Pollinator-Mediated Selection and the Evolution of Floral Traits Inorchids

Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1677

Pollinator-mediated selection and the evolution of floral traits in orchids

JUDITH TRUNSCHKE

ACTA UNIVERSITATIS UPSALIENSIS ISSN 1651-6214 ISBN 978-91-513-0352-9 UPPSALA urn:nbn:se:uu:diva-349139 2018 Dissertation presented at Uppsala University to be publicly examined in Zootissalen, Villavägen 9, Uppsala, Thursday, 14 June 2018 at 10:00 for the degree of Doctor of Philosophy. The examination will be conducted in English. Faculty examiner: Professor Allan G. Ellis (Stellenbosch University).

Abstract Trunschke, J. 2018. Pollinator-mediated selection and the evolution of floral traits in orchids. Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1677. 43 pp. Uppsala: Acta Universitatis Upsaliensis. ISBN 978-91-513-0352-9.

In this thesis, I combined manipulations of traits and pollination environment with analysis of phenotypic selection to examine causes of variation in strength and mode of selection on floral traits, and I conducted a reciprocal sowing experiment to test for local adaptation in germination success. I tested the following predictions (1) the opportunity for selection, and the strength of pollinator-mediated and net selection increase with increasing pollen limitation, (2) the effects of traits affecting pollinator attraction and traits affecting pollination efficiency are non-additive and this leads to pollinator-mediated correlational selection, (3) the effects of spur length on pollen removal, pollen receipt, and female fitness differ between populations with short-tongued and populations with long-tongued pollinators, and (4) local adaptation at the stage of germination contributes to the maintenance of ecotypes growing in grasslands and woodlands, respectively. A study including natural populations of 12 orchid species that varied widely in pollen limitation showed that opportunity for selection, pollinator-mediated selection and net selection were all positively related to pollen limitation, whereas non-pollinator-mediated selection was not. In the moth-pollinated orchid Platanthera bifolia, experimental reductions of plant height and spur length decreased pollen removal, pollen receipt and fruit production, but non-additive effects were not detected. Effects of plant height translated into pollinator-mediated selection for taller plants via female fitness, but there was no current pollinator-mediated selection on spur length. An experiment using artificial nectar spurs demonstrated that in P. bifolia pollen receipt saturated at shorter spur length in a population with short-tongued pollinators than in a population with a long-tongued pollinator. Effects of spur length on pollen receipt did not translate into current pollinator-mediated selection indicating that also plants with the shortest spurs for the most part received sufficient pollen for full seed set. Reciprocal sowing of seeds from grassland and woodland populations detected no evidence of local adaptation at the germination stage between ecotypes of P. bifolia. Taken together, the results illustrate how a combination of trait manipulation and analysis of strength and causes of selection can throw light on both the functional and adaptive significance of trait variation within and among natural populations.

Keywords: Among-population variation, Ecotype, Field experiment, Floral evolution, Germination, Local adaptation, Orchidaceae, Platanthera bifolia, Phenotypic selection, Pollination, Reciprocal transplant, Trait manipulation

Judith Trunschke, Department of Ecology and Genetics, Plant Ecology and Evolution, Norbyvägen 18 D, Uppsala University, SE-752 36 Uppsala, Sweden.

ISSN 1651-6214 ISBN 978-91-513-0352-9 urn:nbn:se:uu:diva-349139 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-349139)

“If you don´t climb the mountain, you can´t see the view”

(Harvey Mackay)

For my parents and my sister

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Trunschke, J., Sletvold, N. and Ågren, J. (2017) Interaction intensity and pollinator-mediated selection. New Phytologist, 214: 1381-1389

II Trunschke, J., Sletvold, N. and Ågren, J. The independent and combined effects of floral traits distinguishing two pollination ecotypes of a moth-pollinated orchid. (Manuscript)

III Trunschke, J., Sletvold, N. and Ågren, J. Pollinator-mediated selection and effects of spur length on pollination success in short-spurred and long-spurred populations of the orchid Platanthera bifolia. (Manuscript)

IV Trunschke, J., Sletvold, N. and Ågren, J. No evidence for adaptive population differentiation during seed germination between woodland and grassland populations of the orchid Platanthera bifolia. (Manuscript)

Reprints were made with permission from the respective publishers.

Contents

Introduction ...... 9 Biodiversity and the importance of ecology for evolution ...... 9 Quantifying phenotypic selection ...... 10 Opportunity for selection and strength of selection ...... 11 The importance of experimental manipulations for understanding selection ...... 12 Pollinator-mediated selection ...... 13 Traits important for pollinator attraction and pollination efficiency ... 14 Aims of the thesis ...... 16 Material and Methods ...... 17 Deceptive and rewarding orchids ...... 17 Platanthera bifolia and its ecotypes ...... 18 Pollen limitation and strength of pollinator-mediated selection (I) ..... 19 Manipulation of plant height and spur length, and correlational selection (II) ...... 20 Manipulation of spur length and pollinator-mediated selection (III) ..... 22 Reciprocal transplant (IV) ...... 23 Results and Discussion ...... 24 Pollen limitation and strength of pollinator-mediated selection (I) ..... 24 Effects of plant height and spur length on pollinator interactions (II) ... 26 Pollinator-mediated selection and the effects of spur length on pollination success (III) ...... 27 Local adaptation in germination success (IV) ...... 29 Conclusions ...... 31 Svensk sammanfattning ...... 32 Acknowledgements ...... 36 References ...... 38

Introduction

Biodiversity and the importance of ecology for evolution Eight point seven million is an impressively large number. It is the currently estimated number of species inhabiting our planet (Mora et al. 2011), and a numerical expression of the tremendous diversity of living forms and func- tions surrounding us. Understanding the processes generating and maintaining differences among populations and ultimately speciation are fundamental question addressed by evolutionary biologists. Many plant and animal species are widely distributed, occur in a variety of habitats and show substantial variation among populations. Local adaptation has been documented for both plants and animals across several spatial scales (Leimu and Fischer 2008, Hereford 2009). However, in most cases the selective agents and the traits contributing to local adaptation are poorly known (Wade and Kalisz 1990, MacColl 2011, Caruso et al. 2017). Selection and adaptive differentiation can act through several different components of fitness and at different stages in the life-history of an organ- ism (Mojica and Kelly 2010, Postma and Ågren 2016). Many studies have quantified selection and adaptation based on estimates of survival, mating success and fecundity, while few attempted measuring selection through lifetime fitness by including also the earliest stages of the life cycle (germination and seedling establishment; McGraw and Antonovics 1983, Schemske 1984, Campbell 1998, Postma and Ågren 2016). Studying the mechanisms shaping and maintaining among-population variation in natural populations provides essential insights to the processes of divergence on a micro- evolutionary scale and contributes to our understanding of macro- evolutionary patterns.

In my thesis, I combine studies of phenotypic selection with experimental manipulations of traits and selection environment in natural populations of animal-pollinated plants to examine the functional and adaptive significance of differences in trait expression and mechanisms responsible for current selection. In the paragraphs below, I present the aims and content of this thesis. I begin by identifying gaps of knowledge and discussing how field experiments can be used to advance our understanding of selection and evolutionary processes in contemporary populations. Finally, I describe how the

9 study of flowering plants can importantly contribute to the development of evolutionary concepts and theory.

Quantifying phenotypic selection Controversial at the time the theory was published, natural selection is now widely accepted as one of the major processes shaping phenotypic diversity (Darwin 1859, Dobzhansky 1973). Nevertheless, for long it was not thought to be possible to directly observe in nature, because the process is weak and slow (Hoekstra et al. 2001). Only about a hundred years after its postulation in Darwin´s “Origin of Species”, the appropriate statistical tools to quantify (Crow 1958, O´Donald 1970, Price 1970, 1972, Lande and Arnold 1983, Arnold and Wade 1984, Endlers 1986, Schluter 1988, Conner 1996, Henshaw and Zemel 2017) and visualize (Phillips and Arnold 1989, Brodie et al. 1995) phenotypic selection on quantitative traits in natural populations were developed. Because phenotypic selection is defined as the change in relative fitness with unit change in phenotype, it can be quantified by re- gressing relative fitness against a standardized quantitative character (Robertson 1966, Price 1970, 1972). Selection on a given character is a function of its direct effect on fitness and indirect effects caused by correlated characters. Multiple regression can be used to control for covariance structure among multiple phenotypic characters and identify their independent effect on fitness (Lande and Arnold 1983). Finally, selection can take many forms as reflected in the relationship between trait and fitness (linear, stabilizing, disruptive) and act on combinations of traits (correlational selection), and these forms of selection can be identified by including quadratic and interaction terms in the multiple regression model. Taken together, these tools advance our ability to examine the causes, targets and modes of natural selection. Pioneer studies (Kettlewell 1955, Grant and Grant 1989) demonstrated that selection in natural populations is empirically detectable and may be stronger than has been previously assumed encouraging researchers to pur- sue this path (Hoekstra et al. 2001, Kingsolver and Pfennig 2007, Kingsolver et al. 2001). Today, phenotypic selection has been quantified in a variety of animal and plant species and the number of studies is exponentially increasing. Finally, we begin to synthesize and understand the patterns and strength of phenotypic selection in natural populations (Kingsolver et al. 2001, 2012, Kingsolver and Pfennig 2007, Siepielski et al. 2013, Caruso et al. 2017). However, despite these advances in our understanding of natural selection many questions still remain: (1) What determines the strength of selection? (2) What are the targets and agents of selection? (3) To what extent does selection vary in time and space? (4) To what extent does selection through different components of fitness differ in strength and direction (e.g. between male and female function, and between life-history stages)?

10 In papers I, II, and III, I quantify phenotypic selection in natural populations and explore variation in agents, targets, and strength of selection through reproductive fitness. In paper I, I explore variation in selection among populations of different species, and in paper III among populations within a species. In paper II, I discuss correlational selection, and in papers II and III, how selection through pollen removal and female fitness may differ. Fi- nally, in paper IV, I examine whether selection at another life-history stage (germination) contributes to divergence between two ecotypes within species.

Opportunity for selection and strength of selection Natural selection is the non-random sorting of individuals within a population, based on the association between variation in phenotype with variation in fitness (Darwin 1859). Natural selection requires sufficient variation in both phenotype and fitness to be detectable. The opportunity for selection is defined as the variance in relative fitness (Crow 1958). For biotic interactions such as herbivory and pollination, Benkman (2013) showed that the intensity of both antagonistic and mutualistic interactions are important determinants of the strength of selection, because of their effects on the variance in fitness. The strength of selection also depends on the magnitude of variation in trait expression in a population relative to the optimal phenotype. Selection can be expected to be stronger when mean trait expression is far from the optimum compared to when it is close to the optimal value. For example, Pauw et al. (2009), Nattero et al. (2010) and Paudel et al. (2016) measured the patterns of selection in three mutualistic interactions between flowering plants and their pollinators and found that selection was particularly strong in populations where the mismatch in functional morphology of the flowers (important for reproduction through pollen transfer) and of pollinators (important for nectar reward gain as energy source) was large.

In paper I, I examine the effect of pollen limitation, which is a measure of the intensity of plant-pollinator interactions, on opportunity for selection and strength of pollinator-mediated and net selection across 12 flowering plant species. In papers II and III, I discuss how the strength of current selection is affected by the magnitude of variation in traits and fitness.

11 The importance of experimental manipulations for understanding selection A full understanding of the causes and targets of phenotypic selection requires experimental approaches (Wade and Kalisz 1990, McColl 2011, Caruso et al. 2017). This includes the identification of the agents of selection and of the traits subject to selection, and why and how these particular traits affect fitness. Because of the complexity of environmental factors interacting in natural populations, the direct causes of selection may not be clearly evi- dent and easily distinguished from indirect effects. Further, consistent directional or stabilizing selection may reduce phenotypic variation within populations, which complicates, or makes it impossible, an evaluation of the fitness effects of trait expression without trait manipulation. Manipulation of environmental factors can determine their importance as agents of selection in a given habitat (Wade and Kalisz 1990). For example, herbivore exclusion can be used to reveal selection on plant traits due to interactions with herbivores (Gomez 2003, Lehndal and Ågren 2015), and supplemental hand-pollination to quantify pollinator-mediated selection (Sandring and Ågren 2009, Sletvold and Ågren 2010), because they remove variation in fitness caused by these interactions and quantify non-herbivore- mediated and non-pollinator-mediated selection, respectively. Trait manipulation can be used to reveal the functional and adaptive significance of phenotypic characters (Campbell 2009, Johnson 2012). Trait manipulations have been used in plant (e.g. Peakall and Handel 1993, Campbell 2009, Dudash et al. 2011, Fenster et al. 2015) and animal systems (e.g. Kingsolver 1999, Theis et al. 2012) to quantify the effects of focal traits on fitness. Moreover, trait manipulation can identify characters important for reproductive isolation (e.g. Xu et al. 2011, Campbell et al. 2016) and in- volved in population or species divergence (e.g. Boberg and Ågren 2009, Bischoff et al. 2015). Furthermore, combined manipulation of more than one trait can reveal possible non-additive effects on fitness and shed light on the evolution of phenotypic integration (Benitez-Vieyra et al. 2006, Campbell et al. 2014, Bischoff et al. 2015, Fenster et al. 2015). When trait variation is limited within populations, it can also be increased through crosses between diverged populations or species (Andersson and Widén 1993, Schemske and Bradshaw 1999, Toräng et al. 2017). The reciprocal transplant experiment is the lithmus test of local adaptation (Kawecki and Ebert 2004, Blanquart et al. 2013), and can be used to examine selection at several stages of the life cycle (McGraw and Antonovics 1983, Postma and Ågren 2016). However, by itself a reciprocal transplant does not reveal which traits contribute to selection or which environmental factors are responsible for differences in selection among sites. This requires analysis of targets and agents of selection as described above.

12 In papers I, II and III, I manipulate the pollination environment to disentan- gle pollinator-mediated and non-pollinator-mediated selection. In papers II and III, I manipulate trait expression to study the independent and combined effects of correlated characters on components of reproductive fitness (II), and to characterize the relationship between flower morphology and pollination success (III). In paper IV, I use a reciprocal transplant experiment to test for evidence of local adaptation at the germination stage between diver- gent populations.

Pollinator-mediated selection Flowering plants are one of the most diverse groups on earth (currently 350 000 accepted species names; The plant list, The state of the World´s Plants Report 2017), colonizing a diversity of habitats (Kier et al. 2005, Körner 2011). As sessile organisms, plants are particularly interesting because of their need to evolve mechanisms to adapt to spatial and temporal heterogeneity in environmental conditions including a vector mediating sexual reproduction. Biotic interactions can result in strong natural selection and ecological speciation (Dobzhansky 1950, Thompson 1994), and animal-pollinated plants provide some classic examples of diversification in response to selection mediated by pollinators (Kay and Sargent 2009, Schiestl and Johnson 2013, Van der Niet et al. 2014). The pollinator-shift model of Grant and Grant (1965) and Stebbins (1970) provides a fundamental framework for our current understanding of adaptation and speciation driven by pollinator interactions in angiosperms. It is built on the observations that, (a) pollinator species differ in their preferences and in pollination efficiency, and (b) the relative importance of pollinator species vary across the geographic distribu- tions of plants. As a consequence, many plant species are likely to experi- ence spatial variation in pollinator-mediated selection favoring different floral phenotypes in different parts of their range (e.g. Chapurlat et al. 2015, Paudel et al. 2016). The importance of variation in interactions with pollinators for floral divergence and establishment of reproductive isolation has provided strong support for the role of geographic heterogeneity in selection for speciation in general (Stebbins 1950, Grant 1981). Thus, the study of population divergence and adaptation in flowering plants can advance our understanding of mechanisms driving the evolution and maintenance of biodiversity. The family Orchidaceae presents an extraordinary floral diversity, which in many cases is attributable to their interactions with pollinators. Orchids therefore represent an attractive system to study the importance and underly- ing mechanisms of pollinator-driven floral evolution (Darwin 1862, Fay and Chase 2009, Micheneau et al. 2010). One of the key innovations suggested to have fostered diversification and speciation in orchids is the evolution of deceptive flowers (about one-third of the species do not offer a reward to

13 their pollinators; Cozzolino and Widmer 2005, Tremblay et al. 2005). Floral deception can have various forms including food-, sexual- and brood-site, but they are all associated with typically low pollination success. Deceptive orchids have on average lower fruit production and are more strongly pollen limited compared to rewarding orchids (e.g. Tremblay et al. 2005).

Throughout my thesis, I use orchids as model systems to examine selection on floral traits, population divergence and adaptation. In paper I, I include both deceptive and rewarding species to explore how differences in pollen limitation affect variation in fitness and strength of selection. In the remaining papers II, III and IV, I study among-population variation in selection in the rewarding moth-pollinated orchid Platanthera bifolia, which shows con- siderable geographic co-variation in morphology and pollinator assemblages.

Traits important for pollinator attraction and pollination efficiency In search of reward, pollinators respond to several floral characters including traits that are important for pollinator attraction (visual and olfactory signals) and those affecting the efficiency of pollination. Pollinator-mediated selection has been demonstrated in a variety of species and on various traits (Sandring and Ågren 2009, Sletvold and Ågren 2010, Sletvold et al. 2010). Traits affecting the rate of pollinator visitation and traits affecting the efficiency of pollen transfer are expected to be subject to correlational selection because of their non-additive effects on fitness (Sletvold and Ågren 2011). Yet, the combined effect of functionally different traits is little investigated (but see Fenster et al. 2006, Boberg and Ågren 2009, Sletvold and Ågren 2011) and their independent and combined effects for reproductive success poorly understood. In many plant species, nectar is produced at the bottom of the flower, and the depth of the floral tube or nectar spur should critically influence the morphological fit between flower and pollinator, and thereby the efficiency of pollen transfer (Darwin 1862, Nilsson 1988, Muchala and Thomson 2012, but see Vlašánková et al. 2017). Experimental studies have demonstrated that if floral depth is too short compared to pollinator proboscis length, floral reproductive organs do not contact with the important pollinator body parts and flowers are not effectively pollinated (Nilsson 1988, Johnson and Steiner 1997, Boberg and Ågren 2009, Ellis and Johnson 2010). Pollination success is expected to increase asymptotically with increasing nectar spur length, because the contact between pollinator and flower reproductive parts should intensify and pollination efficiency increase only until the spur is longer than the mouth parts of the pollinator (Darwin 1862, Nilsson 1988). The preva-

14 lence of among-population co-variation in floral tube depth or nectar spur length with pollinator proboscis length (Anderson et al. 2009, Nattero et al. 2011, Newman et al. 2014, Boberg et al. 2014) may thus be a consequence of variation in pollinator morphology leading to variation in the shape of the relationship between nectar spur length and pollination success and thus in pollinator-mediated selection.

In paper II, I study the combined and independent effect of a trait influencing pollinator attraction (plant height) and a trait influencing pollination efficiency (spur length). In paper III, I characterize the shape of the relationship between nectar spur length and pollination success, and examine whether it varies among populations with different pollinator communities.

15 Aims of the thesis In this thesis, I have used orchids as model systems to investigate pollinator- mediated selection including the identification of targets, strength and direction of selection, and the importance of among-population variation in pollinator assemblages for population divergence in plant and floral characters. Further, I have examined whether two ecotypes within species show evidence of local adaptation in traits affecting germination success. More specifically, I have addressed the following questions:

1. What are the targets, and the strength and mode of pollinator-mediated selection in natural orchid populations (I, II, III)?

2. How does the strength of selection relate to the magnitude of trait variation and opportunity for selection (I, II, III)?

3. Do floral traits important for pollinator attraction (plant height) and pollination efficiency (spur length) independently or jointly affect pollination success and are they subject to pollinator-mediated correlational selection (II)?

4. Does the functional relationship between spur length and pollination success differ between populations experiencing different pollination environments, and if so, does this translate into differences in current pollinator-mediated selection (III)?

5. Do differences in traits affecting germination success contribute to local adaptation among populations occupying grassland and woodlands, respectively (IV)?

16 Material and Methods

Deceptive and rewarding orchids In the study reported in chapter I, I included six deceptive and six rewarding orchid species to obtain a wide range of pollen limitation. The study included one population each of the deceptive species Anacamptis morio (L.) R.M.Bateman, Pridgeon & M.W.Chase, Dactylorhiza incarnata (L.) Soó, D. sambucina (L.) Soó, Orchis mascula (L.) L., O. militaris (L.), and Neotinea ustulata (L.) R.M.Bateman, Pridgeon & M.W.Chase (Fig. 1a-f), and of the rewarding species Dactylorhiza viridis (L.) R.M.Bateman, Pridgeon & M.W.Chase, Epipactis palustris (L.) Crantz, Gymnadenia conopsea (L.) R.Br., Neottia ovata (L.) Bluff & Fingerh., Platanthera bifolia (L.) Rich., and P. chlorantha (Custer) Rchb. (Fig. 1g-l). All 12 species are terrestrial, perennial herbs with insect pollinators (Delforge 2005) including bumble bee queens, bumble bee workers, solitary bees, tachinid flies, ichneumonid wasps and sawflies, eumenid wasps, beetles, butterflies, and noctuid moth and hawkmoth (Nilsson 1978a,b, 1980, 1981a,b, 1983a,b, 1984, Boberg et al. 2014, Chapurlat et al. 2016; see Supplemental material of paper I for detailed information). Study populations were located on the island Öland in southeastern Sweden.

Figure 1. Flowers of the 12 orchid species included in paper I. (a) Anacamptis morio, (b) Dactylorhiza sambucina, (c) D. incarnata, (d) Orchis mascula, (e) Neotinea ustulata, (f) O. militaris, (g) D. viridis, (h) Epipactis palustris, (i) Gymnadenia conopsea, (j) Neottia ovata, (k) Platanthera bifolia, and (l) P. chlorantha. (a-f) are deceptive species, whereas (g-l) are rewarding.

17 Platanthera bifolia and its ecotypes In chapters II-IV, I used the nocturnal-pollinated, long-lived, terrestrial orchid Platanthera bifolia (L.) Rich. as a model system. The species is wide- spread in the temperate regions across Eurasia and grows in a variety of habitats ranging from grasslands, moorlands and marshes to open woodlands, and can be found at elevations up to 2500 m asl (Delforge 2005). In Scandi- navia, P. bifolia typically flowers between early- to late-June. Individual plants produce a single inflorescence containing 12-25 flowers. Flowers are white, and at night-time emit a strong scent dominated by benzenoids and linalool (Tollsten and Bergström 1989, Tollsten and Bergström 1993). The long spur produces sugar-rich nectar throughout the lifetime of the flower (Stpiczynska 1997). The species attracts a variety of sphingid and noctuid moths (Nilsson 1983, Claessens et al. 2008, Boberg et al. 2014, Paper III). Flowering time and morphological traits such as plant height, flower production, the size of individual flowers and the length of the nectar spur vary considerably among populations across Scandinavia (Boberg et al. 2014) and Europe (Bateman et al. 2012). On the Baltic island Öland in southeastern Sweden, P. bifolia occurs as a grassland and a woodland ecotype (Boberg 2010, Boberg et al. 2014; Fig 2a). The grassland ecotype begins to flower about two weeks later, and produces markedly shorter inflorescences with smaller flowers compared to the woodland ecotype (Boberg et al. 2014; Fig. 2a and b, Paper III). Most no- ticeable, is the distinct difference in spur length: Nectar spurs in grassland populations are on average 21 mm long (mean based on 7 population means), whereas spurs in woodland populations are on average 36 mm long (mean based on 5 population means, Boberg et al. 2014; Fig. 2b, Paper III). Variation in spur length is associated with variation in the proboscis length of the dominating pollinator in each habitat: various short-proboscid moths and hawkmoths are common in grassland populations (Deilephila porcellus, Hyles gallii, Cucullia umbratica; Nilsson 1983, Boberg et al. 2014, Paper III), while woodland populations are predominantly pollinated by the long- proboscid hawkmoth Sphinx ligustri (Boberg et al. 2014). For my thesis I have worked in two large long-spurred woodland populations (Gråborg: N 56.669068, E 16.598443; and Vedby: N 56.769556, E 16.645412) and two large short-spurred grassland populations (Melösa: N 56.859528, E 16.856478; and Strandtorp: N 56.659951, E 16.712848) in the central part of Öland (Fig. 2c).

Figure 2. (a) Illustration of the two ecotypes of the orchid Platanthera bifolia, that can be found on the island Öland in southeastern Sweden: the long-spurred woodland ecotype and the short-spurred grassland ecotype. (b) Among-population variation in four flowering characteristics (plant height, flower number, flower size and spur length) for the two woodland populations (blue colours; N = 2) and the two grassland populations (orange colours; N = 2) used in the papers II, III and IV. (c) Locations of the four study populations.

Pollen limitation and strength of pollinator-mediated selection (I) To test the prediction that the opportunity for selection, pollinator-mediated selection, and net selection increase with increasing pollen limitation, I used data collected in populations of 12 orchid species in 2013. Six of the species were deceptive, which is expected to be associated with strong pollen limitation, and six were rewarding and expected to be characterized by less strong pollen limitation. I quantified pollen limitation (PL) as 1 – FitnessC/FitnessHP, where FitnessC is the female fitness of open-pollinated plants and FitnessHP is the female fitness of plants receiving supplemental hand-pollination. To test whether the opportunity for selection increased with increasing pollen limitation, and whether this represents a causal relationship, I used a linear regression model with pollen limitation in the control treatment, pollination treatment (open-pollinated control vs. supplemental hand-pollinated), and their interaction as independent variables (ANCOVA). The pollen limitation × pollination treatment interaction was significant, and I analysed regression models separately by treatment to determine whether the strength of the relation-

19 ship between pollen limitation in the control treatment and opportunity for selection was reduced after supplemental hand-pollination. I quantified phenotypic selection through female fitness on five flowering characters (start of flowering, plant height, number of flowers, flower size and spur length). I used multiple regression analysis following the procedure of Lande and Arnold (1983) to estimate linear, quadratic and correlational selection gradients separately for open-pollinated plants (βOpen-pollinated = βNet selection; N ~ 120) and plants receiving supplemental hand-pollination (βHand- pollinated = βNon-pollinator-mediated; N ~ 120). To quantify selection mediated by the interaction with pollinators, I subtracted selection gradients estimated for plants receiving supplemental hand-pollinations from selection gradients estimated for open pollinated plants (βPoll = βOpen-pollinated - βHand-pollinated). For all estimated selection gradients, I used a non-parametric bootstrapping procedure to assess statistical significance. I used ANCOVA with relative female fitness as response variable to examine whether selection gradients varied among species, traits and pollination treatments. To test whether the strength of pollinator-mediated selection increased with increasing degree of pollen limitation, I used ANCOVA with the degree of pollen limitation, trait and their interaction as predictor variables and the absolute value of pollinator-mediated selection as response (|βPoll|) variable. I used ANCOVA to examine whether the degree of pollen limitation influences the strength of net selection among species and whether this is a causal relationship using pollen limitation, trait, pollination treatment and all interactions as predictor variables and the absolute values of net (|βOpen-pollinated|) and non-pollinator- mediated selection (|βHand-pollinated|) as response variables.

Manipulation of plant height and spur length, and correlational selection (II) To test the prediction that two traits likely to affect pollinator attraction (plant height) and pollination efficiency (spur length) are subject to pollinator-mediated correlational selection because of non-additive effects on pollination success and female fitness, I combined a factorial trait manipulation experiment (conducted in 2015) with an analysis of phenotypic selection (conducted in 2016) in a large woodland population of Platanthera bifolia. First, to examine whether pollination success and fruit production increase with both plant height and nectar spur length in the woodland habitat where the surrounding vegetation is tall and pollinators have a long proboscis, and whether their effects are non-additive, I manipulated both traits in a factorial design. I selected about 200 bolting plants, bagged them individual- ly to avoid pollinator visitation prior to the experiment and assigned them to blocks of four plants (one for each of four trait manipulation treatments; N = 23) based on their number of available open flowers. To manipulate plant

20 height, I cut inflorescences and mounted them at either 27cm (short) or 38 cm (tall) on a bamboo stick. To manipulate spur length to either 21 mm (short spur) or 35 mm (long spur) I gently folded the spur and fixed it with elastic tape. Experimental treatments corresponded to mean trait values recorded in previous studies of the grassland (short plants with short spurs) and woodland ecotype (tall plants with long spurs), respectively. All nectar spurs on experimental plants were manipulated. I presented plants to pollinators for three nights and estimated pollination success as the total number of pollinia removed and the total number of massulae received. After three nights, plants were bagged to prevent further pollination, and at fruit maturation, I scored the number of fruits produced by each inflorescence. To test whether effects of the two traits on pollination success and fruit production translate into correlational selection mediated by pollinators, I estimated phenotypic selection for open-pollinated plants (N = 119) and plants receiving supplemental hand-pollination (N = 58). At peak flowering, I measured plant height, the number of flowers produced, and the size and spur length of one flower on each plant. All open flowers on plants belong- ing to the hand-pollination treatment received supplemental hand- pollinations on 2-3 occasions. At the end of flowering, I scored for all individuals the number of pollinia removed. At fruit maturation, I scored the number of flowers producing a fruit and harvested up to three fruits per plant to estimate mean fruit mass. I used the product of number of fruits and mean fruit mass to quantify total female fitness. I calculated the degree of pollen limitation in the population (PL) as described above. I estimated the opportunity for selection as the variance in relative pollen removal and in relative female fitness, respectively. In the trait manipulation study, I used a mixed-effect model to examine the effect of plant height and spur length (fixed factors) on pollen removal, the number of flowers receiving pollen, and the average pollen receipt per pollinated flower, and the number of flowers forming a fruit (block was included as random factor). I used posthoc Tukey analysis to test for differences between treatments. In the selection study, I used F-test for dependent samples (Lee 1992, Pitman 1939) to examine whether the opportunity for selection via pollen removal and via female fitness differed. Following Lande and Arnold (1983), I estimated directional, quadratic and correlational selection gradients from multiple regression models in which relative fitness was regressed on standardized trait values. I used added-variable plots to visualize significant directional and quadratic selection, and contour plots to visualize correlational selection (Phillips and Arnold 1989, Wood et al. 2017).

21 Manipulation of spur length and pollinator-mediated selection (III) To characterize the pollination environment in grassland and woodland populations of P. bifolia, I used previously published data of pollinator species and their proboscis lengths, and I caught additional pollinators and deter- mined their proboscis length in the two grassland populations used in this study. To investigate whether the effect of spur length on pollination success dif- fers between populations visited by pollinator species of different morphology, I conducted a spur manipulation experiment in one grassland and one woodland population in summer 2017. In each population, I marked about 100 plants at the bolting stage and bagged them shortly before their flowers opened to avoid pollination prior to spur manipulation. I constructed artificial spurs using elastic transparent tygon tubing and cut it to pre-defined length (13 categories ranging from 4 to 52 mm). I manipulated spur length on three flowers on each experimental plant by replacing natural spurs with artificial spurs (on a given plant all artificial spurs were of the same length). I repeated my experiment on three different days across the flowering period of the population (total N = 6 plants of each spur length in the woodland population and N = 5 plants of each spur length in the grassland population). I quantified pollination success as the total number of pollinia removed and the total number of massulae received in experimental flowers after three nights of exposure to pollinators. To test whether differences in the effects of spur length on pollination success translate into differences in pollinator-mediated selection on spur length, I estimated phenotypic selection in two grassland and two woodland populations in 2016. Data used for selection analysis in one of the woodland populations were the same as those reported in the study reported in paper II, and I followed the same procedure for quantifying phenotypes and fitness, and for hand-pollinations as described in the previous section. To characterize the functional relationship between spur length and pollination success, I first calculated for each individual the mean number of pollinia removed and the mean number of massulae received in the three flowers with artificial spurs. Because, I expect a saturating function of pollination success with increasing spur length, I used a non-linear least-square regression (nlstools R package, Baty et al. 2015) to examine the effects of spur length on pollen removal and pollen receipt, respectively. I fitted a neg- ative exponential function of the form y = a(1 – exp(-bx), where y is mean number of pollinia removed per flower, or mean number of massulae received per flower, respectively, and x is the length of the artificial spur of a given plant. In this model, a is an estimate of the asymptotic value of the dependent variable, and b reflects the rate at which the asymptote is reached. I defined the “saturation point” as the spur length for which the predicted

22 number of pollinia removed or number of massulae received was 95 % of the predicted maximum. For each of the four study populations, I quantified mean pollen removal, pollen limitation of female fitness, opportunity for selection, pollinator- mediated selection, non-pollinator-mediated selection, and net selection as described above for paper II.

Reciprocal transplant (IV) To test whether among-population variation in germination success contributes to local adaptation between grassland and woodland populations of P. bifolia, I conducted a fully crossed seed-burial experiment including two grassland and two woodland populations. In each population, I collected one fruit from each of 30 plants that had received supplemental hand-pollination, but also visits by natural pollinators. I pooled seeds by population. I then prepared 120 seed packages (~ 100 seeds per package) for each population following the method of Rasmussen and Whigham (1993). I established at each site three random transects, each containing 10 plots at a distance of 3 m. In each plot, I placed one seed package per population (in total, N = 30 seed packages per site per population of origin). About 10 months later, I retrieved all seed packages, cleaned them, and counted the number of seeds that had germinated. For each seed package, I calculated germination success as the proportion of seeds that had germinated. Because differences in germination success can be affected by differences in seed viability, I used a seed staining test following Vujanovic et al. (2000) to estimate proportion of seeds that were viable. To do that, I prepared for each population seed packages containing ~ 60 seeds. After staining with acid- fuchsin, I counted the number of red colour-stained (unviable) and uncol- oured (viable) seeds. I quantified seed viability as the proportion of seeds that were viable. Because soil pH has been shown to influence germination success in other orchids (Batty et al. 2001, Diez 2007, Jacquemyn et al. 2015), I measured pH from soil samples collected for each plot at each site at the time when seeds where buried. I used a mixed-effect model to test for evidence of local adaptation. The model included germination rate as response variable and population of origin, transplant site and their interaction as explanatory variables. Plot within site was included as a random factor. For each site separately, I examined the effect of soil pH on proportion of seeds germinating (plot means) with linear regression.

23 Results and Discussion

Pollen limitation and strength of pollinator-mediated selection (I) Pollen limitation of female fitness varied significantly among the twelve studied orchid species (ranging from zero to 0.92) and was bimodally distributed. Seven species (six rewarding and one deceptive) had PL < 0.4, and the five remaining deceptive species had PL > 0.8. The opportunity for selection (variance in relative female fitness) was positively related to pollen limitation in the population in the open-pollinated control, but not in the hand-pollination treatment (ANCOVA; significant pollen limitation × pollination treatment interaction, F = 15.60, P < 0.001), demonstrating that this was a causal relationship. The targets and direction of selection varied among species. Across the twelve orchid populations, I found significant pollinator-mediated selection on flowering time, plant height and flower number, whereas selection on spur length approached significance (ANCOVA; significant trait × pollination treatment interaction, or trait × species × pollination treatment interactions in ANCOVA). Pollinator-mediated selection translated into net selection to varying degree. The strengths of pollinator-mediated selection and net selection increased with increasing pollen limitation (Fig. 3). By contrast, no significant relationship was recorded between pollen limitation in the control treatment and strength of non-pollinator-mediated selection

In conclusion, this chapter experimentally shows that the degree of pollen limitation influences the strength of selection acting on floral characters, and more generally that the intensity of biotic interactions can strongly determine selection regimes.

Figure 3. Relationship between pollen limitation in a population (PL; quantified in the control treatment) and strength of (a) net selection (selection among open- pollinated control plants, |βC|), (b) pollinator-mediated selection |ΔβPoll|), and (c) non-pollinator-mediated selection(selection in the hand-pollination treatment,

|βHP|), on flowering start, plant height, number of flowers, flower size and spur length in populations of 12 orchid species on Öland, SE Sweden. Statistical significance: *, <0.05; ***, <0.001; ns, not significant.

25 Effects of plant height and spur length on pollinator interactions (II) Experimental reductions of plant height (expected to influence pollinator attraction) and spur length (expected to influence pollination efficiency) showed that both traits are important for pollen removal and total female fitness (Fig. 4). Contrary to my hypothesis, I did not detect any non-additive effects of the two characters (non-significant interaction terms in two-way ANOVA, all P > 0.05). In the study of phenotypic selection, both pollen removal and fruit set were high (82 ± 17.9 % of pollinia were removed, and 75 ± 20.6 % of flowers produced a fruit), and the opportunity for selection through pollen removal was lower than through female fitness (0.14 vs. 0.62). There was pollinator-mediated selection through female fitness for taller plants (selection gradient ± SE, βPoll = 0.213 ± 0.094, P = 0.004), whereas no selection was detected on spur length via pollen removal or female fitness. The results demonstrate that both plant height and spur length affect pollination success and female fitness in the woodland habitat, where the surrounding vegetation is tall and the dominant pollinator has a long proboscis. However, only plant height was sufficiently variable for current pollinator- mediated selection to be detected. A history of consistent directional or stabilizing selection on plant height and spur length may have reduced variation in the study population. Moreover, pollen limitation and opportunity for selection were relatively low reducing expected strength of selection.

The results demonstrate that a combination of experimental trait manipulation and analysis of phenotypic selection can be required to understand fully the adaptive significance of trait variation.

Figure 4. Effects of plant height and spur length on (a) total number of pollinia removed, (b) the number of flowers receiving pollen, (c) number of massulae received by pollinated flowers, and (d) number of fruits produced after exposure to pollinators for three consecutive nights in a woodland population of P. bifolia on the island Öland, SE Sweden, in 2015. Least-square means ± SE extracted from a model of untransformed data are given. Letters above bars indicate significant differences between treatments in post-hoc Tukey tests.

Pollinator-mediated selection and the effects of spur length on pollination success (III) Spur length of P. bifolia and the proboscis lengths of dominating pollinators differed between the grassland and woodland populations. In the two grassland populations, mean (± SE) spur length was 20.5 ± 2.34 mm (Melösa) and 20.8 ± 3.96 mm (Strandtorp), and the proboscis length of abundant pollinators ranged from 18.0 ± 0.7 mm to 25.1 ± 0.5 mm (Fig. 2). In the woodland populations, spurs were much longer (33.9 ± 3.79 mm in Gråborg and 35.2 ± 3.96 mm in Vedby), and the proboscis length of the dominant pollinator was 39.1 ± 2.2 mm (Fig. 2). In plants that offer nectar at the bottom of the flower, floral depth is expected to influence pollination success as long as it is shorter or equal to the length of the mouthparts of the pollinator. As expected, I found pollen receipt to saturate at a shorter spur length in the grassland population (34 mm) compared to the woodland population (> 52 mm). However, there was no difference between the two populations in the spur length at which pollen removal saturated (23 and 22 mm in the grassland and woodland population,

27 respectively; Fig. 5) indicating that spur length is less critical for pollen removal than for pollen receipt. I found no evidence for directional or stabilizing pollinator-mediated selection on spur length through pollen removal or female fitness. Several factors may explain the lack of current selection on spur length. First, in all four populations, rates of pollen removal were high and pollen limitation of female fitness was relatively low. However, the opportunity for selection was still sufficient for the detection of selection for larger flowers through pollen removal in the two woodland populations, and of pollinator-mediated selection through female fitness for more flowers in two populations and for taller plants in one population. Second, in the natural populations, variation in spur length was limited to ranges where effects on pollen success were weak ac- cording to the results of the experimental spur length manipulations. Interestingly, I detected strong non-pollinator-mediated selection on spur length in the two grassland populations indicating that long spurs are not associated with a trade-off. However, the causes of such selection await further exploration, as well as the mechanisms behind the detected difference in selection through pollen removal and female fitness on traits other than spur length.

To summarize, in this chapter I used artificial nectar spurs to determine the effect of spur length on pollination success and fruit production in a grassland and a woodland population that are pollinated by different pollinators. I use the results to interpret current pollinator-mediated and net phenotypic selection on spur length.

Figure 5. The effects of spur length on mean number of pollinia removed and mean number of massulae received in flowers with artificial spurs in a short-spurred grassland (left panel; orange symbols and line; N = 55) and a long-spurred woodland (right panel; blue symbols and line; N = 78) population of the orchid P. bifolia in 2017 modeled with non-linear regression. Each grey symbol represents the mean of three experimental flowers of a given plant, and coloured symbols are means per spur length (N = 5 plants per spur length in the grassland population, and 6 plants per spur length in the woodland population). The distribution of spur length in the two natural populations in the year phenotypic selection was estimated (2016) is presented in the bottom panel. The saturation points (see Methods) of the functional relationships are indicated with a star in panels a-d, and with arrows in panels e-f. No saturation point was detected for pollen receipt in the woodland population.

Local adaptation in germination success (IV) In the reciprocal seed burial experiment, I found that the relative germination success of the four population differed among sites (significant transplant site × population of origin interaction in mixed-effect model; P = 0.035), but there was no evidence of local adaptation (Fig. 6). Instead, germination success was consistently highest for one of the grassland populations (Melösa), and, for all populations of origin, germination success was highest at one of the woodland sites (Gråborg). The results of the seed viability test (Melösa < Vedby < Strandtorp < Gråborg) suggest that variation in germination success

29 was not a simple function of differences in seed viability. The germination of orchid seeds is affected by a complex interaction of appropriate physical properties of the soil and the availability of a suitable mycorrhiza symbiont. I discuss several factors that potentially could contribute to the differences in germination success between populations and sites observed in this study, and I suggest that more data including the description of environmental abiotic conditions and characterization of mycorrhiza associations would be needed for a full understanding of the factors causing differences in germination.

Although the stage of germination is crucial in the life history of plants and differences in seedling establishment have been shown to strongly contribute to population divergence and adaptation, I found no evidence for local adaptation in germination success in P. bifolia.

Figure 6. The effect of transplant site and population of origin on germination success of seeds originating from two grassland (grey symbols; Melösa and Strandtorp) and two woodland (open symbols; Gråborg and Vedby) populations on the island Öland, south-eastern Sweden. For each site the local population is represented by a dark symbol. Least-square means and SE are indicated.

30 Conclusions Understanding mechanisms behind population divergence and adaptation is fundamental to explain speciation and the origin and maintenance of biodiversity. In this thesis, I have used plant-pollinator interactions in orchids to examine the importance of pollinator-mediated selection for the evolution and divergence of floral characters. More specifically, I have shown that pollinators can exert selection on several traits expected to influence pollinator attraction and pollination efficiency, and that the strength of selection in a given population is affected by the level of pollen limitation (I), the variance in relative fitness (I, II, III), and current phenotypic variation (II, III). I have shown with trait manipulations that both plant height and spur length affect pollination success and female fitness, but only plant height was sufficiently variable for current pollinator-mediated selection to be detected. In an experiment with artificial nectar spurs, I showed that the effects of spur length on pollination success differed between a grassland and a woodland population. The results are consistent with the hypothesis that differences in pollinator community have contributed to divergence between the short-spurred grassland and the long-spurred woodland ecotype of P. bifolia (III). However, a full understanding of population divergence and adaptation requires the integration of fitness responses to heterogeneity in selective environments throughout the life cycle. In my reciprocal seed burial experiment, I found no evidence of local adaptation between grassland and woodland populations during the germination phase (IV). Taken together, my work demonstrates the value of combining experimental manipulations of traits and environment to explore the mechanisms responsible for variation in selection, population divergence and adaptation.

31 Svensk sammanfattning

"De flesta av de grundläggande idéerna om vetenskap är väsentligen enkla och bör som regel uttryckas på ett språk som alla förstår."

Albert Einstein (tysk teoretisk fysiker)

Ta dig ut i naturen och du är omgiven av en mängd olika livsformer och funktioner. Det finns mer än 8,5 miljoner arter som delar planeten med oss. Och varje år upptäcks och beskrivs fler än 2000 nya arter. Det är lätt att för- undras och tänka: Varför finns det så många olika arter? Och hur har de upp- stått? Det här är frågor som evolutionsbiologier söker besvara. I min dok- torsavhandling har jag undersökt processer som bidrar till uppkomst och bevarande av biologisk mångfald.

Arter skiljer sig åt i olika egenskaper, men det gör många gånger också populationer av samma art när de förekommer i olika livsmiljöer. Det är den observation som var grunden för Darwins grundläggande idé om artbildning: genom naturligt selektion och anpassning till olika levnadsförhållanden. Om effekten av olika egenskaper på överlevnad och reproduktionsframgång skiljer mellan olika miljöer, och om variation i egenskaper har en genetisk grund, kan populationer utveckla skillnader som gör dem särskilt väl anpas- sade till den lokala miljön.

I en given population beror det naturliga urvalets styrka och form på flera faktorer. En övre gräns för selektionens styrka sätts av hur mycket överlev- nad och reproduktionsframgång varierar. I en population där överlevnad och reproduktionsframgång uppvisar stor variation finns förutsättningar för stark selektion, medan det omvända gäller i en population med liten variation i överlevnad och reproduktion. Den naturliga selektionens styrka beror också på hur omfattande variation som finns i egenskaper som påverkar överlevnad och reproduktionsframgång. Är variationen stor och omfattar ett intervall där effekter på överlevnad och reproduktion är kraftiga, kan vi förvänta oss stark selektion. Omvänt om variationen är begränsad eller endast omfattar ett intervall där effekter är svaga, kan vi förvänta oss svag selektion.

För att komplicera bilden ytterligare kan effekter på olika komponenter av reproduktiv framgång variera. De flesta växter är hermafroditer och fortplan-

32 tar sig både genom att bilda frö (honlig funktion) och genom att befrukta andra växter i populationen med sitt pollen (hanlig funktion). De egenskaper som gör en växt till en effektiv fröproducent är inte nödvändigtvis desamma som de som gör en växt effektiv vad gäller spridning av pollen.

Flera miljöfaktorer kan påverka om en egenskap är förknippad med hög eller låg reproduktionsframgång. Hos en växt kan en given egenskap vara gynn- sam för att den ökar förmågan att attrahera pollinatörer, men vara skadlig för att den ökar risken för skador från växtätande djur. Slutresultatet beror i en sådan situation på den relativa styrkan hos positiva och negativa effekter. För att ta reda på varför ett samband mellan egenskap och reproduktions- framgång ser ut som det gör krävs experiment där miljöförhållanden varieras och effekter på sambandets form utvärderas.

I min avhandling har jag undersökt faktorer som påverkar det naturliga urvalets styrka och form i naturliga populationer av orkidéer. Jag har fokuserat på egenskaper som påverkar pollinationsframgång, och på naturligt urval som beror på växters samspel med pollinatörer. Men jag har också studerat skillnader i groningsframgång mellan populationer i gräsmarker och skog. Väx- ters förmåga att locka till sig pollinatörer beror på en rad egenskaper, såsom blomningstidpunkt, blomstorlek, blomfärg och doft, men också på hur många blommor de samtidigt skyltar med samt hur högt över marken blommor presenteras. Pollinationsframgång beror dessutom av blommans form, som påverkar hur effektivt pollen överförs till och från blombesökare.

För att dokumentera selektionens styrka och form har jag samlat data på enskilda växters egenskaper och reproduktionsframgång i naturliga orkidé- populationer på Öland. För att testa om de korrelationer jag dokumenterat mellan egenskap och reproduktionsframgång representerar orsakssamband har jag manipulerat egenskaper som sporrelängd och planthöjd. Till sist, för att undersöka i vilken utsträckning naturlig selektion beror på samspel med pollinatörer respektive andra miljöfaktorer, har jag undersökt om selektionens styrka påverkas av att växter förses med ett överskott av pollen. Bland handpollinerade växter har egenskaper som normalt sett påverkar pollinat- ionsframgång inte längre någon betydelse för om växten blir välpollinerad eller inte.

I mina studier har jag testat följande förutsägelser: (1) Variation i reprodukt- ionsframgång och styrka av selektion orsakad av samspel med pollinatörer ökar med ökad grad av pollenbegränsning. (2) Egenskaper som påverkar förmåga att attrahera pollinatörer och egenskaper som påverkar pollenöver- föringens effektivitet har synergistiska effekter på pollinationsframgång och fröproduktion. (3) Effekten av nektarsporrars längd på pollinationsframgång och fröproduktion skiljer mellan populationer som pollineras av kort- re-

33 spektive lång-tungade pollinatörer. (4) Skillnader i egenskaper som påverkar frögroningsförmåga bidrar till lokal anpassning mellan nattviolpopulationer som växer i gräsmarker respektive skog.

För att undersöka betydelse av pollenbegränsning för variation i reproduktiv framgång och styrkan hos selektion på florala karaktärer utförde jag en stu- die som omfattade naturliga populationer av 12 orkidéarter. Jag fann att med ökande pollenbegränsning ökade variation i reproduktiv framgång, samt selektion orsakad av pollinatörer och total selektionsstyrka. Däremot var selektion orsakad av andra miljöfaktorer inte kopplad till grad av pollenbe- gränsning.

Orkidén nattviol förekommer i form av två ekotyper på Öland: (a) en kort- sporrad, kortvuxen ekotyp som förekommer i gräsmarker och som pollineras av svärmare med förhållandevis korta sugsnablar, och (b) en långsporrad, högvuxen ekotyp som förekommer i skogsmiljöer och som pollineras av ligustersvärmare som har en lång sugsnabel. Jag har i fältexperiment studerat betydelsen av skillnader i nektarsporrars längd (som kan förväntas påverka pollinationseffektivitet) och växters höjd (som kan förväntas påverka växters förmåga att attrahera pollinatörer) i skogs- och gräsmarkspopulationer.

Genom att experimentellt reducera sporrelängd och växters höjd kunde jag visa att båda dessa egenskaper påverkar pollinationsframgång och fröpro- duktion i en skogspopulation, men någon synergistisk effekt observerades inte. I den population som experimentet utfördes kunde jag detektera selektion som gynnade högvuxna plantor och jag kunde experimentellt visa att det berodde på att högvuxna plantor blev bättre pollinerade.

I ett annat experiment ersatte jag blommors nektarsporrar med konstgjorda sporrar för att jämföra hur långa sporrar måste vara för att blommor ska bli effektivt pollinerade i en gräsmarkspopulation och i en skogspopulation av nattviol. Som förväntat krävdes det längre sporrar i skogspopulationen där pollinatörerna har längre sugsnablar. Mina resultat visade dessutom att pre- cision i pollenöverföring är viktigare för växters förmåga att motta pollen än för deras förmåga att avsätta pollen på pollinatörer. Pollinationsframgången ökade med ökad sporrelängd, men någon selektion på sporrelängd orsakad av pollinatörer kunde inte påvisas. Avsaknad av selektion på sporrelängd i den naturliga populationen beror sannolikt på att variationen i sporrelängd var begränsad och att också individer med relativt korta sporrar blev tillräck- ligt pollinerade.

Populationer som förekommer i olika miljöer kan utveckla skillnader som ger dem fördelar som uttrycks under flera olika stadier av livscykeln. Or- kidéers frön är små och är kända för att ha väldigt specifika krav för fram-

34 gångsrik groning. I ett såddexperiment undersökte jag om gräsmarks- och skogspopulationer av nattviol uppvisar adaptiva skillnader i groningsför- måga. Jag begravde frö från två gräsmarks- och två skogspopulationer på alla fyra ursprungslokalerna på hösten, och kontrollerade groningsframgång följande år. Resultaten gav inte stöd för hypotesen att skillnader i gronings- krav bidrar till lokal anpassning mellan gräsmarks- och skogspopulationer. I stället var det en av gräsmarkspopulationerna som genomgående hade den högsta groningsprocenten.

Sammantaget illustrerar mitt avhandlingsarbete hur en kombination av fält- experiment och empiriska studier av naturlig selektion kan bidra till en ökad förståelse av de mekanismer som styr variation i det naturliga urvalets form och riktning. Det är information som är viktig för att tolka variation i egenskaper inom och mellan populationer, men också för att kunna förutsäga konsekvenser av pågående miljöförändringar.

35 Acknowledgements

First of all, I am exceptionally grateful to my supervisor and mentor Jon Ågren. Thank you for all your support from the first moment that I contacted you as an immature Master student to the stage I reached today. I am always impressed by your outstanding ability to capture details and I hope to bring some of that with me when I am going my own way from now. I am thankful for your patience with me, for putting me back on track when I lost it, and for lifting me up when I was frustrated. I am also thankful to Nina Sletvold, my co-supervisor on this journey. Since the days we first met you played an important role for my development as a scientist and person, and I will always remember your open ear in the difficult moments and your support to realize this dream.

Three quarter of this thesis would never have happened without the helping hand of Lovisa D. Thank you so much for all your great help in the field, and always keeping a fresh mind for the important things when I got tired at the end of a long field season. These long days and nights would not have been as enjoyable without you. For the first quarter of my thesis I am greatly thankful to Linnus V. and Malin B.

I would like to thank the entire Department of Plant Ecology and Evolution, particularly all former and current PhD colleagues, for the many wonderful years working there. I want to thank Elodie C. for sharing the enthusiasm for orchids and their pollinators, and the many fun stories that can happen during fieldwork on Öland. I am grateful to Fia B. for her help with realizing so wonderful my imagination for the cover layout. I want to thank Karin G., Amy P. and Magne F. for insightful discussion, inspiration and mental support. Tom E. for taking all my statistic questions, and Lina L. for sharing office during the first year and Ingvar B. for giving me new company in the final years.

Thanks to all from station Linné, Dave K., Kenneth P., Mareike K., Catrin and Mathias J., Jessica M. and most of all Anne W. for helping with all the small and big practical issues during fieldwork and communicating my needs and thanks to the farmers. Finally, I want to thank all farmers, who kindly allowed me to work on their land: Niklas O., Börje K., Greger J., Kjell and Mari J., Markus L. and Tore J.

36 The Svensk Växtgeografiska Sällskapet, the former Gradschool for Genomes and Phenotypes, and the Gertud Thelin, Sernanders and Linnéska stipend- stiftelse financially supported me for fieldwork, and conference and educa- tional travel.

I am grateful to Johan Hammer for joining me during fieldwork in 2015, and documenting my work in such beautiful pictures (cover pictures for Paper I- III).

I should not miss to thank Charly “Olof” K. and Merima M. from Cam- pus1477 for keeping me in physically and mentally sustainable shape during these years.

Tack alla i Team Nordic Trail Uppsala och på högkvarteret i Stockholm, Miranda K. och Robin J., för denna underbara tid ute i skogen och bergen, som har varit så viktig för mig under dessa år. Ett stort tack till mina ledar- kompisar Jenni J., Sandra L., Jesper A. och Robert I. för förståelse om jag har behövt stanna längre på jobbet eller har missat någon aktivitet. Sofia A., Alexandra P., Rose-Marie S., Per O. och Hendrik R. utan ni skulle ultra löp- ning i Uppsala vara bara hälften så fin och jag är världigt tacksam att har träffat ni. Martin S. och Magnus V. tack för driv när det kom till lite fart.

Ich möchte natürlich auch Julia W., Diana V., Jenni H. und Matthias Q. danken für euer Verständnis und eure Unterstützung all die Jahre seit wir zu- sammen im Biologieunterricht die Schulbank gedrückt haben. Weiterhin geht ein grosser Dank an Maria T., Linda. B. und Sieglinde K. dafür, dass ihr mich so herzlich immer wieder aufgenommen habt, wenn ich müde war und in die Berge entfliehen musste um neue Kraft zu tanken.

Der größte Dank von Allen gilt ohne Zweifel euch, Mama, Papa und Anne. Ich bin euch so dankbar für all die Jahre Unterstützung in jeglicher Hinsicht mit allen Höhen und Tiefen, die ihr mit mir durchgemacht habt. Egal wo ich gerade war, ich konnte mich immer darauf verlassen, dass ihr für mich da seid. Selbst in den Momenten in denen nichts mehr geklappt hat, habt ihr mir gezeigt, dass den Kopf hängen zu lassen zu keiner Lösung führt und Aufge- ben keine Alternative ist. In diesem Sinne, möchten ich ganz besonders dir, Papa, danken: auch ohne das man es beim Namen nennen muss bist du das größte Leitbild dieser Arbeit.

37 References

Anderson, B. and Johnson, S.D. 2009. Geographical covariation and local convergence of flower depth in a guild of fly-pollinated plants. New Phytol. 182: 533-540. Andersson, S., and B. Widén. 1993. Pollinator.mediated selection on floral traits in a synthetic population of Senecio integrifolius (Asteraceae). Oikos 66: 72-79. Arnold, S.J. and Wade, M.J. On the measurement of natural and sexual selection: Theory. Evolution 38: 709-719. Batty, A.L., Dixon, K.W., Brundrett, M. and Sivasithamparam, K. 2001. Constraints to symbiotic germination of terrestrial orchid seed in a mediterranean bushland. New Phyt. 152: 511-520. Baty, F., Ritz, C., Charles, S., Brutsche, M., Flandrois, J.-P. and Delignette, M.-L. 2015. A toolbox for non-linear regression in R: The package nlstools. J. Stat. Softw. 66: 1-21. Benitez-Vieyra, S., Medina, A.M., Glinos, E. and Cocucci, A.A. 2006. Pollinator- mediated selection on floral traits and size of floral display in Cyclopogon elatus, a sweat bee-pollinated orchid. Funct. Ecol. 20: 948-957. Benkman, C.W. 2013. Biotic interaction strength and the intensity of selection. Ecol. Letters 16: 1054-1060. Bischoff, M., Raguso, R.A., Jürgens, A., and Campbell, D.R. 2015. Context- dependent reproductive isolation mediated by floral scent and color. Evolution 69: 1-13. Blanquart, F., Kaltz, O., Nuismer, S.L. and Gandon, S. 2013. A practical guide to measuring local adaptation. Ecol. Letters 16: 1195-1205. Boberg, E. and Ågren, J. 2009. Despite their apparent integration, spur length but not perianth size affects reproductive success in the moth-pollinated orchid Platanthera bifolia. Funct. Ecol. 23: 1022-1028. Boberg, E. 2010. Evolution of spur length in a moth-pollinated orchid. Uppsala University, Acta Universitatis Upsaliensis. Boberg, E., Alexandersson, R., Jonsson, M., Maad, J., Ågren, J. and Nilsson, L.A. 2014. Pollinator shifts and the evolution of spur length in the moth-pollinated orchid Platanthera bifolia. Ann. Bot. 113: 267-275. Brodie, E.D. III, Moore, A.J., Janzen, F.J. 1995. Visualizing and quantifying natural selection. Trends Ecol. Evol. 10: 313-318. Campbell, D.R. 1998. Variation in lifetime male fitness in Ipomopsis aggregata: Tests of sex allocation theory. Am. Nat. 152: 338-353. Campbell, D.R. 2009. Using phenotypic manipulations to study multivariate selection of floral trait associations. Ann. Bot. 103: 1557-1566. Campbell, D.R., Forster, M. and M. Bischoff, M. 2014. Selection of trait combinations through bee and fly visitation to flowers of Polemonium follosissimum. J. Evol. Biol. 27: 325-336.

38 Campbell, D.R., Jürgens, A. and Johnson, S.D. 2016. Reproductive isolation between Zaluzianskya species: the influence of volatiles and flower orientation on hawkmoth foraging choices. New Phyt. 210: 333-342. Caruso, C.M., Martin, R.A., Sletvold, N., Morrissey, M.B., Wade, M.J., Augustine, K.E., Carlson, S.M., MacColl, A.D.C., Siepielski, A.M. and Kingsolver, J.G. 2017. What are the environmental determinants of phenotypic selection? A meta-analysis of experimental studies. Am. Nat. 190: 363-376. Chapurlat, E., Ågren, J. and Sletvold, N. 2015. Spatial variation in pollinator- mediated selection on phenology, floral display and spur length in the orchid Gymnadenia conopsea. New Phyt. 208: 1264-1275. Claessens, J., Gravendeel, B. and Kleynen, J. 2008. Cucullia umbratica L. als Bestäuber von Platanthera x hybrida Brugg. in Süd-Limburg (Niederlande). J. Eur. Orch. 40: 73-84. Conner, J.K. 1996. Understanding natural selection: An approach integrating selection gradients, multiplicative fitness components, and path analysis. Ethol. Ecol. Evol. 8: 387-397. Cozzolino, S., and Widmer, A. 2005. Orchid diversity: an evolutionary consequence of deception? Trends Ecol. Evol. 20: 487-494. Crow, J.F. 1958. Some possibilities for measuring selection intensities in man. Human Biol. 30: 1-13. Darwin, C.M.A. 1859. The origin of species by means of natural selection, or the preservation of favoured races in the struggle of life. John Murray, London. Darwin, C.M.A. 1862. The various contrivances by which orchids are fertilzed by insects. University of Chicago Press. Delforge, P. 2005. Orchids of Europe, North Africa and the Middle East. A&C Black Publishers Ltd., Delachaux et Niestlé, Paris. Diez, J.M. 2007. Hierarchical patterns of symbiotic orchid germination linked to adult proximity and environmental gradients. Ecology 95: 159-170. Dobzhansky, T. 1937. Genetic nature of species differences. Am. Nat. 71: 404-420. Dobzhansky, T. 1950. Evolution in the tropics. Am. Sci. 38: 209-221. Dobzhansky, T. 1973. Nothing in biology makes sense except in the light of evolution. American Biology Teacher 35: 125-129. Downhower, J.F., Blumer, L.S. and Brown, L. 1987. Opportunity for selection - an appropriate measure for evaluating variation in the potential for selection? Evolution 41: 1395-1400. Dudash, M.R., Hassler, C., Stevens, P.M., and Fenster, C.B. 2011. Experimental floral and inflorescence trait manipulations affect pollinator preferences and function in a hummingbird-pollinated plant. Am. J. Bot. 98: 275-282. Ellis, A.G. and Johnson, S.D. 2010. Gender Differences in the effects of floral spur length manipulation on fitness in a hermaphrodite orchid. Int. J. Plant Sci. 171: 1010-1019. Endler, J.A. 1986. Natural selection in the wild. Princeton University Press. Fay, M. F. and Chase, M.W. 2009. Orchid biology: from Linnaeus via Darwin to the 21st century. Ann. Bot. 104: 359-364. Fenster, C.B., Cheely, G., Dudash, M.R. and Reynolds, R.J. 2006. Nectar reward and advertisment in hummingbird-pollinated Silene virginica (Caryophyllaceae). Am. Nat. 93: 1800-1807. Fenster, C. B., Reynolds, R.J., Williams, C.W., Makowsky, R. and Dudash, M.R. 2015. Quantifying hummingbird preference for floral trait combinations: The role of selection on trait interactions in the evolution of pollination syndromes. Evolution 69: 1113-1127.

39 Gomez, J.M. 2003. Herbivory reduces the strength of pollinator-mediated selection in the Mediterranean herb Erysimum mediohispanicum: Consequences for plant specialization. Am. Nat. 162: 242-256. Grant, B.R., and Grant, P.R. 1989. Natural-selection experiments in a population of Darwins finches. Am. Nat. 133: 377-393. Grant, V. and Grant, K.A. 1965. Flower pollination in the Phlox family. New York, Columbia University Press. Grant, V. 1981. Plant speciation. Columbia University Press, New York. Henshaw, J.M. and Zemel, Y. 2017. A unified measure of linear and non-linear selection on quantitative traits. Methods Ecol. Evol. 8: 604-614. Hereford, J. 2009. A quantitative survey of local adaptation and fitness trade-offs. Am. Nat. 173: 579-588. Hoekstra, H.E., Hoekstra, J.M., Berrigan, D., Vignieri, S.N., Hoang, A., Hill, C.E., Beerli, P. and Kingsolver, J.G. 2001. Strength and tempo of directional selection in the wild. Proc. Nat. Acad. Sci. USA 98: 9157-9160. Jacquemyn, H., Waud, M., Merckx, V., Lievens, B. and Brys, R. 2015. Mycorrhizal diversity, seed germination and long-term changes in population size across nine populations of the terrestrial orchid Neottia ovata. Mol. Ecol. 24: 3269-3280. Johnson, S.D. and Steiner, K.E. 1997. Long-tongued fly pollination and evolution of floral spur length in the Disa draconis complex (Orchidaceae). Evolution 51: 45-53. Johnson, S.D. 2012. Using model and manipulated flowers to study pollinator- mediated selection. SA J. Bot. 79: 191-191. Kawecki, T.J., and Ebert, D. 2004. Conceptual issues in local adaptation. Ecol. Letters 7: 1225-1241. Kay, K.M., and Sargent, R.D. 2009. The role of animal pollination in plant speciation: Integrating ecology, geography, and genetics. Ann. Rev. Ecol. Evol. Syst. 40: 637-656. Kettlewell, H.B.D. 1955. Selection experiments on industrial melanism in the Lepidoptera. Heredity 9: 323-342. Kew Royal Botanic Garden. 2017. State of the World´s Plant List 2017. Retrieved from: https://stateoftheworldsplants.com. (accessed 20th April 2018). Kier, G., Mutke, J., Dinerstein, E., Ricketts, T.H., Kuper, W., Kreft, H. and Barthlott, W. 2005. Global patterns of plant diversity and floristic knowledge. J. Biogeography 32:1107-1116. Kingsolver, J.G. 1999. Experimental analyses of wing size, flight, and survival in the western white butterfly. Evolution 53: 1479-1490. Kingsolver, J.G., Diamond, S.E., Siepielski, A.M. and Carlson, S.M. 2012. Synthetic analyses of phenotypic selection in natural populations: lessons, limitations and future directions. J. Evol. Ecol. 26: 1101-1118. Kingsolver, J.G., Hoekstra, H.E., Hoekstra, J.M., Berrigan, D., Vignieri, S.N., Hill, C.E., Hoang, A., Gibert, P. and Beerli, P. 2001. The strength of phenotypic selection in natural populations. Am. Nat. 157: 245-261. Kingsolver, J.G., and Pfennig, D.W. 2007. Patterns and power of phenotypic selection in nature. Bioscience 57: 561-572. Körner, C. 2011. Coldest places on earth with angiosperm plant life. Alp. Bot. 121: 11-22. Lande, R. and Arnold, S.J. 1983. The measurement of selection on correlated characters. Evolution 37: 1210-1226. Lee, J. 1992. Comparison of variance between correlated samples. Comp. Appl. Biosciences 8: 405-406.

40 Lehndal, L. and Ågren, J. 2015. Herbivory differentially affects plant fitness in three populations of the perennial herb Lythrum salicaria along a latitudinal gradient. PloS ONE 10: 13. Leimu, R. and Fischer, M. 2008. A meta-analysis of local adaptation in plants. PloS ONE 3. Mayr, E. 1947. Ecological factors in speciation. Evolution 1: 263-288. Mayr, E. 1963. Animal species and evolution. Harvard Univ. Press, Cambridge, Ma. McColl, A.D.C. 2011. The ecological causes of evolution. Trends Ecol. Evol. 26: 514-522. McGraw, J.B. and Antonovics, J. 1983. Experimental ecology of Dryas octopetala ecotypes. I. Ecotypic differentiation and life-cycle stages of selection. Ecology 71: 879-897. Medel, R., Botto-Mahan, C. and Kalin-Arroyo, M. 2003. Pollinator-mediated selection on the nectar guide phenotype in the Andean monkey flower, Mimulus luteus. Ecology 84: 1721-1732. Micheneau, C., Johnson, S.D. and Fay, M. F. 2010. Orchid pollination: from Darwin to the present day. Bot. J. Linn. Soc. 162: 10-28. Mojica, J.P., and Kelly, J.K. 2010. Viability selection prior to trait expression is an essential component of natural selection. Proc. R. Soc. Sci. 277: 2945-2950. Mora, C., Tittensor, D.P., Adl, S., Simpson, A.G.B. and Worm, B. 2011. How many species are there on earth and in the ocean? PLOS Biol. 9: 8. Morrisey, M.B. 2014. In search for the best methods for multivariate selection analysis. Methods Ecol. Evol. 5: 1095-1109. Muchhala, N. and Thomson, J.D. 2009. Going to great lengths: selection for long corolla tubes in an extremely specialized bat-flower mutualism. Proc. R. Soc. B. 276: 2147-2152. Nattero, J., Sérsic, A.N. and Cocucci, A.A. 2010. Patterns of contemporary phenotypic selection and flower integration in the hummingbird-pollinated Nicotiana glauca between populayions with different flower-pollinator combinations. Oikos 119: 852-863. Nattero, J., Sérsic, A.N. and Cocucci, A.A. 2011. Geographic variation of floral traits in Nicotiana glauca: Relationships with biotic and abiotic factors. Acta Oecol. 37: 503-511. Newman, E., Manning, J. and Anderson, B. 2014. Matching floral and pollinator traits through guild convergence and pollinator ecotype formation. Ann. Bot. 113: 373-384. Nilsson, L.A. 1978a. Pollination ecology and adaptation in Platanthera chlorantha (Orchidaceae). Botaniska Notiser 131:35-51. Nilsson, L.A. 1978b. Pollination ecology of Epipactis palustris (Orchidaceae). Bot. Notiser 131: 355-368. Nilsson, L.A. 1980. The pollination ecology of Dactylorhiza sambucina (Orchidaceae). Bot. Notiser 133: 367-385. Nilsson, L.A. 1981. The pollination ecology of Listera ovata (Orchidaceae). Nord. J. Bot. 1: 461-480. Nilsson, L.A. 1983a. Anth ecology of Orchis mascula ssp mascula (Orchidaceae). Nord. J. Bot. 3: 157-179. Nilsson, L.A. 1983b. Processes of isolation and introgressive interplay between Platanthera bifolia (L.) RICH and Platanthera chlorantha ((CUSTER) REICHB (Orchidaceae). Bot. J. Linn. Soc. 87: 325-350. Nilsson, L.A. 1984. Antecology of Orchis morio (Orchidaceae) at its outpost in the north. Acta Universitatis Upsaliensis Nova Acta Regiae Societatis Scientiarum Upsaliensis Series V C 3:167-180.

41 Nilsson, L.A. 1988. The evolution of flowers with deep corolla tubes. Nature 334: 147-149. O´Donald, P. 1970. Change in fitness by selection for a quantitative character. Theor. Popul. Biol. 1: 219-232. Paudel, B.R., Shrestha, M., Burd, M., Adhikari, S., Sun, Y.-S. and Li, Q.-J. 2016. Coevolutionary elaboration of pollination-related traits in an alpine ginger (Roscoea purpurea) and a tabanid fly in the Nepalese Himalayas. New Phyt. 211: 1402-1411. Pauw, A., Stofberg, J. and Waterman, R.J. 2009. Flies and flowers in Darwin´s race. Evolution 63: 268-279. Peakall, R. and Handel, S.N. 1993. Pollinators discriminate among floral heights of a sexually deceptive orchid - implications for selection. Evolution 47: 1681- 1687. Phillips, P.C. and Arnold, S.J. 1989. Visualizing multivariate selection. Evolution 43:1209-1222. Pitman, E.J.G. 1939. A note on normal correlation. Biometrika 31: 9-12. Postma, F. M. and Ågren, J. 2016. Early life stages contribute strongly to local adaptation in Arabidopsis thaliana. Proc. Nat. Acad. Sci. USA 113: 7590-7595. Price, G. R. 1970. Selection and covariance. Nature 227: 520. Price, G.R. 1972. Extensions of covariance selection mathematics. Ann. Hum. Genetics 35: 485. Rasmussen, H.N. and Whigham, D.F. 1993. Seed ecology of dust seeds in-situ - a new study technique and its application in terrestrial orchids. Am. J. Bot. 80: 1374-1378. Reed, D.H. and Bryant, E.H. 2004. Phenotypic correlations among fitness and its components in a population of the housefly. J. Evol. Biol. 17: 919-923. Robertson, A. 1966. Artificial selection in plants and animals. Proc. Nat. Acad. Sci. 164: 341. Sandring, S. and Ågren, J. 2009. Pollinator-mediated selection on floral display and flowering time in the perennial herb Arabidopsis lyrata. Evolution 63: 1292- 1300. Schemske, D.W. 1984. Population-structure and local selection in Impatiemns pallida (Balsaninaceae), a selfing annual. Evolution 38: 817-832. Schemske, D. W. and Bradshaw, H.D. 1999. Pollinator preference and the evolution of floral traits in monkeyflowers (Mimulus). Proc. Nat. Acad. Sci. USA 96: 11910-11915. Schemske, D.W. 2000. Understanding the origin of species. Evolution 54: 1069- 1073. Schemske, D.W. 2010. Adaptation and the origin of species. Am. Nat. 176: 4-25. Schiestl, F. P. and Johnson, S.D. 2013. Pollinator-mediated evolution of floral signals. Trends Ecol. Evol. 28: 307-315. Schluter, D. 1988. Estimating the frm of natural selection on a quantitative trait. Evolution 42: 849-861. Schluter, D. 2001. Ecology and the origin of species. Trends Ecol. Evol. 16: 372- 380. Siepielski, A.M., Gotanda, K.M., Morrissey, M.B, Diamond, S.E., DiBattista, J.D. and Carlson, S.M. 2013. The spatial patterns of directional phenotypic selection. Ecol. Letters 16: 1382-1392. Sletvold, N. and Ågren, J. 2010. Pollinator-mediated selection on floral display and spur length in the orchid Gymadenia conopsea. Int. J. Plant Sci. 171: 999-1009. Sletvold, N. and Ågren, J. 2011. Nonadditive effects of floral display and spur length on reproductive success in a deceptive orchid. Ecology 92: 2167-2174.

42 Sletvold, N., Grindeland, J.M. and Ågren, J. 2010. Pollinator-mediated selection on floral display, spur length and flowering phenology in the deceptive orchid Dactylorhiza lapponica. New Phyt. 188: 385-392. Sobel, J.M., Chen, G.F, Watt, L.R. and Schemske, D.W. 2010. The biology of speciation. Evolution 64: 295-315. Stebbins, G.L. 1950. Variation and evolution in plants. Columbia University Press, New York. Stebbins, G.L. 1970. Adaptive radiation of reproductive characteristics in angiosperms. I. Pollination mechanisms. Ann. Rev. Ecol. Syst. 1: 307-326. Stpiczynska, M. 1997. The structure of nectary of Platanthera bifolia L. Orchidaceae. Acta Societatis Botanicorum Poloniae 66: 5-11. The Plant List (2010). Version 1. Published on the Internet; http://www.theplantlist.org/ (accessed 20th April 2018). Theis, A., Salzburger, W. and Egger, B. 2012. The Function of Anal Fin Egg-Spots in the Cichlid Fish Astatotilapia burtoni. PloS ONE 7: 8. Thompson, J.N. 1994. The geographic mosaic of coevolution. Chicago University Press. Tollsten, L. and Bergström, J. 1989. Variation and post-pollination changes in floral odors released by Platanthera bifolia (Orchidaceae). Nord. J. Bot. 9: 359-362. Tollsten, L. and Bergström, L.G. 1993. Fragrance chemotypes of Platanthera (Orchidaceae) - the result of adaptation to pollinating moth. Nord. J. Bot 13: 607-613. Toräng, P., Vikström, L, Wunder, J., Wotzel, S., Coupland, G. and Ågren, J. 2017. Evolution of the selfing syndrome: Anther orientation and herkogamy together determine reproductive assurance in a self-compatible plant. Evolution 71: 2206-2218. Tremblay, R.L., Ackerman, J.D, Zimmerman, J.K and Calvo, R.N. 2005. Variation in sexual reproduction in orchids and its evolutionary consequences: a spasmodic journey to diversification. Biol. J. Linn. Soc. 84: 1-54. Van der Niet, T., Peakall, R. and S. D. Johnson. 2014. Pollinator-driven ecological speciation in plants: new evidence and future perspectives. Ann. Bot. 113: 199- 211. Vujanovic, V., St-Arnaud, M., Barabe, D. and Thibeault, G. 2000. Viability testing of orchid seed and the promotion of colouration and germination. Ann. Bot. 86: 79-86. Vlašánková, A., Padyšáková, E., Bartoš, M., Mengual, X, Janečková, P. and Janeček, S. 2017. The nectar spur is not only a simple specialization for long- proboscid pollinators. New Phyt. 215: 1574-1581. Wade, M.J. and Kalisz, S. 1990. The causes of natural selection. Evolution 44: 1947- 1955. Wood, S.N. 2017. Generalized additive models: An introduction with R (2nd edition). Chapman and Hall/CRC. Xu, S., Schlüter, P.M., Scopece, G., Breitkopf, H., Gross, K., Cozzolino, S. and Schiestl, F.P. 2011. Floral isolation is the main reproductive barrier among closely related sexually deceptive orchids. Evolution 65: 2606-2620.

43 Acta Universitatis Upsaliensis Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1677 Editor: The Dean of the Faculty of Science and Technology

A doctoral dissertation from the Faculty of Science and Technology, Uppsala University, is usually a summary of a number of papers. A few copies of the complete dissertation are kept at major Swedish research libraries, while the summary alone is distributed internationally through the series Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology. (Prior to January, 2005, the series was published under the title “Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology”.)

ACTA UNIVERSITATIS UPSALIENSIS Distribution: publications.uu.se UPPSALA urn:nbn:se:uu:diva-349139 2018