Taxonomical use of floral scent data in apomictic taxa of Hieracium and Sorbus derived from hybridization

Taxonomical use of floral scent data in apomictic taxa of Hieracium and Sorbus derived from hybridization Dissertation zur Erlangung des Doktorgrades Dr. rer. nat. an der Fakultät Biologie/Chemie/Geowissenschaften der Universität Bayreuth vorgelegt von Martin Feulner, aus Bayreuth Bayreuth, 2013 Die vorliegende Arbeit wurde von Mai 2006 bis Januar 2013 am Lehrstuhl Pflanzensystematik der Universität Bayreuth unter Betreuung von Frau Prof. Dr. Sigrid Liede-Schumann und Herrn Prof. Dr. Stefan Dötterl angefertigt. Vollständiger Abdruck der von der Fakultät für Biologie, Chemie und Geowissenschaften der Universität Bayreuth genehmigten Dissertation zur Erlangung des akademischen Grades eines Doktors der Naturwissenschaften (Dr. rer. nat.). Dissertation eingereicht am: 08.02.2013 Zulassung durch die Prüfungskommission: 13.02.2013 Wissenschaftliches Kolloquium: 04.07.2013 Amtierender Dekan: Prof. Dr. Beate Lohnert Prüfungsausschuss: Prof. Dr. Sigrid Liede-Schumann (Erstgutachter) Prof. Dr. Bettina Engelbrecht (Zweitgutachter) Prof. Dr. Stefan Dötterl Prof. Dr. Heike Feldhaar PD Dr. Gregor Aas This dissertation is submitted as a ‘Cumulative Thesis’ that includes four publications: two published articles, one submitted article, and one article in preparation for submission. List of Publications 1) Feulner M., Schuhwerk F., Dötterl S. 2009: Floral scent analysis in Hieracium subgenus Pilosella and its taxonomical implications. Flora 204: 495–505. 2) Feulner M., Schuhwerk F., Dötterl S. 2011: Taxonomical value of inflorescence scent in Hieracium s. str. Biochemical Systematics and Evolution 39: 732–743. Submitted (revision under review): 3) Feulner M., Liede-Schumann, S., Meve, U., Weig A., Aas, G.: Genetic structure of Sorbus latifolia (Lam.) Pers. taxa endemic to Northern Bavaria. Submitted to Plant systematics and evolution, PLSY-D-12-00168. In preparation for submission to Organisms Diversity & Evolution: 4) Feulner, M., Pointner, S., Heuss, L., Aas, G., Dötterl, S.: Floral scent and its correlation with genetic data in Sorbus taxa. Publications not included into this thesis: 5) Feulner, M., Möseler, B.M., Nezadal, W. 2001: Introgression und morphologische Variabilität bei der Blauen Himmelsleiter, Polemonium caeruleum L. in Nordbayern, Deutschland, Feddes Repertorium, 112: 231–246. 6) Feulner, M., Konnert, M. 2007: Autochthone Weißtannenvorkommen in den Schluchten Fränkischer Keupergebiete. Diskussionsbeitrag zu deren genetischer Struktur, Artenausstattung, waldbaulicher Behandlung und Kartierung in Natura 2000, Waldoekologie online 4: 91–110. Declaration of contribution to publications The thesis contains four research articles. Most of the research work presented in this thesis was carried out by myself at the University of Bayreuth including all sample collections, most analytic work and the most statistics under profound support of Prof. Dr. Stefan Dötterl, PD Dr. Ulrich Meve, Dr. Alfons Weig, PD Dr. Gregor Aas and Prof. Dr. Sigrid Liede-Schumann. I prepared the manuscripts under consideration of the comments of all coauthors. 1st publication Feulner M., Schuhwerk F., Dötterl S. (2009): Floral scent analysis in Hieracium subgenus Pilosella and its taxonomical implications. Flora 204: 495–505. The field work was done by myself, data analysis was done by myself under the profound support of PD Dr. Stefan Dötterl. Norbert Meyer and Dr. Franz Schuhwerk contributed to species selection, gave profound advice about localities of endemic Hieracium taxa and helped with species identification in the field and of herbarium specimens. I prepared the manuscript by recognizing the profound comments of my co-authors. 2nd publication Feulner M., Schuhwerk F., Dötterl S. (2011): Taxonomical value of inflorescence scent in Hieracium s. str. Biochemical Systematics and Ecology 39: 732–743. The field work was conducted by myself, data analysis was done by myself under the profound support of PD Dr. Stefan Dötterl. Dr. Jochen Müller, Norbert Meyer and Dr. Franz Schuhwerk gave profound advice about localities of endemic Hieracium taxa and helped with species identification in the field and of herbarium specimens. I prepared the manuscript by recognizing the comments of my co-authors. 3rd publication Feulner, M., Liede-Schumann, S., Meve, U., Weig A., Aas G.: Origin and genetic structure of three Sorbus latifolia (Lam.) Pers. taxa endemic to Northern Bavaria. Submitted to Plant systematics and evolution, PLSY-D-12-00168-1. The plant material collection was conducted by myself under the support of Dr. Gregor Aas (EBG Bayreuth) who had the idea for this research. AFLP Laboratory work was conducted by Michaela Hochholzer and Dr. Alfons Weig (both DANECO Bayreuth). Chromosome counts were conducted under supervision of PD Dr. Ulrich Meve. Data analysis was done by myself under the support of Dr. Alfons Weig and Prof. Dr. Sigrid Liede-Schumann. I prepared the manuscript by recognizing the comments of my coauthors. 4th publication Feulner, M. Pointner S., Heuss L., Aas G., Dötterl S.: Correlation between taxonomic groupings of Sorbus microspecies based on floral scent and genetic data. In preparation for submission to Organisms Diversity & Evolution. The scent collection was done by the bachelor candidates Lisa Heuss, Stefan Pointner and myself. Data analysis was done by myself under the profound support of PD Dr. Stefan Dötterl. I prepared the manuscript by recognizing the comments of my co-authors. Content Summary 1 Zusammenfassung 3 1. General Introduction 5 2. Aims of research 8 3. Synopsis 9 3.1 Material and Methods 9 3.2 Results and Discussion 17 3.3 Conclusion and perspectives 23 3.4 References 25 Publications 31 1. Floral scent analysis in Hieracium subgenus Pilosella and its taxonomical implications. 31 2. Taxonomical value of inflorescence scent in Hieracium s. str. 43 3. Genetic structure of three Sorbus latifolia (Lam.) Pers. taxa endemic to Northern Bavaria. 56 4. Floral scent and its correlation with genetic data in Sorbus taxa. 78 Danksagung 95 Eidesstattliche Erklärung 96 Summary Summary Scent investigations are mainly carried out in the context of the ecological function of scent components for pollinator attraction, to study their geographical variability or their evolution. In contrast, profiles of inflorescence scent compounds were rarely used for phylogenetic analyses and taxonomy. So far no investigations are available that focus on scent of apomictic plant groups and its value for the taxonomy of these groups. Apomicts produce fertile seeds without pollination, either from somatic cells of the nucellus or unreduced embryo sac cells. Overall, apomixis occurs scattered over the whole Angiosperm tree in early as well as late branching families. Rosaceae and Asteraceae are particularly rich in apomictic taxa. In these families genera such as Hieracium and Sorbus contribute a lot to local species biodiversity in Central Europe because of their high number of apomictic species and taxa, which are often endemic. Most members of these genera built up polyploid microspecies swarms initiated by hybridization events. The reticulate structure of these taxon complexes leads to taxonomic difficulties that can not be solved by morphology alone. Therefore, molecular or chemical markers are needed to investigate the parentage of such taxa and their collective species membership. An important starting point of this research was the finding that artificial hybrids of Citrus produce a combination of the leaf and peel volatiles of their parents. This led to the hypothesis that natural hybrids might likewise produce scent patterns combined of the volatiles of their parents. Inflorescence scents were investigated from 64 mainly Bavarian taxa of the genus Hieracium (Subgenus Pilosella and Subgenus Hieracium). In Sorbus (Rosaceae) we focussed on three taxa of the S. latifolia agg., endemic to Northern Bavaria, S. adeana, S. cordigastensis and S. franconica and their parental species from the S. aria aggregate (agg.), as well as S. torminalis. Samples were collected with dynamic headspace method. Substances were identified by gas chromatography coupled to mass spectrometry (GC-MS). Scent data were evaluated by using various cluster methods. In Hieracium, additional reticulation analyses were applied that trace conflicting signals in a phylogram which can be interpreted as hybridizations between the taxa involved. In Sorbus additional AFLP (Amplified Fragment Length Polymorphism) analyses were carried out from the populations and individuals that were studied for scent and genetic and scent data were correlated. In Hieracium, the inflorescense scent consisted mainly of sesquiterpenes, monoterpenes, aliphatics and aromatics. In the flower scent of Sorbus besides these substances also nitrogene-containing substances were found. 1 Summary In Hieracium (both subgenera) as well as in Sorbus, taxa of hybrid origin showed a mixed scent pattern compared with the parental taxa. In many cases, the parental taxa that had been suggested by morphological investigations or revealed by genetic investigations could be confirmed by scent. In Hieracium subgenus Pilosella however, based on scent data, some critical subspecies could be shown to belong to another collective species (e.g. H. bauhini ssp. hispidissimum) than the one that had been proposed for morphological reasons. In general, scent patterns correlate well with morphological or genetical groupings, both in Hieracium and in Sorbus. Reticulate scent analyses in Hieracium showed that some taxa are probably derived from more than two ancestors. Scent data reflected even taxonomic patterns on a higher level, i.e. the sectional level. In Hieracium s.str. two main groups were found, the high growing and late flowering taxa such as H. umbellatum and H. laevigatum on the one hand, and the low growing earlier flowering taxa such as H. murorum and H. bifidum on the other hand. The AFLP study revealed that the selected members of the Sorbus latifolia agg. are genetically clearly differentiated and mostly of clonal structure. They are more closely related to S. aria than to S. torminalis. The S. aria agg. has a complex structure. Besides S. aria s.str. and S. pannonica there are also intermediate plants with affinity (aff.) to either the one or the other of these taxa. S. cordigastensis was derived from intermediates aff. S. aria s.str., and both S. adeana and S. franconica derived from intermediates aff. S. pannonica or S. pannonica itself. Floral scent of the same Sorbus taxa was investigated and compared with AFLP data on individual as well as on population level. Correlation analysis revealed a very high correlation between scent and AFLP data on individual and population level. Overall, this work shows that in two unrelated plant complexes, Hieracium and Sorbus, which both harbour a high number of apomictic species derived by hybridization, scent is of high taxonomical value. The main reason for this correlation may be that most taxa in the investigated groups possess mixed scent patterns from their parents or progenitors because they are of hybrid origin. In addition, intraspecific variability of scent patterns is low within apomictic taxa due to their clonality, simplifying the taxonomic use of scent data. Furthermore, the role of pollinator mediated selection of scent is reduced because of apomixis. 2 Zusammenfassung Zusammenfassung Schwerpunkt der Untersuchungen von Blütenund Infloreszenzdüften ist ihre ökologische Funktion bei der Anlockung von Bestäubern, ihre geografische Variabilität, oder es werden evolutive Aspekte des Duftes untersucht. Nur wenige Studien beschäftigen sich dagegen mit Düften und ihrem Potential zur Aufklärung taxonomischer Fragestellungen. Insbesondere bei apomiktischen Pflanzengruppen gibt es außer den hier vorgelegten Untersuchungen bei Hieracium und Sorbus bisher keine weiteren Untersuchungen zu diesem Thema. Bei Apomikten erfolgt die Embryobildung ohne Befruchtung aus einer somatischen Nucelluszelle oder einer unbefruchteten, unreduzierten Embryosackzelle. Apomixis findet man in Europa gehäuft bei Asteraceen und Rosaceen und hier insbesondere bei den Gattungen Hieracium (Asteraceae) und Sorbus (Rosaceae). Viele apomiktische Taxa dieser Gattungen gehören zu polyploiden Kleinartenschwärmen, die durch Hybridisierung entstanden sind. Aufgrund retikulater Merkmalsverteilung sind bei solchen Pflanzenarten die Ausgangssippen sowie die Zugehörigkeit von Kleinarten zu größeren Einheiten oft aus der Morphologie allein nicht zuverlässig abzuleiten. An künstlich erzeugten Citrus-Hybriden konnte nachgewiesen werden, dass sie neben wenigen neuen Düften eine Mischung aus den jeweiligen elterlichen Düften besitzen. Daher erschien es interessant, bei Pflanzenkomplexen, die zu großen Teilen auf natürliche Hybridisierung zurückgehen, die Zusammensetzung der Düfte zu untersuchen und deren Nutzen für systematische Fragestellungen zu erforschen. In der vorliegenden Arbeit wurden Blütenund Infloreszenzdüfte von 64 vielfach endemischen Taxa der Gattung Hieracium (Subgenus Pilosella und Subgenus Hieracium) untersucht. Bei Sorbus wurden drei in Nordbayern endemische Vertreter der S. latifolia-Gruppe (Sorbus adeana, S. cordigastensis, S. franconica) untersucht, die durch Hybridisierung zwischen S. aria und S. torminalis entstanden sind. Duft wurde im Gelände mittels der „Dynamic Headspace“ Methode abgesaugt und mit Gaschromatographie gekoppelt mit Massenspektrometrie (GCMS) analysiert. Die Düfte wurden aufgrund von Ähnlichkeiten in Beziehung gebracht, dabei kamen bei Hieracium auch Retikulationsanalysen, welche HybridisierungsEreignisse aufdecken können, zum Einsatz. Bei Sorbus wurden bisher fehlende genetische Untersuchungen mittels AFLP (amplified fragment length polymorphism) Analysen durchgeführt. Dies ermöglichte eine direkte Korrelation von Duftund genetischen Daten auf Populationsund teilweise auch Individuenebene. Bei Hieracium wurde der Infloreszenzduft vor allem von Sesquiterpenen, Monoterpenen, Fettsäurederivaten, und einigen Aromaten bestimmt. Im Duft von Sorbus 3 Synopsis – Material and Methods Hieracium harzianum ssp. pseudofranconicum Harz et Zahn, endemic to Northern Franconia on limestone rocks, Walberla. Hieracium caesium Fr., a rare relict species of the Southern Franconian Jura on limestone rocks, Essing. Hieracium schneidii x pilosella, a rare spontaneous hybrid between H. schneidii Schack et Zahn and H. pilosella L. H. schneidii Schack et Zahn, endemic to the Northern Frankonian Alb. Plate 1: Examples of investigated Hieracium taxa. 10 Synopsis – Material and Methods Furthermore, subspecies attribution to the one or the other collective species sometimes seems to be subjective (i.e. H. bauhini ssp. hispidissimum). Such units are doubtful because it is not clear yet whether they constitute natural units or are of polytopic origin (cf. Schuhwerk 2002). Therefore, it is important to investigate to which collective species some doubtful subspecies belong. The holarctic genus Sorbus (Plate 2) comprises a large variety of intermediate species evolved from hybridization between common and widely distributed taxa such as Sorbus torminalis, Sorbus aucuparia and Sorbus aria. In some areas of Europe endemic hybrids have been developed that are stabilized by apomixis (Kárpáti 1960, Düll 1961, Meyer et al. 2005, Lepší et al. 2009, Rich et al. 2010, Robertson et al. 2010). These hybrid species are distributed mostly in calcareous areas of Europe. In Sorbus, besides diplospory and apospory (Jankun and Kovanda 1987), another apomixis type, pseudogamy occurs, in which pollination is necessary to induce fruit set (Jankun and Kovanda 1987, Campbell and Dickinson 1990). According to Meyer et al. (2005) selfing is sufficient to induce fruit set in pseudogamous Sorbus. Here, Sorbus adeana, S. cordigastensis and S. franconica were investigated which are endemic in Northern Bavaria (Meyer et al. 2005, Aas and Kohles 2011). They have a small distribution area and typically occur along forest margins or in very open forest stages. Their parental lineages are S. torminalis and S. aria agg. The latter has a complicated phylogenetic structure. It comprises besides obligatory sexual taxa such as S. aria s.str., also several facultative or obligate apomictic lineages such as S. pannonica and S. graeca. Sorbus pannonica is a xeromorphic member of S. aria agg. and more widespread than S. aria s.str. in the northern Franconian Alb (Kutzelnigg 1995, Meyer et al. 2005). It is a non-lectotypified taxon which comprises presumably apomictic morphotypes filling the morphological gap between S. aria s.str. and S. graeca (Spach) Loddiges ex Schauer (Kárpáti 1960, Kutzelnigg 2005, Meyer et al. 2005). Sorbus graeca is another xeromorphic member of the S. aria agg., mainly distributed in the Mediterranean floral region. It reproduces sexually or is a facultative apomict (Kutzelnigg 1995). It is uncertain whether S. graeca occurs in the study area at all (Düll 1961, Kutzelnigg 1995, 2005), but individuals that can be attributed morphologically to S. graeca were found in the Northern Franconian Alb (own obs.), yet, it is difficult to delimitate this element against S. pannonica. 11 Synopsis – Material and Methods Sorbus pannonica Kárpáti, on the edge of pine forests in the Northern Franconian Alb, Oberailsfeld. Sorbus fanconica Bornm. ex Düll, on the edge of pine forests, endemic to Northern Franconia, Oberailsfeld. Fruits of Sorbus pannonica Kárpáti, in the Southern Franconian Jura, Deining. Sorbus adeana N. Mey., endemic to the Northern Franconian Alb, Modschiedel. Plate 2: Examples of investigated Sorbus taxa. 12 Synopsis – Material and Methods In the present thesis three main methods were applied: Scent collection and analysis using dynamic headspace and gas chromatography coupled to mass spectrometry (GC-MS), amplified fragment length polymorphism (AFLP) analyses and morphological analyses. Volatile collection and chemical analyzes of floral/inflorescence scent (publication 1, 2 and 4) Floral scent was collected in the field using the dynamic headspace method described by Dötterl and Jürgens (2005), and Dötterl et al. (2005a,b). Capitula (Hieracium) or inflorescences (Sorbus) were enclosed within a polyester oven bag (Toppits) and the emitted volatiles were trapped in an adsorbent tube through the use of a membrane pump (ASF Thomas, Inc.). As absorbent tube, we used ChromatoProbe quartz microvials of Varian Inc. (length: 15 mm; inner diameter: 2 mm), cut the closed end, filled them with a mixture (1:1) of 3 mg Tenax–TA (mesh 60–80) and Carbotrap (mesh 20–40), and fixed the adsorbent mixture in the vial with glass wool. Simultaneous collections of both the flower scent and surrounding air are used to distinguish between floral compounds and ambient contaminants. In Sorbus we used green leaf samples as blank, so we could reveal the floral scent by substracting these substances, whereas in Hieracium we collected the inflorescence scent because we used the surrounding air as blank. For each taxon, one sample of two to six individuals was collected. In Hieracium, sampling was carried out on fresh and newly opened inflorescences (the capitula of Hieracium are composed of many florets with most (90%) of them open at time), between 11 a.m. and 3 p.m, the period with the most intensive scent emission (as determined by the human nose; Feulner, unpublished data). In Sorbus, the 100–150 flowers of the pseudoumbels bloom in parallel. Here, headspaces covered single inflorescences, each. In both study plants, scent was collected for 3 to 5 minutes after a time of 3 to 10 minutes, where the scent accumulated in the closed bag. Headspace samples were analyzed on a Varian Saturn 2000 mass spectrometer coupled to a Varian 3800 gas chromatograph equipped with a 1079 injector (GC-MS) as described earlier (Dötterl and Jürgens 2005, Dötterl et al. 2005). The GC-MS data were processed using the Saturn Software package 5.2.1. Component identification was carried out using the NIST 02 mass spectral database, or MassFinder 2.3, and confirmed by the comparison of retention times with published data (Adams 1995). Identification of individual components could be confirmed by the comparison of both mass spectrum and GC retention data with those of authentic standards. 13 Synopsis – Material and Methods Scent data analysis (Publication 1,2 and 4) For both Hieracium and Sorbus data sets, pairwise qualitative similarities were calculated using the Jaccard similarity index. The significance of differences in scent profiles among taxa was assessed by ANOSIM with 10,000 random permutations based on these similarity matrices using Primer Version 5 and 6 (Clarke and Gorley 2001, 2006). In Hieracium, a reticulation network analysis was conducted with the program trex, version 4.0a1 (Makarenkov 2001) to analyse the relationships among the taxa. This method allows to visualize relationships of species interconnected with more than one ancestor (Legendre and Makarenkov 2002), which is important for analysing groups, such as Hieracium, with many taxa of hybrid origin (see Feulner et al. 2009). In this approach, a neighbour joining tree was constructed using a dissimilarity matrix (1-Jaccard), and homoplasies were made visible by so-called reticulation lines. Those homoplasies point towards hybridization or introgression (Legendre and Makarenkov 2002). In Hieracium, in addition to the presence and absence of compounds, we also calculated the average relative (percentage of total) amount of scent compounds of the single taxa. In Sorbus, the Jaccard matrix was used to cluster the scent data with UPGMA using Primer Version 5 and 6 (Clarke and Gorley 2006). Additionally intraspecific variability of scent data was compared among species using PERMDISP in PRIMER Version 6 (Clarke and Gorley 2006). Molecular methods, DNA marker Sample collection and DNA extraction (Publication 3) Leaf samples of Sorbus were taken in May and June 2010. Immediately after harvesting they were placed in plastic bags and put in a box with ice for transportation. At the same day, leaves were washed with ethanol in the laboratory and frozen in an extraction tube at –80 ° C until extraction. Frozen leaf samples (40 – 70 mg, 1 – 2 cm2) were extracted using widespread extraction systems and plant kits (NucleoMag 96 Plant kit; Machery-Nagel, Düren, Germany, FastPrep®-24 Tissue Homogenizer (MP Biomedicals Europe, Illkirch, France). The purified genomic DNA was diluted tenfold and used for all subsequent PCR reactions. 14 Synopsis – Material and Methods AFLP (Publication 3) AFLP is a fingerprinting method that allows discrimination between individuals (Vos 1995). It is helpful for closely related plant groups and can detect clonal structures derived i.e. by apomictic reproduction (Vos 1995). For AFLP is of fundamental importance to find appropriate specific primers for the second specific PCR step. For the preliminary primer search 24 primer combinations were tested, and the following six combinations were then selected for this study because they yielded the best results in species differentiation: MCAA/E-ACG, M-CAC/E-ACG, MCAC/E-ACA, M-CAT/E-ACG, M-CTC/E-ACG, MCTT/E-ACG. AFLP Data analysis (Publication 3) The reactions were separated on a vertical electrophoresis system (4200 Sequence Analysis System, Li-Cor Biosciences, Bad Homburg) together with DNA size markers (50–700 bp Sizing Standard, Li-Cor Biosciences, bad Homburg). AFLP banding patterns were evaluated using GeneMarker1-95 software (SoftGenetics). Band classes were calculated with a tolerance factor of 0.1 %. A neighbour joining (NJ) analysis of the presence and absence matrix was conducted (Nei-Li distance), followed by a bootstrap (BS) analysis after internode rooting with 1000 replicates using the program TREECON (Van de Peer and De Wachter 1994). For data of S. aria agg., we additionally applied model-based clustering (Pritchard et al. 2000) using the program STRUCTURE (http://pritch.bsd.uchicago.edu/structure.html) in order to retrieve the most likely number of groups within the S. aria agg. In order to investigate genetic variability, the number of polymorphic loci and Nei´s gene diversity “NGD” (Nei 1972) were calculated with Popgene (Yeh and Yang 1999). As a measure for the genetic distance between taxa we calculated Nei´s standard genetic distance (Ds) using the program POPGENE (Yeh and Yang 1999). Chromosome counts (Publication 3) In Sorbus, chromosome numbers were counted from root tip meristems of cultivated progeny of S. cordigastensis, S. adeana, S. franconica, and S. pannonica (one seedling per taxon), grown in the Ecological Botanical Garden and harvested in May 2010. Some of the seedlings were also included in the AFLP analysis. The fresh root tips were pretreated in 0.002 hydroxychinoline (4hrs), fixed in CARNOY´s solution and stained in carmine 15 Synopsis – Material and Methods following Snow (1963). From the stained root tips we prepared squash preparations in 45% acetic acid, and observed somatic metaphase in the microscope. Correlation analyses between scent and AFLP data (Publication 4) Correlation analyses between in scent and AFLP data were made on individual level and population level. For correlation analyses on population level presence-absence data occurring at least in one individual from both data sets - AFLP (see Feulner et al. 2013, submitted) and scent - were used. Similarity matrices (Jaccard) were calculated and these were the input for the RELATE correlation analysis (Spearman Rank correlation, 10,000 permutations) in PRIMER Version 6 (Clarke and Gorley 2006). 16 Synopsis – Results and Discussion 3.2 Results and Discussion Floral scent analysis in Hieracium subgenus Pilosella and its taxonomical implications (Publication 1). In Hieracium subgen. Pilosella floral scents of 27 predominantly Bavarian intermediate species, mostly of the collective species Hieracium calodon, H. zizianum and H. densiflorum were investigated with dynamic headspace method. Reticulate analyses were applied to depict hybrid speciation by visualizing relations between samples placed far from each other in a tree (Makarenkov 2001). Altogether, 56 floral scent compounds were identified, mainly aromatics, fatty acid derivatives and mono-, homoand sesquiterpenes. The chemical patterns were found to be taxon-specific and are thus of taxonomical value. The result that the basic (non-hybrid) species of Hieracium subgen. Pilosella such as H. piloselloides, H. echiodes or H. cymosum were well separated by scent underlined the utility of scent for taxonomical investigations in Hieracium. For many species of presumed hybrid origin, reticulation analyses of scent data allowed insights into their parentage. One example is H. fallax ssp. durisetum. This taxon is an intermediate between H. cymosum and H. echioides. In the scent tree it clusters close to H. cymosum, but the reticulation analysis connected it with H. echioides. The reason is that it has scent compounds of both taxa, on this basis it is possible to detect parental taxa (see above). In the scent tree, different subspecies of one taxon (e.g. H. zizianum) often clustered together. Exceptions have to be evaluated and can have taxonomical implications. Hieracium bauhini ssp. hispidissimum did not cluster with other members of H. bauhini but with H. densiflorum. In consequence it should be assigned to H. densiflorum and not to H. bauhini. There are morphological characters that support this placement, suggesting that the evaluation of some morphological characters needs to be reconsidered in the classification of Hieracium (in this case the cymose inflorescence structure). Another interesting result is that in the cluster analysis some subspecies of H. densiflorum do not group together with the other members of this group, but rather with the H. echioides derivates, such as H. calodon and H. fallax. They also show some morphological affinities to section Echioides such as dense, thick and curved bristle-like hairs on the stem. The subspecies H. densiflorum ssp. cymosiforme and H. densiflorum ssp. psammotrophicum may have been derived from taxa of section Echioides, as could been deviated from their strong thick and curved bristle-like hairs, too. Hieracium subgen. Pilosella taxa strongly deviated in their numbers of scent compounds (12-30 compounds). In hybrids this may lead to an overestimation of the 17 Synopsis – Results and Discussion influence of one of the parents. Nevertheless, this can make influences of species detectable which are not apparent in morphology. An example of this aspect is the primary hybrid H. schneidii x pilosella. This species is hard to differentiate against H. piloselliflorum by morphology alone. This taxon was found and identified here by floral scent and unpublished RAPD-markers (Gebauer and Feulner 2008, unpub.) for the first time. It grows as a spontaneous hybrid among its parents. Overall, scent patterns implicate that only a low number of taxa may be the ancestors of most of the hybrid taxa. Interestingly, this has been suggested already by Zahn (1921-1923, 1930-35), and the presumably parental species involved have been named as basic species by him (cf. Zahn 1921–1923, 1930–35). The placement of H. caespitosum in the scent tree close to H. zizianum (an intermediate species) speaks against its status as “basic species”. It shares also morphological traits with H. zizianum such as the straight hairs on the stem. Also Tichomirov (2000) considered H. caespitosum as hybrid between H. onegense (syn: H. caespitosum ssp. brevipilum, an eastern distributed species) and H. lactucella. It would be interesting in further studies to investigate the scent of H. onegense to confirm this hypothesis. Taxonomical value of inflorescence scent in Hieracium s. str. (Publication 2) Publication 2 deals with the taxonomical value of inflorescence scent in Hieracium s. str. in Central Europe. Hieracium s. str. comprises a vast number of mostly apomictic taxa presumably originated from hybridizations in the past. Inflorescence scents of 37 taxa from seven sections of Hieracium subgen. Hieracium were investigated by headspace analyses. Overall, 58 different scent compounds belonging to aromatics, sesquiterpenes, homoterpenes, monoterpenes and fatty acid derivatives were found. As in H. subgen. Pilosella (publication 1) inflorescence scent was found to be highly taxon-specific in Hieracium subgen. Hieracium. Taxonomy suggested by scent patterns was compared with results from genetic studies (Fehrer et al. 2009) that include many taxa investigated here by scent. Fehrer et al. (2009) identified two main groups, termed “western clade” and “eastern clade” in Hieracium s.str. using sequence data of chloroplast and mitochondrial markers (Fehrer et al. 2009). This differentiation was explained by different glacial refugia (Fehrer et al. 2009). The scent study identified the same main groups as in Fehrer et al. (2009). However, we found that these groups are identical with two distinct morphoand flowertypes, the high-growing and late-flowering-one, such as H. umbellatum and H. laevigatum and the low-growing and earlier flowering morphotype, such as H. murorum 18 Synopsis – Results and Discussion and H. bifidum. Some substances such as linalool and linalool oxide were found dominantly in species with high growth, whereas monoterpenes such as terpinolene were rather typical for the low-growing morphotypes. The low-growing scent group comprises sections such as Oreadea, Hieracium and Bifida, the high-growing group comprises the sections Drepanoidea, Tridentata and Hieracioides. Members of the low-growing species groups such as H. wiesbaurianum or H. glaucinum may be of polytopic origin since their taxa often clustered intermingled in the scent cluster. In contrast, in the scent tree most subspecies of H. murorum are placed next to each other and therefore are most likely monophyletic. Also H. bifidum is mainly placed in a group of its own despite some morphologically deviating members (i.e. Hieracium bifidum ssp. stenolepis var. valdefloccosum). Interestingly, the investigated H. bifidum taxa of H. bifidum grex bifidum and grex subcaesium are nested between H. murorum (low growing) and H. glaucum (high growing) in the scent tree. This intermediate position for a whole species group was proposed already by Koch (1838) and Zahn (1906) due to its morphological intermediacy and is confirmed here by scent data. This finding shows that floral scent can confirm the taxonomical position of species even at higher rank (i.e. sectional level). In H. franconicum (intermediate between H. murorum and H. bupleuroides) interpopulation differences could be found. Populations from Baden-Württemberg were closer to H. bupleuroides whereas populations from Franconia were closer to H. murorum. The same results were revealed by AFLP studies (Feulner, unpublished data). This result was unexpected and shows that hybrid taxa in Hieracium s.str. could be influenced by introgression. The scent study reveals an intermediate position for some taxa hitherto considered as basic species such as H. lachenalii and H. laevigatum, supporting the results of Fehrer et al. (2009) based on molecular data. The scent study shows that the intermediate taxa such as H. lachenalii, H. saxifragum, H. caesium were derived by multiple hybridization events between a very restricted number of members of the two morphological groups (highgrowing and low-growing). This imitates a clinal variation derived by stepwise evolution as was described by so-called reduction lines (i.e. Hieracium umbellatum - H. laevigatum - H. lachenalii - H. murorum, comp. Zahn 1921–1923, 1930–1935). Although the species of Hieracium subgen. Hieracium are older hybrids, their phylogeny is not concealed by potential mutations. Thus, floral scent composition is a highly conserved trait in Hieracium s.str. 19 Synopsis – References Dötterl, S., Jürgens, A. (2005): Spatial fragrance patterns in flowers of Silene latifolia: Lilac compounds as olfactory nectar guides? Plant Systematics and Evolution 255: 99–109. Dötterl, S., Füssel, U., Jürgens, A., Aas, G. (2005a): 1,4-Dimethoxybenzene, a floral scent compound in willows that attracts an oligolectic bee. Journal of Chemical Ecology 31: 2993–2998. Dötterl, S., Wolfe, L., W., Jürgens, A. (2005b): Qualitative and quantitative analyses of flower scent in Silene latifolia, Phytochemistry 66: 203–213. Düll, R. (1961): Die Sorbus-Arten und ihre Bastarde in Bayern und Thüringen. Berichte der Bayerischen Botanischen Gesellschaft 34: 11–65. Faegri, K., van der Pijl, L. (1979): The principles of pollination ecology, Oxford, New York, Toronto, Sydney, Paris, Frankfurt, Pergamon Press Ltd. Fehrer, J., Gemeinholzer, B., Chrtek, J. Jr., Bräutigam, S. (2007): Incongruent plastid and nuclear DNA phylogenies reveal ancient intergeneric hybridization in Pilosella hawkweeds (Hieracium, Cichorieae, Asteraceae). Molecular Phylogenetics and Evolution 42: 347–361. Fehrer, J., Krak, K., Chrtek, J. Jr. (2009): Intra-individual polymorphism in diploid and apomictic polyploid hawkweeds (Hieracium, Lactuceae, Asteraceae): disentangling phylogenetic signal, reticulation, and noise. BMC Evolutionary Biology 9: 239. Fenster, C.B., Armbruster, W.S., Wilson, P., Dudash, M.R., Thomson, J.D. (2004): Pollination syndromes and floral specialization. Annual Review of Ecology, Evolution and Systematics 35: 375–403. Feulner, M., Schuhwerk, F., Dötterl, S. (2009): Floral scent analysis in Hieracium subgenus Pilosella and its taxonomical implications. Flora 204: 495–505. Feulner, M., Schuhwerk, F., Dötterl, S. (2011): Taxonomical value of inflorescence scent in Hieracium s. str. Biochemical Systematics and Ecology 39: 732–743. Füssel, U., Dötterl, S., Jürgens, A., Aas, G. (2007): Interand intraspecific variation in floral scent in the genus Salix and its implication for pollination. Journal of Chemical Ecology 33: 749–765. Gadella, T.W.J. (1984): Cytology and the mode of reproduction of some taxa of Hieracium subgenus Pilosella. Proceedings of the Koninklijke Nederlandse Akademie van Wetenschappen 87: 387–399. 26 Synopsis – References Gancel, A.L., Ollé, D., Ollitrault, P., Luro, F., Brillouet, J.M. (2002): Leaf and peel volatile compounds of an interspecific citrus somatic hybrid [Citrus aurantifolia (Christm.) Swing. + Citrus paradisi Macfayden]. Flavour and Fragrance Journal 17: 416–424. Gögler, J., Stökl, J., Sramkova, A., Twele, R., Francke, W., Cozzolino, S., Cortis P., Scrugli, A., Ayasse, M. (2009): Ménage à trois - two endemic species of deceptive orchids and one pollinator species. Evolution 63: 2222–2234. Heuss, Lisa (2011): Bestäuber und Blütendüfte verschiedener Sorbus Taxa. Bachelor thesis am LS Pflanzensystematik Universität Bayreuth, unpublished. Hörandl, E., Hojsgaard, D. (2012): The evolution of apomixis in angiosperms: A reappraisal. Plant Biosystems 146: 681–693. Jankun, A., Kovanda, M. (1987): Embryological studies of Sorbus 2. apomixis and origin of Sorbus bohemica. Preslia 59: 97–116. Junker, R.R., Blüthgen, N. (2008): Floral scents repel potentially nectar-thieving ants. Evolutionary Ecology Research 10: 295–308. Junker, R.R., Loewel, C. Gross, R., Dötterl, S., Keller A., Blüthgen, N. (2011): Composition of epiphytic bacterial communities differs on petals and leaves. Plant Biology 13: 918–924. Kárpáti, Z. (1960): Die Sorbus-Arten Ungarns und der angrenzenden Gebiete. Feddes Repertorium 62: 71–334. Kessler, D.I., Baldwin, T. (2007): Making sense of nectar scents: the effects of nectar secondary metabolites on floral visitors of Nicotiana attenuate. The Plant Journal 49: 840–854. Knudsen, J.T., Tollsten L. (1993): Trends in floral scent chemistry in pollination syndromes: floral scent composition in moth-pollinated taxa. Botanical Journal of the Linnean Society 113: 263–284. Knudsen, J.T., Tollsten, L. (1995): Floral scent in bat-pollinated plants: a case of convergent evolution. Botanical Journal of the Linnean Society 119: 45–57. Koch, D.W.D.J. (1838): Synopsis der deutschen und Schweizer Flora., Wilmans, Frankfurt a. M. 840 p. Koltunow, A., Grossniklaus, U. (2003): Apomixis: A developmental perspective. Annual Review of Plant Biology 54: 547–574. Koltunow, A., Johnson, S.D., Okada, T. (2011): Apomixis in hawkweed: Mendel’s experimental nemesis. Journal of Experimental Botany: 1–9. 27 Synopsis – References Krahulcová, A., Krahulec, F., (2000): Offspring diversity in Hieracium subgen. Pilosella (Asteraceae): new cytotypes from hybridisation experiments and from open pollination. Fragmenta Floristica et Geobotanica Polonica 45: 239–255. Krahulcová, A., Krahulec, F., Chapman, H.M. (2000): Variation in Hieracium subgen. Pilosella: what do we know about its sources? Folia Geobotanica 35: 319–338. Kutzelnigg, H. (1995): Sorbus. In: Conert H.J. et al. (eds): Gustav Hegi, Illustrierte Flora von Mitteleuropa 4 (2B): Spermatophyta: Angiospermae: Dicotyledones 2 (3). Rosaceae 2. Blackwell, Berlin: 328–385. Kutzelnigg, H. (2005): Sorbus L. em. Crantz – Mehlbeere, Eberesche, Elsbeere, Speierling. In: Jäger, E.J., Rothmaler, K.W. (eds): Exkursionsflora von Deutschland 4, Gefäßpflanzen, Kritischer Band, 9th edn. Spektrum, Heidelberg: 403–406. Legendre, P., Makarenkov, V. (2002): Reconstruction of biogeographic and evolutionary netrworks using reticulograms. Systematic Biology 51: 199–216. Levin, R.A., McDade, L.A., Raguso, R.A. (2003): The systematic utility of floral and vegetative fragrance in two genera of Nyctaginaceae. Systematic Biology 52: 334– 351. Lepší, M., Vít, P., Lepší, P., Boublík, K., Kolář, F. (2009): Sorbus portae-bohemicae and Sorbus albensis, two new endemic apomictic species recognized based on a revision of Sorbus bohemica. Preslia 8: 63–89. Makarenkov, V. (2001): T-REX: Reconstructing and visualizing phylogenetic trees and reticulation networks. Bioinformatics 17: 664–668. Merxmüller, H. (1982): Hieracium schneidii – Ein unbekannter Bayerischer Endemit. Berichte der Bayerischen Botanischen Gesellschaft 53: 91–95. Meyer, N., Meierott, L., Schuwerk, H., Angerer, O. (2005): Beiträge zur Gattung Sorbus in Bayern. Berichte der Bayerischen Botanischen Gesellschaft Sonderband: 5–216. Mráz, P., Chrtek, J.J., Fehrer, J., Plačková, I. (2005): Rare recent natural hybridization in Hieracium s.str. - evidence from morphology, allozymes and chloroplast DNA. Plant Systematics and Evolution 255: 177–192. Nägeli, C.V., Peter, A., (1885): Die Hieracien Mittel-Europas. Monographische Bearbeitung der Piloselloiden mit besonderer Berücksichtigung der mitteleuropäischen Sippen. München. Nei, M. (1972): Genetic distance between populations. American Naturalist 106: 283–291. Nogler, G.A. (1984): Gametophytic apomixis. In: Johri, B M. (ed.): Embryology of angiosperms. Springer-Verlag, Berlin: 475–518. 28 Synopsis – References Plepys, D., Ibarra, F., Löfstedt, C. (2002): Odour-mediated nectar foraging in the silver Y moth, Autographa gamma (Lepidoptera: Noctuidae): behavioural and electrophysiological responses to floral volatiles. Oikos 99: 75–82. Pritchard, J.K., Stephens, M., Donnelly, P. (2000): Inference of population structure using Multilocus Genotype Data. Genetics 155: 945–959. Raguso, R.A. (2001): Floral scent, olfaction, and scent driven foraging behaviour. Cognitive Ecology of Pollination; Animal Behaviour and Floral Evolution (eds L. Chittka & J.D. Thomson). Cambridge University Press, Cambridge: 83–105. Raguso, R.A., Schlumpberger, B.O., Kaczorowski, R.L., Holtsford, T.P. (2006): Phylogenetic fragrance patterns in Nicotiana sections Alatae and Suaveolentes. Phytochemistry 67: 1931–1942. Rich, T., Houston, L., Robertson, A. (2010): Whitebeams, Rowans and Service Trees of Britain and Ireland A Monograph of British and Irish Sorbus L. BSBI Handbook 14, London. Robertson, A., Rich, T.C.G., Allen, A.M., Houston, L, Roberts, C., Bridle, J.R., Harris, S.A., Hiscock, S.J. (2010): Hybridization and polyploidy as drivers of continuing evolution and speciation in Sorbus. Molecular Ecology 19: 1675–1690. Rosenbaumová, R., Krahulcová, A., Krahulec, F. (2012): The intriguing complexity of parthenogenesis inheritance in Pilosella rubra (Asteraceae, Lactuceae). Sexual Plant Reproduction 25: 185–196. Schäffler, I., Balao, F., Dötterl, S. (2012): Floral and vegetative cues in oil-secreting and non-oil-secreting Lysimachia species. Annals of Botany 110: 125–138. Schiestl, F.P., Dötterl, S. (2012): The evolution of floral scent and olfactory preferences in pollinators: coevolution or pre-existing bias? Evolution 66: 2042–2055. Schuhwerk, F., (2002): Some thoughts on the taxonomy of Hieracium. Berichte der Bayerischen Botanischen Gesellschaft 72: 193–198. Schuhwerk, F. (2008): Zur Gliederung der Hieracium bifidum-Gruppe in Bayern und benachbarten Gebieten. – In: Dengler, J., Dolnik, C. & Trepel, M. (eds.): Flora, Vegetation und Naturschutz zwischen Schleswig-Holstein und Südamerika – Festschrift für Klaus Dierßen zum 60. Geburtstag. Mitteilungen der Arbeitsgemeinschaft Geobotanik für Schleswig-Holstein und Hamburg 65: 375– 392. Snow, R. (1963): Alcoholic hydrochloric acid-carmine as stain for chromosomes in squash preparations. Stain Technology 38: 9–13. 29 Synopsis – References Steiner, K.E., Kaiser, R., Dötterl, S. (2011): Strong phylogenetic effects on floral scent variation of oil-secreting orchids in South Africa. American Journal of Botany 98: 1663–1679. Talent, N. (2009): Evolution of gametophytic apomixis in flowering plants: an alternative model from Maloid Rosaceae. Theory in Biosciences 128: 121–138. Tichomirov, V.A. (2000): The genus Pilosella (Asteraceae) in the flora of belarus. Botanicheskii Zhurnal 5: 104–114. Van de Peer, Y., De Wachter, Y. (1994): TREECON for Windows: a software package for the construction and drawing of evolutionary trees for the Microsoft Windows environment. Computer Applications in the Biosciences 10: 569–70. Vereecken, N.J., Cozzolino, S., Schiestl, F.P. (2010). Hybrid floral scent novelty drives pollinator shift in sexually deceptive orchids. BMC Evolutionary Biology 10:103. Vos, P., Hogers, R., Bleeker, M. (1995): AFLP: a new technique for DNA fingerprinting. Nucleic Acids Research 23: 4407–4414. Whitehead, M.R., Peakall, R. (2009): Floral scent in a whole plant context. Integrating floral scent, pollination ecology and population genetics. Functional Ecology 23: 863–874. Yeh, F., Yang, R. (1999): POPGENE. Microsoft Window-based Freeware for Population Genetic Analysis. Version 1.31, University of Alberta: Edmonton, Alberta, Canada. Zahn, K. H. (1906): Die Hieracien der Schweiz. Neue Denkschriften der Allgemeinen Schweizerischen Gesellschaft für Gesammten Naturwissenschaften 40: 165–728, Zürich. Zahn, K. H. 1921–1923. Compositae – Hieracium. In: Engler, A. (ed.), Das Pflanzenreich. Regni vegetabilis conspectus. Verlag von Wilhelm Engelmann, pp. 1–1146. Zahn, K. H. 1930–1935. Hieracium 2. In: Ascherson, P. et al. (eds), Synopsis der mitteleuropäischen Flora, zwölfter Band, zweite Abteilung. Verlag von Gebrüder Borntraeger, pp. 1–790. 30 Publications Publication 1 1. Floral scent analysis in Hieracium subgenus Pilosella and its taxonomical implications. Published in Flora 2009 31 Author's personal copy Flora 204 (2009) 495–505 Floral scent analysis in Hieracium subgenus Pilosella and its taxonomical implications Martin Feulner a, , Franz Schuhwerk b , Stefan Do¨tterl a a Department of Plant Systematics, University of Bayreuth, Universita ¨tsstr 30, 95440 Bayreuth, Germany b Botanische Staatssammlung Mu ¨nchen, Menzinger Straße 67, D-80638 Mu ¨nchen Received 5 April 2008; accepted 30 June 2008 Abstract Species-rich Hieracium subgen. Pilosella is well-known for a high degree of endemism and infra-specific differentiation including many subspecies (‘‘microspecies’’) of very restricted distribution. In Hieracium subgen. Pilosella floral scents of 27 predominantly Bavarian species, mostly of Hieracium calodon, H. zizianum and H. densiflorum, are investigated here. Floral scent compositions were studied by GC-MS analysis of dynamic headspace samples. Altogether, 56 floral scent compounds were identified, mainly benzenoids, fatty acid derivatives, monoterpenes, homoterpenes and sesquiterpenes. The chemical patterns were found to be taxon-specific and are thus of taxonomical value. The data support some rearrangements at subspecific level, such as the inclusion of H. bauhini subsp. hispidissimum in H. densiflorum. These rearrangements are supported by morphological data. The traditional species concepts, however, are mostly corroborated by our scent data. r2008 Elsevier GmbH. All rights reserved. Keywords: Plant terpenes; Chemotaxonomy; Microspecies; Reticulation; Apomictic plants Introduction The main question of floral scent investigation regards interactions of flowering plants and their pollinators. Floral scent components are well-known for playing an important role as attractants for pollinators (e.g., Dobson et al., 2005;Do¨tterl et al., 2006;Plepys et al., 2002). Floral scent compounds can also function as repellent for herbivores or pathogens. However, Levin et al. (2003) point out that these compounds may not be functional in every case, because they are often by-products of the metabolism. Floral scents can be species-specific, helping to maintain species integrity via their effects for pollinator preselection. Differences among species in floral scent composition were detected in a large range of plant families (Knudsen et al., 2006). However, profiles of floral scent compounds were rarely used for phylogenetic analyses and taxonomy. An objection to the use of floral scent patterns in systematics is that convergence of floral scents may play an important role in plants in general, caused by similar pollinator pressures acting on plants of independent origin (Dobson et al., 2005). Nevertheless, comparative studies of floral scent and DNA data in orchids (Barkman, 2001;Williams and Whitten, 1998) as well as in Nyctaginaceae (Levin et al., 2003) revealed that the tree topologies generated by ARTICLE IN PRESS www.elsevier.de/flora 0367-2530/$ - see front matter r2008 Elsevier GmbH. All rights reserved. doi:10.1016/j.flora.2008.06.003  Corresponding author. E-mail addresses: [email protected] (M. Feulner), [email protected] (F. Schuhwerk), [email protected] (S. Do¨tterl). 32 Author's personal copy ARTICLE IN PRESS Table 1. Voucher and locality information for plant material used in this study. Taxon Voucher Number of samples per population H. lactucella Wallr. Neusta¨dtlein, 491580N, 111250; 15.7.2006, Feulner 1 (UBT) 2 H. pilosella L. Neusta¨dtlein, 491580N, 111250; 30.6.2004, Feulner 2 (UBT) 1 Oberailsfeld, 491490N, 111210O; 21.6.2005, Feulner 3 (UBT) 2 O ¨kol. Bot. Garten Bayreuth, 491560N, 111350O; 25.6.2004, Feulner 4 (UBT) 3 H. bauhini Schult. subsp. bauhini Bindlach, 501000N, 111370O; 12.6.2006, Breitfeld and Feulner 5 (UBT) 2 H. bauhini Schult. subsp. hispidissimum (Rehm.) Zahn Go¨ßweinstein, 491460N, 111190O; 10.6.2005, Feulner and Bolze 6 (Herbar Bolze 2) 3 H. piloselloides Zahn subsp. praealtum (Vill. ex Gochnat) Pottenstein, 491460N, 111250O; 25.6.2005, Feulner 7 (UBT) 2 Velburg, 491130N, 111400O; 23.6.2005, Feulner 8 (UBT) 1 H. cymosum L. subsp. cymosum Etterzhausen, 491020N, 111590O; 1.7.2006, Feulner 91 O ¨kol. Bot. Garten Bayreuth, 491560N, 111350O; 10.6.2005, Feulner 10 (UBT) 2 H. caespitosum Dumort. subsp. caespitosum O ¨kol. Bot. Garten Bayreuth, 491560N, 111350O; 15.6.2006, Feulner 11 (UBT) 2 Weiden, 491400N, 121100O; 1.7.2006, Feulner 12 (UBT) 1 Zeil, 501010N, 101350O; 25.6.2006, Feulner 13 (UBT) 1 H. echioides Lumn. subsp. echioides Brandenburg, Brodowin, 521550N, 131560O; 15.7.2006, Feulner 15 (UBT) 3 H. densiflorum Tausch subsp. psammotrophicum (Schack and Zahn) Altdorf, 491330N, 111350O; 10.6.2006, Feulner 16 (UBT) 2 H. densiflorum Tausch subsp. bauhinifolium (NP.) Velburg, 491130N, 111400O; 11.6.2005, Feulner and Meyer (Herbar Meyer 1) 2 Hieracium densiflorum Tausch subsp. umbelliferum (Na¨geli & Peter) Gottschl. Pottenstein, 491460N, 111250O; 1.7.2005, Bolze (Herbar Bolze 5) 3 H. densiflorum Tausch subsp. ochrocephaloides (Harz and Zahn) Lochau, 491580N, 111230O; 15.7.2006, Feulner 18 (UTB) 2 H. densiflorum Tausch subsp. cymosiforme (NP.) Velburg, 491130N, 111400O; 7.2005, Meyer99-13.3a (M) 2 H. calodon phyllophorum NP. Go¨ßweinstein, 491460N, 111190O; 1.7.2005, Feulner and Bolze (Herbar Bolze 1) 3 H. calodon Tausch ex Peter subsp. pseudofallax Touton Neudorf, 501040N, 111160O; 15.7.2006, Feulner 19 (UTB) 3 H. calodon ‘‘Ravensburg’’ Thu¨ngersheim, 491520N, 91500O; Schuhwerk 95/27 & Meierott (M) 3 H. schneidii Schack and Zahn Oberailsfeld, 491490N, 111210O; 8.6.2005, Feulner 21 (UTB) 3 Waischenfeld, 491500N, 111210O; 8.6.2005, Bolze (Herbar Bolze 4) 2 Zauppenberg, 491490N, 111230O; 9.6.2006, Bolze (Herbar Bolze 5) 1 H. fallax Froel. subsp. durisetum NP. Regensburg, 491010N, 121070O; 2005, Schuhwerk s. nr. (Merxmu ¨ller 33371) (M) 2 H. zizianum Tausch subsp. pachyphyes Zahn Etterzhausen, 491020N, 111590O; 1.7.2006, Feulner 24 (UTB) 2 H. zizianum Tausch subsp. zizianum Steifling, 491490N, 111240O; 29.5.2005, Feulner 25 (UTB) 2 Pegnitz-Bahnhof, 491450N, 111320O; 17.7.2006, Feulner 26 (UTB) 1 H. zizianum Tausch subsp. adenocymigerum Gerstl. and Zahn Kirchahorn, 491500N, 111230O; 15.7.2007, Feulner 28 (UTB) 3 H. spurium Chaix subsp. tubulatum (Vollm.) Zahn Weltenburg, 481530N, 111490O; 3.6.2006, Schuhwerk 87/36 and Lippert (M) 3 H. fallacinum F. W. Schultz subsp. fallacinum U ¨ttingen, 491470N, 91430O; 20.5.2005, Feulner and Meyer 29 (UTB) 4 H. fallacinum F. W. Schultz Steifling, 491490N, 111240O; 15.6.2005, Bolze and Feulner (Herbar Bolze 3) 3 H. schneidii x H. pilosella Oberailsfeld, 491490N, 111210O; 7.2005, Feulner 31 (UTB) 2 Pottenstein, 491460N, 111250O; 7.2005, Feulner 32 (UTB) 1 H. glomeratum Froel. subsp. glomeratum Hof ,501180N, 111540O; 2006, Feulner 33 (UTB) 1 Coburg, 501150N, 101570O; 2006, Feulner 34 (UTB) 1 H. aurantiacum L. subsp. aurantiacum Oberstdorf , 471240N, 101160O; 2006, Feulner 35, (UTB) 2 O ¨kol. Bot. Garten Bayreuth, 491560N, 111350O; 2005, Feulner 36 (UTB) 3 M. Feulner et al. / Flora 204 (2009) 495–505496 33 Author's personal copy non-coding DNA markers are in some cases congruent with those based on floral scent data. The methods to analyse scent patterns for taxonomical questions are manifold (Barkman, 2001;Levin et al., 2003). The simplest method is the use of presence/ absence data of different flower volatiles. For taxonomical questions, differentiation between compounds of floral and green parts of the plants seems to be negligible (cf. Levin et al., 2003). Considerations on the taxonomic value of floral fragrance analysis concentrated so far on outbreeding groups, while inbreeding groups and groups with reticulate relationships have not been considered. Reticulation in Scandinavian species of Hieracium subgen. Pilosella was proven by Tyler’s (2005) investigations using isoenzyme markers. He found no specific patterns of isoenzymes useful for species discrimination, and concluded that gene flow must be common among the species. Reticulation in Hieracium subgen. Pilosella comprises a high degree of putative hybrid species with intermediate characters often genetically isolated by apomixis. Further, different reproductive modes, such as allogamy, autogamy or apomixis can occur together in the same capitula (Krahulcova´ et al., 2000). Apomictic elements even can introgress as Krahulcova´ and Krahulec (2000) have found in artificial crossing experiments involving pentaploid apomictic Hieracium species. In this case, the apomictic pentaploids just serve as pollen donors, whereas the tetraand diploids act as recipients. A large number of species in Hieracium subgen. Pilosella are poorly characterized morphologically. Difficulties to identify the elements of the complex have stimulated Hieracium taxonomists to describe ‘‘collective species’’–binding together formal subspecies or microspecies–although many of these elements behave in nature as fixed apomictic species (cf. Schuhwerk, 2002). Na¨geli and Peter (1885) proposed the first taxonomic concept for Hieracium which is still in use in Central Europe, whereas in other parts of Europe (e.g., Scandinavia and Russia) alternative concepts are applied (cf. Schuhwerk, 2002). The concept of Na¨geli and Peter (1885) is based on the idea that there are some well distinguishable species, the so called ‘‘basic species’’, with unique morphological characters. However, most of the elements in Hieracium subgen. Pilosella show intermediate characters, therefore, they are treated as ‘‘intermediate species’’ by Na¨geli and Peter (1885). The authors considered some of them to be of hybrid origin. In this study, we investigate 27 taxa mostly of Bavarian origin (Table 1), representing about one third of the Bavarian species of subgen. Pilosella. In the taxa studied, both spontaneous hybrids growing together with at least one parental species only, and putative hybrids showing independent traits in morphology, ecology or distribution are included. Here, we focus especially on H. densiflorum, H. zizianum, H. fallax and H. calodon,all intermediates between the main species H. cymosum, H. echioides, H. piloselloides and H. bauhini. Some taxa, e.g., H. densiflorum subsp. cymosiforme, H. densiflorum subsp. bauhinifolium or H. bauhini subsp. hispidissimum display morphological characters questioning their membership in the described collective species. Furthermore these taxa are very rare or even endemic to Bavaria. Members of the ‘‘Echinina’’ species group (H. fallax, H. calodon and H. schneidii) are well-known glacial relicts (Merxmu¨ ller, 1982) and, therefore, of special interest. In Hieracium subgen. Pilosella many species are expected to be of hybrid origin. Against this background, it will be tested whether floral scent substances are useful markers for detection of hybrids and their origin. Material and methods Study plants This paper deals with taxa of Hieracium subgen. Pilosella in Germany with special attention to speciesrich Bavaria. All accessions, authors of taxa, voucher specimens and localities are given in Table 1. Taxa of a subspecies level were given when their state was confirmed by previous studies, i.e. Schuhwerk (2002); Schuhwerk and Lippert (1997, 2002) and Gottschlich (1996). Volatile collection Floral scent was collected in the field using the dynamic headspace method described by Do¨ tterl and Ju¨rgens (2005), and Do¨tterl et al. (2005). Capitula were enclosed within a polyester oven bag (Toppits s ) and the emitted volatiles were trapped in an adsorbent tube through the use of a membrane pump (ASF Thomas, Inc.) for eight minutes. As absorbent tube, we took ChromatoProbe quartz microvials of Varian Inc. (length: 15 mm; inner diameter: 2 mm), cut the closed end, filled them with a mixture (1:1) of 3 mg Tenax-TA (mesh 60–80) and Carbotrap (mesh 20–40), and fixed the adsorbent mixture in the vial with glass wool. Simultaneous collections of both the flower scent and surrounding air were used to distinguish between floral compounds and ambient contaminants. Sampling was carried out on fresh and newly opened capitula, between 11 a.m. and 3 p.m., the period with the most intensive scent emission (Feulner, unpublished data). In multi-headed synflorescences more than one open capitulum was sampled. ARTICLE IN PRESS M. Feulner et al. / Flora 204 (2009) 495–505 497 34 Author's personal copy Chemical analysis The samples were analysed using a Varian Saturn 2000 mass spectrometer, and a Varian 3800 gas chromatograph with a 1079 injector, that had been fitted with the ChromatoProbe kit. This kit allows the thermal desorption of small amounts of solids or liquids contained in quartz microvials (Micro-SPE; cf. Amirav and Dagan, 1997). The injector split vent is opened (1/20) to flush all air from the system and closed after 2 min; simultaneously, the injector is heated from 40 1C (temperature during the first 2 min) with a rate of 200 1C/min to 200 1C; this temperature is held for 4.2 min, after which the split vent opens (1/10) and the injector cools down. A ZB-5 column (5% phenyl polysiloxane) was used for the analyses (60 m long, inner diameter 0.25 mm, film thickness 0.25 mm, Phenomenex). Electronic flow control was used to maintain a constant helium carrier gas flow of 1.8 ml min 1 . The GC oven temperature was held for 7 min at 40 1C, then increased by 6 1C/min to 250 1C and held for 1 min. The MS interface was 260 1C and the ion trap worked at 175 1C. The mass spectra are taken at 70 eV (in EI mode), with a scanning speed of 1 scan s 1 from m/z 30 to 350. Data analysis The GC-MS data were processed using the Saturn Software package 5.2.1. Component identification was carried out using the NIST 02 mass spectral database, or MassFinder 2.3, and confirmed by the comparison of retention times with published data (Adams, 1995; Davies, 1990). Identification of individual components could be confirmed by the comparison of both mass spectrum and GC retention data with those of authentic standards. Statistical analysis For statistical analyses, a similarity matrix (So¨rensen similarities) was constructed using a presence/absence matrix of scent data. The significance of differences in scent profiles among taxa was assessed by ANOSIM (Clarke and Gorley, 2001), with 10,000 random permutations. For further analyses, the taxa samples (mostly 2–5 individuals per taxon) were merged by counting a single substance, if it was represented in at least half of the samples. To analyse the relationships among the taxa, we conducted a reticulation network analysis with the program t-rex Version 4.0a1 (Makarenkov, 2001). For analysis of groups with many species of hybrid origin, normal tree models are not suitable, because they cannot depict relationships of species interconnected with more than one ancestor (Legendre and Makarenkov, 2002). Therefore, these authors developed the software t-rex Version 4.0a1 (Makarenkov, 2001;Legendre and Makarenkov, 2002), which is applied for this study. The latter program is calculating a distance reduction between the preliminary joined arrangements by a special algorithm made visible by adding reticulation lines. The dashed lines are symbols for homoplasy in the data set (Makarenkov, 2001;Legendre and Makarenkov, 2002). T-rex cannot distinguish between the reasons of homoplasy, be it gene flow via hybridisation events or convergence. Results A total of 56 different floral scent compounds were found in the 27 investigated Hieracium taxa (Table 2). There are remarkable differences among the taxa concerning the number of scent components ranging from 19 components in H. pilosella to 31 in H. bauhini subsp. bauhini. The identified compounds belong to benzenoids, sesquiterpenes, monoterpenes and fatty acid derivatives. The most commonly occurring compounds were (Z)-3-Hexen-1-ol (found in all investigated species), D-Limonene, (E)-b-Ocimene, Ylangene (all found in 27 of 28 taxa), (Z)-Ocimene (26 taxa), (E)-4,8-Dimethyl1,3,7-nonatriene and Methylsalicylate (in 25 taxa). Most scent samples were dominated by (E)-b-Ocimene (average amount 20%), (Z)-3-Hexen-1-ol (15%) and (E)-4,8 Dimethyl-1,3,7-nonatriene (11%). The relative amount of (Z)-3-Hexen-1-ol was highest in H. zizianum subsp. pachyphyes with 67%; (E)-4,8-Dimethyl-1,3,7-nonatriene reached the highest relative amount in H. calodon x H. fallacinum (36%), and (E)-b-Ocimene in H. densiflorum subsp. psammotrophicum (49%). Members of H. densiflorum emitted a broader range of monoterpenes, and -Phellandrene, l-Fenchone and Linalool reached high relative amounts in these taxa. For H. cymosum subsp. cymosum or H. zizianum subsp. zizianum high values of sesquiterpenes as Copaene and Ylangene were found. In H. zizianum subsp. zizianum, H. piloselloides subsp. praealtum and H. fallacinum subsp. fallacinum higher amounts of acids (Hexanoic acid and Octanoic acid) were identified. (Table 2) The present-absent data of scent are highly specific in the investigated Hieracium taxa (ANOSIM R-value ¼0.769, po0.01), allowing the identification of taxa by floral scent data. Therefore, the samples of one taxon were be merged for the reticulation analysis. The result of the reticulation analysis is shown in Fig. 2. The tree consists of three main groups, in which the taxa are arranged in most cases as expected by morphology. However, some differences to the concept ARTICLE IN PRESS M. Feulner et al. / Flora 204 (2009) 495–505498 35 Author's personal copy Swing.+Citrus paradisi Macfayden]. Flavour Fragrance J. 17, 416–424. Gottschlich, G., 1996. Hieracium. In: Sebald, O., Seybold, S., Philippi, G., Wo¨rz, A. (Eds.), Die Farnund Blu¨tenpflanzen Baden-Wu¨rttembergs, vol. 6. Ulmer, Stuttgart, pp. 393–535. Ju¨rgens, A., Do¨tterl, S., Meve, U., 2006. The chemical nature of fetid floral odours in stapeliads (Apocynaceae–Asclepiadoideae–Ceropegieae). New Phytol. 172, 452–468. Knudsen, J.T., Eriksson, R., Gershenzon, J., St( ahl, B., 2006. Diversity and distribution of floral scent. Bot. Rev. 72, 1–120. Krahulcova´, A., Krahulec, F., 2000. Offspring diversity in Hieracium subgen. Pilosella (Asteraceae): new cytotypes from hybridisation experiments and from open pollination. Fragm. Flor. Geobot. 45, 239–255. Krahulcova´, A., Krahulec, F., Chapman, H.M., 2000. Variation in Hieracium subgen. Pilosella (Asteraceae): what do we know about its sources? Folia Geobot. 35, 319–338. Legendre, P., Makarenkov, V., 2002. Reconstruction of biogeographic and evolutionary networks using reticulograms. Syst. Biol. 5, 199–216. Levin, R.A., McDade, L.A., Raguso, R.A., 2003. The systematic utility of floral and vegetative fragrance in two genera of Nyctaginaceae. Syst. Biol. 52, 334–351. Makarenkov, V., 2001. T-REX: reconstructing and visualizing phylogenetic trees and reticulation networks. Bioinformatics 17 (7), 664–668. Merxmu¨ller, H., 1982. Hieracium schneidii—Ein unbekannter bayerischer Endemit. Ber. Bayer. Bot. Ges. 53, 91–95. Na¨geli, C.V., Peter, A., 1885. Die Hieracien Mittel-Europas. Monographische Bearbeitung der Piloselloiden mit besonderer Beru¨cksichtigung der mitteleuropa¨ischen Sippen, Mu¨nchen. Plepys, D., Ibarra, F., Lo¨fstedt, C., 2002. Volatiles from flowers of Platanthera bifolia (Orchidaceae) attractive to the silver Y moth, Autographa gamma (Lepidoptera: Noctuidae). Oikos 99, 69–74. Schuhwerk, F., 2002. Some thoughts on the taxonomy of Hieracium. Ber. Bayer. Bot. Ges. 72, 193–198. Schuhwerk, F., Lippert, W., 1997. Chromosomenzahlen von Hieracium (Compositae, Lactuceae) Teil 1. Sendtnera 4, 181–206. Schuhwerk, F., Lippert, W., 2002. Chromosomenzahlen von Hieracium (Compositae, Lactuceae) Teil 4. Sendtnera 8, 167–194. Tyler, T., 2005. Patterns of allozyme variation in Nordic H pilosella.. Plant Syst. Evol. 250, 133–145. Westrich, P., 1989. Die Wildbienen Baden-Wu¨rttemberg. Ulmer, Stuttgart, pp. 373–374. Williams, N.H., Whitten, W.M., 1998. Molecular phylogeny and floral fragrances of male euglossine bee-pollinated orchids: a study of Stanhopea (Orchidaceae). Plant. Spec. Biol. 14, 129–136. Zahn, K.H., 1921–1923. Compositae-Hieracium. Das Pflanzenreich 4 (280). W. Engelmann, Leipzig. Zahn, K.H., 1930–1935. Hieracium. Synopsis der mitteleuropa¨ischen Flora. Gebr. Borntra¨ger, Leipzig, 12 (1). ARTICLE IN PRESS M. Feulner et al. / Flora 204 (2009) 495–505 505 42 Publication 2 2. Taxonomical value of inflorescence scent in Hieracium s. str. Published in Biochemical Systematics and Evolution 2011 43 Author's personal copy Taxonomical value of inflorescence scent in Hieracium s. str. Martin Feulner a , * , Franz Schuhwerk b , Stefan Dötterl a a Department of Plant Systematics, University of Bayreuth, Universitätsstr. 30, 95440 Bayreuth, Germany b Botanische Staatssammlung München, Menzinger Straße 67, D-80638 München, Germany article info Article history: Received 7 January 2011 Accepted 25 June 2011 Available online 5 August 2011 Keywords: Inflorescence volatiles Reticulation analysis Asteraceae Apomictic species complex Ancient hybridization abstract In Central Europe Hieracium s. str. comprises a vast number of mostly apomictic taxa presumably originated from hybridization in the past. Inflorescence scents of 37 taxa from 7 sections of Hieracium subgenus Hieracium were investigated by headspace analysis. Overall, 58 different scent compounds belonging to benzenoids, sesquiterpenes, homoterpenes, monoterpenes and fatty acid derivatives were found. The scent patterns were used to perform a neighbour joining and reticulation analysis and the results are discussed against the background of current taxonomy. The scent clustering revealed a clear segregation between sections Drepanoidea, Tridentata and Hieracioides against members of the sections Hieracium, Oreadea and Bifida. The scent tree reflected distinct morphological and phenological groups in Hieracium and were congruent with actual genetic groupings. Actual section circumscriptions were supported with some exceptions concerning the sections Oreadea and Vulgata. Reticulation analyses of scent data reflected the hybrid status of intermediates, such as Hieracium franconicum,Hieracium caesium, as well as taxa of Hieracium wiesbaurianum and Hieracium glaucinum. The data also pointed towards a hybrid origin of members of some putative non-hybrid taxa such as Hieracium lachenalii, which is in accordance with recent molecular studies. The taxonomical usefulness of scent data in dominantly apomictic taxa is discussed. Ó2011 Elsevier Ltd. All rights reserved. 1. Introduction Investigations of floral scent deal mainly with questions concerning the interaction between flowers and pollinators, and many studies were conducted in order to identify substances attractive for pollinators (e.g. Plepys et al., 2002; Dötterl et al., 2006). In contrast to this field of research, there are only few studies dealing with the use of scent data for taxonomy (Williams and Whitten, 1998; Barkman, 2001; Levin et al., 2003). Feulner et al. (2009) demonstrated for the first time that such analyses are useful for investigating reticulate complexes as those of Hieracium subgen. Pilosella. In this species group, where recent hybridization occurs very often, inflorescence scent patterns were found to be taxon-specific and useful for taxonomic considerations (Feulner et al., 2009). Furthermore, parental taxa of hybrids could be identified on the basis of scent compounds (Feulner et al., 2009). Subgenus Hieracium is distributed in temperate areas of North America, Asia and Europe (Bräutigam, 1992), however, it is most diverse in Europe (Zahn, 1922–1938). It is a reticulate complex, but in contrast to Hieracium subgen. Pilosella, recent hybridization is rare and was only reported from Southern and Eastern Europe so far (Mráz et al., 2005; Mráz and Paule, 2006; Fehrer et al., 2007). In Central *Corresponding author. Tel.: þ49 921 55 2459; fax: þ49 921 55 2786. E-mail addresses: [email protected] (M. Feulner), [email protected]wn.de (F. Schuhwerk), [email protected] (S. Dötterl). Contents lists available at ScienceDirect Biochemical Systematics and Ecology journal homepage: www.elsevier.com/locate/biochemsyseco 0305-1978/$ –see front matter Ó2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.bse.2011.06.012 Biochemical Systematics and Ecology 39 (2011) 732–743 44 Author's personal copy Europe, most taxa are triploid apomicts and seem to have arisen from ancient hybridization (Schuhwerk, 2002; Fehrer et al., 2009). Here, we tested whether scent patterns are useful for taxonomic considerations in a presumable ancient hybrid complex using Hieracium s. str. as a model. In Middle Europe nearly all Hieracium taxa are polyploid apomicts (triand tetraploids), which produce seeds without pollination (Bräutigam and Greuter, 2007). In sexual plants pollinator mediated selection of floral scent appears to be important (Salzmann et al., 2007), such processes should become redundant in apomicts, where floral scents have no function in pollinator attraction and therefore may be under relaxed selection. Therefore it is likely that mutations could lead to a loss or gain of scent compounds, except those scent substances which have functions others than pollinator attraction (e.g. repellents against florivores, anti-pathogenes). Other floral features may also be under relaxed selection in Hieracium, and indeed, many apomicts are known to be male sterile and don’t produce pollen at all ( Storchová et al., 2002, own investigations). Such changes may especially be evident in taxa that are ancient hybrids and apomicts for many generations. Therefore, it may be more difficult to identify parental taxa in Hieracium s. str. by scent data than in groups such as Hieracium subgen. Pilosella where recent hybridization still occurs (comp. Feulner et al., 2009). Taxonomy in Hieracium s. str. is complicated. Many authors separate the two subgenera of Hieracium into two genera (Bräutigam and Greuter, 2007). The taxonomic concept of “basic species”und “intermediate species”is still used for Hieracium, which according to Zahn (1922–1938), helps in interpreting the large number of taxa with characters intermediate among two or more species. It further assigns the high number of subspecies or microspecies to so-called “collective species”. However, collective species are only a theoretical construction, because the microspecies or subspecies are the real taxa (Schuhwerk, 2002). Taxa investigated here belong to “basic”species of Hieracium such as H. schmidtii,H. bupleuroides,H. murorum and Hieracium bifidum and “intermediate”taxa such as H. franconicum,Hieracium glaucinum and H. wiesbaurianum. We investigated whether scent patterns reflect the current taxonomic concept and the interpretation of taxa as intermediate, and can the composition of volatiles contribute to understanding a putative hybrid origin. Furthermore, we ask whether scent similarities are in accordance with actual genetic investigations (Fehrer et al., 2009) and current section delimitation. 2. Method 2.1. Study plants All accessions, authors of taxa, voucher specimens and localities of the investigated Hieracium s. str. taxa are given in Table 1. Taxon identification follows Zahn (1922–1938). As additional information growth form and hair types of the involucrum (bristled, stellate and glandular) are given in Table 1.H. wiesbaurianum taxa of Thuringia were determined by Jochen Müller, Jena. The recent nomenclature changes in this subgenus (Greuter, 2007) were not applied, since not all taxa used in this study were published following the new nomenclature. Section circumscription (Table 2) follows Gottschlich (2009), for taxa not included in the latter study, we used the concept of Sell and West (1976). 2.2. Volatile collection Inflorescence scent was collected in the field using a standard dynamic headspace method as described in Feulner et al. (2009). For each taxon two to six individuals were collected. Sampling was carried out on fresh and newly opened capitula (one capitulum per plant and sample), between 11 a.m. and 3 p.m, the period with the most intensive scent emission (Feulner, unpublished data). 2.3. Chemical analysis The samples were analysed on a Varian Saturn 2000 mass spectrometer, and a Varian 3800 gas chromatograph with a 1079 injector, that had been fitted with the ChromatoProbe kit. This kit allows the thermal desorption of small amounts of solids or liquids contained in quartz microvials (Micro-SPE; cf. Amirav and Dagan, 1997). The injector split vent was opened (1/20) to flush any air from the system and closed after 2 min; the injector was heated with 40  C for 2 min, and temperature was then increased with a rate of 200  C/min to 200  C; this temperature was held for 4.2 min, after which the split vent opened (1/10) and the injector cooled down. A ZB-5 column (5% phenyl polysiloxane) was used for the analyses (60 m long, inner diameter 0.25 mm, film thickness 0.25 m m, Phenomenex). Electronic flow control was used to maintain a constant helium carrier gas flow of 1.8 ml min 1 . The GC oven temperature was held for 7 min at 40  C, then increased by 6  C per min to 250  C and held for 1 min. The MS interface was 260  C and the ion trap worked at 175  C. The mass spectra are taken at 70 eV (in EI mode) with a scanning speed of 1 scan s 1 from m/z30 to 350. M. Feulner et al. / Biochemical Systematics and Ecology 39 (2011) 732–743 733 45 Author's personal copy Table 1 Voucher and locality information for plant material used in this study. All accessions originate from Germany. Additional information about growth form and hair cover of the involucrum is given. Taxon Species code Voucher Number of scent samples per population Growth form and flowering time Hair types of involucrum H. bupleuroides C.C. Gmel. ssp. bupleuroides bb Bavaria, Gößweinstein, 49  46 0 N, 11  20 0 E, 30.7.2008, Feulner 95 (UBT) 2 tall-growing and late flowering stellate, slightly bristled Streitberg, 49  48 0 N, 11  11 0 E, 30.7.2008, Feulner 40 (UBT) 2 H. glaucum ssp. isaricum (Nägeli ex J. Hofm.) Nägeli et Peter gli Bavaria, Ruhpolding, Fischbachtal, 47  42 0 N, 12  39 0 E, 10.8.2007, Feulner 41 (UBT) 3 tall-growing and late flowering stellate H. franconicum (Griseb.) Zahn fw Bavaria, Forchheim, Walberla, 49  42 0 N, 11  09 0 E, 30.7.2008, Feulner 101 (UBT) 3 tall-growing and late flowering stellate, bristled and glandular fh Eggloffstein, Hardt, 49  42 0 N, 11  15 0 E, 30.7.2008, Feulner 42 (UBT), 2 fb Beuren, 48  32 0 N, 9  15 0 E, 20.7.2008, Feulner 43 (UBT) 3 H. harzianum Zahn h Bavaria, Forchheim, Walberla, 49  42 0 N, 11  09 0 E, 30.7.2008, Feulner 43 (UBT) 3 tall-growing and late flowering stellate, bristled and glandular H. schmidtii ssp. kalmutinum (Zahn) Gottschlich sk Bavaria, Karlstadt, Kalbenstein, 49  58 0 N, 9  46 0 E, 29.5.2008, Schuhwerk 86/181 (M). 3 low-growing and early flowering stellate, bristled and sparsely glandular H. schmidtii ssp. comatulum (Jord. ex Boreau) Gottschlich sc Bavaria, Naila, Höllental, 50  20 0 N, 11  41 0 E, 21.6.2007, Feulner 45 (UBT) 4 low-growing and early flowering slightly stellate, bristled and glandular H. glaucinum ssp. medium (Jord.) O. Bolòs & Vigo gm Bavaria, Karlstadt, Kalbenstein, 49  58 0 N, 9  46 0 E, 29.5.2008, Feulner 46 (UBT) 2 low-growing and early flowering glandular H. glaucinum ssp. similatum (Jord. ex Boreau) Zahn gs Bavaria, Knetzgau, 49  59 0 N, 10  37 0 E, 29.5.2008, Feulner 100 (UBT) 1 low-growing and early flowering bristled and glandular Üttingen, Steinbruch, 49  46 0 N, 9  43 0 E, 29.5.2008, Feulner 47 (UBT) 1 H. glaucinum ssp. oegocladum (Jord. ex Boreau) Soó go Bavaria, Schmidmühlen, Aichaer Berg 49  16 0 N, 11  56 0 E, 5. 6. 2009, Feulner 72 2 low-growing and early flowering glandular H. glaucinum Jord. gpo Bavaria, Pottenstein, Felsenbad 49  46 0 N, 11  24 0 E, Feulner 48 (UBT) 2 low-growing and early flowering bristled and glandular H. glaucinum grex cinerascens (Jord.) Zahn gci dol Bavaria, Döhlau, 49  57 0 N, 11  39 0 E, 20.5.2008, Feulner 49 (UBT) 2 low-growing and early flowering glandular H. glaucinum grex cinerascens (Jord.) Zahn gci mun Bavaria, Münnerstadt, 50  15 0 N, 10  11 0 E, 15.5.2008, Feulner 50 (UBT) 2 low-growing and early flowering glandular Hieracium glaucinum ssp. prasiophaeum (Arv.-Touv. et Gautier) Greuter gp Bavaria, Forchheim, Ehrenbürg, 49  42 0 N, 11  09 0 E, Feulner 73 (UBT) 3 low-growing and early flowering sparsely bristled and glandular Weismain, Neudorf, 50  34 0 N, 11  15 0 E, Feulner 77 (UBT) 3 H. onosmoides Fr. o Bavaria, Karlstadt, Kalbenstein, 49  58 0 N, 9  46 0 E, 15.5.2008, Feulner 51 (UBT) 2 tall-growing and late flowering bristled, sparsely glandular and stellate H. wiesbaurianum ssp. semicinerascens Bornm. et Zahn ws Bavaria, Münnerstadt, 50  15 0 N, 10  11 0 E, 29.5.2008, Feulner 52 (UBT) 2 low-growing and early flowering stellate, bristled and glandular Hieracium [wiesbaurianum] parvimaculatum Jochen Müll. wp Thuringia, Jena, Haselberg, 50  55 0 N, 11  32 0 E, 5.6.2008, Feulner 53 (UBT) 2 low-growing and early flowering stellate, bristled and glandular H. [wiesbaurianum ssp. jenzigense var.] euwiesbaurianiforme (Schack et Zahn) Jochen Müll. we Thuringia, Jena, Haselberg, 50  55 0 N, 11  32 0 E, 5.6.2008, Feulner 54 (UBT) 2 low-growing and early flowering stellate, bristled and glandular M. Feulner et al. / Biochemical Systematics and Ecology 39 (2011) 732–743734 46 Author's personal copy Table 1 (continued ) Taxon Species code Voucher Number of scent samples per population Growth form and flowering time Hair types of involucrum H. wiesbaurianum ssp. jenzigense Bornm. et Zahn wj Thuringia, Jena, Haselberg, 50  55 0 N, 11  32 0 E, 5.6.2008, Feulner 55 (UBT) 2 low-growing and early flowering stellate, bristled and glandular H. wiesbaurianum ssp. niphanthodes Bornm. et Zahn wn Thuringia, Schirnewitz, 50  51 0 N, 11  32 0 E, 5.6.2008, Feulner 56 (UBT) 2 low-growing and early flowering stellate, bristled and glandular H. [sommerfeltii] crinicaesium (Schack et Zahn) Jochen Müll. wc Thuringia, Schirnewitz, 50  51 0 N, 11  32 0 E, 5.6.2008, Feulner 57 (UBT) 2 low-growing and early flowering stellate, bristled and glandular H. wiesbaurianum ssp. apertorum Bornm. et Schack ex Zahn wap Thuringia, Leutratal, 50  52 0 N, 11  34 0 E, 5.6.2008, Müller 9794 2 low-growing and early flowering stellate, bristled and glandular H. wiesbaurianum ssp. arnoldianum Zahn war Bavaria, Eichstätt, Steinbruch, 48  54 0 N, 11  10 0 E, 19.7.2008, in cult. Meyer, 20.7.2008 1 low-growing and early flowering stellate, bristled and glandular Altdorf Titting, 48  59 0 N, 11  17 0 E, in cult. Meyer, 20.7.2008 1 Mühlheim Gailachtal, 48  51 0 N, 10  59 0 E, in cult. Meyer, 20.7.2008 1 H. wiesbaurianum s l. Uechtr. ex Baenitz wsch Bavaria, Arnsberg, Schambachtal, 48  54 0 N, 11  22 0 E, 5.6.2009, in cult. Meyer 2 low-growing and early flowering stellate, bristled and glandular Hieracium bifidum ssp.stenolepis var.valdefloccosum (Vollm.) Zahn bsv Bavaria, Heitzenhofen, Öder Grainberg, 49  08 0 N, 11  56 0 E, 5.6.2009, Feulner 74 (UBT) 1 low-growing and early flowering stellate, bristled and sporadic glandular Deuerling, 49  02 0 ,11  54 0 , 5.6.09, Schuhwerk 09/100 2 H. bifidum Kit. ex Hornem.grex bifidum bbf Bavaria, Ruhpolding, Fischbachtal, 47  41 0 N, 12  39 0 E, 10.8.2007, Feulner 61 (UBT) 2 low-growing and early flowering stellate, bristled H. bifidum Kit. ex Hornem.grex bifidum bbk Bavaria, Krögelstein, 49  58 0 N, 11  16 0 E, 6.6.2007, Feulner 62 (UBT) 2 low-growing and early flowering stellate, bristled H. bifidum ssp. basicuneatum Zahn bsc Bavaria, Forchheim, Ehrenbürg, 49  42 0 N, 11  09 0 E, Feulner 63 (UBT) 3 low-growing and early flowering stellate, bristled and glandular H. murorum ssp. silvularum (Jord.) Zahn msi Bavaria, Betzenstein, Burgruine 49  41 0 N, 11  25 0 E, 20.6.2008, Feulner 64 (UBT) 2 low-growing and early flowering glandular H. murorum L. mk Bavaria, Kupferberg, 50  08 0 N, 11  35 0 E, 15.6.2007, Feulner 65 (UBT) 2 low-growing and early flowering glandular H. murorum L. mw Bavaria, Forchheim, Ehrenbürg, 49  42 0 N, 11  09 0 E, Feulner 80 (UBT) 2 low-growing and early flowering glandular H. murorum L. mne Bavaria, Neustädtlein, Horlache, 49  58 0 N, 11  25 0 E, 10.6.2008, Feulner 66 (UBT) 2 low-growing and early flowering glandular H. murorum L. mb Bavaria, Neudorf, Bärental, 50  34 0 N, 11  15 0 E, 20.6.2006, Feulner 90 (UBT) 1 low-growing and early flowering glandular H. saxifragum Fr. ssp. dufftii Zahn sd Bavaria, Steinbruch Guttenberg, 50  09 0 N, 11  34 0 E, 15.6.2007, Feulner 67 (UBT) 2 tall-growing and late flowering slightly bristled and glandular Hieracium caesium Fr. ca Bavaria, Gräfenberg, Almos 49  40 0 N, 11  21 0 E, 20.6.2008, 15. 6. 1996, Wagenknecht s. n.(M) 2 tall-growing and late flowering stellate, bristled, slightly glandular Hieracium caesium Fr. ssp. caesium cc Bavaria, Essing, 48  56 0 N, 11  47 0 E, 20.8.2008, Feulner 68 (UBT) 2 tall-growing and late flowering stellate, bristled (continued on next page) M. Feulner et al. / Biochemical Systematics and Ecology 39 (2011) 732–743 735 47 Author's personal copy 2.4. Data analysis The GC–MS data were processed using the Saturn Software package 5.2.1. Component identification was carried out using the NIST 08 mass spectral data base, or MassFinder 3, and confirmed by comparison of retention times with published data (Adams, 2007). Identification of individual components was confirmed by comparison of both mass spectrum and GC retention data with those of authentic standards. 2.5. Statistical analysis A similarity matrix (Sörensen similarities) was constructed based on the presence/absence of compounds. The significance of differences in scent profiles among taxa was assessed by ANOSIM with 10,000 random permutations using Primer (Clarke and Gorley, 2001). For further analyses, the samples of a specificspecies(2–6 individuals per taxon) were merged. A single substance was treated as present in a taxon if it occurred in at least half of the individual samples. To analyse the relationships among the taxa we conducted a reticulation network analysis with the program t-rex, version 4.0a1 (Makarenkov, 2001). This method allows to visualize relationships of species interconnected with more than one ancestor (Legendre and Makarenkov, 2002), which is important for analysing groups, such as Hieracium, with many taxa of hybrid origin (see Feulner et al., 2009). In this approach, a neighbour joining tree was constructed using a dissimilarity matrix (1-Jaccard), and homoplasies are made visible by so-called reticulation lines. Those homoplasies point towards hybridization or introgression (Legendre and Makarenkov, 2002). In addition to the presence and absence of compounds, we also calculated the average relative (percentage of total) amount of scent compounds of the single taxa (see Table A.1). Table 2 Sections of Hieracium s. str., according to Gottschlich (2009) and Sell and West (1976) for the investigated species. Basic species printed in bold. For intermediate species the formula of the taxa to which they are morphologically intermediate (comp. Zahn,1922–1938) are given in brackets. Section taxa Hieracium sensu Gottschlich (2009) H. murorum Oreadea (Fr.) Arv.-Touv. sensu Gottschlich (2009) H. schmidtii H. glaucinum (schmidtii-murorum) Oreadea (Fr.) Arv.-Touv. sensu Sell and West (1976) H. saxifragum (lachenalii >schmidtii) H. onosmoides (lachenalii <schmidtii) Bifida (Arv.-Touv.) Clapham sensu Gottschlich (2009) H. bifidum H. wiesbaurianum (bifidum-schmidtii/glaucinum) Vulgata (Griseb.) Willk. & Lange sensu Sell and West (1976) H. lachenalii H. caesium (lachenalii-bifidum) Drepanoidea Monnier sensu Sell and West (1976) H. glaucum H. bupleuroides H. franconicum (bupleuroides-murorum) H. harzianum (laevigatum-franconicum) Hieracioides Dumort. sensu Sell and West (1976) H. umbellatum Tridentata (Fr.) Arv.-Touv. sensu Sell and West (1976) H. laevigatum Table 1 (continued ) Taxon Species code Voucher Number of scent samples per population Growth form and flowering time Hair types of involucrum (continued on next page) Hieracium lachenalii C. C. Gmel. l Bavaria, Neustädtlein, Horlache 49  58 0 N, 11  25 0 E, 20.6.2007, Feulner 69 (UBT) 2 tall-growing and late flowering glandular Mistelbach, Buchstein, 49  55 0 N, 11  32 0 E, 12.6.08, Feulner 70 (UBT) 2 Naila Höllental 50  20 0 N, 11  41 0 E, 9.6.08, Feulner 71 (UBT) 2 Hieracium laevigatum Willd. lae Bavaria, Neustädtlein, Horlache, 49  58 0 N, 11  25 0 E, 14.7.2008, Feulner 72 (UBT) 2 tall-growing and late flowering slightly short glandular Rehauer Forst, 50  16 0 N, 12  40 0 E, Feulner 73 (UBT) 1 Hieracium umbellatum L. u Bavaria, Neustädtlein, Horlache 49  58 0 N, 11  25 0 E, 20.8.2007, Feulner 78 (UBT) 2 tall-growing and late flowering glabrous M. Feulner et al. / Biochemical Systematics and Ecology 39 (2011) 732–743736 48 Author's personal copy 3. Results 3.1. Scent compounds and patterns A total of 58 different inflorescence scent compounds were present in the Hieracium taxa (Table A.1). The compounds belong to benzenoids, sesquiterpenes, homoterpenes, monoterpenes and fatty acid derivatives. Most numerous substances were sesquiterpenes with nearly half of the number of identified substances. There were remarkable differences among the taxa concerning the number of scent components, ranging from 22 components in H. glaucinum ssp. prasiophaeum to 40 in Hieracium umbellatum. The scent patterns (presence/absence of compounds) were highly taxa specific (ANOSIM: R¼0.86; p¼0.01), allowing the identification of taxa by scent data. Low-growing taxa with zero to one stem leaf (except H. lachenalii, H. onosmoides and H. caesium) and an earlier flowering time Tall-growing taxa with two to many stem leaves and a later flowering time schmidtii kalmutinum glaucinum cinerascens gci mun wiesbaurianum semicinerascens wiesbaurianum apertorum glaucinum similatum wiesbaurianum jenzigense crinicaesium wiesb. s. l. glaucinum cinerascens gci dol glaucinum oegocladum onosmoides glaucinum medium wiesbaurianum niphanthodes wiesbaurianum arnoldianum bifidum stenolepis valdefloccosum glaucinum gpo wiesbaurianum wsch murorum silvularum murorum mk murorum mb lachenalii glaucinum prasiophaeum murorum mw murorum mne parvimaculatum wiesb. s. l. euwiesbaurianiforme wiesb. s. l. schmidtii comatulum bifidum bifidum bbf bifidum bifidum bbk caesium ca bifidum basicuneatum caesium caesium saxifragum dufftii umbellatum harzianum laevigatum franconicum fw franconicum fh franconicum fb glaucum isaricum bupleuroides Fig. 1. Neighbour joining tree (1-Jaccard) with reticulations added to the basic additive tree limited to 17, number of objects n¼40 (by t-rex V. 4.1, Makarenkov, 2001). Basic species are printed in bold. M. Feulner et al. / Biochemical Systematics and Ecology 39 (2011) 732–743 737 49 Author's personal copy The most abundant and commonly occurring compounds were (Z)-3-hexen-1-yl acetate (average relative amount 27%), (Z)-3-hexen-1-ol (19%), methyl salicylate (10%), (E)-4,8-dimethyl-1,3,7-nonatriene (6%), (E)- b -ocimene (4.6%), a -copaene (4.4%), 2-phenylethyl alcohol (3.4%), linalool, limonene, and phenylacetaldehyde (Table A.1). 3.2. Grouping of taxa in the neighbour joining tree on the basis of scent similarity In the neighbour joining tree (Fig. 1) the investigated 37 taxa clustered in four mainand several subgroups, based on the presence and absence of compounds. Overall, scent data were –despite some exceptions (see below) –congruent with the growth form and flowering time (tall growth, many stem leaves and later flowering time versus low growth, zero to one stem leaf and earlier flowering time, see Table 1) of the taxa studied. We did not find exclusive compounds for the tall-growing or for the low-growing taxa group as a whole. However, (Z)-Linalooloxid furanoid occurred in most of the tall-growing taxa and only in a few of the other morphotypes. p-Methylanisole occurred only in the tall-growing taxa of section Drepanoidea (H. franconicum, H. bupleuroides and H. glaucum ssp. isaricum; comp. Table A.1). In the majority of the low-growing taxa terpinolene was present, but missing in most of the tall-growing morphotypes. The tall-growing group comprised members of four sections, among them Drepanoidea (e.g., H. bupleuroides,H. franconicum), Tridentata and Hieracioides (see Fig. 1,Table 2). A well segregated second group was built by low-growing taxa (e.g., H. bifidum,H. schmidtii, H. murorum,H. glaucinum and H. wiesbaurianum), inwhich members of H. bifidum and Hieracium caesium were sister to the remaining taxa. However, this second group also contains a few high-growing morphotypes, i.e., Hieacium lachenalii, H. caesium,andH. onosmoides (see Fig. 1). H. lachenalii clustered with H. murorum, whereas the putative derivatives of H. lachenalii (i.e., H. onosmoides,H. caesium) clustered with H. glaucinum and H. bifidum, respectively. 3.3. Identification of scent homoplasies and reticulate relationships of taxa Reticulations linked in most cases morphologically intermediate hybrid taxa with their probable parents even if they did not built a group in the neighbour joining tree. As an example, H. lachenalii groups together with H. murorum, but is connected by a reticulation line to Hieracium laevigatum. Indeed, H. lachenalii shares its involucrum-indument (dense glandular hairs) with H. murorum while its taller growth form resembles H. laevigatum (see Table 1). H. caesium ca (see Table 1) clustered with H. bifidum but a reticulation line linked it to H. lachenalii, indicating scent homoplasies in accordance with its morphological intermediary between H. bifidum and H. lachenalii. It possesses stellate hairs on the involucrum like H. bifidum, but two stem leaves, as it is often found also in H. lachenalii (Table 1). The putative origin of the endemic species H. franconicum by hybridization between H. bupleuroides and H. murorum (see Table 2) was partly confirmed by our analysis, because a reticulation line linked the group with one taxon of H. murorum but additionally also with some members of the H. bifidum/H. caesium group. Taxa of a specific section (e.g. Bifida, Oreadea or Hieracia) did not cluster together in all cases, instead, sectional subgroups occurred in different places of the tree (Fig. 1). However, if reticulations between taxa (e.g. between H.wiesbaurianum and H. bifidum or H. glaucinum and H. murorum) are taken into consideration, scent results fit better with the section concept according to Sell and West (1976) and Gottschlich (2009). 4. Discussion 4.1. Subgenus specific components In Hieracium subgen. Hieracium many components were identified which were already found in our previous study of subgenus Pilosella (Feulner et al., 2009). Substances such as (Z)-3-hexen-1-yl acetate or (Z)-3-hexen-1-ol dominate the scent in both subgenera (conf. Feulner et al., 2009), monoterpenes, such as fenchone could be found only in subgenus Pilosella (see Feulner et al., 2009). Substances that could be found so far only in subgenus Hieracium are among others p-methylanisole, carvone, pyroids and furanoids of linalool oxid and g -terpinene. Some substances are much more widespread in Hieracium s. str. than in H. subgen. Pilosella (e.g. benzeneacetaldehyde, linalool). 4.2. Conformity between scent grouping and current taxonomy Members of the section Drepanoidea cluster closely together (Fig. 1) which is in strong accordance with current sectional classification (cf. Zahn, 1922–1938, Stace, 1998; Gottschlich, 2009) as well as molecular investigations (Fehrer et al., 2009). Only the intermediate H. harzianum clustered closer to its putative second parental taxon H. laevigatum. The close grouping of the low-growing taxa, including putative derivatives of H. lachenalii, is in accordance with high morphological and genetic similarity among these types (Fehrer et al., 2009). Furthermore, the weak correlation of scent groups and sectional groups within the low-growing taxa points towards a complex and presumably polyphyletic evolution. Nevertheless, within this group, scent data revealed a clear segregation of H. murorum against members of H. bifidum. Therefore, our data confirm the actual taxonomical treatment of Gottschlich (2009), assigning H. bifidum and H. murorum to different sections, and not combining it in one, as proposed by Sell and West (1976). Scent data confirm the result of Fehrer M. Feulner et al. / Biochemical Systematics and Ecology 39 (2011) 732–743738 50 Author's personal copy et al. (2009), suggesting that H. lachenalii is not a true “basic”species but a hybrid between a member of the “Western clade” (H. murorum,H. bifidum, etc.) and the H. umbellatum group (H. umbellatum,H. laevigatum and others). Similarly, our data point towards a hybrid origin of H. lachenalii with participation of H. murorum and –fitting with the genomic data of Fehrer et al. (2009) - the H. umbellatum/H. laevigatum group (Fig.1). From scent data, H. lachenalii could not retrieved as parent to taxa like H. onosmoides, H. saxifragum ssp. dufftii or H. caesium ssp. caesium, which had previously been considered as intermediates of H. lachenalii.H. saxifragum ssp. dufftii according to scent data has a closer relationship with H. umbellatum/H. laevigatum, which is supported by its relatively poor-hairy involucrum cover, corresponding more to H. laevigatum than to H. lachenalii. Other examples for the conformity between scent-grouping and molecular sequence data are the vicinity of H. bupleuroides/H. glaucum and H. umbellatum/H. laevigatum, all belonging to the “Eastern clade”sensu Fehrer et al. (2009) as well as the close grouping of H. schmidtii, H. murorum and H. bifidum, which all are members of the “Western clade”(Fehrer et al., 2009). Overall, the conformities of our results with results based on genetic analyses give strong support for the taxonomic reliability of scent data in Hieracium. The distant placement of H. schmidtii ssp. comatulum and ssp. kalmutinum indicates divergent evolution of both subspecies (see Fig. 1), in accordance with their considerably deviating morphology. H. schmidtii ssp. kalmutinum, which clusters together with H. wiesbaurianum and H. glaucinum taxa, has indeed more common morphological features with H. wiesbaurianum taxa (e.g. presence of bristled hairs on the upper leaf side and densely stellate hairs on the involucrum) than with H. schmidtii ssp. comatulum (e.g. glabrous upper leaf side and slightly stellate hairs on the involucrum), making this clustering reliable. Furthermore, the two H. schmidtii subspecies also behave strongly different according their ecology: H. schmidtii ssp. kalmutinum grows on calcareous and H. schmidtii ssp. comatulum only on acidic soil (e. g. Schuhwerk, 1990). As a consequence H. schmidtii ssp. kalmutinum will be keyed out as H. wiesbaurianum by Bräutigam and Schuhwerk (in press). All investigated H. glaucinum taxa which are morphologically either closer to H. murorum (e.g. ssp. oegocladum, ssp. medium)ortoH. schmidtii (ssp. prasiophaeum, ssp. similatum) group next to H. murorum and H. schmidtii ssp. comatulum. Therefore, scent clustering does not favour any sectional delimitation, neither that of Sell and West (1976), uniting H. murorum and H. glaucinum in section Hieracium, nor that of Gottschlich (2009),affiliating it to sect. Oreadea (Table 2). H. glaucinum and most H. wiesbaurianum taxa are close neighbours or cluster partly intermingled. This might be due to their common parent, H. schmidtii (Table 2). Some authors (e.g. Zahn,1922–1938), however, discuss H. glaucinum instead of H. schmidtii as parental taxon of H. wiesbaurianum. Both ideas are supported by our data, since some of the H. wiesbaurianum taxa (e.g. H. parvimaculatum) neighbour H. schmidtii ssp. comatulum, others (e.g., H. wiesbaurianum ssp. semicinerascens, ssp. niphanthodes)H. glaucinum. One obvious example for the latter scenario is H. wiesbaurianum ssp. semicinerascens which neighbours H. glaucinum ssp. cinerascens mun (Fig. 1). Both taxa share unique morphological features, such as the multiheaded inflorescences and the stem leaf morphology. The other putative parent of H. wiesbaurianum is H. bifidum (Table 2). With exception of H. bifidum ssp. stenolepis var. valdefloccosum (bsv), both taxa groups are well separated by scent: H. wiesbaurianum taxa and bsv emit (E)- b -ocimene, which is absent from the other investigated H. bifidum taxa, whereas benzeneacetaldehyde is only emitted by H. bifidum ssp.bifidum and ssp. basicuneaum (Table A.1). Interestingly, also in AFLP analyses (Reisch/Meyer, ined.) H. bifidum ssp. stenolepis var. valdefloccosum groups closer to taxa of H. wiesbaurianum than to taxa of H. bifidum. In conclusion, our data confirm the hybrid origin of morphologically intermediate taxa such as H. glaucinum, H. wiesbaurianum, H. franconicum and also suggest a hybrid origin of H. lachenalii so far considered as ”basic”taxon. One common assumption is that many Hieracium s. str. taxa originated during and shortly after the last ice age, a time when hybridization between sexual “basic”species appeared to be frequent (Zahn, 1922–1938, Fehrer et al., 2009). Most of the species studied are apomicts in which inflorescence/flower scent does not longer have a function as pollinator attractant. Therefore, scent profiles may be faced with mutational chances that do not underlay conserving pollinator mediated selection (see chapter 1). However, scent mutations did not seem to occur in such rates that they conceal the tracks of phylogeny. Otherwise it would not have been possible to identify intermediates by scent or to find a correlation with taxonomy. Therefore, scent profiles as well as other floral features (e.g. inflorescence morphology) are rather conserved in apomictic Hieracium. This may reflect the relatively young age of many apomictic taxa. Also Fehrer et al. (2009) concluded that, because of an extremely low level of ITS variation, most species evolved in the Quaternary, a time of rapid speciation in Hieracium s. str. Another reason for scent conservation may be that a large range of substances may have functions others than pollinator attraction (e.g. repellents against florivores, anti-pathogenes; comp. Pichersky and Gershenzon, 2002). Nevertheless, there are some rare reports about recent gene flow in putative apomictic Hieracium (see Chapman et al., 2004; Tyler and Jönsson, 2009), which can only be mediated by insect pollinators. In such cases, pollinator mediated selection of inflorescence scent may still be of some importance. Acknowledgement We thank Prof. Sigrid Liede-Schumann for supporting this study. We thank Dr. Jochen Müller for determining the H. wiesbaurianum taxa of Thuringia. Norbert Meyer and Dr. Alfred Bolze helped identifying taxa in the field. Dr. Ulrich Meve gave valuable comments on the manuscript. M. Feulner et al. / Biochemical Systematics and Ecology 39 (2011) 732–743 739 51 data confirmed the hybrid origin of the S. latifolia taxa, they were 1.3 to 1.5 times more closely related to S. aria agg. than to S. torminalis. The S. aria agg. showed a complicate genetic structure and fell into four main groups, two intermediate groups besides S. pannonica and S. aria s.str. Some progeny of S. pannonica was more variable than expected and clustered partly with other groups indicating gene flow within S. aria agg. Different subgroups of the S. aria aggregate may be parental for the S. latifolia taxa, contributing to the remarkable genetic distances between them. Key words Sorbus, Northern Bavaria, genetic structure, parental species identification, AFLP Introduction In Central Europe, the genus Sorbus L. (Rosaceae) includes several widespread, diploid (2n = 34; Liljefors 1955, Düll 1959) and sexually reproducing species, namely S. aria (L.) Crantz, S. torminalis (L.) Crantz, and S. aucuparia L. Spontaneously sexually reproducing and out-crossing hybrids can be found between S. aria and S. torminalis and between S. aria and S. aucuparia; however, these hybrids are rare (comp. Aas et al. 1994, Meyer et al. 2005). Besides such unstable hybrids (i.e. S. x tomentella Gand., S. x pinnatifida (Sm.) Düll), an impressive number of stable hybrids have been described as endemic apomictic microspecies from many areas of Europe. A high diversity of endemic Sorbus microspecies has been reported especially from Britain (Rich et al. 2010, Robertson et al. 2010), the Czech Republic (Karpati 1960, Lepší et al. 2009), and from parts of southern Germany, in particular Thuringia and Northern Bavaria (Düll 1961, Meyer et al. 2005). Most of these microspecies show a limited distribution, but contribute considerably to the local species diversity and therefore attract increasing notice of species protection efforts (Meyer et al. 2005). The S. latifolia (Lam.) Pers. aggregate comprises microspecies derived by hybridization between S. aria agg. and S. torminalis (Düll 1961, Challice and Kovanda 1978, Aas et al. 1994). Members of this aggregate presumably originated polytopically in the postglacial period (Düll 1961). The origin of the Northern Bavarian S. latifolia taxa was not yet investigated in detail (see Meyer et al. 2005). Parental species identification is complicated by the fact that in the study area the S. aria aggregate (agg.) consists of a wider range of taxonomically not sufficiently investigated forms (see below). In Bavaria, so far 17 microspecies of S. latifolia agg. have been recognized (cf. Meyer et al. 2005). Here, we focus on three microspecies endemic to the northern Franconian Jura (Germany, 58 Bavaria), namely Sorbus franconica Bornm. ex Düll, S. cordigastensis N. Mey. and Sorbus adeana N. Mey. These taxa are very similar, they differ to some extent in leaf and fruit morphology, mainly in the size and shape of the leaves, the number of lateral veins and the color and shape of the fruit (for details see Meyer et al. 2005). Morphologically, they resemble Sorbus aria agg. more than Sorbus torminalis (Meyer et al. 2005). All three investigated S. latifolia-taxa are distributed parapatrically in the study area (see Meyer et al. 2005). The distribution areas of S. adeana and S. cordigastensis are situated close to each other (distance about 10 km) and are restricted to a few square kilometers only (Meyer et al. 2005, Aas and Kohles 2011), whereas S. franconica has a much wider range in the Franconian Alb more southward and distant from the two other taxa. The taxa regularly grow sympatrically with species from the Sorbus hybrida group (hybrid taxa between S. aria and S. aucuparia), S. torminalis or with members of S. aria agg., such as S. pannonica Kárpáti. Sorbus pannonica is a xeromorphic member of S. aria agg., and is more widespread in the northern Franconian Alb than S. aria s.str. (Kutzelnigg 1995, Meyer et al. 2005). It is a non-typified taxon, which comprises presumably apomictic morphotypes filling the morphological gap between S. aria s.str. and S. graeca (Spach) Loddiges ex Schauer (Kárpáti 1960, Kutzelnigg 2005, Meyer et al. 2005). Sorbus graeca, another xeromorphic member of S. aria agg., is mainly distributed in southern, southeastern and eastern Europe. It can reproduce sexually or facultatively apomictic (Kutzelnigg 1995). It is uncertain whether S. graeca occurs in the study area (Düll 1961, Kutzelnigg 1995, 2005), but individuals that are very similar to S. graeca have been found in the northern Franconian Alb (own obs.); yet, it is difficult to delimitate S. graeca against S. pannonica. In this paper the AFLP technique is used to investigate the genetic structure of S. latifolia taxa and its probable parents in northern Franconia. Questions addressed include (1) how wide is the genetic distance between the S. latifolia taxa; (2) do S. latifolia taxa have a clonal structure and do they reproduce as apomicts; (3) which member of S. aria agg. - besides S. torminalis - is most likely parental for these S. latifolia taxa? Additional chromosome counts give insights in the cytology of the taxa. 59 Material and Methods Plant material For AFLP-analyses we collected leaf material from Sorbus cordigastensis (one of one locality) and S. adeana (at one of the three known localities, comp. Meyer et al. (2005)) and from two populations of S. franconica. Sampled individuals were chosen randomly. Leaf material was also collected from four to seven populations of parental species cooccurring or coming close to the microspecies distribution area. Seven plants of S. torminalis from different localities were harvested. From the S. aria agg. we collected samples at two populations of S. pannonica. Plants from one site (Neudorf) are similar to S. graeca (roundish leaves as broad as long, serration as long as broad, comp. Düll 1961, Kutzelnigg 1995) and plants from the locality "Kordigast" are affiliated to the typical form of S. pannonica termed “tennis racket” by local botanists because it has obovate oblong leaves. Furthermore, we collected plants from four populations in the contact area of S. aria and S. pannonica in the north-western part of the Franconian Alb. There, populations included plants that could be clearly affiliated to S. aria s.str., but also plants with morphological similarity to S. pannonica. Such intermediates were also found within the range of S. cordigastensis. These intermediates were morphologically deviating from the thin-leaved S. aria s.str. by rough leaves and variable leaf shapes ranging from ovate to orbicular. From S. pannonica they differed in usually having more veins on their leaves. Additionally, material from seedlings of the investigated taxa were included into this study (Table 1). To this purpose, seeds of three mother trees of S. adeana, S. cordigastensis, S. franconica, S. pannonica and S. aria-S. pannonica intermediates were harvested in autumn 2009 from the same populations chosen for the investigation of adults. Seeds were germinated and plants were grown in the Ecological-Botanical Gardens (University of Bayreuth, EBG). Four seedlings from each mother tree were analyzed. Table 1 Taxa, site locality and voucher information of the individuals analysed. Taxon Locality // Gauss Krüger coordinates Individuals Taxon code, Voucher number Autobahn Rossdorf // 4437013/ 5539920 5 aria70-74 Wattendorf // 4434614/ 5543670 5 aria75-79 Grafenhäusling // 4437746/ 5542273 5 aria80-84 Sorbus aria (L.) Crantz Rossdorf // 4437884/ 5540559 5 aria85-89 60 Kordigast // 4443782/ 5551720 5 panK8-panK10, panK13, panK34 S. pannonica Kárpáti Neudorf // 4447194/5546549 4 panK22, panK24, panK28, panK29 Offspring S. pannonica Kordigast // 4443782/ 5551720 12 (MT 7) panOff154-157 (MT 8) panOff150-153 (MT 9) panOff146-149 S. adeana N. Mey. Neudorf // 4447194/5546549 7 ade004, ade005, ade006, ade023, ade025, ade027, ade030 Offspring S. adeana Neudorf // 4447194/5546549 11 (MT 3) adeOff162adeOff165 (MT 4) adeOff166, adeOff167, adeOff 169 (MT 6) adeOff090adeOff093 S. franconica Bornm. ex Düll Brünnberg // 4457310/ 5520910 2 franc067, franc069 Muggendorf // 4447515/ 5518390 5 franc044, franc047, franc 051, franc053, franc064 Offspring S. franconica Muggendorf // 4447515/ 5518390 12 (MT 2) francOff102francOff 105 (MT 3) francOff122franc Off125 (MT 4) francOff126francOff 129 S. cordigastensis N. Mey. Kordigast // 4443782/ 5551720 8 cord003, cord011, cord014, cord017, cord018, cord020, cord031, cord039 Offspring S. cordigastensis Kordigast // 4443782/ 5551720 12 (MT 1) cordOff110cordOff113 (MT 2) cordOff106cordOff109 (MT 7) cordOff130cordOff133 Gottelhof // 4451879/5530456 1 tor001 Neudorf // 4447194/5546549 1 tor021 Hainbach // 4451892/ 5530457 1 tor002 Muggendorf // 4447515/ 5518390 2 tor042, tor046 S. torminalis (L.) Crantz Kordigast // 4443782/ 5551720 2 tor012, tor033, tor043 61 Molecular methods DNA extraction Leaf samples were taken in May and June 2010. Immediately after harvesting they were placed in plastic bags and put in a box with ice for transportation. At the same day, leaves were washed with ethanol in the laboratory and frozen in an extraction tube at –80 ° C until extraction. Frozen leaf samples (40 – 70 mg, 1 – 2 cm2) were blended in 200 µl extraction buffer (NucleoMag 96 Plant kit; Machery-Nagel, Düren, Germany, containing 5 µl RNAse A) with a FastPrep®-24 Tissue Homogenizer (MP Biomedicals Europe, Illkirch, France) for 40s at a speed of 6m/s. Insolubles were pelleted at 15,000 x g for 5 min at room temperature. Genomic DNA was prepared from the supernatant using the NucleoMag 96 Plant kit adapted to the KingFisher automated purification system (Thermo Scientific, Langenselbold, Germany). Details of the nucleic acid purification procedure are presented in Table A1 of the supporting information. The purified genomic DNA was diluted tenfold and used for all subsequent PCR reactions. AFLP analysis AFLP analysis was conducted following the method of Vos (1995) using the IRDye Fluorescent AFLP Kit for Large Plant Genome Analysis (Li-Cor-Biosciences, Bad Homburg, Germany). All reactions were conducted as described in the Li-Cor application manual, but adapted in the following manner: a) 200 ng genomic DNA were used for the EcoRI/MseI restriction digestion, which was extended to 16 hours; b) a 1:4 (instead a 1:10) dilution of the adaptor-DNA ligation mixture was used for preamplification reactions; c) a 1:40 dilution of the preamplification reaction was used for selective amplification. For a preliminary primer search 24 primer combinations were tested, and the following six combinations were then selected for this study because they yielded the best results in species differentiation: M-CAA/E-ACG, M-CAC/E-ACG, M-CAC/E-ACA, M-CAT/EACG, M-CTC/E-ACG, M-CTT/E-ACG. Image collection and analysis The reactions were separated on a vertical electrophoresis system (4200 Sequence Analysis System, Li-Cor Biosciences, Bad Homburg) together with DNA size markers (50–700 bp Sizing Standard, Li-Cor Biosciences, bad Homburg). AFLP banding patterns were evaluated using GeneMarker1-95 software (SoftGenetics) and a presence-absence matrix 62 was constructed. Also, unique bands were scored and were not left out from analysis. Band classes were calculated with a tolerance factor of 0.1 %. Chromosome counts Chromosome numbers were counted from root tip meristems of one seedling, respectively from three to four mother trees (MT) of S. cordigastensis (MT 2,5,6,7), S. adeana (Neudorf MT 3,4,6), S. franconica (MT 2,3,4, s.n.) and S. pannonica (locality Neudorf, MT 2,5, locality Kordigast MT 8,9), grown in the EBG and harvested in May 2010. Some of the seedlings were also included in the AFLP analysis (Table 1). The fresh root tips were pretreated in 0.002 hydroxychinoline (4hrs), fixed in CARNOY´s solution and stained in carmine after Snow (1963). From the stained root tips we prepared squash preparations in 45% acetic acid, and observed somatic metaphase plates in the microscope. Statistical analyses A neighbour joining (NJ) analysis of the presence - absence matrix was conducted (Nei Li distance), followed by bootstrap (BS) analysis after internode rooting with 1000 replicates using the program TREECON (Van de Peer and De Wachter 1994). The tree was rooted with a S. torminalis individual. The taxon clades or subgroups (in case of S. aria agg.) revealed were used for all further calculations (Fig. 1). For data of S. aria agg., we additionally applied model-based clustering (Pritchard et al. 2000) using the program STRUCTURE (http://pritch.bsd.uchicago.edu/structure.html) in order to retrieve the most likely number of groups within the S. aria aggregate. For AFLP data the recessive allele criterion was used and set to 1. Data were analysed as diploid because we did not know the exact ploidy levels of the plants, which varies between diand tetraploid. A total of 10 independent runs with K set to 2-10 using the admixture model option with correlated frequencies (prior mean FST ¼ 0.1 equal for all populations) were performed. The most likely number of groups is characterized by a maximum posterior probability Ln P(D) and the highest stability of results revealed from each of ten runs (comp. Pritchard et al. 2000; Falush et al. 2003, Gugerli et al. 2008). A burn-in of 50,000 steps followed by 50,000 iterations gave stable results after testing different burn-in periods and iterations. STRUCTURE calculates also the proportion of an individual genotype originating from each of the K groups (= q). The individuals were assigned to each of the K groups using a threshold of q of 0.3-0.8 or higher. 63 A Principal Coordinate analysis (PCo) of the data was conducted with PRIMER (Jaccard Index) (Clarke and Gorley 2001). In order to investigate genetic variability, the number of polymorphic loci and Nei´s gene diversity “NGD” (Nei 1972) were calculated with POPGENE (Yeh and Yang 1999) setting the program routines for a diploid, dominant marker data set. It was assumed that the NGD of probable apomictic taxa should be clearly lower than the one of sexual taxa (comp. Nybom and Bartish 2000), and the NGD of seedlings of apomicts should not exceed the NGD of the adults. As a measure for the genetic distance between taxa we calculated Nei´s standard genetic distance (Ds) using the program POPGENE (Yeh and Yang 1999). For the S. latifolia taxa the proportions between the genetic distances to S. aria agg. and to S. torminalis were calculated to find out to which parent they are genetically more closely related. For the S. aria agg. we calculated the distances to the four subgroups revealed in the NJ tree (see Fig. 1). We also tested the calculation using more subgroups as indicated by Bayesian clustering, however, the results did not deviate and therefore we not show them here. Voucher study Vouchers (lateral shoots), that were simultaneously collected with the material for AFLP analyses from individuals of each S. aria agg. subgroup were deposited in the herbarium UBT, and morphologically analysed regarding shape of broadest leaves, number of veins and serration (Fig. 3). Results AFLP analyses 578 markers were identified from six primer combinations. The percentage of polymorphic loci of the S. latifolia taxa varied between 6.75 % and 22.32 % (Table 3). With 52.42 % the intermediate S. aria - pannonica group aff. S. aria s.str. reached the highest percentage of polymorphic loci. S. aria s.str. reached 43.25 %, S. torminalis 26.99 % (Table 2). Neighbour joining-tree and Bayesian clustering for S. aria agg. In the neighbour joining tree (Fig. 1) the S. latifolia taxa grouped between S. aria agg. and S. torminalis, but closer to S. aria agg. The individuals of S. aria agg. and S. torminalis were much more strongly dissimilar than the individuals of the putative microspecies S. adeana, S. cordigastensis and S. franconica (Fig. 1). All S. latifolia accessions constituted 64 highly supported groups of their own (BS 100 for S. adeana, 84 for S. cordigastensis, and 97 for S. franconica), and were clearly separate from each other. The offspring of the S. latifolia taxa clustered with the adults and the progeny of a single mother tree clustered together. The Sorbus aria agg. clade was strongly supported (BS 99) and could be divided into four groups, the S. pannonica group, the S. aria s.str. group and two groups of intermediates (Fig. 1). One group of intermediates (aria76, aria77, aria79, panOff150, panOff 151, panOff153, panOff156) was retrieved more closely to the S. pannonica group and is therefore called “intermediate group affine (aff.) pannonica” (Fig. 1). Another group was retrieved more closely to S. aria s.str., it is therefore referred to as “intermediate group aff. S. aria”. This group consists of two subclades (the one with aria074, aria075, aria082, aria086 and the other one with aria070, aria073, aria078) (Fig. 1). The offspring of two mother trees of S. pannonica is partly grouped within S. pannonica and partly within the intermediate group aff. pannonica (Fig. 1). In the Bayesian clustering of the S. aria agg. the Ln P(D) was highest and standard deviation of posterior probabilities was lowest for seven groups, (Ln P(D) for K = 7 between -7915 and -8085; 10 runs). The assignment of individuals to seven groups was very constant and the groups were in general in good accordance with the groups and subgroups revealed by the NJ tree. Bayesian groups were identic to the NJ tree subgroups for aria074, aria075, aria082 (see above), aria087, aria088, aria089 or aria076, aria077, aria079, pannoff150, pannOff151, pannOff153. Individual aria086 formed a group of its own. 65 Fig. 1 Neighbour joining tree with Nei Li distance and bootstrap values with 1000 replicates using TREECON. Offspring individuals are shadowed dark-gray. 66 Voucher study Voucher studies of S. aria agg. revealed that the intermediates aff. S. aria s.str. deviated from S. aria s.str. (Fig. 3a) by a thicker leaf texture. Their leaf shape was variable, ovate or obovate (Fig. 3c). The intermediates aff. S. pannonica (Fig. 3d) had more leaf veins (about 10–11) than S. pannonica (8 veins, Fig. 3b) and the serration of the leaf margins extended to the leaf base; furthermore, the leaf shape was roundish instead of obovate and the leaf base was rounded and not cuneate as in S. pannonica. Nei´s gene diversity Nei´s gene diversity (NGD) of the taxa and subgroups of S. aria agg. is shown in Table 2. Gene diversity values of the S. latifolia group (adults and seedlings) were rather similar. They varied for S. adeana between 0.029 and 0.083, for S. cordigastensis between 0.051 and 0.088, and for S. franconica between 0.041 and 0.088 (see Table 2). The genetic diversity in the S. aria aggregate differs remarkably. It was high for the plants identified as S. aria s.str. (0.129) and for plants of the intermediate group with affinity to S. aria s.str. (0.177). It was lower for the intermediate S. aria-S. pannonica group with affinity to S. pannonica (0.098) and very low for S. pannonica (between 0.027 and 0.070). Genetic diversity was also relatively low in the second putative parental taxon S. torminalis (NGD 0.089). None of the progeny was found to be identical – possibly a consequence of band reproducibility, because also unique bands were scored (cf. Material and Methods). However, the genetic diversity of the progeny of S. franconica and S. adeana did not exceed the values of the adults. NGD exceeded somewhat the values of the adults in case of S. pannonica progeny of mother tree 8, (NGD 0.070) and S. cordigastensis progeny of mother tree 2 (NGD 0.088) (Table 2). 67 a) S. aria s.str. (aria087*) 12 veins, leaf texture thin, leaf margin serrated from the base, leaf base rounded. b) S. pannonica (pannN029*) 8 veins, leaf texture thick, broader than long serrated, leaf base cuneate, not serrated. c) Intermediate aff. S. aria (aria075*) 11 veins, leaf texture moderately thick, leaf margin serrated from the base, leaf base cuneate. d) Intermediate aff. S. pannonica (aria079*) 11 veins, leaf texture thick, leaf margin serrated from the base, leaf base rounded. Fig. 3 Leaf shape of the broadest leaf of lateral shoots from selected vouchers of S. aria agg. investigated in this study incl. additional information about leaf characters and group affiliation (*for abbreviations see Table 1). Acknowledgments The financial support of this study by the Oberfrankenstiftung Bayreuth (OFS Project-No. 03086) is gratefully acknowledged. We thank Michaela Hochholzer (DNA Analytics, University of Bayreuth) for conducting AFLP laboratory work, and Francesco Balao for supporting statistical analysis. Martin Lepší and Norbert Meyer deserve thanks for determination of herbarium material. 74 Literature Aas G, Kohles M (2011) Verbreitung, Häufigkeit und Verjüngung von Sorbus cordigastensis (Kordigast-Mehlbeere) in der nördlichen Frankenalb. Tuexenia 31:59 - 71 Aas G, Maier J, Baltisberger M, Metzger S (1994) Morphology, isozyme variation, cytology, and reproduction of hybrids between Sorbus aria (L.) Crantz and S. torminalis (L.) Crantz. Bot Helv 104:195–214 Challice J, Kovanda M (1978) Flavonoids as markers of taxonomic relationships in the genus Sorbus in Europe. Preslia 50:305-320 Clarke KR, Gorley RN (2001) Primer v5: User Manual/Tutorial. Primer-E Ltd. Düll R (1959) Unsere Ebereschen und ihre Bastarde. Wittenberg. Ziemsen. Düll R (1961) Die Sorbus-Arten und ihre Bastarde in Bayern und Thüringen. Ber Bayer Bot Ges 34:11-65 Gugerli F, Brodbeck S, Holderegger R (2008) Utility of multilocus genotypes for taxon assignment in stands of closely related European white Oaks from Switzerland. Ann of Bot-London 102:855-863 Kárpáti Z (1960) Die Sorbus-Arten Ungarns und der angrenzenden Gebiete. Feddes Repert. 62:71-334 Kutzelnigg H (1995) Sorbus. In: Conert HJ et al (eds): Gustav Hegi, Illustrierte Flora von Mitteleuropa 4 (2B): Spermatophyta: Angiospermae: Dicotyledones 2 (3). Rosaceae 2. Blackwell, Berlin, pp 328-385 Kutzelnigg H (2005) Sorbus L em Crantz – Mehlbeere, Eberesche, Elsbeere, Speierling. In: Jäger E J, Rothmaler KW (eds) Exkursionsflora von Deutschland 4, Gefäßpflanzen, Kritischer Band 9th edn. Spektrum, Heidelberg, pp 403-406 Lepší M, Vít P, Lepší P, Boublík K, Kolář F (2009) Sorbus portae-bohemicae and Sorbus albensis, two new endemic apomictic species recognized based on a revision of Sorbus bohemica. Preslia 8:63–89 Liljefors A (1955) Cytological studies in Sorbus. Acta Horti Bergiani 17:47-113 Meyer N, Meierott L, Schuwerk H, Angerer O (2005) Beiträge zur Gattung Sorbus in Bayern. Ber Bay Bot G Sonderband:5-216 Nei M (1972) Genetic distance between populations. American Naturalist 106:283–291 Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using Multilocus Genotype Data. Genetics 155:945-959 75 Rich T, Houston L, Robertson A (2010) Whitebeams, Rowans and Service Trees of Britain and Ireland A Monograph of British and Irish Sorbus L. BSBI Handbook 14, London. Robertson A, Rich TCG, Allen AM, Houston L, Roberts C, Bridle JR, Harris S A, Hiscock JS (2010) Hybridization and polyploidy as drivers of continuing evolution and speciation in Sorbus. Mol Ecol 19:1675–1690 Snow R (1963) Alcoholic hydrochloric acid-carmine as stain for chromosomes in squash preparations. Stain Technol 38:9-13 Talent N (2009) Evolution of gametophytic apomixis in flowering plants: an alternative model from Maloid Rosaceae. Theory Biosci 128:121-38 Van de Peer Y, De Wachter Y (1994) TREECON for Windows: a software package for the construction and drawing of evolutionary trees for the Microsoft Windows environment. Comput Applic Biosci 10:569-70 Vos P, Hogers R, Bleeker M (1995) AFLP: a new technique for DNA fingerprinting. NAR 23: 4407–4414 Yeh F, Yang R (1999) POPGENE. Microsoft Window-based Freeware for Population Genetic Analysis. Version 1.31, University of Alberta: Edmonton, Alberta, Canada 76 Supporting information Table S1: Plate layout and instrument settings for DNA purification via BindIT 3.1 KingFisher software. After step 7, eluted DNA was transferred to plastic cups and kept at -20°C. Settings Step 1 Bind MC2 Step 2 Wash MC3 Step 3 Wash MC4 Step 4 Wash Ethanol Step 5 Wash MC5 Step 6 Elution MC6 Step 7 Disposal of magnetic beads Plate A B C D E F G Layout 92 µl supernatant 8 µl magnetic beads 100 µl buffer 200 µl buffer 200 µl buffer 200 µl 80% Ethanol 200 µl buffer 20 µl buffer Precollect no no no no no no Release time [mm:ss] no release 00:30 00:30 00:30 no release 00:15 00:20 Beginning Release speed - fast fast fast - fast fast Pause for manual handling no no no no no no Mixing time [mm:ss] 05:00 01:00 01:00 01:00 01:00 10:00 Mixing/ pause Mixing speed medium fast fast fast medium medium Postmix no no no no no no Collect count 3 3 3 3 3 6 End Collect time [s] 1.5 1.5 1.5 1.5 1.5 1.5 77 Publication 4 4. Floral scent and its correlation with genetic data in Sorbus taxa. In preparation for submission to Organisms Diversity & Evolution 78 Floral scent and its correlation with genetic data in Sorbus taxa Martin Feulnera,*, Stefan Pointnera, Lisa Heussa, Gregor Aasb, Stefan Dötterl a,c aDepartment of Plant Systematics, University of Bayreuth Universitätsstr. 30, 95440 Bayreuth, Germany * Author for correspondence: [email protected] b Ecological-Botanical gardens, University of Bayreuth Universitätsstraße 30, 95447 Bayreuth, Germany email: grego[email protected] c Present address: Universität Salzburg, FB Organismische Biologie Hellbrunnerstrasse 34, 5020 Salzburg email: [email protected] Abstract Statistical analyses between floral scent and genetic marker data for testing their taxonomical correlation are rare. We investigated inflorescence scent patterns of apomictic Sorbus latifolia microspecies, Sorbus franconica, S. adeana and S. cordigastensis endemic to northern Bavaria, originated by hybridization and their parental taxa with dynamic headspace method. The scent data (presence-absence of compounds) were used to construct an UPGMA tree, and to calculate a similarity matrix to correlate them, both on individual as well as population level, with AFLP data published in an earlier study. Scent analyses showed a total of 68 chemical substances, among them aromatic compounds, monoand sesquiterpenes, aliphatics, and nitrogen containing compounds. Scent patterns were taxon specific, and the number of scent components differed among taxa. Correlations with AFLP data on population and individual level are highly significant, indicating that scent and genetic data are highly congruent in the plants studied. Scent therefore is a useful marker for taxonomical questions in Sorbus. Keywords Floral scent, apomixis, AFLP, Sorbus, taxonomy, correlative analysis; 79 Introduction Studies of sexual (i.e. Levin et al. 2003; Raguso et al. 2006), but also of apomictic species complexes (Feulner et al. 2009; 2011) revealed that taxonomic conclusions based on scent can be highly congruent with those based on genetic markers. The main function of floral scent is the attraction of pollinators (Dötterl et al. 2006; Plepys et al. 2002). Therefore, scent data may be influenced by pollinator mediated selection (Knudsen and Tollsten 1993; Plepys et al. 2002; Dötterl et al. 2005). One consequence of this is the evolution of pollination syndromes which means that plant species pollinated by the same guild of animals have similar phenotypes of their floral characteristics including scent (Faegri and van der Pijl 1979; Fenster et al. 2004; Dobson et al. 2005). Although scent is influenced by pollinator-mediated selection and coevolution, in most studies it was shown that only a limited number of substances have key functions in attracting pollinators (Dötterl et al. 2006, Svensson et al. 2010, Burger et al. 2012), whereas other substances may be more determined by phylogeny than by pollinator-mediated selection (Steiner et al. 2011; Schäffler et al. 2012). There are some examples in which scent data supported taxonomy and revealed a good conformity with DNA data (i.e. Levin et al. 2003; Raguso et al. 2006). However, to our best knowledge, a statistical approach of the taxonomical value of scent by detailed correlations between scent data and data from genetic markers was so far only presented once in a study dealing with Ophrys (Orchidaceae, Stökl et al. 2008). Here, however, significant correlation between both data sets was not found (Stökl et al. 2008). Scent data in apomicts may behave differently compared to sexual species, because apomictic plants produce seeds without fertilisation and do not rely on pollination (Nogler 1984; Jankun and Kovanda 1987; Talent 2009). Therefore, pollinator-mediated selection influencing scent patterns is of minor importance in apomicts (comp. Feulner et al. 2009; 2011). Furthermore, the intra-individual genetic variability of apomicts is extremely low, therefore, scent patterns also may be strongly identical between individuals and populations of the same apomictic taxon. Apomixis is often coupled with hybrid speciation (Talent 2009) as it is the case in the Sorbus latifolia group (Rosaceae). Sorbus latifolia taxa originated from hybridization between S. aria agg. and S. torminalis (Düll 1961; Rich et al. 2010, Feulner et al. 2013; submitted). Among members of the Sorbus latifolia agg. are many taxa endemic to restricted regions in i.e. Great Britain, Czech Republic, and Germany (Düll 1961; Lepší et al. 2009, Meyer et al. 2005; Robertson et al. 2010; Rich et al. 2010). 80 Interestingly, it has been shown in other hybrid complexes such as Citrus, Ophrys and Hieracium, that the scent consists mainly of a mixture of scent components of the parental species (Gancel et al. 2002; Vereecken et al. 2010; Feulner et al. 2009; 2011), and only a low number of new compounds. Therefore, scent analyses may be a valuable tool for parental species identification, and indeed, the taxonomic reliability of scent patterns was shown to be high in groups originated by hybridization (Feulner et al. 2009; 2011). Here, we investigate the scent of the apomictic microspecies S. adeana, S. cordigastensis, and S. franconica belonging to the S. latifolia aggregate endemic to Northern Bavaria and occupying very small parapatric distribution areas. In a former study the hybrid state, the intraspecific variability and the genetic structure of parental taxa was investigated with AFLP analyses (Feulner et al. 2013; submitted). In the present study, based on the same material investigated in the AFLP study (Feulner et al. 2013; submitted), we investigated floral scent composition and correlated scent clustering with the AFLP tree, to estimate the correlation between both data sets. Material and method Study plants We collected scent from S. adeana (one of one known population, comp. Meyer et al. 2005), S. cordigastensis (one of approximately three known populations), S. franconica (two of > 50 known populations) as well from parental taxa such as S. aria s.str., S. pannonica and S. torminalis (comp. Feulner et al. 2013; submitted). From the S. aria agg. also intermediates between S. aria s.str. and S. pannonica with affinity to S. aria s.str. (comp. Feulner et al. 2013; submitted) were included into the study. These plants are appelated as aff. aria. For all taxa, AFLP data were available from the same populations (Feulner et al. 2013, submitted) and in 11 cases AFLP data and scent data were collected from the same individuals (Table 1, Feulner et al. 2013, submitted). For further information about taxonomy, population structure and ecology of the taxa investigated, see Feulner et al. (2013, submitted). 81 Table 1: Taxa, locality and voucher information of the individuals analysed. (* compare Feulner et al. 2013, submitted). Taxon Locality / Gauss Krueger coordinates / Voucher number of individuals for scent sampling number of AFLP samples from the same population/individual as used for scent sampling* Sorbus aria (L.) Crantz Grafenhäusling 4437746/ 5542273, Feulner 200–207 (UBT) 2 5/0 Intermediates aff. aria s.str. Autobahn Rossdorf 4437013/ 5539920, Feulner 208–217 (UBT) 4 5/0 Kordigast 4443782/ 5551720, Feulner 218–227 (UBT) 4 5/1 Neudorf 4447194/5546549, Feulner 228–234 (UBT) 1 4/0 S. pannonica Kárpáti Brünnberg 4457310/ 5520910, Feulner 235 (UBT) 1 – S. adeana N. Mey. Neudorf 4447194/5546549 Feulner 236 (UBT) 4 7/0 S. franconica Bornm. ex Düll Brünnberg 4457310/ 5520910 Feulner 237–239(UBT) 1 2/0 Mugggendorf 4447515/ 5518390 Feulner 240–245 (UBT) 4 5/3 S. cordigastensis N. Mey. Kordigast 4443782/ 5551720, Feulner 246–253 (UBT) 8 7/8 Neudorf Bärental 4447194/5546549, Feulner 254 (UBT) 1 1/0 Hainbach // 4451892/ 5530457, Feulner 255 (UBT) 1 0/1 S. torminalis (L.) Crantz Kordigast 4443782/ 5551720, Feulner 256–259 (UBT) 2 3/0 Volatile collection Inflorescence scent was collected in the field using a standard dynamic head-space method as described in Feulner et al. (2009). For each taxon, two to six individuals were sampled. 82 Sampling was carried out on fresh and newly opened inflorescences (one inflorescence per plant and sample), between 11 a.m. and 3 p.m, the period with the most intensive scent emission (as determined by the human nose; Feulner, unpublished data). Scent samples of leaves and surrounding air were collected as control for each locality and population investigated. Chemical analysis The samples were analysed on a Varian Saturn 2000 mass spectrometer, and a Varian 3800 gas chromatograph with a 1079 injector, that had been fitted with the ChromatoProbe kit. This kit allows the thermal desorption of small amounts of solids or liquids contained in quartz microvials (Micro-SPE; cf. Amirav and Dagan, 1997; Dötterl et al. 2005). The injector split vent was opened (1/20) to flush any air from the system and closed after 2 minutes; the injector was heated with 40 °C for 2 min, and the temperature was then increased with a rate of 200 °C/min to 200 °C; this end temperature was held for 4.2 min, after which the split vent opened (1/10) and the injector cooled down. A ZB-5 column (5% phenyl polysiloxane) was used for the analyses (60 m long, inner diameter 0.25 mm, film thickness 0.25 µm, Phenomenex). Electronic flow control was used to maintain a constant helium carrier gas flow of 1.8 ml min-1. The GC oven temperature was held for 7 min at 40 °C, then increased by 6 °C per min to 250 °C and held for 1 min. The MS interface was 260 °C and the ion trap worked at 175 °C. The mass spectra are taken at 70 eV (in EI mode) with a scanning speed of 1 scan s-1 from m/z 30 to 350. Data analysis The GC-MS data were processed using the Saturn Software package 5.2.1. Component identification was carried out using the NIST 08 mass spectral data base or MassFinder 3, and confirmed by comparison of retention times with published data (Adams 2007). Identification of individual components was confirmed by comparison of both mass spectrum and GC retention data with those of authentic standards. Statistical analysis A cluster analysis (UPGMA) based on Jaccard similarity index (calculated using the presence-absence of compounds) was constructed in PRIMER Vers. 6 (Clarke and Gorley 2006). The taxon specificy of scent was tested with ANOSIM (10.000 pemutations) in 83 reduced influence of pollinator selection this fact is leading to a strong correlation between scent data and genetic data, as has been shown above. Most studies comparing scent and molecular data are based on Orchidaceae. However, this is a strongly contrasting system including many deceptive plants where pollinator selection plays a key role (Salzmann et al 2007; Vereecken et al. 2010). Deceptive plants usually have an increased variability of scent (Salzman et al. 2007), likely to avoid that pollinators can easily learn to discriminate between the reward and the mimic (Ackerman et al. 2011). Therefore it is not astonishing that in Ophrys no correlation of scent with genetic data was found (Stökl et al. 2008). Table 3: Presence/absence of floral scent volatiles, occurring in n of all individuals investigated (n/n) of 7 Sorbus taxa. aria aff. aria aria s.str. pannonica torminalis adeana cordigast -ensis franconica aromatics benzaldehyde 4/4 2/2 6/6 4/4 4/4 8/8 5/5 benzeneacetaldehyde 3/4 1/2 3/6 3/3 4/4 4/8 - methyl benzoate 4/4 2/2 6/6 - 4/4 8/8 5/5 2-phenylethyl alcohol 4/4 2/2 6/6 4/4 4/4 8/8 5/5 methyl phenylacetate 4/4 2/2 6/6 - 4/4 8/8 3/5 methyl salicylate 4/4 2/2 6/6 4/4 4/4 8/8 5/5 anisaldehyde 2/4 1/2 2/6 1/4 - - - aliphatics methyl isovalerate 4/4 2/2 5/6 - 4/4 - 2/5 2,3-butandiol 4/4 2/2 6/6 - - - - (Z)-3-hexen-1-ol 4/4 2/2 6/6 3/4 4/4 8/8 5/5 methyl hexanoate - - 4/6 - 4/4 8/8 3/5 methyl (3Z)-hex-3-enoate - - 2/6 - 4/4 8/8 2/5 methyl 2-hydroxy-3-methylpentanoate 4/4 2/2 6/6 - 4/4 - - (Z)-3-hexen-1-ol acetate 4/4 2/2 6/6 4/4 4/4 8/8 5/5 acetic acid hexyl ester - - 4/6 3/4 1/4 8/8 - (E)-2-hexen-1-ol acetate - - 4/6 - - - 5/5 octanal 4/4 2/2 6/6 4/4 4/4 - 5/5 Homoterpenes (E)-4,8-Dimethyl-1,3,7-nonatriene 4/4 2/2 6/6 4/4 4/4 8/8 5/5 Irregular monoterpenes 4-oxoisophorone epoxide 4/4 2/2 6/6 4/4 4/4 8/8 4/5 4-oxoisophorone 4/4 2/2 6/6 4/4 4/4 8/8 5/5 dihydrooxoisophorone 4/4 2/2 6/6 4/4 4/4 8/8 5/5 monoterpenes α-pinene 4/4 2/2 6/6 4/4 4/4 8/8 5/5 camphene 3/4 2/2 6/6 2/4 4/4 8/8 - β-pinene 4/4 2/2 6/6 3/4 4/4 8/8 5/5 (Z)-ocimene 4/4 2/2 6/6 4/4 4/4 8/8 5/5 limonene 4/4 2/2 6/6 4/4 4/4 8/8 5/5 90 eucalyptol 4/4 2/2 6/6 4/4 4/4 8/8 5/5 dihydro-5-methyl-5-vinyl-2(3H)- furanone 4/4 2/2 5/6 1/4 4/4 7/8 - (E)-ß-ocimene 1/4 1/2 5/6 3/4 - 5/8 5/5 (Z)-arbusculone 4/4 2/2 6/6 1/4 1/4 - - (E)-arbusculone 4/4 2/2 6/6 1/4 - - - (Z)-linalol-oxid furanoid 3/4 1/2 4/6 3/4 3/4 - 5/5 (E)-linalol-oxid furanoid 4/4 2/2 6/6 3/4 4/4 - 1/5 lilac aldehyde A 4/4 2/2 5/6 - - - - Lilak aldehyd B+C 4/4 2/2 6/6 - 3/4 8/8 - lilac aldehyde D 4/4 2/2 6/6 - 3/4 3/8 - unkn MT1402 3/4 - 6/6 - - 1/8 4/5 lilac alcohol A 4/4 2/2 2/6 - 4/4 2/8 - lilac alcohol BC 4/4 2/2 6/6 - 4/4 4/8 - lilac alcohol D 4/4 2/2 4/6 - 4/4 2/8 2/5 isomenthone 4/4 1/2 4/6 4/4 3/4 4/8 1/5 lilac derivative 4/4 2/2 6/6 - 4/4 - - linalool 4/4 2/2 6/6 4/4 4/4 - 3/5 N-containing substances 3-pyridinecarboxaldehyde 4/4 2/2 5/6 4/4 4/4 8/8 - amyl/isoamyl-pyrrole - - - 3/4 4/4 2/8 - phenylacetonitrile 4/4 2/2 6/6 4/4 4/4 8/8 5/5 methyl nicotinate 2/4 - - 2/4 4/4 8/8 - unk-N1364 m/z 125, 81, 39 4/4 2/2 6/6 3/4 4/4 8/8 1/5 unk-N1377 m/z 151, 94 - - - 2/4 2/4 - - Unk-N1498 m/z 117,91,65,50,39 4/4 2/2 6/6 - 3/4 8/8 2/5 unk-N1530 m/z 117,91,59,50 4/4 2/2 6/6 - 4/4 7/8 2/5 Indole 4/4 2/2 6/6 - 3/4 - - 1-nitro-2-phenylethane 4/4 2/2 6/6 - 4/4 8/8 3/5 Sesquiterpenes cf. α-longipinene 2/4 - 6/6 - - - - unk-ST1684 m/z 204,161, 91, 69, 55 1/4 - 2/6 - - - - unk-ST1690 m/z 161,119, 85, 73, 58 - - 6/6 1/4 1/4 4/8 5/5 α-Copaene 3/4 - 4/6 - - 3/8 - β-bourbonene 3/4 2/2 6/6 1/4 4/4 5/8 5/5 unk-ST1711 m/z 161,123, 81, 67, 55 2/4 - 6/6 - - 4/8 - unk-ST1732 m/z 161,139, 93, 79 2/4 - - - - - - Longifolene 4/4 2/2 6/6 4/4 3/4 8/8 1/5 Isocomene 2/4 2/2 6/6 - 1/4 8/8 - (E)-β-caryophyllene 4/4 2/2 6/6 4/4 3/4 6/8 5/5 α-gurjunene 4/4 2/2 6/6 - 3/4 5/8 3/5 unk-ST1808 m/z 204,161,143,133,105 - - 2/6 1/4 1/4 4/8 1/5 unk-ST1831 m/z 204,189,161,133,119, 105 cf. Germacrene D 3/4 1/2 6/6 1/4 2/4 6/8 4/5 unk-ST1838 m/z 161,93,41 2/4 1/2 6/6 - 1/4 5/8 5/5 unidentified unk m/z 112,140,181 4/4 2/2 5/6 - 4/4 - - 91 References Aas, G. & Kohles, M. (2011). Verbreitung, Häufigkeit und Verjüngung von Sorbus cordigastensis (Kordigast-Mehlbeere) in der nördlichen Frankenalb. Tuexenia 31, 59–71. Ackerman, J. D., Cuevas, A. A. & Hof, D. (2011). Are deception-pollinated species more variable than those offering a reward? Plant Systematics and Evolution 293: 91–99. Adams, R. P. (2007). Identification of Essential Oil Components by Gas Chromatography/Mass Spectrometry, fourth ed. Allured Publishing Corporation, Carol Stream, IL. Amirav, A. & Dagan, S. (1997). A direct sample introduction device for mass spectrometry studies and gas chromatography mass spectrometry analyses. European Mass Spectrometry, 105–111. Campbell, C. S. & Dickinson T. A. (1990). Apomixis, Patterns of Morphological Variation, and Species Concepts in subfam. Maloideae (Rosaceae). Systematic Botany 15, 124–135. Clarke, K. R. & Gorley, R. N. (2006). Primer v6: User Manual/Tutorial. Primer-E Ltd. Dötterl, S. & Jürgens, A. (2005). Spatial fragrance patterns in flowers of Silene latifolia: Lilac compounds as olfactory nectar guides? Plant Systematics and Evolution 255, 99–109. Dötterl, S., Wolfe, L., W. & Jürgens, A. (2005). Qualitative and quantitative analyses of flower scent in Silene latifolia. Phytochemistry 66, 203–213. Dötterl, S., Jürgens, A., Seifert, K., Laube, T., Weißbecker, B. & Schütz, S. (2006). Nursery pollination by a moth in Silene latifolia: the role of odours in eliciting antennal and behavioural responses. New Phytologist 169, 707–718. Düll, R. (1959). Unsere Ebereschen und ihre Bastarde. Wittenberg. Ziemsen. Düll, R. (1961). Die Sorbus-Arten und ihre Bastarde in Bayern und Thüringen. Berichte Bayerische Botanische Gesellschaft 34, 11–65. Feulner, M., Liede-Schumann, S., Meve, U., Weig A. & Aas, G. (2013). Genetic structure of Sorbus latifolia (Lam.) Pers. taxa endemic to Northern Bavaria – submitted to Plant systematics and evolution (PLSY-D-12-00168). Feulner, M., Schuhwerk, F. & Dötterl, S. (2009). Floral scent analysis in Hieracium subgenus Pilosella and its taxonomical implications. Flora 204, 495–505. Feulner, M., Schuhwerk F. & Dötterl S. (2011). Taxonomical value of inflorescence scent in Hieracium s. str. Biochemical Systemetic and Evolution 39, 732–743. 92 Gancel, A. L., Ollé, D., Ollitrault, P., Luro, F. & Brillouet, J. M. (2002). Leaf and peel volatile compounds of an interspecific citrus somatic hybrid [Citrus aurantifolia (Christm.) Swing.+Citrus paradisi Macfayden]. Flavour and Fragrance Journal 17, 416–424. Jankun, A. & Kovanda M. (1987). Embryological studies of Sorbus 2. apomixis and origin of Sorbus bohemica. Preslia 59, 97–116. Kárpáti, Z. (1960). Die Sorbus-Arten Ungarns und der angrenzenden Gebiete. Feddes Repertorium 62, 71–334. Knudsen, J. T. & Tollsten, L. (1993). Trends in floral scent chemistry in pollination syndromes: floral scent composition in moth-pollinated taxa. Botanical Journal of the Linnean Society 113, 263–284. Kutzelnigg, H. (1995). Sorbus. In Conert, H. J. et al. (Eds.) Gustav Hegi, Illustrierte Flora von Mitteleuropa 4 (2B): Spermatophyta: Angiospermae: Dicotyledones 2 (3). Rosaceae 2. (pp. 328–385). Blackwell, Berlin. Kutzelnigg, H. (2005). Sorbus L em Crantz – Mehlbeere, Eberesche, Elsbeere, Speierling. In Jäger, E. J & Rothmaler, K. W. (Eds.) Exkursionsflora von Deutschland 4, Gefäßpflanzen, Kritischer Band 9rd edn. (pp 403–406). Spektrum, Heidelberg. Lepší, M., Vít, P., Lepší, P., Boublík, K. & Kolář, F. (2009). Sorbus portae-bohemicae and Sorbus albensis, two new endemic apomictic species recognized based on a revision of Sorbus bohemica. Preslia 8, 63–89. Levin, R. A., McDade, L. A. & Raguso, R.A. (2003). The systematic utility of floral and vegetative fragrance in two genera of Nyctaginaceae. Systematic Biology 52, 334– 351. Meyer, N., Meierott, L., Schuwerk, H. & Angerer, O. (2005). Beiträge zur Gattung Sorbus in Bayern. Berichte Bayerische Botanische Gesellschaft Sonderband, 5–216. Nogler, G. A. (1984). Gametophytic apomixis. In Johri, B M. (Ed.) Embryology of angiosperms. (pp. 475–518). Springer-Verlag, Berlin. Plepys, D., Ibarra, F. & Löfstedt, C. (2002). Odour-mediated nectar foraging in the silver Y moth, Autographa gamma (Lepidoptera: Noctuidae): behavioural and electrophysiological responses to floral volatiles. Oikos 99, 75–82. Raguso, R. A., Levin, R. A., Foose, S. E., Holmberg, M. W. & McDade, L. A. (2003). Fragrance chemistry, nocturnal rhythms and pollination‘‘syndromes’’ in Nicotiana. Phytochemistry 63, 265–284. 93 Rich, T., Houston, L. & Robertson, A. (2010). Whitebeams, Rowans and Service Trees of Britain and Ireland A Monograph of British and Irish Sorbus L. BSBI Handbook 14, London. Robertson, A., Rich, T. C. G., Allen, A. M., Houston, L., Roberts, C., Bridle, J. R., Harris, S. A. & Hiscock, J. S. (2010). Hybridization and polyploidy as drivers of continuing evolution and speciation in Sorbus. Molecular Ecology 19, 1675–1690. Salzman, C. C., Nardella, A. M., Cozzolino, S. & Schiestl, F.P. (2007). Variability in floral scent in rewarding and deceptive orchids: The Signature of pollinator-imposed selection? Annals of Botany 100, 757–765. Stökl, J., Schlueter, P. M., Stuessy, T. F., Paulus, H. F., Assum, G., & Ayasse, M. (2008). Scent variation and hybridization cause the displacement of a sexually deceptive orchid species. American Journal of Botany 95, 472–481. Svensson, G.P., Okamoto, T., Kawakita, A., Goto, R. and Kato, M. (2010). Chemical ecology of an obligate pollination mutualism: testing the ‘private channel’ hypothesis in the Breynia-Epicephala association. New Phytologist 186: 995–1004. Vereecken N. J., Cozzolino S. & Schiestl F.P. (2010). Hybrid floral scent novelty drives pollinator shift in sexually deceptive orchids. BMC Evolutionary Biology 10, 103. 94 Danksagung Danksagung Ohne die Hilfe zahlreicher Personen wäre diese Arbeit nicht möglich gewesen. Den nächstgenannten Kollegen bin ich für ihre Unterstützung bei dieser Arbeit besonders zu Dank verpflichtet: Dr. Andreas Jürgens, Prof. Dr. Stefan Dötterl, PD Dr. Ulrich Meve, Prof. Dr. Sigrid Liede Schumann, PD Dr. Gregor Aas. Dr. Andreas Jürgens führte mich in die Methode der Blütenduftuntersuchungen ein. Er bot mir zusammen mit Ulrich Meve, Sigrid Liede-Schumann und Stefan Dötterl am Lehrstuhl Pflanzensystematik die Möglichkeit, meine Forschungen über Blütendüfte zu beginnen. Großer Dank dafür. Ebenso groß ist der Dank an Stefan Dötterl, der mich die Promotionszeit hindurch immer mit Rat und Tat anleitete in allen Fragen rund um das, was man unter Headspacern „Dreck“ nennt, dessen Anwesenheit aber manchmal auch durch mehrmaliges Kopfwaschen nicht gleich „blank“ geputzt werden konnte. Darüber hinaus unter anderem für den sogenannten statistical support, ohne den es nichts geworden wäre. Großer Dank an Ulrich Meve, der durch stetige Unterstützung entscheidend zur Verbesserung der taxonomischen Qualität der Arbeiten beitrug und mit Rat und Tat zur Seite stand. Prof. Dr. Sigrid Liede-Schumann Dank dafür, dass sie mir meine Arbeit am Lehrstuhl Pflanzensystematik durchzuführen ermöglichte, mich immer unterstützte und die Endredigierung meiner Publikationen übernahm. PD Dr. Gregor Aas danke ich für die Zusammenarbeit in Sachen Mehlbeere und so manches Tröpfchen Elsbeerenlikör, manchmal auch Kaffee. Dank an Dr. Franz Schuhwerk, Norbert Meyer, Dr. Günther Gottschlich und Dr. Jochen Müller bei der Unterstützung zu den Hieracien und viele interessante Exkursionen. Dank an Michaela Hochholzer und Dr. Alfons Weig für die Durchführung der AFLP Analysen und das stete offene Ohr. Dank an Dr. Irmgard Schäffler, meiner Zimmerkollegin für die viele praktische Hilfe und Tipps während meiner Arbeit. Ihr, Anne Heiduk und Sabrina Sattler bin ich auch für das Korrekturenlesen dankbar. Dank an Dr. Hannah Burger für die Hilfe bei den schwäbischen Hieracien. Dank an die Regierung von Oberfranken (Herr Neumann), die meine Arbeit zu seltenen endemischen Hieracium und Sorbus-Arten förderten und unterstützten. Dank an die Oberfrankenstiftung für die Finanzierung der AFLP-Analysen. Ganz herzlich möchte ich mich bei dem Lehrstuhlteam für die kollegiale Bereitschaft bedanken. Danke an Angelika Täuber, Barbara Meth, Petra Kraus, Margit Gebauer und Tina Leistner. Die Gärtner des Vertrauens waren Reinhard Krug, Heike Deinlein, Sabine Barthel, und Georg Seidler. Besonders möchte ich mich auch bei meiner Mutter für die Unterstützung bedanken. 95 Eidesstattliche Erklärung Eidesstattliche Erklärung Hiermit erkläre ich, dass ich die Arbeit selbständig verfasst und keine anderen als die von mir angegebenen Quellen und Hilfsmittel benutzt habe. Ferner erkläre ich, dass ich anderweitig mit oder ohne Erfolg nicht versucht habe, diese Dissertation einzureichen. Ich habe keine gleichartige Doktorprüfung an einer anderen Hochschule endgültig nicht bestanden. Bayreuth, den 96