78° 38.98' N 45° 39.54' W 78° 57.86' N 45° 38.86' W 78° 59.08' N 45° 35.22' W 78° 22.19' N 45° 27.71' W 78° 27.65'  Number of fish sampled (numbers of individuals with known COI haplotypes are shown in parentheses) P. eos P. neogaeus P. eos-neogaeus Cybrids Total 0 14 (5A, 1J) 15 (2A) 94 (2A, 1K) 123 (9A, 1J, 1K) 0 2 (2A) 17 (1A) 1 (1A) 20 (4A) 0 7 (4A) 4 (1A) 0 11 (5A) 0 1 (1A) 1 1 3 (1A) 0 4 (3A) 4 0 8 (3A) 0 0 17 (1A) 0 17 (1A) 0 0 17 11 (1A) 28 (1A) 0 0 27 0 27 0 0 5 (2A, 1I) 23 28 (2A, 1I) 0 12 (2A) 28 (1A) 0 40 (3A) 0 1 (1A) 31 (2A) 0 32 (3A) 0 0 24 (1A) 17 41 (1A) 0 0 39 (1A) 1 40 (1A) 0 0 32 4 36 0 3 (2A) 18 29 50 (2A) 0 1 (1A) 28 (4A) 35 (1A) 64 (6A) 0 1 (1A) 3 29 33 (1A) 0 0 0 32 32 0 8 (3A) 3 (2A) 0 11 (5A) 0 0 23 (1A) 5 28 (1A) 0 0 7 (1A) 14 (1A) 21 (2A) 0 8 (2A) 1 (1A) 11 (2A) 20 (5A) 0 1 (1A) 26 (4H) 1 (1H) 28 (1A, 5H) 12 5 (4A) 0 0 17 (4A) 4 9 (3A, 1G) 5 0 18 (3A, 1G) 10 12 (2A) 2 (2A) 2 (2A) 26 (6A) 10 4 (3A) 7 (1A) 2 23 (4A) 15 (1 eos) 0 32 (2A) 0 47 (2A, 1 eos) 7 (2 eos) 0 25 (2A) 3 (1A) 35 (3A, 2 eos) 17 (2 eos) 0 23 (1A) 0 40 (1A, 2 eos) 9 0 2 (1A) 1 12 (1A) 14 0 8 0 22  134  Table B1 (continued).  Location PEI N NB N NB N NB S NB  Waterbody Morell River Catamaran Lake Little McKendrick Lake Rocky Brook Lake Adelaide Lake  Coordinates N 46° 17.84' W 62° 44.90' N 46° 51.10' W 66° 20.51' N 46° 51.11' W 66° 20.48' N 46° 59.51' W 65° 34.38' N 45° 19.14' W 66° 37.87'  Number of fish sampled (numbers of individuals with known COI haplotypes are shown in parentheses) P. eos P. neogaeus P. eos-neogaeus Cybrids Total 0 0 0 5 (2L) 5 (2L) 0 0 35 (2A) 8 (1A) 43 (3A) 0 0 17 (2A) 9 (2A) 26 (4A) 0 0 13 (2A) 11 (2A) 24 (4A) 0 36 (8F) 0 11 (4A) 47 (4A, 8F)  135  

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.24.1-0072467/manifest

Comment

Related Items