Population genetic analyses in the orchid genus Gymnadenia: a conservation genetic perspective

Full text

(1)

(2)

(3)

(4)

(5)

(6)

(7) .

(8)

(9) .

(10)

(11)

(12)

(13) . !. "##$ %"&#. "" #'($)'"' "*'$#' "*" +,,-.

(14) Dissertation for the Degree of Doctor of Philosophy in Conservation Biology presented at Uppsala University in 2003 ABSTRACT Gustafsson, S. 2003. Population genetic analyses in the orchid genus Gymnadenia – a conservation genetic perspective. Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 799. 43 pp. Uppsala ISBN 91-554-5517-4. Small populations are facing a particular risk of extinction due to a lack of appropriate genetic diversity and associated negative effects, factors dealt with in the discipline of conservation genetics. Many orchid species exhibit characteristics that make them a perfect study object in the scope of conservation genetics. The aim with this thesis was to investigate genetic structure at different levels in two orchid species Gymnadenia conopsea, geographically widespread, although diminishing and G. odoratissima with a long history of being rare. Microsatellite markers, developed in and used in studies of G. conopsea were also used in the study of G. odoratissima. Populations of G. conopsea expressed high levels of genetic variation and a certain amount of gene flow, although investigated mating pattern in a small population indicated non-random mating among individuals, with the majority of pollen exchange between near neighbours, and noticeable levels of geitonogamous pollinations. Further a pronounced year to year variation in flowering frequency among individuals was found. It was also discovered that flowering time variants (early and late) within the species G. conopsea were highly differentiated and seem to have had a more ancient historical separation than the separation between the two different species, G. conopsea and G. odoratissima. Levels of genetic variation in the rare congener, G. odoratissima differed between island and mainland populations where the more numerous island populations expressed larger levels of genetic variation and were less differentiated compared to the few remaining and genetically depauperated mainland populations. Key words: Orchidaceae, Gymnadenia, conservation, microsatellites, genetic structure, mating pattern. Susanne Gustafsson, Department of Conservation Biology and Genetics, Evolutionary Biology Centre, Uppsala University, Norbyvägen 18D, SE-752 36 Uppsala, Sweden. © Susanne Gustafsson 2003 ISSN 1104-232X ISBN 91-554-5517-4 Printed in Sweden by Uppsala University, Print & Media, Uppsala 2003.

(15) PAPERS INCLUDED IN THE THESIS This thesis is based on the following papers, which will be referred to in the text by their Roman numerals:. I. Gustafsson, S. and Thorén, P. (2001). Microsatellite loci in the Gymnadenia conopsea, the fragrant orchid. Molecular Ecology Notes, 1:81-82.. II. Gustafsson, S. Mating pattern in a small population of the Lepidopteran pollinated orchid, Gymnadenia conopsea, as revealed by microsatellites. Manuscript.. III. Gustafsson, S. (2000). Patterns of genetic variation in the fragrant orchid (Gymnadenia conopsea). Molecular Ecology 9: 1863-1872.. IV. Gustafsson, S. and Sjögren-Gulve, P. (2002) Genetic diversity in the rare orchid, Gymnadenia odoratissima and a comparison with the more common congener, G. conopsea. Conservation Genetics 3: 225-234.. V. Gustafsson, S. Flowering frequency and genetic diversity in a small population of Gymnadenia conopsea - a five year study. Manuscript.. VI. Gustafsson, S. and Lönn, M. Genetic differentiation and habitat preference of flowering-time variants within Gymnadenia conopsea. Manuscript.. Papers I and III are reproduced with kind permission of Blackwell Publishing. Paper IV are reproduced with kind permission of Kluwer Academic Publishers..

(16) Contents. Introduction.....................................................................................................2 Conservation genetics ................................................................................2 Population genetics ....................................................................................3 Mating systems and gene dispersal ............................................................5 Orchids .......................................................................................................6 Material and methods......................................................................................9 Genetic markers, microsatellites ................................................................9 Development of microsatellite loci ..........................................................10 Genetic and statistical analyses ................................................................11 Population structure.............................................................................12 ҏAssignment methods ...........................................................................13 Genetic differentiation within species .................................................13 Study species............................................................................................14 The genus Gymnadenia .......................................................................14 Results and discussion ..................................................................................17 Microsatellite loci in Gymnadenia conopsea (I) ......................................17 Mating pattern (II) and flowering frequency (V).....................................20 Genetic structure in rare and diminishing orchid species (III), (IV) .......22 Restricted gene flow within species (VI).................................................26 Conclusions...................................................................................................32 Acknowledgements.......................................................................................34 References.....................................................................................................36. 1.

(17) Introduction. Conservation genetics During the 20Th century, mankind has had a widespread and devastating impact on the survival of many species. Factors directly or indirectly associated to different human activities, like pollution, habitat destruction, over-exploitation and introduced species have, besides a rapid and direct extinction of numerous species, also through fragmentation events and reduced population size, deteriorated the conditions for long-time survival in an even larger number of species. The area of conservation biology, concerning preservation of biodiversity, has therefore been a huge and rapidly growing discipline of research in the last decades. In preserving biodiversity, the strategy of conservation biology has customarily been to treat the species as the unit of conservation. However, even though the emphasis has been and still will be on rare species, more recently the view has widened to also involve other levels of organization, ranging from genes to ecosystems. Together with demographic and environmental factors, one part of conservation biology therefore includes the influences of genetic factors. Even though the significance of the first two is undisputed, the importance of genetic factors in conservation biology has been questioned (Frankham et al, 2002 and references therein). Responsible for raising the subject of conservation genetics was Sir Otto Frankel, who, in the beginning of the 1970s wrote important papers covering the issue (Frankel 1970, 1974). The area of conservation genetics deals with genetic factors affecting a species risk of extinction and with management regimes that could be acquired to minimize these risks (Frankham et al, 2002). Small populations are, due to their reduced population size, supposed to also express reduced genetic diversity (Frankham, 1995). Different studies, both in plants and animals, have found support for the fact that small populations face a particular risk of extinction due to a lack of appropriate genetic diversity and associated negative effects (Young & Clarke, 2000, Frankham et al, 2002 and references therein). A primary element in conservation management has therefore been preservation of genetic diversity and minimizing inbreeding. However, the significance of genetic factors contributing in species extinction has not been without debate and some will instead emphasize demographic and environmental factors whereas genetic factors have been 2.

(18) suggested to be more a symptom than a cause to species extinction (Lande, 1988; Caughley, 1994; Holsinger & Vitt, 1997; see also Lande, 2000). As stressed by Hedrick (2000a) genetic processes do not operate in isolation but influence extinction probabilities indirectly by their effect on reproductive success, behaviour, viability, disease resistance etc. Frankham et al (2002) also point at the difficulties in isolating the effects of genetic processes and imply that genetic threats form one component of the endangering process, and population extinction is due to interactions between genetic, demographic and environmental factors. Quite a number of plant species have been studied in the area of conservation genetics and examples of some genera are:– Caesalpinia (Cardoso et al, 1998), Brighamia (Gemmill et al, 1998), Saxifraga (Hollingsworth et al, 1998), Vicia (Black-Samuelsson & Lascoux, 1999), Goodyera (Wong & Sun 1999), Rutidosis (Brown & Young, 2000), Gymnadenia (Gustafsson & Sjögren-Gulve, 2002), Leucopogon (Zawko et al 2001), Triunia (Shapcott, 2002) Epipactis (Squirrell et al, 2002). The studies describe demographic and genetic status of the investigated species, comparisons with other rare plant species and often a conservation strategy. Even though generalisations about and trends in genetic structure and other life history traits in rare and diminishing species are vital as numerous plant species are presently threatened (review Kunin & Gaston 1993, Hedge & Ellstrand 1999), comparisons to an average comprising “all” rare species have suggested to be of little value (Gitzendanner & Soltis 2000). The reason for this was found when comparing available genetic data of rare and widespread congeners. Even though rare species on average had less genetic variability than their widespread congeners, there were many exceptions and a large range of values (Karron, 1987; Gitzendanner & Soltis, 2000). Plant species express low genetic diversity for various reasons and not necessarily as an effect of small population size (Karron, 1987; Gitzendanner & Soltis, 2000). Instead, levels of diversity were found to be highly correlated within a genus and the relationship between the geographic range and genetic diversity was suggested to have high significance in congeneric comparisons (Felsenstein, 1985, Silvertown and Dodd, 1996, Gitzendanner & Soltis, 2000). Below, genetic processes and theories behind the proposal of small populations facing an increased extinction risk are presented.. Population genetics All natural populations are exposed to a number of genetic factors or forces affecting the amount and kind of genetic variation. Such factors are mutation, chance events like genetic drift and founder events, selection, migration and a species’ mating system (Hedrick, 2000b). The first factor, 3.

(19) mutation, always increases genetic variation, although most of the times with neutral and sometimes with negative effects. Genetic drift, on the other hand, in the end always reduces a population’s genetic variation due to random loss of alleles. The other factors may in different situations either increase or reduce genetic variation. The relative importance and strength of each force differ among different species and populations (Barret & Kohn, 1991). In general, even though all factors are responsible for evolution in large as well as small populations, the conditions for mutation, migration and selection are more favourable in large populations at the same time as chance effects, like genetic drift, are small. In contrast, chance effects predominate in small populations where the impact of selection on allele frequencies is reduced, resulting in a loss of alleles and thereby reducing genetic variation (Hedrick, 2001). Endangered species are, as mentioned above, in general supposed to have lower levels of genetic variation than non-endangered species (Frankham et al, 2002, but see Gitzendanner & Soltis, 2000). The main reason is found in their small populations, where loss of genetic variability, primarily due to chance effects like genetic drift, will be more rapid than in large populations. Species with small populations are also exposed to another genetic consequence, inbreeding. The opportunities for mating between close relatives are higher in small than in large populations. This could result in an increased homozygosity, reduced fitness and inbreeding depression. However, some populations with a history of being small may have adapted to the negative consequences of inbreeding (Barret & Charlesworth, 1991; Ouborg & van Treuren, 1994) or developed more or less advanced systems against close inbreeding. Inbreeding might therefore, at least in theory, be most harmful to populations or species that recently have become small (Huenneke, 1991). The genetic impact of the threats described above, especially in small and patchily distributed populations, is critically dependent on the amount of existing gene flow (Ellstrand & Elam, 1993; Frankham et al, 2002). With restricted or no gene flow, the negative effects of inbreeding and genetic drift may increase and consequently increase the risk of extinction (but see Caughley, 1994; Holsinger & Vitt, 1997; Lande, 2000). However, as the level of gene flow depends on the abilities of dispersal and on the geographic distribution of each species, the outcome will vary. In theory, one migrant per generation would, in long term, be sufficient to counter-balance the effects of genetic drift (Wright, 1931; Allendof, 1983; Hartl & Clark, 1988). On the other hand, gene flow could also be seen as a constraining factor in the area of evolution (Mayr, 1970; Slatkin, 1987). In natural populations, selection acts to adapt populations to conditions in the local environment. Therefore, gene flow which in small populations counteracts the negative effect of inbreeding and genetic drift will, in the form of immigrants from other populations with genes adapted to other conditions, simultaneously 4.

(20) counteract the ongoing selection process (Slatkin, 1987; Lenormand, 2002). With interrupted or restricted gene flow on the other hand, populations could, both by chance and as a result of different environmental conditions, evolve independently and as a consequence eventually become incompatible with each other, leading to the evolution of new species (e.g Mayr, 1970; Turelli et al., 2001).. Mating systems and gene dispersal In the development of efficient conservation strategies, processes affecting patterns of gene flow, dispersal abilities, and mating patterns are important to study (Ellstrand, 1992). Plants display a variety of breeding systems and different mechanisms of seed and pollen dispersal which have a large impact on intraspecific genetic structure (Barrett & Kohn, 1991; Hamrick et al., 1991). Genetic variability within plant populations as well as genetic differentiation among populations will be determined by their mating system and probabilities of dispersal (Brown, 1989; Hamrick & Godth, 1989). The fact that flowering plants, at least as “adults”, are sessile implies that dispersal events are performed by some kind of vector (wind, water or animals) and the way these vectors transport the genetic material will have great impact on both the direction and distance of gene flow. In insect mediated pollen dispersal, the efficiency of gene flow will be dependent on the behaviour of the pollinator (Proctor et al., 1996). While some insects are highly mobile or even migratory, providing opportunities of long-distance dispersal, others are quite stationary, resulting in a more restricted gene flow. The pollinators can either transfer pollen within populations or between populations, between individuals or within an individual. The relationship between flowering plants and their pollinators will be highly involved in the evolution of floral features like flower colour, fragrance, nectar and inflorescence size (Waser, 1983). Individuals attracting more pollinators could thereby have the benefit of contributing more genes to the next generation. This gives small populations a great disadvantage as they have found to be less attractive to pollinators than large populations (Sih & Baltus, 1987; Sowig, 1989). Subsequently the result might be a decrease in reproductive performance either through reduced seed set due to a smaller number of pollinations or increased inbreeding resulting from mating among close relatives. It has been shown that the amount of selfpollination increases as the number of possible mates decreases (Murawski et al, 1990). However, the advantage of attracting more pollinators could be counteracted by increased geitonogamous pollinations, self-pollination between flowers within an inflorescence. The frequency of geitonogamy 5.

(21) seem to depend on a number of factors like the number of available flowers on the plant, where larger inflorescences may promote pollinator visitation (Oashi & Yakara, 1999; Johnson & Nilsson, 1999, but see Maad & Reinhammar, 2002); presence or absence of reward, where nectariferous species are found to have larger amounts of geitonogamy (Dafni, 1987; Nilsson et al., 1992a); and pollinator behaviour (Johnson & Nilsson, 1999). As self-fertilization increases the amount of homozygosity and the risk of higher frequencies of detrimental alleles, the appearance of inbreeding depression, several ways of avoiding this have evolved among the hermaphrodites. Some examples are given by species with spatial separation of anthers and stigma, species where male gametes are released before (or after) the female part is susceptible to pollen receipt; and the remarkable mechanism of self-incompatibility, in which a plant rejects its own pollen. However, the majority of plants, covering more than 70% of the angiosperms, are hermaphrodites and in this group many plant species are self-compatible and thus exposed to the risk of self-fertilization (Silvertown & Charlesworth, 2001). Gene dispersal in plants also includes the component of seed dispersal. The distance with which a seed is transported depends both on the mother plant’s ability and on the characteristics of the vectors (Silvertown & Charlesworth, 2001). Some examples of the widely different mechanisms with which seeds are dispersed includes expulsion from exploding fruits, winged or tiny seeds transported by the wind, hooked seeds trapped by animals and seeds carried in the guts of animals (Silvertown & Charlesworth, 2001). The shape of the dispersal-distance distribution depends on the dispersal mechanism and is often difficult to quantify, especially for long-distance dispersal (Silvertown & Charlesworth, 2001). Commonly though, the highest concentrations of seeds are found around the mother plant irrespective of the dispersal mechanism (Clark et al, 1999). The distributions of dispersal distances of wind-dispersed seeds are predicted to have long tails, i.e. slowly decreasing seed densities with increasing geographic distance, whereas the curve shape of animal-dispersed seeds, on the other hand, is steeper and has a more pronounced decrease of seed densities with the geographic distance (Clark et al, 1999).. Orchids Many orchid species exhibit characteristics that make them a perfect study object in the scope of conservation genetics. There are numerous endangered species and species diminishing in number. Many species have had a long history with small population size and a scattered distribution whereas others have quite recently become threatened due to anthropogenic habitat loss and fragmentation processes. Furthermore, members of the orchid family have 6.

(22) “always” fascinated people with their extraordinary beauty, their tremendous variation in size, colour and shape and their, sometimes extremely advanced, pollination systems. Another characteristic is their association with fungi, mycorrhiza, which serves the endosperm-less seed with nutrients. The family Orchidaceae is one of the most species rich families in the world, with the estimated number of species ranging from 19 500 (Dressler, 1993) up to perhaps 25 000 (Nilsson, 1992a) of which only about 200 are growing in Europe. The vast majority are found in tropical areas, where they often are epiphytic, whereas all European orchid species are grounddwelling. Most of the species are perennial and many are long-lived, ages of 15-30 years has been recorded in some northern species (Tamm, 1972; see also Wells & Williams, 1991). Orchids occupy different types of habitats. Epiphytes are primarily found in trees, and tropical species often occur high up in the tree tops (Mossberg & Nilsson, 1985). The majority of the grounddwelling species grow in open areas like grazed meadows, marshes and shores, even though some species also are found deep in the forest (Mossberg & Nilsson, 1985). Tropical orchid species exhibit a tremendous variation in size, shape and colour, whereas ground-living species have a more or less uniform construction with flowers on a leafy stem, either singleor multi-flowered organized as a spike or a raceme. (Mossberg & Nilsson, 1985). Orchid flowers have six perianth segments, one of which is larger than the others: the lip (Proctor et al, 1996; van der Cingel, 1995). The lip appears to be positioned at the bottom of the flower. However the ovary is often twisted 180o so in fact the lip will be the uppermost petal (Proctor et al, 1996). It functions as a platform for visiting animal pollinators which are attracted by flower colour, shape or scent. The pollinators are in most orchid species rewarded by nectar from the spur or nectaries, although about 1/3 of the species deceit the visiting pollinator and offer no reward (Gill, 1989; Nilsson, 1992a). Often, the only remaining fertile stamen (in some species two or rarely three stamens) and the stigma are fused into a column, positioned in the centre of the flowers, above the lip (Proctor et al, 1996). This structure prevents visiting insects from transferring the flower’s own pollen to the stigma, promoting cross-fertilization (van der Cingel, 1995) (but see geitonogamy). Unlike many other plant families, the pollen grains are not loose and powdery in the orchids but bound together more or less tightly by elastic threads, forming two pollen packages or pollinias (Proctor et al, 1996; Johnson & Edwards, 2000). The mechanisms of pollen transfer from the plant to the pollinator differ between the different subfamilies. For example in the subfamiliy Orchidoideae each pollinium is attached to a viscid disc, which sticks somewhere at the pollinator, often tongue/proboscis, eye, head or leg (Proctor et al, 1996). Some orchid species are highly specialized to a few or even a single pollinator species, whereas others are visited by many different pollinators. 7.

(23) After pollination and fertilization, the seeds are matured in a fruit, a capsula. Orchid seeds are tiny, light and dispersed by the wind, having the ability of long-distance dispersal. It has been suggested that long-lived and outbreeding plants with possibilities to long-distance dispersal often express higher genetic variation within than among populations (Hamrick et al., 1991). On the other hand, a potential for long-distance seed dispersal may also entail founder events and geographic isolation among populations (Dodson & Gillespie, 1967; Gill, 1989; Case, 1994) resulting in a reduced genetic diversity within populations.. 8.

(24) Material and methods. Genetic markers, microsatellites The majority of the results in this thesis are based on information from genetic markers known as microsatellites. Microsatellites or simple sequence repeats (SSR) have, since their first description been widely used as genetic markers in different kinds of studies (Litt & Luty, 1989; Tautz, 1989; Weber & May, 1989; see Zane et al, 2002). Their wide applicability span over areas like genetic mapping for which they were originally utilised, studies in conservation genetics, forensic DNA studies, paternity analyses, identifying individuals, studies in population structure, reconstruction of human origin, studies in hybridisation etc (Goldstein & Schlötterer, 1999 and references therein). The major advantage in using microsatellites as genetic markers is their usually high variability especially in long and uninterrupted sequences, which is due to a high mutation rate of about 10-3 events per locus per generation (Hancock, 1999). Microsatellites are composed of 1-6 base pairs which are arranged in the genome as tandem repeated motifs. The variability is mostly due to changes in the number of copies of the microsatellite repeat. In general, mutation in a microsatellite allele generates changes in size of one repeat, but sometimes several repeat units could be changed (Weber & Wong, 1993; Di Rienzo et al, 1994; Primmer et al, 1996). Microsatellites have been detected in every organism so far analysed, in eukaryotic as well as prokaryotic genomes (Zane et al, 2002). Their highest abundance is found in non-coding regions, simplifying a high mutation rate (Hancock, 1999; Zane et al, 2002). Mutation mechanisms proposed to be involved which could explain such high mutation rates are slipped-strand mispairing and unequal crossing-over, of which the former seems to have gained most support (Eisen, 1999; Hancock, 1999; Zane et al, 2002). In slipped strand-mispairing the two DNA strands misanneal during replication where the nascent strand either will be shorter or longer compared to the template. In unequal crossing-over, the two chromosome strands are misaligned during crossing-over, which will give a deletion in one DNA molecule and an insertion in the other. The possible functions of microsatellites have been debated, but are still unclear. There are some evidence of microsatellites having functional roles as coding or regulatory elements (Kunzler et al, 1995; Kashi et al, 1997) and highly expanded trinucleotide repeats have also found to be involved in 9.

(25) some human diseases (Rubinsztein, 1999). However, most microsatellites are regarded as neutral markers, i.e. mutation does not influence the fitness of the organism (Goldstein & Schlötterer, 1999). To fully utilize information revealed by microsatellite loci it is essential to get a thorough understanding of the mutational processes shaping this variation. The estimation of population parameters such as genetic differentiation, number of migrants per generation etc., are dependent on which mutation model could be applied to the genetic markers. As the sensitivity to the mutation model increases with the markers mutation rate, this is especially important in microsatellites (Estoup & Cornuet, 1999). Unfortunately, in spite of the increasing use of microsatellites in many different types of studies, the mutational process in microsatellites is still not fully understood (Garza et al, 1995). Models suggested to explain microsatellite mutation are mainly the Infinite Allele Model (IAM) (Kimura & Crow, 1964) and different variants of the Stepwise Mutation Model (SMM) (Ohta & Kimura, 1973; Valdes et al, 1993; Shriver et al, 1993; see also Garza et al, 1995). The main difference between the two models is whether mutation results in unique alleles or not. Under the IAM, mutation of a microsatellite allele could be the gain or loss of any number of tandem repeats and result in a new allele state not previously found in the population. The SMM on the other hand, involves a gain or loss of a single repeat unit, and as a consequence alleles identical in state (IIS), homoplasy, could be quite common. Other mutation models that have been introduced are often found to be variants of SMM or IAM, for example the two-phase model (TPM) (DiRienzo et al, 1994) and the K-allele model (KAM) (Crow & Kimura, 1970). If a major advantage with microsatellites is their high variability, a major drawback could be that they often are species-specific, implying that they need to be isolated de novo from most species, even though there are reports of primers isolated from highly conserved regions allowing crossamplification between different species (Primmer et al, 1996 and references therein) and sometimes even between species diverging millions of years ago (see Zane et al, 2002).. Development of microsatellite loci Traditionally, the development of microsatellite loci consists of screening genomic libraries of the species of interest. The procedure initially involves fragmentation of high quality DNA using restriction enzymes. Short fragments, with sizes around 300-700 base pairs (bp), are selected. The selected DNA fragments are ligated into plasmid vectors and the products are transformed to bacterial cells. In the next step the recombinant clones are screened for the presence of microsatellite sequences, generally with a 10.

(26) technique called Southern hybridization, in which repeat-containing probes are hybridized to corresponding bacterial colonies. The positive colonies are then transferred to new plates, either by picking single colonies on new plates or by replica plating. As the frequency of microsatellites varies in different taxa and in different parts of the genome, the outcome of clones containing microsatellites, positive clones, may vary considerably across groups of taxa (Zane et al, 2002). In different studies, ranges from 12% to less than 0.04% have been described (Zane et al, 2002). Due to the sometimes ineffective outcome and time consuming procedure in developing microsatellite loci, different enrichment protocols, in which the genomic libraries are highly enriched for specific microsatellite repeats, have been devised (Fleischer & Loew, 1995; Zane et al, 2002). The clones containing a repeat are sequenced and visualized with different available techniques, for example radioactivity, silverstaining or fluorescence. Sequences containing microsatellites are further analysed and at the last step, primers are designed on both sides of the microsatellite repeat.. Genetic and statistical analyses The eight available microsatellite loci in this thesis, Gc02, Gc17, Gc29, Gc31, Gc42, Gc49, Gc51 and Gc77, developed in the orchid species Gymnadenia conopsea (see paper I for further information) were, depending on variability, amplification and occurrence of potential null alleles, used in different combinations in different studies. Paper II, focussing on mating pattern in G. conopsea, included all loci except Gc31 which expressed very low polymorphism. In paper V, about flowering frequency in a population of G. conopsea, Gc31 and Gc02 were excluded since they gave no further information. Gc02 had most likely null alleles as all individuals did not amplify, but the locus was included in paper II although not in the statistical analyses. In paper III, concerning genetic variation in G. conopsea, the available loci were from the unenriched method (Gc17, Gc29 and Gc49) since this study was performed before the enriched library was accomplished. Gc29, Gc42, Gc49, Gc51 and Gc77 were the five loci amplifying in G. odoratissima and thereby included in paper IV. Paper VI were based on Gc29, Gc31, Gc42 and Gc51 which were the loci amplifying in the late-flowering-time variant of G. conopsea and the last three loci were diagnostic according to flowering-time. All investigations were based on a combination of field studies and laboratory work. Leaf material and in paper II seed material as well, were sampled and stored in freezer (-20oC) or dried using silica grains before DNA extraction. In paper II, the direction of pollen removal and deposition were analysed by manually scoring patterns of microsatellite multilocus 11.

(27) genotypes from seeds. All seeds in a fruit were analysed together in a capsule seed-mass, due to their minute size. The genotypic pattern revealed from each capsule seed-mass was compared with pattern from all possible combinations of alleles from analysed leaf material of every flowering individual in the population. The analyses were considered at each locus with the most easily scored and second most variable locus (Gc49) being decisive. The remaining putative pollen-donors were then matched locus by locus, excluding non-possible fathers. Prior to using microsatellites in a study of mating pattern (II) the markers were tested whether they showed Mendelian inheritance. Ten individuals were caged and cross-pollinated manually in a controlled mating experiment. At fruitset, DNA were extracted from the seeds in all fruits and analysed at each of the eight microsatellite loci. The aim of paper V was to evaluate the use of genetic markers, microsatellites as individual genetic markers to distinguish among individual plants in different years. The markers discriminative strength was calculated as: 1 . ¦ P

(28) 2 ij. mainly following Kloosterman et al (1993).. Population structure In analyses of population structure, most analyses were performed using the software program package Genepop 1.2 or 3.1b (Raymond & Rousset, 1995). Genepop tests for Hardy Weinberg equilibrium, non-random association of genotypes among loci and perform estimates of population parameters like FST and RST, and the effective number of migrants (Nem). The two different methods, available by Genepop, to quantify genetic structure between populations and among all pairs of populations are based on different mutation models, FST (Wright, 1969; 1978) based on IAM and RST (Slatkin, 1995) based on SMM, or more accurate, Genepop perform the unbiased estimates of FST (T) (Weir & Cockerham, 1984) and RST (U) (Michalakis & Excoffier, 1996) estimated by a weighted analyses of variance. As it is still debated which mutation model (IAM or SMM) best fits microsatellite mutation, results from unbiased estimates of both FST (T) and RST (U) were compared in paper III, although in paper IV, only FST (T) was estimated. With accumulating data, it have been quite clearly stated that no single model could explain mutation dynamics in all microsatellite loci (Gaggiotti et al, 1999). In the study of Slatkin (1995), it was shown that if microsatellite loci strictly follow a SMM, RST in general gave better 12.

(29) estimates of the effective number of migrants (Nem). However, if SMM is not appropriate for the investigated microsatellite loci, the performance of RST will suffer accordingly (Slatkin, 1995). Estimates of FST were fully applicable if the mutation rate not exceeded 10-3 and if the time scales of interest was sufficiently short that mutations were of no immediate importance, i.e. in time periods shorter than tens or hundreds of generations. In the study by Gaggiotti et al (1999) better estimations of Nem based on FST were found when sample sizes were small (d10

(30) and when the numbers of scored loci were few (20

(31) In paper (III) I also estimated whether FST were correlated with geographic distance involving the regression of FST. 1 FST

(32) estimates for pairs of subpopulations on geographic distances (Rousset, 1997). Further the effective number of migrants (Nem) was calculated from the FST and RhoST estimates, as (Slatkin, 1995) FST. 1 4 N e m 1

(33). Nem was also estimated in an alternative way using the occurrence of private alleles, where the average frequency of alleles found only in one population was calculated per location (Slatkin, 1985).. ҏAssignment methods Another approach, instead of estimating population divergence, applied in paper III and IV is to classify individuals into populations. Such an assignment test will determine how indicative an individual’s genotype is of the population in which it was sampled (Paetkau & Strobeck, 1995). The suggestion is that individuals coming from the same population exhibit more similar genotypes. The computer program GeneClass, version 1.0.02 (Cornuet et al. 1999) supply different methods for the assignment of individuals to populations.. Genetic differentiation within species In paper VI, concerning differentiation in flowering-time variants within G. conopsea, the results were based on microsatellites and sequences in the internal transcribed spacer (ITS). In this paper we also analysed differences 13.

(34) in habitat preference within sympatric populations of the two phenological variants. We described habitat differentiation between early and lateflowering individuals using a linear discriminant analysis (LDA) and the significance of the habitat variables that discriminate between the flowering types was determined using a generalized linear model.. Study species The genus Gymnadenia The 10 members in the orchid genus Gymnadenia are distributed around Eurasia. All members are terrestrial, have a leafy stem and multi flowered spike. They are fragrant flowers, attractive to different species of moths and butterflies which are rewarded by nectar (van der Cingel, 1995). The aggregations of pollen grains in Gymnadenia, in contrast to pollinia in more advanced orchid groups, are produced by elastic threads holding the lumps of pollen, or massulae, together (Johnson & Edwards, 2000). This means that an insect with attached pollinia might be able to visit and pollinate several flowers successively, either within an individual or between individuals, with the same pollinia. At pollinia removal, the viscidia become fixed to the proboscis of the pollinating insect (Proctor et al, 1996). The fruit contains thousands or even millions of tiny, wind spread seeds. In Sweden, there are two species in the genus Gymnadenia, namely, G. conopsea, the fragrant orchid and G. odoratissima, the lesser fragrant orchid. Gymnadenia conopsea (L.) R.Br. (Figure 1a) is found in cultivated areas like grasslands and grazed meadows, sometimes close to marshes and fens. The geographic distribution covers the Eurasian area (Komarov, 1935). It is not a rare orchid, although during the recent decades, the number of populations has diminished in many areas, due to human influences. G. conopsea is quite variable and a large number of subspecies, variants and forms are described (Hegi, 1909). Most commonly the species is divided into subspecies ssp conopsea and ssp densiflora. The division is based on morphological and phenological characters with ssp densiflora being larger, more densely flowered and late-flowering (in July in northern Europe) compared to the main form which flowers in June (northern Europe). However, the morphological variation within the late-flowering variant is not distinct and individuals morphologically similar to the early-flowering ssp conopsea but late-flowering as ssp densiflora do occur. The different flowering time variants can either be found in separate or in mixed populations.. 14.

(35) a. b. Figure 1 a) Gymnadenia conopsea (Orchidaceae) and b) G. odoratissima (illustration by Ditte Werner). G. conopsea has a slender spike, about 20-60 cm high, although some individuals of densiflora could reach 1 meter. It has 4-8 linear-lanceolate, keeled leaves. The colour varies from pale pink to cerise or lilac and rarely to pure white (Mossberg & Nilsson, 1985). The labellum is trilobed with a long, basal spur and the column is short and erect (Proctor et al., 1996). It is frequently visited by pollinators, different species of moths and butterflies (Proctor et al., 1996), and has usually a high fruitset. G. conopsea is selfcompatible, but seems to be dependent on a pollinator for fruit set (Gustafsson, unpublished data). The fruit matures and the seeds are spread by the wind in July, in densiflora in August (northern Europe). 15.

(36) Gymnadenia odoratissima (L.) L.C.M. (Figure 1b) is a smaller plant compared to G. conopsea and has a more sparsely flowered raceme. The distribution is restricted to Europe, where it mostly has a scattered distribution within calcareous chars or grasslands (Mossberg & Nilsson, 1985). In Sweden the occurrence is limited and it is only found on the island of Gotland where it still is fairly abundant and in two remaining mainland provinces (Västergötland and Östergötland) (Mossberg & Nilsson, 1985). The height is 15-45 cm and it has 4-8 linear-lanceolate, keeled leaves. The colour varies from dark pink to lilac and rarely pure white (Mossberg & Nilsson, 1985). It has a three lobed lip and blunt side lobes (Proctor et al, 1996). The spur is shorter than in G. conopsea and the pollinators are mostly thought to be night-flying moth species. As the name intends, G. odoratissima is extremely fragrant and the odour reminds clove. In the northern parts of Europe it begins to flower in July.. 16.

(37) Results and discussion. In this thesis I have studied genetic structure at different levels in the orchid genus Gymnadenia using microsatellite markers developed in Gymnadenia conopsea. I analysed patterns of genetic variation within and made comparisons between two congeners, G. conopsea, geographically widespread although diminishing and G. odoratissima having a long history of being rare. In G. conopsea I also made studies at a more detailed level. I investigated mating pattern during one year and variation in flowering frequency during five years in a small population and finally I studied genetic structure between early- and late-flowering time variants of G. conopsea.. Microsatellite loci in Gymnadenia conopsea (I) Microsatellites have become one of the most popular molecular markers since the last few years. They are used in a wide variety of different genetic fields. In a computer search from 1989-2002, about 13 450 documents matched the word microsatellites (http://isi2.isiknowledge.com). In 2002 about 2200 matches were found. Of the 13 450 documents matching the word microsatellite, 250 were concerning plants and only eight of them orchids (http://isi2.isiknowledge.com). In Gymnadenia conopsea, due to a low outcome of clones containing microsatellites, two genomic libraries were constructed one without enrichment and one enriched library. In the procedure without enrichment of the genomic library, 20 positive clones out of 5000 recombinants contained microsatellites (0.4%), whereas for the enriched genomic library 30 microsatellites out of 1000 recombinant colonies were detected (3%). Of the 50 detected positive clones, only eight were both variable and gave scorable results (Gc17, Gc29, Gc42, Gc49, Gc51, Gc77, Gc02 and Gc31) and the first six are included in Table 1. In a more recent study in G. conopsea, Campbell et al. (2002), found about 6.9% of the recombinant clones appearing positive. The relative abundance of microsatellites and motives has often been found to be taxon specific and varies a lot among different species. In plants, the occurrences of microsatellites are, in general, up to five times less compared to mammals (Lagercrantz et al, 1993). 17.

(38) (CT)16. (CT)17. (CT)16. (CT)20. 132. 147. 18. 81. 171. 183. 199. Size (bp). Above line from un-enriched library; below line from enriched genomic library *GC17 consist of a complicated repeat motif with several short stretches of (CT)n. Gc77. Gc51. Gc42. Gc49. (CT)10. (CT)n*. U - GCCATAAATGCTCAGAAATGC L - GAGCTCATGCCCTTCTCC U - CATCTACACAATCATCCTAAGAAG L - CTAGACGCCATGACTTACATG U - TCTTTAACAGTTAACAATCTTATCTC L - CATTTAGAAGCAGGAGCAG U - GAGTGAAGTGTCTTTAATCGATAAC L - GGGAGAAAGAGTGTGCATGT U - GATCCTAGCTTTCGTTTCAT L - AGTAATCGAGGCAACCTG U - TCTTACAACATTTAGGACTC L - GCACAAGAATCTGTCATTA. Gc17. Gc29. Repeat. Primer sequences 5´-3´orientation. Locus. 6. 10. 6. 9. 8. 4. No. alleles. 52. 60. 52. 57. 63. 64. 1.5. 1.5. 1.5. 1.5. 1.2. 1.0. Tao MgCl2 mM. 0.54. 0.90. 0.75. 0.85. 0.90. 0.65. Hobs. 0.79. 0.88. 0.72. 0.83. 0.84. 0.54. Hexp. 0.09. 0.36. 0.89. 0.49. 0.52. 0.51. HW (p). AF319990. AF319989. AF319988. AF319987. AF319986. GenBank accession no. AF319985. Table 1 Primer sequences of six microsatellite loci in the fragrant orchid. Repeat motif and size in no. of basepairs refers to the alleles of the sequenced clones. The number of observed alleles, observed heterozygosity (Hobs), expected heterozygosity (Hexp) and probabilities of deviation from Hardy-Weinberg equilibrium (HW) were determined from 20 individuals originating from one population of G. conopsea located in Eneby, southeast Sweden. Given are also optimal annealing temperature (Tao), MgCl2 concentration used in the PCR reactions and GenBank accession numbers..

(39) One of the highest densities of dinucleotide repeats has been reported from Hymenoptera. In Vespula rufa, (CT)n repeats were found every 2,5 kb and (GT)n every 8 kb (Thorén et al, 1995). On average, though, Beckmann and Weber (1992) found a microsatellite repeat every 6 kb in the eukaryotic genome. In plants, the most abundant repeat motif is AA/TT followed by AT/TA and GA/CT, whereas in mammals the GT/CA motif is most frequent (Lagercrantz et al, 1993). The commonest microsatellite repeat found in G. conopsea during the development procedure was GA/CT followed by GT/CA although AT/TA repeats could not be compared as they were not screened for due to their palindrome nature and therefore more complicated screening procedure. As well as a large variation in the abundance of microsatellites, the level of polymorphism varies a lot among different loci within and between species, indicating that there is a great variation in the mutation rate (Estoup & Cornuet, 1999 and references therein). In general, dinucleotide repeats seem to have the highest mutation rate, followed by (non-disease-related) triand tetranucleotide loci. The polymorphic repeats found in G. conopsea were exclusively dinucleotides. Primers were constructed for two trinucleotides, of which both were monomorphic. Other factors suggested to influence the variability is the composition of the repeat type (Gastier et al, 1995; Sheffield et al 1995) and the number of repeats, where the mutability appears to increase with an increasing number of repeat units (but see below about restrictions in allele size) (Estoup & Cornuet, 1999 and references therein). In the G. conopsea test material (based on 20 individuals), the locus with the largest number of repeats was the second most polymorphic (Table 1). The purity of the microsatellite repeat could also influence the stability of the repeat. Perfect repeats (e.g. (CA)n, (GT)n and (AT)n) seem to have a higher mutability than interrupted repeats (e.g. (CA)nGA(CA)n). The interrupting bases could have a stabilizing effect, reducing the possibility of misalignment (Richards & Sutherland, 1994; Pépin et al, 1995). In G. conopsea, one locus consisted of an interrupted repeat motif and this locus was the least polymorphic with 4 alleles (Table 1). The number of alleles found in the other five loci varied from 6 to 10, determined from 20 investigated individuals (Table 1). The observed heterozygosity in the two most variable loci was 0.9 (Gc29 and Gc49) and in the least variable microsatellite locus 0.54 (Gc77) (Table 1).. 19.

(40) Mating pattern (II) and flowering frequency (V) In this study, mating pattern or gene flow and fruit set accomplished by pollinator movement were analysed in a small G. conopsea population. Data were based on information from 7 microsatellite loci, Gc17, Gc29, Gc42, Gc49, Gc51, Gc77 and Gc02 - of which the first 6 were described in Table 1 (I). Pollination of G. conopsea is mainly performed by different Lepidopteran species, which are supposed to fly quite long distances (Schmitt, 1980; Waser, 1982). In this small (25 flowering individuals) population of G. conopsea though, individuals were non-randomly pollinated and the majority of pollinations were near neighbour-matings (Figure 2). Furthermore, there was a clear pattern with pollinators visiting additional flowers in one inflorescence, i.e. many fruits in an inflorescence were sired by the same pollen-donor. With no mechanisms prohibiting self-fertilization, this could result in high levels of self pollinated seeds through geitonogamous pollinations and the level of geitonogamy has, in plants showing multifloral inflorescences, sometimes found to be quite frequent (Geber, 1985; de Jong 1992, 1993). In G. conopsea, there is a pollinarium bending mechanism, which has been suggested to reduce selfing through geitonogamous pollinations (Johnson & Nilsson, 1999; Johnson & Edwards, 2000). Despite that, seeds in about 10% of the fruits were fertilized by pollen from the mother plant solely, although the total level of self-fertilization could not be determined as all seeds in a fruit were analysed jointly, as a capsule seedmass.. 20.

(41) Figure 2 Number of matings according to the relative distance between flowering individuals The revealed results with near-neighbour matings and geitonogamous pollinations could be a consequence of the large amount of available nectar tempting the pollinators to fly short distances for foraging and to do repeatable visits within the same inflorescence. Predominant short distant pollinations have also been found in other rewarding plant species pollinated by Lepidoptera (Nilsson et al, 1992b; Broyles & Wyatt, 1991). With such a mating pattern, one could expect a genetic structure with detectable levels of non-random mating. However, the present population was investigated during five years in a study concerning flowering frequency (V), where low inbreeding values (FIS) were found over all loci although the values varied among years and loci. There was also a high tendency for an individual to remain vegetative or rest underground the year after flowering (V) and analysing the number of different multilocus genotypes in the population during five years, the population was actually comprised of totally 85 different individuals, although just a small part was flowering each year (2448%). This means that the population almost every year were greatly reconstructed by a different composition of individuals. Such a strategy, with the randomized effect resulting from different compositions of flowering individuals each year and the high potential for long-distance seed dispersal in G. conopsea, through wind dispersed seeds, will probably minimize the risk of an increased of inbreeding resulting from near neighbour mating and geitonogamous pollination.. 21.

(42) Genetic structure in rare and diminishing orchid species (III), (IV) Genetic structure was analysed in two congeners, G. conopsea and G. odoratissima, one more widespread, although diminishing and one rare orchid species. Population structure and genetic differentiation appears to be in accordance to a generalized pattern with higher levels of genetic variation within and lower levels of genetic differentiation between populations in the more widespread G. conopsea, compared to the geographically restricted G. odoratissima (Table 2, 3). G. odoratissima have had a long history of being rare in Sweden and it is presently categorised as Near Threatened in the Swedish red list (Gärdenfors, 2000). The distribution is restricted to the island of Gotland, where the species still is fairly abundant, and two mainland provinces, Västergötland and Östergötland. There was a clear distinction in genetic structure among G. odoratissima populations, with in comparison lower levels of genetic variation within the remaining mainland populations, whereas populations on the island of Gotland showed quite high levels of genetic variation (Table 3) (IV). Also a strong genetic structure was detected between the two mainland provinces as well as between mainland and island populations. In contrast low genetic differentiation, indicating high levels of gene flow, was found among the more numerous and geographically larger populations on the island of Gotland (Table 4) (IV). The observed pattern, resulting from differences in allele frequencies between populations and not from unique alleles, could be due to foundation event where mainland populations were founded by individuals originating from the island populations which thereafter remained quite isolated. Another explanation could be by isolation as a vicariant event, where the relatively large number of interconnected populations on the island most likely reduces the effects of genetic drift, in contrast to the small mainland populations.. 22.

(43) 10 13 12 20 19 20 20 20 20 20 174. Ruddmossen** Bråfors** Lämpesbo** Eneby Väddö Tistbrottet Gruvan Karsbo Styrars Norrängen All samples. 2 5 3 4 4 7 4 4 5 5 10. AO 2.00 2.62 2.70 2.13 2.10 3.57 2.69 2.03 2.28 3.24. AE* 0.40 0.54 0.72 0.65 0.26 0.75 0.60 0.40 0.45 0.45. HO. *AE=1 / 6 pi2, pi is the frequency of the ith allele **All flowering individuals 1995 were investigated. N. Populations. Locus Gc17 0.52 0.64 0.65 0.54 0.54 0.74 0.64 0.52 0.49 0.71. HE 4 6 6 8 8 6 4 5 6 5 8. AO. 23. 2.04 2.68 4.31 5.41 3.54 3.83 2.57 3.67 2.96 3.45. AE* 0.50 0.69 0.83 0.90 0.68 0.70 0.70 0.95 0.60 0.75. HO. Locus Gc29 0.50 0.65 0.79 0.84 0.74 0.76 0.63 0.74 0.68 0.73. HE 6 7 10 10 10 16 15 11 10 13 24. AO 4.88 5.12 6.88 5.30 6.56 10.1 7.34 7.89 5.63 8.16. AE* 0.80 0.69 0.92 0.85 0.68 0.70 0.85 0.53 0.70 0.80. HO. Locus Gc49 0.84 0.84 0.99 0.83 0.87 0.92 0.89 0.90 0.84 0.90. HE. 4.00 6.00 6.33 7.33 7.33 9.67 7.67 6.67 7.00 7.67 14.3. AOL. 0.56 0.64 0.82 0.80 0.54 0.72 0.72 0.62 0.58 0.67. HOL. All loci. observed (AO) and effective (AE) number of alleles, and observed (HO) and expected (HE) heterozygosity at loci Gc17, Gc29, Gc49 and over all loci (AOL, HOL and HEL). 0.62 0.71 0.78 0.74 0.72 0.81 0.72 0.72 0.67 0.78. HEL. Table 2 Localities of Gymnadenia conopsea listed according to approximate population size, number of individuals investigated/population (N),.

(44) 0.22 0.47 0.37 0.47 0.44 0.10 0.05. Hoburgsmyr* 20 Lojsta* 20 Horsan* 11 Besteträsk* 20 Omberg** 20 Skogatorp*** 20 Svartarp*** 20 Total 131. 0.15 0.20 0.09 0.35 0 0.10 0.05. Ho. * = Populations from Gotland ** = Population from Östergötland *** = Populations from Västergötland. He. N. Population. Gc29 2 4 2 2 2 2 2 5. Ao 0.80 0.79 0.87 0.76 0.63 0.58 0.58. He 0.75 0.90 0.91 0.65 0.50 0.42 0.60. Ho. Gc42 6 7 7 7 3 3 6 14. Ao. Ho 0.85 0.55 0.73 0.95 0.10 0.35 0.70. 24. 0.84 0.86 0.74 0.84 0.10 0.32 0.79. He. Gc49. Locus. 8 9 8 10 2 4 6 17. Ao 0.60 0.73 0.40 0.71 0 0.50 0.45. He 0.50 0.65 0.36 0.65 0 0.65 0.25. Ho. Gc51 6 5 4 5 1 3 2 7. Ao 0.58 0.36 0.55 0.76 0.33 0 0.24. He 0.58 0.30 0.40 0.25 0.05 0 0.25. Ho. Gc77 6 5 4 6 4 1 4 8. Ao. 0.61 0.62 0.59 0.71 0.30 0.30 0.42. 0.57 0.52 0.50 0.57 0.13 0.30 0.37. 5.6 6.0 5.0 6.0 2.4 2.6 4.0 10.2. HEL HOL AOL. Over all loci. Table 3 Localities of Gymnadenia odoratissima, number of individuals investigated/population (N), expected (HE) and observed (HO) heterozygosity and observed (AO) number of alleles at loci Gc29, Gc42, Gc49 Gc51 and Gc77 and over all loci (HEL, HOL and AOL).

(45) Table 4 Pairwise FST over all loci below diagonal and pairwise geographic distances (km) above diagonal between investigated populations of Gymnadenia odoratissima Population Hoburgsmyr Lojsta Horsan Besteträsk Omberg Skogatorpskärret Svartarpskärret. Hob 0.04 0.01 0.04 0.33 0.23 0.14. Lo 63 0.05 0.03 0.29 0.22 0.16. Hor 4.5 66 0.04 0.30 0.25 0.18. Be 9 80 6.5 0.28 0.25 0.19. Om 283 285 284 290 0.29 0.40. Sk 332 326 335 338 54. Sv 331 325 334 338 56 6. 0.20. According to revealed results from microsatellite markers in G. conopsea, the high genetic variation within and low genetic variation among populations could be interpreted as if the species is relatively unaffected of the current reductions in population number. However, significant genotypic differentiation and significant isolation by distance indicated restriction in gene flow between populations, i.e. the populations can not be regarded as one panmictic unit (III). Further, as only a small number of generations have passed since fragmentation in the region commenced, the species could be in an early stage of losing genetic variability, and due to a time lag in the genetic markers exhibiting a scarcely noticeable pattern (see also Ehlers & Pedersen, 2000). In addition, G. conopsea is a perennial species with overlapping generations and, according to Vöth (1980), a fairly long generation time (up to ten years) which might slow down the loss of genetic variation. In an earlier study of G. conopsea, by Scacchi and Angelis (1989) based on 11 allozyme loci, data from 16 Italian populations showed a similar genetic pattern as was revealed by microsatellites (III). No comparable genetic studies could be found in G. odoratissima. It seems like the observed differences in overall genetic pattern between G. conopsea and G. odoratissima could be explained by generalized genetic theories, like differences in geographic distribution, the number of populations and population size. The conditions for gene flow by seed dispersal will also be similar with tiny, wind dispersed seeds. However, gene dispersal by pollen will most likely differ between the two congeners. Pollinators of G. conopsea among others are strong flying, sometimes even migrating Lepidopteran species whereas G. odoratissima is mainly pollinated by more sedentary butterflies. In G. odoratissima, the Baltic Sea thereby might function as a barrier to gene flow by pollen, and could explain the mainland – island differentiation. A more homogenous genetic pattern was found among populations of G. conopsea, pollinated by highly mobile 25.

(46) butterflies and moths. Even if the relative importance of long distance dispersal by seed versus pollen was not investigated in this study, orchids are in general considered to be excellent long-distance dispersers due to their light, numerous and wind dispersed seeds, thereby promoting the abilities of a geographic structure with low levels of differentiation between populations. Despite that, the Baltic Sea seems to be a barrier even to seed dispersal in populations of G. odoratissima.. Restricted gene flow within species (VI) The relatively high levels of genetic differentiation found among island and mainland populations of G. odoratissima, FST ranging from 0.14-0.33 in the pairwise population comparisons (IV), pointed at some sort of dispersal barrier. A genetic pattern with an even larger amount of genetic divergence, than between island and mainland populations of G. odoratissima, was actually found between flowering-time variants (early and late) of G. conopsea, investigated with microsatellites and ITS-markers (VI). Differentiation in flowering-time is an example of a mechanism preventing gene flow and promoting genetic divergence and has been observed within several other plant taxa of which a few examples are Euphrasia, (Karlsson, 1984), Beta (Van Dijk et al, 1997), Gentianella (Lennartsson, 1997), Silene (Hauser & Weidema, 2000), Capsella (Neuffer & Hurka, 1999); see Lennartsson (1997) for additional taxa. Commonly, G. conopsea is divided into two subspecies, the earlyflowering ssp conopsea and the late-flowering ssp densiflora (Mossberg and Stenbeck, 1992). Besides phenological variation, the division is based on morphological characters with ssp densiflora being higher, more densed flowered and with broader leaves. In a study by Soliva and Widmer (1999), large levels of genetic divergence between ssp conopsea and ssp densiflora was described and Scacchi and Angelis (1989) detected high allozyme divergence between 16 Italian populations of G. conopsea divided into humid and dry ecotypes respectively. Unfortunately Scacchi and Angelis (1989) did not study the variation in flowering-time. In the present study, among the late-flowering form, we observed individuals morphologically similar to the early-flowering ssp conopsea, although late-flowering as ssp densiflora. Early- and late-flowering variants of G. conopsea were separated in the ITS haplotypes as well as microsatellite allele lengths and frequencies and exhibited some differences in habitat preference, indicating extremely low levels of genetic exchange (Table 5, Figure 3a-d). The late-flowering ssp densiflora could not be distinguished from the late-flowering ssp conopsea neither in ITS sequences 26.

(47) nor microsatellite allele frequencies. The early-flowering variant was genetically the most variable, showing the largest number of alleles at the microsatellite loci, whereas the late-flowering ssp conopsea and ssp densiflora were much less variable (Table 6). Additionally, individuals of its rare congener G. odoratissima were found to be almost identical in ITS sequences to those of the early-flowering time variant of ssp conopsea. Table 5 Positions and base substitutions in ITS 1 and 2 in the two flowering types of Gymnadenia conopsea, G. conopsea ssp densiflora and G. odoratissima.. Gymnadenia 64 odoratissima T ssp con early T late ssp con C C ssp den. Position* and base Position* and base substitution substitution ITS1 ITS2 108 178 187 202 38 58 79 81 144 186 192 A A A G G C C G A T A A A A G G C C G A T A G G C C A T T T T A T G G C C A T T T T A T. *According to Gymnadenia conopsea EMBL database Accession no GCZ94067 and 68.. Table 6 The total number of alleles (AO) and observed (HO) heterozygosity per locus and flowering-time variants of Gymnadenia conopsea.. Flowering types Gymnadenia early-flowering late-flowering. Gc29 AO HO 8 0.72 6 0.53. Gc31 AO HO 3 0.14 2 0.03. 27. Gc42 AO HO 17 0.90 2 0.41. Gc51 AO HO 35 0.91 3 0.32.

(48) 28.

(49) 29.

(50) Questions asked in the study by Soliva and Widmer (1999) as well as in the present study is what processes could be involved in performing and maintaining such a strong genetic differentiation and such differences in levels of genetic variation between the two flowering-time variants? Soliva and Widmer (1999) proposed the difference could be a consequence of ssp densiflora growing in more moist habitats - habitats that have been strongly reduced over the last century. Consequently, populations of ssp densiflora have been reduced in size and would thereby be more exposed to random genetic drift. This could of course be a possible explanation, but if genetic drift in a relatively polymorphic regional population of the late-flowering variant alone had created the low genetic diversity, local populations should have a mosaic variation with different alleles fixed. This was not the case, since all late-flowering individuals in three out of four loci had more or less unique allele lengths found within the total length distribution of the earlyflowering type alleles. The information from ITS sequences indicated the occurrence of an early historical split between the two phenological conopsea variants, more ancient than between the two different species, G. conopsea and G. odoratissima. It is commonly accepted that founder events usually result in a loss of genetic diversity (Frankham et al 2000). As early colonization history in Sweden is not known and we have no information about number of founders or duration of time between gene flow events, the Swedish regional populations could have been founded by only few individuals. Thus, the low diversity in late-flowering populations could be explained by founder effect. However, a question will still be why mutation, which have resulted in 12 base pair substitutions in ITS between early- and late flowering time variants have not increased the genetic variation in the investigated microsatellite loci also in the late flowering type. Reasons could be selection or/and genetic drift if the late-flowering form have a more narrow habitat amplitude and geographic distribution. Even though differentiation in flowering-time will be an important mechanism maintaining the genetic separation between early- and lateflowering time variants of G. conopsea, flowering-time will, at least some years, overlap in the middle of the season and we did also find a few possible hybrids between the flowering time types which combinations had not spread within populations. Thereby, the strong genetic differentiation, with very low exchange of genes between the two flowering-time variants occurring in sympatry, probably do suggest additional mechanisms or processes preventing crosses or making hybrid seed less fertile. Some examples could be ecological isolation due to habitat preferences, which was also indicated in our study as well as in Soliva and Widmer (1999); isolation due to different species of pollinators (different sets of Lepidoptera - see 30.

(51) Nilsson, 1983), genetic incompatibility between flowering-time types and/or differences in chromosome numbers. According to the last two examples, genetic incompatibility could be unlikely given the number and frequency of interspecific and integenic hybrids in which G. conopsea have found to be involved. With the microsatellite markers used in our study and by allozymes in the study of Soliva and Widmer (1999), there was no evidence for ssp densiflora being tetraploid, indicating no simple ploidy level differences between the two flowering-time variants. To be able to clarify which genetic processes that are maintaining the separation between the early and late-flowering types we need to investigate mixed populations in more detail, screening for hybrids, doing additional cross-pollinations, germinating "hybrid" seed etc. It is also important from a conservational point of view to examine the actual distribution of mixed and distinct populations and to revise the taxonomic status of the two floweringtime variants. At present, populations of the early-flowering type are endangered due to reduced mowing and cultivation, whereas late-flowering populations and ssp densiflora which are less dependent on management slowly increase their acreage. This situation could actually lead to the extinction of the genetically unique early-flowering type while the species G. conopsea seem to thrive in Sweden. In this case we would also loose a substantial part of genetic diversification and a large part of the evolutionary potential in the genus Gymnadenia in Sweden.. 31.

(52) Conclusions. Altogether, according to information revealed by microsatellite markers, populations of Gymnadenia conopsea in the investigated area (parts of Sweden) seem to be connected by a certain amount of gene flow and harbour quite high levels of genetic variation. A generally high fruitset in G. conopsea and the large amounts of wind-spread seeds in each fruit do create good conditions for long-distance dispersal, even though the Baltic Sea could be acting as geographic barriers to long-distance seed dispersal, as was seen in G. odoratissima. However, the fact that the highest level of pollen exchange was between neighbours could, if neighbours are more closely related than more distanced individuals, as well as the possibility of geitonogamous pollinations, increase the risk of inbreeding in the future. In studies using genetic markers it is also important to be aware of the potential effect of time delays as G. conopsea is perennial and the time since habitat fragmentation commenced in the region could be too short to yield any noticeable genetic pattern so far. In addition, a long generation time will probably slow down the loss of genetic variation. The high levels of genetic variation within G. conopsea were though restricted to the early-flowering variant, whereas the late-flowering type was comparatively depauperate. The great genetic differentiation found between the two flowering-time variants within G. conopsea is probably due to a more ancient historical separation than the separation between the two different species, G. conopsea and G. odoratissima. At present, populations of the early-flowering time variant are diminishing due to reduced mowing and cultivation, whereas the late-flowering form, which is less dependent on cultivation, slowly increases its acreage. This situation could actually lead to the extinction of the early-flowering variant while the species G. conopsea seem to thrive. In this case we could lose a substantial part of genetic diversification and probably the main part of evolutionary potential in the genus Gymnadenia in Sweden, equal to the loss of a recognized species. Therefore it is important to design future conservation work using a "two separate species" approach, aiming at conserving all forms. In the rare congener G. odoratissima, the level of genetic variation varied considerably among populations. Populations on the island of Gotland showed quite high levels of genetic variation whereas reduced genetic 32.

(53) variation was found in the mainland populations. Thus, there is a potential genetic resource in the more numerous populations on the island of Gotland. Since mainland populations of G. odoratissima were clearly differentiated from the Gotland populations, it could be that some of the genetic variation actually is specific and adapted to mainland conditions whereby introduction of genetic material from the island of Gotland to mainland populations could result in lower survival or establishment of the introduced material. If such supportive actions for mainland populations are planned, this would have to be considered and tested experimentally.. 33.

(54) Acknowledgements. First of all I want to thank my two supervisors (for some PhD students one is not enough), Prof Pekka Pamilo and Doc. Per Sjögren-Gulve. Without your support and help this thesis would not have been possible to do. Pekka, I realize that I have been extremely privileged to have you as a supervisor, but I have not totally forgiven you for not letting me complete my first publication about genetic structure in G. conopsea with additional loci. For the first time, and the last, he did not answer as I expected him to do: “Hmm, one could either include more loci or one could not include more loci”. The answer I got was: “I think three loci is enough”. Guess if I was surprised… So, to the opponent: It is not my fault; don’t shoot the piano player. Per, you have been great. Every time you came to the department I got a “spark I baken”? And I can tell you in confidence that I needed that. You and your lovely family are also interested in the conservation work concerning old Swedish domestic breeds, something that also is very close to my heart. In the next “First of all” I want to thank my dearest Prof Martin Lascoux. I will remember our discussions about what is the most ethical – to eat “djyr som har lidit eller djyr som inte har lidit”? Of course you as well (as all Frenchmen?) will have this inverse logic that it’s more ethic to kill animals that have suffered than to kill the happy ones. Well, I guess it’s hopeless to teach some Frenchmen in animal ethics. Besides that, you have helped me a lot with this thesis, both reading bad manuscripts and trying to make me understand horrible statistics. You have never let me feel that you are too busy for helping with a problem. Hmmm, maybe you are never busy… First of all I also want to thank Prof. L. Anders Nilsson for always being kind and reading and improving my manuscripts. Anders, sorry for not doing these interesting ecological studies we discussed in the beginning, but as time and luck are factors influencing the rate of success these studies were unfortunately left behind. Thanks also to everyone at the department of Conservation genetics and especially to my both room mates, Anna and Marita. By the way, I guess it’s soon time for ordering new batches of seeds for the garden. With all my heart I also want to thank Marianne Heikenskiöld and Mats Block for helping me with very important things. Without your kindness, things would have been extremely complicated. Rose-Marie Andersson, for helping with 34.

(55) the most various things and for administrating all (!) my money. Unfortunately I always hoped that there would be more left. Kerstin Santesson, you are the sunshine in the lab. Without you being so extremely positive the sometimes very boring lab work would have been unbearable. Please tell me your secret; could it be the magic key to the room downstairs? Thanks also for letting me have nice dog talks and sending me funny emails. Cia Olsson, for being one of the most pleasant persons I´ve ever known and for extremely skilful lab work. Nisse, you are such a cheerful person. It is always a pleasure to meet and talk to you in the corridor. Aha, now I know the secret, both you and Kerstin originate from the southern and warmer parts of Sweden (or could it be the key anyway?). Vaktmästarna for helping me with cages and other things to my dear orchids. People at the old “Genetikcentrum”. Tack Siv och Inger för att ni alltid har varit så fantastiskt trevliga och gulliga. Det kändes sorgligt att flytta därifrån. Thanks to everyone in the “trädgårdsvillan” for letting me share this building with you and especially Urban Gullberg for making things possible. I´m especially grateful to Micke L for your encouragement, our nice discussions and for also being interested in this fascinating species. You’re simply the best. Ronny A for giving me cheerful mails and phone calls when I was totally “into the black” (it helped a lot, as you can see I almost managed to finish. By the way, I´m still so sorry for the failure with the Norna material). Johanne M, for your generous advices and kindness, although I wish you would have stayed in Uppsala. Karin Bengtsson, for a lot of help with the Gotland populations; Ann Smithson for all advice and generosity; Mikael Hedrén, Anna-Lena Fritz, Bo-Göran Johansson and several botanists for help with the localities and other things as well; Svante Malmgren who helped in germinating orchid seeds; Minna M for a lot of laughs (I remember… no, takes to much space); Anna L for many nice trips, dog walks, shoppings and for teching me a lot about gardening; Mats I, you are a “missing person”; Måns Stefansson, my dear nephew, for help in the field. Unfortunately, I guess this trip was the reason why you changed your plans of being a “green biologist” to something very different; Ditte W for drawing the lovely cover illustration; and at last but not least, my dear beloved husband for… well, you know what I mean. I would also like to thank the Oscar and Lili Lamm foundation, the Nilsson-Ehle foundation, the Ebba and Sven Schwartz foundation and finally the Sven and Lilly Lawski foundation. Without your generosity this work would not have been possible.. 35.

No results found