Species Delineation Using Bayesian Model-Based Assignment

SPECIES DELINEATION USING BAYESIAN MODEL-BASED ASSIGNMENT

TESTS: ARE PHRYNOCEPHALUS VLANGALII AND PHRYNOCEPHALUS

PUTJATIA BOTH VALID SPECIES?

A Thesis

Presented to

The Faculty of Graduate Studies

The University of Guelph

DANIEL W.A. NOBLE

In partial fulfillment of requirements

for the degree of

Master of Science

July, 2009

395 Wellington Street 395, rue Wellington OttawaONK1A0N4 OttawaONK1A0N4 Canada Canada

Your file Voire reference ISBN: 978-0-494-56964-1 Our We Notre reference ISBN: 978-0-494-56964-1

NOTICE: AVIS:

The author has granted a non L'auteur a accorde une licence non exclusive exclusive license allowing Library and permettant a la Bibliotheque et Archives Archives Canada to reproduce, Canada de reproduire, publier, archiver, publish, archive, preserve, conserve, sauvegarder, conserver, transmettre au public communicate to the public by par telecommunication ou par I'lnternet, preter, telecommunication or on the Internet, distribuer et vendre des theses partout dans le loan, distribute and sell theses monde, a des fins commerciales ou autres, sur worldwide, for commercial or non support microforme, papier, electronique et/ou commercial purposes, in microform, autres formats. paper, electronic and/or any other formats.

The author retains copyright L'auteur conserve la propriete du droit d'auteur ownership and moral rights in this et des droits moraux qui protege cette these. Ni thesis. Neither the thesis nor la these ni des extraits substantiels de celle-ci substantial extracts from it may be ne doivent etre imprimes ou autrement printed or otherwise reproduced reproduits sans son autorisation. without the author's permission.

In compliance with the Canadian Conformement a la loi canadienne sur la Privacy Act some supporting forms protection de la vie privee, quelques may have been removed from this formulaires secondaires ont ete enleves de thesis. cette these.

While these forms may be included Bien que ces formulaires aient inclus dans in the document page count, their la pagination, il n'y aura aucun contenu removal does not represent any loss manquant. of content from the thesis.

1+1 Canada ABSTRACT

SPECIES DELINEATION USING BAYESIAN MODEL-BASED ASSIGNMENT

TESTS: ARE PHRYNOCEPHALUS VLANGALII AND PHRYNOCEPHALUS

PUTJATIA BOTH VALID SPECIES?

Daniel W.A. Noble Advisor: University of Guelph, 2009 Dr. Jinzhong Fu

I use microsatellite DNA markers and Bayesian assignment tests to discern between species level hypotheses (one vs. two species) in toad-headed agamas of the

Phrynocephalus vlangalii complex. Eight microsatellite loci were amplified and assignment test analyses revealed two genotypic clusters east and west of Qinghai Lake,

Qinghai province, China. The genomes of the two clusters remained distinct in sympatry.

In addition, a fragment of the mitochondrial ND2 gene was amplified and a gene tree was constructed. The two microsatellite clusters largely formed reciprocal monophyletic groups, although several historical mitochondrial introgression events between the two groups were also revealed. These results support the two-species hypothesis that both P. vlangalii and the second controversial species, P. putjatia, are valid species. Other collaborative evidence supports this conclusion, including morphological and chromosomal differences between the two species. AKNOWLEDGEMENTS

There are a number of people who have been integral to my development as a scientist and to them I express much gratitude and respect. I owe much of this development to my advisor, Dr. Jinzhong Fu. He has allowed me to explore research ideas independently, while always providing critical and constructive input into their development. His encouragement of independent learning, collaborative research and discussion has shown me the exciting opportunities and challenges associated with scientific research. I am also thankful to Dr. Jim Bogart, who has provided me with valuable advice, opportunities and encouragement over the years and to Dr. Tom Nudds who has taught me the importance of formulating clear predictions and hypotheses while also providing insightful guidance on projects that I have developed. The development of this thesis benefitted greatly from their input.

I would also like to thank lab mates and colleagues throughout the department who have provided me with stimulating discussions, constructive feedback and advice over the years. I am particularly in debt to Dr. Rob McLaughlin, Dr. Elizabeth Boulding,

Dr. Andreas Heyland, Ke Bi, Kate Crosby, John Urquhart, Allan Edelsparre, Heather

Freamo, Kaitlyn Reid, Ashley Miller and Evan Timusk.

Finally, I would like to thank my family and friends for their positive feedback and interest in what I have been doing. In particular, Monika and Tony Noble (my parents), who have always encouraged my desire to get into herpetology; Yian Yian

Dam, Quinn Dam and Grant Wilkinson, who have all helped with projects and provided advice and Jonathan Choquette, who has been a great friend and voice of reasoning. TABLE OF CONTENTS

ACKNOWLEDGEMENTS i

TABLE OF CONTENTS ii

LIST OF TABLES iii

LIST OF FIGURES iv

INTRODUCTION 1 Species Criteria and Methods of Species Delineation 1 Assignment Tests for Species Delineation 4 Hypothesis and Predictions 7 MATERIALS AND METHODS 8 Sample Collection and Location 8 DNA Extractions and Microsatellite DNA Amplification 9 Microsatellite DNA Data Analysis 11 Mitochondrial DNA 15

RESULTS 16 Microsatellite DNA 16 Mitochondrial DNA 18 Congruence between Microsatellite and Mitochondrial DNA 19

DISCUSSION 19 Validity of Phrynocephalus putjatia 19 Assignment Tests as a Tool for Species Delineation 23 Inter-specific Hybridization between P. vlangalii and P.putjatia 25 High Genetic Diversity inP. vlangalii andP. putjatia 26

CONCLUSIONS 27

REFERENCES 28

TABLES AND FIGURES 38

APPENDIX I- Microsatellite DNA alleles for 192 individuals at eight loci: a) Diploid samples; b) Triploid sample 46 APPENDIX II- Log likelihood ratio tests of linkage disequilibrium in three populations across eight microsatellite DNA loci: a) Xinghai; b) Haiyan; c) Wulan. Significant deviations after Bonferroni correction (a=0.002) are bolded 56

u LIST OF TABLES

Table 1- Sampling localities and individual sampling numbers for specimens. Population labels correspond to those on Figure 1 38

Table 2- Percentage of genotypes [mitochondrial (mtDNA) and nuclear DNA] at each sampling site. EEE represents individuals with a nuclear genome composition and mtDNA haplotype from Phrynocephalus vlangalii (E clade). EEA represents individuals with a nuclear genomic composition from Phrynocephalus vlangalii (E clade) and a mtDNA haplotype from P. putjatia (A clade). "AD" refers to admixed nuclear genomes 39

in LIST OF FIGURES

Figure 1- Map of collecting sites (4-23) and populations (1-3) in Gonghe Basin, Qinghai province, China, where Phrynocephalus vlangalii (Clade E; dashed line) and a proposed species, Phrynocephalus putjatia (Clade A; dotted line), are sympatric (Jin et al, 2008). Population abbreviations are as follows: Wulan (WL), Haiyan (HY) and Xinghai (XH) 40

Figure 2- Boxplots of the 13 lowest likelihood values of 100 independent runs for K=l-4 from STUCTURE: a) admixture model; b) no admixture model 41

Figure 3- Individual assignment probabilities from STRUCTURE for: a) K=2 and b) K=3 . Each vertical line represents one individual and each colour represents a single cluster. The vertical height of each colour represents the probability of being associated with that colour. Population abbreviations are as follows: Wulan (WL), Haiyan (HY) and Xinghai (XH) 42

Figure 4- Spatial clustering output from TESS (vers. 2.1). Population coordinates were permuated using a standard deviation of 0.015 to more accurately depict individuals. Each dot represents a single individual. Red corresponds to Phrynocephalus vlangalii (Clade E) and green corresponds to Phrynocephalus putjatia (Clade A) in Figure 6. Grey polygon represents a lack of information 43

Figure 5- Interpolated admixture levels in each of the two clusters identified by TESS: a) Admixture in cluster 1 corresponding to Phrynocephalus vlangalii (Clade E); b) Admixture in cluster 2, corresponding to Phrynocephalus putjatia (Clade A) in Figure 6 44

Figure 6- Bayesian phylogenetic tree for 152 individuals using an 850bp fragment of the ND2 mitochondrial DNA gene. A total of 17 sequences of the ND2 fragment from Jin et al. (2008) were also included in this analysis and are labeled as p#. Only posterior probabilities associated with deep nodes are included 45

IV INTRODUCTION

Species Criteria and Methods of Species Delineation

Species form the foundation of many studies in ecology, evolution and conservation and yet the past 30 years have witnessed substantial disagreement regarding how they should be classified and defined (Mayden, 1997; Harrison, 1998; Hey, 2001;

Hey et al, 2003; Sites and Marshall, 2004; Sanders et al, 2006; de Queiroz, 2007). The debate surrounding what comprises a species and how we decide what species are, is not a new topic. Darwin, himself, did not clearly present the characteristics of a species in the

Origin of Species, due to the controversies present during his time (Darwin, 1859; Hey et al, 2003). Recognition of this problem quickly led to the proliferation of different

'species concepts', which were created to define the properties of a species (reviewed by

Mayden, 1997 and see de Queiroz, 2007 for a more recent review). Rather to the contrary, they stimulated intense debates among advocates of each concept, and therefore did little to resolve the species problem. Synthesizing similarities between different

'species concepts', Mayden (1997) and de Queiroz (1998) recognized that most 'species concepts' are not primary definitions describing the necessary qualities of a species, but rather secondary definitions or 'species criteria', which describe particular patterns and/or processes that take place during speciation (de Queiroz, 1998; Hey et al, 2003; de

Queiroz, 2007).

In a recent review, de Queiroz (2007) argued that all contemporary 'species concepts' equate species to separately evolving meta-population lineages or segments of lineages. This theme has been iterated in the past by numerous authors who agree that

1 species are spatio-temporally bounded entities that take part in evolutionary processes

(Hull, 1978; Mayden, 1997; de Queiroz, 1998; de Queiroz, 1999; Hey et al, 2003; Sites and Marshall, 2004) and are held together by gene flow (Mayr, 1942; Weins and Penkrot,

2002). Under this framework individual 'species concepts' act as operational definitions rather than primary definitions and are used as evidence in assessing lineage separation

(speciation) (Mayden, 1997; de Quieroz, 2007). For example, using the biological species criteria (i.e. species are reproductively isolated; Mayr, 1942), a species would be considered a meta-population lineage that shows reproductive isolation and genetic cohesion from other such lineages (de Queiroz, 2007). Contemporary species description requires the use of novel tools, designed to quantify these operational criteria. As such, there has been a shift in attention from conceptually defining species to more practical methods of species delineation (Weins and Penkrot, 2002; Hey et al, 2003; Sites and

Marshall, 2004; de Queiroz, 2007).

Species delineation can be broken up into what has been deemed 'Non-tree based' and 'Tree based' methods (Weins and Penkrot, 2002; Sites and Marshall, 2004). Tree based methods vary, but all make use of either morphological or molecular data to generate phylogenetic relationships or groupings of individuals that share a common evolutionary history or have shared attributes (Sites and Marshall, 2004). In contrast, non-tree based methods have traditionally relied on establishing large morphological discontinuities between populations or groups of populations, using statistical procedures such as numerical clustering or multivariate analyses (Coyne, 1994; Sites and Marshall,

2004). However, the high variability of morphological characters prompted recognition of their inadequacy in establishing species boundaries (Futuyma, 2009). With low gene

2 flow being a major mechanism promoting speciation (Dobzhansky, 1937; Mayr, 1942), molecular techniques were quickly adopted to quantify indirectly these measures and test species hypotheses.

The biological species criterion recognizes that reproductive isolation plays a pivotal role characterizing species boundaries (Dobzhansky, 1937; Mayr, 1942) and is widely accepted among evolutionary biologists (Futuyuma, 2009). Low gene flow between groups of populations is taken as strong evidence for species level differences

(Sites and Marshall, 2004; Futuyma, 2009). For sexually reproducing species, reproductive isolation has always been and is still the most practical species identification criterion and is a necessary pre-requisite for the establishment of independently evolving meta-population lineages (Futuyma, 2009). To assess reproductive isolation between populations, gene flow has traditionally been characterized using protein based allozyme markers to assess the level of genetic divergence between pre-defined populations

(Highton, 1989; 1990; Porter, 1990; Good and Wake, 1992). Fixed allelic differences at multiple loci were considered strong evidence for reproductive isolation. Although effective, this requires the sacrifice of whole organisms, which for many groups is no longer feasible. Moreover, allozyme markers are often too conservative to delineate closely related species. Important innovations in molecular genetics and greater computational power over the last few decades have allowed researchers to make use of model-based assignment tests and highly variable microsatellite DNA markers in population genetics (Sunnucks, 2000; Manel et al, 2005), providing powerful tools for species delineation.

3 Assignment Tests for Species Delineation

Assignment tests have become popular tools for assessing a multitude of questions of both applied and theoretical importance (Manel et al., 2005). These tests make use of Bayesian or likelihood statistics to cluster individuals based on linkage disequilibrium in a sample of individuals from multiple source populations (Manel et al,

2005). Linkage disequilibrium occurs when a population has a non-random association between genotypes at one locus and genotypes at another locus (Hartl and Clark, 1997;

Nordborg and Tavare, 2002). Natural selection, mutation, genetic drift, and population admixture (or mixing between two populations) are known to create linkage disequilibrium within populations (Hartl and Clark 1997). If populations are genetically differentiated, have a large population size and the genetic loci used are not under selection, we would expect them to be in linkage equilibrium. Therefore, in a mixture of individuals sampled from multiple source populations that are strongly genetically differentiated, assignment tests should be capable of accurately assigning individuals to their true source population by clustering them into groups that minimize linkage disequilibrium (Manel et al, 2005). Traditionally, these tests have been used in fisheries science to identify the origin of particular fish, however, they have since diversified to include the identification of illegal harvests (Manel et al., 2002), establishing the origins of bioinfestations (Bonizzoni et al, 2001), population differentiation assessment (Crosby et al, 2008), and determining the number of migrants from other populations (Berry et al, 2004). Other approaches treat individual alleles (i.e. alternate forms of a gene) as the unit of analysis and thus are capable of detecting individuals that arose from crosses between two distinct gene pools (Manel et al, 2005). These types of studies are applied

4 in hybrid zones where two parental species are known to coexist and reproduce. Such analyses allow one to identify Fl hybrids (first generation crosses between the two parental taxa) (Congiu etal, 2001; Randi and Lucchini, 2002).

The power of assignments tests lies not only in the rigorous statistical techniques used but also in the type of markers that are used in conjunction with such tests. Short tandem repeats (e.g. microsatellite DNA) have gained popularity among both ecologists and evolutionary biologists because they are co-dominant, selectively neutral, and are variable enough to pick up subtle biological processes. Furthermore, they are readily available, cost effective and amplify across closely related species (Sunnucks, 2000;

Selkoe and Toonen, 2006). The use of DNA based markers has also removed the need to kill whole organisms, as genomic DNA can be readily extracted from a minute amount of tissue (Sunnucks, 2000; Selkoe and Toonen, 2006).

These important features circumvent some of the problems with traditional population genetic analyses in assessing species boundaries and will provide the tools necessary for detecting even recently diverged taxa. Studies combining assignment tests and microsatellite DNA to delineate species will be very useful because: 1) they do not require large sample sizes; 2) there is no need to establish populations a priori (Mank and

Avise, 2004; Manel et al, 2005); 3) they treat individuals as the units of analysis, therefore increasing power to detect subtle biological patterns (Mank and Avise, 2004);

4) they allow for the detection of admixed or hybrid individuals (Mank and Avise, 2004;

Manel et al, 2005) and 5) they allow for the incorporation of geographical information to establish congruencies between tree based methods (Chen et al, 2007). To date few studies have directly used assignment tests for species delineation and none has formally

5 discussed their general potential as a tool for this purpose, despite their successes

(Maingon et al., 2003; Drummond and Hamilton, 2007). Assignment tests may provide

researchers with a powerful means of establishing species boundaries because they are

more easily applied than traditional population genetic analyses, remove biases and

address numerous species criteria, such as genetic cohesion, geographical association of

genetic clusters and reproductive isolation (Harrison, 1998). However, studies of this

nature are necessary to determine their general applicability.

Toad-headed lizards of the P. vlangalii complex provide an excellent model

system to test species boundaries using such approaches because they are widespread,

have numerous primers that cross amplify microsatellite DNA from closely related

species, have a well established mtDNA phylogeny and have received substantial

taxonomic attention. The high morphological variability within P. vlangalii has generated

intense debate over species designations providing an interesting and difficult case to test

species level boundaries. One specific case occurs between P. vlangalii and P. putjatia.

Bedriaga (1909) elevated populations of 'P. vlangalii' southeast of Qinghai Lake to P. putjatia based on fixed morphological differences. However, the status of P. putjatia has

been contentious; while Wang et al. (2002) considered it a valid species based on

morphological and chromosomal differences, most authors regard it as a synonym of P.

vlangalii (Zhao, 1997; Zhao, 1999; Barabarnov and Ananjeva, 2007). A recent study by

Jin et al (2008) revealed two deeply diverged mitochondrial DNA (mtDNA) clades east

(A clade) and west (B, C, D, E clade; herein referred to as clade E) of Qinghai Lake

(Figure 1), which appears to correspond to P. putjatia and P. vlangalii in this area,

respectively. Molecular clock estimates suggest that the initial divergence between these

6 two clades took place approximately 6.43-4.75 million years ago. South of Qinghai Lake these two clades overlap providing an excellent opportunity to test the proposed species hypothesis.

Hypothesis and Predictions

Using microsatellite DNA data and Bayesian assignment methods, I test the hypothesis that populations east and west of Qinghai Lake form one species, P. vlangalii, while P. putjatia is not a valid species. If P. putjatia is a synonym of P. vlangalii and lacks reproductive isolation and genetic cohesion from P. vlangalii then I would expect there to be one genotypic cluster (identified from the microsatellite DNA markers) with no geographic congruency with tree-based methods. This would suggest that significant gene flow occurs across the Gonghe Basin to keep the A and E lineages together.

Alternatively, if P. putjatia is reproductively isolated from P. vlangalii then two genotypic clusters are predicted east and west of Qinghai Lake (Figure 1), which should correspond to the previously identified mtDNA clades by Jin et al. (2008). Furthermore, individuals of each species should maintain their genetic identity where they live sympatrically. This would provide evidence to suggest that lineages are independently evolving and show contemporary genetic clustering and cohesion.

7 MATERIALS AND METHODS

Sample Collection and Location

One hundred and ninety-six individuals were collected in Qinghai province, of the

Peoples Republic of China in 2008. To test for the presence of null alleles (non-amplified

alleles due to mutations in the primer region) and the assumptions of Hardy-Weinberg

(HWE) and linkage equilibrium, 20 individuals from three populations (Xinghai, Wulan

and Haiyan) collected in 2007 were also used (Figure 1). Toe clippings were taken from

all samples collected in 2007. Specimens collected in 2008 were euthanized by an

intracardial injection of sodium barbital. Liver and heart muscle were excised and stored

in 95% ethanol and all voucher specimens were preserved in the Chengdu Institute of

Biology. A full list of samples and their collecting locations is provided in Table 1.

Three to sixteen individuals were collected at 21 different sites along two transects (east to west and north to south) where two previously identified mitochondrial

DNA (mtDNA) clades are known to persist in sympatry (Clades E and A; from Jin et al.

2008). These two clades provide a unique opportunity to test the utility of assignment tests in assessing the reproductive isolation between these groups because: 1) they are geographically close together, reducing the possibility that genetic divergence is due to isolation by distance and 2) they are found in a small flat area, reducing the possibility that genetic divergence is a result of a topologically complex landscape.

8 DNA Extractions and Microsatellite DNA Amplification

Whole genomic DNA was extracted from liver, heart muscle or toe clippings.

Minced tissue was incubated overnight at 55°C in a solution containing 75uL of 10%

SDS (Sodium Dodecoly Sulfate), 600uL of STE buffer and 17.5uL of proteinase kinase

(20mg/mL). Samples were washed twice with phenol:chloroform:isoamyl alcohol

(25:24:1) and once with chloroforrrdsoamyl alcohol (25:1). Genomic DNA was precipitated with 3M NaAc, washed with 95% ethanol twice and re-suspended in PCR grade water (Fisher Scientific) overnight.

I used microsatellite DNA primers developed by Urquhart et al. (2005) and Zhan and Fu (2009). These loci have been used successfully in previous genetic studies of P. vlangalii (Wang et al, 2009). Loci that had a close match between their expected and observed heterozygosity were chosen since this is indicative of a low occurrence of null alleles. This was necessary since null alleles can increase the degree of genetic differentiation between groups leading to inflated FST values (Carlsson, 2008). A total of eight microsatellite loci were used (Table 2), all of which have been shown to successfully amplify microsatellite DNA repeats across closely related species (Zhan and

Fu, 2009). PCR amplification was performed in lOuL reaction volumes containing 0.5-

0.7uL of extracted DNA, O.luL Taq polymerase (Takara; rtaq), 0.6uL of Mg2+(25 mM), l.OuL of 10X universal PCR Buffer (Takara), 0.2uL dNTP (10'mM of each dNTP; Roche

Diagnostics), and 0.25uL of each primer (10 pmol uL"1). All forward primers were labeled with tetrachloro-6-carboxy-fluoresine (TET). Reactions took place in a thermocycler (PTC-200, MJ Research) with an initial denaturation of 95°C for 5 minutes

9 followed by 30 cycles of 95°C for 45 seconds (sec), primer specific annealing temperature (Urquhart et al, 2005 and Zhan and Fu, 2009) for 30 sec and 72°C for 45 sec. Products were separated on 6% polyacrylamide gels and visualized with an FMBIO laser scanner (Hitachi). The base pair length of each allele was determined by running all samples with three marker individuals and a Genescan™-350 TAMRA size standard

(Applied Biosystems). Analyses were conducted using FMBIO analysis software

(Hitachi).

Microsatellite DNA genotyping is known to have a high occurrence of genotyping errors (Hoffman and Amos, 2005; Morrissey and Wilson, 2005; DeWoody et al. 2006), and as such the following precautions were taken to ensure as little error during the scoring process.

Firstly, scoring within and across gels was standardized. Since marker individuals were run multiple times, the most commonly computed value was taken as the true value for the base pair length of these individuals' alleles. Based on sequencing data, computed alleles were shown to be quite accurate; within two base pairs of the true repeat length.

Marker individuals, the TAMRA ladder and the slippage from incomplete PCR amplification were utilized to compute the base pair length of all other alleles. 'Slippage' or 'studder bands' are common mistakes made during PCR amplification, where the Taq polymerase causes a mutation in the number of repeat motifs during the replication process (Hoffman and Amos, 2005). This is particularly common among dinucleotide repeats. Although these can be problematic for the scoring process, they can be used as a two base pair ladder, providing alleles are scored consistently.

10 Secondly, 24 individuals (of known genotype) were re-run to ensure that their alleles were scored correctly. Four individuals containing an allele scored as being the same base pair length were chosen and run beside each other on the same polyacrylamide gel. I tested whether allele j from individual X is the same as allele j from individual Y, Z and W at locus n. Alleles placed beside each other can easily be detected as being the same or different. When possible, secondary alleles were also re-scored to ensure that their predicted value, based on their relationships with other alleles on the same gel, was consistent with previous scoring efforts. Any incorrectly scored alleles were changed accordingly.

Finally, MICRO-CHECKER (Ver. 1.0; Van Oosterhout et al, 2004) was used to test for the presence of scoring errors, large allele dropouts (where alleles of large repeat length compete with alleles with fewer repeats during the DNA replication process, causing the large allele to go undetected; Dewoody et al, 2006) and/or null alleles.

MICRO-CHECKER requires population level data in order to assess the aforementioned criteria and so populations Xinghai and Haiyan were used for these tests. The Wulan population was excluded because it had too much missing information.

Microsatellite DNA Data Analysis

Departures from HWE and LE were tested with ARLEQUIN (Ver. 3.1: Excoffier et al, 2005) using populations Xinghai, Haiyan and Wulan. For HWE an exact test with

10,000 burn-in runs and 1,000,000 post burn-in runs was used and for LE an EM algorithm with 10,000 permutations was used. The LE tests were conducted to identify

11 consistent associations between loci across populations, which would suggest that the loci are physically linked. Alpha was adjusted for multiple comparisons (a =0.002).

The Bayesian model-based clustering programs STRUCTURE (Ver. 2.2;

Pritchard et al, 2007) and TESS (Ver. 2.1; Chen et al, 2007) were employed to detect species level structure and assign individuals to groupings. These methods use multilocus genotypes and a predefined number of clusters (K) to generate groupings that minimize the deviation from Hardy-Weinberg and linkage equilibrium (Pritchard et al,

2000; Francois et al, 2006; Chen et al, 2007). Individuals are then assigned probabilistically to one or more clusters based on their multi-locus genotype.

STRUCTURE implements a non-spatial, Bayesian clustering method that uses a

Markov Chain Monte Carlo (MCMC) approach to explore the parameter space for the most likely estimates of the parameters Z(l) (individual i's population of origin), P (allele frequency in all populations) and a (admixture proportions for each individual; admixture model only) (Pritchard et al, 2000). The two models used in STRUCTURE differ slightly in their sampling criteria used to estimate the posterior probability distribution of

Z, P and a. The admixture model uses a Metropolis-Hasting sampling algorithm whereas the 'no admixture model' uses a Gibbs sampling algorithm (Pritchard et al, 2000).

The Metropolis-Hastings update works by selecting a random parameter combination [e.g. (Z1, P1, a1) in the multi-variate space]. From (Z1, P1, a1) a new combination is randomly drawn [e.g. (Z2, P2, a2)]. If the likelihood ratio of (Z2, P2, a2) to

(Z1, P1, a1) is larger than 1 it accepts (Z2, P2, a2) and repeats the process again starting from (Z2, P2, a2). In contrast, if the likelihood ratio of (Z2, P2, a2) to (Z1, P1, a) is smaller than 1 it is rejected and the original parameter combination is retained. In

12 contrast, the Gibbs sampling criteria always accepts the newly generated parameter combination while sampling the multivariate space. In both cases, samples are taken from repeated iterations until the most likely parameter set is found. An important feature of these update criteria is that they have no memory of past parameter combinations. Each parameter state depends only on the previous parameter state.

TESS implements a spatial Bayesian clustering method using a Hidden Markov

Random Field (HMRF) approach to model spatial dependency at the cluster membership level (Francois et al, 2006). The HMRF concept accounts for the idea that individuals close together are more likely to be genetically similar with each other than individuals further apart and introduces a spatial dependency parameter ty (Francois et al, 2006;

Chen et al, 2007). When i|> is set to zero the models used in TESS are virtually the same as STRUCTURE (Chen et al, 2007). Therefore, TESS was used to determine whether the spatial location of the two clusters corresponded to the geographic areas of the two mtDNA clades identified by Jin et al. (2008) and to assess the level of admixture between clusters.

STRUCTURE was used to identify the most likely number of clusters within my dataset (KMAX). Ascertaining the true value of K is not a trivial task and much discussion has been devoted to this issue (Pritchard et al, 2000; Evanno et al, 2005). Because Wang et al. (2009) identified strong population genetic structure in Phrynocephalus vlangalii of this area I restricted the range of K from 1-4 as I was only interested in detecting

'species' level divergences. A total of 100 independent runs with 10,000-50,000 burn-in iterations and 100,000 post burn-in iterations were conducted at each value of K using default settings. This was a sufficient number of iterations to guarantee convergence. One

13 hundred independent runs were completed to ensure that the multi-dimensional parameter space was sufficiently explored, removing the possibility of an MCMC chain getting stuck on one local optima. Both an 'admixture' and 'no admixture' model were run because the 'admixture model' has a tendency to under-estimate the true value of K

(Durand et al., 2009). Ten independent runs with the largest LnP(D) were averaged and plotted using the statistical software package R (Ver. 2.7.1; R Foundation for Statistical

Computing, 2008). This method is similar to Evanno's AK (Evanno et al, 2005; Durand, pers. comra.). Individual assignment probabilities, LnP(D) and convergence between runs were all used to assess the most likely value of K. In general, higher lnP(D) and large changes in the LnP(D) values between successive changes in the K parameter indicate a better fit to the data. Furthermore, a suitable K value should yield an individual assignment plot with a high probability of individual assignment. If greater than 80% of the individuals had a probability above 90% of being assigned to the identified clusters then the K parameter was considered to have sufficient power in explaining the data.

Admixed individuals were those with less than 90% individual assignment probability.

Using the identified number of clusters from STRUCTURE, 100 independent iterations were run in TESS with individual spatial information. An admixture model was run with 50,000 burn-in iterations and 100,000 post bum-in iterations with ip=0.6 and oc=1.0. This was sufficient to ensure convergence within a single iteration. I ran 10 pilot runs and found that varying values of i|>=0.6-2.0 and a=l.0-6.0 did not change the overall assignment probabilities. Therefore, default parameter settings were used. TESS requires unique individual coordinates to accurately depict spatial clustering. Since there were only site specific coordinates, each latitude and longitude were permutated slightly by a

14 standard deviation of 0.015 using the "Generate Spatial Coordinates" function in TESS.

This was necessary to more accurately depict individual clustering and its geographical association. CLUMPP 1.1 (Jakobsson and Rosenberg, 2007) was used to average the 15 lowest DIC runs and to produce the admixture (Q) matrix. The level of admixture in each cluster was displayed graphically using the statistical package R (Durand et al., 2009).

Mitochondrial DNA

DNA extractions from 152 of 196 (samples collected during 2008) individuals were selected, which had greater than 90% individual assignment probabilities or which showed evidence for an admixed genome (i.e. < 90% probability to one cluster). An

850bp ND2 fragment was amplified in order to determine whether clusters correspond to the two mtDNA lineages outlined by Jin et al. (2008). Reactions took place in 25uL volumes with luL of DNA, luL of ND2 forward and reverse primers (10 pmol uL"1)

(L4447 5'-aag cag ttg ggc cca tgc ccc aaa aac gg- 3' and H5622- 5'- tat ttt aat taa aat ate tga gtt gca-3'; Wang and Fu, 2004), 2uL dNTP (10 mM of each dNTP; Roche

Diagnostics), 1.5uL Mg2+(25 mM), 2.5uL 10X loading buffer (Takara), and 0.25uL Taq polymerase (rTaq: Takara). Thermal cycling was performed with an initial denaturation at 95 °C for 5 minutes followed by 30 cycles of 95°C for 30 sec, 50°C annealing for 30 sec, 72°C for 45 sec and a final extension of 72°C for 5 minutes. All PCR products were run on 1% agarose gels and purified using QIAquick PCR purification kits (Qiagen). The cleaned products were directly cycle sequenced in the forward direction using the L4447 primer. All DNA sequencing reactions were performed using BigDye terminator sequencing chemistry with an ABI 3730 sequencing machine (Applied Biosystems).

15 Sequences were visualized and corrected using Sequencher (Vers. 4.5; Genecode Corp)

and aligned using MacClade (Vers. 4.08; Maddison and Maddison 2003).

A phylogenetic tree was generated to determine whether my data recovered the

same clades as Jin et al. (2008) and to assess if the mtDNA lineages correspond to

genotypic clusters found using the nuclear DNA. A Bayesian inference approach using

MrBayes (Vers. 3.2; Ronquist and Huelsenbeck 2003) was used. Akaike's Information

Criterion (AIC) in MrModeltest (Vers. 2.1; Nylander, 2004) was used to determine the

best-fit model. I used a "flat" prior and four Markov Chain Monte Carlo (MCMC) chains with 10,000,000 generations to ensure convergence. Tracer (Vers. 1.4; Rambaut and

Drummond 2007) was used to plot likelihood values and determine whether the Markov chains had reached convergence. Trees were sampled every 500 generations and the last

5,000 trees were used to estimate the consensus tree and Bayesian posterior probabilities.

All prior trees were designated as "burn-in".

RESULTS

Microsatellite DNA

Allelic diversity ranged from 36 (PVMS 38) to 71 (Phry 75) alleles per locus with an average of 47.5 alleles/locus. No loci showed consistent linkage across all three populations after Bonferroni correction (P>0.002). Phry 79 and Phry 75 showed significant linkage in Haiyan and Wulan suggesting that they may be physically close to each other in the genome (Appendix II). In all three populations tested, Phry 75 and Phry

79 deviated significantly from HWE while PVMS 35 deviated from HWE in the Wulan

16 and Xinghai populations. PVMS 63 deviated from HWE only in the Haiyan population.

In all cases, deviations were a result of heterozygote deficiency. MICRO-CHECKER did

not detect evidence for large allele drop out or scoring errors, however, it did detect the

potential for null alleles at Phry 79, Phry 75 and PVMS 35. One individual (IOZ 6011

from site 4) appeared to have a triploid (3n) genome with 6/8 loci showing three alleles.

Average LnP(D) values from STRUCTURE increased by 3.3% from K=l to K=2,

0.19% from K=2 to K=3 and 0.87% from K=3 to K=4 (Figure 2). Individual assignment

probabilities declined as K increased from 2 to 3, with no significant proportion of an

individual's genome assigned distinctly to cluster 3 (Figure 3b). The large jump in

LnP(D) values between K=l and K=2 coupled with small changes between K=2 and K=3

suggest that K=2 is the most parsimonious. Due to the large fragment size and high

allelic diversity at locus Phry 75,1 removed it from the analysis to determine whether it

influenced assignment probabilities. Re-running the analysis without this locus caused no

change in assignment probabilities. Two distinct clusters were detected when K=2

(Figure 3a) with 93% (232/247) of the individuals assigned with >90% probability to one

of the clusters. The two clusters had an average FST=0.0524. Seven percent of individuals

showed evidence of an 'admixed' genome.

TESS aggregated the two clusters in the geographical areas associated with the

mtDNA lineages. The red cluster was west of Qinghai Lake and a green cluster was

mostly situated east of Qinghai Lake (Figure 4). It would have been desirable to obtain

more samples southeast of Qinghai lake to compare more thoroughly the extent to which the nuclear DNA showed geographical congruence with hypothesized distributions.

However, this was not possible due to the remoteness of the area and because the area

17 bordering the Yellow River is characterized by agriculture, providing un-suitable habitat for these lizards. Nonetheless, these clusters persisted in sympatry (Xinghai, Haiyan) and in closely situated sites with no geographical barriers (21 and 22). Furthermore, admixture levels within the clusters showed an east-west cline (Figure 5).

Mitochondrial DNA

A total of 152 sequences were successfully aligned along with 19 sequences from

Jin et al. (2008). My sequence data had an approximately 350bp overlap with Jin et al.

(2008)'s data. Overall, 76 unique haplotypes were identified including 72 ingroup individuals and four outgroup individuals {Phrynocephalus theobaldi, Phrynocephalus zetangensis, Phrynocephalus axillaris and Phrynocephalus versicolor).

A GTR+G model was selected to be the best-fit model based on Akaike's

Information Criteria. The Bayesian tree along with posterior probabilities is shown in

Figure 6. In general, my tree conformed to Jin et al. (2008) and identified clades A, B, C,

D and E. However, one new clade was identified within Phrynocephalus vlangalii, corresponding to individuals from site 21, southeast of Lake Qinghai (100% posterior probability). Clade E was primarily situated in the western part of the Gonghe Basin and west of Qinghai Lake, except for site 21. In contrast, clade A was situated east of

Qinghai Lake with three sites (sites 12, 13, 20; Table 2) found in the west of Gonghe

Basin near the Yellow River. One site, 16, was found within an area of the Gonghe Basin that was primarily occupied by clade E.

18 Congruence between Microsatellite and Mitochondrial DNA

There have been one or more instances of hybridization between clade E and clade A, both of which caused introgression of the mtDNA haplotypes into populations of each clade (Table 2). Population 13 was composed entirely of EEA [Nuclear genome from clade (E) and mitochondrial genome from clade (A)] individuals. Site 12 (3/8 individuals) and site 16 (3/8 individuals) also had individuals with EEA genotypes. In contrast, site 23 had two individuals with AAE genotypes. Haiyan had a mixture of pure

EEE and pure AAA individuals living in sympatry. Site 20 and 21 were dominated by pure

AAA (11 individuals) and EEE (11 individuals), respectfully. There were two individuals at site 20 with EEA and AAE genotypes. Overall, 81.79% of all individuals contained pure genomes (AAE and EEE), 14.8% of all individuals had pure nuclear genomes and the opposite mitochondrial genomes and 3.38% of all individuals contained an admixed nuclear genome with an A or E mtDNA haplotype.

DISCUSSION

Validity of Phrynocephalus putjatia

I reject the hypothesis that all populations are P. vlangalii because the mtDNA lineages identified by Jin et al. (2008) show evidence of reproductive isolation and genetic cohesion. Two lines of evidence support this conclusion: i) microsatellite DNA distinguishes two distinct genetic clusters corresponding to the geographical locations of the A and E mtDNA clades (Alternative hypothesis; Prediction 1) and ii) pure P.

19 vlangalii (EEE) and P. putjatia (AAA) live in sympatry (populations 20, 21 and Haiyan) while still maintaining their genetic integrity (Alternative hypothesis; Prediction 2).

The largest change in LnP(D) values occurred between K=l and K=2 suggesting that two very distinct gene pools exist in the dataset. Furthermore, individuals could be assigned to one of the two identified clusters with greater than 90% probability further supporting discrete differences at the genie level. To make the prediction of two distinct clusters more stringent a geographical constraint was imposed on the identified clusters indicating that they must correspond to the geographical distributions of P. vlangalii and

P. putjatia. Indeed these clusters exhibited geographical cohesiveness (Figure 4) with the levels of admixture (degree to which gene pools exchange) showing an east to west and west to east cline in cluster 1 and cluster 2, respectively (Figure 5). Jin et al. (2008) suggested that P. putjatia diverged from all other populations of P. vlangalii during the

Late Miocene when global temperatures decreased and the climate became more cool and arid. Populations of P. vlangalii isolated during this time would have diverged substantially from other populations.

Although two distinct clusters were identified, the FST value between these clusters was low (average FST = 0.0524). This finding is in conflict with Wang et al.

(2009), who found high pairwise FST (>0.11) between Haiyan and Xinghai /Wulan populations. Genetic divergence among species is known to be extremely variable in more slowly evolving allozyme markers, with species level divergence estimates ranging from 0.05 to over 0.77 (Avise, 2004). One reason for the lack of genetic differentiation found could be due to the high mutation rates among the loci used in this study. Loci with high mutation rates likely experience high levels of homoplasy (reversals) decreasing the

20 differentiation between groups and leading to low FST values (O'Reilly et al, 2004). This is a plausible explanation given the high allelic diversity (47.5 alleles/locus) I observed.

O'Reilly et al. (2004) showed that microsatellite DNA markers with high polymorphism caused a decline in FST estimates between Walleye Pollock (Theragra chalcogramma) populations. Simulation studies also suggest that increased mutation rates significantly decrease estimates of FST (Estoup et al, 2002; Kalinowski, 2002). Hedrick (1999) also analytically demonstrated that when the number of alleles is high, FST is necessarily low.

Other explanations such as historical introgression through hybridization and the small sample sizes of P. putjatia are also plausible. Greater sampling south of the Yellow river will be necessary to differentiate between these explanations.

The maintenance of pure P. vlangalii and P. putjatia at sites 20, 21 and populations Haiyan and Xinghai, suggest that reproductive isolation mechanisms exist preventing the breeding between these two species. Sites 20 and 21 are situated extremely close to each other with no barriers to dispersal. Furthermore, P. vlangalii and P. putjatia live sympatrically in Haiyan and Xinghai. If these were not two species one would predict most of the individuals examined to have admixed or hybrid genomes, suggesting random mating between the two genotypes. However, there is a lack of complete isolation between the groups (A total of 3.38% of all individuals showed evidence for an admixed genome, suggesting recent hybrids between P. vlangalii and P. putjatia).

Complex social behaviour, differing ecologies, genomic incompatibilities and hybrid selection may all be responsible for the maintenance of these species in the area

(Futuyma, 2009). The genus Phrynocephalus is well known for its complex social structure and tail-waving behaviours, which may play an important role in mate choice

21 and intrasexual communication. To date, the ecology and behaviour of these species remain poorly understood, and future work addressing the mating behaviour of these lizards will help elucidate potential isolation mechanisms between these species.

One individual, IOZ 6011 from site 4, had three alleles at six out of the eight loci screened, indicating that this individual likely contains three haploid genomes. The mtDNA of this individual is nested within the E mtDNA clade suggesting that the maternal parent was P. vlangalii. Since IOZ 6011 had three alleles it could not be included in the global assignment test because STRUCTURE is not capable of incorporating both diploid and triploid individuals in the same analysis. Therefore, the origin of the alleles at each locus for this individual is not known. Autopolyploidy or allopolyploidy could explain the existence of this triploid individual. Although autopolyploidy, through polyspermy or non-disjunction at meiosis I is plausible, these events are extremely rare in vertebrates. Reptiles are only known to exhibit physiological polyspermy, where two sperm enter the zygote with only one sperm nucleus fusing with the egg nucleus (Dale and Monroy, 1981; Elinson, 1986). Alternatively, allopolyploidy

(the union of two distinct sets of genomes) could also explain this individual (Comai,

2005; Otto, 2007). Under this scenario a diploid hybrid (between a female P. vlangalii and a male P. putjatia) would have to breed to either a P. vlangalii or P. putjatia male, elevating the ploidy of the offspring (Comai, 2005). This would require the Fl hybrid to be female and to be able to produce unreduced eggs.

Polyploidy is rare in animals (Mable, 2004). In reptiles, polyploidy has been reported to occur infrequently in the families Teiidae, Lacertidae, Xantusidae,

Gekkonidae, Agamidae, Chamaeleonidae and Typhlopidae (Vrijenhoek et al, 1989;

22 Avise, 2008). To date only two species in the family Agamidae (Leiolepis triploida

(Vrijenhoek et al, 1989; Otto and Whitton, 2000) and Amphibolurus nobbi nobbi (Otto and Whitton, 2000)) are known to have polyploid individuals. The occurrence of eggs that have not gone through non-disjunction in hybrids would suggest that genomic incompatibilities exist, facilitating improper segregation during meiosis. This would add more support for the validity of P. putjatia because it suggests that the genomes of these two species have diverged making them less compatible. Individuals with admixed genomes were detected suggesting that diploid hybrids do exist; however, greater sampling coverage and larger sample sizes will be necessary to discriminate between the aforementioned hypotheses.

The evidence presented in this study, together with evidence from morphology and chromosomes suggests that populations east of Qinghai Lake, corresponding to the A mtDNA clade, identified by Jin et al. (2008), should be recognized as P. putjatia.

Assignment Tests as a Tool for Species Delineation

Bayesian model-based assignment tests delineated two distinct clusters corresponding to P. vlangalii and P. putjatia. Drummond and Hamilton (2007) applied assignment tests in a sympatric zone between two varieties (i.e. sub-species) of Lupis microcarpus. They found strong evidence to suggest two distinct clusters corresponding to Lupis microcarpus horizontalis and Lupis microcarpus densiflorous, with the largest difference in likelihood values between K=l and K=2. They concluded that the two varieties should be recognized as separate species because they maintain their distinctiveness in sympatry and show low levels of admixture (only 2.1% of individuals

23 showed evidence of an admixed genome) (Drummond and Hamilton, 2007). Maingon et

al. (2003) also used assignment tests to discern between sandfly (Lutzomyia longipalpis)

populations exhibiting differences in pheromone composition. They too found strong

evidence that sandflies with different pheromone compounds (9MGB and cembrene) are

distinct species, with <10% of all individuals being incorrectly assigned.

The use of assignment tests as a tool for species delineation will likely only apply

to closely related species for which microsatellite DNA markers have been successfully

characterized and cross amplify (Sunnucks, 2000). Microsatellite DNA markers are

praised for their high sensitivity in detecting subtle population genetic structure

(Sunnucks, 2000) and thus will be important markers in conjunction with assignment

tests for recently diverged taxa. I do caution against the use of assignment tests without a priori reasons to suggest the existence of two species. Strong population genetic structure

within species could easily be mistaken as species level evidence between groups, and

therefore these tests should not be used for exploratory species identification.

Furthermore, complications arise when applying these tests to allopatric populations and

under these circumstances the use of many molecular and non-molecular tools should be

used concurrently to assess evidence of species level designations. Isolation by distance

and geography are known to cause differentiation between populations within species and

can confound interpretations of species with assignment tests. In addition, assignment test

models assume that populations are in Hardy-Weinberg and linkage equilibrium

(Pritchard et al, 2000; Francois et al, 2006; Chen et al, 2007). These conditions are not

satisfied by most empirical studies and greater effort should be put towards understanding how violations of such assumptions affect conclusions drawn from assignment test

24 analyses. Such deviations could be a result of null alleles (Carlsson, 2008), which may be more common when cross amplifying between closely related species (Selkoe and

Toonon, 2006). However, the ability of some methods to now incorporate linked loci

(Pritchard et al, 2007) and identify the presence of null alleles (Van Oosterhout et al,

2004) circumvents some of these issues.

Inter-specific Hybridization between P. vlangalii and P. putjatia

Incongruence between mtDNA and nuclear DNA suggests at least one or more hybridization events between P. vlangalii and P. putjatia in the Gonghe basin and east of

Qinghai Lake. Furthermore, hybridization appears to have been in both directions with introgression of the A (site 12,13 and 16) and E (23) mtDNA haplotypes into populations of each species.

Sites 12, 13 and 16 contained mtDNA haplotypes from P. putjatia yet individual assignment probabilities suggest that the nuclear genome was entirely made up of P. vlangalii. Jin et al. (2008) provides evidence that P. vlangalii from the Qiadam Basin exhibited a rapid eastward range expansion as the Yellow River drained Gonghe Lake, approximately 0.15 million years ago. Under this scenario P. vlangalii would have come into secondary contact with P. putjatia populations isolated during this time.

Hybridization would have resulted in introgression of mtDNA haplotypes from both of these species. Continual expansion of the E clade through the Gonghe basin would have provided consistent, directional gene flow from clade E, which evidently would have homogenized the nuclear genome of these hybrid populations (Jin et al. 2008). Indeed, historical hybridization and introgression of mtDNA haplotypes, through range

25 expansions, has been reported in a number of studies. Hybridization between two collard

lizard species in the Rio Grande Valley, Texas has resulted in the replacement of

Crotophytus reticulatus mtDNA genomes with that of C. collaris over two-thirds of its

range (McGuire et al, 2007). Melo-Ferreira et al, (2008) also report extensive

introgression in the Iberian Hares from Spain with 32% of Lepis granatensis containing

an mtDNA lineage from Lepis timidus.

High Genetic Diversity in P. vlangalii and P. putjatia

The observed allelic diversity in this study (47.5 alleles/locus) was greater than

that found by Wang et al. (2009), who found an average of 32.5 alleles per locus. The

greater average number of alleles per locus in this study is likely due to the different loci

used and differences in microsatellite scoring between studies. Wang et al. (2009) found

that populations from Xinghai and Wulan had the greatest number of alleles on average than any other populations (23.8 and 16.5, respectively). Since this study had much more

extensive sampling in this area it is not surprising that a greater diversity of alleles was

found. Wang et al. (2009) speculated that the high genetic diversity in this group is likely

due to the age of the lineage, the high population density of these lizards, isolation with migration and the lack of bottleneck events.

One new clade was discovered, composed of individuals IOZ 6148 and 6188-191

(Figure 6) that was not identified by Jin et al. (2008) in their phylogeographic study. The posterior probabilities of this clade strongly support its positioning in the phylogeny and it was highly diverged from other clades. The complex geological and environmental changes in this region suggest that new clades will be common, although greater

26 sampling coverage and a more complete understanding of the geological history of this area are necessary to understand these patterns.

CONCLUSIONS

The results from the assignment test and phylogenetic analyses support the validity of P. putjatia as being a distinct species. The two clusters corresponded to the hypothesized distributions of P. vlangalii and P. putjatia and were reciprocally monophyletic lineages. However, there was evidence for both contemporary and historical hybridization between P. vlangalii and P. putjatia.

Species delineation has always been an important scientific endeavor. The characterization of independently evolving lineages is necessary for addressing fundamental theories in ecology and evolution, while also guiding our efforts in conserving biological diversity. An evidence-based approach to species discovery requires the recruitment of a wide array of tools that allow a researcher to effectively discriminate between competing species hypotheses. Assignment methods provide a novel tool to allow researchers to test between such species hypotheses and will be beneficial when working with sister species.

27 REFERENCES

Avise JC (2004) The history of interest in genetic variation. In: Molecular markers,

natural history and evolution (2nd edition), Sinauer Associates Inc. Massachusents,

pp 24-54.

Avise JC (2008) Reproduction by the Chaste: Parthenogenesis. In: Clonality: The

genetics, Ecology, and Evolution of Sexual Abstinence in Vertebrate Animals,

Oxford University Press, pp 35-79.

Barabanov AV and Ananjeva NB (2007) Catalogue of the available scientific species-

group names for lizards of the genus Phrynocephalus Kaup, 1825 (Reptilia, Sauria,

Agamidae). Zootaxa, 1399, 1-56.

Bedriaga JV (1909) Amphibian und reptilien. In: Wissenschaftliche Resultate der von

N.M. Prezewalski nach Central-Asien unternomenen Reisen (Zoologischer Theil).

Band 3, Abt. 1 Annual reports of the Zoological Museum of the Emperor Academy

of Sciences (St. Petersburg).

Berry O, Tocher MD and Sarre SD (2004) Can assignment tests measure dispersal?

Molecular Ecology, 13, 551-561.

Bonizzoni M, Zheng L, Guglielmino CR, Haymer DS and Gasperi G (2001)

Microsatellite analysis of medfly bioinfestations. Molecular Ecology, 10, 2515-

2524.

Carlsson J (2008) Effects of microsatellite null alleles on assignment testing. Heredity,

99, 616-623.

28 Chen C, Durand E, Forbes F and Francois O (2007) Bayesian clustering algorithms

ascertaining spatial population genetic structure: a new computer program and a

comparison study. Molecular Ecology Notes, 7, 747-756.

Crosby MKA, Licht LE and Fu J (2008) The effect of habitat modification on fine scale

population structure of wood frogs (Rana sylvatica). Conservation Genetics, in

press.

Coyne JA (1994) Ernst Mayr and the origin of species. Evolution, 48, 19-30.

Comai L (2005) The advantages and disadvantages of being polyploidy. Nature Reviews

Genetics, 6, 836-846.

Congiu L, Dupanloup I, Patarnello T, Fontana F, Rossi R, Arlati G and Zane L (2001)

Identification of interspecific hybrids by amplified fragment length polymorphism:

the case of sturgeon. Molecular Ecology, 10, 2355-2359.

Dale B and Monroy A (1981) How is polyspermy prevented? Gamete Research, 4, 151-

169.

Darwin CR (1859) Mutual affinities of organic beings: Morphology: Embyrology:

Rudimentary organs. In: The Origin of Species, (Ad bridged and introduced by RE

Leakey), pp 206-216. de Queiroz K (1998) The general lineage concept of species, species criteria,

and the process of speciation: A conceptual unification and terminological

recommendations. In: Endless forms: Species and speciation (Howard DJ and

Berlocher SH, eds.), Oxford University Press, New York, pp 57-75.

29 de Queiroz K (1999) The general lineage concept of species and the defining properties

of the species category. In: Species: New interdisciplinary essays (Wilson RA, ed.),

MIT Press, Cambridge, Massachusetts, pp 49-89. de Queiroz K (2007) Species concepts and species delineation. Systematic Biology, 56,

879-886.

DeWoody J, Nason JD and Hipkins VD (2006) Mitigating scoring errors in microsatellite

data from wild populations. Molecular Ecology Notes, 6, 951-957.

Dobzhansky TD (1937) Genetics and the Origin of Species. Columbia

University Press, New York, pp 303-321.

Drummond CS and Hamilton MB (2007) Hierarchical components of genetic variation at

a species boundary: population structure in two sympatic varieties of Lupinus

microcarpus (Leguminosae). Molecular Ecology, 16, 753-769.

Durand E, Chen C and Francois O (2009) Tess version 2.1- reference manual.

Elinson RP (1986) Fertilization in Amphibians: The ancestry of the block to polyspermy.

International Review of Cytology, 101, 59-100.

Estoup A, Jarne P and Cornuet J-M (2002) Homoplasy and mutation model at

microsatellite loci and their consequences for population genetic analysis.

Molecular Ecology, 11, 1591-1604.

Evanno G, Regnaut S and Goudet J (2005) Detecting the number of clusters of

individuals using the software STRUCTURE: a simulation study. Molecular

Ecology, 14, 2611- 2620.

30 Excoffier L, Laval G and Schneider S (2005) Arlequin ver. 3.0: An integrated software

package for population genetics data analysis. Evolutionary Bioinformatics Online,

1, 47-50.

Francois O, Ancelet S and Guillot G (2006) Bayesian clustering using Hidden Markov

Random Fields in spatial population genetics. Genetics, 174, 805-816.

Futuyma DJ (2009) Species. In: Evolution: Second Edition, Sinauer Associates, Inc, pp

445-469.

Good DA and Wake DB (1992) Geographic variation and speciation in the torrent

salamanders of the genus Rhyacotriton (Caudata: Rhyacotritonidae). University

California Publications in Zoology, 126, 1-91.

Harrison RG (1998) Linking evolutionary pattern and process: The relevance of species

concepts for the study of speciation. In: Endless forms: Species and speciation

(Howard DJ and Berlocher SH, eds.), Oxford University Press, New York, pp 19-

31.

Hartl DL and Clark AG (1997) Organization of genetic variation: Linkage and Linkage

disequilibrium. In: Principles of population genetics, Sinauer Associates, Inc,

Sunderland Massachusetts, pp 71-109.

Hedrick PW (1999) Perspective: Highly variable loci and their interpretation in evolution

and conservation. Evolution, 53, 313-318.

Hey J (2001) The mind of the species problem. Trends in Ecology and Evolution, 16,

326-329.

31 Hey J, Waples RS, Arnold ML, Butlin RK and Harrison RG (2003) Understanding and

confronting species uncertainty in biology and conservation. Trends in Ecology and

Evolution, 18, 597-603.

Highton R (1989) Biochemical evolution in the slimy salamanders of the Plethodon

glutinosus complex in the eastern United States. Part I. Geographic protein

variation. 111. Biological Monographs, 57, 1-78.

Highton R (1990) Taxonomic treatment of genetically differentiated populations.

Herpetologica, 46, 114-21.

Hoffman JI and Amos W (2005) Microsatellite genotyping errors: detection approaches,

common sources and consequences for paternal exclusion. Molecular Ecology, 14,

599-612.

Hull DL (1978) A matter of individuality. Philosophy of Science, 45, 335-360.

Jakobsson M and Rosenberg NA (2007) CLUMPP: a cluster matching and permutation

program for dealing with label switching and multimodality in analysis of

population structure. Bioinformatics, 23, 1801-1806.

Jin Y-T, Brown RP and Liu N-F (2008) Cladogenesis and phylogeography of the lizard

Phrynocephalus vlangalii (Agamidae) on the Tibetan plateau. Molecular Ecology,

17, 1971-1982.

Kalinowski ST (2002) Evolutionary and statistical properties of three genetic distances.

Molecular Ecology, 11, 1263-1273.

Mable BK (2004) 'Why polyploidy is rarer in animals than in plants': myths and

mechanisms. Biological Journal of the Linnean Society, 82, 453-466.

Maddison WP and Maddison DR (2003) MacClade: Analysis of Phylogeny and

32 Character Evolution. Sinauer Associates, Sunderland, Massachusettes.

Maingon RDC, Ward RD, Hamilton JGC, Noyes HA, Souza N, Kemp SJ and Watts PC

(2003) Genetic identification of two sibling species of Lutzomyia longipalpis

(Diptera: Psychodidae) that produce distinct male sex pheromones in Sobral, Ceara

State, Brazil. Molecular Ecology, 12, 1879-1894.

Manel S, Gaggiotti OE and Waples RS (2005) Assignment methods: Matching biological

questions with appropriate techniques. Trends in Ecology and Evolution, 20, 136-

142.

Manel S, Berthier P and Luikart G (2002) Detecting wildlife poaching: Identifying the

origin of individuals with bayesian assignment tests and multilocus genotypes.

Conservation Biology, 16, 650-659.

Mank JE and Avise JC (2004) Individual organisms as units of analysis: Bayesian

clustering alternatives in population genetics. Genetical Research, 84, 135-143.

Mayden RL (1997) A hierarchy of species concepts: The denouement in the saga of the

species problem. In: Species: The units of biodiversity (Claridge MF, Dawah HA

and Wilson MR, eds.) Chapman and Hall, London, pp 381-424.

Mayr E (1942) Systematics and the Origin of Species. New York: Columbia Univ. Press,

pp 102-122.

McGuire JA, Linkem CW, Koo MS, Hutchison DW, Lappin AC, Orange DI, Lemos-

Espinal J, Riddle BR and Jaeger JR (2007) Mitochondrial introgression and

incomplete lineage sorting through space and time: Phylogenetics of crotaphytid

lizards. Evolution, 61, 2879-2897.

33 Melo-Ferreira J, Boursot P, Suchentrunk F, Ferrand N, and Alves PC (2005) Invasion

from the cold past: extensive introgression of mountain hare (Lepus timidus)

mitochondrial DNA into three other hare species in northern Iberia. Molecular

Ecology, 14, 2459-2464.

Morrissey MB and Wilson AJ (2005) The potential costs of accounting for genotypic

errors in molecular parentage analyses. Molecular Ecology, 14, 4111-4121.

Nordberg M and Travare S (2002) Linkage disequilibrium: What history has to tell us.

Trends in Genetics, 18, 83-90.

Nylander JAA (2004) MrModeltest version 2.1. Computer program distributed by the

author. Uppsala University, Uppsala.

O'Reilly PT, Canino MP, Bailey KM and Bentzen P (2004) Inverse relationship between

FST and microsatellite polymorphism in the marine fish, walleye pollock (Theragra

chalcogramma): implications for resolving weak population structure. Molecular

Ecology, 13, 1799-1814.

Otto SP and Whitton J (2000) Polyploid incidence and evolution. Annual Review of

Genetics, 34, 401-437.

Otto SP (2007) The evolutionary consequences of polyploidy. Cell, 131, 452-462.

Porter AH (1990) Testing nominal species boundaries using gene flow statistics:

taxonomy of two hybridizing admiral butterflies (Limenitis: Nymphalidae).

Systematic Zoology, 39, 131-47.

Pritchard JK, Wen X and Falush D (2007) Structure, version 2.2.

http://pritch.bsd.uchicago.edu/software.

34 Pritchard JK, Stephans M and Donnelly P (2000) Inference of population structure using

multilocus genotype data. Genetics, 155, 945-959.

Randi E and Lucchini V (2002) Detecting rare introgression of domestic dog genes into

wild wolf {Canis lupus) populations by Bayesian admixture analysis of

microsatellite variation. Conservation Genetics, 3, 31-45.

Rambaut A and Drummond AJ (2007) Tracer: MCMC Trace Analysis Package (version

1.4). Computer programs distributed by the authors.

Ronquist F and Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogenetic inference

under mixed models. Bioinformatics, 19, 1572-1574.

R Development Core Team (2008) R: A language and environment for statistical

computing. R Foundation for Statistical Computing, Vienna Austria ((ISBN) 3-

900051-07-0), http://www.R-project.org.

Sanders KL, Malhotra A and Thorpe RS (2006) Combining molecular, morphological

and ecological data to infer species boundaries in a cryptic tropical pitviper.

Biological Journal of the Linnean Society, 87, 343-364.

Selkoe KA and Toonen RJ (2006) Microsatellites for ecologists: a practical guide to

using and evaluating microsatellite markers. Ecology Letters, 9, 615-629.

Sites JW Jr. and Marshall JC (2004) Operational criteria for delimiting species. Annual

Review of Ecology, Evolution and Systematics, 35,199-227.

Sunnucks P (2000) Efficient genetic markers for population biology. Trends in Ecology

and Evolution, 15,199-203.

35 Urquhart J, Ke B, Gozdzik A, and Fu J (2005) Isolation and characterization of

microsatellite DNA loci in the toad-headed lizards, Phrynocephalus przewalskii

complex. Molecular Ecology Notes, 5, 928-930.

Van Oosterhout C, Hutchinson WF, Wills DPM and Shipley P (2004) Micro-checker:

Software for identifying and correcting genotyping errors in microsatellite data.

Molecular Ecology Notes, 4, 535-538.

Vrijenhoek RC, Dawley RM, Cole CJ and Bogart JP (1989) A list of the known

unisexual vertebrates. In: Evolution and Ecology of Unisexual Vertebrates,

(Dawley RM and Bogart JP, eds.) The University of the State of New York

Museum Bulletin 466, pp 19-23.

Wang YZ, Zeng XM, Fang ZL, Wu GF, Liu ZJ, Papenfuss TJ and Macey JR (2002) A

valid species of the genus Phrynocephalus: P. putjatia and a discussion on

taxonomy of Phrynocephalus hongyuanensis (Sauria: Agamidae). Acta

Zootaxonomica Sinica, 27, 372-383.

Wang Y and Fu J (2004) Cladogenesis and vicariance patterns in the Toad-Headed

Lizard Phrynocephalus versicolor species complex. Copeia, 2004, 199-206.

Wang Y, Zhan A and Fu J (2009) Testing historical phylogeographic inferences with

contemporary gene flow data: Population genetic structure of the Qinghai toad-

headed lizard. Journal of Zoology, in press.

Weins JJ and Penkrot TA (2002) Delimiting species using DNA and morphological

variation and discordant species limits in spiny lizards (Sceloporus). Systematic

Biology, 51, 69-91.

Zhan A and Fu J (2009) Microsatellite DNA markers for three toad-headed lizard

36 species (Phrynocephalus vlangalii, P. przewalskii and P. guttatus). Molecular

Ecology Resources, in press.

Zhao KT (1997) Introduction and key to Chinese species of toad-headed agamids (genus

Phrynocephalus). Cultum Herpetolica Sinica, 6, 86-91.

Zhao KT (1999) Phrynocephalus kaup. In: Fauna Sinica, Reptilia, Vol 2 (Zhao EM, Zhao

KT and Zhou KY, eds.), pp 153-193. Science Press, Beijin.

37 TABLES AND FIGURES

Table 1- Sampling localities and individual sampling numbers for specimens. Population

labels correspond to those on Figure 1.

Population Catalogue Latitude Longitude

Numbers

Xinghai-XH (1) 10(#) N35°80'006 E 099°87'933 Haiyan-HY (2) 2(#) N37°01'030 E 100°63'975 Wulan-WL (3) 7(#) N 36°65'243 E 099°37'298 4 IOZ 06008-6015 N 36°30'566 E 100°54'564 5 IOZ 06016-6022 N 36°29'575 E 100°41'498 6 IOZ 06023-6030 N 36°26'527 E 100°33'526 7 IOZ 06033-6041 N 36°26'327 E 100°22'209 8 IOZ 06042-6046 N 36°20'682 E 100°53'485 9 IOZ 06048-6061 N 35°96'003 E 100°13'024 10 IOZ 06062-6071 N 36°02'903 E 100°16'232 11 IOZ 06072-6085 N 36°08'624 E 100°06'871 12 IOZ 06086-6095 N36°03'330 E 100°30'091 13 IOZ 06096-6104 N36°10'402 E 100°40'946 14 IOZ 06105-6116 N 36°34'046 E 100°14'559 15 IOZ 06117-6132 N36°36'355 E 100°00'814 16 IOZ 06135-6145 N36°32'911 E 099°83'822 17 IOZ 06148-6162 N36°36'917 E099°64'797 18 IOZ 06163-6168 N36°31'429 E099°70'199 19 IOZ 06169-6171 N 36°35'032 E 100°28'450 20 IOZ 06173-6187 N36°55'759 E 100°73'343 21 IOZ 06188-6192 N 36°65'764 E 100°80'575 22 IOZ 06194-6200 N36°15'355 E 101°57'399 23 IOZ 06201-6207 N 36°06'605 E 101°45'580

38 Table 2- Percentage of genotypes [mitochondrial (mtDNA) and nuclear DNA] at each

sampling site. EEE represents individuals with a nuclear genome composition and mtDNA haplotype from Phrynocephalus vlangalii (E clade). EEA represents individuals

with a nuclear genomic composition from Phrynocephalus vlangalii (E clade) and a mtDNA haplotype from P. putjatia (A clade). "AD" refers to admixed nuclear genomes.

Population EEE EEA AAA AAE ADE ADA Samp Size Xinghai-•XH 0) 100% - - - - - 4 Haiyan - HY (2) 20% - 60% - - 20% 5 Wulan-WL (3) 100% - - - - - 3 4 100% - - - - - 5 5 100% - - - - - 6 6 83% - - - 16% - 6 7 100% - - - - - 8 8 66% - - - 33% - 3 9 100% - - - - - 9 10 100% - - - - - 9 11 100% - - - - - 10 12 62.5% 37.5% - - - - 8 13 - 100% - - - - 9 14 100% - - - - - 6 15 100% - - - - - 4 16 62.5% 37.5% - - - - 8 17 100% - - - - - 10 18 100% - - - - - 4 19 100% - - - - - 3 20 - 7.69% 84.5% 7.69% - - 13 21 100% - - - - - 11 22 - - 83.3% - - 16.6% 6 23 . - 50% 50% - - 4

39 Figure 1- Map of collecting sites (4-23) and populations (1-3) in Gonghe Basin, Qinghai

province, China, where Phrynocephalus vlangalii (Clade E; dashed line) and a

proposed species, Phrynocephalus putjatia (Clade A; dotted line), are sympatric

(Jin et ah, 2008). Population abbreviations are as follows: Wulan (WL), Haiyan

(HY) and Xinghai (XH).

40 3

•s aj 1

1 1 1 I 1 1 oo !-£(.- oozet- oom- oom- oogei.- oo9ei.-

(a)dui -O

OOLCI.- oozet- ooeei.- OOKI-- oosei- 009EI-- (a)dui 1? !U S3 00

CoO -t--; C3 x'3 3 O es e >> t> tH r^ X (3 cS O <—j *M KS cx ^-^ o u ffi l-H X > en £ •i-H X ctf o Dc X ca CO 60 W <

re p T3 (3 ,—, _so X "o o 5P1 1cd XoI K= 2 aiya n o f e a ,— +^ erfs X ffi top ^^^ o '53 h-l <+-< X ca l s n( W p '-g J2 H £ lo STR U er . Th e -4-» S CO o J3 1 * "o c<3 1/)

sin g a X C3 cd o X xn O •f-i _J l-i 'is g lent p pres e fi abbr c > ui-H X H G

•I—< '-4-» ou r t» *—( z c3 o X ^^ o 3OH c3 X O o CH o o O o * i CO our . T- div i CD o CD CD T—1 CI ande ; l-H o ^^ o f»i 3 c3 0) X S-H 3 -*-> S '> X •3

2" E wi t _o '•*-» a >* Gr e dev : VD c3 W *co cs o Ec t>0 o

si n

(D la d P y, p •2 S3 a> p -a^ PH -5 ••to wer e ndiv i us pu o <-*«a* -t-» -s; S3 "bb a 5r g W •3 KS o «o c o to C o p -« 1/3 a.

io n a> o +-» l-H ~t-t ca EX w en P f-H T2 ft -*-> P o o O AH T3

^^ XI es p »—i (N co Ea c o »j P (vers , gre e dua l GT3

GO iv i J3 m T3 cd •&i H ,—s _•4-gt S o rm a fro ade E epi nf o T3 3 U, <4-H a, j>% o -4-» M P **** o o l-H ^ bo p _g o cd 't-i o 1 w o CO —> * +-» clus t alus resel l ,__< mor e -c o- •S o -i c3 io o p D. f-H o o GO o 5 bD •4 o ol y

ft 02 *•

^ 0.0 a)

1.0

h 0 8

* * X * 0.4 K 0.2

Figure 5- Interpolated admixture levels in each of the two clusters identified by TESS: a)

Admixture in cluster 1 corresponding to Phrynocephalus vlangalii (Clade E); b)

Admixture in cluster 2, corresponding to Phrynocephalus putjatia (Clade A) in

Figure 6.

44 GCHYfT) GCHY(18)+GCHY(27)+GCHY(42) GCHY(f2) » 0.005 substitutions/site -GCHY07) P03 p04 p05 p57 6088+6096+6097+6099+6101+6104 6093+6094 P. putjatia 6100+6102+6103 6143+6194-186+6201+6202+6204 6142 6197+6199+6200 1-6141+6173-175+6177-179+6180-186 ,XH(4)«XH(19) JXH<6) fe|XH(21) >XH<43) WL|1S)+6075 WL<17)+WL(37i+6145+6155-158 WL{16)+WL<36)+6139+6154 61S7 6144+6148+6149+6151+6203+6206 6164 8165 ,6117 16166 6118 WL(39)+WL(40) P48+6152 -6008+60O9+6011»6012-6014-6016-6013+602a+5021»6024+6029+6043 6153 6038+6048+6056+6057+6067+6068+6071+6073+6077+6082+6083*6138 6022 ,6027+6035+6105+6113+6116 6074+6084+6066

6062+6169 6025*6030+6038+6040 6124 6111 6112 6041+6107+6170 6063 6053-6055+6059+6091+6092 6095 6086 6049+6060+6064-6066 6087 6069 P. vlangalii 6140 6166 6037 6034+6171 6044+6046 6119 p49 ,p28 1-P42 GCHY{25) p17 -p21 Lp23 ,6188+6190 C+D+New 16148+6189+6191 P12 p16 p62 B _j~ P. f/ieofiawi *—P. zetangensis —- Plvynoceplraltis axSlaris Outgroup • Phtynocepluilus iwsfcotor

Figure 6- Bayesian phylogenetic tree for 152 individuals using a 850bp fragment of the

ND2 mitochondrial DNA gene. A total of 17 sequences of the ND2 fragment from

Jin et al. (2008) were also included in this analysis and are labeled as p#. Only

posterior probabilities associated with deep nodes are included.

45 eor^t^i/oeocot—imwor-ieocoi—i>neot^t^- >—i co ro m Os coeo^-cNcocsicNeo'^i-^HeofNsooscNr--^- oo ro co O CN m eoeoeoeoeocoeoeoco''d-coeoeocScocNco co co co co co r~ i. .e eo eo Os eo m eo >—i v> UO CM ntnNNNNNN r—1 co 33 5 33 3 30 7 30 7 33 3 32 3 31 9 27 7 32 3 27 7 33 5 corococococococo 36 3 co CO

Os m co >—< m co Os t~~ Os vOsOOsSOt^OOSOOO i o

o\ 26 1 27 5 30 7 31 7 26 3 26 3 26 3 26 7 28 3 26 3 28 3 26 3 co t- u J3 i—i Os i—iosos-^osmt^r~- i so SOSO-^SOSOOOTJ-OO 25 3 30 7 29 7

a- 25 7 24 9 28 9 28 3 26 3 26 7 C-4 NMNMNNMN wo «/-> wo wo os os f- i—i wocoOsOs^wooswowoost"-t"~-r--co m O f~- t^ O I> ' 0000l^r-—it~-r--c--i-~t--^Hi—< OH t~- co co •*}• ro ' oo(^CNOeococoOsOsr--eoeoeoC--ooos N M N (S M NNNwNNMMNINNNfNNNN CO S N*Ot(0\0WMNaiÔON*M0000NOOV0 w 00 Os 00 Os r-< Os O ' t~- O 'SOOOOOOsOsOOOst--OOOsOsOO o t*i J en ^ 2 Z ^ •5j-00CS«NTfS00sO^t0sTj-00rf(S|S0(NO00CNOO00 o & soôcôsosos • so r- isosor~ooTtoooosoooosr-oo •aM* .—iuowot^wol^coOswowowoosOsOswoC"-t^>——i i—i Os C- Os wo t—it^^Hpoost^r-i-^r-os co Os PN TJ- ^- wo wo t-» wo TT « oo oo h io ' so •*!- wo co • oo f- wo N (S N N (S N M (N r- 'wowowOr-s m m n ^ • co ro ro co 'mTfroi—imr- ON H f~ OS WO CO 00 '^-^••^'^•'^•fSO-'j-mO'-H l-H oeN^cocNOO-* wo SO in CZ2 § Os t~~ wo wo en en >—iOsr^wot-~coOsOswot^wooswor~-osr-- > Tf rf Tt so m (S 0-

—i (S ro co •^ •"* ,TT wo wo so t-» t~~ r~- t~~ oo os os O 04 wo •- ea CZJ ooooooooooooooooooo cs Q N N N t--ONC---^t^t^t~-r-Ht^^H^Hi/oco^H^H'—ieo^HONr-/or-->/omco ON f*i^t>—ivOi-^csosoinvorovoOf^vooo^i—i ' ON -r-< ^o en oo co t> eo 36 5 encn'^ten'^-cncncnenenenenenencnencn•^• m * co co co eo ro

co <—> co t~- eo r-V) t> ON c~i f- •—i eo i—i t~~ i—i co C~~ ON r-n ON C"- C~~ C"- en en O ^H O oo o oo o o cs (S o en O CO O 00 t~~ CO ' NO oo r- t~- en en vo 33 5 36 5 co co en (N co tN eo cn c~i co co eo co en en fN CN en en en en en en en en

IH t^ ci h o\ n t~- wo wo ON wo r-> wo wo wo ON en en C~- ON1 I—I ON ON I—I i—i t~- o oo o o 'oo H oo Oi oo H H N ^H ^H ON O O ON O ON ON *—< (—i ON 32 5 27 3 en CN eo eo (S eo

PH H o\ vi o> n en wo ON wo in wo t*> wo wo ON en en t^- ON1 r- ON r- o ^H t> o ON o oo « oo oo oo oo h H a o ri ^H ON O O •* "t OO OO 00 H Ol

29 3 27 3 eo

~~o\mm^rtHH(t)i>>a\0(iflO\0\H(fio\îflmHrtHnvir»i •^ôoNtôooNoooot^tôoN ' vor~-t~~ONt~-Not~-ovoor~ÔNo 28 3 26 9~~

~~O_N w-o . t- ON >—HWoONcowoONONwoONON^^Ht^-eoenwowoeo^coONt^ 'N-st-wovo-^i-wooo-^-ooNOt^-t'-^oT V) \l*l ^r w ) uu -^" ou M*> l^- l^ Wo ' ^tt^hint^^^^^vovomin 26 3~~

~~00 ON OOOTJ-NONOOONOVONOOONVOO ON NO O O NO O ON ON ON ON "t ON NO OO i 0NOO00O000N00000N 'ONO 1 ON ON O ON O OO ' 00 rHrHNrHN^pHrHrHrH -—I CN -H rt CS r-H CS~~

~~OONNOOO'tfOOONOONONOOOONONOONOCNOOrNONCNONON O ON O ON 'OOt^OOt^ONOOCNOOOOOO 'ON00 ' ON 00 NO ON 00 ' ' ' ' VO ' NO~~

~~^CNco^r-woeowoi—ii—i n ci r-1 n H ^Ht^tnfirt^Wr-i wo f- ON ON r-voor-NOt^ooooooooooooc^t^ (^ t^ NO NO NO t"^ NO ' OO Id ^t if l (S M N N N M CN M tS (S N~~

CO t"- ON ON rtfiinnnrtHt'iHHt'i eoONcoONr-HONONONeotêncnON l> •* W0 •<*• t~~t>ôr~-t^t^NONovovoNO NO NO wo co wo eo wo ! Ô m Tf in ' N N (S (S M C4 (S N M M (S N N (S (N t~- <—i en C~~ ONtÔNC^WOt^WOONCNONON r—iwol-ÔNWOt^^t^-ONON^HONCN rH VO Tf IO TfTfifiviviinvOinrt^MJoo* N N N M •^•rNcnrNenen'srenTfenen NMNNM(NNNMMNM(N eowoONeot^woeot^t~~woONtênt~~cn wo r~~ C^ ON wo t^ >—i wo ON ON ON ON p-* CNenwo-^-cnwoNOen>—H^H^HWO-^ (NMNMNMMNMNMNM(SM en wo ON wo , — lONOUOO'^fV-lOVO"* '

^HcnwocntÔNaNCNentÔNwowoenenwoent~-i—iC"--C~--cnwoeO'-ênco ON O**OW0t--W0O00^HO'—iNONOW0W0NOW0l~-00NONOWOO000C--0000 coONr~~enenencococneoeocococoONONeoONUowor~^^t~~'-Ht~-vo^ûo oocnNOOenooooenenoocnenenenC^(SootôowowoONenONONONt-~oo enONcnwoON'—irnen^Hi—iment—ienenON^HON^HV0"rNONt~~ON^HONt-~t> cnencoONC-lNOoococnwoeocncncneorNeoeot^wowoNOeoeol>ooeoco

NO OOi/or~-^-vênNO^Hc"!rnONr~-mC^t,,~ONt^î—imt"-v>r«"ir«'iC,"-r«")^>nr«-)ONr<"> •^-ooÔNONONONOroinrot-~ONtôoooONOOCNl^H'-*Oooc-10Noocnoo in r- in en m ro m ON t~~ r~~ m in -—i c-»(• - •—irî-^mmmONtîni—i ON <—i i—i m oo vi m NO NO NO NO m m cs n m ^H ON ON ON ON i—i o ^ o r~ M VO »0 h t VO O CN •—I(NOONONOOOONONOOO M is N n en ro c-> CN M (S M N N N h H in in t^ vi n m C~- in en m ON r«-) i—i r>- ON f~ i—imc-iONcif-ONm—-• •——« OO NO ON NO oo oo o t^ cs NO r- t— TJ- NO OCN)ÔCSIONONONCN|OCN^|-ONOI> (N

~~ON in ON m l> m en en en mf^r-Hint^-ONONmcomr-~t^vNinr--Hint^-ON • NO ' NO ON r- NO r~ vo*ONOl"~NONONOl~-NOt^NONONOV">NONOOOt-Ot > N N N M N (N>NNNtN,NNMNNN(SNNNNN(N(S~~

ON l> ON *—I Tj- ' NO ON O NO NO m mNONONOTj-NONONONONOV->mr]-NONONOmir>>/"> CN CN N N N M N C^CNICNCNICNCNCNCNCNICNCNCNCNCNCNCNCNCNCN 00 NO o o 00 OO o 00 00 NO NO 00 NO 00 It o 00 ON ON ON ON 00 00 00 ON ON 00 ON NO ON ON 00 00 OO 00 00 NO 20 0 20 8 20 8 00 00 00

~~CN CN CN 00 00 00 NO CN 00 o 00 o CN 00 CN CN 00 OO NO NO 00 NO 00 NO 00 o NO NO OO 00 00 00 NO 00 00 OO in 00 IT) 00 00 NO 00 NO 00 NO NO 00 00 00 NO NO~~

~~^^r^rnONrnt^c^t ~t^>nt^cnt^ooNot^t^NONO'^-t^ooooin 00 NN(S(N|Nr4Mn(S(NltNlrl~~

•—c ON i—ii/ii—ir-^ONi—<^-ONroinirim^Hint~-v-iromr-~NOinminmmininmr»")CNi/">''trou">comm N N M (S N N M(SNMNMNM(SNNM(N(SNfSM(SM(N(S c«imr<->minmONmint^rnC~-^Hin^H^HONc«iONC"-- m N m vi ^ m CN CN CN CN CN CN CNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCN

~~ON ON m ON m ON en en~~

~~en ON in ON en en en en en ON en ON ON ON ON en ON ON en en r- 00 NO 00 NO NO en en en en en in CN en ON 00 NO as 00 F^en ON en~~

~~ooaoMn^viw^oowo csm^-'tinNOC^ooONOro o rHrHrlNN(SNNNMM(SM(SC<1M o O r-! ,-( NONONONONONONONONONONONONO*ONONONONONONONONONONo o o o o o o OOOOOOOOOOOOOOOOO oooooooooooooooooooooooo~~

r-H ON ON in —< ^-< m r-~ m co en ^H m i—< in coONmON^ONmt~-C--mmr--cs o NOONON^-^-OOincNOcO en co i—i m NO NO co CN -* ON ' co (S r-< I - 'fi ^ m cocNCNroroeorocococo co co ro co co ro co ro co CN co co ro co co co co r-HONONmro~Hrot—>t^>—i I—1 i-H r—H ON ,—I F-NONOCOOONONOOOO oo o t-~ 00 ON r_voo^o\^v£)ON^^

~~0\t>t"~-ONror^l>ONr-< l> f- t- ON ^H O r—ii—iroromi—icomi—icococot~~ CNCNCNCN"*NOCNCNt>CN CN m CN CO m N N (S N M~~

~~t^comintÔNtÔNin'—< .-H ^H ^H t^ co m t^ ON m r-i^HCNcot--t^mmC-- r~t~~f~ooc~-t^tîovioo oo oo ON ON ON r^c^NomcÔNNOF^mrorot^ro N N N (S (S~~

~~co t- r~ t-> ON r- (SMCSMMtSNNMM N N tN M N~~

~~^O o NO NO NO NO CN CN 00 NO CN o 00 00 CN o 00 oo oo 00 00 ON 00 ON 00 00 ON ON 00 ON oo ON 00 00 ON 20 8 ON 20 8 20 8 m ON ON~~

~~NO 00 00 o o 00 NO o NO NO NO CN NO NO o o O NO CN CN CN 00 00 o CN NO oo in oo 00 00 00 00 m m ON r- r- NO 00 o ON 00 ON 00 ON~~

~~ONt--c-~t-~-cor--cococoON ON ro ro ^ m ONt^roO\rorO'-HONroONmt'"--ON CNONONmOmt^ONt^ON ON r^ ON TJ oo ON CNCNCNCNrOCNCNCNCNCN CN CN CN CN CN CNCNCNCNCNCNCNCNCNCNCNCNCN~~

~~ONromC^-t^-inr^^ÔNtÔNmromONinONmrotÔNON^HONONONON CNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCN~~

~~CNCNCNCNCNCNCNCNcNCNCNCNCNCNCNroCNCNCNCNCNCNCNCNCNCNCNCN~~

~~U-)r-Hf~-<^N^HON~~

~~CNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCN t--ONro^c--cocomcot"~ONr-—i H OI ^OO^CNinNO"3-inNOO^ONOCNCNcO^'-~~

~~- - ON in ON ON in r- ro f—1 CO ON ON in in - i—i in m H f—H f—1 i—i i—i r— in CN NO NO o CN £ CN CN £ •* o o CN in o~~

ON in ON co ON ON CO CO in ON ON in CO ON CO in ro CO CO ON m VO CO 00 ON o NO 00 F^ 00 r- NO o NO in 00 NO ^o in ON ON ro in in ON in CO CO CO ON in CO in in m ON CO CO co NO NO in CO ro NO oo in ON in in NO CO ro NO in in NO l> NO NO in r-

TtininNOCôoONO^HCNcNco-<*invoooONO^HCNco'î-inNot---t~-oooN ooooooooooooooooooooooooooococorococococo'^-'^l-TfTj-TfTj-'rfTa-Tj-'rtininininininininininion oooooooooooooooooooooooooooNONONONONONONOÔNONONONONONONOÔNONONO^NONOÔNONOÔNONoO NNNNNNNNNNNNNNNNNNNNNNNNNNNN OOOOOOOOOOOOOOOOOOOOOOOOOOOO >—iooo^-cot-~Tj-co>—imtN—1«/1 ON 0> t^- ^H ON t^ i—ir-icoONcoint~--ONcomC"-ONONm co^foo-^-coNO'fco ^HcooomcocNlcooocom^l-

~~co co co ON •—i i—ICX-O>F-~COIX-ONCOCOI—• C— t— •—i r— «——• TfrOOO^0>N0N0OOr-*00O0Ntx-OO0NCN|0N OlNNNNNNN~~

~~.—< >—i h o\ t~- in H tri mmco^tx-cocom>—i >—' co t^- co >—i H h m o\ h ^ •n-TtoNr>ot~~,«i-r<-iONt^oi>ONmr-i^»oooo(N^o NMN(SWMNM(SMnr)NN(SNM(S(NN~~

'-^cococoinco'—it-~ON inONt"~-ONCO^^ÔNCNONCx-^HONl"'--mONON^H (StS(S(NfSMM(SM oo NO • e» NONO(x-tx-tx-mmr-~NONONOt--.r-.mmô NNMN(SNNN(SM(NNMMN(NCSN(N co in co co ON >—imm^H NOcoNOr--NONONONOco minNONONONOmmîninNO^J-mNOtx-coco-^-

~~VJO Ti 00 NO o NO NO o NO o 00 00 TT ON 00 ld 00 00 00 00 00 00 00 00 00 ON oo 00 OO m 00 00 00 r- 20 0 ON o> 00 ON 20 8 20 0~~

~~O es 00 o ON NO o o 00 o 00 00 00 NO o NO o o NO O o 00 m 00 m~~

CN^HONint^mC^t^ i—i ON co co m i—i t~- ^HfTl>nn^Hl>^-lr-HON mOOf-inONOONOin NO NO in NO NO NO 00 t>Not^-ôr~-NOr^NOt>^-Not-- • O M N M N (N N N in hHt^mhtnO\in in f» >o f» (*•• ^H in CÔNmt^r^rÔNtÔNinONinON mf-NomNooomm co in co -* •<*• NO NO roint^mrnNOmNO'^-m'*-^- ' tStNMNNNNN N (S N M fS N N fNtN*• co •*}-•<*•

~~ON l> CO ON in in CO in in CO ON t- CO CO CO CO in CO NO in <—i CO NO 00 •* CO in •* NO in in in in CO in~~

~~CO CO ON co ON ON o CO CO CO ON in ON CO ON ON co CO CO CO NO NO ON CO CO CO CO CO NO CO OO ON CO CO NO 00 NO co NO~~

OHNntmahl!0aOHN(riîniohlX)0\OHNn"tin*0h NONONONONONONONONONOtx-Cx-tx-tx-tx-t-~f^t^.tx-t--.00000000000000 00 oooooooooooooooooooooooooooo oooooooooooooooooooooooooooNONONONONONONONONONONONONOÔNONONONOÔNOVONOÔNONONONOV©o N N N N N N N NNNNNNNNNNNNNNNNNNNNN OOOOOOOOOOOOOOOOOOOOOOOOOOOO m n 0\ -H w n m i—i .—i in co ON in ON NO •* h h oo> t 39 5 29 5 40 7 cocococococococo cococococococococo 41 3 co co co co co co co

~~m i—i m h n a ^ oo minmc^-minco!—*^ a a m vi h n r- "* CN NO m t*» cs oo N m N n ^O rf 34 5 36 9 36 3 30 5 co co co co co co co CN cococococococococo 29 5 co co co co co co~~

~~CO r^ CO r-H U-) i—i t^- co co -—< ON co ON ON ON i—i C~~ m »—i r- co ON M CS ^ o r- oo CN r» i> r-~ t> NO t^- NO i> 'oo i o ON NO ON (N ON 00 29 7 tS (N m co ts CN)~~

~~>—i r- m co ^H os t~~ in i—iinmr-nONtx-inONtx. ON co C-- ON t~- ON ON r~- in N h m Tf \o (S 'oo i O (N O~~

~~cof^inr-moNt^c^ ON t*» t~» r-1 in O i—i ON co co co co i—i i—i ON •nr-NoinNO'^i-t^co in NO in NO~~

cocoininco ON in l-» i—< in C~~ •—i co co ON I—I ON CO in in in i—i co in co -3-NOin^-NO'n in in in NO in NO NO NO NO NO NO in NO (NMNM(S(StSN N (S N (S N M N (S M N tS M M (N N fS N CS CS N OOO'+^-TtNOOvO Tf t 0O t N C O ^ ft O^ O *0 NO Tf 00 —I

~~ON ON ON f—ICOON—IONI—I—•lONinONt^i—i C*- i—i co m co t~- m 1 •*NOUONOt^NONOONNONONONOONt~--^-NOCOOO 27 9 25 7 26 3 25 9 26 5 25 5 NN(SNNNNMNMN(NlNMN(S(Smm~~

~~ON O^ ON ONONONCOmr—iCOmC^-—~~

~~ON t^ONinONinONincoin ON t-- ONmr--in^^Hr-~~

~~ON CO ONt^r~-t^cocoONcoint^inr—iincoON-—i^H^Hin l 21 9 20 7 24 9 24 9 24 9 22 3 20 9 r^i(Nf^cNcsrqrNicN|fSrN|(N(NrNicN|cNicNCNi~~

~~in ON co CO CO co r~ ONt~incoinininincoincoincoONinONf-~co—<~~

~~in in in in ON CO I—I ON in ON ON in in CO in in ON in ON in o in in CO NO c-4 ON CO CO ON 00 NO CO CO tx- CO CO in NO in NO~~

~~CO ON ON ON ON CO~~

~~ON in ON in CO fx ON i—i ON CO ON CO in NO F^ in ON 00 ^ ON ON ON NO ON ON o F- NO co o ON in ON O F-~~

OOONO—tNw^^vohooo\o^Mtn •<*inNor~-ooONO^H(sico^j- 00 OOONONONONONONONONONONOOOO o OOOOOOOOOOO-^^Hrt OOOOOOr-H^r^^r-< NO NONONONONONONONONONONONONONONONONONONONONONONONONONONO o_ ooooooooooooooooooooooooooo OOOOOOOOOOOOOOOOOONNNNNNNNNNNNNNNNNNNNNNNNNNNO O O O O O O O O NO rt O N o NO NO NO in in CO r- o CO cs CO mOONONONOr-ONOO r~~r-ONONco'—irom 255 255 r- o 0 263 275 233 293 327 343 O N o NO •—' r- 0 ON ON CO CO 0 0 cs

~~H 263 265 275 257 265 289 309 295 295 O N o NO ^ 0 o in r- CO in CS in o o 0~~

rt 229 255 255 257 261 227 297 387 387 O s$z N o NO ON r- in co r- 0 CO o CO CS 0 0 269 —i i—coONmr- r- Tt r- 0 N M w Ôc4«vo ^ inmo\H N (SM N «^0Oh in f-ONI—me ON ^H0t~~ f- ClWHIfr0\ t~~ r-i>CO cscscscscscs cscsccsc NO T)-m'r'•st-st o incoinONONt— ONin^înONinin cscscscscscscscscscscscscs r-cnmcncnenNONOmmmm-st ^-HinONr-t~~C~-COCOONONCOONin ONr-NOincONONONONONOinNONO co^Ht-t— r~^Hrot~-^HinONcoin in N (SM NO in< C*- coONin CS o NOmr- i—l C~-IONC ON l-»t~~co co—i r-mcot co r~mine 1 1 r-CSmNO Oincst>^ O N o NO CO in in in in cs in ON CS N M 0 NO CS cso N O NO CO NO NO NO r- CO in r- 0 CS 0 O N o NO CO r» r- r~- ON 0 CS 0 0 N o NO CO 0 r- CO CO in r- cs in o O N o NO CO ON in NO in CO ON CS ON 1 ' 0 ON O N o NO T o r- r- in in in O N o NO T —H NO r- CO in ON 0 0 cs — ON r~-in O N o NO T cs NO NO CO ON CS CS NO ON ON 210 263 1 363 O N o NO Tj- CO in in in in in r—i o 0 o 0 ON 255 267 293 307 1 O N o ô •<* •*t cs in 0 ON "vt r- r -si ON 259 267 295 i 299 O N o NO •"I- in 0 CO in in CO 0 cs 0 es 259 267 299 297 CO o O N o NO ^r 0 o cs in cs o in r- m 0 0 r- o 255 263 289 287 415 in cs ^H m ON Tf H OO co NO ro C~~ NO t,^ N_O • co^t^cocNi'~>Nor--'Trininoocoooo m •* m co co co co co co co co^cococococococococo'^cocococoeo

ONmONt--ONcoint^mt~ONco comNOcocoOtncococN 'NO cot~--cN--*inTj-coco'^"''tf'coOONONOO cocococococococococo co COCOCOCOCOCOCOCOCOCOCOCOCNCNCOCO r-- m i—i ON .—H c-- c^ ON ON *—i t-» m i—I in t^ ^H t"~ t^- ON i—iU~>ONi—imminent^ ON ON r-H ON t—i M ^ N ' 00 CN 00 ONONONOOOONOOI—ICOO"—-H NONOt--i--c^mNONot^t--^NONONONONONOt'"--inNONONONooooooot~-oo

~~ONmcoC"^<—<>—'ONmcot^minONtnl^i—it"-cot"~--C~-Ot~--coONmmONin incoNOco^OminmNONONOcomNOmcominmmNONONOooooooNO'^-~~

~~•* NO CN CNOOCNONOOCNOOON OOONNOO^tOTtOONONOONONONO ON oo r^ ON 00 OO ON ' 00 ON 00 ' O00 't~-0N00 0N00OOt^0N0N0N0N0N~~

~~00 0O CN OVONMO\NlOtaiNtW^NOOOOtN'tO' , i—I NO OO r-H a iO tO rt a « Nom^Ht^mNoinNOcoaNt"--t~'-c ~-ooooooNor'- CO CN CN CN CO cN CN en CN CN co m~~

~~CO ON in IT) in f» t- r~- ^- ^—< r—~~

CO CO ON in r-- t~~ to in .—i in ro ON <—i i—i inONt^-t^t^r^t— < t~~ >—i co co co in in m cN ' •«*• t ^ m 1- ^ * CN ' m « N •^-CN-^^-com^i-^-cocococNcN CN CN CN N M N M N M CN CN CN CN CNCNCNCNCNCNCNCNCNCNCNCNCN co in f- co f—i co co ON co in •—i CO CO CO ON COCOONCOONONONinr—lONf^intO in in in CN in CN CN m C-- oo m ' • t-~ m oo oo (-- oo O H O O in o O O t~~ r- CO CO in ON in CO CO in CO CO ON ON CO ON ON F^ F^ F^ in F^ NO CN NO CN CO (N CN in in o o in in in

CO CO CO ON CO CO CO in CO ON in CO CO in ON CO co in r- CO i> ^ in in NO O I ON CO CO 00 ON o CN CO CO CO NO co CO in CO in CO CO in CO in ON in CO CO ON ON ON ON ON ON CO 00 ON oo 00 in o ON o oo 00 in CO CO NO CO NO t-- CN CN CN CN CO

ON © ,—i CN CO ^i- in NO r^ 00 ON o r—1 CN CO • t- i> t~- f- r- NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO o o o o o o o o o o o o O o O o o o o o O o o O o o o o N N N N N N N N N N N N N N N N N N N N N N N N N N N N O O O O O o O O O O o O o O O ooooooooooooo ONOoo ' Cs(^oooomO<7\«^m^,^-roOOOiri

l/l lO I/") oooooooÔTfTj-(S|"*ôo(sr-o • oo o ^H o ^H ^ CN (N (N (SI fN m r^Hi/->r'NONU">C'\C-- —' n m 0O 00 t~- 0\ 1 •tÔOOOl^-'frcNCNCNcNÔOOtÔO ' l> t~- O ON —* t> CN CN CN voTj-voo>ôr~tÔNON\ooo^Ht^t^^vor^r-t~~ôr-~<îri*iôco

~~»/")(r>^Hr-t^rv">CN'—iw-)r^immt~-t^-mt~-u^ r^00r~-<^TfvOvOVOr~-'<*0000iriir)VOVOiOvO'«tt---vovo^0u-iir)r<-ivOr^~~

~~OOTj-ONTtO^O.VO^tOCN^-OvooO O-^J-tNONCNONOOvOONONONVO~~

~~"st-rsooooofscsoeNvooooôoôcNONcsoNOOôoNONô oo ON t~~ ir--o\~~

~~t-~-C-~mt^»/"iONco>/"ir^î—iON»r>rot'~-rn'/">u">^0-\t--c<-)r<->0\ t^r-oooot^ôoooovoOi—ivooooovooooooooovocôow-)V£) 28 3 27 7 28 5 28 3 IT)~~

~~t~-r^t^t~-t^--t^>OCT\>ot^f^ONromr^mmmONu-)ONONr~- 22 9 23 7 25 5 23 7~~

~~t> as m f~ 22 5 22 9 22 9 22 9 N(S(SNNMNNNNNN NNNNNNNNMNM~~

~~ro rt -H in m H îi—i m m r«i co •—i M r^ (^ ON n a H v> OO lO >/~>l/-){^t^->Ol/->l/->OO^rtO00O0000000000CNCNl-- ' 00 noomôoôohoo«inHr-HTJ-^,—ioooooooooo~~

~~^aimaoi^rih^r-iv^f^r-iioONONfomt-~t~~t'^t~^t~~t~--t~~t~~v^'—i cNcNm'* ' oooTrt^^3f~NOC^Ti-t--r--t^-vot^ON~~

ONhONtÔNfflrHÎ—<,_it"«t*->—1>—li—li—'or-c~-r--r-t--O.. v N_ . C-4 NO <> ^mnnnmiftm«iinini,Kih\O't^^f\o^\0voh ooONO^H^HtNm-^-ioôr~ooONO^HONot--oooNO'^tNmioô t^-rÔOOOOOOOOOOOOOOOOOOOOOONONONONONONONONONOOOOOO l l oooooooooooooooooooooooooooNO*0*OVO\OVO'O ONO*OVOÔ*0^0\0>0\OC)*0*ONONO'ONO»O ONONoO NNNNNNNNNNNNNNNNNNNNNNNNNNNN OOOOOOOOOOOOOOOOOOOOOOOOOOOO o-> r- fO VI i> P ON Ju3 m CM

~~m 9 3 34 7 ff\~~

~~ON 00~~

r*4 3 23 5 f> t^ VO 00 £ CM NO Q NO -+-«» mm t—1 r^ *4*) « r- © in 255/ 2 NO (J JS in S CM 5 1 23 IT) CS O in ON VI •* CO (N >s CM 4 9 24 1 00 ON 1-1 ,_< ~3\ OQ m § > ON CM 6 5 10 3 ON i—i —

~~~~O t-H o -a ON ON NO o o (a NO o 's •-. o XI o QH OH •a~~~~

1/3 2 o 3 o o oo C la >-i O S~ o .—i J3 o o X! o o o OH OH '•§ c 3 o ex o •S o. t-» i—< u v~, O (N 00 1 us si o o o o £S o o o o CD H OH 0- C/3 c o SP ••§" •> co C~~ © © © o o o o o a 0) i—n Tj- O NO -^f (N ON Tf NO •<*• • F—< T3 © © c-> © © O rH Tt O (S © o o o o" o o o o o o

~~~~CN NO u-i u-i O~~~~

~~~~00 W 00 N SO Ifi Q •a *-H *4 C5 TH >H ffj 3 •a tzi (Z5 Xfl (Z! fi ON Xfl Xlilfl z S u 3 NO t-- a, o > -S J3 Ph r OH stf -J ON 0- OH OH OH OH Ph r I PH OH OH OH OH If} s. JS PL;~~~~

~~~~«"> O S- CN O o o~~~~

~~~~in N >o ^ oo •* o CO vo ^H ^H Tt m o (N o ^H ^H o rr ^ —i s o ©'©'©'©' © © I 00 CO &0 CO s «s 5; \o r» r- u ip t. i. s- o ^ > ^ JC JS Cu CL CH ON CM~~~~