THE EVOLUTION OF GEOGRAPHIC PARTHENOGENESIS AND THE PERSISTENCE OF ASEXUALITY IN TIMEMA WALKING-STICKS

Jennifer Heather Law

B.Sc., University of Waterloo, 1998

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE

in the Department of Biological Sciences

O Jennifer Heather Law 200 1 SIMON FRASER UNIVERSITY July 2001

Al1 rights reserved. This work may not be reproduced in whole or in part, by photocopier or other means, without permission of the author. The author tias granted a non- L'auteur a accordé une licence non exchive licence dowing the exclusive permettant 9 la National Library of Canada to Bibliothèque nationale du Canada de reprduce, loan, disttiiute or sen reproduire, prêter, distribuer ou copies of bis thesis in microform, vendre des copies de cette thèse sous paper or electronic formats. la forme de microfiche/nlm, de reproduction sur papier ou sur fonnat électronique.

The author retains ownershrp of the L'auteur conserve ta propriété du copyright in tbis thesis. Neither the droit d'auteur qui pdgecette thèse. thesis nor substantiai extracts fiom it Ni la thèse ni des extraits substantiels may be printed or otherwise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation. Abstract - Phylogenetic studies of parthenogenetic Lineages and their sexual counterparts have been invaluable to analysing questions sumiuading the evolution of sex. Understanding the geographic patterns of parthenogenesis, as weU as determining the evolutionary age of asexuality, should help us infer critical selective factors and the temporal deunder which asexuality and sex evolve. With five known parthenogens and well-studied ecology, Tirnema waking-sticks are a useful system for studying these questions. Endemic to California,

Timemn exhibit the common pattern of geographic parthenogenesis, and previous phylogenetic work suggested that asexual limages were ancient. Neighbour-joining and maximum parsimony analyses of 4Mbp of mitochondrial COI were used to infer general phylogenetic relationships, resulting in three major geographic subdivisions, a Northem clade, a Santa Barbara clade, and a Southem clade. A nested cladistic analysis, comparing intra- and interspecific haplotypic variation on a geographic scale, revealed that the overall pattern of geographic partheniigenesis could be actributed to historical range expansion. Two parthenogens, I: douglosi and T. sheprdi, were very dosely related to each other (0.31~

0.35% divergence) and were evolutionarily young (O. 1- 0.5 million years). Although T. moriikensis and T. tahoe were each monophyletic, evidence for mtiquity, or a lack thereof, was weak since only one population of each was sampled, and within population divergence was relaiively low (0.63 f 0.41 and 0.51 f 0.25 respectively). By contrast, the parthenogen

T. genevieve was inferred to be an ancient asexud (1.0 - 2.2 million years). Timem is exceptional in having both short-term and long-rem persistence of parthenogens, as well as exhibiting strong geographic patterns, and thus it provides the opportunity to critically compare hypotheses about dispersal rates, cornpetitive ability, and conditions that favout sex over asexuality. Acknowledgements

Bernie Crespi, my senior supervisor, guided me through this degree. Cristina

Sandoval was instrumental in showing me the ropes in California, and introduced to me the wonderful world of Timema walking-sticks. Mary Betbee provided me with valuable comments that helped to enhance my work. Mike Smith could always be counted on for words of encouragement, someihing I needed more often than not.

1 give my utmost thanks to my drivers and field assistants Tahirih Rempel(1999) and

Ayala Knott (2000), without whom 1 would not have ken able to conduct the extensive sampling that 1 did. 1 am amazed that we managed to survive thousands of kilometres cramped in a van, often with no idea when or where we would be camping next.

Members of the Behavioural Ecology Research Grcrup not only provided me with scientific insight, but also contributed greatly to my social life here at SFU. Tearn Crespi supported me through thick and thin - may the sigh jar be emptied for a good cause. In particular, Tom Chapman, with his infinite words of wisdom and uncanny ability to tell stories, made me laugh more than 1 tbought possible.

Finally, 1 thank Stevo DeMuth for his endless patience, support, and love. Table of Contents .. Approval Page ...... il ... Abstract ...... ui Acknowledgements ...... iv

Table of Contents ...... v

List of Tables ...... vi

List of Figures ...... vii

Chapter 1: General Introduction...... 1

Chapter 2: The Evolution of Geographic Parthenogenesis in Timema ...... 5

Chapter 3: The Persistence of Asexuality in Timema ...... 7 1

Chapter 4: Concluding Remarks ...... 87

Appendix A ...... 91

Appendix B ...... 97 List of Tables

Table 1: Sexuai: asexual (parthenogenetic) Timema pairs and respective host plant usage ...... 11

Table 2: Sexual Timema species (that have no known asexual counterpart) and respective host plant usage ...... 14

Table 3: Chi-square test for geographic structure for the Northern cladograrn with associated probabilities ...... 41

Table 4: bference chain for nested geographical analysis of the Northern cladogram ...... 44

Table 5: Chi-square test for geographic smctuiie for the Santa Barbara cladograrn with associated probabilities ...... 46

Table 6: inference chah for nested geographical analysis of the Santa Barbara cladograrn ...... -49

Table 7: Chi-square test for geographic structure for the Southem cladogram with associated probabilities ...... 50

Table 8: inference chain for nested geograpbical anaiysis of the Southem cladogram...... 53

Table 9: Average sequence divergence within parthenogenetic lineage and to closest sexual relative for mtDNA COI ...... 77

Table 10: Estimated age range of parthenogens based on 2% divergence per million years ...... 8 1 vii

Li of Figures

Figure 1: Geographic distributions of Timema ...... 12

Figure 2: General phy logenetic relationships within the Thema genus obtained through neighbour joining (NJ) and maximum parsimony (MP) analyses ...... 24

Figure 3: Phylogenetic relationships among Timema species in the Northem clade ...... 26

Figure 4: Phylogenetic relationships between Tbcristinue and T. monikensis in the Santa Barbara clade ...... -28

Figure 5: Phylogenetic relationships among Timema species in the Southern clade ...... 31

Figure 6: The Nonhern cladogram for Timemu ...... 34

Figure 7: The Santa Barbara cladogram for Timema ...... 36

Figure 8: The Southern cladogram for Timema ...... 38

Figure 9: Results of the nested geographic analysis of Timema COI haplotypes for the Northem cladogram ...... 42

Figure 10: Results of the nested clade maiysis for T. cristinue and T. monikensis in the Santa Barbara cladogram ...... 47

Figure 1 1: Results of the nested geographic analysis for Timema taxa in the Southem cladograrn ...... 51 Theoreticaliy, an asexual population cmteproduce at twice the rate of a sexual population, yet most organisms reproduce sexualiy (Be& 1982; Barton and Charlesworth,

1998). The result is a pamdox - why sex? - one of the outstanding questions in evolutionaq biology. How and why did sex corne to predominate? Understanding the geographic distributions of asexuaiity and sexuality and how variability, both genetic and ecological, affects sexual and asexual populations may provide some insight into the paradox of sex.

Timema walking-sticks (Insecta: Phasmatodea: Timernatodea) are an ideal system for studying the paaâox of sex for the reasons: 1) There is a high incidence of both sexual and asexual species (22 described species, 5 of which are parthenogenetic), and each asexual species has a morphologicaily similar close sexual relative; therefore multiple cornparisons can be perfonned (Sandoval et al., 1998; Vickery and Sandoval, 2001, submitted). 2)

Timema exhibits geographic parthenogenesis, a common pattern whereby asexual species have a more northerly distribution with respect to sexual counterparts (Sandovai et al, 1998).

3) Timema ecology is well characterised (Vickery, 1993; Sandoval et al., 1998) and the phyIogeny of 13 species has ken inferred using mitochondnai COI sequence data (Sandoval et al., 1998).

Timema walking-sticks are herbivorous uisects found along the West Coast of North

America in geographicaiiy isolated, mountainous areas (Vickery, 1993; Saudoval et al.,

1998). In Arizona, species specialise on juniper, wMe in California Timema feed primarily on chaparral vegetation with some species found on fir (Sandoval and Vickery, 1999). Each

Timema species is highly cryptic to its respective host plant vegetation and most species are geographicaily locaiised (Vickery, 1993). Since these organisms are wingless, they have limited dispersal compared to other insects, moving primady at night, and king stationary during the day (Vickery, 1993; Sandovai, 1994).

Male taxonomy, in particular male genitalia, is the primary factor used to distinguish species (Vickery, 1993). Consequently, the identification of ail-female taxa has been difficult. Body colour and morphology, host plant use, and lack of males (male mate guarding is prevalent in sexual species) have ken used to identify parthenogenetic species

(Vickery, 1993). However, body colour cm be quite variable, with many species having

very different colour morphs (Vickery, 19931, and it may not be a good measure for

determining species relationships (Vickery and Sandoval, 1997). Therefore, the first goal of

my thesis is to expand on previous phylogenetic work (Sandoval et al., 1998) and further

resolve the phylogenetics of Timemu. If taxonomie classifications are accwate, then each

described species, both sexual and asexuai, should fonn a monophyletic clade. This

phylogeny is then used to address questions in two main areas of research on the evolution of

sex debate: 1) the geographic patterns of asexuality relative to sex, and 2) the evolutionary

persistence of asexuality.

Chapter 2 presents the overall phylogenetic relationships and uses these relationships

to evaluate the different spatial scales of sex and asexuality. Geographic parthenogenesis

refers to the observation that parthenogens tend tr, be more northerly, more widespread, at

higher elevations, and in harsher habitats han their sexual counterparts (Glesener and

Tilman, 1978; Bell, 1982; Peck er al., 1998). In Thema, asexuals are more northerly, and in

some cases more widespread, than iheir morphological sexual counterparts. Ecological

competitive hypoiheses predict Timema asexuds are more northerly is due to differing cornpetitive abilities between parthenogens and sexuals (Glesener and Tian, 1978; Barton and Charlesworth, 1998). Conversely, demographic hypotheses predict that Timemu parthenogens are more northerly because they have a two-fold reproductive advantage and are therefore better dispersers (Bell, 1982; Peck et al., 1998). Since Timema are wingless, spatial patterns are more likely due to history than to present &y effects. The genus Timema is thought to have onginated in the southeni part of its current distribution and since expanded northwards (Sandovai et al., 1998). Under a demographic scenario, Timema asexuals should be at the forefront of this expansion and the pattern of geographic parthenogenesis may therefore be a result of range expansion. To test this hypothesis 1 used a novel statisticai method, nested cladistic analysis, to assess the evolutionary history of

Timema.

Chapter 3 expands on the phylogenetic work presented in Chapter 2 to evaluate the antiquity of asexuals. The evolutionary age of asexuais has ken a long-standing conundrum for the evolution of sex debate (ludson and Nomark, 19%). In the short term, asexuality confers a strong two fold reproductive advantage (Bell, 1982; Maynard Smith, 1992).

However, long term persistence is puzzling since pabenogens lack the genetic variability that sexuality confers (Bell, 1982). Consequently, asexuality should be an evolutionary dead end. Previous phylogenetic work suggested that Timemu asexuals are ancient (Sandoval et al., 1998). Two measures of evolutionary age are used: 1) a minimum age for each parthenogen is calculated based on intraspecific divergence; and 2) a maximum age is caiculated from each parthenogen to its closest sexual relative.

Fmaily, in my concluding remarks in Chapter 4,I provide the larger context of my study to the evolution of sex debate. Barton, N.H., and B. Charlesworth. 1998. Why sex and recombination? Science 281: 1986- 1989.

Bell, Graham. 1982. The Masterpiece of Nature: The Evolution and Genetics of Sexuality. Croom Helm, London.

Glesener, R. R. and D. Tilman. 1978. Sexuality and the components of environmental uncertainty: ches from geographic parthenogenesis in terrestrial animals. The American Naturalist 112: 659-672.

Judson, O. P., and B. B. Normark. 1996. Ancient asexual scandais. Trends in Ecology and Evolution 11: 4 1-46.

Maynard Smith, J. 1992. Age and the unisexual lineage. Nature 356: 661-662.

Peck, J. R., J. M. Yearsley, and D. Waxman. 1998. Explaining the geographic distributions of sexual and asexual populations. Nature 391: 889 - 892.

Sandoval, C. P. 1994. The effects of the relative geographic scales of gene fiow and selection on the morph frequencies in the walking-stic k Timema cristinae. Evolution 48: 1866- 1879.

Sandovai, C. P. and V. R. Vickery. 1996. Timema douglasi (Phasmatoptera: Timematodea), a new parthenogenetic species form southwestern Oregon and northem California, with notes on other species. The Canadian Entomologist 128: 79-84.

Sandoval, C. P. and V. R. Vickery . 1999. Timema comani (Phasmatoptera: Timematodea) a new species from Arizona and description of the female of Timema ritensis. Journal of Orthoptera Research 8: 49-52.

Sandoval, C., D.A. Carmean, and B. J. Crespi. 1998. Molecular phylogenetics of sexual and parthenogenetic Timema walking-sticks. hoceedings of the Royal Society of London B 265: 589-595.

Vickery, V. R.. 1993. Revision of Timema Scudder (Phasmatoptera: Timematodea) including three new species. Canadian Entomologist 125: 657-692.

Vickery, V. R. and C. P. Sandoval. 1997. Timema bartmani (Phasmoptera: Timematodea: Timematidae), a new species from southem California. The Canaàian Entomologist.129: 933-936.

Vickery, V. R. and C. P. Sandoval. 2001. Description of three new species of Timema (Phasmoptera: Timematodea: Timernatidae). Submitted to the Lyman Entomological Museum and Research Laboratory. Chapter 2: The Evolution of Geagraphic ParknogenesLP in Tirnenia

Phylogeographic studies have considerably deepened our understanding of geographicai and evolutionary relationships between and within species (Bermingham and

Moritz, 1998; Avise, 1994; Avise, 2000). Traditionai snidies of geography and phenotypic diversity have relied mainly on correlative data, resuiting in poor inference about the underlying genotypic patterns (Templeton and Sing, 1993). Extensive molecular and phylogenetic data as it relates to species distributioas dlows us to finally look at genetic variation on a spatial scale. However, only with the advent of coaiescence theory and cladistic analyses have we been able to critically test hypotheses about the history of species distributions.

The geographic pattern of asexuals relative to sexual counterparts has been a long- standing conundrum in the evolution of sex debate. Geographic parthenogenesis, fust noted by Vandel in 1928, refers to the common pattern of asexuals king more northerly, more widespread, in hmher habitats, or at higher elevations than sexuai relations (Bell, 1982;

Stearns, 1987; Peck et al., 1998). Why is geograpbic parthenogenesis a common pattern?

This question has been very dificult to address since the best explanation is one that combines historical geographical, ecdogical, and glaciation eveots with the assumption that asexuals have better dispersai abiity than sexuais and are better able to persist at the edges of a geographic range (Beil, 1982; Stems, 1987). Genedy speaking, events or processes leading to speciation and the production of asexuals, coupled with cornpetitive and demographic differences between sexuals and asexuals, have led to the geographic pattems we observe today. Understanding the distribution of asexuality relative to sex should be important in determining why sex is maintained in most species, and why it evolved in the first place.

Tests to distinguish between altemate hypotheses for the advantages of sex have been difficult to formulate (Barton and Charlesworth, 1998). Cansequendy, many cases of geographic parthenogeaesis have been explained qualitatively by comparing geographic patterns of asexuals relative to sexuals. However, simply noting relative distributions does not help distinguish htween possible explanadons for the evolution of sex. Hypotheses for the pattern of geographic parthenogenesis can be grouped into two alternative categories: 1) ecological competition hypotheses, and 2) demographic hypotheses. Competition between asexuality and sex is thought to play an important role in the establishment of asexuality

(Cuellar, 1977; Glesener and Tilrnan, 1978; Barton and Charlesworth, 1998). Under an ecological scenario, asexuals are better cornpetitors or have an advantage over sexuals under certain ecological conditions. Asexuality is maintained in close proximity to sex due to competition (Cuellar, 1977). in this case, the patterns of geographic parthenogenesis are due to differences in cornpetitive abiiity as they relate to ecology.

Demographic effmts are linked to the two-fold advantage of asexuality. if a population is expanding and an individual has a reproductive advantage, then the "new" area

WUhave more of that individuai's offspring. As weli, parthenogenesis may be an advantage in a colonising species since a single individual can establish a new population (Bell, 1982;

Gemstsen, 1980; Stems, 1987; Peck et al., 1998). Asexuals may also be better able to persist at the edges of a geographic range or in marginal habitats (Bell, 1982, Stems, 1987, Peck et al., 1998). When population densities are low and cesult in mate limitation, for example at the edge of a geographic range, there should be a higher proportion of individuals with parthenogenetic ability because asexuais are not Limited by the necessity of finding a mate (Bell, 1982; Peck et al., 1998). Those individuais found where population density is high, for example at the centre of a range, would have no such mate limitation and consequently sex would predorninate (Bell, 1982, Peck et al., 1998). With demographic effects, the pattern of geographic parthenogenesis is due to the ability of asexuals to spread to new areas faster than sexuals.

Combined with competitive or demographic effects, selection due to catastrophic events rnay contribute to the realised location of asexuality (Cuellar, 1977). As speciation events give rise to asexuals, where, geographically, will these lineages be located?

Glaciation, fire, flooding or changes in climate may promote speciation and allow asexuality to flourish in sites devoid of sexual counterparts (Cuellar, 1977). It has been noted that asexuals appear to colonise once-glaciated areas (Bell, 1982; Stems, 1987). Geologic occurrences such as mountain building or changes in sea level may also be critical to speciation events by opening up new habitat for species to occupy. After glaciation or geologic events, sexuais and asexuais may invade new areas, but due to cornpetitive or demographic effects, asexuality may occupy areas more rapidly (Cuellar, 1977; BeU, 1982).

Thema walking-sticks are an ideal system for studying questions about geographic parthenogenesis. There are five described parthenogenetic species, each with a morphologically close sexual relative (see Table 1 for list of sexual: asexual pairs) (Vickery,

1993; Sandoval and Vickery, 1996; Sandoval et al., 1998; Vickery and Sandovai, 1999). The ecology of this system is well characterised - each parthenogen feeds on similar host plants as its sexual counterpart, with asexual populations being ail female. Of the parthenogens, four (T. douglasi, T. shepardii, T. genevieve, and, T. tahw) fit the criteria of king more northerly in distribution than sexual counterparts, with the fiftb, T. monikensis, being only slightly south-east of its closest sexual (Figure 1). T. douglasi and T. shepardii are aiso geographically widespread relative to their respective sexual counterparts.

Previous phylogenetic work suggested that the genus Timema originated in the southern part of its cunent distribution about 20 million years ago and expanded northward

(Sandovai et al., 1998). Since most Timema are restricted to mouniainous areas, the ages of mountain ranges provided the basis for an upper age limit on Timema (Sandoval et ~1.~1998).

The northward expansion of the genus over time is thought to have followed changes in distributions of host plants. Mountain building would have resulted in the formation of physical barriers and new habitats for Timema to occupy, thus promoting the process of speciation (Sandoval et al., 1998).

Asexuals tend to colonise once-glaciated areas, leaving their closest sexual relative behind in the process (Bell, 1982; Stems, 1987). The onset of the most recent series of glaciations in North Arnerica 2 - 3 million years ago may have been a chmatic trigger for speciation (Sandovai et al., 1998). Though the glaciers did not reach California, glaciation events would have caused extensive cwling in California, resulting in host plants moving to lower altitudes (Sandoval et al., 1998). Consequently, Timema would have been able to disperse between mountain ranges during this tirne, with subsequent warm periods isolating

Tirnemu populations on different mouutains (Sandovai et al., l998), and leading to speciation. Sexual and asexual species should both exhibit range expansion under this scenario. if asexuals have higher dispersal ability and are better able to persist at the edges of a geographic range, as theory suggests, then Timensn parthenogens should be at the leading edge of the overall genus expansion northward. The pattern of geographic parthenogenesis may therefore be a result of range expansion.

The purpose of this paper is twofold. First, phylogenetic studies of Sandovai et al.

(1998) are extended to include recently collected and newly discovered species, and by the inclusion of extensive intraspecific sarnples. This phylogeny is then used to further evaiuate the evolutionary history of this genus and describe general biogeographic patterns. Each described asexual species is expected to represent an independent origin of parthenogenesis, king monophyletic and closely related to respective sexuai counterparts. Second, to test the hypothesis that geographic parthenogenesis in Timema is the result of range expansion, a nested cladistic analysis is conducted. This type of analysis has the ability to discriminate between the alternative hypotheses of range expansion, allopatric fragmentation, and restricted dispersal (restricted gene flow in sexual species) (Templeton et al., 1995).

Furthemore, haplotype networks are powerful tools for estimating intraspecific relationships

(Posada and Crandail, 2001). Since Thema parthenogens should be closely related to their sexual counterparts, each sexual: asexual pair can be treated as a separate intraspecific relationship. The information contained wiihtn haplotype networks, as a measure of the genetic structure between closely related individuals, can be used to test for historical geographic associations. Each sexud: asexuai pair can therefore be assessed for evolutionary

patterns leading to geographic parthenogenesis. Materiais and Metbods

. .. . rimema Taxonomv and Geoeraohic Distributiow

Timema are herbivorous stick insects, most of which inhabit chaparral vegetation

(Vickery, 1993; Sandoval et al., 1998). This genus belongs to the order Phasmatoptera and is a sister group to the rest of the phasmatids (Vickery , 1993; Vickery and Sandovai, 200 1 subrniîted). Timema taxonomy has been based mostly on differences in male genitalia

(Vickery, 1993; Vickery and Sandoval, 2001 submitted). Consequently, the identification c parthenogenetic species has relied on a combination of host plant use, colour morphology, and lack of maies (Vickery, 1993; Sandovai and Vickery, 1996; Vickery and Sandovai, 200 submitted). For clarity, sexuals species are defmed by the biological species concept while asexuals species are parthenogenetically reproducing moaophyletic populations.

Each of the five parthenogenetic Timema species has a close morphological sexual counterpart (Table 1) and each asexual species is found on a sirnilar host plant as its sexuai.

T.douglasi had previously been paired wiih T. californicum (Sandoval et al., 1998), but a closer morphological sexual, T. poppensis, bas since been sampled (Vickery and Sandoval,

1999). The T. poppensis: T. douglasi pair feeds pnmarily on Douglas-fir (Pseudotsuga mentiessi) and females of these taxa are morphologically indistinguishable. T. douglasi is geographically widespread and more northerly in distribution than T. poppensis; although these taxa are both found in coastal mountain areas and overlap in distribution (Figure 1). T. shepardii ,a recently described parthenogen, has been paired with T. califomicum (Vickery and Sandoval, 1999). Although similar in body morphology and predominant colouration, Table 1: Sexual: asexual (parthenogenetic) Timemu pairs and respective host plant usage. Sexuai and asexual counterparts are morphologically sidar and highly cryptic to primary host plants (the fust host plant Listed).

Sexual Species Sexuai Species Morphological Asexual Host Plant Parthenogenetic Species Counterpart Host Plant T. poppensis A, B T. douglasi A T. califomicum C, D, ci T. shepardii C T. cristinae D, E, H T. monikensis D, 1 T. podura E, G T. genevieve E T. bartmani F T. tahoe F

Host Plants A = Pseudotsugu menziesii (douglas-fir) B = Sequoia sempervirens (Califocnian redwood) C = Arctostaphylos spp. (manzanita) D = Ceanothus spp. (ceanothus) E = Adenostoma fasiculatum (chamise) F = Abies concolor (white fir) G = Quercus spp. (oak) H = Heteromeles arbutifolia (toyon) 1 = Cercocarpus spp. 1 California

North

A-T ntensis -

Figure 1: Geographic distributions of Timema (adapted fiom Sandoval et al., 1998). The newly descnbed species T. morongensis bas been coiiected West of T. chumash but its distribution is unknown. T. shepardii is a specialist on manzanita (Arctostaphylos spp.) while T. cal$micum has broader host plant use. T. shepardii ranges geographically from just above San Francisco nortb to the Oregon border and its southernmost limit is more northerly than the distribution of T. califomicum (Figure 1). These species are also found in coastal mountainous areas but distributions do not overlap for this species pair.

T. monikensis has been described as the asexual relative of T. cristinae (Vickecy and

Sandoval, 1998). T. monikensis is the only parthenogen that does not fit the pattern of geographic parthenogenesis - its distribution is highiy restricted and slightly south-east of T. cristinae (Figwe 1). Both species feed on Ceamthus spp., but T. monikensis also feeds on

Cercocarpus spp., while T. cristirta ahfeeds on chamise (AdenostomfusticuIatum} and toyon (Heteromeies arburiolia). T. cristinue and T. monikensis are geographically restricted to the Santa Ynez Mountains and Santa Munica Mountains respectively. As stated in

Sandoval et ai. (1998), T. genevieve is the asexuai relative of ?: podura and T. tahoe is the asexuai relative of T. hrtmani. The parthenogen T,genevieve is geographically more northedy than T. podura and found east of San Francisco (Figure 1). T. podrrra bas mainly been sampled from the San Bernardino Mountain Range, aithough its disicibution ranges from the MexicanlCalifornia border to northsast of Los Angeles. Both T. genevieve and T. pudura feed on chamise, and although this host plant is widely distributed, these species have not been sampled hmthe geographic areas between their mappd distributions. Females of

T. tahoe and T. bartmani are indistinguishable from one another and highly cryptic to kir host plant, white fu (Abies concolor). Both are geographically restricted, with T. rahoe king ailopatric with, and far more northerly than, T. 6u-i (Figure 1). T. bartmani is Table 2: Sexual Timema species (that have no known asexual counterpart) and their host plant usage. Note: Species marked by ** are not included in the phylogenetic analyses. T. co@ni proved difficult to sequence and T. morongensis was discovered after this work was perfomed. ***Although T. morongensis was collected on Eriogonum, it is unknown if this plant is its host.

Sexuai Species Host Plant(s) T. knulli B, D T. landelsensis C T. petita D T. boharti D, E T. chumash G, 1 T. ritensis K T. nevadense K T. dorotheae D T. nakipa C, D, E, G T. co%ni** K T. morongensis** J***

Host Plants A = Pseudorsuga menziesii (douglas-fir) B = Sequoia sempervirens (Califomian redwood) C = Arctostaphylos spp. (manmita) D = Ceanothus spp. (ceanothus) E = Adenostoma fasicularum (chamise) F = Abies concolor (white fu) G = Quercus spp. (oak) H = Hereromeles arbutifolia (toyon) 1 = Cercocarpus spp. J = Eriogonum sp. K = Juniperus spp. (juniper) restricted to the San Bernardino Mountain Range while T. tahoe is found only around Lake

Tahoe.

In addition to the sexual: asexual pairs described above, the host plant use and geographic distributions of other sexual species, including the recently discovered sexuals, T. rbiulli, T. petita, and T. landelsensis, are given in Table 2 and Figure 1. T. knulli Strohecker is a previously described species known only from pinned specimens, leading Sandoval and

Vickery (1996) to suggest it was a synonym of T. califomicwn. However, in the spring of

1999, this species was collected at Big Creek Reserve and T. knulli has since been re- described as a valid species (Vickery and Sandovai, 2001, submitted). T. knulli is found on two different host plants - Sequoia sempervirens (Californian redwood) and Ceanothus thyrsiflorus (Ceanothus). T. landelsensis and T. petita are new, previously un-described species. T. peiita was collected on the coastai region slightly south of T. knulli. The host plant of T. petita is also Ceanothus thyrsijlorus. T. petita is one of the srnailest sexual species (14-24mm), dark green with brown stripes (Vickery and Sandoval, 2001, submitted).

The male genitalia of T. petita and T. knulli are very sirnilar, with T. petita having smaller genitaiia (Vickery and Sandoval, 2001, submitted). T. landelsensis was also collected at Big

Creek Reserve but at a higher elevation than T. knulli. Morphologically, however, these species are quite distinct - T. knulli is as described above, wMe T. landelsensis is sdl(18-

21mm), pale green, with distinct male genitalia and fernale subgenital plates thai were quite different from any other Timema species (Vickery and Sandoval, 2001, subrnitted). Tirnenia walking sticks were collecied throughout California from as many geographically widespread sites as possible. The facus of coiiecting was on extensively sampling the ssexual: asexual pairs from as many geographically separate populations as possible. Since the nested cladistic analysis depends critically upan sampling, as well as finding the closest sexual relative for each asexuai, the goal was to accurateiy represent extant genetic variation. See Appendix B for a list of coilection sites. For each site, Tirnema were shaken from th& respective host plants using a sweep net. Long term mak mate guarding bebaviour is quite prevaient in rhis genus and adult females of sexual species are rarely found without a male nding on their back (Bartman and Brock, 1995; Sandoval and

Vickery, 1996; Vickery and Sandoval, 2001, submitted). Generdly, a minimum of five adult females with no males is suficient to conclude the speçies is parthenogenetic (Sandoval and

Vickery, 1996). Therefore, populations where only femaies and no males were found after one hour of sampling or where females were known to reproduce parthenogenetically were deemed asexud. if fewer than five females but no males were sarnpled from a previously un-sampled locality, geographic location (e.g. a norîheily population) and host plant use were used to categorise as asexud. Four males (and 16 females) were collected in the T. monikensis population but it is not known if these males were viable. It is not uncornmon for partheuogenetic species to occasionally produce sterile males (Soumalainen et d., 1987).

However, since T. monikensis is known to reproduce parthenogenetically (c biblio >), this species was still included in comparisons. AU sexual species collecred had an equal or higher proportion of desthm females and it was assumed that these males were reproductively viable.

Walking-sticks were kept alive in jars until they could be processed for DNA sampling. For each collected insect, 1-3 legs were removed with forceps and dried in silica gel, to be used for DNA samples. If the legs were small, the head was also removed and stored in silica gel. The body was kept in 75% ethanol as a voucher specimen and for species identification.

DNA Sarn~les

DNA was extracted from silica gel stored legs. Usually one leg was sufficiently large enough for this procedure, however in the case of smailer species or juvenile individuals multiple legs from the same individual or the head may have been used. The tissue was crushed with a sterile glass pipette and suspended in 900 CIL Lifton buffer (0.2 M sucrose,

0.05 M EDTA, O. 1 M Tris, 0.5% SDS). The DNA was extracted using a phenol chloroform proiocol and 70% ethanol precipitation. PCR was performed using combinations of the mitochondriai COI primers S2183 (CAA CAT TTA TIT TGA TTï TTï GG) and S2 195

(TTG Am 'ITT TGG TCA TCC WGA AGT) with A2566 (CCT ATA GAI ART ACA TAA

TG) and A3014 (TCC AAT GCA CTA ATC TGC CAT ATT). PCR product was processed using EX01 and SAP. Big Dye Cycle Sequencing was used to sequence a fragment about

450 bp long. Seauence Analvsis and Phvlogenetic Reconstruction

Sequences were aligned by eye using the program Se-AI version 2.0 (Rambaut, 1996-

2000). PAUP 4.0b8 (Swofford, 2000) was used to analyse the data by neighbour joining

(NJ) (under a Kimura two-parameter mode1 of evolution) and maximum parsimony (MP) searching (heuristic search, simple addition of sequences, TBR branch swapping). Maximum likelihood was too computationally intense for the total dataset. Trees were assessed for robustness using bootstrapping (1000 for NJ and 100 for MP) and by comparing major clade divisions under different models of evolution. As in Sandoval et al. (1998) theoutgroups were used - the phasmatids Baculum extradentatm and Anisomorpha birprestoidea, and the coc kroac h Blatella germanica.

Nested Cladistic Analvsis

Nested cladistic analysis (NCA) is a method that relates genetic and geographic distance. The first step is to detemine the 95% statistical parsimony limit and estimate the haplotype network, as given by the algorithm in Templeton et al. (1992). The program TCS

(Clement and Posada, 2000) was used to estimate the mtDNA haplotype network for each sexual: asexual pair and to calculate the 95% statistical parsimony plausible bit. This limit is a measure of how many steps cm separate two haplotypes with confidence that parsimony is supported.

Once the haplotype netwodcs are consûucted, the next step is to construct a nested cladogram, the full desof which are given in Templeton et al. (1987), and Templeton and Sing (1993). The following is a synopsis of this method: starting at the tips, move in one step and join those taxa to form 1-step clades. If two or more haplotypes converge on the same internai haplotype then ail become joined together into a 1-step clade. Once this initial step is done, any internai haplotypes one step in fiom nested clades becomes tips and the procedure is repeated until al1 haptotypes are contained within 1-step clades. This process is repeated with the 1-step clades to form 2-step clades, and so on, until the whole cladogram is connected into an n-step clade. if there are any loops in the haplotype network (cladogram uncertainty) then al1 x-step clades within the loop are connected into a singe x+l step clade.

If there are haplotypes that cannot be joined under the 95% statistical parsimony, the rules are fust applied to that data which meets the 95% criteria to form b-step networks. The 6-step networks can then be assessed for evolutionary smcture using maximum parsimony and joined using predictions from coalescence theory as well as evaiuating non-parsimonious connections (Templeton et al, 1992; Templeton and Sing, 1993; pers. comm. with

Templeton, 200 1).

Following the nesting procedure, each clade is tested for geographicai structure,

against the null hypothesis that no geographical associations exist (see Templeton et al.,

1995, for full methodology). The two main test statistics are the clade distance, Dc (X),and

the nested clade distance, Dn (X). Dc (X)is a meam of how geographically widespread

haplotypes are within the nested n-step clade X, and Dn (X) is a measure of how

geographicaiiy distant haplotypes are within the n-step clade X to al1 haplotypes in the clade

it is nested within (Le. the clade at the next higher level) (Templeton et al., 1995). The

average distance of intenor to tip (1-T) clades is ahmeasured for Dc and Dn (Templeton et

al., 1995). The sîatistical ~ig~canceof each of these test statistics is determined at the 5% level with 1000 permutations. The interpretation of these results is conducted using an inference key (Templeton et al., 1995; Templeton, 1998; Febmary 2001 updated version available at: http://bioag.byu.eddzoology/crmdaU~lab/ges.h).These statistics were calculated using GeoDis 2.0 (Posada et al., 2000).

There are three possible scenarios that could explain the observed patterns of geographic parthenogenesis - range expansion, allopatric fragmentation, and restricted dispersal. The methodology describeci above is unique in its ability to discriminate between these alternative hypotheses. First, if Timemn parthenogens have experienced a range expansion north, then northerly populations should be younger (tip haplotypes) than southerly ones (interior haplotypes). As well, there should be fewer widespread haplotypes in the expanded populations to northern areas and more haplotype variation in southerly pre- expansion regions (Templeton et al., 1995; Templeton, 1998). If geographic sarnpling is adequate, it is possible to discriminate between continuous range expansion versus colonisation (abrupt establishment of populations in new ma).

A second possible mode1 that could explain geographic distributions is allopatric fragmentation (Templeton er al., 1995; Templeton, 1998). Immediately after the split, each new sexual population will reflect the pre-fragmented population and thus the populations wiil be indistinguishable from one another. if ase!xuality arose in one of the sexual populations and &ove this sexual progenitor extinct, then, as time increases, mutations occur independently in the isolated populations and therefore sexuals and asexuals should become genetically differentiated.

Third, under a scenario of restricted dispersai (restricted gene flow in sexual species),

the geographical extent of a haplotype tends to correspond with age; an older haplotype WU be more widespread (Templeton et al., 1995; Templeton, 1998). This situation wiii be reflected in the nested haplotype design, with young asexuai haplotypes king less widespread than older sexual ones. When a new asexual haplotype starts to spread, it will often remain within the geographic range of its ancestors, especially under an isolation-by- distance model. In addition, because the ancestral sexual haplotype is expected to be most frequent near its geographical origin, most asexual derivatives of these haplotypes will aiso occur near this area.

The ability to differentiate between these alternative hypotheses is dependent upon adequate sampling. None of the above described events are mutuaily exclusive, and the nested cladistic analysis can assess these different events at various hiecarchical levels within the cladogram. This methodology has been applied extensively to infemng intraspecific relationships and to a lesser degree, evolutionarily close interspecific relationships. Since a sexual: asexual pair should share a recent common ancestor, this connection can be treated as intraspecific. Therefore, the nested cladistic analysis was applied to each Timema sexuai: asexuai pair separately, and in the case of evolutionary close species, to those connections at the 95% statistically parsimonious level. mtDNA variation and General Phvlogmetic Relationshios

The analysis of 416 bp of rnitochondnal COI from 168 Timema individuals resulted in 122 haplotypes (Appendix A). New sequences (154) have been deposited at GenBank under the accession numbers AF4û9998 - AF4 1015 1 (Appendix A). Sequences from

Sandoval et al. (1998) can be found under accession nurnbers AF005330 - AF005345

(Appendix A). This segment of COI is globaüy AT rich and GC deficient. as estimated from the dataset (A; 27.3%. C; 16.1%; G: 19.2%; T: 37.4%). About half (207) of the sequenced sites were constant. Of the 209 variable sites, 174 were parsimony informative. Third- position changes accounted for most of the informative sites (132 sites, 75.9%), followed by first and second position changes (32 sites, 18.4%; 10 sites, 5.7% respectively). Most changes were synonymous (89.5%)while non-synonymous changes were few (10.5%).

The topologies of the neighbour joining and the strict consensus parsimony tree were the same, resulting in three major phylogeographic subdivisions - a Northern coastal clade

(hence called Northern), a Santa Barbara clade, and a Southern interior clade (hence called

Southem) (Figure 2). The southernmost sexual species T.chumash and T. nakipa, the

Arizona species T. ritensis and T. dorotheae, and the Nevada species T. nevadense, were not contained within these subdivisions and were basal to these groupings. (Figure 2). The bootstrap values for these divisions were high. Divergence between the Northem and Santa

Barbara clades ranged from 12- 1596, with the Southem clade king 16- 19% from either the

Northern or Santa Barbara clades. The two sexual: asexuai pairs, T. poppensis: T. douglasi, and, T. califomicum: T. shepardii, in the Northem clade were sampled in the Coastal Mountain Range fmn just below San Francisco to the Oregon border (Figure 1). Unexpectedly, the parthenogenetic taxa T,douglasi and T. shepardii were very closely related, having almost identicai or identical haplotypes (Figure 3). These asexual species both grouped with some individuals

€rom the most northen T. poppensis populations. T. californicum and T. poppensis, the two sexual species, were polyphyletic. Furthemore, T. poppensis and T. califomicum found in the same geographic area, in particular just below San Francisco Bay, were very closely related (0- 1% divergence) and in some cases had exactiy the same haplotype.

In addition to these sexuai: asexual pairs, the newly sampled species T. knulli, T. petita, and, T. landelsensis, demonstrated interesting phylogenetic relationships. Populations of T. knulli on different host plants (redwood and Ceanothus) were genetically distinct for

COI, with a divergence of 4.13 f 0.34%. T. knulli on Ceanorhus was more closely related to

T. petita than it was to T. knulli on redwood. Conversely, T. knulli on redwood was genetically closer to T. landelsensis found on Archytosraphylos (manzanita).

The Santa Barbara clade contained the sexuai: asexual pair T. cristinae: T. monikensis. The parthenogen formed a rnonophyletic group that was nested within its maternai sexuai progenitor (Figure 4). ûeographicaily, T. cristinae was found in the mountain ranges surrounding Santa Barbara, with T. monikensis found at the north- westenunost edge of Los Angeles (Figure 1). This pair was the only one that does not fit the pattern of geographic parthenogenesis - T. monikensis is slightiy south-east of its closest sexual.

The geographic range of the Southem clade was frorn the Mexico border to just north Figure 2: General phylogenetic relationships within the Timerna genus obtained through neighbour joining (NJ) and maximum parsimony (MP)analyses of mitochondnal COI. Al1 169 Timemu individuds that were sequenced are not show in this tree; rather, general species positions are given within triangle tips. Asexual lineages are ksignated by fernale symbols. Baculaum exrradentatum, Anisomporpha buprestoi&u, and Blatella gemnica were used to root the tree. Nurntierç above branches are NJ bootstrap values (1000 replications) and numbers below show MP bootstrap values (100 replications). Branches with les than 50% bootstrap support have no values shown. See Figures 3-6 for detailed views of the Northern, Santa Barbara, and Southem clades. General geography for each subdivision is dso shown.

Figure 3: Phylogenetic relationships among Timema species in the Northern clade. Species codes are as foiiows: pop = T. poppensis; doug = T. douglusi; cali = T. californicum; shep = T. shepardii; knull = T. knulli (redwood); knul2 = T. knulli (ceanothus); lan = T. landelsensis; petita = T. petita. The fmt nwnber dereach species code designates a population location while the second number refers to a different individuai. For example, poplnol is T. poppensis from locality one, individual number one. NovScalCV and mar3calLP aiso refer to T. califomicum but were individuais sequenced by another mearcher. See Appendix A for detailed locaiity information. Parthenogenetic lineages are designated by female symbols. a) neighbour-joining tree; b) neighbow-joining and maximum parsimony tree, with numbers above branches indicating NJ bootstrap values (1000 replications), while numbers below are MP bootstraps (LOO replications). No value is shown on branches with less than 50% bootstrap support.

Figure 4: Phylogenetic relationships between T. cristinae and T. monikensis in the Santa Barbara clade. Species codes are as follows: cris = T. cristinue, and mon = T. monikensis. The fmt number refers to a locality while the second number designates individual. The codes 3marlcris, para, laurel, house, and poppy also refer to T. cristinae but were coilected and sequenced by other researchers. See Appendix A for detaiied sampling information. The parthenogen, T. monikensis, is designated by fernale symbols. a) neighbour-joining tree; b) neighbour-joining and maximum parsimony tree, with NJ bootstrap values (1000 replications) indicated above branches and MP bootstraps (100 replications) given below branches. Branches with less than 50% bootstrap support have no numbers. of San Francisco and to Lake Tahoe (Figure 1). Unlike taxa in the Northern clade, species in this range are not geographically situated along the coast but rather are found in regions that are more interior. This clade contained the sexuai: asexual pairs T. podura: T. genevieve and

T. barnnani: T. tahoe, as weli as the sexual T. boharti (Figure 5). There was a bifurcation in this group, with T. boharti king basal to the rest of the taxa in this clade. There were two genetically distinct populations of T. genevieve that were geographically separated €rom one another by 200 km. Both of these populations were located north-east of San Francisco and were geographicaily far frorn the closest sexud relative T. podura. T. genevieve, however, formed a monophyletic group. T. pudura populations from different locations generaily had haplotypes that were most sirnilar to other individuais from the same locality. The asexual T. tahoe, sarnpled only in the White Fir (Abies concolor) surrounding Lake Tahoe, was also monophyletic and closely related to its closest morphologicai sexuai progenitor, T. barmani.

Three distinct and separate networks were constructed which agree with the above- described phylogenetic subdivisions. The 95% statistical parsimony lirnit for the dataset was

8 steps. The number of mutationai steps between each one of these subdivisions was a minimum of 30 steps. Since this method was designed IO infer intraspecific processes, each sexual: asexual pair was first allocated to a single nenvork and the analyses were conducted as such. However, the Northern clade coniained many closely related haplotypes for the sexuai: asexual pairs T. poppensis: T. douglasi, and, T. califomicum: T. shepardii, as well as haplotypes for T. knulli, T. petiia, and T. landelsensis. Because these taxa were not separable Figure 5: Phylogenetic relationships among Timema species in the Southem clade. Species codes are as foiiows: pod = T. podura; gen = T. genevieve; bart = T. batmani; ta& = T. rahoe; bohar = 7'. boharti; with numbers referring to sampling location and individual respectively. The designations novl lpodu and jun6podSJ refer to T. podura; mar4genev to T. genevieve, and mar6bohar to T. boharti* these individuals, as well as ttahoexxx and tbartmani, were sequenced by other researchers. Parthenogenetic species are designated by female symbols. a) neighbour-joining tree; b) neighbow-joining and maximum parsimony me, with numbers NJ bootstrap values (1000 repiications) shown above branches while numbers below are MP bootstraps (100 replications). Bootstraps under 50% are not shown. based on the mtDNA dataset, the analysis was performed using al1 of these species to form the Northem cladograrn (Figure 6). The Santa Barbara cladogram (Figure 7) contains T. cristinae and T. monikensis as in the phylogenetic tree. Even though T. podura: T. genevieve and T. bartmani: T. tahoe were joined within the sarne network (agreeing with the Southem phylogenetic subdivision) at the 8 step level (Figure 8), it made no difference to the inferred results if they were separated into different analyses since each species pair grouped with its closest respective relative.

The Northem Cladogram

The nested design for species in the Northem cladogram is shown in Figure 6. As in the NJ and MP phylogeny, T. douglasi and T. shepardii were both closely related to T. poppensis (clades 3-6 and 3-7) while separated from T. californicum (clade 3-5). T. knulli was separated into two disjoint b-step networks, with clade 3-2 containing i? knulli on redwood and clade 3-4 having T. knulli on Ceanothus. T. poppensis and T. californicum from various localities were nested within aU three higher level clades (5- 1,5-2, and 5-3).

There were two types of arnbiguities within ths structure - circularities and separation of clades by more steps than given under 95% statisticd parsimony. Circularities where there were unobsewed haplotypes (zeros) between represented haplotypes were broken and haplotypes were grouped according to the niles in Templeton and Sing (1993) and Crandaiî

(2000). Circularities in clades 1-12, 1-18, and 1-32 could not be resolved and haplotypes were therefore grouped at that level. These circularities comsponded to the result that these haplotypes were almost identical or identical to one another. The second arnbiguity was the Figure 6: The Northem cladogram for Tirnema. Numbers and fonts represent àifferent haplotypes and species as follows: 1-28 = T. poppemis; 29-32 = T. douglasi; 33-49 = T. califomicum; Sb55 = T. shepardii; 112-16 = T. knulli (redwood); 1ôô-107 = T. knulli (ceanothus); 188- 118 = T. Iandehemis; 111 = T. petita. See Appendix A for detailed location information. Haplotypes that are italicised contain more than one species, as follows: 13,38,41= T. poppensis, and T. caIifornicum; 30 = T.douglasi and T. shepardii. Lines between haplotypes represent a single mutational step supported at the 95% statistical parsimony level. Zeros are infemd intermediates. Numbers beside heavy solid iines denote that many mutational steps between clades. The nested clade level is given in a hieraschical manner; 1-n for 1-step clades, 2-n for 2 step clades, ..., 5-n for 5 step clades. The whole cladogram is a 6- 1 step clade.

Figure 7: The Santa Barbara cladograrn for Timema. Numbers and fonts represent different haplotypes and species as follows: 56-73 = T. cristinae; 74-76 = T. monikensis. See Appendix A for detailed location data. Lines between haplotypes are inferred mutational steps at the 95% parsimonious level, with zeros repmenting hypothetical intemediates. Heavy lines with nurnbers denotes that number of mutational steps between clades. Clades levels are designated as for the Northem cladograrn (Figure 6). The whole cladogram is a 5-4 step clade.

Figure 8: The Southern cladogram for Timemu. Numbers and fonts represent different baplotypes and species as follows: 77-87 = T. podura; 88-95 = T. genevieve; 96 - 9 8 = I: barmrani; 99- 101 = T. tahoe. See Appendix A for iwdity information. Solid lines represent one mutational step, with zeros being intermediate haplotypes not sampled. Statistical parsimany is supported at the 95%level. Heavy lines have a greater number of mutational steps benveen clades, as given by the number beside. Clade levels are designated as in the Northern cladogram (Figure 6). The whole cladogram is a 5-5 step clade. separation of clades 4-L,3-44-2,2-3,4-3, and 3- 10 by more steps than given for 95% statistical parsimony. However, these hepnetworks matched the inferred relationships in the phylogeny above, with each of those clades having a high bootstrap value, The agreement between the phylogeny and this network provided confidence that these subdivisions were valid under maximum parsimony.

The nested contingency analysis for the Northem cladogram network showed statistical significance (p< 0.05) mainly for higher level clades (Table 3). No tests were significant at the 1-step clade level, two were significant at the Zstep level(2-8 and 2-14), four were significant at the 3-step level(3-4,3-6,3-7,3-IO), and al1 were significant at higher levels (4-2,4-3,s-1.5-2,5-3-6-1). The results of the nested geographic analysis are given in Figure 9, with the chah of inference results for significant clades shown in Table 4.

The clades containing both sexuai and asexual species gave results of restricted dispersal by distance (2- l4), contiguous range expansion (3-6), contiguous range expansion or long distance colonisation (4-2), and long distance colonisation (5-2). Fragmentation or isolation by distance was inferred for clade 4-1 (containing T. knulli on redwood, T. landelsensis, T. poppensis, and T. califomicum). Overall, the sampling design was inadequate to differentiate between contiguous range expansion and long distance colonisation for al1 taxa within the

Northem Cladogram. Table 3: Chi-squared test for geographic structure for the Nonhern cladogram witb associated probabilities

Clade x2 Statistic Pmbability 1-2 8.00 1.O00

6- 1(total) 174.48 O.OOO*

* indicates significant geographical association at the 5% level Figure 9: Results of the nested geographic analysis of Thema COI haplotypes for the Northem cladogram. The hierarchy of haplotypes and clade levels is the same as that in Figure 6. Columns represent increasing clade level àesignations, from left to right. Bracketed groupings indicate the nesting structure. The clade @J and nested clade (Dm) distances are given for each clade. With the nul1 hypothesis of no geographical associations, significant clade distances are given by a superscripted 'SI for smaller distances than expected, and a superscripted 'L'for larger distances than expected. For clades that have identifiable interior (I)or tip (T) status, the 1-T distances are given in boxed areas. In these cases, interior clades are designated with bold type.

Table 4: Inference chah for nested geographical analysis of the Northem cladogram usbg the inference key given in Templeton (1998). The relative localities for sexual species are given by the following letter designations: N= north of San Francisco, SF = San Francisco, B = below San Francisco. The parthenogens T. douglasi and T. shepardii are each more northerly than San Francisco.

Clade Species Chain of Inference lnferred Pattern 2-8 T. -poppensis -- (SF, N) 1-2-3-4-no Restricted gene flow with T. califomicum (SF). . isolation by distance T. popPensis (N) Restricted dispersai by distance T. douglasi in non-sexual species; restricted T. shepardii gene flow in sexuai species T,poppensis (N) Contiguous range expansion T,douglasi T. shepardii T. poppensis (N) Inconclusive T. douglasi T. poppensis (N) Geographic sampling scheme T. califomicum (SF) inadequate to discriminate T. knulli redwood (B) between fragmentation or T. landelsensis (B) isolation by distance T. poppensis (N) Sarnpling design inadequate to T. douglasi discriminate between contiguous T. shepardii range expansion and long T. califomicum (B) distance colonisation T. poppensis (SF) inconclusive T. califomicum (SF) T. poppensis (N) Long distance colonisation T. douglasi T. shepardii T. califomicum (B) T. poppensis (SF, N) Sampling design inadequate to T. douglasi discriminate between contiguous T. califomicum (SF, range expansion and long B) distance colonisation T. shepardii T. knulli redwood (B) T. knulli ceanothus (BI T. landelsensis (B) T. petita (B) The Santa Barbara Cladograrn

The chi-square test for geographic structure was significant for most higher clade levels in this cladogram (clade 4-44-5, and 5-4; Table 5). As in the other cladograms, there were more steps than statistically signifcant between 6-step networks (Figure 7). The nested anaiysis results are given in Figure 10. Each pping, however, reflected the structure seen in the NJ and MP phylogenetic tree. The inferred pattern of dispersal for al1 significant clades was range expansion (Table 7). However, T. monikensis, the parthenogen, grouped together with only one T. crisrinue haplotype (clade 4-6). Since the contingency analysis was not significant for this clade, there was insufficient evidence to infer geographic structure within this clade. At the total cladogram levej, the overall inference was one of contiguous range expansion.

The Southern Cladogram

As for the other species pairs, higher clade levels were significant for the nested contingency analysis (clade 3-18,4-9,4- IO, 5-5; Table 5). There were few ambiguities in this nested clade analysis; some connections were a greater number of steps than ailowed under 95% statistical parsimony. However. the topology of the cladogram shown matched that of the phylogenetic tree (Figure 8). The nested results and chain of inference are given in Figure 11 and Table 8 respectively. Clade 3-18 contained T. genevieve from two geographically distant populations (clades 2-44,245). The chain of inference for this analysis is consistent with contiguous range expansion within the asexual heage (Table 8). Table 5: Chi-squared test for geographic structw for ihe Santa Barbara cladogram with associated probabilities

Clade x2 Sfafistic Probability 2-37 3.00 0.324 3-1 1 10.00 0.066 3-12 4.00 0.340 3-13 3.00 0.362 3-14 1.33 1.O00 44 24.00 0.00 1* 4-5 12.00 0.03 1* 4-6 6.0 O. 173 5-4 (total) 32.00 0.003*

* indicates significant geographical association at the 5% level Figure 10: Results of the nested clade analysis for T. cristiw and T. monikensis in the Santa Bahara cladograrn. The hierarchy of this analysis foilows that given in Figure 7. Statistical designations are the same as outiined in Figure 9.

Table 6: inference chah for nested geographical analysis of the Santa Barbara cladogram using the inference key given in Templeton (1998).

Clade Species Chain of Inference Inferred Pattern 4-4 T. cristinae 1-2-1 1-12-no Contiguous range expansion 4-5 T. cristinae 1-2-1 1-12-no Contiguous range expansion 5-4 T. cristinae 1-2-1 1-1240 Contiguous range expansion T. monikensis Table 7: Chi-squared test for geographic structure for the Southem cladogram with associated probabilities

Clade x2 Statistic Probabiiity 1-65 2.00 1.000

5-5 (total) 50.75 O.OOO*

* indicates significant geographical association at the 5% level Figure Il: Results of the nested geographic analysis for Tirnema taxa in the Southem cladogcam. Haplotypes and clade designations are the same as given in Figure 8. The format of this figure follows that given in Figure 9.

Table 8: inference chah for nested geographical analysis of the Southern dadogram using the inference key given in Templeton (1998).

Clade Species Cbain of ~nxrence lnferred Pattern 3- 18 T. genevieve 1-2-1 1-12-110 Contiguous range expansion 4-9 T. piidura 1-2-3-5-15-110 Pmt fngrnen tation T. genevieve 4-10 T.barûnani 1-2- 1 1 - 12- 13- 14-yes Long distance colonisation T. tahoe 5-5 T. podura 1-2-1 1-12-110 Contiguous range expansion T. genevieve T. barmani T. iakoe At the next higher clade level, 4-9, T. genevieve nested with T. podura from the Sierra Madre mountains and Sequoia National Park, and the Uiference was consistent with past fragmentation. The geographic smture of the T. bartniani: T. tahoe clades were ~ig~cant only at the 4- 10 level (Table 5). As in the Northem cladograrn, the parthenogenetic T. ta& haplotypes were nested interior to their sexud counterparts. The nested clade analysis for this clade indicates range expansion with long distance colonisation. At the total cladogram level, the inference was contiguous range expansion (Table 8).

General Phvlo~eneticRelationshios

Previous phylogenetic work with Timema suggested that this genus originated in the southern part of its current distribution appmximately 20 million years ago (Sandovd et al.,

1998). The subsequent expansion northwards and speciation of Timema over time was inferred to follow the changes in host plant disiribution and concur with major construction periods of the Sierra Nevada, Coastai, and Transverse mountain ranges (Sandoval et al.,

1998). The phylogenetic relationships inferred in this paper generaily agree with these inferences. There are thedistinct phylogenetic subdivisions that correspond well with the geography of California. The most southern Californian and Arizonian sexual species are basal to these subdivisions, consistent with the inferred origination and age of Timema. One notable difference however, is the placement of T. chumash. Previous work suggested this species was a close sister group to T. poùura sampled from the same geographic area. The pbylogeny presented in this paper demonstrates that T. chiunosh is actually a basal species

and that does not cluster with T. podura.

The Northem clade contains numerous polytomies with low divergence, suggesting

that these taxa are of very recent origin. Unexpectedly, T. shepardii appears more closely

related to T. poppensis (and T. douglaso than to T. califomicum, its presumed sexuai

counterpart. There are six possible hypotheses that could account for this discrepancy: 1) T.

poppensis is actually the closest sexual relative, 2) there is a speciedgene tree ambiguity for

rnitochondrial COI, 3) the closest T. califomicum was not sampled, 4) the closest sexual T.

califomicum was driven or has gone extinct and thereby not sampled, 5) T. shepardii is a

hybrid of T. poppensis and T. califomicum, exhibiting the morphology and host plant use

simila.to T. califomicum but having mtDNA close to T. poppensis, and, 6) T. poppensis and

T. califonzicum (and thereby T. douglasi and T. shepardii) are not different sptxies. It is not

easy to discem between these hypotheses without further data.

All populations of T. shepardii and T. douglasi are very closely related, in some

instances having identical haplotypes, irrespective of the fact that these species are

geographically widespread. T. shepardii and T. douglasi have also ken sampled side by

side from one geographic area (OnSprings Rd) on their respective host plants, manzanita

and Douglas-fir. Intensive sampling in other areas may reveal that these species coincide or

overfap in further localities. The geneticaiiy closest (4%diverged) T. poppensis to these

asexuals geographicaliy overlaps the southenunost distribution of both asexuals. Conversely,

the geograpbically and geneticaily closest (2.5% diverged) T. californicum is 250 km south

of either asexual and separated by the San Francisco Bay. The geography of these taxa is

suggestive that T. poppensis is the actual sexual counterpart of T. shepardii. However, if the dosest T. califomicum counterpart was not sampled, either due to inadquate sampling effort or extinction, then this proposal may be presumptive. It does not seem iikely that the closest T. californicum was missed since hrewere T. califomicum populations coliected between the geneticaiiy closest T. cal~omicwnand T. shepardii.

Extinction is a definite possibiiity, due either to parthenogens driving sexual cornterparts extinct, or to development in the San Francisco Bay area,

Evidence that the sexuals T. poppensis and T. culifornicm have similar mtDNA haplotypes cornes fiom populations around San Francisco, where these species have identical haplotypes or have very low divergence from one another. However, rnorphology and host phiUS; ye quite distinct between T. californicum and T. poppensis. T. poppensis is dark green with dom- lateral stripes and is highly cryptic on its host plant Douglas fir

(occasionally found on redwood). Coaversely, tbe most common T. californicm mqhis a fiai light green (occasionally pink or brownish) and this species is found on mamanita, oak, and Ceanothics. Although male genitalia are distinct between these species, it is unknown if these T. poppensis and T. cal~ornicmare actually reproductively isolated. It is possible, although unlikely, that tbese "species" are just different morphs of the same species.

Conseqwntly, T. shepardii and T. douglasi would just be different morphs of the same parthenogen.

These results could also be exptaioed if T. poppemis and T. californiciun occasionally mate with one another. Under dis scenario, T. shepardii could have originated by such a hybridisation event. Many asexuals are thought to be hybrids (Vrijenhoek, 1989; Vrijenhoek et al., 1989) and there is evidence within Timenza that hybridisation may occur (describeci below for T. rnornikensis), leading to asexuality. In the case of T. psppemis and T. californicum, since both species exhibit similar mtDNA haplotypes, hybridisation may be recent and non-directional. However, T. shepardii bas mitochondnal haplotypes most similar to T. poppensis, strongly suggestive that 'I: poppensis is the materna1 ancestor under a hybridisation hypothesis.

Thylogenetic and host plant evidence from the re-described species T. knulli suggests that T. knulli on Ceanothus and T. knulli on redwood are quite distinct. Whether or not these populations interbreed is not known. Although T. petita has ken described as a new species, it is closely related to T. Rnulli on Ceanothus. Since these species fced on the same host plant species, and male genitalia only differs in size, these taxa may not be different species. The size variation, and possibly the colour morphology variation, could be due to differences in habitat. T. petita was collected dong an unsheltered coastal area and would experience high winds and rain associated with this region. Conversely, T. knulli (Ceanothus) inhabits a coastal region surrounded by mountains and would be protected from harsh conditions. No intermediate variants have been collected but the geographic area between T. petita and T. knulli has not been extensively sampled. T. landelsensis does appear to be a sister group to

T. baulli (redwood).

The Santa Barbara clade is geographicaily limited to the coastai region around Santa

Barbara. During the 1999 collection season, four males were found in the T. monikensis population. Although it is not uncornmon for parthenogenetic species to oçcasionally produce stede males (Soumalainen et al., 1987), these males exhibited morphological characteristics consistent with a hypothesis of hybridisation. The mtDNA phylogeny

indicates definitively that T. cristinue is the maternai ancestor of T. monikensis. The collected males, however, have genitalia and behavîour more similar to T. chumash than to T. cristinae or other described species (Vickeq and Sandovai, 2001, submitted). T. monikensis also feeds on similar host plants to both species (Ceanothus and Cercocarpus). This evidence suggests that T. monikensis may be a hybrid of T. chwnash and T. cristinae. In the case of T. monikensis, hybridisation appears to be directional, with T. cristinae king the maternai ancestor and T. chumosh the patemal. As well, the proposed hybridisation event would have occurred a long tirne ago. As mentioned, T. monikensis is the only parthenogen that does not clearly fit the pattern of geographic parthenogenesis. Evidence that T. monikensis is a hybrid, as presented above, could potentially explain why this species does not fit the pattern of geographic parthenogenesis. Geographically, T. monikensis is located in the mountain range between T. cristinae to the north and T. chumash to the south.

In the Southern clade each asexuai, T. genevieve and T. tahoe, is monophyletic.

Since the base of this clade is unresolved, it is difficult to infer relationships between the sexual: asexual pairs T. podura: T. genevieve and T. bartmni: T. tahoe. Host plant use and morphology agree that these species pairs are each other's closest relatives, as does the nested cladogram presented below. T. boharti is basai to the Southem clade and is geographically located at the southernmost lirnits of the genus, consistent with the idea that Timema originated in the south and has since expanded north. Except for one sampled population of

T. podura, the sexuals T. podura and T. bartmani are geographicaily located in the 100 km surrounding T. boharti, indicating that these species may not have expanded their range very much. By contrast T. genevieve is separated from the geographically closest T. podura by

250 km and T. rahoe is >500km norîh of T. barrniani, suggestive that asexuals have better dispersal ability than their sexual counterpsuts. NCA. Inferred Geomhic Patterns. aûeogmhicPwnesis

The Northem Cladogram

Overall, the analysis for higher clade levels in the Northem cladogram shows range expansion, consistent with previous phytogeographic work that the genus has expanded.

However, the major implication of this suggestion was that Timema ociginated in the south and the direction of expansion was only towards the north. Contras, to this suggestion, there are two significant indications that ancestral haplotypes occur in the north. First, the inferred pattern for clade 2-8 is restricted gene flow with isolation by distance. Tip haplotypes in this clade belong to both T. poppensis and T. califomicum geographically located near San

Francisco. The interior haplotypes are those of T. poppensis from San Francisco and from

200 km north. Consistent with a delof restricted gene flow, the tip (young) haplotypes have a geographic range nested within the range immediately interior to it (Templeton et al.,

1995). This result suggests that some northern T. popperrsis haplotypes are ancestral.

Second, the overall inference for the Northern cladogram is range expansion. Under a mode1 of range expansion, old haplotypes should be sampled hm the pre-expansion areas and should be interior to the cladogram (Templeton et al., 1995). Again, some northern T. poppensis haplotypes are interior, suggestive that they may be ancestral. One hypothesis that wodd be consistent with ancestral T. poppensis haplotypes is that T. poppensis has expanded from northerly located glacial refugia populations.

A pecdiarity in the NCA is the observation îhat the asexuals T. douglasi and T. shepardii are found interior to the semai species, T. poppensis. Usually, haplotypes of recent origin and low frequency will occur at the tips of a cladogram (Golding, 1987; Excoffier and

Langaney, 1987). Assuming that parthenogenetic haplotypes wouid be of more recent origin since they arise from their sexual progeaitors, asexuals should occur preferentially at the tips of the cladogram. Since the asexual haplotypes faIl into circulatities with, and are closely related to, T. poppensis, these asexuals may have arisen from interior sexual haplotypes and are so recent in origin that their haplotypes have not had time to diverge, and consequently are "aciing" like old haplotypes within the nested design. Consistent with this hypothesis is the result of restricted dispersal by distance for clade 2-14. When a new haplotype arises it will often remain within the geographic range of it progenitor (Templeton et al., 1995). If these haplotypes are of very ment origin, there will have been insufficient time for divergence and consequently these haplotypes will nest irnmediately adjacent to ancestral haplotypes.

Although range expansion has occurred, there is inadequate sampling to discriminate between contiguous range expansion and long distance colonisation for most clades. In temof evaluating the hypothesis that geographic parthenogenesis is the result of range expansion, the evidence indicates that both restricted dispersal (clade 2-14) and contiguous range expansion (clade 3-6) play a role in the geographic structure. Further support that parthenogen distribution is the resuit of range expansion is the observation that despite the fact that T. douglasi and T. shepardii have few haplotypes, both of these taxa are remarkably widespread in distribution. The Santa Barbara Cladogram

Within the T. cristirne: T. monikensis pair there is an overall inference of contiguous range expansion. T. monikensis, the parthenogen, groups together with only one T. cristinae haplotype, suggestive that there may be other un-sampled baplotypes. Furthemore, there are as many as 16 steps between T. cristinae haplotypes sampled from the same geographic location. There are two plausible explanations fer this result: 1) Sampling effort was not adequate to characterise al1 T. cristinae haplotypes; 2) T. crisrinue has remarkably high within species divergence that corresponds to spatial variation of different colour morphs

(Sandovai, 1994). This latter hypothesis is the subject of ongoing work. Even though this pair does not fit the pattern of geographic parthenogenesis, inferred contiguous range expansion does explain the geographic structure. The primary inferred direction of this range expansion is south-east, from clade 4-5 with T. cristinae co 4-6 with T. monikensis haplotypes. If T. chumash were the patemal ancestor of T.moriikensis, the expectation would be to see range expansion for a cladistic anaiysis using a neutral nuclear marker.

The Southem Cladogram

Within the parihenogen T. genevieve (clade 3- 18), contiguous range expansion is inferred. Haplotypes within the nested clades 2-44 and 245 are geographically separated from one another by 150 km, suggesting that T. geneview is a good disperser or that therie are missing intermediates. With respect to its sexual counterpart T. podura, however, the inference is one of fragmentation (clade 4-9). This result suggests that the geographic distribution of T. genevieve is due to a past fragmentation event and that the pattern of geographic parthenogenesis is not due to range expansion. At the next higher level (total cladogram), the inference is one of contiguous range expansion, regardless of whether or not the T. bartmani: T. rahoe pair is included. There are two plausible scenarios that could account for different inferences at different levels. in the fmt scenario, T. podura, overall, has expanded northward. At some point during this expansion, the parthenogen T. genevieve originated and both species continued to expand north. Since T. genevieve is expected to be a better disperser, this asexual would be at the leading edge of this expansion. By definition, the ongin of asexuality can be considered a fragmentation event since the two lineages are no longer able to exchange genes. T. genevieve continued to expand north while T. podura lagged. The second scenario is sirnilar except that the fragmentation event could be inferred withm T. podura More the origination of T. genevieve. in this case, when T. genevieve arose, it drove its progenitor extinct and continued to expand (and diverge) northwards. Both hypotheses could account for the pattern of geographic parthenogenesis since the actual position of T. genevieve is the result of range expansion and not fragmentation. Under either supposition, the observation that T. genevieve has some internai haplotypes in the nested design can be exarnined. In the fmt scenario, T. genevieve would be expected to be old since the Lineage originated before the fragmentation event. Under the second hypothesis, T. genevieve haplotypes would be ancestral if the progenitor extinction event occurred a long time ago. The observation that there are two highly diverged and geographically separated populations of T. genevieve supports this notion.

Long distance colonisation is inferred for the T. bartmani: T. rahoe clade (4- 10). As in other species pairs, T. rahoe is internai to its sexual counterpatt T. bartmani. This internalisation falls between T. barrmani haplotypes hmtwo different, but geographically close, populations, suggesting that the T. tahoe haplotypes are ancestral. Examination of the clade distance values for the 4- IO clade indicates significance with T. bartmani from only one population (haplotypes 96 and 97), and that T. tahoe is distantly located from this population. These T. bartmani haplotypes are more intemal, and thereby presumably older, than T. tahoe. It is diffmlt to accurately determine relationships within this part of the cladogram since there are not many sampled populations or haplotypes. Since the NCA indicates that the asexual distribution is the result of long distance colonisation, the pattern of geographic parthenogenesis can be attributed to range expansion.

Conclusions

Parthenogenesis has arisen independently at least four times, possibly five, in the genus Timema. Evolutionary relationships in the northem sexual: asexual pairs (T. poppensis: T. douglasi, T. cal@omicum:T. shepardii) are ambiguous. Although morphologically distinguishable, T. doughi and T. shepardii are not respectively monophyletic; consequently, these taxa cannot be interpreted as independent asexuai species using this dataset. T. monikensis, T. genevieve, and T. tahoe are each monophyletic and closely related to their respective sexual counterparts, indicating each species has an independent origin of asexuaiity. The overail phylogeography of Timema is consistent with previous findings that this genus originated in the south and has since expanded north, with both sexual and asexuals species showing range expansion. However, there are indications in the Northern clade that some T. poppensis haplotypes are ancestral and have expanded south. Furthemore, asexuals appear to bave undergone more extensive northerly range expansion than sexuals, leading to geographic parthenogenesis.

Observations that parthenogenesis tended to be geographicaily more northerly in distribution has been explained by assuming a higher dispersal rate of asexuality at the leading edge of species range expansion (Bell, 1982; Stearns, 1987). This study is the fmt of its kind that has tried to address this pattern of geographic parthenogenesis in a rigorous and testable fashion. in each case of parthenogenesis, the major inference is that geographic structure is the result of range expansion. Regardless of whether or not T. douglasi and T, shepardii are different species, the observation that there are few widespread haplotypes supports the notion of a high asexual dispersal rate, as well as inferences of range expansion.

Although T. monikensis does not fit the pattern of geographic parthenogenesis, this species may be a hybrid with its distribution determined by both of its parental types. The T. genevieve lineage has been fragmented from its sexual counterpart but the overall more northerly pattern is consistent with a hypothesis of range expansion. T. tahoe is geographically north and far from its sexual counterpart, the apparent result of long distance colonisation. The cases of T. genevieve and T. tahoe, since each is geographically far from respective sexuals, are also consistent with theory that suggests asexuals have better dispersal ability than sexuais.

The phylogenetic split of Tirnemu into the Northem, Santa Barbara, and Southern clades matches the geography of California remarkably well, with the distributions of

Timema correspondhg to the mountain ranges. Few phylogeographic studies have been conducted on California taxa but there are many species known to exhibit geographic patterns defmed by the mountain ranges (e.g, Tan and Wake, 1995; Gervais and Shapiro, 1999; Rodriguez-Robles et al., 1999). As in Timema, phylogeographic analyses of the

Californian Newt, Tarichu forosa, showed that northern populations had no sequence divergence while southern and central populations were quite differentiated, suggesting that northem Newts are reiatively young (Tan and Wake, 1995). Furthemore, the Tan and Wake

(1995) study interpreted phylogeographic patterns to reflect dispersal of subspecies from south to north, and from north to south, comsponding to geologic changes in the Caiifomia coastal region. The observation that T. poppensis and T. califomicum are polyphyletic may ais0 reflect dispersal correspondhg to changes in sea level and land formation around the

San Francisco region. The phylogeny of the Californian mountain kingsnake (iumpropelris zonafa) also shows geographical structuring between northem and southern clades that roughly corresponds to that seen in Timema (Tan and Wake, 1995). In addition, the southern subspecies are basal and ancestral to more northem populations. These studies, as with this study of Timema, suggest that species distributions in California have been highly affected by dispersal ability.

Studies of the geographic distributions of asexuality relative to sex have long been thought to be able to provide insight into the advantages of sex. Geographic distributions of

Timema, in generd, match the biogeographic patterns seen in other Californian taxa.

Although Newts and kingsnakes do not have asexual lineages, evidence suggests that the ability to disperse may ôe contribute to observed biogeographic patterns in this region. Since the demographic effects of king asexual make them better dispersers than sexuals, and the overaii phylogeographic patterns in Timemu are sirnilar to other taxa in California, the pattern of geographic parthenogenesis in Timema is most likely the resdt of these effects. Other studies on asexual geographic distributions, not just geographic paxthenogenesis, have also supported the idea that the ability to capidly colonise to vacant habitats is important (Chaplin

and Ayre, 1997; Schon et al., 2000). The ability to disperse essentially lirnits competition

between asexuais and sexual progenitors, aihugh this outcome may only confer an evolutionary short tenadvantage. Avise, J. C. 1994. Molecular Markers, Natural history, and Evolution. Chapman and Hall, New York.

Avise, J. C. 2000. Phylogeography: the History and Formation of a Species. Harvard University Press, Cambridge, Mas.

Bartman, G. and P. D. Brock. 1995. Observations on the appearance and behaviour of species of the stick-insect genus Tirnemu Scudder (Phasmida: Timematodea). American Entomological Society Bulletin 54: 197-203.

Barton, N. H., and B. Charlesworth. 1998. Why sex and recombination? Science 281: 1986- 1989.

Bell, Graham. 1982. The Masterpiece of Nature: The Evolution and Genetics of Sexuality. Crwm Helm, London.

Birmingham E. and C. Moritz. 1998. Comparative phylogeography: concepts and applications. Molecular Ecology 7: 367-369.

Chaplin J. A. and D. J. Ayre. 1997. Genetic evidence of widespread dispersal in a parthenogenetic freshwater osiracod. Heredity 785767.

Clary, D. 0. and D. R. Wolstenholme. 1985. The mitochondrial DNA molecule of Drosophila yakuh: nucleotide sequence, gene organization, and genetic code. Journal of Molecular Evolution 22: 252-27 I .

Clement, M. and D. Posada. 2000. TCS: a computer program to estimate gene genealogies. Molecular Ecology 9: 1657- 1660.

Crandall, K. A. and A. R. Templeton. 1993. Empirical tests of some predictions from coalescent theory with applications to intraspecific phylogeny reconstruction. Genetics 134: 959-969.

Cuellar, 0. 1977. Animai parthenogenesis: A new evolutionacy model is needed. Science 197: 837-843.

Gaggiotti, O. E. 1994. An ecologicd model for the maintenance of sex and geographic parthenogenesis. Journal of Theoretical Biology 167 201-22 1.

Gemtsen, J. 1980. Sex and parthenogenesis in sparse populations. The Arnerican Naturaiist 115: 7 18-742.

Gervais, B. R. and A. M. Shapiro. 1999. Distribution of edapbic-endemic buttedies in the Sierra Nevada of California Global Ecology and Biogeography 8: 15 1- 162. Glesener, R. R. and D. Tian. 1978. Sexuality and the components of environmental uncertainty: clues from geographic parthenogenesis in terrestrial animais. The American Naturaikt 112: 659-672.

Lynch, M. 1984. Destabilizing hybridization, general-purpose genotypes and geographic parthenogenesis. Quarterly Review of Biology 59: 257-290.

Morrone, J. J. and J. V. Crisci. 1995. Historical biogeography: introduction to methods. Annual Review of Ecological Systernatics 26: 373-401.

Peck, J. R., J. Yearsley, G. Barreau. 1999. The maintenance of sexual reproduction in a structured population. Proceedings of the Royal Society London B 266: 1857- 1863.

Peck, J. R., J. M. Yearsley, and D. Waxman. 1998. Explaining the geographic distributions of sexual and asexual populations. Nature 391: 889 - 892.

Posada, D. and K. A. Crandall. 2001. Intraspecifîc gene genealogies: trees grafting into networks. Trends in Ecology and Evolution 16: 37-45.

Posada, D., K. A. Crandall, and A. R. Templeton. 1999. GeoDis: a program for the cladistic nested analysis of the geographical distribution of genetic haplotypes. Molecular Ecology 9: 487-488.

Rambaut, A. 1996-2001. Se-AI: Sequence alignment editor v2.0. University of Oxford. available at: http://evolve.zw.ox.ac.uWsoftware/Se-SVmain.htrnl. (21 June 2001).

Rodriguez-Robles, J., D. F. Denardo, and R. E. Staub. 1999. Phylogeography of the California mountain kingsnake, Lampropeltis zonata (Colubridae). Molecular Ecology 8: 1923- 1934.

Sandoval, C. P. 1994. The effects of the relative geographic scales of gene flow and selection on the morph frequencies in tbe walking-stick Timema cristinae. Evolution 48: 18661879.

Sandoval, C. P. and V. R. Vickery. 1996. Timema douglasi (Phasmatoptera: Timematodea), a new parthenogenetic species fom southwestern Oregon and northem California, with notes on other species. The Canadian Entomologist 128: 79-84.

Sandoval, C. P. and V. R. Vickery. 1999. Thema comani (Phasmatoptera: Timematodea) a new species from Arizona and description of the female of Timema ritensis. Journal of Orthoptera Research 8: 49-52.

Sandoval, C., D. A. Carmean, and B. J. Crespi. 1998. Molecular phylogenetics of sexual and parthenogenetic Tinaema walking-sticks. Proceedings of the Royal Society of London B 265: 589-595. Scknck, R. A., and R. C. Vrijenhoek. 1986. Spatial and temporal factors affecthg the coexistence among sexual and clonal forms of PoeciIiopsis. Evolution QO: 1060- 1070.

Schon, I., R. K. Buth, H. I. Gfliths, and K Martens. 1998. Slow molecular evolution in an ancient asexual ostracod. Proceedings of the Royal Society of London B 265: 235- 242.

Soumalainen, E., A. Saura, and J. Lukki. 1987. Cytology and Evolution in Parthenogenesis. CRC Press, Boca Raton, FL, USA.

Stems, S. C., ed. 1987. The Evolution of Sex and its Consequences. Birkhauser Verlag, Boston.

Strimmer, K., and A. von Haeseler. 1996. Quartet puzzhng: A quartet maximum likelihood method for reconstructing tree topologies. Molecular Biology Evolution 13:964-969.

Swofford, D. L. 2000. PAUP*.Phylogenetic analysis using parsimony (*and other methods). Version 4. Sinauer Associates, Sunderland, Massachusetts.

Tm,A. and D. B. Wake. 1995. MtDNA phylogeography of the California newt, Taricha torosa (Caudata, Salamandridae). Molecular Phylogenetics and Evolution 4: 383- 394,

Templeton, A. R. 1982. The prophecies of parthenogenesis, in: Evolution and Genetics of Life Histories, pp. 75- 10 1. eds. H. Dingle, and J. P. Hegmann. Springer-Verlag, NY, USA.

Templeton, A. R. 1989. The meaning of species and speciation: A genetic perspective. In: Speciation and its Consequences. eds D. Otte and J. A. Endler. Sinauer, Massachusetts. pp 3-27.

Templeton, A.R. 1998. Nested clade analysis of phylogeographic data: testing hypotheses about gene flow and population history. Molecular Ecology 7: 38 1-397.

Templeton, A. R. and C. F. Sing. 1993. A cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping. IV Nested analyses with cladogram uncertainty and recombination. Genetics 134: 659-669.

Templeton, A. R., E. Boerwinkle, and C. F. Sing. 1987. A cladistic analysis of phenotypic associations with haplotypes inferred hmrestriction endonuclease mapping. 1. Basic theory and an analysis of alcohol dehydrogenase activity in Drosophila. Genetics 117: 343-35 1. Templeton, A. R., K. A. CrandaIl, and C. F. Sing. 1992. A dadistic analysis of phenotypic associations with baplotypes infened from restricted endonuclease mapping and DNA sequence &ta. m. Cladogram estimation. Genetics 132: 619-633.

Templeton, A.R., E. Routman, and C.A. Phillips. 1995. Separating population structure from population history: a cladistic analysis of geographical distribution of mtDNA haplotypes in the Tiger Salamander Ambystoma tigrinum. Genetics 140: 767-782.

Vickery, V. R.. 1993. Revision of Timema Scudder (Phasmatoptera: Timematodea) including three new species. Canadian Entomologist 125: 657-692.

Vickery, V. R. and C. P. Sandoval. 1997. Timema bartmani (Phasmoptera: Timematodea: Timematidae), a new species from southern Californio. The Canadian Entomologist.l29: 933-936.

Vickery, V. R. and C. P. Sandoval. 1998. Timema monikensis, species nov. (Phasmoptera: Timematodea: Timematidae), a new parthenogeneiic species in California. Note, Lyman Entomological Museum and Research Laboratory 22: 1-3.

Vickery, V. R. and C. P. Sandoval. 1999. Two new species of Timema (Phasmoptera: Timematodea: Timematidae), one parthenogenetic, in California. Journal of Orthoptera Research 8: 4 1-43.

Vickery, V. R. and C. P. Sandoval. 2001. Description of three new species of Timema (Phasmoptera: Timematodea: Timematidae). Submitted to the Lyman Entomological Museum and Research Laboratory.

Vrijenhoek, R. C. 1989. Genetic and ecological constrainr on the origins and establishment of unisexual vertebrates, in: Evolution and Ecology of Unisexual Vertebrates, pp. 24- 3 1. eds. R. M. Dawley and J. P. Bogart. New York State Museum, Bull. no. 466: Albany, NY, USA.

Vrijenhoek, R. C., R. M. Dawley, C. J. Cole, J. P. Bogart. 1989. A list of the known unisexual vertebrûtes, in: Evolution and Ecology of Unisexual Vertebrates, pp. 19-23. eds. R. M. Dawley and J. P. Bogart. New York State Museum, Bull. no. 466: Albany, NY, USA. Chapter 3: Tbe pemisîence of asexuality in TIniema

Introduction

Since sex is the prevaient mode of reproduction, why do some organisms maintain

asexuality? What is the adaptive significance of sex? In the short term, advantages of sex

are unclear given the two-fold reproductive advantage that asexuality confers (Bell, 1982;

Maynard Smith, 1992; Taylor et al., 1999). Conversely, the long-term advantages

encompassed by sexual genetic variability are generally accepted (Bell, 1982; Hwst and

Peck, 1996). The appearance of ancient asexuals is a direct affront to the many theones of

sex; understanding why these ancient lineages persist will potentially help explain why most

organisms are sexual (Judson and Normark, 1996; Little and Hebert, 1996; Johnson and

Bragg, 1999; Normark, 1999).

There are few exiunples of ancient obligate asexuals (Judson and Normark, 1996;

Johnson and Bragg, 1999). The best evidence for long-lived asexuaiity comes from the

bdelloid rotifers, organisms where no males have been sampled, there is no evidence of sex,

and there is high morphologicai diversity between sublineages (Judson and Normark, 1996;

Welch and Meselson, 2000). Although fossil evidence for darwinulid ostracods supports

dates of 6ûMy (miiiion years) for these asexuals, there is littie genetic variability within

lineages, suggestive that rare sex may be occurring or that these organisms have extremely

low evolutionary rates (Schon et al., 1998). Lasaea clams exhibit some of the highest

asexual divergences (16.2-22.9%) from their sexual counterparts, but the evidence is

incomplete to unambiguously make a clah for antiquity (6Foighil and Smith, 1995). Timema walking-sticks are a usehl system for studying the question of antiquity since there are five parthenogenetic (asexual) species, each one with a morphologically close sexual relative (Sandoval et al., 1998). Previous phylogenetic work suggested that Timema asexuds were evolutionarily long lived, dating to around the start of the latest round of glaciations approximately 2 -3 million years (My) ago (Sandoval et al., 1998). However, this phylogeny contained Lirnited inter- and intra- specific sampling; two parthenogens, T. shepardii and T. monikensis, as weil as four recently discovered sexuds T. poppensis, T. knulli, T. petita, and T. landelsensis were lacking from this dataset (Vickery, 1993; Vickery and Sandoval, 2001, submitted). As well, T. poppensis, instead of T. califomicum as reported in Sandoval et al., 1998, is a more closely related sexual relative of T. douglasi, based on morphology and host plant use. Addition of these taxa to the phylogeny is needed to better estimate the ages of Timema parîhenogens and further evaluate relationships in this group.

More recent dates of parthenogenesis in Timema than estimated in Sandoval et al.

(1998) may be plausible for two reasons. First, asexuality could have a more recent origin if the asexud species drove its sexual progenitors to extinction, if asexuals replaced sexuals, or if a doser sexual relative has not been sampled (Sandoval et al., 1998; Normark, 1999). The presence of morphologicaiiy similar sexual relatives seems to preclude this latter idea

(Sandoval et al., 1998). Second, determining the age of parthenogens should rely on more than just genetic divergence from the closest sexual relative (Judson and Normark, 1996;

Littie and Herbert, 1996). To make a strong daim for ancient parthenogenesis, ali extant taxa should be analysed with dates based on divergence within a putative asexual clade

(Judson and Normark, 1996; Little and Herbert, 1996). Materials and Methods

Collection and DNA Samvhg

Refer to Chapter 2 (pages 10- 17).

Percent Divergence and Phvlosnetic inferences

Sequences fiom 168 sampled Timema individuais representing 19 narned species were aligned by eye using the program Se-Ai version 1.0 alpha 1 (Rarnbaut, 1996). Three outgroups were used as in Sandovai et al. ( 1998) - the phasmids Baculum extradentatum and

Anisomorphu buprestoidea, and the cockroach Blatella germanica.

The overall dataset was analysed with neighbour joining (NJ) (under Kimura 2- parameter model) and maximum pmimony (MP) (heuristic search) in PAUP 4.0b8

(Swofford, 2000). Node support was assessed using bootstrapping. To estimate ages of parthenogenesis, the validity of a molecular clock was tested. However, since an analysis of the whole dataset was too computationally extensive under maximum likelihood (ML), the dataset was pruned to contain only sexual: asexual pairs and ML was performed separately on each pair. This p~ningcorresponded with the sexual: asexual relationships as determined under NJ and MP. ML was performed using quartet puzzling (QP) with the program TREE-

PUZZLE 5.0 (Strimmer and Haeseler, 1996). Rate heterogeneity was gamma distributed under an HKY model of evolution and internal nodes were assessed with QP reliabiiity

(Stnmmer and Haeseler, 1996). For each sexual: asexual pair, a ML phylogenetic tree was constructed and the likiihood ratio test was used to test the assumption of a local molecular clock (Strimmer and Haeseler, 1996).

The number of independent origins of asexuality in Timema was determined by assessing the monophyly of each asexual taon (T. douglasi, T. shepardii, T. monikensis, T. genevieve, and T. tahw). Origination of asexuality was assumed to be irreveaible. To estimate the age of parthenogenesis in Timema, two measures were used. A minimum age was determined using sequence divergence within each asexud based upon an arthropod rnolecular dock for mtDNA of 2% per million years (Brower, 1994; Juan et al., 1995, 1996;

Sandoval et al, 1998). A maximum age was estimated at the node corresponding to the most recent ancestor of the asexual and its closest sexual relative.

Ph! loeenetic Inferences

The strict consensus parsirnony tree and neighbow joining tree were very similar in topology, with high bootstraps (>70%) for most of nodes (Figures 2-5). The main clades

(referred to here as the Northem, Santa Barbara, and Southem clades; Figures 2-5) corresponded to the geography of California (as discussed in Chapter 2). The Northem clade contained the sexual: asexual species pairs T. poppensis: T. dougbsi, and T. cal~omicum:T. shepardii, as well as the sexual species T. knulli, T. landelsensis, and T. petita. The Santa

Barbara clade contained the T. cristinae: T. monikensis pair. The Southem clade contained the sexual: asexual pairs T. poàum: T. genevieve and T. bartmani: T. tahoe, as weU as the sexual T. boharti.

Evaluation of a Molecular Clock for Each Parthenom

Within the Northem clade, T. douglasi and T. shepardii had the same or very closely related haplotypes, and these haplotypes were both closely related to those of T. poppensis.

These asexuals did not meet the criteria of monophyly. A molecular clock was rejected for this grouping, regardless of whether T. shepardii was paired with T. californicm (its close morphologicai sexual relative) or with T. poppensis (its closest mtDNA ancestor). In the

Santa Barbara clade, T. monikensis was monophyletic and clustered with its close sexual relative T. cristinae. A molecular clock was not rejected for T. cristinae: T. monikensis pair

(-log likelihood without clock: -1345.85; with clock: -1357.63: x2= 23.56; d.f. = 23; p>0.05).

In the Southern clade, the sexual: asexual pairs T. podura: T. genevieve and T. bamani: T. rahoe can be separated to respective sexual: asexual pairs based on morphology and host plant use. Under ML, the T. podura: T. genevieve pair was clocWike (-log likelihood without clock: -797.69; with clock: -805.32; x2= 15.25; d.f. = Il; pAI.05). The T. barnani: T. tahoe pair was also clocklike (-log likelihood without clock: -749.10; with clock: -756.12; $=

14.03; d.f. = 7; p>0.05). Aee and Diversitv of Parthenogens

Under the assumption that asexuality cannot revert back to sex, there were four or five independent origins of asexuality in Timemu. T. douglasi and T. shepardii, the two parthenogens in the Northern clade, were very closely related (0.3 11 0.35% divergence;

Table 9). In fact, there was less divergence between T. douglasi and T. shepardii than within T. douglasi (0.771 0.58%; Table 9). Consequently, it is difficult to determine, using this dataset, if these species represent independent origins. As a conservative estimate, T. shepardii does not represent an independent origin of asexuality. T. monikensis, T. genevieve, and, T. tahoe were each monophyletic and nested with their respective sexual counterparts (T. cristinae, T. podura, and T. bartnani). Each of these events was inferred to represent an independent origin of parthenogenesis.

Average sequence divergence and estimated ages for each asexual are given in Tables

9 and 10. In the Northern clade, both T. douglasi and T. shepardii were closely related to T. poppensis (0.98 I0.64% divergence; Table 9). Although a molecular clock was not valid for this clade, if there was rate heterogeneity clonal ages can be roughly estimated using sequence divergence (Quattro and Avise, 1998). There was so little divergence within this group that it is reasonable to infer that these asexuals are relatively young (O. 1- 0.5 My;

Table 10). If however, T. califomicum turns out to be the me sexual relative of T. shepardii, then an age of O. 1- 1.3 My (Table 10) could be applied to this asexual.

T. monikensis fit the criteria of monophyly. There was a relatively low percent divergence within T. monikensis (0.63f 0.41%; Table 9) population but a high divergence between T. monikensis and T. cristinae (3.8% 0.59%; Table 9). The best evidence for Table 9: Average sequence divergence wiihin parthenogenetic lineages and to closest sexual relative for rntDNA COI. Since T. douglasi and T. shepurdii were closely related for this dataset, the sequence divergence between these taxa is also shown. The nearest sexual relative is the genetically closest materna1 lineage, as shown in the phylogenetic me. In the case of T. shepardii, two divergence measures are given - the fmt * is to the closest morphological relative (T.califomicm), and the second **, with T. douglasi, is to the mtDNA closest relative (T. popperisis). See text for details.

Parthenogen Average within group Divergence to closest divergence (%) sexual relative (%) T. douglasi 0.77 f 0.58 1.O3 f 0.63 T. sheiardii * 0.25 f 0.25 2.57 f 0.78 T. douglasi and T, shepardii ** 0.31 f 0.35 0.98 f 0.64 T. monikensis 0.63 f 0.4 1 3.89 f 0.59 T. genevieve 1.17f 0.73 4.35 f 0.91 T. tahoe 0.51 f 0.25 2.93 I 0.36 Table 10: Estimated age range of parthenogens based on 2% divergence per million years. The minimum age was based on divergence within a parthenogenetic lineage. The maximum age was calculated to the closest sexual relative. In the case of T. shepardii, two age estimates are given - the first * range is within only T. shepardii and to its closest rnorphological relative (T.californiciun). The second ** estimated age range, with T. douglasi, is to the mtDNA closest relative (T.poppensis).

Parthenogen Estimated age range (My) T. douglasi 0.2 - 0.5 T. shepardii * 0.1 - 1.3 T. douglasi and T. shepardii ** O. 1 - 0.5 T. monikensis 0.3 - 1.9 T. genevieve 1 .O - 2.2 T. tahoe 0.3 - 1.5 antiquity in this genus cornes from the sexual: asexual pair T. podura: T. genevieve. The average T. genevieve inîraspecific divergence was 1.17f 0.73% (Table 9). However, there were two geographically and genetically distinct T. genevieve populations that diverge from one another an average of 2.1 1%. The estimated age range of T. genevieve was 1.0- 2.2 My

(Table 10). T. tahoe was also monophyletic with a low intraspecific divergence (0.5If 0.25;

Table 9). This species diverged 2.932 0.36% (Table 9) hmits closest sexual relative, T. bamnuni. From this data, an age range of 0.3-1.5 My (Table 10) was inferred for T. tahoe.

Discussion

One of the problems with determining antiquity is that the term has not been well defined - it is debatable what ancient actuaily means in an evolutionary sense. Maynard

Smith (1992) States that 100,000 years is not very long in evolutionary terms (referring to

Poeciliopsis). However, Vrîjenhoek (1993)points out that this time range means 200 000 generations in Poeciliopsis, and that 200 000 generations is sufficiently long for asexuals to accumulate deleterious mutations. More recently, age estimates have focused on millions of years even though it is biologicaily more relevant to discuss age in terms of number of generations. What constitutes a relevant timescaie is open to debate (Griffith and Butlin,

1995). For the purpose of rhis paper, as a conservathe estimate, 500 000 - 1 million generations will be considered long enough to warrant a daim of antiquity. It seems ceasonable to assume that this timescale is long enough to weed out the evolutionary dead ends, as asexuals are so ofien termed. At one generation per year, if Timemu parthenogens are ancien&then individuais in an asexuai heage sfiould fonn a monophyletic clade that diverges from its closest sexual relative on the or&r of 106 My. If Timema parthenogens are not ancient then individuals in an asexual Iineage should have a more recent time of divergence.

Al1 methods of tree reconstruction agree with one another for major clades, and bootstrap support was generally good, indicaihg that our estimate of the evolutionary history for Timema is robust. Timema asexuals show remaicable differences in ages. The parthenogens T. douglasi and T. shepardii appear to be quite young (O. 1- 0.5 My) while T. genevieve meets the criteria for antiquity (1- 2.2 My). Al1 sampled populations of T. douglasi and T. shepardii are closely related to one another, and bath are closely related to

T. poppensis. Hypotheses such as a species treelgene tree ambiguity. the inability to detect the closest T. californicum, or that T. poppensis is the tme sexual relative, are difficult to distinguish between without a neutral nuclear marker (discussed in Chapter 2, pages 5 1-53).

The evidence for antiquity, or a lack thereof, in T. monikensis and T. tahoe is weak since for each of these parthenogens only one population was collected, and the divergence within that population was relatively low. Both of these asexual were substantially diverged from respective sexual counterparts. The inferred age range for T. monikensis is 0.3- 1.9 My and for T. tahoe is 0.3 - 1.5 My. There are three possible expianations that could account for this limited intraspecific divergence. First, these parthenogens are of recent origin. In this case, the high divergence from respective sexual counterparts couid be due to an inadequate sampling scheme or to extinction of sexual progenitors. Second, sampling effort could bias or iimit the results. Generally, specimens were coiiected from easily obtainable locations and often individuals were sampled from a single plant. Since a single female would produce geneticaily identical offspring, it is reasonable to assume that individuals sampled geograpùically close to one another could be the products of one (or few) females. Where and when a new clone arises is not known; it is also unknown how different clones may compete with one another for similar ecological resources. Spatial variation and non-random distribution of clone frequencies has ken shown in other taxa (Christensen and Noer, 1986;

Schenck and Vrijenhoek, 1986) and asexual populations may be geographically structured in such a way that limited sampling wodd reduce the arnount of detected genetic variability.

Many authors agree that sampling effort can influence age estimate if the closest sexual relative is not found (Sandoval et al., 1998; Johnson and Bragg, 1999; Normark, 1999) but the possibility that inadequate sampling could reduce the arnount of observed divergence within an asexual clade has largely been ignored.

Third, rare sex may occur. A claim for antiquity should be based upon cornpletely asexual clades and problematically, asexuality is often based upon negative evidence (usuaily lack of males) (Judson and Normark, 1996; Normark, 1999). Males were sampled in the T. monikensis population in 1999 (four males io 16 fernales). It is not know if these males were viable although it is not uncornmon for parthenogens to occasionally produce males

(Soumalainen et al., 1987). The presence of males could be taken as evidence for sex

(Judson and Nonnark, 1996; Little and Herbert, 1996; Normark, 1999). Hunt and Peck

(1996) suggest that a little bit of sex may be enough to drive a lineage to only sex. However,

the evidence for T. monikensis is inconsistent wiih this idea. T. monikensis is monophyletic

and substantially diverged from T. cristirne (1.9 My) yet it still retains predorninantly

parthenogenesis.

Studying taxa that have both sexual and asexual lineages gives us the oppoctunity to

look for reasons why asexuals pecsist and in tum, why sex ever evolved (Johnson and Bragg, 1999). The distribution of asexuality versus sex has often been considered in terms of different processes acting at the short term versus the long tenn (Moritz, 1991; Griffiths and

Butlin, 1995; Norrnack, 1999). Timema walking-sticks are unique in that there appear to be both evolutionarily young and old asexual species. The question then becomes - what is different between young and ancient parthenogens with respect to their sexuai counterparts?

Differences in young versus old asexuality generally appear to correlate with geography - northern asexuais are young while ones that are more southern are older. The overail phylogeography is consistent with the idea that Thema originated in the south and has since expanded north (see Chapter 2). In the north, T. douglasi and T. shepardii are recently diverged and geographically widespread, consistent with a hypothesis of recent range expansion. These asexuais also roughly overlap in geographic distribution by 200k.m at their southemmost lirnits with the closest genetic sexual T. poppensis. Although T. poppensis is more coastally situated, the close proximity of the taxa suggests that interactions between asexuals and sexuals may be important. Conversely, T. genevieve, the inferred ancient parthenogen, is geographically separated fiom its closest sexual relative T. podura by more than 250km. This observation suggests that Z'. genevieve may have dispersed far enough to limit competition with its sexual cornterpart. This evidence suggests that the northern asexuals could be young not only to having arisen more recently but also due to cornpetitive interactions with sexual counterparts, where as T. genevieve is old because it is geographicaiiy separated and not in direct competition with its sexual counterpart.

This work is of importance because asexuals are generaily expected to be evolutionarily short lived (Bel, 1982; Judson and Nomark, 1996, Little and Herbert, 1996).

Ancient asexuais must have evolved mecbaaisms to avoid effects of deleterious mutations or escape from parasites and if they exist they may provide clues as to why sex is prevaient in so many organisms (Judson and Normark, 1996, Little and Herbert, 1996). Timema is exceptional in having both short-term and long-tem persistence of parthenogens, and thus it

provides the opportunity to critically compare hypotbeses about dispersal rates, cornpetitive ability, and conditions that favour sex over asexuality. Bell, Graham. 1982. The Masterpiece of Nature: The Evolution and Genetics of Sexuality. Croom Helm, London.

Brower, A. V. 2. 1994. Rapid morphological radiation and convergence among races of the butterfly Heliconius erato inferred from patterns of mitochondrial DNA evolution. Proceedings of the National Academy of Science USA 91: 649 1-6495.

Christensen, B. and H. Noer. 1986. Spatial and temporal components of genetic variation in triploid parthenogenetic Trichoniscus pusillus (Isopoda: Crustacea). Hereditas 105: 277-285.

Cuellar, 0. 1977. Animal parthenogenesis: A new evolutionary mode1 is needed. Science 197: 837-843.

Gerritsen, J. 1980. Sex and parthenogenesis in sparse populations. The American Naturalist 115: 7 18-742.

Griffiths, H. I., and R. K. Butlin. 1995. A timescale for sex versus parthenogenesis: evidence from subfossil ostracods. Proceedings of the Royal Society London B 260: 65-7 1.

Hurst, L. D. and J. R. Peck. 1996. Recent advances in understanding of the evolution and maintenance of sex. Trends in Ecology and Evolution 11: 46-52.

Johnson S. G.and E. Bragg. 1999. Age and polyphyletic origins of hybrid and spontaneous parthenogenetic Campelorna (Gastropoda: Viviparidae) from the southeastem United States. Evolution 56: 1769- 178 1.

Juan, C., P. Oromi and G. M Hewitt. 1995. Mitochondrial DNA phylogeny and sequential colonization of the Canary Islands by darkling beetles of the genus Pimelia (Tenebrionidae). Proceedings of the Royal Society London B 261: 173- 180.

Juan, C., P. Oromi and G. M. Hewitt. 1996. Phylogeny of the genus Hegeter (Tenebrionidae, Coleoptera) and its colonization of ihe Canary Islands deduced form Cytochrome Oxidase 1 mitochondrial DNA sequence. Heredity 76: 392-403.

Judson, O. P., and B. B. Nomark. 1996. Ancient asexual scandais. Trends in Ecology and Evolution 11: 41-46.

Little, T.J., and P. D. N. Hebert. 1996. Ancient asexuals: scanda1 or artifact? (letter) Trends in Ecology and Evolution 11: 296. Maynard Smith, J. 1992. Age and the unisexual lineage. Nature 356: 661462.

Moritz, C. 199. The origin and evolution of parthenogenesis in Heteronotia binoei (Gekkonidae): evidence for recent and locaiized origins of widespread clones. Genetics 129: 2 1 1-219.

Normark, B. B. 1999. Evolution in a putatively ancient asexual aphid lineage: recombination and rapid karyotype change. Evolution 53: 1453- 1469.

6 FoigM, D., and M. J. Smith. 1995. Evolution of asexuality in the cosmopolitan marine clam Lusaeu. Evolution 49: 140- 150.

Rarnbaut, A. 1996-201. Se-AI: Sequence alignment editor v2.0. Umvcisity of Oxford. availablr: at: http://evolve.z~.ox.ac.uW~~ftware/Se-SVmain.html(21 lune 2001).

Sandoval, C. P. 1994. The effects of the relative geogriipbic scales of gene flow and selection on the morph frequencies in the walking-stick Timema cristinae. Evolution 48: 1866-1879.

Sanduvai, C., D.A. Camean, and B. I. Crespi. 1998. Molecular phylogenetics of sexual and parthenogenetic Timema walking-sticks. Raceedings of the Royal Society of London B 265: 589-595.

Schenck, R. A,, and R. C. Vrijenhoek. 1986. Spatial and temporal factors affecting the coexistence among sexual and clonal fumof Poeciliopsis. Evolution 40: 1060- L070.

Schon, I., K. Martens, and V. Rossi. 1996. Ancient asexuals: scanda1 or artifact? (letter) Trends in Ecology and Evolutionll: 2% - 297.

Schon, L, R. K. Butlin, H. 1. Griffith, and K Martens. 1998. Slow rnolecular evolution in and ancient asexuai ostracod. Proceedings of the Royal Society of London B 26s: 235-242.

Soumalainen, E., A. Saura, and J. Lokki. 1987. Cytology and Evolution in Parthenogenesis. CRC Press, Boca Raton, FL, USA.

Stems, S. C., ed. 1987. Tbe Evolution of Sex and its Consequences. Birkhauser Verlag, Boston.

Strimmer, EL, and A. von Haeseler. 1996. Quartet puzzling: A quartet maximum iikelihood mehifor reconstnicting tree topologies. Molecular Biology Evolution 133964-969.

Swofford, D. L. 2000. PAUP*. Phylogenetic analysis using parsimony (*and other methods). Version 4. Sinauer Associates, Sunderland, Massachusetts. Taylor, D. J., T. J. Crease, and W. M. Brown. 1999. Phylogenetic evidence for a single long lived clade of crustacean cyclic parthenogens and its implications for the evolution of sex. Proceeding of the Royal Society London B 266: 79791-797.

Vickery, V. R. 1993. Revision of Timema Scudder (Phasmatoptera: Timematodea) including three new species. Canadian Entomologist 125: 657-692,

Vickery, V. R. and C. P. Sandoval. 2001. Description of three new species of Timema (Phasmoptera: Timematodea: Timematidae). Submitted.

Vrijenhoek, R. C. 1993. The origin and evolution of clones versus the maintenance of sex in Poeciliopsis. Journal of Heredity 84: 388-395 Chapter 4: Concludiag Remah

Using phylogenetic methods to study the evolution of geographic parthenogenesis and the persistence of asexuality has resulted in the following inferences for Timema walking-sticks. First, parthenogenesis has arisen four, possibly five, times in the genus.

Second, Timema appears to have originated in the south and since expanded north, consistent with previous findings (Sandoval et al., 1998). Third, major phylogenetic subdivisions, the

Northem, Santa Barbara, and Southern clades, correspond with the geography of California.

Fourth, the pattem of geographic parthenogenesis, namely the observation that Timema asexuals are more northerly in distribution, appears to be the result of range expansion.

Finally ,Timema appears to have both short-term and long- tenn persistence of asexuality .

This study is the first to critically address the pattern of geographic parthenogenesis in a testable manner. In each case of parthenogenesis, geographic patterns are the result of range expansion. Both sexuals and asexuals have expanded their ranges northwards, but asexuais have expanded further. This inference suggests that demographic effects, specificaily a higher dispersal ability of asexuais, are critical factors leading to geographic patterns of sex versus asexuality. However, the phylogeographic subdivisions, combined with the inference that Northem clade asexuais are evolutionariiy young and geographically close to sexual counterparts while at Southern clade asexuals are older and distantly located from sexual relatives, suggests that ecological cornpetitive factors may also be important.

What are the critical ecological and genetic factors that dow some asexuals to persist while others do not? The Timema system, with young and old, north and south, splits provides the unique opportunity to test alternative hypotheses for the evolution of sex. Future work that may provide a greater understanding of these conclusions by:

1) Examining whether or not asexuality in Timema actudy confers a iwo-fold reproductive advantage, and how this advantage Links to dispersai ability. It is generally assumed that asexuals have a higher reproductive rate but it is unknown in Timemu what this rate is actually. Understanding îhis advantage rnay aüow us to more naiisticaily infer dispersai abiiity. 2) bvestigating competitive abilities of asexual species with respect to sexual counterparts, and studying in more detail ecological conditions that favour sex over asexuality. AIthough the general ecology of Timema is known, possible differences in habitat and competitive abiiity between sexual and asexual species has not been well-studied.

3) Conducting fine-scde sampling of asexual and sexual species to better understand genetic variability on a geographic scaie. In particular, sampling biases may lirnit my conclusions, especiaily if a more closely related sexual species exist for each asexuai. 4) Using a faster evolving genetic marker to resolve species ambiguities. Although the sexud: asexual pairs

T. poppensis: T. douglasi and T. californicum: T. shepardii are described species and distinguishable based on morphology and host plant use, these species are closely related based on my analyses using mtDNA. Using different neutrai markers may clarify this situation. Nuclear markers cm also be used to determine if asexuality arises via hybridisation in Timema. 5) Comparing T. genevieve to other old asexuals to elucidate factors ailowing asexuality to persist. The Messelson rnethod can also be used to compare asexuai and sexuai heterozygosity.

in summary, my thesis has pmvided a strong basis for further study in this system.

The unexpected conclusion that there are distinct differences between north and south, yowig and old, asexual species provides a unique opportwiity to determine ecological and genetic factors leading to the evolution of sex. Sandoval, C., D.A. Cannean, and B. J. Crespi. 1998. Molecular phylogenetics of sexual and panhenogenetic Thema walking-sticks. Proceedings of the Royal Society of London B 265: 589-595. Appendix A: Locality and GPS information for Tirnema haplotypes.

Host Plants A = Pseudotsuga menziesii (Douglas-fir) B = Sequoia sempervirens (Califocnian redwood) C = Arctostaphylos spp. (manzanita) D = Ceanothus spp. E = Adenostoma fasiculatm (chamise) F = Abies concolor (White fis) G = Quercus spp. (Oak) H = Heteromeles arbutifolia (Toyon) 1 = Cercocarpus spp. J = Juniperus spp Location GPS Coordimîea Poppe Rd., Howell Min 3835725N1222666QW Poppe Rd., Howell Mtn 3835725N lZ22666SW Poppe Rd.. Hwvell Mtn 3835725N12226689W Ink Grade Rd., Howell Min 3835725N 12226669W Ink Grade Rd.. Howell Mtn 3835725N1222666QW Ink Grade Rd., HoweU Mtn 3835725N12226669W Ink Grade Rd., Howell Min 3835 725 N l222666QW Santa Clara Loma Priete Way 3706252N 121 WïïOW Santa Ciara Loma Prieta Way 37 06 252 N 121 52 770 W Santa Clara Loma Prieta Way 37 06 252 N 121 52 770 W SMKHna Tin Barn Rd. 3837093N 12317535W Sanoma Tin Barn Rd. 3837093N 123 17535 W Çanta Ciara Loma Prieta Way 3708252N 121 5277OW Santa cm8 Loma Priete Way 37062S2Nl2l52ïïOW Santa ClaraiCfuz Summit Rd. 3713340N12205270W Santa CtarelCniz Sununii Rd. 3713340N 12205270W Humbddi King Mbi 4008223N 12404358W Humbdd King Mtn 40 08 223 N 124 04 358 W Humboldt Shwaly 4026759Nl2358581 W Humboldt Bald Hilis Rd. mile 9 41 12264N l2357508W Humbddt Bald Hilis Rd. mile 9 41 12264N 12357508W Humboldt King Mtn 4008223N 1240435ôW Humboldt King Mtn 4008223N 12404358W King Min 2 4008589N 12405047W King Mtn 2 40 08 589 N 124 05 047 W Sonoma FiRock Rd. 3 3850317 N 12331 029W Smma Fish Rock Rd. 3 3850317N 12331 029 W Sonoma Tin Barn Rd. 38 37 093 N 123 17 535 W Sonoma Fish Rock Rd. 3 3850317N 12331 029 W Somma Fish Rock Rd. 2 38 49 389 N 123 34 885 W Sonoma Fish Rock Rd. 4 3853117N12322505W Sonama Fidi Rock Rd. 4 3853117N12322505W Somma Fish Rock Rd. 4 3853 ll7N 12322505 W Somma Fish Rock Rd. 4 3853117N12322505W Hoot Pknt County Locatlon popl3nol A Sonoma Fish Rock Rd. 1 popl4nol B Santa ClaraCruz Summiî Rd. pop14no2 B Santa Ctara/Cruz Summiî Rd. pop14no3 B Santa ClaralCniz Summit Rd. douglnol A Mendocino Orr Springs Rd. dwglno2 A Mendocino On Springs Rd. shep1nol4 C Mendocino On Springs Rd. shepl no5 C Mendocino Orr Springs Rd. douglno3 A Mendocino OR Springs Rd. dwglna4 A Mendocino Orr Springs Rd. doug2no2 A Humboldt Baki Hilb Rd. mile 9 CalEnol C, O, E San Luls Obispo Cuesta RiQge, Santa Lucia Mtns ca112no2 C, D, E San Luis OMspo Cuesta Rklge, Santa Lucia Mtns caIi2no3 C. D, E San Luis 0- Cueata Riâge, Santa Luci Mtns calbnol Montery Arroyo Seco NovûcaiCV Montery Arroyo Seco cau4nol Santa Clara /Cruz Loma Prieta caU5nol Santa Clara Corraiiios Canyon pop7nol A Santa Ciara /Cruz Loma Prieta x Boche pop7no2 A Santa Clara /Cruz Loma Prieta x Boche PoP7no3 A Santa Clara /Cruz LmPrieta x Boche calisnol C, E, G Santa ClaraCruz Summit Rd. cali6no2 C. E, G Santa ClardCruz Summit Rd. cali6no3 C, E, G Santa ClaraCruz Summit Rd. popSn02 B Santa ClaraCruz Summit Rd. caIi7no2 Santa ClaralCniz Loma Priela x Boche ca117no3 Santa ClaralCruz Loma Prieta x Boche ~all71104 Santa ClaraiCruz Loma Prieia x Boche rnar3call.P Santa Clara/Cruz Loma Prieta cali9nol C, G Santa ClardCruz Summit Rd. caliünol G Santa Clara Lkk Obs, Mi Hmailîon cali8nO2 G Santa Clara Lkk Obs, Mt Hmaiîion ca118no3 G Santa Clara Lik Obs, Mt Hrnaiiîon novlOcali G Santa Clara Lkk Obs, Mt Hmailton sheplnol C, G Mendocino Orr Springs Rd. noit c-& Plant County Location GPS CoordiMt.. shepl no2 Mendocino On Springs Rd. 3811 559N l23lS707W shepl no3 C, G Mendocino On Springs Rd. 39 11 559 N 123 15 707 W shep5nol C, G Mendocino Elk Mtn 39 16 729 N 122 55 546 W shep5no2 C Mendocino Eik Mtn 39 16729 N 12255 546 W shep2no4 C Mendocino Hopland Research Stn 3857aNl2307479W s-1 C Sonoma King Ridge Rd. 3835734N12309569W shep3no3 C Sonoma King Ridge Rd 3835734N l2309589W shep4no2 C Mendocino near Leytonville 39W2ll N 12307948 W sheprl- C Mendocino near Laytonville 3904211 N l2327948W shepeno2 C Dei Norte Big Flat Rd. 41 42 035 N 12348 574 W C Del Norte Big Flat Rd. 41 42-N 12348574W shep2nol C Mendocino Hopland Research Sin 3857320N 12307479W C Mendocino Hopland Research çtn 38 57 320 N 123 O7 479 W -2- C Mendocino Hopland Research Stn 38 57 320 N 123 O7 479 W Wnol C Mendocino near Laytonville 3QW2llN12327948W shepsnol C Del Norte Big Flat Rd. 41 42-N 12348574W 3marlcri8 Santa Barbara Santa Ynez Min8 Hwy 154 3431 ûOûN11948000W cris1nol D Santa Barbara ojala 3429488NllB18367W Crlslno2 O Santa Barbara Ojaia 3429488N11918367W crisln03 D Santa Barbara OWa W29488N 119 l8367W cris3nol 0,H Santa Barbara Santa Ynez Mtns Gavloie 342926QN 120 13589 W crls3no2 0,H Santa Barbara Santa Ynez Mtns Gavrota 34 29 269 N 120 13 569 W cris2nol O Santa Barbara Santa Ynez Mtns Refugio Rd.. 34 30 950 N 120 04 389 W cris2no2 D Santa Barbara Santa Ynez Mtns Refugio Rd.. 34 30 950 N 120 04 389 W crisclno1 E Santa Barbara Santa Ynez Mtns Refugkï Rd.. 343089ïN l2004278W cris41102 E Santa Barbara Santa Ynez Mtns Refugb Rd.. 34 30 897 N 120 04 278 W laureil 3431 OM)N 11948000W laure12 3431 ûOûN 11948000W laure13 3431 000N 11948000W WPY 1 3428ûOûN11946111 W WPy2 3428000N 11946 111 W para1 3431500Nl1951OWW para2 3431500N11951000W housel 3430200N11950100W house2 3430200N11950100W sp.ck. County Location monlnol T. monikensls Los Angeles Santa Monka Min rnMilno3 T. moniken- Los Angeles Santa Monka Mtn Inonln04 T. monikensls Los Angeles Santa Monka Mtn monlm2 T. tnonikensls Los Angeles SantaMonica Mm monln05 T. monikends Los Angeies Santa Monidta Mtn novl 1podu T. Mura Tulare Sequda NP podlnol T. podura Sanla Barbara Hwy 166 podt- T. padura Santa Barbara Hwy 166 pDdlno2 T. podura Santa Barbara Hwy 166 pod2nol T. podura San Diego Teicate DMde pod2no2 T. podura San Diego Tecab DM& pomno3 T. podura San Dlego Tecale Divide T. podura Riverside Hwy 243-1 pod5nol T. podura RiverMe Hwy 243-3 T. podura Riverside Hwy 243-3 junspodÇJ T. podura Riversade Hw~243-3 podenol T. podura Riverside Hwy 243-2 mm2 T. podura Riverside Hwy 243-2 T. Mura San DIego Pakmar Observatory mm2 T. podura San Diego Palomar Observatory wm T. podura San Diego Pakmar Observatory genlnol T. genevievae Lake Hwy 20 Rd. mk 38.22 genlno2 T. genevievae Lake Hwy 20 rd mk 38.22 genlno3 T. genevievae Lake Hwy 20 rd mk 38.22 genl no4 T. genevievae Lake Hwy 20 rd mk 38.22 gen4nol T. genevievae Cdusa Cook Springs Rd. gen4no2 T. genevievae Cdusa Cook Springs Rd. gen2nol T. genevievae Santa Clara Mine Rd. gen2m2 T. genevievae Santa Clara Mine Rd. gen2no3 T. genevievae Santa Clara Mine Rd. gen-1 T. genevievae Santa Clara Del Puerto Rd. gen-2 T. genevievae Santa Clara Del Puerto Rd. mar4genev T. genevievae Santa Clara San Ardd Del Puerto Rd. tbartrnani T. bartmani San Bernadino Camp Meadow bartlnol T. barfmani San Bernadino Camp Meadow bartlno2 San Bernadino Camp Meadow

Appendix B: Detailed collection information for Timema specimens collected during 1999 and 2000. Species Host Plant GPS co-ordinates Location Numbcr of Date(8) lndividuals Collccted

T. cristinae Adenostonla Santa Barbara Co.: Santa Ynez Mtns. on 4 male March 23. 1999 fasicularuni sides of Refugio Rd. 14 female 13 juvenile

T. cristinae Ceanothus spp. Santa Barbara Co.: Santa Ynez Mtns, on 12 male March 23,1999 si&s of Refugio Rd. but ai higher 6 female elevation than chamise T. cristinae 6 juvenile collected in t his area

Ceanothus spp., Santa Barbara Co.: Santa Ynez Mins. near 17 male March 23. 1999 Heterorneles Gaviota (Gaviota Pass) 14 female arbutifolia l juvenile

Ceanothus spp. Santa Barbara Co.:on Hwy 33 northwest l male March 23, 1999 of Ojala and Ojai 4 female 2 juvenile

T. podura Adenos roma Santa Barbara Co.: on Hwy 166 E, called 3 male March 23, 1999 fasicularum Santa Maria population by C. Sandoval 8 female

T. califomicum Adenostoma San Luis Obispo Co.: Santa Luçia Mtns, I l male March 24.1- fasicularum, northwest of San Luis Obispo, Cuesta 8 female Arctosraphylos Ridge Boianical Area, near radio station spp., Ceanorhus Iower SPP. T. perira Ceanoihus olygarhus San Luis Obispo Co.: on Hwy I Coast I l male March 24, 1999 north of Simeon 9 female

T. knulli Sequoia Monierey Co.: Big Creek Resewe, along 8 male March 24, 1999 semprrvirens raad within reserve 12 female

T. knulli Ceanothus spp. Monterey Co.: Big Creek Resewe. at 3 male March 25.1999 higher elevation than redwood population, 3 female natchv ceanoihus alone trails Spccics Host Plant GPS CO-ordinates Location Number of Individuab

T. landelsensis Arctostaphylos spp. 36 1 1 365 N Monterey Co.: Big Creek Reserve. al top 2 male March 25, 1999 121 33 100 W of mountain in manzanita grove 5 female

T. chumash Quercus spp., 34 16 182 N Los Angeles Co.: on Hwy 2 just south of 10 juvenile March 27,1999 Ceanothus 118 10 105 W junciion with N3, vista poini turnout olyganthus

T. chumash Quercus spp., 34 1 1 433 N Los Angeles Co.: travelling nonh on Mt. 14 juvenile March 28. 1999 Ceanothus spp. 11740737 W Baldy Rd. towards Mt. Baldy, rd mk 3.53

T. chunush Quercus spp. 34 06 420 N San Bemadino Co.: on Hwy 38 rd mk 4 juvenile March 28, 1999 1 16 58 387 W 16.27

T. bartmani Abies concolor 3409 813 N San Bernadino Co.: on Jenks Lake Rd. l female March 28, 1- 1 16 54 377 W off Hwy 38, at YMCA Camp Meadow 14 juvenile

T. chuniash Ceanothus spp, 33 48 907 N Riverside Co.: on Hwy 243 (stop 1). near -20 individuals March 29, 1999 11647462 W James Reserve, 0.5 miles from Bay Street Springs

Ceanoihus spp. 33 48 907 N Riverside Co.: on Hwy 243 (stop I), near 4 male March 29, 1999 Il6 47 462 W James Reserve. 0.5 miles from Bay Street 8 fde Springs 14 juvenile

T. chumash Ceanothus spp. 3351 096N Riverside Co.: on Hwy 243 (stop 2). near 10 juvenile March 29, 1999 II649 568 W "icy chain" sign tumout nrea

T. poduru Ceanoihus spp., 33 51 096 N Riverside Co.: on Hwy 243 (stop 2), near 4 male March 29.1999 Adenostoma 116 49 568 W "icy chain" sign turnoui area 10 female fasiculaium 4 juvenile

Cearrurhus spp. 33 47 842 N Riverside Co.: on Hwy 243 (stop 3). ai -15 individuais March 29, 1999 11646591 W "lndian Vista" turnout area

Ceur~orhusspp. 33 47 842 N Riverside Co.: on Hwy 243 (stop 3). at 5 juvenile March 29, 1999 11646591 W "lndian Vista" iumout ma Species Host Plant GPS CO-ordinates Location Number of Date(s) Individuals Collected

T. podura Ceanorhus spp. San Diego Co.: at Palomar Observatory 2 male March 29, 1999 off S7lS6 3 fernale

T. podura Adenosronia San Diego Co.: Tecate Divide. on Old 3 male March 30, 1999 fasicularum Hwy 80 l female 5 juvenile T. boharti Ceanorhus spp., San Diego Co.: on Laguna Mtn ai I male March 30, 1999 Adenostonta Monument Peak radio faciliiy (off SI) 5 female fasicularum T. ntonikensis Ce~OWPUS Los Angeles Co.: on Wesilake Blvd., 4 male April 2. 1999 Santa Monica Mtns 16 female

T. poppensis collectai on between Humboldi Co.: on borh si&s Shively Rd. 1 male June 26. 1999 Ceanothus spp. but 40 26 759 N ,going from Shively back to Hwy 1, 1 fernale may have been from 123 58 581 W relatively close IO town, "tïcld area ndyPseudorsuga and overlooking river, between forested ntenziesii or 40 27 235 N (Dwglas fir and redwmd) areas Sequoia 123 59 2% W sempervirens

T. poppensis Pseudorsuga 40 O8 223 N Humboldt Co.: on Kings Rmge Rd., a 4 male June 28. 1999 menziesii l24OQ 358 W grave1 rdgoing from Wilder Ridge Rd. 4 female and 10 Saddle Min, large relatively 40 OB 589 N homogenous paich of Douglas Gr, wiih a note: I male was 124 05 047 W bit of redwd found on redwd

Arcrostaphylos spp. 39 1 1 559 N Mendocino Co.: on Orr Springs Rd. near 9 bmaie June 29, 1999 ad 123 15 707 W Ukiah. just below "Bob and Joy's" Heleronie/es and houselfarm arburjfolia 39 1 1 493 N 123 15691 W and 39 1 1 635 N 123 16092 W Species Hast Plant GPS CO-ordinates Location Numbcr of Datds) Individumls Collected

T. douglasi Pseuâorsuga 39 12 047 N Mendocino Co.: on Orr Springs Rd. 14 female June 29, 1999 menziesii 123 17 636 W going towards Ukiah. just before rd mk ad 38.00 39 11 950 N 123 17 362 W

T. poppensis PseiIdotsuga 37 06 252 N Santa C'!ara Co.: off Imna Rieta Way, in 3 male June 30. 1999 menziesii 121 52 770 W siand of Douglas fir on right side of road 1 female heading towards top of mouniain, on a side Rd. going ioward house #26980

Arctostaphylos spp. 37 13 343 N Santa Clara Co.: on Summit Rd. going 3 male June 30, 1999 ad 122 O5 271 W from Loma Prieta towards San Francisco, 7 female Hereromeles stopped ai 'no parking' pull off, approx. arbutifolia 50m from a cail box, left side of rd. mixed note: 2 female were vegetation on manzanita. 5 femalel3 male on Heteromeles

T. poppensis Pseudoisuga 38 35 725 N Napa Co.: on Ink Grade Rd., in from 3 male July 3, 1999 menziesii 122 26 669 W north end of White Cottage Rd., l female approximately 200111. Howell Mtn T. genevieve Adenosroma 37 25 431 Santa Clara Co.: on Mines Rd., - 28 I female July 2, 1999 fasicularum 121 30 336 miles south of Livermore, large homogenous patch of chamise

T. poppensis Sequoia 37 01 692 N Santa ClardSanta Cruz Co.: on right side 12 male May 24,2000 sempervirens 121 44 415 W of Summit Rd. , going nonhwest from 8 femaie Mt. Madonna Rd. (no1 far from 5 juvenile iniersection. - I mile and 2 mile), in relatively young stand of redwoods

Santa ClardSanta Cruz Co.: un Summit l male May 24,2000 Rd. (travelling northwesi from Mt. 2 female Madonna Rd.) Spccies Host Plant GPS co-ordinates Location Number of Datc(s) Individuals ~otle~ted

T. poppensis Pseudotsuga 37 06 252 N Santa Clara Co.: off Loma Prieta Way, in 3 male May 24, 2000 menziesii 121 52 770 W stand of Douglas fir on right side of road 2 female heading towards top of mounrain, on a l juvenile side rd going toward house U26980

T. californicunt Quercus spp. 37 20 590 N Santa Clara Co.: on HWY 130 (San 2 male May 25,2000 121 38 188 Antonio Valley Rd.), just after the Lick 7 female Obsewatory. mainly on south side of road

T. genevieve Adenosrtrma 37 23 433 N Santa Clara Co.: on Del Puerto Rd.. large 6 female May 25,2000 fasiculatum 121 28 403 W homogenous patch of chamise, a few miles east of intersection with San Antonio ValleyMines Rd. T. genevieve Adenos toma 37 25 431 N Santa Clara Co.: on Mines Rd. - 28 4 female May 25.2000 fasiculatum 121 30 336 W miles south of Livermore. large homogenous patch of chamise

T. poppensis Pseudorsuga 38 35 725 N Napa Co.: on Ink Grade Rd.. in from 5 male May 26.2000 menziesii 122 26 669 W north end of White Cottage Rd., 4 female approximately 200113,Howell Min

T. shepardii Arctosraphylos spp. 38 35 734 N Sonoma Co.: near Guerneville, on King 8 female May 27,2000 123 09 589 W Ridge Rd., travelling northwest. large turnout area with good stand of manzanim. both sides of road, after mileU8 (if travelling north)

T. poppensis Pseudorsuga 38 37 093 N Sonoma Co.: on Tin Barn Rd. (- 1 mile 7 male May 27,2000 menziesii 123 17 535 W northwest from intersection with King 5 female Ridge Rd.), on east side of road before 2 juvenile mile ü4 (if travelling north)

T,poppensis Pseudotsuga 38 49 389 N Mendocina Co.: on Fish Rock Rd., 1 male May 28,2000 menziesii 123 34 885 W travelling easi from Fish Rock, at house 2 female M515 1 (also ac house #4457) Spccies Host Plant GPS CO-ordinates Location Number of Date(#) Individuals Collected

T. poppensis Pseudotsuga Mendocino Co.: on Fish Rock Rd. 3 male May 28.2000 menziesii travelling east, after tums into din road, l female road mark 6.82 1 juvenile

T. poppensis Pseudotsuga Mendocino Co.: on Fish Rock Rd. fi male May 28.2000 rnenziesii travelling easi, after tums back into paved 7 female road, before road mark 20.62

T. shepardii ArctosfuphyIos spp. Lake Co.: on Elk Min Rd., large stand of 15 female May 29,2000 manzaniia between road mark 12.62 and 12.73

T. genevieve Adenosloma Lake Co.: norîh side of HWY 20 ai road 8 female May 29.2ûûO fasicularum, + mark 38.22. around Cache Creek bed

T. genevieve Adenostowia Colusa Co.: on Cook Springs Rd. (after 2 female May 29.2000 fasicularum Vickery's revision), along Indian Creek bed?

T. shepardii Arctostaphylos spp. Mendocino Co.: near Hopland, on wesi 1 female May 30.2000 side of HWY 101, Lafichi Rd. (din pari) near #760

T. shepardii Arcfostaphylos spp, Mendacino Co.: near Laytonville. on 9 female May 31,2000 Layionvillc Dos Rios Rd. easi, large siand of manzanita on lefi side of road before road mark 1.70 (if travelling West)

T. shepardii Arctostaphylos spp. Del Norte Co.: on Big Rai Rd. near top 3 juvenile June 2,2000 of mountain (Hardin Min??) jusi before mile 20, large siand of manmita

T, douglasi Pseudotsuga Curry Co (Oregon): near Brookings, off 1 juvenile June 3, 2000 menziesii Rd. 784 towards "packers cabin". in relatively old siand of Douglas fir encircling a field area