This item is only available as the following downloads:
1 HISTORICAL BIOGEOGRAPHY AND CONSERVATION GENETICS OF CENTRAL FLORIDA SCRUB ENDEMICS By CHARLOTTE C. GERMAIN AUBREY A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 201 2
2 2012 C harlotte C. G ermain A ubrey
3 To my p arents, Chantal and Bob and my wonderful husband Thomas for letting me keep my head in the clouds, and to my children Eliott and Rose for keeping my feet on the ground.
4 ACKNOWLEDGMENTS I would like to thank my chair and co chair, Matt Gitzendanner and Pam Soltis for suppo r ting me througho ut the whole process of this dissertation, especially in the more difficult moments of doubts and set backs; the rest of my committee, Jim Austin, Steve Manchester, Doria Go rdon and Eric Menges for their support and ideas to make this project successful. A lso, I would like to thank Doug Soltis, Gordon Burleigh, Reed Beaman, Kent Perkins and other member of the Biology Department and the Florida Museum of National History for their sharing their expertise so generously at different points in my project. I w ould like to thank everyone who has helped me in the collections of plants outside of Florida: Bryan Connolly, John Nelson, Josh Clayton, Ashley Morris, Robert Peet, Wesley Knapp, Irene Kadis, Steve Leonard, Otto Gockman, Heather Alexander, Anne Malatesta, Deborah White, Lucas Majure and Richard Carter. Within Florida, I thank Jason griffin, Beatrix Pace, Eric Menges, Paul Corogin, Richard Abbott, Anne Malatesta, Wendy Poag, Lucas Majure, Keith Clanton, Eric Ergensteiner and William Carromero for their help in the field. For permits, I thank Bryan Benson of the Bureau of Plant Inspection, Division of Plant Industry, The Ocala National Forest, and The Florida Department of Environmental Protection, Division of recreation and Parks. For access to sequences, I thank Richard Abbott, Joey Shaw, Alexandra Gottlieb, Jens Rohwer and Andre Chanderbali. I thank all my lab mates, especially Vaughan Symmonds and Monica Arakaki, the masters of microsatellites, Lucas Majure and the computer savvy Stein Servick for their h elp with chromosome counts and unforeseen polyploidy issues, Kurt Neubig, Lorena Endara, Julie Allen Emily Saarinen, Maggie Hunter and Josh Clayton for their help with analyses and interpretation of data.
5 Also I would like to thank the different agencie s and organizations that have funded this research and its dissemination: the Florida Native Plants Society, the American Society of Plant Taxonomists, the Women in Science, the UF Graduate Student Council, the Olowo and Riewald Memorial Fund, and the HHMI GATOR program for funding Taylor Thurston and Cory Nelson as undergraduate s to be mentored in the lab for one year as well as Patricia Soria and Khusbu Shakafi for volunteering on my project. I would like to thank all the people who, knowingly or unknow ingly, have inspired and mentored me in becoming a better scientist and more rounded person, especially my female role models Doria Gordon and the wonderful Pam Soltis. Finally, I would like to th ank my family. My grandparents Pierre and Mar ie Louise, par ents Chantal and Bob, and brothers Sebastien and Julien for always encouraging me in my ambiti ons, being happy for my successes and sad for my failures. Most of all, I would like to thank my infallibly supportive and loving husband Thomas, and my two beaut iful children Eliott and Rose for never complaining when I had to work late.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURES ................................ ................................ ................................ ........ 11 ABSTRACT ................................ ................................ ................................ ................... 13 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 15 The Central Florida Scrub ................................ ................................ ....................... 15 Ju stification of the Study ................................ ................................ ......................... 16 The Study Species ................................ ................................ ................................ .. 17 Project Outline ................................ ................................ ................................ ........ 20 2 USING COMPARATIVE PHYLOGEOGRAPHY TO RETRACE THE ORIGINS OF AN ECOSYSTEM: THE CASE OF FOUR PLANTS ENDEMIC TO THE CENTRAL FLORIDA SCRUB ................................ ................................ ................. 23 Materials and Methods ................................ ................................ ............................ 30 Sample Collections ................................ ................................ ........................... 30 DNA Extractions, Amplification and Sequencing ................................ .............. 31 Data Matrices, Sequence Alignment and Phylogeny Reconstruction ............... 32 Hypothesis Testing ................................ ................................ ........................... 34 Results ................................ ................................ ................................ .................... 35 Phylogenetic Analyses ................................ ................................ ..................... 35 Biogeographic Hypotheses ................................ ................................ ............... 36 Discussion ................................ ................................ ................................ .............. 37 Phylogenetic Relationships of the Focal Species ................................ ............. 37 Biogeography and the Origin of the Central Florida Scrub ............................... 39 Implications for Community Assembly and Conservati on ................................ 42 Conclusions ................................ ................................ ................................ ............ 44 3 IS MICROSATELLITE FRAGMENT LENGTH VARIATION THE BEST MARKER FOR POPULATION LEVEL STUDIES? THE CASE OF POLYGALA LEWTONII ................................ ................................ ................................ .............. 54 Microsatellites and Flanking Regions ................................ ................................ ..... 54 The Central Florida Scrub ................................ ................................ ....................... 57 Polygala lewtonii ................................ ................................ ................................ ..... 58 Material and Methods ................................ ................................ ............................. 60 Plant Collections ................................ ................................ ............................... 60
7 Population on the Nature Conservancy Saddle Blanket (108) .......................... 60 Development and Amplification of Micro satellite Primers ................................ 61 Characterization of Markers ................................ ................................ ............. 62 Comparison of Markers for Molecular Diversity Indices ................................ ... 63 Bottleneck Detection ................................ ................................ ........................ 63 Population Structure ................................ ................................ ......................... 64 Sequence Analysis ................................ ................................ ........................... 65 Results ................................ ................................ ................................ .................... 66 Characterization of Markers ................................ ................................ ............. 66 Comparison of Markers for Molecular Diversity Indices ................................ ... 67 Hardy Weinberg equilibrium ................................ ................................ ....... 67 Overall genetic diversity indices ................................ ................................ 68 Bottleneck detection ................................ ................................ ................... 68 Population structure ................................ ................................ ................... 69 Polygala lewtonii ................................ ................................ ........................ 69 Genetic Diversity of Polygala lewtonii ................................ ............................... 70 Discussion ................................ ................................ ................................ .............. 72 Markers for the Study of Polygala ................................ ................................ .... 72 Marker Comparisons and Implications for the Use of Microsatellite Fragment Lengths ................................ ................................ ......................... 73 Genetic Diversity in Polygala lewtonii and its Implications for the Conservation of this Federally Listed Species ................................ ............... 76 Comparison of Polygala lewtonii and P. polygama ................................ .... 76 Genetic diversity and partitioning within Polygala lewtonii ......................... 78 Conclusions ................................ ................................ ................................ ............ 80 4 FINE SCALE POPULATION GENETIC STUDY OF THREE PLANTS ENDEMIC TO THE CENTRAL FLORIDA SCRUB ................................ ................................ 101 Introduction ................................ ................................ ................................ ........... 101 Materials and Methods ................................ ................................ .......................... 104 Study Species ................................ ................................ ................................ 104 Collections, Microsatellite Amplification and Genotyping ............................... 106 Microsatellite Data Analysis ................................ ................................ ............ 108 Results ................................ ................................ ................................ .................. 110 Asimina obovata ................................ ................................ ............................. 110 Ilex opaca var arenicola ................................ ................................ ................. 112 Pr unus geniculata ................................ ................................ ........................... 114 Discussion ................................ ................................ ................................ ............ 116 Asimina obovata ................................ ................................ ............................. 116 Ilex opaca ................................ ................................ ................................ ....... 117 Prunus geniculata ................................ ................................ ........................... 121 Comparisons ................................ ................................ ................................ .. 123 Impacts of Anthropogenic Activity on Endemic Species .......................... 127 Conclusions ................................ ................................ ................................ .......... 128
8 5 SYNTHESIS OF RESULTS: CONSERVATION IMPLICATIONS FOR THE CENTRAL FLORIDA SCRUB ................................ ................................ ............... 151 Taxonomic Units and Conservation ................................ ................................ ...... 151 Conservation Genetics ................................ ................................ .......................... 152 Genetic Diversity and Partitioning ................................ ................................ .. 152 Allelic Richness ................................ ................................ .............................. 153 Gene Flow and Inbreeding ................................ ................................ ............. 153 Materials and Methods ................................ ................................ .......................... 154 Central Florida Scrub Species ................................ ................................ ........ 154 Comparison of Relative Measures of Genetic Diversity ................................ 155 Results ................................ ................................ ................................ .................. 155 Taxic Resolution and Relationship ................................ ................................ 155 Genetic Partitioning and Gene Flow in the Central Florida Scrub .................. 156 Discussion ................................ ................................ ................................ ............ 1 58 Taxic Resolution and Relationship ................................ ................................ 158 Genetic Partitioning and Gene Flow in the Central Florida Scrub .................. 159 Commonalities with Already Published Studies ................................ .............. 160 Conclusions ................................ ................................ ................................ .......... 163 6 MICROSATELLITE MARKER DEVELOPMENT FOR THE FEDERALLY LISTED PRUNUS GENICULATA (ROSACEAE) ................................ .................. 167 Methods and Results ................................ ................................ ............................ 168 Conclusions ................................ ................................ ................................ .......... 170 7 CONCLUDING REM ARKS ................................ ................................ ................... 173 LIST OF REFERENCES ................................ ................................ ............................. 178 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 197
9 LIST OF TABLES Table page 2 1 List of primers used for all species in this study ................................ .................. 45 2 2 Results of the AU tests on different phylogeographical hypotheses for all 4 c entral Florida endemic species. ................................ ................................ ....... 46 2 3 All accessions used, herbarium voucher number and location, sample location, reference, and all GenBank accessi ons ................................ ............... 47 3 1 Populations, Location, Number of plants collected and population voucher for all Polygala lewtonii and P. polygama. ................................ ............................... 82 3 2 List of markers developed for Polygala lewtonii ................................ ................ 83 3 3 Characterization of Polygala markers for fragment length and repea t number ... 84 3 4 Characterization of polymorphisms in flanking region sequences for the 4 loci. ................................ ................................ ................................ ..................... 85 3 5 Molecular diversity of Polygala lewtonii populations for all marker types. ........... 86 3 6 Summary of Chi Square tests for Hardy Weinberg Equilibrium for Polygala lewtonii populations for all loci ................................ ................................ ........... 87 3 7 Overall genetic diversity in Polygala lewtonii and P. polygama ......................... 88 3 8 Wilcoxon bottleneck detection test results for Polygala lewtonii ......................... 89 3 9 Among populations F ST and R ST values for Polygala lewtonii .......................... 90 4 1 Sampled populations for the Asimina study. ................................ .................... 130 4 2 Sampled populations for the Ilex stu dy. ................................ ............................ 131 4 3 Sampled populations for the Prun us study. ................................ ...................... 132 4 4 Overall molecular diversity indices for Asimina obovata and A. incana ........... 133 4 5 Overall molecular diversity indices for Ilex opaca and I. cassine ...................... 133 4 6 Overall molecular diversity indices for Prunus geniculata and P. maritima ....... 133 4 7 Molecular diversity of populations of Asimina obovata ................................ ... 134 4 8 Molecular diversity of populations of Ilex opaca ................................ ............... 135
10 4 9 Molecular diversity of populations of Prunus geniculata ................................ .. 136 4 10 Wilcoxon bottleneck detection tes t results for Asimina obovata ...................... 136 4 11 Wilcoxon bottleneck detection test results for Ilex opaca ............................... 137 4 12 Wilcoxon bottleneck detection test results for Prunus geniculata .. ................... 138 4 13 Gene flow e stimation between populations of Asimina obovata. ...................... 138 4 14 Gene flow estimation between populations of Ilex opaca ................................ 139 4 15 Gene flow estimation between populations of Prunus geniculata .................... 140 5 1 Molecular diversity in four endemic to the central Florida scrub. ...................... 164 6 1 Loci developed for P. geniculata and P. maritima and their characterization .. 171 6 2 Characteri zation of populations of Prunus geniculata and P. maritima var. maritima ................................ ................................ ................................ ............ 171 6 3 Monomorphic loci that amplify in Prunus genicula ta and/or P. maritima ........... 172
11 LIST OF FIGURES Figure page 1 1 Historical co asts of Florida ................................ ................................ ................ 22 2 1 Phylogeographical hypotheses for the Prunus co mplex. ................................ ... 51 2 2 Prunus chloroplast + ITS combined maximum likelihood ana lysis with bootstrap values ................................ ................................ ................................ 52 2 3 Polygala ITS maximum likelihood analysis with boots trap values. .................... 52 2 4 Persea plastid and ITS combined maximum likelihood analysis with bootstrap values ................................ ................................ ................................ ................. 53 2 5 Ilex ITS maximum likelihood analysis with bootstrap values. ............................ 53 3 1 Sampling of Polygala lewtonii and P. polygama P. lewtonii .............................. 91 3 2 Distribution of Polygala lewtonii populations. ................................ ...................... 92 3 3 Genetic and geographic clustering of both Polygala species. ............................ 93 3 4 Genetic and geographic clustering of P. lewton ii ................................ ................ 94 3 5 Contributions of populations of Polygala lewtonii to allelic richness. ................. 96 3 6 Distribution of Mantel test results ................................ ................................ ...... 100 4 1 Distribution of sampled populations for Asimina obovata and Asimina incana 141 4 2 Distribution of sampled populations for Ilex opaca var. arenicola I. opaca var. opaca and I. cassine ................................ ................................ ........................ 142 4 3 Distribution of populations of Prunus geniculata and P. maritima .................... 143 4 4 Geographic and genetic clustering of Asimina obovata. ................................ ... 144 4 5 Geographic and genetic clustering of Ilex opaca ................................ ............. 145 4 6 Geographic and genetic clustering of Prunus geniculata and P. maritima ....... 146 4 7 Results of Mantel test for Asimina obovata ................................ ..................... 147 4 8 Result of Mantel test for Ilex opaca ................................ ................................ 148 4 9 Result of Mantel test for Prunus geniculata ................................ ..................... 149
12 4 10 Contributions of populations of Asimina o bovata to allelic richness. ................. 149 4 11 Contributions of populations of Ilex opaca to allelic richness ........................... 150 4 12 Contributions of populations of Prunus geniculata to allelic richness. ............. 150 5 1 Conservation genetics synthesis of dissertat ion species. ................................ 165 5 2 Map of gene flow barriers from other species endemic to the Florida scrub.. ... 166
13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy HISTORICAL BIOGEOGRAPHY AND CONSERVATION GENETICS OF CENTRAL FLORIDA SCRUB ENDEMICS By Charlotte C. Germ ain Aubrey May 2012 Chair: Matthew: A. Gitzendanner Co chair: Pamela S. Soltis Major: Botany The central Florida scrub is considered the third biodiversity hotspot in the United States after Hawaii and the California Floristic Province. This hab itat is threatened, with 90% already lost to development For my dissertation research, I took a multispecies, multidisciplinary approach to conservation planning. Five plant species, ranging from herbs to trees, were selected to allow for a multispecies comparison, and to understand if common events shaped the system as it is today: Asimina obovata (Willd.) Nash (Annonaceae), Persea humilis Nash (Lauraceae), Ilex opaca Aiton var. arenicola (Ashe) Ashe (Aquifoliaceae) Polygala lewtonii Small (Polygalaceae ) and Prunus geniculata Harp. (Rosaceae). To provide a historical context, test species delimitation and phylogeographic origins, I reconstructed the phylogenies for each of the genera to which the focal species belong. One central question was that of th e historical origins of the species, which have traditionally been hypothesized to be either Eastern North American (having retreated following Pleistocene glaciations), or Southern US/Northern Mexican (from the continuous xeric belt in the Pliocene). I fo und that both Ilex and Polygala supported the
14 Eastern North American hypothesis, and it was also the more likely origin for Prunus The placements of the Florida scrub species of Persea and Asimina in their respective phylogenies were unresolved, precludin g evaluation of their geographic origins. In addition, my results indicate that Ilex opaca var. arenicola does not form a monophyletic group, but should instead be merged with Ilex opaca var. opaca The Pleistocene, Eastern North American hypothesis was th erefore supported for two species and the more likely hypothesis for one, confirming the last glacial maximum to be one of the major events that has shaped the central Florida scrub. In order to examine the level of diversity within species, I developed m icrosatellite markers for Polygala lewtonii Ilex opaca Asimina obovata and Prunus geniculata and surveyed diversity in these species and their widespread sister species. For Polygala lewtonii I further studied the evolutionary patterns at microsatellite loci, comparing among the commonly used genotypic data based solely on fragment length, a data set made from the actual number of repeat units as determined by sequencing alleles, and sequence variation in the regions that flank the microsatellites. Fragm ent lengths and repeat number gave different results for most measures of diversity. I offer some advice for microsatellite primer design in light of my results. Lastly, I combine the results of my studies with data from the literature to examine broad pa tterns of genetic diversity among Florida scrub endemics. Some general patterns arose: the differentiation between the Lake Wales Ridge and Mount Dora is generally more pronounced for animals than plants, the central part of the Lake Wales Ridge hosts a co mplex network of genetic diversity and partitioning, and the southernmost Lake Wales Ridge populations are hig hly differentiated from others.
15 CHAPTER 1 INTRODUCTION The C entral Florida S crub The central Florida scrub is a n eco system naturally regulated by fire and dominated by xeric adapted species. The characteristic sandy soil hosts a large number of narrow endemics making it a biodiversity hotspot in North America ( Christman and Judd, 1990 ) The Lake Wales Ridge (LWR), lying principally in Polk Highlands, and Orange counties ( Huck et al., 1989 ; Weekley, Menges, and Pickert, 2008 ) and Mount Dora in Lake and Marion counties form the two main, oldest and highest ridges of the region. Several smaller and y ounger ridges flank them (Figure 1 1) The xeric upland habitat along the central sandy ridge of peninsular Florida is a unique and ancient ecosystem rich in endemics, many of which are specially adapted to this dry sandy environment. Tragically, this env ironment is rapidly disappearing: by the mid 1940s, half of the LWR habitat had been converted to agriculture and housing, and today, more than 85% is estimated to have been lost to anthropogenic activity ( Weekley, Menges, and Pickert, 2008 ) up from 82% in 1990 ( MacDonald and Hamrick, 1996 ) Fire suppression is also a continuing problem for the maintenance of the natural ecosystem ( Abrahamson, 1984a b ; Evans, Menges, and Gordon, 2003 ) The combination of the unique habitat and geologic history, and continued human encroachment has left 40 species federal ly listed as threatened or endangered. It is believed that in the Pliocene and Pleistocene interglacial times, most of peninsular Florida was inun dated as sea levels rose, but that the LWR and Mount Dora were emergent and suitable for plant habitation sin ce the late Miocene or early Pliocene (12 million yrs B.P.) with the Ridge s serving as a re fuge for terrestrial species ( Webb
16 and Myers, 1990 ) Plant fossils have been found in Lake Annie at the southern end of the LWR, dating from 37,000 to 13,000 yrs B.P. in dicating the presence of scrub habitat with Ceratiola ericoides Polygonella fimbria ta P. ciliat a Selaginella arenicola and various Asteraceae, species that are still present today ( Watts, 1975 ) Justification of the S tudy One of the main problems for species with small or declining population is the risk of loss of diversity and the potential for fixation of dele terious alleles ( Frankham, 2005 ) This loss of diversity in turn may reduce the ability of a species to evolve with the selective effects of environmental change By quantifying genetic diversity in populations, we can assess overall levels of diversity compared to other populations and related species, patterns of gene flow and population structuring, if past bottlenecks have a ffected present diversity and what steps can be taken to limit further loss ( Frankham, 2003 ) Patterns of diversity among populations can shed light on gene flow, and the biogeographic history of a group. However, not all changes in genetic diversity can be tied to recent anthropogenic events. The evolutionary history of the concerned species must also be considered. It is increasingly recognized that when studying a species with a restricted range, one should also investi gate wider rang ing congeners for comparison and evaluation of diversity. In order to make better conservation management decisions, any rare species should be studied in parallel with one or two of its taxonomic relatives ( Gitzendanner and Soltis, 2000 ; Broadhurst and Coates, 2004 ) In the case of the LWR the gla cial periods and the shifts of habitats have undoubtedly played an important role in shaping the diversity of species. The present populations are the results of these events, being old refugial populations or pre glacial
17 populations, or having originated through long distance dispersal ( Godt, Johnson, and Hamrick, 1996 ; Chang, Kim, and Park, 2003 ) However, some endemic species may have gone through different histories that have differently affected their current level and partitioning of genet ic diversity For example, some central Florida scrub endemics are believed to be neo endemics as they share a common ancestor or are derived from wide ranging Eastern North American taxa while othe r s seem to have Southwestern U.S. relatives ( Huck et al., 1989 ) The S tudy S pecies In order to study the patterns of genetic diversity in the central Florida scrub and begin to understand the historical forces that have shaped this diversity, I studied several endemics and their congeners. This project focuses on five plant species repre sentati ve of the central Florida scrub. Asimina obovata (Willd.) Nash (Annonaceae) is endemic to well drained sandy soils of Central Florida ( Nash, 1896 ; Huck et al., 1989 ; Nelson, 1996 ) It s nearest congener is thought to be Asimina incana widespread in Florida and southe rn Georgia ( K. M. Neubig pers. comm.), but the phylogeny of this genus is complex and still unresolved. Asimina obovata is a perennial shrub to small tree with typical reddish pubescence on the young twigs, petioles and veins of the lower surface of the le aves and buds, making it recognizable in the field. It flowers from March to May with broad, showy flowers with petals that are green at first and then turn white, with a thick, maroon, corrugated tissue towards the bottom of the inner surface of the inner petals. Flowers are fragrant at maturity. Fruit setting is thought to be higher for outcrossed pollination than selfing, as in other species of the genus. Seed dispersers remain
18 unknown, but gopher tortoises have been spotted consuming the fruit, and smal l mammals may be able to disperse seeds ( Norman and Clayton, 1986 ) Prunus geniculata Harp. (Lauraceae), the scrub plum, is a federally endangered species known from only 21 sites, all of them on the LWR ( Harper, 1911 ; USFWS, 1999a ) Although assumed from morphology to be closely related to Prunus texana and P. angustifolia a molecular phylogenetic study placed it as sister to Prunus marimita from the northeastern U.S. ( Shaw and Small, 2004 2005 ) The scrub plum is a shrub up to 2 m tall, heavily branched, with strongly zigzag twigs and spiny lateral branches. Its deciduous leav es are finely toothed. The five petaled white flowers bloom in late winter, when the plant is leafless, and the fruit is a small, bitter, red plum ( Wunderlin, 1998 ) The amount of flowering and fruiting heavily depends on the occurrence of a fire in th e past 3 years ( Weekley et al., 2010 ) Prunus geniculata is andromonoecious, with both male and bisexual flowers on the same plant. Due to the strong fragrance of the flowers, it is pollinat ed by a variety of insects. It is believed to be self incompatible. Polygala lewtonii Small (Polygalaceae), also federally listed, is endemic to the LWR and Ocala National Forest on Mount Dora ( Small, 1898 ; USFWS, 2010 ) yellow sand in sandhill and scrubby areas characterized by longleaf pine and low scrub oaks and on transitional habitats between high pine and turkey oak barrens ( USFWS, 2010 ) It is also found in clearings and roadsides ( CPC ) A general trend of population recruitment 1 3 years after a prescribed burn supports the hypothesis that seed germination is dependen t on regular burns. Even in populations that had disappeared completely, plants reappear a short time after a burn ( Weekley
19 and Menges, submitted ) This seems to show a dormant seed bank that can be activated with heat, dryness or other factors brought by fires ( Slapcinsky and Gordon, 2003 ; Slapcinsky, Pace Aldana, and Gordon, 2005 ) Polygala lewtonii is an amphicarpic species that has both underground cleistogamous (closed obligat ely selfing) flowers and chasmogamous (open pollinated) flowers occurring on aboveground terminal racemes ( Weekley and Brothers, 2006 ) This mode of reproduction is found in only a few dozen species in the world and combined with its limited distribution and declining populations milkwort is morphologically similar to the widespread P. polygama the common milkwort ( FNAI, 2000a ) Persea humilis Nash (syn. Persea borbonia (Linnaeus) Sprengel var. humilis (Nash) Kopp) (Lauraceae) is a species endemic to the Florida scrub. The continental North American clade of Persea is composed of only 3 nat ive species, P. humilis P. palustris and P. borbonia ( the last two wid espread in the southeastern U.S.). M ore phylogenetic work is needed to confirm that these three species form a clade ( Chanderbali, van der Werff, and Renner, 2001 ) A shrub or small tree, the branches are typically appressed and pubescent. The leaves are elliptic, small and the und er surface is densely pubescent and rusty brown. The species flowers in the spring and early summer and produces few flowers and fruits compared to P. palustris and P. borbonia Also, Persea is a genus of special conservation interest since the outbreak of laurel wilt, caused by the fungus Raffaelea sp. carried by the redbay ambrosia beetle Xyleborus glabratus Eichhoff ( Hanula et al., 2008 ; Mayfield et al., 2008 )
20 Ilex opaca Aiton var. arenicola (Ashe) Ashe (A quilofiaceae) is endemic to the Florida scrub. A phylogeny places it and it s nearest congener Ilex opaca var. opaca (widespread in the eastern U.S. ) sister to I. argentina and I. cassine ( Ashe, 1925 ; Gottlieb, Giberti, and Poggio, 2005 ) Ilex opaca var. arenicola is a shrub or small tree with evergreen light green leaves, narrower than those of Ilex opaca var. opaca distinctive ly revolute and serrate, and armed with a spine longer than 1mm at the apex ( Wunderlin, 1998 ) The fruits are green and turn red, typical of Ilex and are dispersed by birds. Project Outline The first goal of the study was to establish an understandin g of phylogenetic relationships within the genera in order to establish the sister taxa of the focal species and address the hypotheses regarding geographic origin of the species. For each of the study species I use DNA sequence data from the internal tran scribed spacer (ITS), and where necessary, several chloroplast regions. The two main goals were: to confirm the closest widespread congener of each Florida endemic to allow comparisons of genetic diversity in subsequent population level studies; and to sta tistically test biogeographic hypotheses on the geographical origi ns of the central Florida scrub As is common in the recent population genetics literature, I wanted to employ microsatellites for my genetic diversity studies. One aspect of microsatellite studies that is often overlooked is the effect that genotyping assumptions have on the data and conclusions drawn from them. It is common for microsatellite genotyping studies to score loci based on inferred allele size alone. Alleles are typically assume d to differ due to insertions and deletions of microsatellite repeats. However, several studies have demonstrated that when the sequences of alleles are investigated, other mutations play
21 roles in generating length variation. Taking Polygala lewtonii as a test case, I genotyped and sequenced microsatellites to compare and contrast the results of genetic diversity and population structure results using three data sets for the same loci: 1) scoring the fragment lengths, the repeat numbers (based on sequence d ata, omitting the flanking region) and the flanking region sequence (omitting the repeats). According to my findings, I provide advice on marker design and steps to avoid misinterpretation of results in a population level genetic study. The implications fo r the biology of Polygala lewtonii are discussed. I used microsatellites on Asimina obovata Ilex opaca var. arenicola and Prunus geniculata for a population level study. After developing microsatellite markers for all three species, I inferred levels of genetic diversity and partitioning for each species and looked for common patterns. The last chapter pooled all species together in a compar ative study that also incorporated inferences regarding genetic distribution of central Florida scrub endemics from the literature. Both plants and animals were included to get a representative set of co distributed species. From these results, conservatio n management advice is given for further studies on other species endemic to the central Florida scrub This will help in the understanding of past events that built this unique and highly threatened ecosystem and the future needs for its preservation.
22 Figure 1 1 Historical coasts of Florida: 20,000 YBP (dark green), present (white), early Pleistocene (light green) ( Webb, 1990 ) and interglacial shorelines (orange) ( Lane, 19 94 ) ; r emnant scrub (red) and oldest ridges (yellow).
23 CHAPTER 2 USING COMPARATIVE PH YLOGEOGRAPHY TO RETR ACE THE ORIGINS OF A N ECOSYSTEM: THE CASE OF FOUR PLANTS ENDEM IC TO THE CENTRAL FLORIDA SCRUB The field of comparative phylogeography was first def geographical patterns of evolutionary subdivision across multiple co distributed species ( Arbogast and Kenagy, 2001 ) Comparative phylogeography enables explor ation of regions of special interest, such as those with high levels of endemism, a particular natural history, or a need for conservation ( Soltis et al., 2006 ; Avise and Riddle, 2009 ) Species with similar ranges have often been molded by the same historical biogeographical forces, resulting in similar genetic architectures. Studying the historical event s that might have influenced a certain region may allow better conserv ation of its genealogical lineages and evolutionary adapta tions. Conservation biogeography applies the principles and theories of biogeography to the problems of conservation of biodiversity ( Whittaker et al., 2005 ) Integrating comparative phylogeography and conservation biogeography provides a multi species perspective on a region of interest. In Europe, for example, congruent phylogeographica l patterns revealed few patterns of migration from eastern to western Europe through the mountain ranges of central Europe and three main Pleistocene refugia the Iberian peninsula, Italy, and the Balkans resulting in extreme bottlenecks for most of the spe cies ( Taberlet et al., 1998 ; Petit et al., 2002 ) Similarly, common patterns of genetic diversity across herbaceous and woody plant species from western North America, ranging from Alaska to California, suggest shared refugia and recolonizatio n routes during and following Pleistocene glaciation (reviewed in Soltis et al., 1998). In eastern North America,
24 however, studies reveal more variance among species in their historical biogeography, certainly due to the fine mosaic of geological and ecolo gical patterns. This has resulted in current high levels of both endemism and widespread distributions in the region. An extensive review of the literature on phylogeographical patterns in the southeast by Soltis et al. (2006) emphasizes several common bar riers in the distribution of lineages: the Atlantic vs. Gulf coast distribution ( Saunders, Kessler, and Avi se, 1986 ; Avise and Nelson, 1989 ; Gurgel, Fredericq, and Norris, 2004 ) the Apalachicola river Tombigbee river ( Avise et al ., 1986 ; Oliveria et al., 2007 ) the Apalachicola river Chattahoochae river ( Mylecraine et al., 2004 ) and the Mississippi river ( Al Rabab'ah and Williams, 2002 ) Some species exhibited a more complex pattern resulting from three different glacial refu gia, one to the east of the Mississippi and two to the west ( Leache and Mulcahy ; Burbrink, Lawson, and Slowinski, 2000 ) All these patterns are thought to be the result of the repetitive glacial cycles throughout the Pleistocene ( Delcourt et al., 1985 ) Within this region, one of the more complex areas is one of the refugia, penin sular Florida. Central Florida has an especially high level of endemism and has been the focus of my studies. The Central Florida scrub consists of an area of approximately 150 km by 15 km, a long strip of dune systems dominated by the older LWR and Moun t Dora, with smaller and younger ridges nearby. More than 20 federally listed plant species and over 40 animal species are endemic to the scrub, with an equal number of narrow endemics with only slightly broader ranges ( Abrahamson, 1984a ; Christman and Judd, 1990 ) This makes the Florida scrub one of the three biodiversity hotspots in the United St ates, rivaling the California Floristic Province and the Hawaiian islands ( Do bson et al., 1997 )
25 This unique habitat is located on both the coastal and central parts of the peninsula. The focus of this study is the inland scrub, corresponding to past shorelines from the Miocene, Pliocene, or early Pleistocene, when sea level was higher and the peninsula was smaller, perhaps even isolated from the continent ( Webb and Myers, 1990 ) The scrub is characterized by well drained and nutrient poor soils, hot and wet summers, and dry and mild winters ( Abrahamson, 1984a ) Intense fires maintain the sand scrub with a recurrence frequency of 5 30 years depending on the type of scrub ( Menges, 2007 ) However, this scrub habitat is highly threate ned, and already an estimated >85 % of it has been lost to agriculture (especially citrus) and estate development ( Weekley, Menges, and Pickert, 2008 ) The suppression of fires also presents a threat to the scrub ende mics, resulting in the formation of oak dominated xeric hammock ( Myers, 1985 ; Menges et al., 1993 ; Menges and Haw kes, 1998 ) Surprisingly, the historical biogeogr aphy of this unique system has seldom been studied and to our knowledge, no comparative analyses have retrace d the biogeography of this region. S everal hypotheses have been proposed to explain the geographical origins of the central Florida scrub ( Berry, 1916 ; Pirkle and Yoho, 1970 ; Watts, 1975 ; Morgan, 1988 ; Myers and Myers, 1990 ; Webb and Myers, 1990 ) The first scenario (western hypothesis) rida between 5 and 2 Ma. By 2.5 Ma, a semi arid biota extended from Florida into the western U.S. ( Myers and Myers, 1990 ) In the mid Pliocene, the scrub of Florida became isolated from the xeric habitats of west ern North America due to increased humidity along the Gulf of Mexico ( Webb and Myers, 1990 ) potentially resulting in the origin of many scrub
26 species. The Florida scrub jay, Aphelocoma coerulescens being most closely related to the western A. ultramarine, illustrates this western hypothesis ( Peterson, 1992 ; MacDonald et al., 1999 ; Peterson, Martinez Meyer, and Gonzalez Salaz ar, 2004 ) How ever, Sceloporus woodii the Florida scrub lizard, was thought to be closely related to S. virgatus and S. undulatus from western North America but in light of molecular studies was found to be sister to S. undulatus undulatus a widespread relative from the northern U.S. ( Jackson, 1973 ; Smith HM, 1992 ; Clark, Bowen, and Branch, 1999 ; Miles et al., 2002 ; Wiens and Donoghue, 2004 ) Most of the inferences of close relationships bet ween Florida scrub endemics and w estern relatives have been based on morpholog ical similarities that might be convergent adaptations to the xeric environments found in Texas, Florida and the western U.S. rather than a reflection of recent common ancestry ( Shaw a nd Small, 2004 ) The second scenario (eastern hypothesis) suggests that the central Florida scrub arose at t he end of the last glaciation, about 10,000 years ago, as species migrated into Florida ahead of the advancing ice sheet Some populations became adapted to the xeric substrate present on the ridges and remained there while other populations migrated back n orth, leading to divergence between the Florida and northern populations and eventually speciation ( Watts, 1975 ) In many species, range expansion occurred without speciation, although speciation may have accompanied northward colonization in some cases. For those species that recolonized via founder events may continue to serve as center s of genetic diversity, as observed for species of
27 Chamaecyparis, Dicerandra, Eryngium a nd Polygonella ( Lewis and Cra wford, 1995 ; MacDonald and Hamrick, 1996 ; Mylecraine et al., 2004 ) A third scenario argues for a much older origin of the scrub endemics. As early as 20 Ma, the oldest sediments of Florida, from Thomas Fa rm ( Pr att, 1990 ) and Alum Bluff ( Berry, 1916 ; Morgan, 1988 ) show that environmental conditions were similar to Florida date back to only 9 Ma, and many not until the Pliocene (5 3 Ma) so this hypothesis seems unlikely and will not be considered here. The f irst two scenarios which we will consider, are both examples of vicariance the process by which some populations become isolated from others leading to speciat ion In a phylogenetic context, v icariance translates into two reciprocally monophyletic group s representing two geographical regions or here where the endemic and its sister species occur ( Avise et al., 1987 ; Knowles and Maddison, 2002 ) In the context of the western hypothesis, I would predict that the central Flori da scrub endemic would be sister to species occurring in the xeric regions of Texas and/or the western U.S. (origin between 5 and 2 My). The eastern hypothesis (origin about 10,000 ya) would predict sister relationships with eastern North American taxa. W h en reconstructing a phylogeny encompassing several populations for multiple species, a lack of reciprocal monophyly would reject the vicariance hypothesis and argue in favor of several long distance dispersal events ( Knowles and Maddison, 2002 ) or the formation of a paraphyletic metaspecies around the more geographically restricted one P seudocongruence ( Cunningham and Collins, 1994 ) can also occur but can be tested for in several ways ( Soltis et al., 2006 ) In the case of pseudocongruence, where the
28 phylogenies are congruent in their topologies, but not in their divergence time s careful calibration is important ( Avise, 2000 ; Donoghue and Moore, 2003 ) However, due to lack of pertinent fossils in the region we cannot test for incongruence in the timing of the events that shaped the phylogeography o f plants in central Florida. In order to test both hypotheses on the geographical origin of the central Florida scrub as well as the processes that might have been involved, our stu dy focuses on four angiosperm taxa endemic to the central Florida scrub habitat. Their different habits and life histories ensure that they are representative of the di versity found in this ecosystem. Prunus geniculata Harp. (Rosaceae), the scrub plum, is a federally endangered scrub species known from only 21 sites, all of them on the LWR ( Harper, 1911 ; USFWS, 1999a ) The scrub plum is a shrub up to 2 m tall, heavily branched, with strongly zigzag twigs and spiny lateral branches. Its deciduous leaves are finely toothed. The five petalled white flowers appear in late winter, when the plant is leafless, and the fruit is a small, bitter, red drupe ( Wunderlin, 1998 ) The amount of flowering and fruiting heavily depends on fire frequency. Prunus geniculata is andromonoecious, with both male and bisexual flowers on the same plant. Due to the strong fragrance of the flowers, it is pollinated by a variety of insects. It is believed to be self incompatible ( Weekley et al., 2010 ) Although assum ed from morphology to be closely related to Prunus texana and P. angustifolia a molecular phylogenetic analysis placed it as sister to Prunus mari tim a from the northeastern U.S. ( Shaw and Small, 2004 2005 ) although the authors admit doubts of this relationship because of limited taxon sampling.
29 Polygala lewtonii Small (Polygalaceae), also federa lly listed, is endemic to the LWR and Ocala National Forest on Mount Dora ( Small, 1898 ; USFWS, 2010 ) and is found on yellow sands in sandhills and scrubs. This short lived perennial, with several annual stems up to 20 cm tall, has dark pink flower s that form racemes. Plants flower in the spring, and although capable of self pollination, flowers are visited by a variety of insects ( Weekley and Brothers, 2006 ) The seeds are dispersed by ants. Interestingly, this species distinguishes itself by the presence of underground cleistogamous flowers that are obligate self ers It is heavily dependent on frequent f ires for reproductive success and growth. Polygala lewtonii is thought to be most closely related to P. polygama widespread in the southeastern U.S. and P. crenata endemic to northern Florida ( Abbott, 2009 ) although no published phylogeny has included more than one accession for P. lewtonii Persea humilis Nash (Lauraceae), the scrub or silk bay, is endemic to the Florida scrub. This small tree has shiny dark green leaves and a very typical dense red pubescence on the abaxial side. This species is of particular interest due to the recent arrival of laurel wilt ( caused by the eastern Asian Redbay Ambrosia Beetle Xyleborus glabratus Eichhoff) in central F lor ida, and the threat i t represents for a species of such restricted range ( Gramling, 2010 ) The continental Nor th American clade of Persea is composed of only three native species, P. humilis P. palustris and P. borbonia (the la tter two widespread in the southeastern U.S. ). No published work so far has included all three species in a phylogeny ( Rohwer, 2000 ; Chanderbali, van der Werff, and Renner, 2001 ; Chen et al., 2009 )
30 Ilex opaca Aiton var. arenicola (Ashe) Ashe (Aquifoliaceae) is endemic to the Florida scrub, a lthough not abundant when found. This variety forms small trees with yellow green spiny revolute leaves on rigid ascending branches. This dioecious species is pollinated by bees, and its seeds are dispersed by birds and other animals in the fall and wint er. A phylogeny places it and its nearest congener, Ilex opaca var. opaca (widespread in the eastern U.S. ), sister to I. argentina and I. cassine but with only one accession per species ( Gottlieb, Gib erti, and Poggio, 2005 ) Ilex opaca var. arenicola and Ilex opaca var. opaca have been treated as either two varieties of the same species or two separate species ( Brizicky, 1964 ; Wunderlin, 1998 ) Comparative phylogeographic analyses of these four species is employed to test the two competing hypothesis of the origin of the central Florida scrub. In this study, we: 1) enhance current phylogenetic knowledge for each species focusing on the endemic and putative relatives 2) statistically test both hypotheses on the geographica l origin of the central Florida scrub, and 3) infer the process by which the endemics differentiated from their sister species, testing the vicariance hypothesis. The congruence of historical biogeography among these four species, and others from the liter ature demonstrates major trends that have led to the formation of this unique but fast disappearing ecosystem. Finally, the impact of our results on our understanding of the processes of community assembly as well as the conservation implications are di scussed. Materials and Methods Sample Collections Study species were selected partially based on the existence of relatively good molecular phylogenies for the genera. However, the Florida endemics were often omitted from or only sampled from one specime n in published phylogenetic studies
31 W e obtained data from authors and/or GenBank, and re analyzed the data sets after adding our own sequence data for the endemics. We sequenced the internal transcribed spacer of the nuclear ribosomal DNA (ITS) for the f ollowing species: Persea humilis Persea borbonia Persea palustris Polygala lewtonii Polygala polygama Prunus geniculata Prunus maritima var. maritima Ilex opaca var. arenicola Ilex opaca var. opaca and Ilex cassine For each species, several populations were sequenced to represent the breadth and diversity of their natural distribution in the wild. For all species, fresh leaf material was collected in the field and stored in silica gel. For the two federally listed plants, Polygala lewtonii an d Prunus geniculata a harvesting permit was issued by the Florida Department of Agriculture and Consumer Services, Division of Plant Industry (permit #714). Populations were chosen according to availability of permits and representativeness of the distrib ution s of the species. When available, at least two populations of the congener s were selected from within Florida. All samples used are listed in supplementary table 1, with both GenBank accessions and voucher information. DNA E xtractions A mplification and S equencing DNA was extracted from all specimens of Ilex Prunus and Polygala following a modified CTAB DNA extraction protocol ( Doyle and Doyle, 1987 ; Cullings, 1992 ) For Persea the protocol yielded mucilage, and an extraction kit (Invitrogen ChargeSwitch gDNA Plant Kit, Life Technologies, Carlsbad, CA) was required to yield DNA of good quality For all genera except Persea primers ITS1 and ITS4 (White et al., 1990) were used for PCR amplification. For these species 1 (~10 100 ng) was added to 49 (200 mM Tris HCl (pH 8.4), 500 mM KCl)
32 2.5 mM MgCl 2 1 M betaine, 0.8 mM dNTPs, 0.5 each primer, and 2.5 units of Taq polymerase PCR conditions were as follows: 95C for 2 min, 5 cycles of 95C for 1 min, 53C for 1 min and 72C for 2 min, 40 cycles of 95C for 1 min, 48C for 1 min and 72C for 2 min, then a 12 min hold at 72C. The product was cooled to and stored at, 4C. For Persea LAUR1 and ITSB were used following other analyses of Lauraceae phylogeny ( Blattner, 1999 ; Chanderbali, van der Werff, and Renner, 2001 ) ( T able 1). In the case of Persea and Prunus ITS alone did not provide the resolution needed, and the chloroplast spacers atpB rbcL, psbC trnS, rpl32 trnL, trnC ycf6, trnH psbA, trnL F and trnS G were sequenced for all three North Ameri can Persea species, and rpl16, trnS trnG trnL intron, trnL F and trnS G were sequenced for the Prunus accessions (Table 1). PCR components were: 0.5 1 x buffer, 2 5 mM MgCl 2 0. 2 Taq polymerase in a 25 solution min. All PCR products were verified on a 1.2% agarose gel for a strong single band and then sequenced at the Interdisciplinary Center for Biotechnolog y Research at the University of Florida using an ABI 3 7 30 DNA sequencer (Applied Biosystems, Carlsbad, CA ). Each product was sequenced with both primers to confirm results. Data M atrices, S equence A lignment and P hylogeny R eco nstruction For all phylogenies, published sequences were obtained from authors of published phylogenies and from GenBank in order to co mplement our own set of sequences (Supplementary Table 1). For Prunus aligned chloroplast sequences were obtained from J. Shaw ( Shaw and Small, 2004 ) and GenBank ( Bortiri et al., 2001 ; Lee and Wen,
33 2001 ; Bortiri and Potter, unpublis hed ) in order to complete our data set of ITS, rpl16 trnL intron, trnL F and trnS G Prunus domestica and P. tomentosa were used as outgroups ( Shaw and Small, 2004 ) For Polygala our own sequences of ITS were added to those of J. R. Abbott from a larger phyloge netic analysis of Polygalaceae and Moutabea Atroxima and Carpolobia were chosen as outgroups ( Abbott et al., in prep ) For Persea, ITS sequences were downloaded fr om GenBank ( Rohwer, 2000 ; Chanderbali, van der Werff, and Renner, 2001 ; Chen et al., 2009 ; Garcia Chavez et al., unpublished ) and ITS and chloroplast regions were generated ( atpB rbcL, psbC trnS, rpl32 trnL, trnC ycf6, trnH psbA, trnL F and trnS G ) for all three North American species and one P. americana accession as an outgroup Finally, sequences of Ilex ITS were obtained from A. Gottlieb ( Gottlieb, Giberti, and Poggio, 2005 ) and added to sequences from GenBank ( Manen et al., 2010 ; Landherr and Higgins, unpublished ) and Ilex cornuta was used as the outgroup Phylogenetic analyses were conducted separately for each of the four genera. Sequences were asse mbled and aligned in Geneious 5.1.7 ( Drummond et al., 2010 ) using MAFFT ( Katoh, Misawa, and Kuma, 2002 ; Kuma and Toh, 2005 ) followed by manual adjust ment t o maximize positional homology. M aximum likelihood analysis was performed in R A xML 7.2.8 ( Stama takis, 2006 ) using the GTRGAMMA model with 10000 bootstrap replicates of a maximum likelihood search with partitioning of data sets into nuclear and chloroplast partitions for both Persea and Prunus These trees were imported into FigTree v.1.3.1 ( Rambaut, 2009 ) and exported in a NEXUS format in order to use PAUP 4.0a114 x86 macosx ( Swofford and Sinauer, 2003 ) for hypothesis testi ng.
34 H ypothesis T esting Following the reconstruction of the phylogenies, we formally tested each scenario on the origin of the central Florida scrub against our data. For each genus, several trees were constrained, placing the Florida endemic as sister to a species that would support either the western or the eastern hypothesis. For example, for Prunus ( Figure 1 ) the western hypothesis would place Prunus geniculata sister to Prunus texana ( Figure 1.a), while the eastern hypothesis would place it sister to P. maritima ( Figure 1.b). For each biogeographic scenario, one or more putat ive sister species were tested. C orresponding tree topologies were manually constrained using MacClade 4.08 OS X ( Maddison and Maddison, 2000 ) to adjust the placement of the Florida scrub endemic In order to statistically test the two hypotheses, the simulated trees were tested against the most likely tree from our data set using the Approximately Unbiased (AU) ( Shimodaira, 2002 ) and the Shimodaira Hasegawa (SH) ( Shimodaira and Hasegawa, 1999 ) tests implemented in PAUP *. We used a R esampling of the Estimated Log Likelihoods (RELL) with 100,000 replicates to test whether each hypothesis could be rejected as b eing identical to the best tree from the likelihood analysis. These tests use a bootstrapping (multi scale bootstrapping in the case of AU) method on the site likelihoods of the different trees to produce a P value. The resulting P value for each hypothesi s was considered significant if less than 0.05. In the cases where more than one sister species had to be tested for a given hypothesis, a Bonferroni correction was applied to the P value. For Prunus we tested the w estern hypothesis with P. texana and P. subcordata as sister to the Florida endemic, and the e astern hypothesis with P. maritima and P. americana Prunus angustifolia was also tested because it has some morphological
35 similarities to the Florida endemic and occurs in Florida For Polygala the w estern hypothesis was tested with P. scoparia P. myrtifolia and P. californica separately as sist er to the endemic, and for the e astern hypothes i s, P. polygama P. incarnata and P. mariana were tested separately. Polygala crenata was also tested, although it does not reflect one particular hypothesis but appears in the trichotomy with Polygala lewtonii and P. polygama For Persea P. palustri s and P. borbonia were tested separately as sister to the endemic, encompassing all three North American Persea spec ies. Finally, no species of Ilex are native to the western U.S. (the only species occurring naturally in the wild are recent establishments of Asian species). Three eastern species, Ilex opaca var. opaca I. cassine and I. myrtifolia were all tested separately as sister to I. opaca var. arenicola the central Florida scrub endemic. Results Phylogenetic A nalyses Sequences of ITS (including 5.8S) varied in length, from 679 bp for Polygala, to 632 bp for Prunus from 500 bp to 684 bp fo r Ilex and 587 bp on average for Persea For Prunus ( Figure 2), the combined ITS + chloroplast tree has a topology somewhat different from that of Shaw et al (2004). A well supported clade (99% bootstrap value (BS) ) contains P. geniculata P. maritima va r. gravesii and some accessions of P maritima var. maritima This coincides with results of the chloroplast only analysis of ( Shaw and Small, 2005 ) However, not all the P. geniculata accessions (identifi ed by the 3 or 4 digits following the name) form a monophyletic group, nor do all of the accessions of P. maritima var. maritima Although one accession of the latter forms a clade with P. geniculata three others form a clade (86% BS) with P. nigra P. umbellata and P. angustifolia
36 T he broad relationships within the large genus Polygala will not be discussed in this paper ( Abbott, 2009 ; Abbott et al., in prep ) Within the clade of interest containing all species occurring in the southeastern U.S. P. lewtonii P. crenata and P. polygama each form s a monophyletic group, all included in a well supported (99% BS) clade. However, the relationships among the three s pecies are unresolved ( Figure 3). For Persea ( Figure 4), the tree combining ITS and chloroplast data places P. borbonia P. palustris and the endemic P. humilis together in a n unresolved trichotomy. Within this clade, the two accession s of P. palustris form a clade (87% BS). However, neither P. borbonia nor P. humilis appears to be monophyl etic F or Ilex ( Figure 5), a clade encompassing I. myrtifolia I. cassine and both varieties of I. opaca shows strong support (96% BS). Within this clade, I. myrtifo lia forms a clade with I. cassine (94% BS), and I. opaca var. opaca forms a clade with the scrub endemic I. opaca var. arenicola (75% BS). However, within those two clades, none of the species are monophyl etic Biogeographic H ypotheses Analyses of two of the species support the eastern hypothesis, and none support the western one (all results of all tests are shown in Table 2 2). For Prunus both eastern (AU p=0.0000; SH p=0.0058 0.0155) and western (AU p=0.0000; SH p=0.0045 0.0056) hypotheses are rejected AU results clearly reject all scenarios, but SH values weakly support P. texana and P. americana being sister to P. geniculata, each respectively representing the eastern and the western hypothesis. These P values are still significantly smaller than 0.0 5. For Polygala the results are very clear: the only hypothe sis that cannot be rejected is the eastern one, with the endemic Polygala lewtonii sister to the e astern
37 widespread P. polygama (AU p=0.4255; SH p=0.8450) or P. crenata (AU p=0.5179; SH p=0.9632). All other hypotheses are rejected at 99%. For Persea neither hypothes i s of the endemic P. humilis being sister to P. borbonia (AU p=0.1399; SH p=0.2826) or to P. palustris (AU p=0.0558; SH p=0.2595) can be rejected. The latt er two species have similar southeastern distribution s but the western limit of their natural distribution reaches Texas. Even though their distributions reflect the eastern hypothesis more than the western one, the rejection of either hypothes i s is there fore not possible. Finally, for Ilex both hypotheses are rejected with AU (p=0.0000) However, with SH, which is usually more conservative ( Shimodaira, 2002 ) the Ilex opaca var. arenicola / Ilex opaca var. opaca hypothesis cannot be rejected ( p=0.353 6 ) Ilex opaca var. opaca has a widespread e astern distribution and these results therefore support the e astern hypothesis. D iscussion Phylogenetic R elationships of the F ocal S pecies All the phylogenies presented here show improved understanding from pr eviously published work even if they sometimes reveal more complexity than previously thought For Prunus the most recently published phylogeny found P. geniculata to be sister to P. maritima ( Shaw and Small, 2005 ) Previously P. geniculata had been considered closely related to P. texana due to their many morphological similarities, or to P. angustifolia due to the resemblance of their flowers ( Harper, 1911 ; Shaw and Small, 2005 ) The addition of our accessions show s several inconsistencies with the Shaw and Small results. First, the multiple accessions of P. geniculata do not form a clade. Instead, the two accessions of P. maritima from the Shaw and Small study form a clade
38 that is part of a trichotomy with P. geniculata acce ssion 106 and a clade with the three other accessions of P. geniculata Additionally our own accessions of P maritima var. maritima were placed in a clade sister to a clade of P. nigra P. umbellata and P. angustifolia The monophyly of both P. geniculata and P. maritima need to be reevaluated with addition al accessions and sequence data Prunus texana however, falls outside of these two clades, making the western hypothesis less likely than the eastern one For Polygala the addition of severa l accessions of P. lewtonii and P. polygama to a larger data set (Abbott et al ., 2009) shows a very well supported clade encompassing P. polygama P. lewtonii and P. crenata all strongly reciprocally monophyletic. This is consistent with previous findings for the group ( Abbott, 2009 ) Polygala crenata is a rare species found in the panhandle of Florida, along the G ulf coast and into eastern Texas. Because the rest of the tree is well resolved, the trichotomy of P. lewtonii P. polygama and P. crenata suggests recent and/or rapid phylogenetic divergence Additional loci either chloroplast spacers or nuclear gene introns might be able to resolve this trichotomy. P ersea borbonia P. palustris and P. humilis formed a trichotomy that could not be resolved and is best explained by a recent and rapid radiation of the lineage. P ersea palustris is the only species for which all accessions form a clade ; however, we have only two accessions of P. palustris Despite very similar morphology between P. borbonia and P. palustris both of which are distinct from P. humilis the latter species is sometimes considered a subspecies of P. borbonia ( Wunderlin, 19 98 ) The phylogeny
39 fails to reject this hypothesis Additional sequence data, or a population level study, should be considered to shed light on the relationship s among th e se three species. T he Ilex phylogeny refines our current understanding of this c omplex genus. Ilex arenicola was first described as a separate species by Ashe ( Ashe, 1924 ) who then merged it with Ilex opaca resulting in two varieties ( Ashe, 1925 ) According to our findings, Ilex opaca var. arenicola and Ilex opaca var. opaca are not reciprocally monophyletic, suggesting that the endemic variety might not be a valid taxonomic entity. More sequence data are needed to evaluate these relationships further The Ilex opaca clade is itself sister to another clade that enc ompasses two species, I. cassine and I. myrtifolia These have similar distributions and were historically considered a single entit y ( Sargent, 1889 ) The placement of I. myrtifolia within a clade of I. cassine However, the sampling of I. myrtifolia should be augmented and more sequen ce data should be added before any firm conclusions can be drawn regarding the distinctness of these two species. Species of Ilex are known to hybridize where distributions overlap ( Ashe, 1925 ; Setoguchi and Watanabe, 2000 ; Lee et al., 2006 ) and the placement of I. myrtifolia within I. cassine could possibly reflect this A population level study would be needed in the case of confirmed hybridization zones. Biogeography and the O rigin of the C entral Florida S crub The results of my phylogenetic tests support an eastern origin of the Florida scrub endemics for both Ilex and Polygala This hypothesis argues in favor of a g lacial refug ium in Florida for species that were previously widespread in the mesic environments of e astern North America. During the last glacial maximum (~10,000 ya) some of the populations dispersed to central Florida and became adapted to the more
40 xer ic environment. Once the ice sheet s retreated and the species migrated northward the scrub specialized populations stayed on the habitat island of Florida and those in the non scrub habitat moved back to the mesic eastern U.S With time, these populations diverged from those that moved northward through allopatric speciation 10,000 years is a very short time for full speciation to occur and so the lack of resolution of some phylogenies might be due to a lack of time for divergent populations to fully spec iate. In the case of Prunus our study cannot statistically support one hypothesis over another The sister relationship of Prunus maritima to P. geniculat a is rejected with both the AU or SH test T he placement of two P. maritima accessions within the Prunus geniculata clade calls for more work to clarify this relationship. The congruence test for the tree with Prunus geniculat a sister to P americana distributed in the eastern half of North America, is not significant nor is the constraint with P. te xana distributed in the w est as the sister taxon. Prunus texana however, falls into a much further clade from the Florida endemic than P. maritima I therefore think it is more likely that additional data will support the eastern hypothesis and reject t he western one. Similar to the findings for the Florida scrub lizard, Sceloporus woodii ( Clark, Bowen, and Branch, 1999 ; Miles et al., 2002 ) P geniculata has been hypothesized to be sister to a western relative, based on morphological similarities, but analyses of chloroplast data shifted this hypothesis with the closest relative being distributed in e astern North America ( Shaw and Small, 2005 ) Morphological similarities can be due to common ancestry, or to homoplasious adaptation to local conditions. The very small, dent ate lea ves, shrub habit and possibly pubescent fruits of both P texana and P. geniculata may be
41 adaptation s to their similar dry sandy environment s and not synapomorph ies At this point, the phylogenetic signal is not sufficient to distinguish among the hypoth eses. For Persea neither hypothes is can be rejected Additionally, both P. palustris and P. borbonia have similar distribution s in the southeastern U.S. extending into Texas preventing us from testing the two phylogeographic hypotheses. Lastly, Ilex opa ca var. arenicola shows no phylogenetic distinction from the widespread southeastern I opaca var. opaca Despite the lack of statistical significance for the eastern hypothesis using the AU test the SH test shows a rejection of all hypotheses except that of Ilex opaca var. opaca/Ilex opaca var. arenicola being closest relatives. In view of the phylogenetic results, we conclude that Ilex opaca var. arenicola and Ilex opaca var. opaca should not be considered separate taxonomic entities. This com plex phylogeny can be interpreted in several ways in terms of biogeography Ilex opaca var. arenicola could be a separate entity, but resulting from very recent events. Also, some Ilex opaca var. opaca populations could just happen to grow in more xeric en vironments and exhibit some different morphological traits as a result of plasticity. Ilex opaca is a morphologically variable species ( Little, 1971 ) ; although the morphological differences from I. opaca var. arenicola are quite dramatic with Ilex opaca var. arenicola having much smaller, very spiny leaves of a lighter green color and a highly revolute margin when compared to Ilex opaca var. opaca ( Wunderlin, 1998 ) Finally, in situ speciation during the Pleistocene glaciation s as for Polygala could have given ri se to I. opaca var. arenicola and subsequent gene exchange could mean that these two entities have not had the necessary time to fully speciate and become reciprocally
42 monophyletic. In either case, the western hypothesis for the origin of Ilex opaca var. arenicola is not supported by molecular evidence. Evidence for two of four species supports an eastern origin of the central Florida scrub endemics. Previous studies for Chamaecyparis, Ery n gium and Polygonella have found similar results of an eastern orig in with a glacial refug ium explanation ( Lewis and Crawford, 1995 ; MacDonald and Hamrick, 1996 ; Mylecraine et al., 2004 ) For Pinus clausa the once held belief that there was a Florida refug ium was not supported ; instead, the refug ium was found to be in an ice free northern part of North America, with subsequent colonization of Florida, accompanied by speciation by isolation of the P. clausa endemic ( Parker et al., 1997 ) Similarly, Ceratiola ericoides is thought to have expanded its range during the Pleistocene gla cial maxima, resulting in two northern populations. A subsequent long distance dispersal event from Florida to coastal Georgia post date s the last glaciation ( Trapnell et al., 2007 ) These last two species illustrate the potentially complex histories of the Florid a scrub endemics relative to their sister species, more complex th a n either of the hypotheses we have tested here. Notably, both studies shed light on complex histories with the help of population level sampling, and of calibration of phylogenies (see belo w ). Further research on our species of focus at the population level will most likely shed light on some of the questions that remain on the origins and processes that have shaped the central Florida scrub biota. Implications for C ommunity A ssembly and C on servation Examining community structure as a means to elucidate the different scenarios of community assembly is not a new exercise ( Sutherland, 1974 ; Drake, 1991 ) However, testing different scenarios in a statistically sound manner can improve our
43 understanding of community structure, and the role of colonization history vs. present ecological interactions governing this community ( Baldwin and Robichaux, 1995 ) Despite the fact that my study only encompassed four species from a targeted community, the molecular evidence presented here gives us a glance at one of the forces that have shaped the central Florida scrub. It is important for community ecology to incorp orate historical, systematic and biogeographical information ( Ricklefs, 1987 ) For each of the clades we examined the present geographical distribution of species is determined by four main characteristics: their ancestral ecological niche, the geographical starting poi nt for dispersal, the biotic and abiotic limitations to their current dispersal ability and their opportunities for niche evolution ( Wiens and Donoghue, 2004 ) If we consider Polygala and Ilex the two endemics may have evolved when abiotic factors forced them to disperse into new niches in Florida most likely during the Pleistocene The ecologically adapted traits therefore evolved in situ reinforcing the idea that the central Florida s crub endemic s should not be expected to be morphologically similar to their congeners. For Ilex the more complex evolutionary history of the scrub endemic could well be due to it being an easily hybridizing species. It is important to note that due to a lack of fossils in the region, associated with the absence of timelines in the published material, we do not have time estimations on the speciation events. We make the assumption that the eastern hypothesis, which was not rejected for all but one of the s pecies, took place during the Pleistocene glaciations. However, rigorous timing of the phylogenies is needed to support this interpretation or potentially reveal a more complex history that would have led to the same topology and inferences of community a ssembly ( Ackerly, 2004 )
44 Conclusions The different phylogenies successfully place three of the initial four Florida scrub endemics in well supported clades identifying one or a few close relatives, using these results to statistically test scenarios of origin of the central Florida scrub. These tend to converge towards the scenario of an eastern North American origin, most likely accompanied by in situ divergence during the Pleistocene glacia l maxima when Florida served a s a refug ium for more temperate species. This scenario h as also been observed in numerous other species endemic to the central Florida scrub, in far more cases than species with western relatives. However, some stories are more complex than either an easte rn or western origin, especially when data are obtained at the population level ( Trapnell et al., 2007 ) T he implications of one period of migration and in situ evolution of ecologically important traits is positive for the ability of the community to potentially adapt to changing conditions. An emerging field in biology is using phylogenetics to understand and predict long term community dynamics ( Willis et al., 2008 ) ecosystem processes ( Cadotte, Cardinale, and Oakley, 2008 ) and response of ecosystems to global change ( Edwards, Soltis, and Soltis, 2006 ) Increasing rates of environmental change, may it be climate, invasions or urbanization mean that it is more important than ever to understand the causes of community structure in concluding the origin of a community in order to understand how species will respond to change ( Cavender Bares, Kozak, and Fine, 2009 ) Additional species endemic to the highly threatened central Florida scrub urgently need to be studied in order to better grasp some of the complexities of the history and
45 formation of this region F urther studies at the population level will also shed light on some of the questions that could not be answered through phylogenetic analysis Table 2 1. List of primers used for all species in this study Taxon Gene region Primer name and sequence Reference Polygala ITS ITS1: (TCCGTAGGTGAACCTGCGG) ( White et al., 1990 ) ITS4: (TCCTCCGCTTATTGATATGC) ( White et al., 1990 ) Ilex ITS ITS1: (TCCGTAGGTGAACCTGCGG) ( Wh ite et al., 1990 ) ITS4: (TCCTCCGCTTATTGATATGC) ( White et al., 1990 ) Prunus rpl16 71F: (GCTATGCTTAGTGTGTGACTCGTTG) ( Jordan, Courtney, and Neigel, 1996 ) 1661R: (CGTACCCATATTTTTCCACCACGAC) ( Jordan, Courtney, and Neigel, 1996 ) trnL F C: (CGAAATCGGTAGACGCTACG) ( Taberlet et al., 1991 ) F: (ATTTGAACTGGTGACACGAG) ( Taberlet et al., 1991 ) trnS G trnG3' UUC: (GTAGCGGGAATCGAACCCGCATC) ( Shaw et al., 2005 ) trnS GCU: (AGATAGGGATTCGAACCCTCGGT) ( Shaw et al., 2005 ) ITS 17SE: (ACGAATTCATGGTCCGGTGAAGTGTTCG) ( Sun et al., 1994 ) 26SE: (TAGAATTCCCCGGTTCGCTCGCCGTTAC) ( Sun et al., 1994 ) Persea atpB rbcL F: (ACATCKARTACKGGACCAATAA) ( Chiang, Schaal, and Peng, 1998 ) R: (AACACCAGCTTTRAATCCAA) ( Chiang, Schaal, and Peng, 1998 ) psbC trnS psbC: (GGTCGTGACCAAGAAACCAC) ( Demesure, Sodzi, and Petit, 1995 ) trnS: (GGTTCGAATCCCTCTCTCTC) ( Demesure, Sodzi, and Petit, 1995 ) r pl32 trnL trnL(UAG): (CTGCTTCCTAAGAGCAGCGT) ( Shaw et al., 2007 ) rpL32 F: (CAGTTCCAAAAAAACGTACTTC) ( Shaw et al., 2007 ) trnC ycf6 ycf6R: (GCCCAAGCRAGACTTACTATATCCAT) ( Shaw et al., 2005 ) trnC (GCA): (CCAGTTCAAATCTGGGTGTC) ( Demesure, Sodzi, and Petit, 1995 ) trnH psbA F: (TGATCCACTTGGCTACATCCGCC) ( Xu et al., 2000 ) R: (GCTAACCTTGGTATGGAAGT) ( Xu et al., 2000 ) trnL F C: (CGAAATCGGTAGACGCTACG) ( Taberlet et al., 1991 ) F: (ATTTGAACTGGTGACACGAG) ( Taberlet et al., 1991 ) trnS G trnG3' UUC: (GTAGCGGGAATCGAACCCGCATC) ( Shaw et al., 2005 ) trnS GCU: (AGATAGGGATTCGAACCCTCGGT) ( Shaw et al., 2005 ) ITS LAUR1: (ACCACCACCGGCAACCA) ( Chanderbali, van der Werff, and Renner, 2001 ) ITSB: (CTTTTCCTCCGCTTATTGATATG) ( B lattner, 1999 )
46 Table 2 2. Results of the AU tests on different phylogeographical hypotheses for all 4 central Florida endemic species. AU results are given by a P value. shows significance at 0.05. This means that the tree of the hypothesis is differ ent from the consensus tree (null hypothesis: trees are the same). Genus Clades hypothesized Hypothesis AU P value SH P value Conclusion Prunus P. geniculata/P. americana Eastern 0* 0.0155* Reject P. geniculata/P. maritima Eastern 0* 0.0058* Reject P. geniculata/P. subcordata Western 0* 0.0045* Reject P. geniculata/P. texana Western 0* 0.0109* Reject P. geniculata/P. angustifolia 0* 0.0056* Reject Polygala P. lewtonii/P. polygama Eastern 0.4255 0.8450 Fail to reject P. lewtonii/P. crenata 0.5179 0.9632 Fail to reject P. lewtonii/P. incarnata Eastern 0* 0* Reject P. lewtonii/P. mariana Eastern 0* 0* Reject P. lewtonii/P. scoparia Western 0* 0* Reject P. lewtonii/P. myrtifolia Western 0* 0* Reject P. lewtonii/P. californica Western 0* 0* Reject Persea P. humilis/P. borbonia 0.1399 0.2826 Fail to reject P. humilia/P. palustris 0.0558 0.2595 Fail to reject Ilex I. opaca var. arenicola/I. opaca var. opaca Eastern 0* 0.3534 Fail to reject I. opaca var. arenicola/I. cassine Eastern 0* 0.0059* Reject I. opaca var. arenicola/I. myrtifolia Eastern 0* 0.0059* Reject
47 Table 2 3. All accessions used, herbarium voucher number and location, sample location, reference, and all GenBank accessions (accessions with are novel and were produced for this study herbarium voucher location reference GenBank accessions Ilex phylogeny ITS I. opaca opaca 1201 I. opaca opaca 202 I. opaca opaca 120 CGA FL I. opaca arenicola 108 CGA FL I. opaca ( Landherr and Higgins, unpublished ) AF20059 0 I. opaca opaca 1401 I. opaca arenicola 116 CGA FL I. opaca opaca 1504 I opaca opaca 1001 CGA FL I. opaca opaca 802 I. opaca opaca 801 I. opaca arenicla 106 CGA FL I. opaca arenicola 121 CGA FL I. opaca opaca 1204 I. cassine (Manen) ( Manen, Boulter, and Naciri Graven, 2002 ) AJ49266 7 I. cassine (Loizeau) ( Manen et al., 2010 ) FJ39466 4 I. myrtifolia 501 I. cassine 702 I. cassine (Landherr) ( Landherr and Higgins, unpublished ) AF20058 8 I. cassine 107 FL
48 Table 2 3. Continued. I. vomitoria ( Gottlieb, Giberti, and Poggio, 2005 ) AF17462 5 I. crenata FL ( Gottlieb, Giberti, and Poggio, 2005 ) AH0071 44 I. glabra ( Gottlieb, Giberti, and P oggio, 2005 ) AJ27534 2 I. cornuta ( Landherr and Higgins, unpublished ) AF20059 1 Prunus phylogeny ITS rpl16 trnS trnG trnL intron trnL F P. unbellata JSh774 003 (TENN) FL ( Lee and Wen, 2001 ; Shaw and Small, 2004 ) AF17949 3 AY5006 47 AY5007 09 AY5007 52 AY500771 P. nigra JSh979 (TENN) VT ( Shaw & Small, 2004 ) AY5006 48 AY5007 10 AY5007 53 AY500772 P. angustifolia JSh785 (TENN) GA ( Shaw & Small, 2004 ) AY5006 44 AY5007 06 AY5007 49 AY500768 P. maritima maritima 902 DE P. maritima maritima 903 DE P. maritima maritima 901 DE P. geniculata JSh989 (TENN) FL ( Shaw and Small, 2004 ) AY5006 51 AY5007 13 AY5007 56 AY500775 P. geniculata 107 CGA FL P. genicualta 105 CGA FL P. maritima maritima JSh877 045 (TENN) MA ( Shaw a nd Small, 2004 ) AY5006 52 AY5007 14 AY5007 57 AY500776 P. maritima gravesii Conn Greenhouse (TENN) ( Shaw and Small, 2004 ) AY5006 53 AY5007 15 AY5007 58 AY500777 P. geniculata 106 CGA FL P. americana JSh038 (TENN) TN ( Shaw and Small, 2004 ) AY5006 38 AY5007 00 AY5007 43 AY500762
49 Table 2 3. Continued. P. texana JSh924 077 (TENN) TX ( Shaw and Small, 2004 ) AY5006 54 AY5007 16 AY5007 59 AY500778 P. subcordata J.Syring (TENN) CA ( Shaw and Small, 2004 ) AY5006 55 AY5007 17 AY5007 60 AY500779 P. domestica DPRU 350 (TENN) ( Shaw and Small, 2004 ) AY5006 59 AY5007 21 P. tomentosa DPRU 2316.4 (TENN) ( Shaw and Small, 2004 ) AY5006 67 AY5007 29 Polygala phylogeny ITS P. polygama 1206 MA P. polygama 101 CGA FL P. polygama (JH) JH P. polygama (JRA) P. polygama ( R) P. polygama 601 WI P. polygama 602 WI P. polygama 402 IL P. polygama 401 IL P. lewtonii 105 CGA FL P. lewtonii 102 CGA FL P. lewtonii 107 CGA FL P. lewtonii (R1) P. lewtonii (R2) P. crenanta ( R) P. crenata (JRA1) P. crenata (JRA2) P. incarnata P. setacea
50 Table 2 3. Continued. P. mariana P. smallii P. galapageia P. scoporia P. vulgaris P. myrtifolia P. californica Carpolobia Atroxima Moutabea Persea phylogeny ITS atpB rbcL psbC trnS rpl32 trnL trnC ycf6 trnH psbA trnL F trnS G P. palustris 112 CGA P. palustris 1401 PA P. borbonia 104 CGA P. borbonia 805 LM MI P. humilis 103 CGA P. humilis 107 CGA P. borbonia 113 CGA P. americana
51 Figure 2 1. Phylogeographical hypotheses for the Prunus complex. Figure 1a (left) represents the western hypothesis, and Figure 1b (right) the eastern hypothesis. These trees were generated in MacClade 4.08 OS X ( Maddison and Maddison, 2000 )
52 Figure 2 2. Prunus chloroplast + ITS combined maximum likelihood analysis with bootstrap values. Accessions are indicated by 3 or 4 numbers, and the central Florida scrub endemic is in bold. Figure 2 3. Polygala ITS maximum likelihood analysis with bootstrap values. Accessions are indicated by 3 or 4 numbers, and the central Florida scrub endemic i s in bold.
53 Figure 2 4. Persea plastid and ITS combined maximum likelihood analysis with bootstrap values. Accessions are indicated by 3 or 4 numbers, and the central Florida scrub endemic is in bold. Figure 2 5 Ilex ITS maximum likelihood analysis with bootstrap values. Accessions are indicated by 3 or 4 numbers, and the central Florida scrub endemic is in bold.
54 CHAPTER 3 IS MICROSATELLITE FR AGMENT LENGTH VARIAT ION THE BEST MARKER FOR POPULATION LEVEL STUDIES? THE C ASE OF POLYGALA LEWT ONII Genetic diversity is one of the three forms of biodiversity recognized by the International Union for the Conservation of Nature (IUCN). It is linked to the fitness of a species, as well as its capacity to adapt to future environmental changes ( Reed and Frankham, 2003 ) Particularly in the case of endangered species, or endemics from threatened habitats, it is especially useful to consider genetic diversity and partitioning for conservation management plans ( Ellstrand and Elam, 1993 ; Petit, El Mousadik, and Pons, 1998 ; Crandall et al., 2000 ; Frankham, 2003 ; Reed and Fran kham, 2003 ; Frankham, 2005 ) Microsatellites and F lanking R egions T he genetic study of plants at the inter or intraspecific level has often been hampered by a lack of molecular markers. Plant cpDNA evolves more slowly than the popular animal mtDNA and is often not sufficiently variable for final scale analyses of plant population diversity R ecombining a utosomal DNA encompasses a bigger pool of genetic variation and information to reconstruct coalescence and recent evoluti onary history. M icrosatellites, or more generally simple sequence repeats (SSRs), have become popular due to their abundance, high pol ymorphism caused by repeat length variation, their co dominance and Mendelian inheritance ( Morgante and Olivieri, 1993 ) But microsatellites, despite all their advantages still present some problems ( Selkoe and Toonen, 2006 ) Sever al studies have examined the different biases in the practice of microsatellite genotyping, namely our sole reliance on fragment length to score these loci despite the complexity of the composition of the repeats ( Culver, Menotti R aymond,
55 and O'Brien, 2001 ; Ramakrishnan and Mountain, 2004 ; Anmarkrud et al., 2008 ) and the influence of flanking region polymorphism on results ( Blankenship, May, and Hedgecock, 2002 ; Selkoe and Toonen, 2006 ; Vali et al., 2008 ) In general, the extent of homoplasy among alleles of the same population, different po pulations of the same species, and between closely related species is still not well understood and its effect has seldom been evaluated. Homoplasy occurs when an identical fragment length or repeat number results from different evolutionary histories. It has been shown in some studies to lead to an underestimation of population subdivision and genetic dive rgence between populations or species ( Goldstein et al., 1995 ; Hedrick, 1999 ; Makova, Nekrutenko, and Baker, 2000 ; Selkoe and Toonen, 2006 ) caused by an inflation of within population polymorphism a nd allelic diversity. In the case of particularly fast evolving loci, the tendency can invert itself and inflate levels of gene flow between populations ( Rouss et, 1996 ; Epperson, 2005 ; Selkoe and Toonen, 2006 ) Other studies estimate that these biases are not significant, especially for populations effective population size ( Estoup, Jarne, and Cornuet, 2002 ) or in the case of interrupted and/or compound microsatellites ( Adams, Brown, a nd Hamilton, 2004 ) One of the ways to decrease the problem of homoplasy in a data set is to sequence the flanking regions of the microsatellite. The flanking regi on is the non repeat nucleotides on either side of the SSR tandem repeat, for about 100 150 bp on each side of the SSR repeat region, or 200 300 bp total. Nucleotide substitution in these regions is believed to be more slowly evolving than repeat number ev olution of the microsatellite itself, but the nucleotide substitution occurring in non coding regions might
5 6 still exhibit some degree of polymorphism, and in some cases significantly so ( Vali et al., 2008 ) Being able to identify different haplotypes in the non repeat flanking sequences enables us to identify homoplasious alleles in microsat ellite variation ( Blankenship, May, and Hedgecock, 2002 ; Vali et al., 2008 ) More recently, the variability in the flanking regions has been the target of some low level phylogenetic studies, and proved useful for reso lving relationships among closely related species ( Zardoya et al., 1996 ; Rossetto, McNally, and Henry, 2002 ; Chatrou et al., 2009 ) At the intraspecific level, very few studies have used microsatellite flanking regions, but those that hav e obtained resolution ( Grimaldi and Crouau Roy, 1997 ; Waters and Wallis, 2000 ; Mogg et al., 2002 ; Won et al., 2 005 ; Ablett, Hill, and Henry, 2006 ) and other s have found the insertions/deletions (indels) in the flanking regions to be a source of homoplasy ( Matsuoka et al., 2002 ) A few studies have been investigating this problem in non model animal species, but none have been conducted to date on plants ( Hey et al., 2 004 ) I investigated the genetic variation in microsatellite flanking regions in a non model species from an endangered ecosystem and compared my findings with analyses of fragment length (as would be used in a straightforward population level study usi ng microsatellites) and strict tandem repeat numbers (inferred from sequence data). I also assessed the genetic variation of this endangered endemic to assess conservation priorities. Genetic diversity in one species has increasingly been recognized as a direct indicator of ecosystem health, the ability of a species to face present and future evolutionary pressures, and high diversity of other associated
57 species ( Whitham et al., 2003 ; Vellend and Geber, 2005 ; Mitchell Olds, Willis, and Goldstein, 2007 ; Wade, 2007 ) The C entral Florida S crub The Central Florida scrub consists of an area of approximately 150 km by 15 km, a long strip of dune systems dominated by the older Lake Wales Ridge and Mount Dora with smaller and younger ridges nearby. More than 2 0 federally listed plants and over 40 animals are endemic to the scrub, and as many are narrow endemics with only slightly broader ranges ( Christman, 1990 ) This unique and fast disappearing habitat is located on both the coastal and central part s of the peninsula. We are focusing on the inland scrub the broader of which corresponds to past shorelines from the Miocene, Pliocene, and e arly Pleistocene, when sea level s were higher and the peninsula was smaller perhaps even isolated from the continent ( Webb, 1990 ) The scrub is characterized by well drained and nutrient poor soils, hot and wet summers and dry and mild winters ( Abrahamson, 1984b ) Intense fires maintain the sand scrub with a recurrence frequency of 20 60 years ( Menges, 2007 ) The suppression of these fires in scrubs from south c entral Florida has resulted in an unusual domination of Quercus species and Carya floridana showing the fragi lity of this system ( Myers, 1985 ; Menges et al., 1993 ) Several species from the Lake Wales Ridge have been studied genetically and found to exhibit low genetic diversity ( Menges, 2001 ; Weekley, Kubisiak, and Race, 2002 ; Gitzendanner et al., 2011 ) and limited gene flow ( Dolan et al., 1999 ) while others encompass higher diversity than expected ( MacDonald and Hamrick, 1996 ; Menges, 2001 )
58 Some studies suggest that the oldest and highest ridges, the Lake Wales and Mount Dora Ridges, are the centers of origin of some species. The mole skink is inferred to have originated from both Mount Dora and the Lake Wales Ridge ( where the genetic diversity is highest ) but with a separation of the two lineages by 4 Ma ( Branch et al., 2003 ) One recurrent trend is the lack of isolation by distance among populations in the Florida scrub species. Many have higher gene flow between the northern and southern Lake Wales Ridge than the geographically closer northern Lake Wales Ridge and Mount Dora ( MacDonald and Hamrick, 1996 ; Clark, Bowen, and Branch, 1999 ; MacDonald et al., 1999 ) The endemic Nolina brittoniana and Warea carteri however, reveal a peninsula r effect, with a strong association between population location on a n orth s outh axis and a cline in allele frequency ( Evans et al., 2000 ) Polygala lewtonii Polygala lewtonii yellow sand predominantly in sandhill, but also in scrubby areas and on transitional habitats between high pine and turkey oak barrens ( USFWS, 1999c ) This short lived herb (2 10 years) depends on fire for seedling recruitment and survival ( Weekley and Menges, submitted ) In Lake County, FL, a survey in 1995 counted 48 occurrences of P. lewtonii in the high pine system of the Ocala National Forest, but most o f the habitat is gone today ( Slapcinsky et al., 2005 ) The latest review of the species records 49 occurrences in 2009, but no improvement in the management of these sites ( USFWS, 2010 ) The distribution of Polygala lewtonii seems to be generally declining on The Nature S 1999). A general trend of population recruitment two years after a prescribed burn confirms the requirement of fire for seed germination ( Weekley and Menges, submitted ) Even in populations that had
59 disappeared completely, plants reappear a short time after a burn. This seems to show a dormant seed bank that can be activated with heat, dryness or other factors associated with fires ( Slapcinsky, Gordon, and progam, 2003 ; Slapcinsky et al., 2005 ) Polygala lewtonii is an amphicarpic species that has both below surface or ground level cleistogamous (closed obligat ely selfing) flowers and chasmogamous (open pollinated) flowers occurring on above ground terminal racemes ( Wunderlin, 1 998 ) This mode of reproduction is found in only a few dozen species in the world and combined with its limited distribution and declining populations makes this species a prime candidate for further investigation. y similar to the widespread P. polygama the common milkwort ( FNAI, 2000b ) Although P. lewtonii was first described by Blake in 1924, C. James was the first to recognize the differences among P. polygama P. crenata (previously P. polygama forma obovata ) and P. lewtonii in 1957. The habitats of P. lewtonii and P. polygama are distinct with the former found in the sandhills and the latter in low wet pine lands. Their range seems to overlap only in Lake County ( P. lew tonii se ems to be extirpated in Orange C ounty where it was last recorded in 1957) ( James, 1957 ) In a phylogenetic analysis using ITS, Polygala lewtonii P. crenata and P. pol y gama form an unresolved trichotomy, arguing for a history of refugia in Florida during the Pleistocene glacial cycles ( Abbott et al., in prep ; Germain Aubrey et al., in prep c ) Here I analyze Polygala lewtonii and its widespread close congener P. polygama for rigorous comparison of species genetic measures ( Gitzendanner and
60 Soltis, 2000 ) and to compare the effects of using microsatellites and their flanking region sequences to estimate population genetic parameters. Material and Methods Plant C ollections I collected 148 and 118 individuals for Polygala lewtonii and P. polygama respectively, from 10 and 8 populations encompassing the range of each species (T able 3 1, Figures 3 1 and 3 2 for location and details of vouchers). Because P. lewtonii is a federally listed species, a permit was obtained from the Division of Plant Industry for each location (permits #714 and 954). I choose individuals within a location to cover as o, because the species can reproduce with cleistogamous flowers occurring on underground rhizomes, I was careful not to collect individuals too close to each other. For each individual, leaves were dried and stored in silica gel and taken back to the lab. DNA was extracted following a CTAB DNA extraction protocol ( Doyle and Doyle, 1987 ) Population on t he Nature Conservancy Saddle Blanket (108) Population 108, in The Nature Conservancy (TNC) Saddle Blanket, was labeled as P. lewtonii when collections were made although there were doubts about its identity. The morphology of the plant is somewhat intermediate betwe en P. lewtonii and P. polygama but looking more like P. polygama It was unexpected at so far south in Florida, and the measurement of seed sizes resulted in a range that overlapped sizes for P. lewtonii and P. polygama I checked for polyploidy of the in dividuals by comparing their genome size with those of other P. lewtonii populations. I then used flanking region polymorphism and genetic clustering with TESS (see below) to infer the status of this population.
61 Development and A mplification of M icrosatel lite P rimers A microsatellite library was constructed using the voucher for population 107 and biotinylated probes (CA) 8 and (GA) 8 The protocol used was the same as that which has proven successful in our laboratory for approximately 10 other species som e from the same ecosystem ( Edwards, Soltis, and Soltis, 2007 ; Edwards et al., 2007 ; Lopez Vinyallonga et al., 2010 ; Germain Aubrey et al., 2011 ) Because we are here interested in studying variation in both repeat number of the microsatellite and sequences of the flanking regions, we designed the primers to encompass a longer flanking region (about 300 bp) on end of the repeat, designing the other primer close to the other end of the repeat. We checked if the primers amplified a comparable region in the sister species, P. polygama and if not, we redesigned the primers. All loci were amplified in all samples of both Polygala lewtonii and P. polygama in a 10 olution containing 1x buffer, 1 M Betaine, 1.5 mM MgCl 2 FAM VIC NED PET labeled M13 primer, and 0.2 unit Taq polymerase. PCR amplification conditions were 3 min at 95C, followed b y 35 cycles of 45 sec at 95 C, 1 min 15 sec at 52C, and 1 min 15 sec at 72C, with a final step of 20 min at 72C. PCR products were stored at 4C. PCR products were pooled in a NED/PET:VIC/FAM 2:1 ratio and then genotyped on an ABI 3730 DNA analyzer (Ap plied Biosystems, Carlsbad, CA USA ) at the Interdisciplinary Center for Biotechnology Research at the University of Florida. Microsatel lites were scored automatically and then checked by eye using Genemapper 1.6 (Soft Genetics, State College, PA, USA). The same PCR conditions were used to amplify loci destined for sequencing, replacing the biotiny lated fluorescent dye with an unlabeled M13 primer. PCR products
62 were sequenced on an ABI 3130 DNA sequencer from the long flanking region towards the repeat. S equences were visualized and edited using Geneious Pro 5.3.4 ( Drummond et al., 2010 ) In the case of clear and unambiguous sequences, the flanking regions were considered homozygous and any fragment length difference was attributed to a difference in repeat num ber. Any ambiguous sequences were discarded, and those with more than two or three ambi guous sites were cloned and resequenced assigning any fragment length difference to either an insertion/deletion event in the flanking region or a difference in repeat number between the two alleles. Sequences at four loci were edited manually and aligned in Geneious Pro 5.3.4 ( Drummond et al., 2010 ) Characterization of M arkers Using the program A RLEQUIN ( Excoffier, Laval, and Schneider, 2005 ) each locus was characterized by the number of alleles per population, the effective number of alleles, the observed and expected heterozygosity, and a Hardy Weinberg Equilibrium test. All microsatellite data were imported in Excel and format t ed for Genalex 6.41 ( Peakall and Smouse, 2006 ) for analysi s and format conversion towards other software packages. Micro Checker 2.2.3 ( van Oosterhout et al., 2004 ) was used to test potential genotyping errors, selection biases of loci and the presence of null alleles. SpaGeDi 1.3a ( Hardy and Vekemans, 2002 ) was used to p er form a randomization test with 10,000 permutations of R ST on alleles for each locus separately and averaged over all loci in order to find the best fitting evolution ary model for these loci. Linkage disequilibrium was tested using GENEPOP 4. 1 ( Raymond and Rousset, 1995 ; Rousset, 2008 ) and because Polygala is a perennial with a few to dozens of annual stems,
63 GENODIVE 2.0b17 ( Meirmans and Van Tienderen, 2004 ) was used to detect th e presence of clones in the data set. Sequences were imported, edited and stored in Geneious Pro 5.3.4 ( Drummond et al., 2010 ) and Genalex 6.41 ( Peakall and Smouse, 2006 ) which was also used for formatting the files for different software packages. Comparison of M ark ers for M olecular D iversity I ndices For all analyses, both the fragment length and the repeat numbers were used in order to compare them. Genalex 6.41 ( Peakall and Smouse, 2006 ) was us ed to infer observed and expected heterozygosity ( H O and H E ) mean number of individuals per population ( N E ), mean number of alleles per locus per population (both absolute ( N a ) and effective ( N eff )), and the inbreeding coefficient F Also, an exact Hardy Weinberg Equilibrium test was implemented with a 1,000,000 generation Markov chain (including a 100,000 burn in) in ARLEQUIN 3.1 ( Excoffier, Laval, and Schneider, 2005 ) All statistics were averaged over all loci for each population for an analys is of Polygala lewtonii alone, and averaged over all populations within each species for a comparison of the endemic and its widespread sister species. Bottleneck D etection The exploration of past bottlenecks and gene flow between populations or sets of p opulations can also help us to better understand the genetic variation and the conservation needs of a particular species. The program BOTTLENECK ( Cornuet and Luikart, 1996 ) detects an excess of heterozygosity i n comparison to the allele frequency reduction following a reduction in the effective number of individuals. It does so by simulating the coalescent process under the infinite allele s model (IAM), the stepwise mutation model (SMM), and a mixed model I set at 90% SMM and allowed for 10% of
64 IAM variance ( Luikart et al., 1998 ) The detection of past bottlenecks is especially important for the Florida scrub to test the impact of anthropogenic and natural events on demography and genetic diversity Population S tructure To determine the number and composition of clusters of indiv iduals, we used TESS 2.3.1 ( Durand et al., 2009 ) which infers the number of clusters (K) based on allelic identity and frequency. It uses a Bay esian likelihood method to minimize departure of populations from Hardy Weinberg equilibrium, in the same way as STRUCTURE ( Pritchard, Stephens, and Donnelly, 2000 ; Falush, Stephens, and Pritchard, 2003 2007 ) The DIC (De viance Information Criterion) stabilizes around the optimal number of clusters ( K ). However, in contrast to STRUCTURE, TESS includes decay of correlation of membership coefficients with distance within clusters. This might lead to a slower stabilization of K and Kmax (with the highest probability) might overestimate the true number of clusters. As advised in the manual, the analysis of the true number of clusters has to be assessed on the basis of DIC and bar plots. In determining the number of clusters, we therefore took into consideration both the curve of DIC against K and the stabilization of the number of clusters in the bar plots. After exploratory runs, I ran 100,000 generations encompassing 10,000 burn in, with admixture, for 10 replicates per K =2 6 for both P olygala species, and K =2 8 for Polygala lewtonii I examined the different runs separately to verify the stability of the clusters over all runs. I then permutated all runs for the same K them in CLUMPP ( Jakobsson and Rosenberg, 2007 ) and subsequently visualized in DISTRUCT ( Rosenberg, 2004 )
65 F ST and R ST AMOVA analyses with 9,999 permutations were implemented within individuals, among individuals within populations, among populations within species and between species through a permutation test that will enable us to infer gene flow between populations and ridges R ST is a statistic that corre sponds to the evolution model of the loci, while F ST is a commonly used measure of population structure that is less variable than R ST but less adapted to the stepwise mutation model ( Hutchinson and Templeton, 1999 ; Hardy, 2003 ) Using Genalex 6.41 ( Peakall and Smouse, 2006 ) a Mantel test was also implemented to test for geographical and genetic correlations throughout the range of P lewtonii and also along each ridge. The correlation between geographical and genetic distances, and its statistical significance, were examined, along with the distribution of the pairwise points on the graph: a more clustered distribution of the points around a flat or non significant line corresponds to a lack of isolation by distance due to strong gene flow between populations, while a more scattered distribution of the individual points is the sign of strong genetic drift in the species ( Hutchinson and Templeton, 1999 ) Lastly, I evaluated the contribution of each population of Polygala lewt onii relative to other populations in the data set to enable the most careful planning of conservation management for this species using the program CONTRIB ( Petit, El Mousadik, and Pons, 1998 ) This allelic contribution is then broken down into its contribution to total allelic diversity and its divergence from other populations. Sequence A nalysis ARLEQUIN 3.1 ( Excoffier, Laval, and Schneider, 2005 ) was used to infer the number of haplotypes, heterozygosity and nucleotide diversity within each population, as well as an F ST AMOVA analysis with 9,999 permutations, comparable to the one
66 conducted for the microsatellite data sets. Also, the same analysis of allelic contribution was done for the sequence data set using CONTRIB ( Petit, El Mousadik, and Pons, 1998 ) Results Characterization of M arkers Ten microsatellites were developed, 8 amplified easily for fragment length, and 5 were selected for sequencing (T able 3 2). Of the latter five loci, two were found to be linked ( Poly D 1 2 and Poly E01 ). I kept both for the flanking region sequences as they were concatenated for analyses, but only used PolyE01 for the fragment and repeat fragment length, as is used in most microsatellite studies. Then, from the sequences, I separated the repeat numbers (also a frequency based data set) from the microsatellite flanking region sequences, ending up with three different data sets. No scoring error or null alleles we re detected for any of the loci, and the R ST permutation tests on the Polygala lewtonii data set revealed that all loci followed a stepwise mutation model (both fragment and repeats data sets) after a sequential Bonferroni correction ( Rice, 1989 ) The average number of alleles pe r population varied from 1.22 to 3.33 for the fragment lengths while the repeats had a similar range of variation of 1.22 and 4.77. The effective number of alleles per population varied more for the repeats (0.97 3.44) than for the fragments (1.01 1.87). F or both data sets, the expected heterozygosity (0.01 0.41 for fragments and 0.26 0.69 for repeats) was significantly higher than the observed (0.09 0.11 and 0.00 0.06 for fragments and repeats, respectively). Only one
67 locus ( Poly49 ) was in Hardy Weinberg e quilibrium, as might be expected for a rare endemic plant with small population sizes ( Table 3 3). Five loci were successfully sequenced for flanking regions for most of the samples. PolyB08 (100 P. lewtonii and 65 P. polygama ) is 211 bp long with 3 nucle otide substitutions; PolyD12 (66 P. lewtonii and 49 P. polygama ) is 280 bp long with 5 substitution sites, 2 deletions and one 4 bp indel; PolyE01 (48 P. lewtonii and 51 P. polygama ) is 367 bp long with five substitutions; Poly1 (108 P. lewtonii and 61 P. polygama ) is 222 long with 7 substitutions and 2 deletions; and Poly5 (135 P. lewtonii and 87 P. polygama ) is 169 bp long with 3 nucleotide substitutions, 1 deletion, and a 30 bp insertion inside which are 1 substitution and 1 deletion (Table 3 4). Compa rison of M arkers for M olecular D iversity I ndices The number of individuals per population in the sequence data set was at least as high as for fragments and repeats, although because all loci were concatenated, this did not take missing data into account t he same way as for frequency based data (Table 3 5). Also, when accounting for the number of polymorphic sites, missing data that might have occurred in the data set increased polymorphism as it was considered a deletion event. However, the number of haplo types per population was higher than the number of effective alleles for both fragments and repeats (4 7), and H E also was consistently higher (Table 3 5). Hardy Weinberg e quilibrium Departure from Hardy Weinberg equilibrium differed among data sets. The populations generally exhibited small differences in the number of significant, monomorphic and/or non significant loci (apart from population 118 which was not significant for fragment length but had 2 of 4 loci highly significant in the repeat data
68 set). However, on close examination of the details of the three shared loci between the two data sets, 15 inferences of 21 (three loci for 9 populations of Polygala lewtonii ) were different. The proportion of monomorphic populations for a locus was consistent f rom one marker to the next, with about 41% of monomorphic populations at any one locus. However, significantly more populations were non significant when using fragment length (~18%) than repeats (~8%) (Table 3 6). Overall g enetic d iversity i ndices The nu mber of alleles differs, with more alleles per population for the endemic P. lewtonii with repeats and fewer in the widespread congener P. polygama (Table 3 7). The effective number of alleles follows the same trend. The expected heterozygosity differs sig nificantly between fragments and repeats (0.199 and 0.213 for fragments and 0.335 and 0.062 for repeats for P. lewtonii and P. polygama respectively), as does the observed heterozygosity. Lastly, F the inbreeding coefficient ( Weir and Cockerham, 1984 ) also showed significant differences, the repeat data set giving much higher estimates of inbreeding than the fragment data set. Bottleneck d etection The Wilcoxon test performed using the IAM, the SMM, and the Two Phased Model (TPM) allowing 10% of infinite allele model like mutations also showed some discrepancies between the fragment and repeats data sets. Using fragment lengths, populations 115 (Catf ish Creek), 124 and 125 (both in the Ocala National Forest on the Mount Dora Ridge) were detected to have gone through a bottleneck under the SMM and the TPM, and 102 (Lake Wales Ridge State Forest), 107 (TNC Tiger Creek) and 126 (Ocala National Forest) we re detected to have gone through a bottleneck, but only
69 under the SMM model. Using repeats only, there was no detection of any population having gone through a bottleneck under any microsatellite mutation model (Table 3 8). Population s tructure Population 108 T he results of TESS for continued analysis of Polygala lewtonii and P. polygama and for P. lewtonii alone both showed that the repeats did not detect as much structure in the populations as the fragment lengths did for K =4. But when looking at the b ar plots for K =2 and K =3 we can see that with both data sets, 108 clusters with all other P. polygama populations to form a solid cluster (population 108 was therefore excluded from other analyses encompassing P. lewtonii only). Visualization of plots for K =2 for each run separately shows consistency in the clustering of population 108 with P. polygama (results not shown). The results of the flow cytometer do not show any difference in the nucleus size of any populations of P. lewtonii with population 108 (results not shown). All mutations that are specific to P. polygama also encompass population 108 (Table 3 4). Polygala lewtonii Within Polygala lewtonii the Deviance Index Criterion (DIC) curves detect an optimal number of clusters of K =3 for both the fragment length and repeats data sets. However, when examining the repeats data set bar plots, there does not seem to be any structure at all, even for K =2. The fragment lengths data set shows a dichotomy between the populations present on the Lake Wales R idge and the populations on Mount Dora (124 127) in the K K =3 bar plot exhibits a purple group containing populations 105 (Clermont), 107 (TNC Tiger Creek) and part of 102 (Lake Wales Ridge State Forest). Howeve r, this group does not hold when increasing the number of clusters to 4, most likely showing that the third purple
70 cluster is an artifact of the algorithm and not a real separate genetic group. K =2 is therefore the optimal number of clusters for the fragme nt data set. From the analyses of molecular variance based on F ST the flanking region sequence data set shows that most of the variance is due to the difference between species, while the repeats data set attributes most of the variance to differences among individuals within populations. The fragment lengths data set shows nearly equal variance between components for species, among populations and among individuals within populations (Figure 3 5). negative but for one or two populations for each marker type, always including population 118 (Catfish Creek). Since this population is only composed of four individuals, it could be disproportionately rich by chance when compared to larger populations. After removing this population from the data sets, I found that the fragment lengths data set includes population 102 (Lake Wales Ridge State Forest) as being solely positive. Removing this one also reveals that the rest of the populations all contribute to the allelic diversity, with populations 107 (Tiger Creek), 115 (Catfish Creek), 124 and 126 (both in Ocala National Forest) contributing more than others. Repeats reveal populations 102, 115 and 127 (Ocala National Forest) as contributing mostly to the species genetic diversity. Sequences, at firs t identifying only population 118 as having a positive allelic diversity index, show 115 and 127 as positive (Figure 3 5). Genetic D iversity of Polygala lewtonii The number of individual plants per population was consistently lower for the repeats data se t than for the fragment data set due to the way we sequenced flanking
71 regions and inferred repeat number from them. The number of alleles per population is comparable with 2.192 alleles per population on average for fragment lengths and 2.074 for repeats. The number of effective alleles also exhibits similar values (1.39 for fragments and 1.48 for repeats), revealing that the fragment length data set encompasses more differences in the allele frequencies within each population than the repeats data set does Generally H O was much lower than H E leading to high values of the fixation index F (0.244 0.574 for fragments and 0.362 1 for repeats, except for population 126) (Table 3 5). Reinforcing these data of high inbreeding, the results of the Hardy Weinberg e quilibrium tests showed that most populations are not in Hardy Weinberg equilibrium at most loci (Table 3 6), even after a sequential Bonferroni correction ( Rice, 1989 ) The fragment data set showed most of the populations to have heterozygosity deficiencies under the SMM model ( the appropriate model based on the R ST permutation test for all loci), and several under the TPM model (Table 3 8). A heterozygote deficiency in a population is, with the Wilcoxon test, interpreted as the past presence of a genetic bottleneck that the popu lation has not yet recovered from, or a sign of inbreeding significant enough to reduce the average heteroygosity. F ST and R ST permutation pairwise tests among populations of Polygala lewtonii reveal high gene flow values throughout both fragment and repe ats data sets, with slightly higher and all significant values between more proximal locations, such as the four populations from the Ocala National Forest (124 127) (Table 3 9). If we consider populations of P. polygama that are as far apart or less than the geographically most distant P lewtonii populations (12 population pairs), the average genetic distance value
72 is consistently higher than the average genetic distance between all population pairs of P. lewtonii (average P. lewtonii F ST / R ST =17.2%/19% and average P. polygama F ST / R ST =43.2%/32% for fragment lengths; for repeats in the same respective order, 6.7%/28.7% and 13.9%/30%). The results of the Mantel test within P lewtonii show slightly different results, with the fragment data set having a significant (p=0.00) but very small correlation (R 2 =0.02) while the repeats data set also gives a small, but non significant (p=0.079), correlation (R 2 =0.03). The contribution of the populations to the total allelic richness was positive for population 118 (Catfish Creek), and for two out of three data sets for populations 115 (Catfish Creek) and 127 (Ocala National Forest). Populations 115 and 118 both occur in the same state park, very close to one another (Figure 3 6). When comparing P le wtonii to its more widespread sister species, P. polygama most measures of genetic diversity are lower for the widespread species than the endemic (number of alleles, effective number of alleles, observed and expected heterozygosity), but their measures o f inbreeding ( F ) are comparable. This is true for both fragments and repeats data sets (Table 3 7). Discussion Markers for the S tudy of Polygala The 8 loci used for the amplification of microsatellite fragments show variation at the intrapopulation level and prove to be useful for the population level genetic study of both Polygala lewtonii and P. polygama Due to the close phylogenetic proximity of P. crenata to the species of focus ( Abbott et al., in prep ; Germain Aubrey et al., in prep c ) it is likely that these primers will amplify for this species, and possibly others in a
73 broader clade. The flanking regions amplified in this study should also prove useful for phylogenetic and phylogeographic studies of the genus due to their high variability among species, and to a lesser extent within species. Marker C omparisons and I mpli cations for the U se of M icrosatellite F ragment L engths Genotyping fragment length does not seem to accurately reflect variation in repeat number within the microsatellite. We can see here that homoplasy can be detected with the use of flanking region seque ncing and the discovery of substitutions (18 substitutions total, including 10 within Polygala polygama but none within P. lewtonii ) and even number indels (one of the two indels is variable within P. lewtonii ). Also, these indels can create homoplasy when for example, a repeat is shorter by two repeats of 2 bp each, but encompasses a 4 bp insertion in its flanking region. The fragment length is then the same, although the microsatellite itself is different. This situation appears in the data set. Single p oint indels, however, cannot reveal or create homoplasy in a data set encompassing 2 bp repeat SSRs (here, 6 single point indels are present in the data set, 4 vary within P. polygama and 1 within P. lewtonii ) (Table 3 4). However, Estoup (2002) estimates and/or smaller effective population size, as is the case for P lewtonii do not affect overall results. This might be due to the fact that recently diverged populations have had less time to accumulate mutations in their flanking regions than populations that diverged further back in time. Between Polygala lewtonii and P. polygama, the presence of a 30 bp insertion, most substitutions and one single base pair insertion in all P. poly gama populations does mean that risks of homoplasy are higher between species than within species. We
74 therefore recommend that a subset of all populations be sequenced when existing primer pairs are used to amplify in a closely related species. The design of primers as close to the repeats as possible would also minimize the risk of interpreting variation in flanking regions as differences in repeat numbers, even if they might be harder to optimize and could result in an increased risk of null alleles in th e data set. The presence of mutations in the flanking regions of microsatellite loci seems to have little effect on some results. The difference between the number of alleles and the effective number of alleles among populations of P. lewtonii is not aff ected by the presence of the flanking region in the fragment genotyped. The absolute values fall within the same range (Table 3 3). The same is true of the separation of populations of the two species, and the placement of the ambiguous population 108 in t he P polygama cluster, although it is more clear cut in the case of whole fragment lengths than repeat numbers (Figure 3 3). Most of the indices of molecular variance differed substantially among data sets, sometimes dramatically. The expected heterozygo sities, observed heterozygosities, and inbreeding coefficients ( F ) varied in terms of absolute values, and also in their relative values when comparing the two species, eventually leading to opposite conclusions on which species is the more diverse or more inbreeding (Table 3 6). The same is true of Hardy Weinberg equilibrium Chi square test results, with repeats detecting less departure from equilibrium than genotypes based on fragment lengths (Table 3 7). Similarly, the repeats data set did not detect any past bottlenecks while the fragment length data set did (Table 3 7). These conclusions are similar to those of other studies where polymorphism in flanking regions had a great influence on results and
75 conclusions ( Blankenship, May, and Hedgecock, 2002 ; Vali et al., 2008 ) even though the direction of the differences might not be the same as what I found here. For example, Epperson (2005) and Rousset (1996) found that allelic diversity is underestimated with fragment lengths compared with repeat number, and that gene flow is overestimated. In the case of Polygala allelic diversity from the absolute and effective number of alleles was higher with fragment lengths than repeat numbers. However, gene flow among P. lewton ii populations was lower with repeats than fragments. These discrepancies are useful if we consider that several authors say that the flanking regions evolve more slowly than, and independently of, mutations in repeat numbers, so that sequence polymorphis ms might reflect a more ancient history of the populations than repeat number variation ( Valdes, Slatkin, and Freimer, 1993 ; Blankenship, May, and Hedgecock, 2002 ; Matsuoka et al., 2002 ; Ablett, Hill, and Henry, 2006 ) In the case of Polygala the fact that the results of analyses of the fragment lengths data set seem to be somewhat intermediary to results from repeats only and flanking regio ns only (as for the AMOVA results) might confirm this hypothesis. In the case of the allelic contribution of the different P. lewtonii populations, the signal from the w ith results from the repeats only data set being different from the other two (Figure 3 5). It is therefore interesting to depict the historical vs. the more recent genetic signal and infer a more detailed population history. In view of these observations it seems more important than ever to include a detailed study of the microsatellite loci used in any population level genetic study. It would include the sequencing of a subset of the samples, and the comparison of the
76 resulting data sets to try to depic t more ancient from recent signal in the case of high polymorphism of the flanking regions. A careful design of microsatellite primers very close to the repeats could also avoid the problem altogether, but, in view of the high polymorphism in the flanking regions, would increase the risk of null alleles due to a mutation in the region of the primer site itself. The sequencing of the flanking regions could be particularly useful for population level studies within a species, and also for phylogeographic stud ies of a clade of closely related taxa. Genetic D iversity in Polygala lewtonii and its I mplications for the C onservation of this F ederally L isted S pecies Comparison of Polygala lewtonii and P. polygama In order to interpret general measures of genetic diversity, P. lewtonii can be compared to its widespread congener, P. polygama This approach accounts for the effects of shared ancestry on measures of genetic diversity and allows the differences in fast evolving marker loci to be interpreted as a consequence of the recent history of the species ( Gitzendanner and Soltis, 2000 ) Most indicators of diversity calculated here are higher for P lewtonii than P. polygama (Table 3 7). This goes against the general patterns of lower genetic diversity in narrow endemics than their more widespread relatives ( Hamrick et al., 1996 ; Dolan et al., 1999 ; Gitzendanner and Soltis, 2000 ) This could be due to one or more of several factors but cannot be due to life history differences between the species. The argument for narrow endemics to have lower ( Godt, 1997 ; Parker et al., 2001 ) or higher ( Lewis and Crawford, 1995 ; Trapnell et al., 2007 ) overall genetic diversity therefore has to be explained by recent historical e vents rather than life history. The argument for a Floridian Pleistocene refugium for P. lewtonii seems like a plausible one ( Webb and Myers, 1990 ; Germain Aubrey et al., in prep c )
77 just as advanced for Ceratiola ericoides which exhibits more diverse populations (significantly higher H E and percentage of polymorphic loci using allozymes) in the central Florida scrub than in more northern populations in Georgia and South Carolina ( Trapnell e t al., 2007 ) The same scenario could very well explain this difference in number of alleles, effective number of alleles, and expected and observed heterozygosity between P. lewtonii and P. polygama The endemic species was here during the glacial maxi ma while the now widespread species recolonized its natural distribution post Pleistocene. The leading edge effect means that the species lost alleles through a series of founder events as it spread northward, and now exhibits lower genetic diversity than the endemic, which remained in the refugium region. It should here be noted that because many populations of P. polygama were sampled in Pleistocene recolonization scenario could lead to the northern populations being genetically poorer, driving the average allele number and heterozygosity down. The inbreeding coefficient, however, is lower for the endemic when using fragments but higher when using repeats. Given tha t fragment lengths also exhibit variation in the flanking regions, this could be interpreted as increased inbreeding in the more recent history of the endemic, reflected in the number of repeats, but not as much in the ical signal. When comparing the levels of inbreeding found here with other published microsatellite studies on endemics of the central Florida scrub, the range seems similar that of the obligately outcrossing but highly clonal Ziziphus celata with an avera ge F per population of 0.77 ( Gitzendanner et al., 2011 ) but different from measures of inbreeding of narrowly endemic lizards from
78 the Lake Wales Ridge ( F ranged between 0.06 and 0.30 for three species of lizards and skinks) ( Schrey et al., 2011 ) In general, high levels of genetic diversity may help species deal with future environmental changes, but high levels of inbreeding might indicate that Polygal a lewtonii has not recently had sufficient gene flow between populations. This could lead to a decrease in genetic diversity in the future. Genetic diversity and partitioning within Polygala lewtonii Within the threatened endemic species, we will interpr et any significant discrepancy among flanking region sequences, fragment length and repeat number data sets as a reflection of the history of the species or population, with flanking regions evolving slightly more slowly than the repeat number mutations an d therefore offering a more historical perspective on the pure microsatellite alleles ( Blanken ship, May, and Hedgecock, 2002 ; Matsuoka et al., 2002 ; Ablett, Hill, and Henry, 2006 ) Within Polygala lewtonii populations cluster into two stable and genetically distinct groups. The first one encompasses all four populations on the Mount Dora ridge, and the second encompasses the Lake Wales Ridge populations (Figure 3 4). This cl ustering is reinforced by the high F ST and R ST values among the Ocala National Forest populations (124 127). Populations 115 and 118 also distinguish themselves by their isolation from each other and from other populations (Table 3 9). Other analyses also distinguish these two populations, as their co ntribution to total allelic diversity is consistently greater than any other population for both repeat number and flanking region sequence data sets (Figure 3 5). These populations contributing to allelic richness are considered important for the conserva tion of genetic diversity, sometimes even more so than other more classical indicators such as heterozygosity or absolute
79 number of alleles ( Petit, El M ousadik, and Pons, 1998 ) reflected by the repeats number data set, populations 102 (Lake Wales Ridge State Forest) and 107 (Tiger Creek) also play a crucial role in the genetic composition of the species, with population 10 7 being the only one contributing to total allelic richness due to its allelic departure from other populations (encompassing the greatest number of private alleles), also an important factor for conservation ( Petit, El Mousadik, and Pons, 1998 ) (Figure 3 6). The other distinct group of populations is present in the Ocala National Forest where four populations were sampled. The high levels of gene flo w among those populations and strong genetic clustering might reflect geographical proximity and definitely reflects a conservation unit to maintain. The Ocala National Forest is a prime site for conservation. This clustering, however, is not the reflectio n of a gene flow barrier between ridges. Polygala lewtonii does not seem to follow an isolation by distance model, with very low or insignificant Mantel test results, but an AMOVA analysis failed to detect any molecular variance due to the variability amon g ridges (0%, p=0.00; results not shown). These results differ from studies of other species endemic to the central Florida scrub ( Clark, Bowen, and Branch, 1999 ; Branch et al., 2003 ) Polygala lewtonii is a short lived species, and the s trong gene flow barrier between the two ridges might be the result of a more recent barrier that is not yet apparent in longer lived species. Lastly, a closer look at the Mantel test results shows the presence of strong genetic drift in P. lewtonii The l ack of correlation and/or significance between genetic and geographic distances, coupled with a scattering of the points throughout the graph, reveals the presence of drift in all populations ( Hutchinson and Templeton, 1999 )
80 (Figure 3 6). The fixation of populations is problematic, and conservation management should urgently be oriented to wards building or emphasizing corridors between populations. Conclusions For Polygala lewtonii the idea that high levels of genetic diversity are coupled with high levels of inbreeding can be interpreted by a recent limitation to gene flow that will s oon impact populations, and eventually the overall level of genetic diversity. Also, the two different types of flowers lead to different levels of inbreeding at different times, which might also explain the lag between the appearance of inbreeding and a l owered heterozygosity. Supporting this same idea is the discrepancy in bottleneck detection between the fragments and repeats data sets (Table 3 8) and AMOVA results (Figure 3 5). Fragment length signal, reflecting a more ancient history of the species, de tects several significant bottlenecks, maybe revealing past fluctuations in population sizes, possibly during Pleistocene glacial cycles. The more recent signal from the repeat numbers does not detect any bottleneck event in any populations, maybe reflecti ng a more stable recent history. However, the evidence for overall drift in the species means that limitations to gene flow, such as the development of human settlement in the past 100 years, has already had an impact on the species. Conservation plans sho uld therefore aim to maintain existing gene flow, and to remediate lower gene flow between some of the populations. An interesting analysis of our data would be to examine the relationship of genetic diversity in space with historical and current human pop ulation densities and anthropogenic activities that could impact gene flow and genetic diversity, as was done for other central Florida scrub species recently ( Menges et al., 2010 )
81 Within Polygala lewtonii the presence of strong genetic clusters, but also significant gene flow among populations, is encouraging and should be taken as a positive sign for the potential of this species for the long term. However, early signs of strong d epartures from Hardy Weinberg equilibrium, high inbreeding, overall drift and some low rates of gene flow between pairs of populations should be taken into consideration. This species is a good candidate for early preventative conservation action rather t han waiting until heterozygosity and the number of alleles has been reduced. Proper, informed and focused action will help maintain this diverse species in a fragmented habitat.
82 Table 3 1 Populations, Location, Number of plants collected and population voucher for all Polygala lewtonii and P. polygama. LWRSF= Lake Wales Ridge State Forest; TNC= The Nature Conservancy; ADBCCSP= Allen David Broussard Catfish Creek State Park; ONF= Ocala National Forest A voucher for pop 118 at ADBCCSP was estimated not viable for the size of the population, and the authorities on site did not permit any ONF pop voucher collections. Species Pop Location No Voucher (herbarium) P. lewtonii 102 LWRSF, Polk Co, FL 16 CGA 11 (FLAS) P. lewtonii 105 Clermont, Lake Co, FL 16 JRA 22703 (FLAS) P. lewtonii 107 TNC Tiger Creek, Polk Co, FL 16 JRA 22685 (FLAS) P. lewtonii/polygama 108 TNC Saddle Blanket, Polk, Co, FL 16 JRA 22690 (FLAS) P. lewtonii 115 ADBCCSP, Polk Co, FL 16 CGA 50 (FLAS) P. lewtonii 118 ADBCCSP, Polk Co, FL 4 P. lewtonii 124 ONF, Salt Spring, Marion Co. FL 16 P. lewtonii 125 ONF, Hu ghes Island, Marion Co. FL 16 P. lewtonii 126 ONF, Hu ghes Island, Marion Co. FL 16 P. lewtonii 127 ONF, Hu ghes Island, Marion Co. FL 16 P. polygama 102 Citrus Co, FL 13 JRA 24633 (FLAS) P. polygama 1301 Mansfield, Connecticut 8 P. polygama 1801 Anoka Co Minnesota 16 P. polygama 18 02 St Louis Co, Minnesota 16 P. polygama 601 Richland Co Wisconsin 15 P. polygama 602 Sauk So Wisconsin 16 P. polygama 805 Lauderdale Co Mississippi 10 P. polygama 301 Ashtabula Co Ohio 8
83 Table 3 2. List of markers develo ped for Polygala lewtonii The library was constructed from the population vouchered JRA22703 (FLAS). Note that Pole 45 and Pole 47 develop from the same markers, but they are not linked and their fragment lengths do not overlap. marker sequence annealing temp. Motif Genbank accession Poly1 Forward TGATGGAGCCAACACGAAG 52C (CA)8 Reverse TTGGAAGGGTGTCATCTCGT 52C PolyE01 Forward TGAGGATTGCACTTGATGCT 52C (CA)9 Reverse ATTTCCAGGAGCACAACACC 52C PolyB08 Forward CACATGCACCTACTGTTCAGG 52C (GT)5(GC)3 Reverse TGCAAGCATCTCCTGTAATCC 52C (GT)6(GA)16 Poly44 Forward GGCATACAAGCCAATTCAGC 52C (GT)6 Reverse CACAACACAAAGGCATCGAC 52C Poly45 + 47 Forward AAAGGCCAGCATACATCAGG 52C (GT)15(GA)8 Reverse AAGCGAGCAGTTTGACAGAT 52C Poly46 Forward GCACACGTTTCCAGTATTGC 52C (TG)9 Reverse AGCCATCAACTCCATTACCG 52C Poly49 Forward GCAGCATGGCAAACTTATCC 52C (TG)7 Reverse TGGATTGGCTTAGAGAACGTG 52C PolyD12 Forward GGGGCAATAATTCAGGCATA 52C Reverse TTGGGATCGGAGAACTGAAG 52C Poly5 Forward AACGAATCTAAGGAACTTGATG G 55C (CT)18(GT)13 Reverse TGGGATCCCGATGAGCTA 55C
84 Table 3 3. Characterization of Polygala markers for fragment length and repeat number (italic). Average over all populations (S.E.) reported. Marker Species N N a N e H O H E HWE P value Poly1 Fragment 14.22(1.299) 3.333(0.687) 1.819 (0.260) 0.071(0.033) 0.349(0.092) 0.00(0.00) Repeats 11.22(1.18) 2.333(0.373) 1.531(0.203) 0.067(0.036) 0.267(0.079) 0.00(0.00) PolyE01 Fragment 14.00(1.28) 2.556(0.444) 1.497(0.178) 0.029(0.015) 0.260(0.078) 0.00(0.00) Repeats 5.11(1.47) 1.222(0.364) 0.976(0.233) 0.010(0.010) 0.267(0.079) 0.00(0.00 ) PolyB08 Fragment 11.11(1.48) 3.111(0.484) 1.871(0.180) 0.064(0.027) 0.416(0.067) 0.00(0.00) Repeats 10.56(2.042) 2.333(0.553) 0.976(0.223) 0.016(0.011) 0.277(0.094) 0.00(0.00) Poly44 Fragment 13.46(1.28) 2.444(0.530) 1. 1802 (0.096) 0.069(0.031) 0.195(0.062) 0.00(0.00) Poly45 Fragment 14.22(1.29) 2.111(0.261) 1.679(0.149) 0.112(0.027) 0.364(0.060) 0.00(0.00) Poly46 Fragment 14.33(1.31) 1.556(0.242) 1.137(0.103) 0.021(0.011) 0.082(0.053) 0.00(0.00) Poly47 Fragment 12.89(1.25) 1.222(0.147) 1.093(0.083) 0.009(0.009) 0.056(0.047) 0.00(0.00) Poly49 Fragment 14.00(1.33) 1.222(0.147) 1.015(0.010) 0.015(0.010) 0.014(0.010) 1.00(0.00) Poly5 Repeats 13.556(1.314) 4.778(0.434) 3.544(0.346) 0.000(0.000) 0.693(0.033) 0.00(0.00)
85 Table 3 4. Characterization of polymorphisms in flanking region sequences for the 4 loci. For the PolyD12 4bp indel, the number represents the number of individuals within the population exhibiting this insertion. For the Poly5 30 b p insert, the checkmark means that all individuals within the population encompass the insertion. Any point mutation signaled as ambiguous means that only some individuals within the population have the mutation. N/A means that the substitution is wit hin an insert that the population does not have. ? is for missing data for that population. locus Poly L102 105 107 108 115 118 124 125 126 127 P102 1301 1802 1801 601 602 805 301 PolyB08 58 C/A C C C C C C C C C C A C C C C C C C 85 G/A G G G A G G G G G G A A A A A A A A 207 G/A G G G G G G G G G G A G G G G G G G PolyD12 4 /A ? A A A A A A A A 60 64 TAAT del 2 0 1 ? 0 0 3 0 0 0 0 0 0 0 0 0 0 0 95 T/G T T T ? T T T T T T G T T T T T G T 112 C/G C C C ? C C C C C C G C C C C C C C 157 A/T A A A ? A A A A A A T T T T T T T T 171 T/A T T T ? T T T T T T T T T T T T A T 176 C/T C C C ? C C C C C C T T T T T T T T 213 /T ? T T T T T T T T E01 1 C/T T T T T T T T T T T T T ? C C C T T 10 T/A A A A T A A A A A A A A ? A A A A A 97 C/G G G G G G G G G G G G G ? G G G C G 129 C/A A A A A A A A A A A A A ? A A A C A 231 A/C C C C C C C C C C C C C ? A A A C C Poly1 20 A/T A A A T A A A A A A T A/T A A A A T A 21 /G /G /G /G G /G /G /G /G /G /G G G G G G G G G 36 C/T C C C T C C C C C C T T T T T T T T 40 C/T C C C T C C C C C C T T T T T T T T 53 G/A G G G A G G G G G G A A G G G G A G 66 A/ A A/ A A A A A A A A A A A A A A A A 84 T/C T T T C T T T T T T T C T T T T C T 87 G/A G G G A G G G G G G A A A A A A A A 90 T/C T T T C T T T T T T C C T T T T C T Poly5 3 G/T G G G G G G G G G G G G T T T T G T 5 A/ G 21 T/A T T T T T T T T T T T T A A A A T A 75 C/T C C C T C C C C C C T T T T T T T T 80 110 30 bp insertion NA NA NA NA NA NA NA NA NA 100 /A NA NA NA A NA NA NA NA NA NA 108 G/T G G G T G G G G G G T T G G G G T G
86 Table 3 5 Molecular diversity of Polygala lewtonii populations for all marker types (Fragment, Repeats, Sequence). Values are averages over all 11 loci. N= number of individuals in the population; N a = number of alleles /number of polymorphic sites ; N e = effective number of alleles /number of haplotypes ; H E = expected heter ozygosity; Ho= observed heterozyg osity /nucleotide diversity F = fixation index. Pop 102 105 107 115 118 124 125 126 127 N Fragment 15.636 15.727 15.727 14.000 3.909 13.182 15.636 14.909 14.818 Repeats 11.333 12.667 10.333 6.667 1.667 10.000 9.667 8.000 8.667 Sequence 16 16 16 15 4 20 15 16 16 N a /polymorphic sites Fragment 2.455 2.000 2.091 2.909 2.273 1.545 2.636 2.091 1.727 Repeats 2.333 2.333 1.667 2.667 1.667 2.667 2.333 1.333 1.667 Sequence 15 12 18 18 9 18 10 15 10 N e /number of haplotypes Fragment 1.446 1.340 1.364 1.446 1.306 1.439 1.553 1.314 1.349 Repeats 1.844 1.612 1.272 1.967 1.533 1.472 1.342 1.044 1.239 Sequence 5 4 6 7 4 7 4 6 3 H O /nucleotide diversity Fragment 0.064 0.126 0.068 0.095 0.097 0.091 0.119 0.117 0.090 Repeats 0.033 0.000 0.000 0.000 0.111 0.024 0.042 0.042 0.089 Sequence 0.296 0.155 0.325 0.433 0.566 0.346 0.303 0.360 0.350 H E Fragment 0.218 0.184 0.193 0.255 0.207 0.217 0.247 0.183 0.179 Repeats 0.370 0.303 0.150 0.443 0.241 0.276 0.226 0.039 0.139 Sequence 0.355 0.233 0.325 0.433 0.629 0.346 0.394 0.432 0.350 F Fragment 0.354 0.244 0.475 0.402 0.574 0.573 0.436 0.449 0.396 Repeats 0.903 1.000 1.000 1.000 0.538 0.913 0.805 0.067 0.362
87 Table 3 6. Summary of Chi Square tests for Hardy Weinberg Equilibrium for Polygala lewtonii populations for all loci, based on fragment length (above) and repeat number (below) M=monomorphic; ns=not significant; P<0.05; ** P<0.01; *** P<0.001 102 105 107 115 11 8 124 125 126 127 Poly1 M M ns *** ns *** *** ** *** PolyE01 *** M *** *** ns *** *** *** M PolyB08 ** *** M ** M *** *** *** Poly44 *** ** ns M *** M M M Poly45 *** ns *** M ns ** ns ** *** Poly47 M M *** M M ns M M M Poly46 ns M M *** ns *** ns M M Poly49 ns M M ns M M M M M 102 105 107 115 118 124 125 126 127 Poly1 M *** ns M M *** ** ns ns PolyE01 M M M M M M M M PolyB08 *** *** M *** *** *** *** M M Poly5 *** *** M *** *** *** *** *** ***
88 Table 3 7 Overall genetic diversity in Polygala lewtonii and P. polygama Results for all but N are means per population and are presented with SE in brackets. N= number of individuals; N a = number of alleles; N e = effective number of alleles; H E = expected heterozygosity; H O = observed heterozy tosity; F = Fixation index. Note that for F ST an average of the P. lewtonii population geographic distance was computed, and only those P. polygama populations within this geographic distance range were co nsidered. P. lewtonii P. polygama N Fragment 148 118 Repeats 133 84 N a Fragment 2.127(0.120) 1.966 (0.162) Repeats 2.667(0.303) 1.095(0.194) N e Fragment 1.363(0.048) 1.339(0.090) Repeats 1.889(0.210) 0.927(0.122) H E Fragment 0.199(0.020) 0.213(0.026) Repeats 0.335(0.050) 0.062(0.034) H O Fragment 0.100(0.016) 0.090(0.020) Repeats 0.023(0.010) 0.003(0.003) F Fragment 0.410(0.046) 0.564(0.055) Repeats 0.888(0.044) 0.741(0.113)
89 Table 3 8. Wilcoxon bottleneck detection t est results for Polygala lewtonii fragment length (above) and repeats only (below) dataset s Analyzes for three models of evolution were performed: the Infinite Allele Model (IAM), the Stepwise Mutation Model (SMM) and the Tw o Phased Model ( TPM). For each population, the P values for heterozygosity deficiency (left), and heterozygosity excess (middle) are reported. The shape of the allele frequency distribution(AFD) is reported as L shaped (LS) or in ShiftedMode (SM). pop IAM SMM TPM AFD 102 0.1015/0.91797 0.018 5/0.9863 0.0820/0.9355 LS 105 0.4218/0.6525 0.0546/0.9609 0.2812/0.7812 SM 107 0.4218/0.6562 0.0390 /0.9865 0.0781/0.9453 LS 115 0.0820/0.9355 0.0097 /0.9931 0.0136 /0.9902 SM 118 0.4062/0.6875 0.3125/0.8906 0.4062/0.6875 SM 124 0.0527/0.9580 0.0034 /0.9975 0.0048 /0.9965 LS 125 0.1875/0.8515 0.0195 /0.9882 0.0390 /0.9726 SM 126 0.2187/0.9218 0.0390 /0.9765 0.0781/0.9453 LS 127 0.9218/0.1093 0.5000/0.5937 0.9218/0.1093 LS pop IAM SMM TPM AFD 102 1.0000/0.0625 0.8750/0.1875 1.0000/0.0625 LS 105 0.9375/0.1250 0.6250/0.8125 0.9375/0.125 LS 107 0.8750/0.2500 0.1250/1.0000 0.2500/0.8750 SM 115 1.0000/0.1250 0.2500/0.8750 0.8750/0.2500 LS 118 1.0000/0.1250 1.0000/0.1250 1.0000/0.1250 SM 124 0.6250/0.8125 0.6250/0.8125 0.6250/0.8150 LS 125 0.8750/0.1875 0.1875/0.8750 0.6250/0.8125 SM 126 0.8750/0.2500 0.2500/0.8750 0.2500/0.8750 LS 127 1.0000/0.1250 0.2500/0.8750 1.0000/0.1250 LS
90 Ta ble 3 9. Among populations F ST and R ST values for Polygala lewtonii F ST values are below diagonal and R ST above, both calculated with 9,999 permutations, for fragments (above) and repeats (below), indicates non significance of permutation test (P value>0.05). 102 105 107 115 118 124 125 126 127 Population 0.000* 0.004* 0.279 0.691 0.007* 0.000* 0.033* 0.199 102 0.222 0.082 0.355 0.732 0.053 0.000* 0.069 0.288 105 0.060 0.154 0.184 0.671 0.000 0.003 0.032 0.126 107 0.209 0.343 0.280 0.308 0.137 0.267 0.140 0.025* 115 0.391 0.521 0.427 0.300 0.575 0.660 0.443 0.395 118 0.073 0.193 0.122 0.148 0.401 0.007* 0.026* 0.106 124 0.131 0.243 0.186 0.177 0.404 0.0140* 0.000 0.177 125 0.095 0.225 0.127 0.171 0.361 0.003* 0.000* 0.034 126 0.172 0.303 0.211 0.192 0.347 0.082 0.046 0.038 127 102 105 107 115 118 124 125 126 127 0.216 0.000* 0.042* 0.275 0.139* 0.000 0.098 0.702 102 0.122 0.203 0.207 0.772 0.198 0.260 0.124 0.958 105 0.126 0.029 0.002 0.266* 0.105* 0.000* 0.078 0.688 107 0.142 0.091 0.119 0.347 0.027 0.037 0.065 0.736 115 0.150 0.222 0.214 0.109* 0.498 0.175 0.593 0.342 118 0.262 0.201 0.159 0.079 0.247 0.158 0.109 0.805 124 0.038 0.056 0.041 0.088 0.079 0.203 0.128 0.619 125 0.135 0.133 0.070 0.158 0.268 0.193 0.034 0.874 126 0.344 0.380 0.360 0.251 0.055 0.341 0.269 0.441 127
91 Figure 3 1. Sampling of Poly g ala lewtonii and P. polygama P. lewtonii is represented with yellow stars and P. polygama with purple diamonds. 1301 301 10 2 805 1801 1802 601 602
92 Figure 3 2. Distribution of Polygala lewtonii populations. The four northern populations are located on the Mount Dora Ridge while the rest are scattered along the Lake Wales Ridge.
93 Figure 3 3. Genetic and geographic clustering of both Polygala species. TESS results are shown as bar plots for number of clusters K=2 4. Fragment lengths data set is above and repeat number data set is below. K=2 K=3 K=4 K=2 K=3 K=4
94 Figure 3 4. Genetic and geographic clustering of P. lewtonii DIC values for each K value tested and their bar plots for fragment length (above) and repeat number (below). 2000 2050 2100 2150 2200 2250 2300 2350 0 2 4 6 8 10 12 Deviance Information Criterion DIC number of clusters K 400 450 500 550 600 650 0 2 4 6 8 10 Deviance Information Criterion DIC Number of clusters K K=2 K=3 K=4
95 Figure 3 4 Continued. K=2 K=3 K=4
96 Figure 3 5. Co ntributions of populations of Polygala lewtonii to allelic richness. Allelic diversity of each population (blue dots) is composed of its contribution to the total diversity (red bar) and its divergence from other populations (green bar). Results are given for fragment length (top flanking region sequence (bottom graphs e and f). -15 -10 -5 0 5 10 102 105 107 115 118 124 125 126 127 Allelic contribution Population a Crd Crs Crt -80 -60 -40 -20 0 20 40 60 102 105 107 115 124 125 126 127 Allelic contribution Populations b Crd Crs Crt
97 Figure 3 5. Continued. -6 -4 -2 0 2 4 6 8 10 12 105 107 115 124 125 126 127 Allelic contribution Populations b' Crd Crs Crt -6 -4 -2 0 2 4 6 8 10 102 105 107 115 118 124 125 126 127 Allelic contribution Populations c Crd Crs Crt
98 Figure 3 5. Continued. -4 -2 0 2 4 6 8 10 12 102 105 107 115 124 125 126 127 Allelic contribution Populations d Crd Crs Crt -20 -15 -10 -5 0 5 10 102 105 107 115 118 124 125 126 127 Allelic contribution Populations e Crd Crs Crt
99 Figure 3 5. Continued.
100 Figure 3 6. Distribution of Mantel test results. Each point represents a pairwise genetic to geographic comparison. Fragment lengths data set above, and repeat number data set below.
101 CHAPTER 4 FINE SCALE POPULATION GEN ETIC STUDY OF THREE PLANTS ENDEMIC TO THE CENTRAL FLORI DA SCR UB Introduction The Florida scrub is a unique but highly threatened ecosystem characterized by sandy, nutrient poor soils, and a very xeric environment ( Webb and Myers, 1990 ) Most of the rainfall occurs in the summer, and the vegetation is regulated by frequen t fires ( Abrahamson, 1984a ) The inland Florida scrub, on which this study is focused, is thought to be the result of the unique geological history of the region, w here in the Pleistocene the peninsula was most likely a refugium ( Chapter 2). This resulting unique habitat hosts an exceptional level of endemism and is considered a biodiversity hotspot in North America ( Christman and Judd, 1990 ; Dobson et al., 1997 ) However, 85% of the central Florida scrub has already been converted to agriculture and residential areas, leaving a highly fragmented landscape with which the endemic species must cope ( Christman and Judd, 1990 ; Weekley, Menges, and Pickert, 2008 ) Previous population genetic studies of plant species endemic to the central Florida scrub have found either a high level of genetic diversity, perhaps attributable to the presence of a Pleistocene glacial refugium ( Lewis and Crawford, 1995 ; MacDonald and Hamrick 1996 ; Menges, 2001 ; Mylecraine et al., 2004 ) or to the contrary, very low genetic diversity ( Evans et al., 2000 ; Menges, 2001 ) likely as a consequence of the fragmentation of this fragile ecosystem Given the partitioning of genetic diversity in some species from the central Florida scrub, it has been hypothesized that the oldest and highest ridges, the
102 Lake Wales Ridge and Mo u nt Dora are the centers of origin of several species. The Florida scrub lizard the sand skink, and the mole skink all show deep evolutionary separation between Mo unt Dora and the Lake Wales Ridge at different times in history, and with further lineage se paration within the larger Lake Wales Ridge ( Clark, Bowen, and Branch, 1999 ; Branch et al., 2003 ) The sand skink Neoseps reynoldsi is composed of four main lineages (Mo u nt Dora, the Northern, Central and Southern Lake Wales Ridge) and h as high levels of genetic diversity. The c entral Lake Wales Ridge is considered the ancestral region and the separation of the two ridges is estimated at 2 M Y ( Branch et al., 2003 ) The mole skink, however, is more widespread (the subspecies, including the one endemic to the central ridges has not been shown to be monophyletic and so cannot be considered here) and is thought t o have originated from both Mount Dora and the Lake Wales Ridg e where the genetic diversity is the highest. The separation of the two lineages however, is estimated to date back 4 M YBP in the Pliocene ( Branch et al., 2003 ) These results directly trans late into inferences of higher gene flow within the Lake Wale s Ridge than between the two ridges, even for geographically closer populations ( MacDonald and Ha mrick, 1996 ; Clark, Bowen, and Branch, 1999 ; MacDonald et al., 1999 ) The endemic plant Warea carteri on the other hand reveals a peninsula effect with a strong association between population location on a n orth s outh axis and a cline in allele frequency ( Evans et al., 2000 ) A peninsula effect may be attributable to a leading edge phenomenon in which the source populations for colonization are in the northern part of the peninsula, with colonization occurring southward
103 throughout Florida and loss of alleles through each founder event. These differences might be due to some life history traits, but also to the origin of the species themselves arguing for different histories for different parts of the biota in central Florida Many rare and t hreatened species have lower genetic diversity than widespread ones ( Hamrick and Godt, 1996 ) although the importance of a phylogenetic context and the benefit of comparing rare species to widespread congeners has been highlighted ( Karron, 1988 ; Gitzendanner and Soltis, 2000 ) A species in a reduced habitat that has already undergone a loss of diversity is anticipated to have generally lower genetic diversity than its more widespread congeners, while a naturally rare or restricted species may not. Additionally, levels of diversity are highly correlated within genera (Gitzend anner and Soltis, 2000). Reduced diversity can lead to a lesser ability to respond to current and future environmental changes and evolutionary pressures, increased risks of inbreed ing and drift, and ultimately extinction ( Charlesworth and Charlesworth, 1990 ; Ouborg and Treuren, 1994 ) In the case of the central Florida scrub, the high levels of anthropogenic pressure and habitat fragmentation ( Weekley, Menges, and Pickert, 2008 ) may already have caused population bottlenecks, resulting in smaller effective population sizes, significant inbreeding, reduced genetic diversity and inc reasingly unstable demographics ( Frankham, Ballou, and Briscoe, 2002 ) Also, t he partitioning and geographical distribution of genetic diversity at the population level are crucial to any conservation efforts and proper long term management (Prance, 1995).
104 I here present an analy sis of the fine scale population structure of three species endemic to the central Florida scrub. The genetic diversity of these endemic species was compared to their sister species, and their within species genetic partitioning and gene flow estimated wit h the goal of better understanding their conservation needs. Materials and Methods Study S pecies The following three species endemic to the central Florida scrub and with close relatives in eastern North America were selected for study. In the case of Ile x opaca variety arenicola is endemic to the central Florida scrub while variety opaca is widespread in eastern North America. Asimina obovata (Willd.) Nash ( Annonaceae ) is a shrub adapted to the well drained sandy soils of c entral Florida ( Nash, 1896 ; Huck et al., 1989 ; Nelson et al., 2008 ) Asimina obovata is a shrub or small tree up to 2 m tall, distinguishable from other Asimina in the region by its bright reddish hairs, especially on the midrib and nerves. The leaves are large (4 10 cm long and 2 5 cm wide), the younger ones often oval and the larger ones obovate The flowers are sess ile, situated at the tips of branchlets, typically with 6 yellowish white petals, 5 6 cm long ( Nash, 1896 ) the inner three petals exhibiting a maroon corrugated base ( Nelson, 1996 ) The flowe r is pollinated by several types of large beetles, all generalists except for one specialist ( Norman and Clayton, 1986 ) The fruit is large and thought to be eaten by gopher tortoises ( Kral, 1960 ) but with no published data. It s nearest congener is Asimina incana widespread in Florida and southern Georgia ( Germain Aubrey et al., in prep c )
105 Ilex opaca Aiton var. arenicola (Ashe) Ashe (Aquifoliaceae) is endemic to the Florida scrub. Ilex opaca is a shrub to small tree with coriaceous evergreen leaves with spinulose serrate or entire leaves, and a typical rigid spine 1 mm long or longer on the leaf apex. Within Ilex opaca dark green leaves without revolute margins characterize variety opaca Li ght green, narrower leaves (1 2.5 cm) and margins distinctly revolute correspond to the distinct variety arenicola ( Wunderlin, 1998 ) A phylogenetic analysis places Ilex opaca (including both var. opaca widespread in the eastern US and var. arenicola ) sister to I. myrtifolia and I. cassine ( Ashe, 1925 ; Gottlieb, Giberti, and Poggio, 2005 ) However, the two vari eties of Ilex opaca do not seem to be distinct biological entities ( Germain Aubrey et al., in prep c ) and sho uld therefore be considered one when investigating molecular diversity indices. Gene flow estimates and within population molecular measures will be discussed for the central Florida scrub populations, regardless of taxonomic variety. Prunus geniculata Ha rp. (Lauraceae) the scrub plum, is a federally endangered species known from only 21 sites, all of them on the Lake Wales Ridge ( Harper, 1911 ; USFWS, 1999b ) The scrub plum is a shrub up to 2 m tall, heavily branched, with strongly zigzag twigs and spiny lateral branches. Its deciduous leaves are finely toothed. The five petalled white flowers appear in late winter, when the plant is l eafless, and the fruit is a small, bitter, red plum ( Wunderlin, 1998 ) The amount of flowering and fruiting heavily depends on fire frequency. Prunus geniculata is functionally andromonoecious, with both male (with a vestigial but not functional gynoecium) and bisexual flowers on the same
106 plant ( Weekley et al., 2010 ) Due to the strong fragrance of the flowers, it is pollinated by a variety of in sects and is believed to be self incompatible. Although assumed from morphology to be closely related to Prunus texana and P. angustifolia a molecular phylogenetic study placed it as sister to Prunus maritima from the northeastern U.S. ( Shaw and Small, 2004 2005 ) Collections, M icrosatellite A mplification and G enotyping Microsatellite primers have been developed for all species ( Germain Aubrey et al., 2011 ; Germain Aubrey et al., in prep b ; Germain Aubrey et al., in prep a ) For each of the endemic species, a widespread close congener (or sister species in the case of Prunus geniculata ) was included in the study in order to compare molecular diversity indices in a rigorous manner, controlling for the evolutionary history of each of the endemic species ( Gitzendanner and Soltis, 2000 ) For Asimina 10 loci were amplified for 84 individuals from 8 populations of A. obovata and 47 individuals from 5 populations of A. incana (Table 5 1 and Figure 5 1). For Ilex 10 loci were amplified for 102 individuals of I. cassine from 12 populations, and 199 individuals of I. opaca (39 individuals of I. opaca var. arenicola from 4 populations and 160 individuals of I. opaca var. opaca from 16 popu lations) (Table 5 2 and Figure 5 2). For Prunus 8 loci were amplified for 56 individuals of P. geniculata from 6 populations and 43 individuals of P. maritima from 6 populations (Table 5 3 and Figure 5 3). All Prunus maritima are of variety maritima as var. gravesii is extinct in the wild (Louise Lewis and Donald Les, pers. comm.). Also, because Prunus geniculata is a federally endangered species, a permit from the Division of Plant Industry was used for each site visited (Permit #714). All populati ons were sampled as evenly as possible within
107 their natural distribution and according to collaboration with landowners, especially for the federally listed Prunus geniculata Within a location, individuals were sampled from throughout the population, with a regular distance between individuals, especially for Ilex which sprouts clonal stems from adventitious roots or callus. All population vouchers are deposited at the Florida Museum of Natural History Herbarium (FLAS). Microsatellite loci were amplified in 10 uL reactions as specified in the respective publications: 1 M Betaine, 1.5 mM MgCl 2 0.1 M dNTPs, 0.5 M forward and reverse primers, 0.5 M of one fluorescent labeled M13 primer, and 0.2 unit Taq polymerase. All loci were amplified under the same optimal conditions, requiring 3 min at 95C, followed by 35 cycles of 45 sec at 95 C, 1 min 15 sec at 52C (55 C for Ilex ) and 1 min 15 sec at 72C, with a final step of 20 min at 72C. PCR products were stored at 4C. A fluorescent labeled M13 tail was added to the forward primer to bind to an ABI fluorescent dye, 6 FAM PET NED or VIC labeled M13 primer. The orange 600 Liz ABI oligo was used as a standard, and all four dyes were pooled in a 20 L solution according to the strength of their band whe n run on a 1.2% agarose gel. All samples were sequenced on an ABI 3730 DNA analyzer (Applied Biosystems, Carlsbad, CA USA ) at the Interdisciplinary Center for Biotechnology Research at the University of Florida. Microsatellites were scored automatically u sing Genemapper 1.6 (Soft Genetics, State College, PA, USA), and then checked by eye.
108 Microsatellite D ata A nalysis For each species, Micro Checker ( van Oosterhout et al., 2004 ) was used to test for the presence of null alleles, scoring errors and selection biases of loci. The program SpaGeDi 1.3a ( Hardy and Vekemans, 2002 ) was used to perform a randomization test with 10,000 permutat ions on alleles for R ST to test the evolutionary model for each locus. Genepop On The Web ( Raymond and R ousset, 1995 ; Rousset, 2008 ) was used to test for linkage disequilibrium between loci. A Bonferroni correction was applied to the resulting P values ( Rice, 1989 ) Fi nally, GENODIVE 2.0b17 ( Meirmans and Van Tienderen, 2004 ) was used to detect clones in our data set, which is especially relevant for Ilex given that new stems often sprout from a common underground root stock ( Ashe, 1925 ) All data were imported in Genalex 6.41 ( Peakall and Smouse, 2006 ) for calculations and formatting of the data sets for other software packages. Genalex was used to implement diversity statistics (number of individuals N, number of alleles N a effective number of alleles N eff observed and expected heterozygosity H O and H E and the inbreeding coefficient F ). Tests for conformance to Hardy ( Fisher, 1922 ) with a 10,000 step Markov chain, including a 1,000 burn in for 100 batches in the program Genepop On The Web ( Raymond and Rousset, 1995 ; Rousset, 2008 ) A Mantel test was performed for each endemic to test for a correlation between genetic and geographic distance, and also for the presence of drift or lack of gene flow in the case of populations fixed ( Hutchinson and Templeton, 1999 )
109 To investigate population structure, Genalex was used for an analysis of molecular variance (AMOVA) among populations, within populations and within individuals, and estimates of pairwise population differentiation, F ST and R ST ( chapter 3 for justification of use of both F ST and R ST ). Also, the allelic contribution of each population was evaluated for conservation purposes using the program CONTRIB 2.02 ( Petit, El Mousadik, and Pons, 1998 ) The contribution is partitioned between contribution of a population relative to others in the same species, and divergence of allelic richness from other populations (presence of unique al leles or abundance of rare alleles), both very important factors for conservation. Finally, for effective conservation genetics of these endemic species, a Bayesian likelihood method ( Pritchard, Stephens, and Donnelly, 2000 ; Falush, Stephens, and Pritchard, 2003 2007 ) was implemented in the program TESS ( Durand et al., 2009 ) to infer the number of genetic and g eographic clusters in each of the endemic species. For this program, the DIC (Deviance Information Criterion) stabilizes around the optimal number of clusters (K). However, in contrast to the program STRUCTURE ( Pritchard, Stephens, and Donnelly, 2000 ) TESS includes decay of correlation of membership coefficients with distance within clusters. This might lead to a slower stabilization of K and Kmax (wit h the highest probability) might overestimate the true number of clusters. As advised in the manual, the analysis of the optimal number of clusters has to be assessed on the basis of DIC and bar plots. In determining the optimal number of clusters, we ther efore took into consideration both the curve of DIC against K and the
110 stabilization of the number of estimated clusters in the bar plots. Additionally, variation of results between the different repetitions within each K value was explored to infer the consistency of cluster composition. Results For all species, no loci were in linkage disequilibrium, showed significant selection bias or the presence of null alleles. No data set showed a significant number of clones, and all loci followed the stepwise mu tation model (SMM). Asimina obovata Overall molecular variance was high for the endemic Asimina obovata ( N a =18.5, N eff =9.9, H E =0.83, H O =0.7) compared to its widespread congener A. incana ( N a =8.2, N eff =4.2, H E =0.62, H O =0.59). F the inbreeding coefficient, was slightly higher but comparable to that for A. incana (0.14 vs 0.11) (Table 4 4). Within the endemic, the number of alleles N a ranged between 4.1 and 9.2, with higher values in more southern populations, but this range was not reflected in the effectiv e number of alleles N eff (range 3.41 to 4.89 with standard error overlapping; population O121, in Paynes Creek State Park (SP), with an N eff of 6.35, had the only value significantly higher than the others). For all populations except O110 in Citrus Co., o bserved and expected heterozygosities were not significantly different from each other. O101 and O110 (Lakes Wales Ridge State Forest) were also in Hardy Weinberg equilibrium. The results of the Wilcoxon test for bottleneck detection were negative except f or O103 (Pine Park, Lake Co.), O107 (Tiger Creek, Polk Co.) and O121 (Paynes Creek SP) for heterozygote excess under the IAM model, an d O103 under the TPM model ( chapter 3 for a definition of the models) (Table 4 10). The total contribution of allelic rich ness
111 was positive for O110, O115 (Catfish Creek SP) and O121, with the contribution (Archbold Biological Station (BS)) and O127 (Ocala National Forest (NF)) and the contribution th rough divergence from other populations also positive for O103 and O110. O115 and O121 all positively contributed to allelic diversity (Figure 4 6). The Mantel test for Asimina obovata showed a significant (P value=0.03812), but very small (R 2 =0.002), corr elation between geographic and genetic distance, with points scattered throughout the graph (Figure 4 7). Overall, the partitioning of genetic diversity was overwhelmingly attributed to variation within individuals (77%) under an IAM model, and to among in dividuals within populations (78%) under the SMM model. The contribution of among population partitioning to total molecular variance was comparable at 15% and 14%, respectively (results not shown). Examining genetic divergence between populations, pairwis e F ST values were surprisingly homogeneous under the IAM model (25 population pairwise values between 0.111 and 0.187, 2 below 0.1 (0.059 and 0.088) and 1 above (0.209)). However, under the SMM model, R ST values showed a wider range of values (0.00 0.640) (Table 4 13). The comparable, but not entirely similar to, those from the TESS run (Figure 4 4). The biggest cluster of populations including O101, O106, O107 and O115, showed little or no differentiation, although populations O106 and O101, and O107 and O106 had a slightly higher R ST (0.206 and 0.251, respectively). O106 was also isolated from O121 by a significant R ST so that O106 was isolated from all
112 populations north of its location. The TESS results also clustered populations O127 and O103, a result consistent with an R ST value of 0.00. The genetically isolated O121 and O110 in TESS did show high levels of genetic differentiati on with several populations, although there were some exceptions, such as O110 and both O101 ( R ST =0.015) and the geographically close O107 ( R ST =0.067), and populations O121 and O103, moderately geographically distant from different ridges ( R ST = 0.064) ( Table 4 15). Ilex opaca var arenicola Overall, the molecular diversity of Ilex opaca was comparable to that of its sister species, I. cassine with slightly more alleles (16.6 vs. 14.1) but slightly fewer effective alleles per population (5.5 vs. 6.6). H eterozygosities also were comparable ( H E =0.65 and 0.57 and H O =0.41 and 0.47 for I. cassine and I. opaca respectively). The inbreeding coefficient was slightly higher for I. cassine (0.39 vs. 0.23). Within Ilex opaca we compared the two varieties and fo und that despite a much larger sample size (160 individuals vs. 39), var. opaca had more alleles per population (14.6 vs. 10.7) but a comparable number of effective alleles per population (5 vs 4.7) than var. arenicola (Figure 4 5). Within Ilex opaca and despite different population sizes (N=1 16), the average number of alleles per population did not differ greatly (1.5 for the population with 1 individual, and N a =2 5.9 with an average of 3.78 and a median of 3.7; N eff = 1.55 3.62 with an average of 2.57 and a median of 2.69). The heterozygosity levels also were homogeneous, and the expected and observed heterozygosities barely differed ( H O =0.27 0.52, average 0.40; H E =0.27 0.5, average 0.40). Populations O1001 (MD), O 115 (Catfish Creek, FL), O1101 (AL),
113 O1204 (MA), O1503 (GA), O701 (SC), O802 (MS), O1201 (MA) and O202 (KY) were all in Hardy Weinberg equilibrium. No populations labeled as var. arenicola were in equilibrium, maybe showing inbreeding or small population s izes. Only one population (O1402 in KY) showed a high inbreeding coefficient (0.24 with S.E. 0.11), and seven other populations showed very low yet significant levels of inbreeding (all under 5%) (Figure 4 8). Population O1402, however, did not exhibit sig ns of a bottleneck under any mutation model, while A106 (Archbold BS, FL) was the only population exhibiting heterozygote deficiency under both SMM and TPM models (Table 4 11). The allelic contributions of populations were all negative except for A106 and O114 (Loblolly Woods, FL). However, many of other populations was negative (Figure 4 11), reveal ing the absence of populations with private alleles. The Mantel test revealed that a correlation between genetic and geographic distance was not significant (P value=0.142, R 2 =0.09378), and points were scattered throughout the graph (Figure 4 8). For the o verall partitioning of the molecular variance, F ST partitioned genetic diversity mostly within individuals (76%) and R ST to among individuals within populations (94%). The variation among populations accounted for 9% of the total variation under the IAM an d for 5% under the SMM model. The clustering of individuals according to their genetic and geographic distance grouped populations A106 (Archbold BS, FL) and A116 (Catfish Creek, FL) together, both of var. arenicola All other Floridian populations (A108, O114, O115, O110)
114 clustered with the two from Mississippi (O801 and O802) and the one from Alabama (O1101). All other Ilex populations with samples along the east coast of North America from Georgia to Maine, clustered together (Figure 4 5). The estimatio n of gene flow from R ST values does not, however, support this clustering of populations exactly. A major barrier to gene flow, revealed by high R ST values, seems to occur between populations in Mississippi and populations in Florida, which contrasts with the TESS cluster. Florida populations appear to tend to exchange genes, regardless of their cluster, except for populations O114 and A116 ( R ST =0.110) (Figure 4 14). Prunus geniculata The overall genetic diversity indices were comparable for the endemic Prunus geniculata and its sister species P. maritima if not a little higher for the endemic, with N a = 8.87 and 8, N eff = 4.4 and 3.94, H E = 0.69 and 0.47 and H O = 0.69 and 0.6, respectively. The inbreeding coefficients, however, differed, with the endemic having very low inbreeding (0.06) while its congener exhibited a fairly low, but substantially higher, F at 0.19 (Figure 4 6). Within Prunus geniculata indice s of molecular diversity were low. Some populations were very small (two populations only contained three individuals each), the number of alleles averaged 3.39 (2.00 4.50), the effective number of alleles 2.48 (1.70 3.61), observed heterozygosities averag ed 0.51 and effective heterozygosities were lower (average 0.40), with three populations with low expected heterozygosities ( H O =1.63, 0.58 and 0.66 and H E =0.43, 0.37 and 0.45 for populations 106 (Archbold BS), 116 (Catfish Creek SP) and 119 (Catfish Cree k SP), respectively). These same three populations exhibited significant
115 results for the Wilcoxon test (0.43, 0.65 and 0.64 for heterozygote excess, respectively) (Table 4 12). Populations 101 (Lake Wales Ridge SF), 105 (Clermont) and 119 (Catfish Creek SP ) were in Hardy Weinberg equilibrium (Table 4 10). From the populations departing from HWE, 116 was the only one exhibiting heterozygote deficiency under all models of evolution, a strong sign of a recent bottleneck. Several others also showed signs of a p ast bottleneck under the IAM model, but this was not sustained under the SMM and TPM models, which were determined to be more appropriate for these loci (Table 4 12). For each population, the allelic diversity was positive or non existent, but all populati ons had a positive contribution to the total diversity of the species due to its positive contribution compared to other populations (populations 101 (Lake Wales Ridge SF), 102 (Lake Wales Ridge SF) and 105 (Clermont)), or due to its divergence from other populations (populations 106 (Archbold BS), 116 (Catfish Creek SP) and 119 (Catfish Creek SP)), or both (population 107 (Tiger Creek))(Figure 4 11). The Mantel test was significant (P value=0.001) but very small (R 2 =0.03373), with points scattered througho ut the graph (Figure 4 9). The overall partitioning of molecular variance was, just as in Asimina and Ilex attributed mostly to the within individual variation for the IAM model and among individuals within populations under the SMM model. The among popul ation contribution was of 8% and 22% for each model, respectively (Figure 4 9). The genetic and geographic clustering of the populations of Prunus geniculata revealed three clusters: one composed of only population 116, one of population 106 (Archbold BS), and the third cluster with all remaining populations
116 (Figure 4 6). Population 116 (Catfish Creek SP), although geographically very close to populations 119 (Catfish Creek SP), 107 (Tiger Creek) and 101 (Lake Wales Ridge SF), shows high levels of F ST with all these populations and of R ST with all populations but 101 (Lake Wales Ridge SF), consistent with its lack of gene exchange with others. Population 106 (Archbold BS) is geographically more isolated and also shows signs of a genetic barrier with populati on 116 (Catfish Creek SP) only with R ST but all populations with F ST Within the largest genetic cluster, no genetic structure exists under the SMM model, but the IAM model shows significant molecular divergence between population 119 (Catfish Creek SP) a nd other populations in its cluster (Table 4 15). Discussion Asimina obovata The endemic species Asimina obovata showed significantly higher allelic diversity and higher heterozygosity than the more widespread A. incana These results go against the commo n pattern that more narrowly distributed species exhibit less genetic diversity than their widespread congeners ( Hamrick and Godt, 1989 ; Gitzendanner and Soltis, 2000 ) al though the opposite has been observed in this region for other endemics ( MacDonald and Hamrick, 1996 ; Menges, 2001 ) Within Asimina obovata measures of genetic diversity reveal greater allelic diversity in O121, the only population on a smaller ridge just west of the Lake Wales Ridge, than any other population. Population O110 is the only one to have significantly lower expected than observed heterozygosity. This population, on
117 the Brooksville Ridge, northwest of the Lake Wales Ridge, and population O101, on the Lake Wales Ridge, have significantly negative inbreeding coefficie nts. Bottlenecks were detected in three populations (O121, O107 and O103), but only in O103, the only sampled population in the northern part of the Lake Wales Ridge, is the detection of heterozygote excess significant under both the IAM and the TPM model s. Population O121 (Paynes Creek SP), having a heterozygote deficiency under the IAM model, also shows signs of isolation (with high R ST values) from neighboring populations O107 (Tiger Creek) and O110 (Citrus Co.) to the north, O101 (Lake Wales Ridge SF) to the east and O115 (Catfish Creek SP) to the south. Interestingly, this population shows the greatest allelic diversity of all sampled populations of Asimina obovata and the only one to have both positive contribution to the total diversity of the speci es and positive contribution due to divergence from other populations (Figure 4 10). Other significant barriers to gene flow revealed by high R ST values are noticeable between the northern populations O127 and O110, located on different ridges, but otherwi se geographically close (Figure 4 1). The Mantel test results, with a lack of isolation by distance and a scattering of the points in the graph, translate into significant drift throughout all populations of the species (Figure 4 7) ( Hutchinson and Templeton, 1999 ) Ilex opaca Ilex cassine and Ilex opaca show similar indices of molecular variance. Within Ilex opaca however, variety arenicola despite a much smaller sample size, shows barely smaller effective number of alleles per population and slightly higher heterozygosity measures (Table 4 5). This result could be explained by
118 the conc entration of var. arenicola in central Florida, a known Pleistocene refugium ( Webb and Myers, 1990 ; Lewis and Crawford, 1995 ; Peterson, Ma rtinez Meyer, and Gonzalez Salazar, 2004 ) and a center of genetic diversity. However, when examination of molecular diversity indices at the population level reveals that none of the central Floridian I. opaca var. o paca populations exhibit higher effective number of alleles or heterozygosity (both expected and observed), arguing against this hypothesis of central Florida being a center of genetic diversity for Ilex opaca as a whole (Table 4 8). Several populations sh owed small but significant levels of departure from F = 0 ( Wright, 1951 ) Five populations had an F value above 0, a sign of inbreeding, and three had a negative F a sign of excess heterozygosity (Table 4 8). Of these populations, O201 (KY) and A106 (Archbold BS) also showed a he terozygote deficiency with the Wilcoxon bottleneck detection test, confirming small but significant levels of inbreeding due to a recent bottleneck. O201 is a population in Kentucky. The other population sampled from this state exhibited a shifted distribu tion of allelic frequency characteristic of a mutation drift disequilibrium, another sign of a bottleneck ( Cornuet and Luikart, 1996 ) Since our sampling of Ilex opaca var. opaca is irregular and given that Kent ucky is the only northern state that is not along the coast (more heavily sampled), it is difficult for us to interpret this result. These results, however, call for further sampling of this species on a landscape level. Population A106 is the southernmost population of var. arenicola to have been sampled, on the southern end of the Lake Wales Ridge, at Archbold Biological Station. This station has been known to have undergone a much
119 improved land management, especially with the implementation of regular pr escribed burns since 1974 ( Main and Menges, 1997 ) after a long period of active burn suppressions. These regular burns benefit scrub species ( Menges and Kohfeldt, 1995 ; Weekley and Menges, 2003 ) but the long suppression of fires on the station might have impacted this population of Ilex opaca var. arenicola more heavily than others. Also, the Wilcoxon test detects bottlenecks within 4Ne generations ( Cornuet and Luikart, 1996 ) and because Ilex opaca is a long lived species, the bottleneck most likely pre dates the newer fire management strategy of the Archbold Biological Station. It would be interesting to increase the sampling of Ilex at Archbold and along the southern half of the Lake Wales Ridge, to detect the extent of this bottleneck in relation to the history of fire management of the region. The results of the Bayesian genetic and geographic approach implemented in TESS prod uced a very interesting separation of the populations in the southeastern US. One cluster groups two populations described as var. arenicola on the Lake Wales Ridge. Another cluster groups all other populations of Ilex opaca in Florida, all situated to the west of the first cluster, with populations in Alabama and Mississippi. The other populations, sampled north and east of Florida along the coast, are clustered together, separate from the two more southern clusters. This distribution is interesting and se ems to correspond to a pattern seen in other species such as the seaside sparrow ( Ammodramus maritimus), the tiger beetle ( Cicindella dorsalis ) and the coastal plain bal m ( Dicerandra linearifolia ) with an Atlantic Coast/Gulf Coast genetic break between
120 pop ulations ( Soltis et al., 2006 ) This pattern can also be found in the Atlantic white c edar Chamaecyparis thyoides (Cupr ess aceae) An allozyme study found three main clusters of populations, one along the Atlantic coast, one in central Florida and one along the Gulf coast ( Mylecraine et al., 2004 ) However, gene flow, inferred from R ST values (Figur e 4 14), shows differentiation between populations in Mississippi versus those in Florida and Georgia. The gene flow barrier between Mississippi and Florida populations is in direct contradiction to the genetic clusters found with TESS. A study on wolverin es found similar discrepancies (using STRUCTURE, not TESS) and considered the Bayesian approach to be more conservative than other methods ( Cegelski, Waits, and Anderson, 2003 ) Also, because TESS includes a decay of correlation of me mbership coefficient with distance within clusters, it may overlook small but real spatial genetic discontinuities within clusters if these are mostly due to geographic distance. This is especially true when sampling is irregular, and given that our sampli ng in Florida is denser than along the remaining Gulf Coast, this might especially be the case. It would be interesting to increase sampling in the region to try and detect finer genetic structure along the coast. Finally, the strongest detection of a gene flow barrier within the Gulf Coast cluster occurs between population O114, in north central Florida, and the population in Alabama ( R ST =0.239), possibly hinting at a barrier at the Apalachicola River, a common pattern of population structure in the regio n ( Soltis et al., 2006 ) Once m ore, further sampling is needed to draw any firm conclusion.
121 Prunus geniculata Prunus geniculata shows higher levels of molecular diversity than its close congener P. maritima It has more alleles, both absolute and effective, higher expected and observed levels of heterozygosity and lower inbreeding (Figure 4 However, it should here be noted that throughout our contact with collaborators in Maine and Delaware, it became increasi ngly obvious that Prunus maritima is a species under considerable anthropogenic pressure and that its distribution is patchy; population sizes are small, and this species might itself be more fragmented than it once was. Nonetheless, an observed and expect ed heterozygosity of 0.69 and an inbreeding coefficient of 0.06 are not alarming values for a federally listed species (Table 4 6). Within the Florida endemic, populations 106 (Archbold BS), 116 (Catfish Creek SP) and 119 (Catfish Creek SP) show low levels of allelic diversity. For population 119, this could be due to the very small sample size (N=3), and to a certain extent for population 116 as well. However, population 106 has a low number of alleles, as well as a significant difference between observed and expected heterozygosities (populations 116 and 119 also exhibit this significant difference) (Figure 4 9). A loss of heterozygosity can indicate that the populations is suffering from high inbreeding, and that it is losing its evolutionary potential ( Frankham, 2003 ) Inbreeding could be dire ctly linked to the noticeably low seedling recruitment in this species, despite abundant flowering ( Weekley et al., 2010 ) However, these three populations have significantly negative inbree ding coefficients, a sign of heterozygote excess (Table 4 9). This observation is further supported in the
122 case of population 116 (Catfish Creek SP) by the detection of heterozygote excess under all evolutionary models for the Wilcoxon test, which is inte rpreted as the presence of a recent bottleneck ( Cornuet and Luikart, 1996 ) Also, the shifted mode of the allele frequency distribution graph is characteristic of a bottleneck (as a disruption of the mutation drift equilibrium). No other population showed signs of a bottleneck under the SMM or the TPM evolution model (Table 4 12 ). However, all populations but 119 (Catfish Creek SP) and 101 (Lake Wales Ridge SF) showed signs of recent past bottlenecks under the IAM model, which seems less conservative. These populations might indeed have gone through a smaller but nonetheless sign ificant recent bottleneck as the IAM model results are thought to be more accurate than others for the Wilcoxon test, no matter which model the loci follow ( Harper, Maclean, and Goulson, 2003 ) The total allelic richness was positive for three populations (101, 106 and 107), with the allelic richness of the southern populati on 106 greatly diverging from other populations (populations 107, 116 and 119 also did to a lesser extent), and the richness of populations 101 and 105 contributing positively to the total diversity (Figure 4 11). The Mantel test results show no isolation by distance but the fixation of populations due to significant drift ( Hutchinson and Tem pleton, 1999 ) Genetic clustering and gene flow measures both show the isolation of population 106, at Archbold Biological Station, from other populations. This makes this inbred population a priority for the conservation of the species. The divergence of its allelic richness from other populations is due to the presence of unique alleles, despite having gone through a recent bottleneck. Due to the
123 combination of all these factors, it is of very high priority to conserve all individuals of this populati on for the species to keep all its evolutionary potential ( Ellstrand and Elam, 1993 ; Petit, El Mousadik, and Pons, 1998 ; Lowe et al., 2005 ) A denser sampling in the southern half of the Lake Wales Ridge might locate some new populations sharing alleles with 106 and could show the population to be less isolated than is shown here. Population 116 (Catfish Cree k SP) also is clustered separately from other populations of the same species. It also exhibits signs of inbreeding, although to a lesser extent than 106 (no bottleneck detection from the Wilcoxon test, but an inbreeding coefficient departing from 0 and a low H E ). Gene flow barriers are also strong around this population, despite the geographic proximity to three other populations (119 (Catfish Creek SP), 107 (Tiger Creek) and 101 (Lake Wales Ridge SF)). This might be explained by factors other than abiotic ones, and the phenology and pollination of this population need to be studied further to try and explain this phenomenon. The last genetic cluster encompasses all other populations of Prunus geniculata Population 107 on The Nature Conservancy land Tiger Creek seems to have the highest and most differentiated population of all. This makes it a very important population for the long term conservation and potential for adaptation for the species. Comparisons All three species reveal different evolutionary histories and patterns of genetic diversity and partitioning. This supports the idea that, despite a narrow endemism common to several species, it is crucial to study species separately to
124 reveal potentially very different stories. However, some common pa tterns should here be highlighted as they are relevant to the species history, and have major implications for conservation. All three species show moderate to high levels of molecular variation when compared to their close congeners. Despite an overwhelm ing number of studies comparing levels of allozyme variation to other species with similar life histories ( Hamrick and Godt, 1989 ) it would not be rigorous to compare levels of diversity measured with allozymes to levels measured with microsatellites, as different markers have shown to reveal slightly different aspects of genetic diversity, partitioning and history ( Young, Boyle, and Brown, 1994 ) Very few studies have used microsatellites for species endemic to the central Florida scrub. Comparing levels of genetic diversity to a close congener makes the most sense evolutionarily ( Gitzendanner and Soltis, 2000 ) All levels of genetic diversity were comparable to ( Ilex ) or higher than ( Prunus and Asimina ) the clos e congener. In a study of several species of Polygonella the two species endemic to the Lake Wales and Mount Dora Ridges, P. myriophylla and P. basiramia, were found to have higher within population genetic diversity than their widespread congeners P. art iculata and P. americana from northeastern North America ( Lewis and Crawford, 1995 ) Results, however, are not t otally comparable to those for our study, however, as they are based on allozymes. The authors attribute these unexpected differences in genetic diversity to the presence of a Pleistocene refugium in peninsular Florida, and the subsequent founder effect of the more widespread species. Since this founder event was recent (10,000 years BP), the
125 widespread population has not yet had time to fully recover and increase its genetic diversity through the evolution of new alleles. Similarly, two Dicerandra species endemic to the Lake Wales Ridge exhibit higher levels of genetic diversity and polymorphism than their more widespread congeners ( MacDonald and Hamrick, 1996 ) The same is true of Liatris ohlingerae a Florida scr ub endemic, although this species was not compared to a close congener, but to other species with similar life histories ( Menges, 2001 ) For all three species, the populations located in the southern part of the Lake Wales Ridge all seem to encompass a special conservation interest, and all were genetically and geographically isolated from other populations. Populations 106 (Archbold BS) of Ilex opaca showed signs of inbreeding, and population 106 of Prunus geniculat a has low diversity and a high contribution to allelic richness due to strong divergence from other populations of the species, which shows a high content of unique alleles. Strong gene flow barriers also exist between these populations and the more norther n ones for Prunus and Asimina and in the case of Ilex population 106 only clustered with population 116 (located in the middle part of the Lake Wales Ridge). This observation is concordant with that from the Florida scrub lizard and the sand skink, both showing lineages in the southern Lake Wales Ridge to be separate from other lineages on this ridge, on other ridges or in coastal scrub habitat ( Branch et al., 2003 ) Another common observatio n is the distinctness of one population in the central Lake Wales Ridge, despite the presence of geographically close populations around. Prunus population 116 in Catfish Creek SP shows low
126 diversity and signs of a strong recent bottleneck. Also, strong ge ne flow barriers surround this population, isolating it from surrounding Prunus populations. In Ilex opaca population 116 is genetically close to the southern population 106, but not to the close population 108 (and despite the fact that they are both var arenicola ). Finally, Asimina population 121 (Paynes Creek SP) is located close to population 101 (Lake Wales Ridge SF), but a strong genetic barrier has isolated population 121 from surrounding populations. This population shows the greatest allelic rich ness, partitioned into both richness compared to the total species richness, and divergence from other populations (presence of private alleles). The sand skink also shows a genetic cluster in the middle of the Lake Wales Ridge, its center of diversity, bu t shows more structure in the rest of the species than any of these three plant species ( Branch et al., 2003 ) Asimina obovata is the only plant species examined so far that seems to exhibit a gene flow barrier between the Mount Dora and Lake Wales Ridges. The mole skink has been shown to have lineages on each ridge which have been separated for 4 million years, and the Florida scrub lizard and the sand skink show the same separation, dating b ack to 2 million years ( Branch et al., 2003 ) Prunus geniculata does not occur on the Mount Dora Ridge, but Ilex opaca var. arenicola does, and it would be interesting to see if those populati ons are isolated from those on the Lake Wales Ridge. Lastly, the presence of significant genetic drift in all three species is consistent with the small population sizes I observed in the field and the inferred lack of gene flow between small populations. Drift means that any change, biotic
127 or abiotic, could dramatically impact the species in its ability to retain rare or private alleles, avoid the fixation of deleterious alleles, and ensure the reproductive ability of the species (especially in the case o f Prunus which already has worryingly low levels of seedling recruitment and a mechanism of self incompatibility ( Weekley et al., 2010 ) ). These findings are important for conservation, and we advise giving priority to more isolated populations, as well as more divergent ones, preserving the potential for future adaptation through the conservation of populations with private alleles ( Petit, El Mousadik, and Pons, 1998 ) The potential for gene flow should also be preserved and, if necessary, enhanced for the long term survival of the species ( Ellstrand and Elam, 1993 ) Impacts of A nthropogenic A ctivity on E ndemic S pecies We should here discuss the fact that despite the lack o f inbreeding, lowered genetic diversity, and increased population structuring, all positive signs for rare endemic species, we should not infer that the recent anthropogenic activity in the region has not impacted the genetics of these species. Habitat los s and degradation may not yet have impacted genetic diversity. Inbreeding and fitness are thought to be impacted faster than genetic diversity, as seen in neotropical trees ( Lowe et al., 2005 ) This could mean that populations that exhibit a significant inbreeding coefficient, but no other signs of reduced genetic diversity (for example Ilex opa ca var. arenicola A106 in Archbod BS, the southernmost population on the Lake Wales Ridge), will, in the near future, have these diversity indices decrease despite conservation efforts. This is an important factor to take into consideration when setting up a conservation management plan, and for interpreting the results of genetic monitoring. Other studies contradict this, finding
128 that habitat loss impacts fitness and genetic diversity before it impacts inbreeding ( Aguilar et al., 2008 ) which seems to be more important for smaller populations ( Leimu et al., 2006 ) All of these differences in response of various indices to habitat fragmentation, bottleneck and disturbance decrease with time ( Prober and Brown, 1994 ; Young, Boyle, and Brown, 1994 ; Leimu et al., 2006 ) Finally, the distance of the population to anthropogenic habitation or activity might be a very important factor to take into consideration when interpreting genetic variability resu lts, gene flow barriers and conservation needs ( Prober and Brown, 1994 ; Menges et al., 2010 ) It might be most informative to try and correlate the results of this study with past and current habitat patches in the landscape of the central Florida scrub habitat as this area has been so heavily affected by anthropogenic activity in the last 100 years ( Cegelski, Waits, and Anderson, 2003 ; Menges et al., 2010 ) Conclusions This study revealed that three taxa endemic to the central Florida scrub did not show reduced levels of genetic variation when compared to their more widespread sister species. This might be due to the fact that anthropogenic activity is recent in the regi on (120 years or less) and its impact on the genetic diversity and fitness of the species might not yet show in genetic parameters. On the other hand, the presence of genetic drift for all three species is a warning of their vulnerability, especially consi dering the small population sizes I encountered in the field. The presence of an isolated and differentiated population on the southern end of the Lake Wales Ridge for all three species shows a common pattern. The
129 isolation of one population in each speci es in the central part of the Lake Wales Ridge, despite the presence of surrounding localities, is a good example that geographic closeness should never be assumed to imply genetic clustering, and these clusters should be considered separate evolutionary l ineages for conservation management. Finally, the presence of a gene flow barrier between the two oldest ridges in Florida, the Lake Wales Ridge and the Mount Dora Ridge, in only one of the three species shows that more sampling needs to be done for Ilex maybe revealing a pattern that is common in lizards and skinks. More species should be included in a broader comparative study of the region as genetic studies using microsatellites are now becoming more commonplace.
130 Table 4 1. Sampled populatio ns for the Asimina study. N is the number of individuals in the population. LWRSF= Lake Wales Ridge State Forest, TNC= The Nature Conservancy, CR= County Road, ADB=Allen David Broussard, SP=State Park. Species Pop Location N Voucher A. obovata O101 LWRSF, Polk Co, FL 12 A. obovata O103 Pine Park, Lake Co, FL 12 A. obovata O106 Archbold, Highlands Co, FL 15 A. obovata O107 TNC Tiger Creek, Polk Co, FL 6 A. obovata O110 CR 486, Citrus Co, FL 4 J.R. Abbott 24670 A. obovata O115 ADB Catfish Creek SP, Polk Co, FL 11 A. obovata O121 Paynes Creek SP, Hardee Co, FL 14 A. obovata O127 Ocala NF, Marion Co, FL 9 A. incana I111 Morriston, Levy Co, FL 16 A. incana I112 Morningside, Alachua Co, FL 15 A. incana I127 Ocala NF, Marion Co, FL 2 A. incana I1506 Echols Co, Georgia 6 A. incana I1507 Lowdes Co, Georgia 7
131 Table 4 2. Sampled populations for the Ilex study. Archbold=Archbold Biological Station; TNC=The Nature Conservancy; ADB=Allen David Broussard; SP=State Park Species Pop Location N Voucher Ilex cassine C103 Pine Park, Lake Co, FL 11 Ilex cassine C104 Palakaha Park, Lake Co, FL 11 Ilex cassine C106 Archbold, Highlands Co, FL 16 Ilex cassine C107 TNC Tiger Creek, Polk Co, FL 8 Ilex cassine C112 Morningside, Alachua Co, FL 5 Ilex cassine C122 Werner Boyce, Pasco Co, FL 16 Ilex cassine C123 Highlands Hammock, Highlands Co, FL 2 Ilex cassine C1505 Old National Highway, Camden Co, GA 6 R. Carter 18219 Ilex cassine C1508 Point Peter, Camden Co, GA 8 R. Carter 18216 Ilex cassine C502 Military Ocean Terminal, Brunswick Co, NC 6 R. Peet n.s. Ilex cassine C702 Louis Ocean Bay, Horry Co, SC 3 Ilex cassine C803 Sarracenia Rd, Stone Co, MS 10 S. Leonard n.s. I. opaca var. arenicola A106 Archbold, Polk Co, FL 15 I. opaca var. arenicola A108 TNC Saddle Blanket, Polk Co, FL 7 I. opaca var. arenicola A116 ADB Catfish Creek SP, Polk Co, FL 16 I. opaca var. arenicola A121 Paynes Creek SP, Hardee Co, FL 1 I. opaca var. opaca O1001 Dorchester Co, MD 5 W. Knapp n.s. I. opaca var. opaca O1002 Adkins bog, Wicomico Co, MD 7 W. Knapp n.s. I. opaca var. opaca O114 Loblolly Woods, Alachua Co, FL 5 I. opaca var. opaca O115 ADB Catfish Creek SP, Polk Co, FL 6 I. opaca var. opaca O801 Bluff Creek Rd, Stone Co, MS 10 S. Leonard n.s. I. opaca var. opaca O1101 Alabama 15 I. opaca var. opaca O1204 Camp Farley, Mashpee Co, MA 9
132 Table 4 2. Continued. I. opaca var. opaca O1401 York Co, PA 16 I. opaca var. opaca O1503 Owens Ferry Rd, Camden Co, GA 8 R. Carter 18270 I. opaca var. opaca O1504 Magnolia Bluff, Camden Co, GA 11 R. Carter 18292 I. opaca var. opaca O201 Klaber Ridge, Menifee, KY 16 I. opaca var. opaca O701 Richland Co, SC 15 Nelson 26944 I. opaca var. opaca O802 Beech Magnolia woods, Perry Co, MS 10 I. opaca var. opaca O1201 Weymouth Great Pond, Weymouth Co, MA 7 I. opaca var. opaca O1402 Wissler Run, Lancaster Co, PA 16 I. opaca var. opaca O202 Whittleton Ridge, Powell Co, KY 4 Table 4 3. Sampled populations for the Prunus study. LWRSF=Lake Wales Ridge State Forest; Archbold=Archbold Biological Station; SP=State Park; ABD=Allen David Broussard; TNC=The Nature Conservancy Species Pop Location N Voucher Prunus geniculata 101 LWRSF, Polk Co, FL 3 Prunus geniculata 102 LWRSF, Polk Co, FL 1 Prunus geniculata 105 Clermont, Lake Co, FL 16 J.R. Abbott 22697 Prunus geniculata 106 Archbold, Highlands Co, FL 15 Prunus geniculata 107 TNC Tiger Creek, Polk Co, FL 12 Prunus geniculata 116 ADB Catfish Creek SP, Polk Co, FL 7 Prunus geniculata 119 ADB Catfish Creek SP, Polk Co, FL 3 P. maritima var. maritima 1202 Wing Island, Barnstable Co, MA 9 I. Kadis 1596 P. maritima var. maritima 1203 Wing Island, Barnstable Co, MA 10 I. Kadis 1596 P. maritima var. maritima 1205 Wing Island Front Dune, Barnstable Co, MA 9 I. Kadis 1597 P. maritima var. maritima 901 Henlopen SP point, Sussex Co, DE 3 W. Knapp n.s. P. maritima var. maritima 902 Henlopen SP dock, Sussex Co, DE 5 W. Knapp n.s. P. maritima var. maritima 903 Seashore SP, Sussez Co, DE 7 W. Knapp n.s.
133 Table 4 4. Overall molecular diversity indices for Asimina obovata and A. incana Asimina obovata Asimina incana N 87 47 N a 18.5 8.2 N eff 9.9 4.2 H E 0.83 0.62 H O 0.7 0.59 F 0.14 0.11 Table 4 5. Overall molecular diversity indices for Ilex opaca and I. cassine and both varieties of I. opaca separately Ilex cassine Ilex opaca Ilex opaca var. arenicola Ilex opaca var. opaca N 102 199 39 160 N a 14.1 16.6 10.7 14.6 N eff 6.6 5.5 4.7 5 H E 0.65 0.57 0.6 0.54 H O 0.41 0.47 0.52 0.45 F 0.39 0.23 0.21 0.21 Table 4 6. Overall molecular diversity indices for Prunus geniculata and P. maritima Prunus geniculata Prunus maritima N 56 43 N a 8.87 8 N eff 4.4 3.94 H E 0.69 0.47 H O 0.69 0.6 F 0.06 0.19
134 Table 4 7 Molecular diversity of populations of Asimina obovata N=number of samples in the population; N a =number of alleles; N eff =effective number of alleles; H O =observed heterozygosity; H E =expected heterozygosity; HWE= P value for the exact Fisher's test Chi squared for Hardy Weinberg equilibrium (P values in italic mean that the population is in equilibrium) ; F =inbreeding coefficient. Pop O101 O103 O106 O107 O110 O115 O121 O127 N 10 13 15 8 4 11 14 9 N a 6.00(1.12) 5.10(0.94) 7.80(1.50) 4.5(0.75) 4.1(0.89) 5.20(0.89) 9.20(1.72) 6.30(1.25) N eff 4.07(0.82) 3.71(0.73) 4.89(1.15) 3.41(0.57) 3.49(0.74) 3.46(0.65) 6.35(1.28) 4.34(0.99) H O 0.72(0.10) 0.56(0.08) 0.63(0.09) 0.58(0.10) 0.72(0.11) 0.62(0.08) 0.70(0.09) 0.60(0.09) H E 0.63(0.09) 0.61(0.09) 0.67(0.08) 0.61(0.08) 0.57(0.09) 0.61(0.08) 0.71(0.09) 0.63(0.09) HWE 0.9 0 0 0.01 0.97 0 0 0.03 F 0.13(0.08) 0.04(0.07) 0.05(0.07) 0.00(0.11) 0.28(0.10) 0.03(0.08) 0.00(0.04) 0.06(0.06)
135 Table 4 8 Molecular diversity of populations of Ilex opaca (both varieties). N=number of samples in the population; N a =number of alleles; N eff =effective number of alleles; H O =observed heterozygosity; H E =expected heterozygosity; HWE= P value for the exact Fisher's test Chi squared for Hardy Weinberg equilibrium (P values in italic mean that the population is in equilibrium) ; F =inbreeding coefficient. Pop N N a N eff H O H E HWE F A106 15 5.20(1.27) 3.04(0.73) 0.44(010) 0.50(0.10) 0 0.12(0.07) A108 7 3.70(0.94) 2.95(0.80) 0.41(0.10) 0.45(0.11) 0.04 0.07(0.11) A116 16 4.10(0.87) 2.28(0.42) 0.42(0.10) 0.44(0.10) 0 0.05(0.06) A121 1 1.50(0.22) 1.50(0.22) 0.60(0.16) 0.30(0.08) NA NA O1001 5 3.40(0.88) 2.89(0.80) 0.43(0.12) 0.43(0.11) 0.46 0.04(0.11) O1002 7 3.80(0.86) 2.79(0.77) 0.51(0.10) 0.46(0.08) 0.03 0.12(0.11) O114 5 2.90(0.58) 2.22(0.48) 0.37(0.11) 0.42(0.09) 0.13 0.12(0.12) O115 6 3.50(0.71) 2.42(0.58) 0.48(0.11) 0.43(0.09) 0.94 0.09(0.06) O801 10 4.30(1.13) 3.09(0.88) 0.42(0.11) 0.45(0.11) 0 0.13(0.12) O1101 15 5.10(1.50) 3.20(0.98) 0.44(0.11) 0.48(0.10) 0.17 0.08(0.08) O1204 9 2.00(0.47) 1.55(0.29) 0.25(0.10) 0.27(0.09) 0.13 0.08(0.15) O1401 16 2.60(0.58) 1.72(0.39) 0.27(0.09) 0.29(0.09) 0.24 0.08(0.10) O1503 8 3.70(0.83) 2.41(0.52) 0.45(0.10) 0.44(0.10) 0.9 0.08(0.05) O1504 11 4.90(1.32) 2.99(0.82) 0.44(0.11) 0.49(0.10) 0.02 0.11(0.07) O201 16 4.30(1.29) 2.69(0.82) 0.30(0.10) 0.35(0.11) 0.04 0.10(0.07) O701 15 5.90(1.66) 3.62(1.22) 0.45(0.11) 0.46(0.11) 0.52 0.01(0.03) O802 10 4.20(1.17) 2.81(0.93) 0.40(0.11) 0.41(0.10) 0.39 0.07(0.08) O1201 7 2.60(0.65) 2.05(0.54) 0.33(0.10) 0.34(0.10) 0.32 0.00(.08) O1402 16 3.10(0.75) 1.99(0.50) 0.27(0.10) 0.33(0.10) 0 0.24(0.11) O202 4 2.70(0.57) 2.13(0.48) 0.43(0.13) 0.38(0.10) 0.98 0.06(0.15)
136 Table 4 9 Molecular diversity of populations of Prunus geniculata N=number of samples in the population; N a =number of alleles; N eff =effective number of alleles; H O =observed heterozygosity; H E =expected heterozygosity; HWE= P value for the exact Fisher's test Chi squared for Hardy Weinberg equilibrium (P values in italic mean that the population is in equilibrium) ; F =inbreeding coefficient. Pop 101 105 106 107 116 119 N 3 16 15 12 7 3 N a 3.50(0.37) 5.87(1.17) 2.50(0.32) 4.50(0.98) 2.00(0.26) 2.00(0.18) N eff 2.96(0.35) 3.61(0.79) 2.00(0.27) 3.03(0.60) 1.70(0.20) 1.76(0.17) H O 0.75(0.13) 0.69(0.06) 0.63(0.15) 0.62(0.11) 0.58(0.16) 0.66(0.14) H E 0.63(0.03) 0.63(0.06) 0.43(0.09) 0.60(0.07) 0.37(0.09) 0.45(0.08) HWE 0.98 0.54 0 0.03 0.02 0.85 F 0.14(0.20) 0.09(0.04) 0.43(0.07) 0.04(0.16) 0.65(0.12) 0.64(0.12) Table 4 1 0. Wilcoxon bottleneck detection t est results for Asimina obovata Analyzes for three models of evolution were performed: the Infinite Allele Model (IAM), the Stepwise Mutation Model (SMM) and the Two Phased Model (TPM). For each population, the P value s for heterozygosity deficiency (left), and heterozygosity excess ( right ) are reported. The shape of the allele frequency distribution (AFD) is reported as L shaped ( normal ) or in Shifted Mode ( shifted ). pop IAM SMM TPM AFD O101 0.615/0.422 0.116/0.903 0.187/0.833 normal O103 1.000/ 0.000 0.903/0.116 0.987/ 0.016 normal O106 0.958/0.052 0.096/0.919 0.187/0.838 normal O107 0.983/ 0.041 0.883/0.137 0.919/0.096 normal O115 0.947/0.065 0.384/0.652 0.500/0.539 normal O121 0.983/ 0.041 0.116/0.903 0.312/0.721 normal O127 0.784/0.246 0.215/0.812 0.384/0.652 normal
137 Table 4 11. Wilcoxon bottleneck detection t est results for Ilex opaca Analyzes for three models of evolution were performed: the Infinite Allele Model (IAM), the Stepwise Mutation Model (SMM) and the Two Phased Model (TPM). For each population, the P value s for heterozygosity deficiency (left), and heterozygosity excess ( right ) are reported. The shape of the allele frequency distribution (AFD) is reported as L shaped ( normal ) or in Shifted Mode ( shifted ). pop IAM SMM TPM AFD A106 0.875/0.150 0.006 /0.995 0.013 /0.990 normal A108 0.972/ 0.037 0.679/0.371 0.769/0.273 normal A116 0.679/0.371 0.191/0.843 0.019/0.986 normal O1001 0.769/0.273 0.769/0.273 0.769/0.273 shifted O1002 0.312/0.721 0.052/0.958 0.065/0.947 normal O114 0.410/0.632 0.212/0.820 0.212/0.820 shifted O115 0.285/0.751 0.082/0.935 0.082/0.935 normal O801 0.714/0.326 0.082/0.935 0.170/0.849 normal O1101 0.787/0.248 0.082/0.935 0.150/0.875 normal O1204 0.972/ 0.039 0.765/0.289 0.593/0.468 normal O1401 0.679/0.371 0.230/0.808 0.156/0.875 normal O1503 0.455/0.589 0.064/0.975 0.013 /0.990 normal O1504 0.714/0.326 0.082/0.935 0.101/0.917 normal O201 0.371/0.679 0.027 /0.980 0.097/0.962 normal O701 0.589/0.455 0.125/0.898 0.179/0.849 normal O802 0.726/0.320 0.097/0.962 0.156/0.875 normal O1201 0.972/ 0.037 0.808/0.230 0.808/0.230 shifted O1402 0.843/0.191 0.097/0.962 0.273/0.769 normal O202 0.472/0.578 0.097/0.962 0.125/0.902 shifted
138 Table 4 12. Wilcoxon bottleneck detection t est results for Prunus geniculata Analyzes for three models of evolution were performed: the Infinite Allele Model (IAM), the Stepwise Mutation Model (SMM) and the Two Phased Model (TPM). For each population, the P value s for heterozygosity deficiency (left), and heterozygosity excess ( ri ght ) are reported. The shape of the allele frequency distribution (AFD) is reported as L shaped ( normal ) or in Shifted Mode ( shifted ). pop IAM SMM TPM AFD 101 0.628/0.421 0.726/0.320 0.769/0.273 shifted 105 0.980/ 0.027 0.230/0.808 0.421/0.628 normal 106 0.980/ 0.027 0.945/0.148 0.945/0.148 shifted 107 0.980/ 0.027 0.527/0.527 0.578/0.472 normal 116 0.992/ 0.015 0.976/ 0.039 0.976/ 0.039 shifted 119 0.945/0.148 0.945/0.148 0.945/0.148 shifted Table 4 13 Gene flow estimation between populations of Asimina obovata F ST is below, and R ST above diagonal. O101 O103 O106 O107 O110 O115 O121 O127 0.000 0.206 0.000 0.015 0.092 0.186 0.084 O101 0.152 0.076 0.000 0.139 0.095 0.064 0.000 O103 0.088 0.174 0.251 0.761 0.397 0.115 0.096 O106 0.059 0.148 0.110 0.067 0.112 0.220 0.087 O107 0.136 0.168 0.191 0.146 0.353 0.640 0.550 O110 0.120 0.187 0.125 0.118 0.212 0.343 0.212 O115 0.162 0.178 0.161 0.139 0.194 0.184 0.095 O121 0.175 0.111 0.176 0.137 0.209 0.181 0.139 O127
139 Table 4 14 Gene flow estimation between populations of Ilex opaca (both varieties). F ST is below, and R ST above diagonal. A106 A108 A116 O1001 O1002 O114 O115 O801 O1101 O1204 O1401 O1503 O1504 O201 O701 O802 O1201 O1402 O202 0.000 0.026 0.051 0.335 0.099 0.000 0.225 0.002 0.076 0.109 0.000 0.000 0.040 0.039 0.310 0.020 0.207 0.148 A106 0.051 0.000 0.049 0.299 0.077 0.017 0.170 0.000 0.050 0.070 0.072 0.000 0.010 0.005 0.272 0.005 0.169 0.174 A108 0.099 0.155 0.000 0.251 0.110 0.000 0.167 0.033 0.043 0.042 0.000 0.000 0.000 0.000 0.217 0.000 0.129 0.018 A116 0.044 0.041 0.100 0.000 0.000 0.030 0.000 0.209 0.000 0.000 0.081 0.000 0.000 0.000 0.000 0.000 0.000 0.000 O1001 0.083 0.045 0.142 0.024 0.000 0.278 0.000 0.457 0.000 0.000 0.329 0.111 0.137 0.124 0.000 0.048 0.000 0.000 O1002 0.074 0.080 0.143 0.065 0.113 0.039 0.000 0.239 0.000 0.000 0.096 0.000 0.029 0.023 0.019 0.000 0.013 0.000 O114 0.047 0.040 0.144 0.043 0.092 0.030 0.154 0.000 0.039 0.058 0.000 0.000 0.001 0.000 0.257 0.000 0.159 0.147 O115 0.059 0.000 0.133 0.017 0.026 0.036 0.035 0.297 0.000 0.000 0.194 0.061 0.091 0.079 0.000 0.007 0.000 0.000 O801 0.068 0.051 0.126 0.051 0.108 0.053 0.051 0.048 0.167 0.155 0.036 0.036 0.084 0.080 0.403 0.121 0.265 0.372 O1101 0.179 0.200 0.206 0.084 0.211 0.131 0.132 0.143 0.139 0.000 0.078 0.000 0.000 0.000 0.000 0.000 0.000 0.000 O1204 0.155 0.170 0.197 0.080 0.157 0.158 0.102 0.142 0.096 0.169 0.086 0.000 0.010 0.002 0.000 0.000 0.000 0.000 O1401 0.054 0.057 0.093 0.016 0.086 0.084 0.042 0.048 0.044 0.124 0.104 0.000 0.026 0.022 0.298 0.028 0.188 0.214 O1503 0.038 0.045 0.105 0.005 0.091 0.072 0.036 0.041 0.052 0.125 0.094 0.019 0.000 0.000 0.113 0.000 0.057 0.000 O1504 0.111 0.115 0.157 0.051 0.130 0.148 0.101 0.112 0.064 0.150 0.066 0.060 0.075 0.000 0.106 0.000 0.031 0.000 O201 0.057 0.044 0.108 0.003 0.067 0.093 0.039 0.039 0.039 0.121 0.091 0.031 0.031 0.053 0.094 0.000 0.024 0.000 O701 0.087 0.018 0.159 0.069 0.069 0.083 0.058 0.000 0.063 0.184 0.172 0.074 0.070 0.128 0.071 0.032 0.000 0.000 O802 0.106 0.052 0.183 0.032 0.092 0.121 0.044 0.046 0.083 0.118 0.131 0.054 0.086 0.079 0.044 0.074 0.000 0.000 O1201 0.137 0.113 0.186 0.039 0.121 0.188 0.105 0.100 0.117 0.166 0.072 0.071 0.090 0.088 0.063 0.141 0.087 0.000 O1402 0.063 0.097 0.099 0.009 0.078 0.064 0.114 0.064 0.029 0.157 0.094 0.053 0.045 0.019 0.041 0.099 0.108 0.135 O202
140 Table 4 15 Gene flow estimation between populations of Prunus geniculata F ST is below, and R ST above diagonal. 101 105 106 107 116 119 0.000 0.000 0.000 0.000 0.000 101 0.000 0.000 0.000 0.305 0.000 105 0.269 0.264 0.000 0.286 0.000 106 0.000 0.034 0.322 0.209 0.000 107 0.236 0.233 0.520 0.255 0.263 116 0.117 0.126 0.456 0.128 0.373 119
141 Figure 4 1. Distribution of sampled populations for Asimina obovata (orange circle) and Asimina incana (blue square).
142 Figure 4 2. Distribution of sampled populations for Ilex opaca var. arenicola (red circle), I. opaca var. opaca (blue square) and I. cassine (yellow triangle).
143 Figure 4 3. Distribution of populations of Prunus geniculata (red circles) and P. maritima (purple squares).
144 Figure 4 4. Geographic and genetic clustering of Asimina obovata K=4 was the optimal number of clusters. The bar plot is showing both species of Asimina for K=5, demonstrating the clear separation of the species. Then, K=2 5 is shown for Asimina obovata only. 6200 6300 6400 6500 6600 6700 6800 6900 0 2 4 6 8 10 Deviance Index Criterion DIC Number of clusters K K=2 K=3 K=4 K=5
145 Figure 4 5. Geographic and genetic clustering of Ilex opaca (both varieties) K=4 was the optimal number of clusters for Ilex opaca and I. cassine (the latter being one cluster not shown in the bar plot). 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.2 0.4 0.6 0.8 1 Deviance Index Creiterion DIC Number of clusters K K =2 K =3 K =4
146 Figure 4 6. Geographic and ge netic clustering of Prunus geniculata and P. maritima K= 5 was optimal. The DIC graph and K=2 4 Prunus geniculata bar plots show that K=3 is optimal for the endemic. 2250 2300 2350 2400 2450 2500 2550 0 2 4 6 8 Deviance Index Criterion DIC Number of clusters K K=3 K=4 K=2
147 Figure 4 7. Results of Mantel test for Asimina obovata Each point corresponds to a pair
148 Figure 4 8. Result of Mantel test for Ilex opaca Graph of all individual pairs (top) and of individuals from populations in Georgia and Florida only (bottom).
149 Figure 4 9. Result of Mantel test for Prunus geniculata Each point represents a pair of distance. Figure 4 10. Figure 4 6: Co ntributions of populations of Asimina obovata to allelic richness. Allelic diversity of each population (blue dots) is composed of its contribution to the total diversity (red bar) and its divergence from other pop ulations (green bar). -0.015 -0.01 -0.005 0 0.005 0.01 101 103 106 107 110 115 121 127 Crd Crs Crt
150 Figure 4 11. Co ntributions of populations of Ilex opaca (both varieties) to allelic richness. Allelic diversity of each population (blue dots) is composed of its contribution to the total diversity (red bar) and its divergence from other populations (green bar). Figure 4 12. Co ntributions of populations of Prunus geniculata to allelic richness. Allelic diversity of each population (blue dots) is composed of its contribution to the total diversity (red bar) and its divergence from other populations (green bar). -0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 Crd Crs Crt -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1 0.12 101 102 105 106 107 116 119 Crd Crs Crt
151 CHAPTE R 5 SYNTHESIS OF RESULTS : CONSERVATION IMPLI CATIONS FOR THE CENT RAL FLORIDA SC RUB Genetic diversity and habitat availability are two of the factors increasing the chances of long term survival of a species. In turn their reduction can lead to an increase in inbreeding, loss of fitness, inability to adapt to future changes, and even tually extinction ( Frankham, 2003 ; Frankham, 2005 ) Taxonomic U nits and C onservation The need to delimitate species in a standardized manner has great implications on conservation, and conservation application. Conservation planning is often based on the concept of species as a conservation unit. The list of threatened species, the species richness estimates, the species covered by legislation, the conservation of species diversity and species genetic diversity are all bases for conservation legislation, and a shortage of taxonomic information will cause serious problems for conservationist s ( Mace, 2004 ) Setting the conservation priority of lineages rather than non phy logenetic species concepts enables to manage evolutionary history and product ( Ryder, 1986 ; Moritz, 1994 ) as well as the potential for evolution in the future. This is a much more promising conservation strategy than one that does not take molecular systematics into consideration ( Soltis and Gitzendanner, 1999 ) Phylogenetic relationship and the diagnostic of ESU cannot be inferred from frequency and distance data, which is what united by shared characters. The sequence data used for phylogenetic inference is a precursor to phylogenetic reconstruction while distance based methods results are outcomes of this evolutionary process. They are unstable and dependent on patterns that may be unrelated to the speciation process ( Goldstein et al., 2000 ) However,
152 frequency based methods can be used to infer intra specific conservation units as they reflect the more rec evolutionary potential ( Ryder, 1986 ; Moritz, 1994 ; Goldstein et al., 2000 ) Addi tionally, the need to identify phylogen etic units to infer the sister taxa of the species of focus an important point of comparison for the levels of genetic diversity as it enables to control for their common evolutionary history ( Gitzendanner and Soltis, 2000 ) Conservation G enetics Conservation genetics is a disciplin e that encompasses genetic management of small populations, resolution of taxonomic uncertainties and management units. The better conserve it. There is evidence that gene tic factors can contribute to extinction risk, and that most threatened species show signs of genetic deterioration before they become extinct ( Frankham, 200 5 ) Genetic D iversity and P artitioning The impact of habitat reduction and fragmentation on the genetic diversity of species is debated in the literature. Several authors converge on the fact that habitat loss impacts fitness quicker than other diversi ty measures ( Lowe et al., 2005 ; Aguilar et al., 2008 ) But while Lowe find s that habitat fragmentation quick ly impacts inbreeding, both Aguilar Leimu and their colleagues did not find this correlation. Instead, they found that genetic diversity in the form of heterozygosity measures is affected by a reduction in population size and rarefaction of the species ( Lowe et al., 2005 ; Leimu et al., 20 06 ; Aguilar et al., 2008 ) These differences attenuate with the number of generations, as the fragmentation effects on g enetic diversity increases with time in plants ( Holderegger,
153 Kamm, and Gugerli, 2006 ; Aguilar et al., 2008 ) These discrepancies in the results of different studies on the effects of habitat loss and fragmentation on species genetic measures are emphasizi ng the importance of individual species studies for conservation. A ll these correlations are always stronger when populations are small and species threatened ( Frankham, 2005 ; Leimu et al., 2006 ) Allelic R ichness Conservation has f or a long time tried to maximize species richness and taxic diversity as a way to select areas of priority to allocate limited funds ( Vane Wright, Humphries, and Williams, 1991 ) However, allelic richness is also a measure of conservation interest and so Petit and his colleagues proposed to prioritize allelic richness in order to maximize the evolutionary response potential of a set of population s ( Petit, El Mousadik, and Pons, 1998 ) Allelic richness is dependent on effective population size, therefore capturing past demographic changes that have genetically affected a population or species. A rarefaction technique is used to compare allelic richness in samples of uneven sizes ( Hurlbert, 1971 ; El Mousadik and Petit, 1996 ) This analysis evaluates the allelic richne ss at each locus, and then averages it across loci. This richness is then partitioned into the allelic richness of the populations relative allelic abundance to others in the dataset, and its divergence from other populations (due to private alleles). Ge ne F low and I nbreeding Genetic exchangeability is an important criterion when planning population management and restoration. Genetic exchangeability is the ability of individuals with shared alleles from one population to replace or augment individuals fr om another population that have been decimated ( Crandall et al., 2000 )
154 Rare and threatened species tend to have small population sizes, and it is the case for our species of study. In the case of a lack of gene flow between populations, inbreeding depression can cause a loss of fitness and reproductive performance, especially in outcrossing species. This has been hypothe sized in the wild and tested in captive colonies experiments ( Frankham, 2005 ) Also, genetic drift can cause the fixation of some deleterious alleles as well as the decline of genetic diversity and loss of private alleles. These scenarios will make populations less likely to be able to respond to the selective pressure of environmen tal change, lowering the chance of the population and the species to adapt and survive on the long term ( Ellstrand, 1992 ; Ouborg and Treuren, 1994 ; Oostermeijer, Luijten, and den Nijs, 2003 ) The presence and maintenance of gene flow between populations of threatened species is therefore crucial to its conservation. Materials and Methods Central Florida S crub S pecies The taxic identity and geographic origin of the species will be discussed briefly for those species who have been studied phylogenetically using molecular tools. There are four species that were studied at the population level for my dissertation. These are Asimina obovata Ilex opaca (both the opaca and arenicola varieties), P olygala lewtonii and Prunus geniculata Additionally, other species endemic to the region have been studied and will be incorporated in this comparative study. Ziziphus celata ( Gitzendanner et al., 2011 ) and Conradina brevifolia ( Edwards, Soltis, and Soltis, 2008 ) were s t udied with microsatellites, the mole skink ( Eumeces egregious ), the Florida scrub lizard ( Sceloporus woodi ) and the sand skink ( Neoseps reynoldsi ) were part of a common
155 study using cyto chrome b ( Branch et al., 2003 ) Warea carteri using allozymes ( Evans et al., 2000 ) Comparison of R elative M easures of G enetic D iversity We will first compare the measures of diversity for the different populations of Asimina, Ilex, Polygala and Prunus according to ridges and a North South gradient. This will ena ble us to identify common areas of high diversity (observed heterozygosity, number of alleles, number of effective alleles, high contribution to allelic richness (relative to the total richness, and as a divergence from other populations) and major gene fl ow barriers that are common to the set of species. Then, we will map the populations of conservation interest, either for their unusually high or low genetic diversity, as well as major gene flow barriers. This will be done for our four species of study, before adding the other 6 species from the literature in trying to seek common patterns of genetic diversity and partitioning. Results Taxic R esolution and R elationship Polygala and Prunus have a Northeastern North America congener. Asimina could not be r econstructed into a resolved phylogeny and most likely has a very complex history of introgression and incomplete lineage sorting (K. Neubig, pers. comm.). This could be resolved by a population level sampling. Asimina was assu m ed to be sister to Asimina incana a very likely scenario given extant molecular work (K. Neubig, pers. comm.). Ilex opaca var. arenicola the central Florida scrub endemic, however, was found not to be a proper taxonomic entity as it was part of the same clade as Ilex opaca var. op aca so that both were treated together for the population genetic study. In this last chapter, focusing on the central Florida scrub conservation needs, only Floridian
1 56 populations of Ilex opaca were considered. Gene flow extending outside of Florida was m entioned, namely in the Apalachicola region, and between Florida and the Atlantic coast (Figure 5 1). For all four species of this dissertation, the overall levels of genetic diversity were equal or superior to the ones from their non threatened widespread congeners. Genetic P artitioning and G ene F low in the C entral Florida S crub Asimina obovata was the species with the most structure. Population 110, in Citrus county, showed signs of heterozygosity excess and gene flow barrier with all popu lat ions but 101 in the Lake Wales Ridge State Forest (LWRSF) and 107, Tiger Creek. The sampling was too small to test for a recent bottleneck. Population 103 (Pine Park, Lake Co.) also showed a reduced genetic diversity and the signs of a recent genetic bottleneck event. On the other hand, population 121 (Paynes Creek State Park, Hardee Co.) contained an exceptionally high number of alleles, but just as population 110, was highly isolated from all other populations. Finally, population 106, in the Archbold Biological Stat ion, the southernmost sampled point on the Lake Wales Ridge, was isolated from any genetic flow with the other populations (Table 6 1) Ilex opaca did not show very much partitioning within Florida, to the exception of the southern tip of the Lake Wales R idge, at Archbold Biological Station, where population 106 showed signs of genetic impoverishment and isolation from the rest of the species. Two major gene flow barriers isolate the Florida populations to the Atlantic coast to the East, and the Alabama an d Mississipi populations to the West (Table 5 1) Prunus geniculata show ed a high degree of complexity considering that it is the species with the smallest population sizes and the narrow natural distribution. Population 106, in the Archbold Biological S tation, despite average levels of diversity,
157 shows a significant isolation from the rest of the populations with F ST but not R ST and is the largest contributor of allelic richness in terms of divergence from other populations (private alleles). Populatio n 116, in the Allen David Broussard Catfi sh Creek Preserve State Park, is also highly isolated from others, even the other population in the same Park, just a very short distance away, and also contains a significant number of private alleles, making it es pecially important for the genetic contribution of the species and its potential for adaptation to future environmental changes (Table 5 1) For the other species studied in already published work, gene flow barriers were the only common measure that was exploitable. Studies used microsatellites for the plant species and Cytochrome b for the animal species. Both Ziziphus celata and Conradina br evifolia showed the same two areas with major among population variance: the Lake Wales Ridge State Forest, and an area around Sebring, a little south (about half way between Lake Wales Ridge State Forest and Archbold Biological Station). These gene flow b arriers occurred between geographically close population s 2). For the Cytochrome b study, all three reptiles showed a strong partitioning between the ridges they were sampled on: the Lake Wales Ridge and the Mount Dora Ridge. Also, both the Florida scrub lizard ( Scleroporus woodii ) and the sand skink ( Neoseps reynoldsi ) had their gene flow limited between the southernmost populations on the Lake Wales Ridge and the rest of the species. Finally, the sand skink had one further partitioning of its populations between the northern most part of the Lake Wales Ridge and the rest of that ridge, so that there were 4 clusters: Mo u nt Dora, and northern, central and sou thern Lake Wales Ridge (Figure 5 2).
158 Discussio n Taxic R esolution and R elationship For both Polygala and Prunus we found that the species endemic to the central Florida scrub was closely related to a species distributed in the Northeast of North America, supporting the hypothesis of a glacial refugium in Florida and the subsequent speci a tion and expansion of the eastern North America congener. If this is true, the two congeners would have been in contact until a fairly recent time (10,000 year ago). For Asimina the lack of molecular resolution in the phylogenetic work of K.M. Neubig and J.R. Abbott at the Florida Museum of Natural History Herbarium yielded very little resolution for the genus. Asimina obovata was thought to be most likely sister to A. incana in light of recent results, but a complex hi story of introgression and incomplete lineage sorting of this young genus in the region did not allow for a high certainty in the matter. Still, we assumed Asimina obovata and A. incana to be sister species, keeping in mind that a widespread congener not t hreatened by anthropogenic activity is a good comparison point to a threatened endemic even if it is not the closest congener, as long as most of their evolutionary history was shared, which we are certain is the case. Finally, Ilex opaca var. arenicola do es not seem to be a valid taxonomic entity and was merged with Ilex opaca var. opaca for the population level study. Their lack of reciprocal monophyly reveals that most likely the obvious and easily recognizable morphological differences between both vari eties are solely due to the plasticity of Ilex Finally, all species endemic to the central Florida scrub, a highly threatened and narrow ecosystem, were equally, or more diverse than their widespread congeners. These high levels of diversity seem to agre e with the hypothesis of a Florida refuge during the Pleistocene glacial cycles and a recolonization North after the last ice sheet
159 retreat 10,000 years ago. The founder effect resulting from the environmental pressure on North America during the Pleistoce ne would be too recent for the widespread temperate species to have recovered from it genetically, especially in longer lived species such as Prunus Ilex and Asimina Genetic P artitioning and G ene F low in the C entral Florida S crub Of all the species in t his study, Asimina obovata is the one that shows the most genetic partitioning and differences in genetic diversity amongst its populations. The two species on secondary ridges (other than Mount Dora or the Lake Wales Ridge), showed high levels or diversit y or heterozygosity, but gene flow isolation from each other and the rest of the sampled populations. The northernmost population on the Lake Wales Ridge shows signs of genetic distress (low diversity, recent bottleneck) while the southernmost population w as isolated from all others. However, we did not detect any major gene flow barrier between the two main ridges, as is observed in other species (see further). Ilex opaca despite a lack of genetic partitioning, reinforced the idea of a genetic isolation o f the southernmost sampled point along the Lake Wales Ridge, at the Archbold Biological Station, with the only populations showing signs of a genetic depauperization and recent past bottleneck. Prunus more complex in its genetic partitioning, confirmed th at the south of the Lake Wales Ridge was isolated from the rest of the distribution of the central Florida scrub endemics with a major gene flow, and a high level of private alleles in the Archbold Biological Station population. Another area of conservatio n interest for this species is the central region of the Lake Wales Ridge where populations show either a high level of diversity (Tiger Creek population), or signs of genetic distress such as population 116, with gene flow isolation from other popula tions low number of effective alleles and observed heterozygosity, and high
160 number of private alleles. This population is situated in the Allen David Broussard Catfish Creek Preserve State Park, and is closely surrounded by other populations in this area. This genetic stress might be due to a high fragmentation of the landscape and the inability of pollinators and seed dispersers to adapt efficiently. Once more, this central Lake Wales Ridge region seem to be an important locality in the conservation of this ce ntral Florida scrub endemic. The last species of this study, Polygala lewtonii also supports the importance of the central LWR with the presence of two populations of special conservation interest in the middle of the Lake Wales Ridge. Two populations, 1 15 and 118, both situated in the Allen David Broussard Catfish Creek Preserve State Park, contribute more than others in terms of allelic richness, but are both genetically isolated from all other populations, including from each other, despite them being geographically very closer. This might be a sign of decline of the pollinator or seed disperser populations in the region and should be investigated further. Moreover, Polygala lewtonii is the only species that exhibit s a strong gene flow barrier between t he Lake Wales and Mount Dora Ridges More sampling of Ilex on the Mount Dora Ridge might also reveal the same barrier. Commonalities with A lready P ublished S tudies O ther species from the literature reinforce s some of my findings Due to the nature of the data and their analyses, only the gene flow barriers could be consistently mapped and interpreted from the published studies (Figure 5 2). Ziziphus celata ( Gitzendanner et al., 2011 ) and Conradina brevifolia ( Edwards, Soltis, and Soltis, 2008 ) both show a gene flow barrier that is found in our species of study, situated in the central part of the Lake Wales Ridge. In the case of Ziziphus and Conradina this gene flow barrier is located within the Lake Wales Ridge State Forest while in the case of Polygala and
161 Prunus it is in the Allen David Broussard Catfish Creek Preserve Sta te Park, but these two locations are geographically near. This might be interpreted as the result of a high fragmentation of the landscape in this part of the Lake Wales Ridge ( Weekley, Menges, and Pickert, 2008 ) with pollinators and seed dispersers struggling to keep the populations connected. Conservation efforts should focus on maintaining or expanding existing, and creating new corridors that would enable gene flow between populations to distribute some priva te alleles to neighboring populations more efficiently. This area also has several other populations that do share their genes with the populations around them, and these corridors should be maintained as much as possible. The other gene flow barrier comm on to both Zizphus celata and Conradina brevifolia is situated in an area slightly south of this, in the Sebring area ( Edwards, Soltis, and Soltis, 2008 ; Gitzendanner et al., 2011 ) This area also seems to be problematic for pollinator and dispersers, and a more intense sampling of our species of study in this region (where it is lacking) could confirm this. The study of three reptiles revealed some common patterns with our species of studies, recovering three main gene flow barriers ( Branch et al., 2003 ) These gene flow barriers are slightly different in the sense that th ey occur between clusters of populations rather than within them, maybe corresponding more to natural barriers due to topography or distance rather than an obvious fine scale fragmentation in densely inhabited areas. The genetic isolation of the southern t ip of the Lake Wales Ridge from the rest of the central Florida scrub, as was observed for Ilex Prunus and Asimina was also found for the Florida scrub lizard and the sand skink. In terms of conservation management, it might be useful to try and purchase more land around the Archbold
162 Biological Station to keep this area genetically sustainable. The high number of private alleles in the plant species makes this area especially precious to the species concerned, and its isolation f ro m other populations coul d be problematic to its long term survival. However, it should not here by re jected the idea of a local adap tation of the populations on this part of the ridge, and so caution should be used before any transplantation or population augmentation is decided as it could result in outbreeding depression ( Ellstrand, 1992 ; Frankham, 2005 ) Another major gene flow barrier observed in all three reptiles is between the two ridges (Lake Wales Ridge and Mount Dora Ridge). This barrier is strong and was also observed in Polygala and to a lesser extent in Asimina In terms of conservation, each ridge should be managed separately and any translocation between ridges should be avoided. The same is true of the minor ridges for Asimina as these have a high content of private alleles, which might be the sign of local adaptation ( Petit, El Mousadik, and Pons, 1998 ) Finally, a strong gene flow barrier between the northern and central Lake Wales Ridge is found in the sand skink, but no other species. However, Ziziphus celata only have 1, is si tuated in the northern half of the Lake Wakes Ridge. This major genetic partitioning, found only in one species out of the nine encompassed here, reinforced the idea that despite the presence of some common patterns of special interest for conservation, th e study of a particular species and an individual conservation manag e ment plan should always be funded before any major decision is taken ( Menges et al., 2010 ) Each species genetic diversity partitioning is the result of some common
163 recent anthropogenic activity (mostly within the last 100 years for central Florida), but also of its unique combination of life history and evolutionary history. Conclusion s From the four species of th is study, and an additional five species from the literature, I have a good representation of the central Florida scrub ecosystem, with one annuals, five perennials and three animal species. I uncovered a very strong partitioning between the Lake Wales Rid ge and the side ridges to the West, which are important populations for the species, possibly the result of local adaptation. Also, a strong partitioning of the southernmost populations of the Lake Wales Ridge has important implications for the conservatio n planning. The central Lake Wales Ridge, especially the Lake Wales Ridge State Forest and the Allen David Broussard Catfish Creek Preserve State Park, seem to have a complex distribution of genetic diversity, with fine scale gene flow barriers, some isola ted populations with private alleles and some others having been severely impacted by anthropogenic activity. The management of this area should be done carefully and might necessitate several species specific population genetic studies with dense sampling Lastly, the partitioning between the two oldest ridges, the Lake Wales Ridge and Mount Dora Ridge is very strong for reptiles, most likely due to their very limited dispersal ability. It was also the case for Polygala maybe for the same reasons. This pa rtitioning was not obvious in other plant species, possibly due in part to an inappropriate sampling for this particular area. For plant species, studies on pollinators and seed dispersers should be encouraged, as it is essential to understand the mechanis ms that drive gene flow in order to truly understand the causal factors of the resulting genetic diversity and partitioning.
164 Table 5 1. Molecular diversity in four endemic to the central Florida scrub. F =inbreeding, N a =number of alleles, N eff =number of effective alleles, H E =expected heterzosygosity, H O = observed heterosygosity, PA=Contribution to allelic diversity due to divergence from other populations (private alleles), AR=Allelic richness, B=Presence of bottleneck. Measures of high or low are relative within each species. Not all species were sampled at all location. R idge P opulation Asimina Ilex Prunus Polygala N Mount Dora 124 B 125 high F B 126 low N eff 127 AR NW Other ridges 114 low H O high F AR SW 110 high H O neg F PA N Lake Wales Ridge 121 high N a ; N eff high H O B, PA, AR high H O 105 High N a N eff B, AR high H O ; low F PA 103 B 115 B, AR 118 high F AR 119 low N a N eff neg F PA 116 low N a N eff H O neg F B, PA 101 high H O neg F AR high H O AR 108 high N eff B, AR 102 AR low H O AR, PA 107 B B, PA PA S 106 AR high N a N eff F B, AR neg F B, PA
165 Figure 5 1. Conservation genetics synthesis of dissertation species. Populations with with weakened genetic diversity (dashed circle) and gene flow barriers (plain line) are represent ed on this map.
166 Figure 5 2. Map of gene flow barriers from other species endemic to the Florida scrub. The populations represented by a purple diamond at the ones sampled for the species from Figure 6 1. Dashed lines represent population variance from geographically close populations, plain lines represent gene flow barriers between regions composed of several populations.
167 CHAPTER 6 MICROSATELLITE MARKE R DEVELOPMENT FOR TH E FEDERALLY LISTED PRUNUS GENICULATA (ROSACEAE) The central Florida scrub is one of the top biodiversity hotspots in North America. However, it is highly threatened, with approximately 90% of its habitat having been urbanized or converted to agricultural lands, This ancient dune ecosystem n ow exists as a series of island ridges that provides habitat formany species of plants and animals. Twenty five and 24, respectively are listed as threatened or endangered ( Christman and Judd, 1990 ) Among th e federallylisted species is Prunus geniculata Harp. (Rosaceae), the scrub plum, known from only a few sites on the Lake Wales Ridge. The scrub plum is a shrub up to 2 m tall, heavily branched, w ith strongly zigzag twigs and spiny lateral branches. Its deciduous leaves are finely toothed. The five petalled whi te flowers bloom in late winter when the plant is leafless, and the fruit is a small red plum ( Wunderlin, 1998 ) The amount of f lowering and fruiting depends heavily on fire frequency and intensity Prunus geniculata is andromonoecious, with both male and bisexual flowers on the same plant. The flowers produce a strong fragrance and are pollinated by a variety of insects The plant s are thought to be self incompatible. A molecular phylogenetic study places Prunus geniculata as sister to P maritima from the northeastern US ( Shaw and Sma ll, 2005 ) Including Prunus maritima in our study of Prunus gen i culata establishes an evolutionary context that allows for more significance in the results. Investigating levels of genetic diversity in Prunus geniculata with Prunus maritima will therefore serve as an important advancement in our understanding of the Lake Wales Ridge endemics and its conservation needs.
168 Methods and Results Leaf samples were collected in the field and dried in silica gel ( Table 6 2 for collection and voucher information). DNA was extracted using a modified CTAB DNA extraction protocol ( Doyle and Doyle, 1987 ) A protocol for constructing a microsatellite library has been designed and optimized in our lab, where we have successfully developed primers for about 10 species ( Edwards et al., 2007 ) Briefly, 5 g of genomic DNA of Prunus geniculata were digested overnight using the restriction enzyme Sau 3AI. A linker, with known sequence, was created and ligated onto the digested DNA. After denaturing t he DNA, a biotinilated (CA) 8 microsatellite probe was added and hybridized fragments enriched by binding to streptavidin coated magnetic beads (Promega Corp., Madison WI, USA). The repeat enriched DNA fragments were then separated from the beads and amplif ied. The resulting PCR product was then cloned using a TOPO TA pCR4 TOPO cloning kit (Invitrogen, Carlsbad, CA, USA). Colonies were screened in two different PCRs, one with M13 F and (CA) 8 and one with M13 R and (CA) 8 Colonies with bands over 200 bp in one reaction were then sent for sequencing. In total, 96 colonies were sequenced, and the resulting sequences were used to design primer CACGACGTTGTAAAACGAC forw ard primer to allow for labeling with a tailed fluorescent dye. Thirty six primer pairs were screened in both Prunus species for amplification and informative polymorphism, both within and between species ( Table 6 3 for a list of monomorphic primers). Ei ght primer pairs were selected and used for population level studies (Table 6 1). For amplification, we used a master mix of 1 M Betaine, 1.5 mM MgCl 2 0.1 M dNTPs, 0.5 M forward and reverse primers, 0.5 M of either 6 FAM
169 VIC NED PET labeled M13 p rimer, and 0.2 unit Taq polymerase in 10 l reactions per sample. All loci were amplified under the same optimal conditions, requiring 3 min at 95C, followed by 35 cycles of 45 sec at 95 C, 1 min 15 sec at 52C, and 1 min 15 sec at 72C, with a final st ep of 20 min at 72C. PCR products were stored at 4C. Products were genotyped on an ABI 3730 DNA analyzer (Applied Biosystems, Carlsbad, CA) at the Interdisciplinary Center for Biotechnology Research at the University of Florida. Resulting peaks were scor ed manually with GeneMarker 1.6 (Soft Genetics, State College, PA). These eight loci were investigated in all individuals sampled for both Prunus geniculata and P. maritima for a total of 96 individuals from 11 populations (Table 6 1 and 6 2). All results were checked for scoring errors and null alleles in MICROCHECKER 2.2.3 ( Hutchinson, Wills, and Shipley, 2004 ) SpaGeDi 1.3 ( Hardy and Vekemans, 2002 ) was used to do a permutation test of R ST across all populations to determine the mutation model for these loci. All loci showed a highly significant result, indicating a stepwise mutation model (P=0.0072). GENEPOP 4.0 ( Raymond and Rousset, 1995 ; Rousset, 2008 ) was used to test for li nkage disequilibrium for all pairs of GenAlEx 6.3 ( Peakall and Smouse, 2006 ) was used to estimat e the average number of alleles per locus for each population (Table 2), which is reasonably high even in the smallest populations. Lastly, all of the loci could also be amplified in Prunus umbellat a P. angustifolia and P. cerasifera and not as strongl y but still producing clear bands in P. americana Prge 19, 23, 26, 27, and 28 amplified well in P. caroliniana P. serotina P. persica and
170 P. campanulata Prge 22 amplified in P. caroliniana as did Prge 29 in P. serotina and P. persica These successfu l amplifications demonstrate that these primers have great potential for future studies of North American Prunus at the population level. Conclusions We developed eight loci that show variability at the population level for both Prunus geniculata and P. maritima These loci will be prime tools for studies of genetic diversity in these two rare species, providing information that will help in their conservation. Also, the successful amplification of most or all loci in several other North American Prunus s pecies will be useful in future population level studies.
171 Table 6 1. Loci developed for P. geniculata and P. maritima and their characterization Name Sequence Repeat Product size (bp) Number of alleles GenBankID Prge 18 F CGACCTGAGACCAGAATCATA (CT) 11 370 27 HQ230802 Prge 18 R AGCAAATAAATAGATGGTTACACG Prge 19 F GAACCTGTTGAGGTACTGCTG (TG) 10 (AG) 4 213 5 HQ230803 Prge 19 R AATTATCCTGCACAAGCACCA Prge 22 F TTTAGCCCTCTTTGGTTCGAG (AC) 8 168 6 HQ230804 Prge 22 R GGGATCAATGTCTGGTTGTGA Prge 23 F TGTCTCTGGCTGAAGATTTGA (TG) 10 163 6 HQ230805 Prge 23 R GCCTGCCATCTATATCCTGG Prge 26 F TGCCATTTATCATTAGCTTCAC (CT) 20 (CA) 17 238 23 HQ230806 Prge 26 R GTGTAGAATCAAGCATAGTTGTGT Prge 27 F ATTTGTGGTTTGTACCTTGCG (CA) 9 286 9 HQ230807 Prge 27 R AGCTTGGGATCATTTGGATGG Prge 28 F AACACTTTGTCTGTCCACTGT (AG) 11 (AC) 3 301 17 HQ230808 Prge 28 R TCCGGCATCAGGAATTGTATC Prge 29 F TTAAACAACGTGTCGTCATCC (CT) 14 TT(CT) 3 (CA) 9 257 18 HQ230809 Prge 29 R GAAGTCATCACCCAGTAGGAC Table 6 2. Characterization of populations of Prunus geniculata and P. maritima var. maritim a Species Pop Pop. location Pop. size Mean number of alleles per locus H E P. geniculata 101 Polk, FL 3 3.5 0.63 105 Lake, FL 16 5.9 0.64 106 Highlands, FL 15 2.5 0.42 107 Polk, FL 12 4.5 0.58 116 Polk, FL 7 2 0.35 P. maritima 1202 Barnstable, MA 9 2.2 0.36 1203 Barnstable, MA 10 4.2 0.55 1205 Barnstable, MA 9 3.9 0.51 901 Sussex, DE 3 1.8 0.33 902 Sussex, DE 5 2.2 035 903 Sussex, DE 7 2.7 0.40 Vouchers for populations (all vouchers are deposited in the FLAS herbarium) : C. Germain Aubrey n.s. (101), J. Richard Abbott 22697 (105), C. Germain Aubrey 44 (106), J. Richard Abbott 226 78 (107), C. Germain Aubrey 45 (116), I. Kadis 1596 (1202), I. Kadis 1597 (1203 a nd 1205), W. Knapp n.s. (901), W. Knapp n.s. (902) and W. Knapp n.s. (903). Locations for populations : L ocations of the populations are approximate, as P. geniculata is a federally listed species. P. maritima voucher for pop1202: 41.75720N 70.11780W; po p1203 + 1205: 41.7847N 70.0312W; pop 901: 38.795 N 51.251389 W; pop 902: 38.786111 N 90.251667 W; pop 903: 38.605 N 22.250833 W
172 Table 6 3. Monomorphic loci that amplify in Prunus geniculata and/or P. maritima Name Sequence Repeat Product size (bp) Prge 14 F CCCTCTCTAGTCTTCACTCCA (CTT) 4 CGT(CTT) 6 339 Prge 14 R CACAAATACTCAATACAAGCCCA Prge 15 F CTCCTCCTACAGACAGACAGA (CTT) 12 241 Prge 15 R AGACTTAGAAAGAAGGCGGTG Prge 16 F AGCTTGGGATCAAAGGTCATC (TG) 12 231 Prge 16 R AAAGGCAGGTATGCAAAGTCT Prge 21 F TCAACTGGGTCTGTAACCAAC (GT) 7 GA(GT) 3 GC(GT) 7 215 Prge 21 R CCCTCTCAACCATTCAACCAT Prge 36 F GCAAATGCAAGTGGAAGAATC (CA) 9 CGCA(CT) 7 194 Prge 36 R AAATCCCGAGCACATCTGAAT
173 CHAPTER 7 CONCLU DING REMARKS Given that Florida is one of the states with the highest rate of growth (almost 18% increase in population in the last decade), it is critical to manage natural areas efficiently, especially in the central Florida scrub, an area of exceptional endemism and a unique ecosystem. In order to best conserv e any ecosystem, it is essential to understand the history of its formation, as well as the more recent impacts of anthropogenic pressure on the species and habitats. This was the ultimate goal of this project. I selected five species representative of the central Florida scrub and reconstructed their phylogenies, testing hypotheses of geographical origins through phylogeny reconstruction, topology modeling and the approximate unbiased test. I found that Prunus geniculata and Polygala lewtonii were placed sister to an eastern North American species. I also found that Ilex opaca var. arenicola was not an appropriate taxonomic entity but polyphyletic to Ilex opaca var. opaca This species was sister to a clade distributed in the eastern half of N orth America as well. Ilex opaca was considered as including both varieties for the rest of the study. I was unsuccessful at reconstructing a phylogeny of Asimina and the Persea tree failed to rejec t either hypotheses tested ( Chapter 3). For three of my f ive species, I therefore successfully advanced our phylogenetic knowledge of their histories. The supported scenario is the one in which the central Florida scrub served as a refuge for more northerly distributed species, which then recolonized their natur al habitats after the ice sheet retreated at the end of the Pleistocene ( Watts, 1975 ; Webb and Myers, 1990 )
174 Seeking to use microsatellites for the population level study of our speci es, I first did a study on the marker itself. For Polygala lewtonii and its nearest congener P. polygama I developed and amplified microsatellites as is widely done in the literature, but also sequenced the fragments. I then compared the fragment length d ata with the repeat number and the flanking region sequence to evaluate how it was influencing the outcome in terms of levels and patterns of genetic diversity, and conservation needs. I found that flanking regions of microsatellites were highly polymorphi c, recovering more alleles and a higher expected heterozygosity than using only fragment size or repeat number. They were especially useful in assigning one ambiguous population to the widespread species with confidence, revealing a great potential for low er level phylogenetic or phylogeography studies ( Chatrou et al., 2009 ) Finally, the comparison of fragment lengths (including the flanking region) and repeat lengths (without the flanking region) uncovered some serious discrepancies in the r esults. A set of molecular diversity indices varied significantly, both in terms of absolute and relative values. A few indices, however, did not seem too affected, at least in terms of relative values, by the presence of the flanking region in the fragmen t genotyped. The combination of these linked markers were most useful to the interpretation of the biology of Polygala lewtonii as they are not believed to reflect the same timespans within their history ( Blankenship, May, and Hedgecock, 2002 ) Recommendations for microsatellite primer design and treatment were listed in order to avoid biasing the results of population level studies. Three additional species were included in a microsatellite population level study of the central Florida scrub, comparing genetic diversity of the endemic to their widespread congener. All s pecies were found to be equally or more diverse than their widespread
175 relatives. Then, the different populations of the endemics were compared to identify areas of greatest contribution to the genetic and allelic diversity of the species, as well as areas bearing the trace of a recent genetic bottleneck, most likely resulting from anthropogenic activity. Finally, major gene flow barriers were located and any commonalities between the species were discussed. The last part of this study applies my findings t o the conservation needs to the region. Pooling the findings from Asimina obovata Ilex opaca Prunus geniculat a and Polygala lewtonii as well as mining the literature for other similar genetic studies of species endemic to the Florida scrub, I added five species to my dataset: Florida Ziziphus, Ziziphus celata ( Gitzendanner et al., 2011 ) short leaved rosemary Conradina brevifolia ( Edwards, Soltis, and Soltis, 2008 ) mole skink Eucemes egregious Florida scrub lizard Scleroporus woodii and sand skink Neoseps reynoldsi ( Branch et al., 2003 ) I found some strong commonalities in the importance of the middle part of the Lake Wales Ridge, which was a separate area of diversity for the sand skink, and had a complex history of allelic di stinction, bottlenecks and fine scale partitioning for all of my study species, as well as Ziziphus and Conradina Furthermore, the Lake Wales Ridge southernmost populations of all species were distinctive in some way (having gone through a bottleneck and/ or having a high private alleles content) and were isolated from the rest of the populations on the ridge. Noticeably also, a gene flow barrier between the two oldest ridges (the Lake Wales Ridge and the Mount Dora Ridge) was present for the annual Polygal a lewtonii and for all three reptiles, but not significantly for any other species. Finally, the two smaller ridges that were sampled for Asimina obovata showed an allelic richness departure from other populations in the form of private alleles
176 and were iso lated from each other and the main ridges, making them of particular interest for the potential of the species to adapt to environmental changes in the future. I believe this multi species approach to the conservation of a particular ecosystem is more app ropriate than a single species study would have been, and my findings will be useful for the advancement of our knowledge on each of the species separately, as well as the central Florida scrub as a whole. I would like to conclude this dissertation by cit ing the great opportunities for further research that this study has revealed. In terms of phylogenetic and phylogeography work, Ilex opaca needs to be revised as a species, as these preliminary results suggest that interesting patterns could be found with additional sampling The Asimina phylogeny is far from being complete, and Kurt Neubig and Richard Abbott from the Florida Museum of Natural History Herbarium have been working on it. A population level study encompassing several species might shed some l ight of a messy phylogenetic history. Prunus geniculata is also a species that requires more phylogenetic work, possibly through the addition of more samples and of a low copy nuclear gene. Concerning the population level part of this project, some additi onal populations should be genotyped, especially in the areas of conservation interest such as the central part of the Lake Wales Ridge. Also, when possible samples from the Mount Dora ridge should be added to the datasets to test for the gene flow barrier between both ridges, as in Ilex opaca for example. Finally, sampling should be sought in unprotected areas. Almost all samples were collected in parks, research station s or preserves, and all had some degree of protection and land management It would be
177 interesting to sample additional populations on unprotected land in order to compare the impact of anthropogenic activity (and protection) on the genetics of populations of endemic species, and help orientate the next land acquisition for conservation purposes.
178 LIST OF REFERENCES A BBOTT J. R. 2009. Phylogeny of the Polygalaceae and a revision of Badiera PhD, University of Florida, Gainesville, FL. A BBOTT J. R., B. S. C ARLSWARD K. M. N EUBIG W. S. J UDD W. M. W HITTEN and N. H. W ILLIAMS in prep. Phylogeny and biogeography of North American Polygalaceae: notes on the disintegration of Polygala with 4 new genera for the Flora of North America. A BLETT G., H. H ILL and R J. H ENRY 2006. Sequence Polymorphism Discovery in Wheat Microsatellite Flanking Regions using Pyrophosphate Sequencing. Molecular Breeding 17: 281 289. A BRAHAMSON W. 1984a. Post Fire Recovery Of Florida Lake Wales Ridge Vegetation. American Journal Of Botany 71: 9 21. ______. 1984b. Species Responses To Fire On The Florida Lake Wales Ridge. American Journal Of Botany 71: 35 43. A CKERLY D. D. 2004. Adaptation, Niche Conservatism, and Convergence: Comparative Studies of Leaf Evolution in the California C haparral. The American Naturalist 163: 654 671. A DAMS R. I., K. M. B ROWN and M. B. H AMILTON 2004. The impact of microsatellite electromorph size homoplasy on multilocus population structure estimates in a tropical tree (Corythophora alta) and an anadrom ous fish (Morone saxatilis). Molecular Ecology 13: 2579 2588. A GUILAR R., M. Q UESADA L. A SHWORTH Y. H ERRERIAS D IEGO and J. L OBO 2008. Genetic consequences of habitat fragmentation in plant populations: susceptible signals in plant traits and methodological approaches. Molecular Ecology 17: 5177 5188. A L R ABAB AH M. A., and C. G. W ILLIAMS 2002. Population synamics of Pinus taeda L. based on nuclear microsatellites. Forest Ecology and Management 163: 263 271. A NMARKRUD J. A., O. K LEVEN L. B ACHMANN and J. T. L IFJELD 2008. Microsatellite evolution: Mutations, sequence variation, and homoplasy in the hypervariable avian microsate llite locus HrU10. Bmc Evol Biol 8: 138. A RBOGAST B. S., and G. J. K ENAGY 2001. Comparative phylogeography as an integrative approach to historical biogeography. Journal of Biogeography 28: 819 825. A SHE W. W. 1924. Notes on Shrubs of the Southeastern S tates. Elisha Mitchell Scientific Society 40: 44.
179 ______. 1925. Ilex opaca var. arenicola Charleston Museum Quaterly Letter 1: 31. A VISE J. 2000. Phylogeography: the history and formation of species. Harvard University Press. A VISE J., and W. N ELSON 19 89. Molecular genetic relationship of the extinct dusky seaside sparrow. Science 243. A VISE J., J. A RNOLD R. B ALL E. B ERMINGHAM T. L AMB J. N EIGEL C. R EEB et al. 1987. Intraspecific Phylogeography: The Mitochondrial DNA Bridge Between Population Gene tics and Systematics. Annual Review Of Ecology And Systematics 18: 489 522. A VISE J. C., and B. R IDDLE 2009. Phylogeography: Retrospect and Prospect. Journal of Biogeography 36: 3 15. A VISE J. C., G. S. H ELFMAN N. C. S AUNDERS and L. S. H ALES 1986. Mitochondrial DNA differentiation in North Atlantic eels: Population genetic consequences of an unusual life history pattern. Proc Natl Acad Sci USA 83: 450 4364. B ALDWIN B. G., and R. H. R OBICHAUX 1995. Historical biogeography and ecology of the Hawaiian silversword alliance (Asteraceae): new molecular phylogenetic perspectives. In W. L. Wagner AND V. Funk [eds.], Hawaiian Biogeography: Evolution on a Hot Spot Archipelago, 259 287. Smithsonian Institution Press. B ERRY E. 1916. The physical cond itions and age indicated by the florida of the Alum Bluff Formations. Geological Survey Professional Papers 98: 41 53. B LANKENSHIP S. M., B. M AY and D. H EDGECOCK 2002. Evolution of a Perfect Simple Sequence Repeat Locus in the Context of Its Flanking Sequence. Molecular Biology and Evolution 19: 1943 1951. B LATTNER F. R. 1999. Direct amplification of the entire ITS region from poorly preserved plant material using recombinant PCR. Biotechniques 27: 1180 1186. B ORTIRI E., and D. P OTTER unpublished. E volution of Morphology in Prunus B ORTIRI E., S. O H J. J IANG S. B AGGETT A. G RANGER C. W EEKS M. B UCKINGHAM et al. 2001. Phylogeny and systematics of Prunus (Rosaceae) as determined by sequence analysis of ITS and the chloroplast trnL trnF spacer DNA. Systematic Botany 26: 797 807. B RANCH L., A. C LARK P. M OLER and B. B OWEN 2003. Fragmented landscapes, habitat specificity, and conservation genetics of three lizards in Florida scrub. Conservation Genetics 4: 199 212. B RIZICKY G. K. 1964. The genera of Celastrales in the southeastern United States. Journal of the Arnold Arbo retum 45: 229.
180 B ROADHURST L., and D. C OATES 2004. Genetic divergence among and diversity within two rare Banksia species and their common close relative in the subgenus Isostylis R.Br. (Proteaceae). Conservation Genetics 5: 837 846. B URBRINK F. T., R. L AWSON and J. B. S LOWINSKI 2000. Mitochondrial DNA phylogeography of the polytypic North American rat snale ( Elaphe obsoleta ): a critique of the subspecies concept. Evolution 54: 2107 2118. C ADOTTE M. W., B. J. C ARDINALE and T. H. O AKLEY 2008. Evolutio nary history and the effect of biodiversity on plant productivity. Proceedings of the National Academy of Sciences 105: 17012 17017. C AVENDER B ARES J., K. K OZAK and P. F INE 2009. The merging of community ecology and phylogenetic biology. Ecology Letters 12: 693 715. C EGELSKI C. C., L. P. W AITS and N. J. A NDERSON 2003. Assessing population structure and gene flow in Montana wolverines (Gulo gulo) using assignment based approaches. Molecular Ecology 12: 2907 2918. C HANDERBALI A., H. VAN DER W ERFF and S. R ENNER 2001. Phylogeny and historical biogeography of Lauraceae: Evidence from the chloroplast and nuclear genomes. Annals Of The Missouri Botanical Garden 88: 104 134. C HANG C. S., H. K IM and T. Y. P ARK 2003. Patterns of allozyme diversity in several selected rare species in Korea and implications for conservation. Biodiversity And Conservation 12: 529 544. C HARLESWORTH D., and B. C HARLESWORTH 1990. Inbreeding Depression with Heterozygote Advantage an d its Effect on Selection for Modifiers Changing the Outcrossing Rate. Evolution 44: 870 888. C HATROU L. W., M. P. E SCRIBANO M. A. V IRUEL J. W. M AAS J. E. R ICHARDSON and J. I. H ORMAZA 2009. Flanking regions of monomorphic microsatellite loci provide a new source of data for plant species level phylogenetics. Molecular Phylogenetics And Evolution 53: 726 733. C HEN J. Q., L. L I J. L I and H. W. L I 2009. Bayesian Inference of nrDNA ITS Sequences from Machilus (Lauraceae) and its Systematic Significanc e. Acta Botanica Yunnanica 2009 02. C HIANG T. Y., B. A. S CHAAL and C. I. P ENG 1998. Universal primers for amplification and sequencing a noncoding spacer between the atpB and rbcL genes of chloroplast DNA. Botanical Bulletin of Academia Sinica 39: 245 2 50. C HRISTMAN S., and W. J UDD 1990. Notes on plants endemic to Florida scrub Florida Scientist 53: 52 73.
181 C HRISTMAN S. A J., WS. 1990. Notes on plants endemic to Florida scrub Florida Scientist 53: 52 73. C LARK A., B. B OWEN and L. B RANCH 1999. Effec ts of natural habitat fragmentation on an endemic scrub lizard ( Sceloporus woodi ): an historical perspective based on a mitochondrial DNA gene genealogy. Molecular Ecology 8: 1093 1104. C ORNUET J., and G. L UIKART 1996. Description and power analysis of two tests for detecting recent population bottlenecks from allele frequency data. Genetics 144: 2001. CPC. Polygala lewtonii Website http://www.centerforplantconservation.org/ASP/CPC_ViewProfile.asp?CPCNum= 3569 2005]. C RANDALL K. A., O. R. B ININDA E MONDS G. M. M ACE and R. K. W AYNE 2000. Considering evolutionary processes in conservation biology. Trends in ecology & evol ution (Personal edition) 15: 290 295. C ULLINGS K. W. 1992. Design and testing of a plant specific PCR primer for ecological and evolutionary studies. Molecular Ecology 1: 233 240. C ULVER M., M. M ENOTTI R AYMOND and S. O B RIEN 2001. Patterns of Size Homo plasy at 10 Microsatellite Loci in Pumas (Puma concolor). Mol Biol Evol 18: 1151 1156. C UNNINGHAM C. W., and T. M. C OLLINS 1994. Developing model systems for molecular biogeography: Vicariance and interchange in marine invertebrates. Experienta Basel Supplementum 69: 405. D ELCOURT H., P. D ELCOURT V. M. B RYANT and R. G. H OLLOWAY 1985. Quaternary Palynology and Vegetational History of the Southeastern United States, Pollen records of late Quaternary North American sediments, 1 37. D EMESURE B., N. S ODZI and R. J. P ETIT 1995. A set of universal primers for amplification of polymorphic non coding regions of mitochondrial and chloroplast DNA in plants. Molecular Ecology 4: 129 131. D OBSON A., J. R ODRIGUEZ W. R OBERTS and D. W ILCOVE 1997. Geographic distribution of endangered species in the United States. Science 275: 550 553. D OLAN R., R. Y AHR E. M ENGES and M. H ALFHILL 1999. Conservation implications of genetic variation in three rare species endemic to Florida rosemary scrub. American Journal Of Botany 86: 1556 1562. D ONOGHUE M. J., and B. R. M OORE 2003. Toward an Integrative Historical Biogeography. Integrative and Comparative Biology 43: 261 270.
182 D OYLE J., and J. D OYLE 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11 15. D RAKE J. A. 1991. Community Assembly Mechanics and the Structure of an Experimental Species Ensemble. The American Naturalist 137: 1 26. D RUMMOND A., B. A SHTON S. B UXTON M. C HEUNG A. C OOPER J. H ELED M. K EARSE et al. 2010. Geneious v5.1. website: http://www.geneious.com D URAND E., F. J AY O. G AGGIOTTI and O. F RANCOIS 2009. Spatial inference o f admixture proportions and secondary contact zones. Molecular Biology and Evolution 26: 1963. E DWARDS C., D. S OLTIS and P. S OLTIS 2006. Molecular phylogeny of Conradina and other scrub mints (Lamiaceae) from the southeastern USA: Evidence for hybridization in Pleistocene refugia? Systematic Botany 31: 193 207. ______. 2007. Isolation, characterization and cross species amplifications of microsatellite loci from Conradina MOLECULAR ECOLOGY NOTES in press. E DWARDS C. E., D. E. S OLTIS and P. S. S OLTIS 2008. Using patterns of genetic structure based on microsatellite loci to test hypotheses of current hybridization, ancient hybridization and incomplete lineage sorting in Conradina (Lamiaceae). Molecular Ecology 17: 5157 5174. E DWARDS C. E., M. A RAKAKI P. F. Q UINTANA A SCENCIO D. E. S OLTIS and P. S. S OLTIS 2007. Isolation and characterization of microsatellite loci from the endangered highlands scrub hypericum (Hypericu m cumulicola). MOLECULAR ECOLOGY NOTES 7: 1135 1137. E L M OUSADIK A., and R. P ETIT 1996. Level of genetic differentiation for allelic richness among populations of the argan tree ( Argania spinosa (L.) Skeels) endemic to Morocco. Theoretical and Applied Ge netics 92: 832 839. E LLSTRAND N., and D. E LAM 1993. Population Genetic Consequences Of Small Population Size Implications For Plant Conservation. Annual Review Of Ecology And Systematics 24: 217 242. E LLSTRAND N. C. 1992. Gene flow by pollen: implicat ions for plant conservation genetics. Oikos 63: 77 86. E PPERSON B. K. 2005. Mutation at high rates reduces spatial structure within populations. Molecular Ecology 14: 703 710. E STOUP A., P. J ARNE and J. M. C ORNUET 2002. Homoplasy and mutation model at microsatellite loci and their consequences for population genetics analysis. Molecular Ecology 11: 1591 1604.
183 E VANS M., R. D OLAN E. M ENGES and D. G ORDON 2000. Genetic diversity and reproductive biology in Warea c arteri (Brassicaceae), a narrowly endemic Florida scrub annual. American Journal Of Botany 87: 372 381. E VANS M. E. K., E. S. M ENGES and D. R. G ORDON 2003. Reproductive biology of three sympatric endangered plants endemic to Florida scrub. Biological Co nservation 111: 235 246. E XCOFFIER L., G. L AVAL and S. S CHNEIDER 2005. Arlequin (version 3.0): An integrated software package for population genetics data analysis. Evolutionary Bioinformatics Online 1: 47 50. F ALUSH D., M. S TEPHENS and J. P RITCHARD 2003. Inference of Population Structure Using Multilocus Genotype Data: Linked Loci and Correlated Allele Frequencies. Genetics 164: 1567 1587. ______. 2007. Inference of population structure using multilocus genotype data: dominant markers and null allele s. MOLECULAR ECOLOGY NOTES 7: 574 578. F ISHER Calculation of P. Journal of the Royal Statistical Society 85: 87 94. FNAI. 2000a. Lewton's milkwort Website http://www.fnai.org/FieldGuide/pdf/Polygala_lewtonii.pdf ______. 2000b. Lewton's milkwort. Florida Natural Areas Inventory F RANKHAM R. 2003. Genetics and conservation biology. Comptes Rendus Biologies 326: S 22 S29. F RANKHAM R. 2005. Genetics and extinction. Biological Conservation 126: 131 140. F RANKHAM R., J. D. B ALLOU and D. A. B RISCOE 2002. Introduction to conservation genetics. Cambridge University Press, Cambridge. G ARCIA C HAVEZ A., C. R INCON L OPEZ R. S ALGADO G ARCIGLIA H. A. M ARQUEZ S ANTACRUZ and C. M. P RIETO B ARAJAS unpublished. Phylogenetic analysis of genus Persea collected in west area of Michoacan, Mexico. G ERMAIN A UBREY C., P. S OLTIS D. S OLTIS and M. G ITZENDANNER 2011. Microsatellite Ma rker Development for the federally listed Prunus geniculata (Rosaceae). American Journal Of Botany Primer Notes and Protocols Accepted. G ERMAIN A UBREY C., P. S OLTIS D. S OLTIS and M. G ITZENDANNER in prep a. Isolation and characterization of microsatelli te primers for both varieties of Ilex opaca (Aquifoliaceae).
184 G ERMAIN A UBREY C., D. S OLTIS P. S OLTIS and M. G ITZENDANNER in prep b. Isolation and characterization of microsatellites primers for the Florida endemic Asimina obovata (Annonaceae). G ERMAIN A UBREY C. C., P. S. S OLTIS K. M. N EUBIG T. T HURSTON D. E. S OLTIS and M. A. G ITZENDANNER in prep c. Using comparative phylogeography to retrace the origins of an ecosystem: the case of four plants endemic to the central Florida scrub. G ITZENDANNER M., and P. S OLTIS 2000. Patterns of genetic variation in rare and widespread plant congeners. American Journal Of Botany 87: 783 792. G ITZENDANNER M. A., C. W. W EEKLEY C. C. G ERMAIN A UBREY D. E. S OLTIS and P. S. S OLTIS 2011. Microsatellite evidence for high clonality and limited genetic diversity in Ziziphus celata (Rhamnaceae), an endangered, self incompatible shrub endemic to the Lake Wales Ridge, Florida, USA. Conservation Genetics in press. G ODT M. 1997. A Population Genetic Analysis of Ziziphus cel ata, an Endangered Florida Shrub. The Journal of Heridity 88: 531 533. G ODT M. J. W., B. R. J OHNSON and J. L. H AMRICK 1996. Genetic diversity and population size in four rare southern Appalachian plant species. Conservation Biology 10: 796 805. G OLDSTEI N D. B., A. R UIZ BARES L. L. C AVALLI S FORZAF and M. W. F ELDMAN 1995. An Evaluation of Genetic Distances for Use With Microsatellite Loci. Genetics 139: 463 471. G OLDSTEIN P. Z., R. D E S ALLE G. A MATO and A. P. V OLGER 2000. Conservation Genetics at th e Species Boundary. Conservation Biology 14: 120 131. G OTTLIEB A., G. G IBERTI and L. P OGGIO 2005. Molecular analyses of the genus Ilex (Aquifoliaceae) in southern South America, evidence from AFLP and its sequence data. American Journal Of Botany 92: 352 369. G RAMLING J. M. 2010. Potential Effects of Laurel Wilt on the Flora of North America. Southeastern Naturalist 9: 827 836. G RIMALDI M. C., and B. C ROUAU R OY 1997. Microsatellite Allelic Homoplasy Due to Variable Flanking Sequences Journal Of Molecular Evolution 44: 336 340. G URGEL C. F. D., S. F REDERICQ and J. N. N ORRIS 2004. Phylogeography of Gracilaria Tikvahiae (Gracilariaceae, Rhodophyta): A Study of Genetic Discontinuity in a Continuously Distributed Species Based on Molecular Evidenc e1. Journal of Phycology 40: 748 758.
185 H AMRICK J., M. G ODT J. A VISE and J. H AMRICK 1996. Conservation genetics of endemic plant species, Conservation Genetics: Case Histories from Nature. H AMRICK J. L., and M. J. W. G ODT 1989. Allozyme diversity in pl ant species. In A. H. D. Brown, M. T. Clegg, A. L. Kahler, AND B. S. Weir [eds.], Plant population genetics, breeding and genetic resources, 43 63. Sinauer, Sunderland, Massachusetts, USA. H AMRICK J. L., and M. J. W. G ODT 1996. Conservation genetics of endemic plant species. In J. C. Avise AND J. L. Hamrick [eds.], Conservation Genetics: Case Histories from Nature. Chapman et Hall. H ANULA J. L., A. E. M AYFIELD S. W. F RAEDRICH and R. J. R ABAGLIA 2008. Biology and Host Associations of Redbay Ambrosia Beetle (Coleoptera: Curculionidae: Scolytinae), Exotic Vector of Laurel Wilt Killing Redbay Trees in the Southeastern United States. Journal of Economic Entomology 101: 1276 1286. H ARDY O. 2003. Estimation of pairwise relatedness between individuals and c haracterization of isolation by distance processes using dominant genetic markers. Molecular Ecology 12: 1577 1588. H ARDY O., and X. V EKEMANS 2002. SPAGEDi: a versatile computer program to analyse spatial genetic structure at the individual or population levels. MOLECULAR ECOLOGY NOTES 2: 618 620. H ARPER G. L., N. M ACLEAN and D. G OULSON 2003. Microsatellite markers to assess the influence of population size, isolation and demographic change on the genetic structure of the UK butterfly Polyommatus bellargus. Molecular Ecology 12: 3349 3357. H ARPER R. 1911. A new plum from the Lake Regio n of Florida. Torreya 11: 64 67. H EDRICK P. 1999. Perspective: Highly variable loci and their interpretation in evolution and conservation. Evolution 53: 313 318. H EY J., Y. J. W ON A. S IVASUNDAR R. N IELSEN and J. A. M ARKERT 2004. Using nuclear haplot ypes with microsatellites to study gene flow between recently separated Cichlid species. Molecular Ecology 13: 909 919. H OLDEREGGER R., U. K AMM and F. G UGERLI 2006. Adaptive vs. neutral genetic diversity: implications for landscape genetics. Landscape E cology 21: 797 807. H UCK R., W. J UDD W. W HITTEN J. S KEAN R. W UNDERLIN and K. D ELANEY 1989. A new Dicerandra (Labiatae) from the Lake Wales Ridge of Florida, with a cladistic analysis and discussion of endemism. Systematic Botany 14: 197 213. H URLBERT S. H. 1971. The Nonconcept of Species Diversity: A Critique and Alternative Parameters. Ecology 52: 577 586.
186 H UTCHINSON D. W., and A. R. T EMPLETON 1999. Correlation of pairwise genetic and geographic distance measures: inferring the relative influences of gene flow and drift on the distribution of genetic variability. Evolution 53: 1898 1914. H UTCHINSON W., D. W ILLS and P. S HIPLEY 2004. MICRO CHECKER: software for identifying and correcting genotyping errors in microsatellite data. MOLECULAR ECOLOGY NOTES 4: 535 538. J ACKSON J. 1973. Distribution and Population Phenetics of the Florida Scrub Lizard, Sceloporus woodi Copeia 1973: 746 761. J AKOBSSON M., and N. R OSENBERG 2007. CLUMPP: a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics 23: 1801. J AMES C. 1957. Notes on the cleistogamous species of Polygala in the southeastern United States. Rhodora 59: 51 54. J ORDAN W. C., M. W. C OURTNEY and J. E. N EIGEL 1996. Low levels of intraspecific genetic variation at a rapidly evolving chloroplast DNA locus i n North American duckweeds (Lemnaceae). American Journal Of Botany 83: 430 439. K ARRON J. D. 1988. A comparison of levels of genetic polymorphism and self compatibility in geographically restricted and widespread plant congeners. Evolutionary Ecology 1: 4 7 58. K ATOH M ISAWA and K UMA 2002. MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic Acids Ressources 30: 3056 3066. K NOWLES L., and W. M ADDISON 2002. Statistical phylogeography. Molecular Ecology 11: 2623 2635. K RAL R. 1960. A revision of Asimina and Deeringothamnus (Annonaceae). Brittonia 12: 233 278. K UMA and T OH 2005. MAFFT version 5: improvement in accuracy of multiple sequence alignment Nucleic Acids Ressources 33: 511 518. L ANDHERR L. L., and W. E. H IGGINS unpublished. A Reevaluation of Nemopanthus (Aquifoliaceae) Based on nrDNA ITS Sequences. L ANE E. 1994. Florida's geological history and geological resources. Published for the Florida Geological Survey, Talahassee, Fl. L EACHE A., and D. M ULCAHY Phylogeny, divergence times and species limits of spiny lizards (Sceloporus magister species group) in western North American deserts and Baja California. Molecular Ecology 0: ??? ???
187 L EE N. S., S. H. Y EAU J. O. P ARK and M. S. R OH 2006. Molecul ar Evidence for Hybridization of Ilex Xwandoensis (Aquifoliaceae) by RAPD Analysis. Journal of Plant Biology 49: 491 497. L EE S., and J. W EN 2001. A phylogenetic analysis of Prunus and the Amygdaloideae (Rosaceae) using ITS sequences of nuclear ribosomal DNA. American Journal Of Botany 88: 150 160. L EIMU R., P. I. A. M UTIKAINEN J. K ORICHEVA and M. F ISCHER 2006. How general are positive relationships between plant population size, fitness and genetic variation? Journal of Ecology 94: 942 952. L EWIS P. and D. C RAWFORD 1995. Pleistocene Refugium Endemics Exhibit Greater Allozymic Diversity Than Widespread Congeners In The Genus Polygonella (Polygonaceae). American Journal Of Botany 82: 141 149. L ITTLE E. L. 1971. Atlas of United States trees. L OPEZ V I NYALLONGA S., M. A RAKAKI N. G ARCIA J ACAS A. S USANNA M. A. G ITZENDANNER D. E. S OLTIS and P. S. S OLTIS 2010. Isolation and characterization of novel microsatellite markers for Arctium minus (Compositae). Am. J. Bot. 97: e4 6. L OWE A. J., D. B OSHIER M. W ARD C. F. E. B ACLES and C. N AVARRO 2005. Genetic resource impacts of habitat loss and degradation; reconciling empirical evidence and predicted theory for neotropical trees. Heredity 95: 255 273. L UIKART G., F. W. A LLENDORF J. M. C ORNUET and W. B S HERWIN 1998. Distortion of Allele Frequency Distributions Provides a Test for Recent Population Bottlenecks. The Journal of Heridity 89: 238 247. M AC D ONALD D., and J. H AMRICK 1996. Genetic variation in some plants of Florida scrub. American Journal O f Botany 83: 21 27. M AC D ONALD D., W. P OTTS J. F ITZPATRICK and G. W OOLFENDEN 1999. Contrasting genetic structures in sister species of North American scrub jays. Proceedings Of The Royal Society Of London Series B Biological Sciences 266: 1117 1125. M AC E G. M. 2004. The role of taxonomy in species conservation. Philosophical transactions of the Royal Society of London. Series B, Biological sciences 359: 711 719. M ADDISON D. R., and W. P. M ADDISON 2000. MacClade 4: Analysis of Phylogeny and Character E volution Version 4.0. Sinauer Associates. M AIN K. N., and E. S. M ENGES 1997. Archbold Biological Station Station Fire Management Plan.
188 M AKOVA K. D., A. N EKRUTENKO and R. J. B AKER 2000. Evolution of Microsatellite Alleles in Four Species of Mice (Genus Apodemus ). Journal Of Molecular Evolution 51: 166 172. M ANEN J., G. B ARRIERA P. L OIZEAU and Y. N ACIRI 2010. The history of extant Ilex species (Aquifoliaceae): evidence of hybridization within a Miocene radiation. M ANEN J. F., M. C. B OULTER and Y. N ACIRI G RAVEN 2002. The complex history of the genus Ilex L. (Aquifoliaceae): evidence from the comparison of plastid and nuclear DNA sequences and from fossil data. Plant Systematics And Evolution 235: 79 98. M ATSUOKA Y., S. E. M ITCHELL S. K RESOVICH M. G OODMAN and J. D OEBLEY 2002. Microsatellites in Zea variability, patterns of mutations, and use for evolutionary studies. Theoretical and Applied Genetics 104: 436 450. M AYFIELD A. E., J. A. S MITH M. H UGHES and T J. D READEN 2008. First Report of Laurel Wilt Disease Caused by aRaffaeleasp. on Avocado in Florida. Plant Disease 92: 976 976. M EIRMANS P. G., and P. H. V AN T IENDEREN 2004. genotype and genodive: two programs for the analysis of genetic diversity of asexual organisms. MOLECULAR ECOLOGY NOTES 4: 792 794. M ENGES E. 2001. Comparative genetics of seven plants endemic to Florida's Lake Wales Ridge. Castanea 66: 98 114. M E NGES E., and N. K OHFELDT 1995. Life history strategies of Florida scrub plants in relation to fire. Journal of the Torrey Botanical Society 122: 282 297. M ENGES E., W. A BRAHAMSON K. G IVENS N. G ALLO and J. L AYNE 1993. 20 years of vegetation change in 5 long unburned Florida plant communities. Journal Of Vegetation Science 4: 375 386. M ENGES E. S. 2007. Integrating demography and fire management: an example from Florida scrub. Australian Journal of Botany 55: 261. M ENGES E. S., and C. V. H AWKES 1998. Interactive effects of fire and microhabitat on plants of Florida scrub. Ecological Concepts in Conservation Biology 8: 935 946. M ENGES E. S., R. W. D OLAN R. P ICKERT R. Y AHR and D. R. G ORDON 2010. Genetic Variation in Past and Current Landscap es: Conservation Implications Based on Six Endemic Florida Scrub Plants. International Journal of Ecology 2010: 1 12. M ILES D., R. N OECKER W. R OOSENBURG and M. W HITE 2002. Genetic relationships among popualtions of Sceloporus undulatus fail to support present subspecific designations Herpetologica 58: 277 292.
189 M ITCHELL O LDS T., J. W ILLIS and D. G OLDSTEIN 2007. Which evolutionary processes influence natural genetic variation for phenotypic traits? Nat Rev Genet 8: 845 856. M OGG R., J. B ATLEY S. H ANL EY D. E DWARDS H. O S ULLIVAN and J. E DWARDS 2002. Characterization of the flanking regions of Zea mays microsatellites reveals a large number of useful sequence polymorphisms. TAG. Theoretical and applied genetics. Theoretische und angewandte Genetik 10 5: 532 543. M ORGAN G. S. 1988. An early Miocene (late Hemingfordian) vertebrate fauna from Brooks Sink, Bradford County, Florida, Southeastern Geological Society Annual Field Trip Guidebook, 53 69. M ORGANTE M., and A. M. O LIVIERI 1993. PCR amplified mic rosatellites as markers in plant genetics. The Plant Journal 3: 175 182. M ORITZ Trends Ecol Evol 9: 373 375. M YERS R. 1985. Fire and the Dynamic Relationship between Florida Sandhill and Sand Pine Scrub Vegetation. Bulletin of the Torrey Botanical Club 112: 241 252. M YERS R., and R. M YERS 1990. Scrub and High Pine, Ecosystems of Florida, 150 194. M YLECRAINE K., J. K USER P. S MOUSE and G. Z IMMERMANN 2004. Geographic allozyme variation in Atlantic white cedar, Chamaecyparis thyoides (Cupressaceae). Canadian Journal of Forest Research 34: 2443 2454. N ASH G. 1896. Revision of the Genus Asimina in North America. Bullet in of the Torrey Botanical Club 23: 234 242. N ELSON E., S. P OLASKY D. L EWIS and A. P LANTINGA 2008. Efficiency of incentives to Proceedings of the National Academy of Sciences N ELSON G. 1996. The shrubs and woody vines of Florida. N ORMAN E. M., and D. C LAYTON 1986. Reproductive Biology of two Florida Pawpaws: Asimina obovata and A. pygmaea (Annonaceae). Bulletin of the Torrey Botanical Club 113: 16 22. O LIVERIA L. O., R. B. H UCK M. A G ITZENDANNER W. S. J UDD D. E. S OLTIS and P. S. S OLTIS 2007. Molecular phylogeny, biogeography and systematics of Dicerandra (Lamiaceae), a genus endemic to the southeastern United States. American Journal Of Botany 94: 1017 1027.
190 O OSTERMEIJER J. G. B., S. H. L UIJTEN and J. C. M. DEN N IJS 2003. Integrating demographic and genetic approaches in plant conservation. Biological Conservation 113: 389 398. O UBORG N., and R. T REUREN 1994. The Significance of Genetic Erosion in the Process of Extinction. IV. Inbreeding Load and Heterosis in Relation to Population Size in the Mint Salvia pratensis. Evolution 48: 996 1008. P ARKER K., J. H AMRICK A. P ARKER and E. S TACY 1997. Allozyme diversity in Pinus virginiana (Pinaceae): intraspecific and interspecific comparisons. Am. J. Bot. 84: 1372 P ARKER K., J. H AMRICK A. P ARKER and J. N ASON 2001. Fine scale genetic structure in Pinus clausa (Pinaceae) populations: effects of disturbanc e history. Heredity 87: 99 113. P EAKALL R., and P. S MOUSE 2006. GENALEX 6: genetic analysis in Excel. Population genetic software for teaching and research. MOLECULAR ECOLOGY NOTES 6: 288 295. P ETERSON A. 1992. Phylogeny and rates of molecular evolution in the Aphelocoma jays (Corvidae). Auk 109: 133 147. P ETERSON A., E. M ARTINEZ M EYER and C. G ONZALEZ S ALAZAR 2004. Reconstructing the Pleistocene geography of the Aphelocoma jays (Corvidae). Diversity And Distributions 10: 237 246. P ETIT R., U. C SAIKL S. B ORDCS K. B URG E. C OART J. C OTTRELL B. VAN D AM et al. 2002. Chloroplast DNA variation in European white oaks:: Phylogeography and patterns of diversity based on data from over 2600 populations. Forest Ecology and Management 156: 5 26. P ETIT R. J., A. E L M OUSADIK and O. P ONS 1998. Identifying Populations for Conservation on the Basis of Genetic Markers. Conservation Biology 12: 844 855. P IRKLE E., and W. Y OHO 1970. The heavy mineral ore body of Trail Ridge, Florida. Economic Geology 65: 17 30 P RATT A. E. 1990. Taphonomy of the large vertebrate fauna from the Thomas Farm locality (Miocene, Hemingfordian), Gilchrist County, Florida. Bulletin of the Florida Museum of Natural History, Biological Sciences 35: 35 130. P RITCHARD J., M. S TEPHENS and P. D ONNELLY 2000. Inference of Population Structure Using Multilocus Genotype Data. Genetics 155: 945 959.
191 P ROBER S. M., and A. H. D. B ROWN 1994. Conservation of the Grazzy White Box Woodlands: Population Genetics and Fragmentation of Eucalyptus al bens Conservation Biology 8: 1003 1013. R AMAKRISHNAN U., and J. M OUNTAIN 2004. Precision and Accuracy of Divergence Time Estimates from STR and SNPSTR Variation. Mol Biol Evol 21: 1960 1971. R AMBAUT A. 2009. FigTree, version 1.3.1. website: http://beast.bio.ed.ac.uk/ R AYMOND M., and F. R OUSSET 1995. GENEPOP (Version 1.2): Population Genetics Software for Exact Tests and Ecumenicism. J Hered 86: 248 249. R EED D. H., and R. F RANKHAM 2003. Correlation between Fitness and Genetic Diversity. Conservation Biology 17: 230 237. R ICE W. R. 1989. Analyzing Tables of Statistical Tests. Evolution 43: 223 225. R ICKLEFS R. E. 1987. Community Diversity: Relative Roles of Local and Regional Pro cesses. Science 235: 167 171. R OHWER J. 2000. Toward a Phylogenetic Classification of the Lauraceae: Evidence from matK Sequences. Systematic Botany 25: 60 71. R OSENBERG N. 2004. DISTRUCT: a program for the graphical display of population structure. MOLE CULAR ECOLOGY NOTES 4: 137 138. R OSSETTO M., J. M C N ALLY and R. J. H ENRY 2002. Evaluating the potential of SSR flanking regions for examining taxonomic relationships in the Vitaceae. Theoretical and Applied Genetics 104: 61 66. R OUSSET F. 1996. Equilibr ium Values of Measures of Population Subdivision for Stepwise Mutation Processes. Genetics 142: 1357 1362. ______. 2008. GENEPOP'007: a complete re implementation of the GENEPOP software for Windows and Linux. Molecular Ecology Resources 8: 103 106. R YDER O. A. 1986. Species Conservation and Systematics: the Dilemma of Subspecies. Trends Ecol Evol 1: 9 10. S ARGENT 1889. Ilex cassine L. var. myrtifolia Sarg. Garden and Forest 2: 616. S AUNDERS N., L. K ESSLER and J. A VISE 1986. genetic variation and geogr aphic differentiation in mitochondrial DNA of the horseshoe crab, Limulus polyphemus Genetics 112: 613 627. S CHREY A. W., K. G. A SHTON S. H EATH E. D. M C C OY and H. R. M USHINSKY 2011. Fire alters patterns of genetic diversity among 3 lizard species in Florida Scrub habitat. The Journal of heredity 102: 399 408.
192 S ELKOE K. A., and R. J. T OONEN 2006. Microsatellites for ecologists: a practical guide to using and evaluating micros atellite markers. Ecology Letters 9: 615 629. S ETOGUCHI H., and I. W ATANABE 2000. Intersectional gene flow between insulat endemics of Ilex (Aquifoliaceae) on the Bonin Islands and the Ryukyu Islands American Journal Of Botany 87: 793 810. S HAW J., and R. S MALL 2004. Addressing the "hardest puzzle in American pomology": Phylogeny of Prunus sect. Prunocerasus (Rosaceae) based on seven noncoding chloroplast DNA regions. American Journal Of Botany 91: 985 996. ______. 2005. Chloroplast DNA phylogeny and p hylogeography of the North American plums ( Prunus subgenus Prunus section Prunocerasus Rosaceae). American Journal Of Botany 92: 2011 2030. S HAW J., E. L ICKEY E. S CHILLING and R. S MALL 2007. Comparison of whole chloroplast genome sequences to choose n oncoding regions for phylogenetic studies in angiosperms: the tortoise and the hare III. American Journal Of Botany 94: 275. S HAW J., E. B. L ICKEY J. T. B ECK S. B. F ARMER W. L IU J. M ILLER K. C. S IRIPUN et al. 2005. The tortoise and the hare II: rela tive utility of 21 noncoding chloroplast DNA sequences for phylogenetic analysis. American Journal Of Botany 92: 142 166. S HIMODAIRA H. 2002. An Approximately Unbiased Test of Phylogenetic Tree Selection. Systematic Biology 51: 492 508. S HIMODAIRA H., and M. H ASEGAWA 1999. Multiple Comparisons of Log Likelihoods with Applications to Phylogenetic Inference. Molecular Biology and Evolution 16: 1114 1116. S LAPCINSKY J. L., and D. R. G ORDON 2003. Monitoring report. Polygala lewtonii The Nature Conservan cy, Gainesville, Fl. S LAPCINSKY J. L., D. G ORDON and S. PROGAM 2003. Monitoring report. Polygala lewtonii S LAPCINSKY J. L., B. P ACE A LDANA and D. R. G ORDON 2005. Monitoring report, Polygala lewtonii The Nature Conservancy, Gainesville, Fl. S LAPCINS KY J. L., B. P ACE A LDANA D. G ORDON and T. C ONSERVANCY 2005. Monitoring report, Polygala lewtonii S MALL J. 1898. Studies in the Botany of the Southern United States. XIII. Bulletin of the Torrey Botanical Club 25: 134 151. S MITH HM, E. 1992. Adaptativ e convergence in the lizard superspecies Sceloporus undulatus Bulletin of the Maryland Herpetological Siciety 28: 123 149.
193 S OLTIS D., A. M ORRIS J. M C L ACHLAN P. M ANOS and P. S OLTIS 2006. Comparative phylogeography of unglaciated eastern North America. Molecular Ecology 15: 4261 4293. S OLTIS P. S., and M. A. G ITZENDANNER 1999. Molecular Systematics and the Conservation of Rare Species. Conservation Biology 13: 471 483. S TAMATAKIS A. 2006. RAxML VI HPC: maximum likelihood based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics 22: 2688 2690. S UN Y., D. Z. S KINNER G. H. L IANG and S. H. H ULBERT 1994. Phylogenetic analysis of Sorghum and related taxa usi ng internal transcribed spacers of nuclear ribosomal DNA. Theoretical and Applied Genetics 89: 26 32. S UTHERLAND J. P. 1974. Multiple Stable Points in Natural Communities. The American Naturalist 108: 859 873. S WOFFORD D., and S INAUER 2003. PAUP*: Phylo genetic Analysis Using Parsimony (and other methods). Version 4.0b 10. T ABERLET P., L. G IELLY G. P AUTOU and J. B OUVET 1991. Universal primers for amplification of three non coding regions of chloroplast DNA. Plant Molecular Biology 17: 1105 1109. T ABER LET P., L. F UMAHALLI A. G. W UST S AUCY and J. F. C OSSON 1998. Comparative phylogeography and postglacial colonization routes in Europe. Molecular Ecology 7: 453 464. T RAPNELL D., J. S CHMIDT P. Q UINTANA A SCENCIO and J. H AMRICK 2007. Genetic Insights into the Biogeography of the Southeastern North American Endemic, Ceratiola ericoides (Empetraceae). J Hered : esm075. USFWS. 1999a. Scrub plum. South Florida multi species recovery plan. ______. 1999b. Crotalaria avonensis Multi Species Recovery Plan for South Florida 2005. ______. 1999c. South Florida multi species recovery plan. ______. 2010. Polygala lewtonii 5 year review: summary and evaluation. V ALDES A. M., M. S LATKIN and N. B. F REIMER 1993. Allele Frequencies at Microsatellite Loci: The Stepwi se Mutation Model Revisited. Genetics 133: 737 749. V ALI U., A. E INARSSON L. W AITS and H. E LLEGREN 2008. To what extent do microsatellite markers reflect genome wide genetic diversity in natural populations? Molecular Ecology 17: 3808 3817.
194 VAN O OSTERH OUT C., W. H UTCHINSON D. WILLS, and P. S HIPLEY 2004. MICRO CHECKER: software for identifying and correcting genotyping errors in microsatellite data. MOLECULAR ECOLOGY NOTES 4: 535 538. V ANE W RIGHT R. I., C. J. H UMPHRIES and P. H. W ILLIAMS 1991. What to Protect? Systematics and the Agony of Choice. Biological Conservation 55: 235 254. V ELLEND M., and M. G EBER 2005. Connections between species diversity and genetic diversity. Ecology Letters 8: 767 781. W ADE M. 2007. The co evolutionary genetics of ecological communities. Nature Reviews Genetics 8: 185 195. W ATERS J. M., and G. P. W ALLIS 2000. Across the Southern Alps by river capture ? Freshwater fish phylogeography in South Island, New Zealand. Molecular Ecology 9: 1577 1582. W ATTS W. 1975. A late Quaternary record of vegetation from Lake Annie, south central Florida. Geology 3: 344 346. W EBB S., and R. M YERS 1990. Historical Biogeography, Ecosystems of Florida, 70 103. W EBB S. D. 1990. Historical Biogeography. In R. Myers, L, AND J. J. Ewel [ed.], Ecosystems of Florida, 70 103. University of Central Florida Press, Orlando. W EEKLEY C., and E. M ENGES 2003. Species and vegetation responses to prescribed fire in a long unburned, endemic rich Lake Wales Ridge scrub. Journal of the Torrey Botanical Society 130: 265 282. W EEKLEY C., and A. B ROTHERS 2006. Failure of reproductive assurance in the chasmogamous flowers of Polygala lewtonii (Polygalaceae), an endangered sandhill herb. American Journal Of Botany 93: 245 253 W EEKLEY C., and E. M ENGES submitted. Burning creates contrasting demographic patterns in Polygala lewtonii (Polygalaceae): a cradle to grave analysis of multiple cohorts in a perennial herb. W EEKLEY C., T. K UBISIAK and T. R ACE 2002. Genetic impover ishment and cross incompatibility in remnant genotypes of Ziziphus celata (Rhamnaceae), a rare shrub endemic to the Lake Wales Ridge, Florida. Biodiversity And Conservation 11: 2027 2046. W EEKLEY C. W., E. S. M ENGES and R. L. P ICKERT 2008. An ecological map of Florida's Lake Wales Ridge: a new boundary delineation and an assessment of post columbian habitat loss. Florida Scientist 71: 45 64.
195 W EEKLEY C. W., D. N. Z AYA E. S. M ENGES and A. E. F AIVRE 2010. Multiple causes of seedling rarity in scrub plum Prunus geniculata (Rosaceae), an endangered shrub of the Florida scrub. American Journal Of Botany 97: 144 155. W EIR B., and C. C OCKERHAM 1984. Estimating F Statistics For The Analysis Of Population Structure. Evolution 38: 1358 1370. W HITE T. J., T. B RUNS S. L EE and J. W. T AYLOR 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In M. A. Innis, D. H. Gelfand, J. J. Sninsky, AND T. J. White [eds.], PCR Protocols: A Guide to Methods and Applications, 315 322. Academic Press, Inc., New York. W HITHAM T., W. Y OUNG G. M ARTINSEN C. G EHRING J. S CHWEITZER S. S HUSTER G. W IMP et al. 2003. community and ecosystem genetics: a consequence of the extended phenotype. Ecology 84: 559 573. W HITTAKER R. J., M. B. A RAJO J. P AUL R. J. L ADLE J. E. M. W ATSON and K. J. W ILLIS 2005. Conservation Biogeography: assessment and prospect. Diversity And Distributions 11: 3 23. W IENS J. J., and M. J. D ONOGHUE 2004. Historical biogeography, ecology and species richness. Trends In Ecology & Evolution 19: 639 644. W ILLIS C. G., B. R UHFEL R. B. P RIMACK A. J. M ILLER R USHING and C. C. D AVIS 2008. Phylogenetic patterns of species loss in Thoreau's woods are driven by climate change. Proceedings of the National Academy of Science s 105: 17029 17033. W ON Y., A. S IVASUNDAR Y. W ANG and J. H EY 2005. On the origin of Lake Malawi cichlid species: A population genetic analysis of divergence. Proceedings Of The National Academy Of Sciences Of The United States Of America 102: 6581 6586 W RIGHT S. 1951. The Genetical Structure of Populations. Annals of Eugenics 15: 323 354. W UNDERLIN R. 1998. Guide to the vascular plants of Florida. University Press of Florida, Gainesville, FL. X U D. H., J. A BE M. S AKAI A. K ANAZAWA and Y. S HIMAMOTO 2000. Sequence variation of non coding regions of chloroplast DNA of soybean and related wild species and its implications for the evolution of different chloroplast haplotypes. Theoretical and Applied Genetics 101: 724 732. Y OUNG A. G., T. B OYLE and T B ROWN 1994. The population genetic consequences of habitat fragmentation for plants. Trends In Ecology & Evolution 11: 413 418.
196 Z ARDOYA R., D. M. V OLLMER C. C RADDOCK J. T. S TREELMAN S. K ARL and A. M EYER 1996. Evolutionary Conservation of Microsate llite Flanking Regions and their Use in Resolving the Phylogeny of Cichlid Fishes (Pisces: Perciformes). Proceeding of the Royal Society, London: Biological Sciences 263: 1589 1598.
197 BIOGRAPHICAL SKETCH Charlotte Germain Aubrey was born and grew up in France. After obtaining her Baccalaureat in Sciences, she spent a year in Beijing studying Chinese. She then integrated the University of Edinburgh in Scotland where s he obtained her b degree in e cological sciences in 2002. After working for several months in Madagascar, she joined the m program in Biodiversity and Taxonomy of Plants at the Royal Botanic Gardens of Edinburgh and graduated in August 2004 Her m on the sustainable logging of mahogany in Belize using mic rosatellites. Charlotte Germain Aubrey integrated the Department of Biology at the University of Florida. Under the supervision of Drs. Matt hew Gitzendanner and Pam ela Soltis, she is based in the Laboratory of Molecular Systematics and Evolutionary Genetic s at the Florida Museum of Natural History. Her interests are in comparative phylogeography of areas of conservation interest and population genetics of threatened plant species After completing her PhD, she is keen on applying her scientific training to the tropical forests of SE Asia and Africa, which she considered as understudied and in great need for conservation in light of illegal logging activities in those regions.